Dissolution Rate and Diffusivity of Lime in Steelmaking Slag and Development of Fluoride-Free Fluxes

Shahriar Haji Amini

A thesis submitted to The Univers ity of New South Wales in total fulfilment of the requirements for admission to the degree of

Doctor of Philosophy
at The University of New South Wales School of Chemical Engineering and Industrial Chemistry & Commonwealth Scientific and Industrial Research Organization Division of Minerals

October 2005

CERTIFICATE OF ORIGINALITY
I herby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, nor material which to a substantial extent has been accepted for the award of any other degree or diploma at UNSW or any other educational institutions, except where due acknowledgment is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis.

I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the projects design and conception or in style, presentation and linguistic expression is acknowledged.

(Signed)(Date)

To my parents:

Professor Ezatollah Haji Amini

and

Touran Nadimpour

If I have seen further... it is by standing upon the shoulders of giants. Issac Newton

ii

ACKNOWLEDGMENT
I would like to express my most sincere thanks to my supervisors; Dr. Sharif Jahanshahi, for his excellent supervision of the work undertaken in this study, A/Prof. Michael Brungs for his guidance and support in facilitating my study as a PhD candidate and Prof. Oleg Ostrovski for his constructive advice throughout the course of this research. I would like to thank Dr. Ling Zhang for many useful discussions. I am also grateful for having had the opportunity to discuss my work with Dr. Shouyi Sun. The work of this project has been aided by the excellent technical support and services provided by staff at; Commonwealth Scientific and Industrial Research organization (CSIRO) division of Minerals, The University of New South Wales and University of Melbourne. I wish to express my appreciation to a number of staff at CSIRO, namely; Rowan Davidson and Justen Bremmel for laboratory assistance, Howard Poynton, Paul Fazey and Daniela Varsamakis for chemical analysis, Dr. Angelica Vecchio-Sadus for OH&S at work place and Damien Hewish for his excellent work on modifying the experimental apparatus. At The University of New South Wales, my thanks to Barry Searle for EPMA analysis and John Sharp for laboratory assistance. At University of Melbourne, my gratitude to Roger Curtin for SEM analysis. I would like to thank my colleagues and friends at CSIRO Minerals, who gave me support and advice during the present work. In particular, I would like to thank Dr. Jakub Bujalski for his valuable suggestions during preparation of my thesis. The author wishes to thank the financial support from The Australian Research Council, The University of New South Wales, CSIRO Minerals and Abel Metal Pty Ltd. for the accomplishment of the present work. I extend the utmost appreciation to my parents and brothers for their love, support and understanding over the course of my PhD studies.

iii

ABSTRACT
A rotating disk technique was used to determine the dissolution rate and diffusivity of CaO and MgO in slags. The dissolution rate was deduced from the measured changes in concentration of oxides in slag with respect to reaction time. The experimental set- up was initially tested with dissolution of magnesia in the CaO 55 wt% Al2 O3 slag at 1430 C and a measured rate of 2.7 10 5 g/cm2 .s was obtained. The dissolution rate was increased by slag chemistry and ranged from 6.5 10 5 to 2 .1 10 4 g/cm2 .s. The dissolution rate of CaO was measured in CaO 42 wt% Al2 O3 8% SiO 2 based slag. The measured dissolution rates were found to be strongly dependent on the slag chemistry and temperature and ranged from 5.03 10 5 to 3.3 10 4 g/cm2 .s. The dissolution rates were strongly dependent on the rotation speed and results indicate mass transfer in the slag phase to be rate- limiting step. The diffusivity of MgO / CaO was c alculated from the dissolution rate and solubility data, using known mass transfer correlations. The diffusivity of MgO in the calcium aluminate slag at 1430 C was found to be about 1.1 10 5 cm2 /s. Additions of 5 and 10 wt% Fe2 O3 increased the diffusivity by a factor ~ 1.5 to 3, respectively. However, with introduction of (CaF2 5 wt% + Fe2 O3 5 wt%) and (CaF2 5 wt% + Fe2O3 10 wt%) in the slag, the diffusivity increased considerably by a factor of about 29 and 11, respectively. The diffusivity of CaO in calcium aluminosilicate was measured to be in the order of 10-6 to 10-5 over a temperature range of 1430 1600 C. CaF2 increased the diffusivity by a factor of 3 to 5 while MnO x and FeO x , ilmenite and TiO 2 increased the diffusivity substantially and SiO 2 had an opposite effect. The measured diffusivities are in accord with published data on comparable systems and are discussed with reference to Eyring theory. It was concluded that MnO x , FeOx and ilmenite in the slag increase the dissolution rate and diffusivity of lime, showing comparable results with respect to CaF2 . iv

TABLE OF CONTENTS
CERTIFICATE OF ORIGINALITY .................................................................I ACKNOWLEDGMENT ...................................................................................III ABSTRACT........................................................................................................IV TABLE OF CONTENTS ................................................................................... V LIST OF FIGURES ...........................................................................................XI LIST OF TABLES .......................................................................................XVIII CHAPTER 1. 1.1 1.2 1.3 1.4 1.4.1 1.5 1.5.1 1.5.2 1.5.3 1.5.4 1.5.5 1.5.6 1.5.6.1 1.5.6.1.1 LITERATURE REVIEW ....................................................... 1

Introduction........................................................................................ 1 Secondary steelmaking ...................................................................... 1 Ladle slag and dissolution of lime in the slag.................................... 4 Fluospar as flux to aid lime dissolution............................................. 5 Use of fluorspar in Australian steel industry ..................................... 7 Factors affecting dissolution.............................................................. 8 A guide for the literature review........................................................ 8 Rate of flow of the molten slag past the solid oxide .......................... 9 Solubility of solid oxide in the slag ................................................... 9 Physical properties of solid oxides .................................................. 12 Formation of a product layer at the solid oxises/slag interface ....... 14 Viscosity of slag............................................................................... 23 Effect of additives on the viscosity of slag...................................... 30 Effect of CaF2 Substitutes on the viscosity of slag...................... 31

1.6 1.6.1 1.6.1.1 1.6.1.2 1.6.1.3 1.7 1.7.1 1.8 1.8.1 1.8.2 1.8.2.1 1.8.2.2 1.8.2.3 1.8.3 1.8.4 1.8.4.1 1.8.4.2 1.8.4.3 1.8.5 1.9

Diffusivity in molten slag ................................................................ 36 Liquid state diffusion models........................................................... 75 Hydrodynamic theory ...................................................................... 76 Hole theory....................................................................................... 78 Eyring theory.................................................................................... 78 General discussion ........................................................................... 82 Questions arising from the literature on diffusivity......................... 83 Methods for measurement of diffusivity.......................................... 83 Instantaneous plane source method.................................................. 85 Capillary - reservoir method ............................................................ 86 Semi infinite capillary................................................................... 87 Finite capillary ................................................................................. 88 Diffusion couple method.................................................................. 90 Electrochemical method................................................................... 93 Controlled forced convection method.............................................. 95 Rotating disk method ....................................................................... 96 Rotating cylinder method............................................................... 101 Applicability of rotating disk/cylinder technique .......................... 105 Selection of experimental technique for the present work............. 106 Objectives of this work .................................................................. 107 EXPERIMENTAL............................................................... 109

CHAPTER 2. 2.1 2.2 2.2.1 2.2.2 2.2.3

The outline of the experimental work ............................................ 109 Material preparation....................................................................... 110 Dense CaO / MgO crucible ............................................................ 110 Chemical reagents.......................................................................... 111 Preparation of calcium aluminosilicate master slag....................... 112 vi

2.2.4 2.3 2.4 2.4.1 2.4.2 2.5

Preparation of calcium aluminate slag........................................... 114 Experimental apparatus.................................................................. 115 Experimental procedure ................................................................. 117 Rotating experiments ..................................................................... 117 Static experiments.......................................................................... 119 Analytical techniques ..................................................................... 119

2.5.1 Scanning Electron Microscopy and Energy Dispersive System analysis (SEM-EDS)...................................................................................... 120 2.5.2 Microprobe analysis ....................................................................... 122 EXPERIMENTAL RESULTS ............................................ 125

CHAPTER 3. 3.1 3.1.1 3.1.1.1 3.1.1.2 slag 3.1.1.3

Rotating experiments ..................................................................... 126 Dissolution of CaO in calcium aluminosilicate slag...................... 126 Effect of rotating speed on dissolution rate ................................... 127 Variation of CaO dissolution at various temperatures in the master 135 Effect of additives on the dissolution of CaO in slag .................... 143

3.1.1.3.1 Effect of CaF2 addition on dissolution of CaO in slag at various temperatures................................................................................................... 144 3.1.1.3.2 Effect of Fe2 O3 addition on dissolution of CaO in slag at various temperatures................................................................................................... 147 3.1.1.3.3 Effect of TiO 2 addition on dissolution of CaO in slag at various temperatures................................................................................................... 149 3.1.1.3.4 Effect of ilmenite addition on dissolution of CaO in slag at various temperatures ...................................................................................... 151 3.1.1.3.5 Effect of Mn3 O4 addition on dissolution of CaO in slag at various temperatures................................................................................................... 153 3.1.1.3.6 Effect of SiO 2 addition on dissolution of CaO in slag at various temperatures................................................................................................... 155 3.1.1.4 Effect of variables on the dissolution rate...................................... 155

vii

3.1.1.5 3.1.2 3.1.2.1 3.1.2.2 3.1.2.2.1 3.2 3.2.1 3.2.1.1 3.2.1.2 3.2.1.3 3.2.1.4 3.2.1.5 3.2.1.6 3.2.1.7 3.2.1.8 3.2.1.9

Effect of basicity on the dissolution of lime at constant temperature 157 Dissolution of MgO in calcium aluminate slag ............................. 160 Effect of rotation speed on the rate of dissolution......................... 161 Effect of Fe2 O3 addition on dissolution of MgO in slag................ 165 Effect of (Fe2 O3 + CaF2 ) addition on dissolution of MgO in slag 166 Static experiments.......................................................................... 167 CaO experiments............................................................................ 167 Solubility of lime in the master slag under various temperatures.. 168 Effect of addition of CaF2 on the solubility of lime in the slag..... 171 Effect of addition of Fe2 O3 on the solubility of lime in the slag.... 172 Effect of addition of TiO 2 on the Solubility of lime in the slag..... 174 Effect of addition of ilmenite on the solubility of lime in the slag 175 Effect of addition of Mn3 O4 on the solubility of lime in the slag.. 176 Effect of addition of SiO 2 on the solubility of lime in the slag...... 177 FactSage thermodynamic modelling.............................................. 178 Formation of a reaction layer on the lime/base slag interface ....... 179

3.2.1.9.1 Effect of basicity on the formation of reaction layer on the lime/slag interface .......................................................................................... 186 3.2.2 MgO experiments........................................................................... 189 DISCUSSION ....................................................................... 192

CHAPTER 4.

4.1 Diffusivity of CaO / MgO in slag and effect of additives on the diffusivity....................................................................................................... 193 4.1.1 4.1.2 4.1.3 Mass transfer from the rotating disk .............................................. 193 Mass transfer from the rotating cylinder........................................ 196 Total mass transfer from the solid oxide specimen ....................... 198

viii

4.2 4.3 4.3.1 4.4 4.5 4.6 4.7 4.8

Diffusivity of MgO in calcium aluminate slags............................. 200 Diffusivity of CaO in calcium aluminosilicate slags ..................... 203 Comparison of CaO diffusivity with literature data ...................... 208 Diffusion in a Mix controlled regime ......................................... 216 Activation energy........................................................................... 230 Relationship of diffusivity with viscosity...................................... 237 Ionic conductivity .......................................................................... 239 Summary of key findings ............................................................... 245 CONCLUSION .................................................................... 247

CHAPTER 5.

5.1 Dissolution rate of MgO in calcium aluminate slag and lime in the calcium aluminosilicate slags ......................................................................... 247 5.2 Solubilities of MgO in calcium aluminate slag and CaO in calcium aluminosilicate slags ...................................................................................... 249 5.3 5.4 Diffusivity of MgO / CaO in slags................................................. 250 Recommendations for future work ................................................ 252

REFERENCES ................................................................................................. 254 APPENDIX A. SOLID OXIDES DISSOLUTION DATA...................... 263

A.1 The effect of rotation speed on dissolution of CaO in calcium aluminosilicate slag at 1430 C ...................................................................... 263 A.2 Effect of CaF2 addition on dissolution of CaO in calcium aluminosilicate slag at various temperatures ................................................. 269 A.3 Effect of Fe2 O3 additio n on dissolution of CaO in calcium aluminosilicate slag at various temperatures ................................................. 274 A.4 Effect of TiO 2 addition on dissolution of CaO in calcium aluminosilicate slag at various temperatures ................................................. 279 A.5 Effect of ilmenite addition on dissolution of CaO in calcium auminosilicate slag at various temperatures................................................... 284

ix

A.6 Effect of Mn3 O4 addition on dissolution of CaO in calcium aluminosilicate slag at various temperatures ................................................. 289 A.7 Effect of SiO 2 addition on dissolution of CaO in calcium aluminosilicate slag at various temperatures ................................................. 294 A.8 slag Effect of Fe2 O3 addition on dissolution of MgO in calcium aluminate 298

A.9 Effect of (CaF2 + Fe2 O3 ) addition on dissolution of MgO in calcium aluminate slag ................................................................................................ 301 APPENDIX B. MODEL FOR ESTIMATING THE SLAG VISCOSITY 304 MODEL FOR ESTIMATING THE SLAG DENSITY 307 ERROR ANALYSIS ........................................................ 310 PRELIMINARY STUDY OF LIME DISSOLUTION IN 317

APPENDIX C. APPENDIX D. APPENDIX E. STATIC SLAG E.2 E.1.1 E.1.2 E.3 E.4

Experimental.................................................................................. 317 Materials ......................................................................................... 317 Experimental Procedure ................................................................. 318 Experimental Results and Discussion............................................ 321 Key findings ................................................................................... 323

LIST OF FIGURES
Figure 1.1: CaO-Al2 O3 -SiO2 phase diagram in Slag Atlas (Eisenhuttenleute (1995)) ..........10 Figure 1.2: CaO Al2 O3 MgO phase diagram in Slag Atlas (Eisenhuttenleute (1995)).....11 Figure 1.3: Schematic diagram of distribution of slag components near the interface according to Matsushima et al. (1977) ..........................................................15 Figure 1.4: Effect of concentration of various fluxes on lowering of melting point of dicalcium silicate Singh et al. (1977) ............................................................35 Figure 1.5: Effect of temperature and slag composition on the chemical diffusivity of Ca2+ introduced as Ca45O into slags of A1, A2 and A3 after Johnston et al. (1974) ........................................................................................................39 Figure 1.6: Diffusion coefficient of calcium (upper line) and silicon (lower line) after Towers et al. (1957). ...................................................................................43 Figure 1.7: Tracer diffusivity of Ca45 in CaO SiO2 melts a function of mole fraction of silica and temperature after Keller et al. (1979b) ...........................................44 Figure 1.8: Electrical conductivity ( 1cm 1 ) of CaO-SiO2 melts as a function of mole fraction of silica and temperature after Keller et al. (1979b) ...........................45 Figure 1.9: Tracer conductivity and computed conductivity of Ca45 in the CaO SiO2 melt as function of SiO 2 at 1600 C after Keller et al. (1979b) ........................47 Figure 1.10: Diffusivities of iron and calcium in silica saturated CaO FeO SiO2 melts at 1600 C after Keller et al. (1986) ..............................................................48 Figure 1.11: Diffusivity of Ca45 in melts as a function of temperature after Hara et al. (1989) ........................................................................................................49 Figure 1.12: Relationship between logarithm of tracer diffusivities of calcium and iron and reciprocal temperature after Goto et al. (1977)........................................50 Figure 1.13: Diffusion coefficients of Ca45 and Si31 as a function of melt composition at 1600 C after Keller et al. (1979a)................................................................51 Figure 1.14: Diffusivities of oxides in CaO-40 wt% SiO 2 -20 % Al2 O3 slag after Ukyo et al. (1982) ...................................................................................................52 Figure 1.15: Diffusivities of oxides in FeOx - 30 wt% CaO - 45 % SiO2 slag after Ukyo et al. (1982) ................................................................................................53 Figure 1.16: The variation of DFe with T in Fe 2 SiO4 after Agarwal et al. (1975) .................54 Figure 1.17: The variation of DFe with 1/T in CaFeSiO 4 after Agarwal et al. (1975)............54 Figure 1.18: Diffusivities of Fe 2+, Ni2+ , Co2+ and Ca2+ in silica saturated MeO CaO SiO2 melts at 1600 C after Nowak et al. (1975) ............................................57 Figure 1.19: Self diffusivities of elements in molten slag for blast furnace CaO-40 wt% SiO2 -20 % Al2 O after Nagata et al. (1982) ....................................................58 Figure 1.20: Self diffusivities of elements in molten slag for steelmaking (25-40) wt% Fe 2 O3 -(30-40)% CaO-SiO2 after Nagata et al. (1982) ....................................59 Figure 1.21: Tracer diffusivities of CaO in slags with various chemistry on the basis of previous publication (B: basicity , C: CaO, A: Al2 O3 , Fe: FeO, M: MO).........61

xi

Figure 1.22: Comparison of alumina diffusivity data according to Henderson et al. (1961) & Cooper et al. (1964) & Taira et al. (1993) & Lee et al. (2001) (B: basicity, A: Al2 O3 ) B is the basicity, A is Aluminium concentration. ...........67 Figure 1.23: Chemical diffusivity of iron oxide in CaO 38 wt% SiO 2 21 % Al2 O3 melts in comparison with the results of other studies at 1300 to 1360 C and approximately the same base melt composition as a function of the average iron concentration. ......................................................................................75 Figure 1.24: Diffusion of large molecule (B) due to the movement of small solvent molecule (A) ..............................................................................................81 Figure 1.25: Apparatus for measuring diffusivity of ele ments dissolved in molten slag by capillary reservoir technique .....................................................................87 Figure 1.26: Diffusion couple, two capillaries..................................................................91 Figure 1.27: Diffusion couple, two capillaries..................................................................92 Figure 1.28: Diffusion couple, one capillary ....................................................................93 Figure 1.29: The relationship between the mass transfer and Reynolds number according to the previous investigations..................................................................... 105 Figure 2.1: Schematic of the experimental apparatus used for the rotating cylinder tests... 116 Figure 2.2: Photo of the CaO/MgO crucible attached with Zirconia paste to the alumina rod........................................................................................................... 117 Figure 2.3: The Philips XL30 used for the SEM analysis ................................................ 121 Figure 2.4: The CAMECA SX-50 Micro Probe used for the EPMA analysis ................... 123 Figure 3.1: CaO-Al2 O3 -SiO2 system phase diagram from Slag Atlas (Eisenhuttenleute (1995))..................................................................................................... 127 Figure 3.2: The concentration of CaO (wt%) in the melt with increasing the rotation speed at 1430 C ....................................................................................... 128 Figure 3.3: Variation of the dissolution rate of CaO versus the square root of rotation speed in air at 1430 and 1600 C ................................................................ 134 Figure 3.4: Variation of the dissolution rate of CaO versus the 0.75-th power of rotation speed in air at 1430 and 1600 C ................................................................ 134 Figure 3.5: Variation of the dissolution rate of CaO versus A 0.5 + B 0.75 of rotation speed in air. A and B are defined at 1430 and 1600 C ................................. 135 Figure 3.6: Concentration of CaO dissolved in base slag at 90 rpm and in air at 1430 C .. 137 Figure 3.7: Concentration of CaO dissolved in base slag at 90 rpm and in air at 1500 C .. 140 Figure 3.8: Concentration of CaO dissolved in base slag at 90 rpm and in air at 1550 C .. 141 Figure 3.9: Concentration of CaO dissolved in base slag at 90 rpm and in air at 1600 C . 142 Figure 3.10: Comparison of CaO Concentrations dissolved in slag at 90 rpm and in air at 1430 1600 C ........................................................................................ 143 Figure 3.11: The lime specimen after dissolution in the slag with 5 wt% CaF2 at 90rpm and after reaction time of 20 minutes ......................................................... 145 Figure 3.12: CaO- Al2 O3 -CaF2 phase diagram according to Mills and Keene (1981)......... 146

xii

Figure 3.13: Comparison of total CaO content in slag with 5 wt% CaF2 at 90 rpm and in air at 1430 1600 C ................................................................................ 147 Figure 3.14: The lime specimen after dissolution in the slag with 5 wt% Fe 2 O3 at 90 rpm and after reaction time of 20 minutes ......................................................... 148 Figure 3.15: Comparison of CaO concentrations dissolved in slag with 5 wt% Fe 2O3 at 90 rpm and in air at 1430 1600 C........................................................... 149 Figure 3.16: The lime specimen after dissolution in the slag with 5 wt% TiO 2 at 90 rpm and after 60 minutes of reaction. ................................................................ 150 Figure 3.17: Comparison of concentrations of CaO dissolved in slag with 5 wt% TiO 2 at 90 rpm and in air at 1430 1570C for 1 hour ............................................ 150 Figure 3.18: The lime specimen after dissolution in the slag with 5 wt% ilmenite at 90 rpm and after reaction time of 10 minutes................................................... 152 Figure 3.19: Comparison of concentrations of CaO dissolved in slag with 5 wt% ilmenite at 90 rpm and in air at 1500 1600 C .......................................... 152 Figure 3.20: The lime specimen after dissolution in the slag with 5 wt% Mn2 O3 at 90 rpm and after reaction time of 20 minutes................................................... 154 Figure 3.21: Concentration of CaO dissolved in slag with 5 wt% Mn3 O4 at 90 rpm and in air at 1430 1600 C ................................................................................ 154 Figure 3.22: Comparison of concentrations of CaO dissolved in slag with additional 5 wt% SiO 2 at 90 rpm in air at 1500 1600 C for 1 hour .............................. 155 Figure 3.23: The rate of dissolution of lime (g/cm2 .s) in the slag at various temperatures and with additives..................................................................................... 157 Figure 3.24: The concentration of lime (wt%) in the slag (basicity of 0.9) at various rotation speed at 1500 C........................................................................... 158 Figure 3.25: The dissolution rate of CaO with speed Figure 3.26: The dissolution rate of CaO with speed
0.5

in slag with basicity of 0.9........... 159 in slag with basicity of 0.9 ......... 159

07.5

Figure 3.27: CaO-Al2 O3 -MgO system phase diagram according to Slag Atlas (Eisenhuttenleute (1995)) .......................................................................... 160 Figure 3.28: Concentration of MgO dissolved in slag at different rotation speed, in air at 1430 C for 1 hour .................................................................................... 162 Figure 3.29: Dependence of rate of dissolution of MgO with 0.5 -th power of speed........ 163 Figure 3.30: Rate of dissolution of MgO with 0.75 -th power of speed............................. 164 Figure 3.31: Variation of the dissolution rate of MgO versus A 0.5 + B 0.75 of rotation speed in air. A and B are defined at 1430 C ................................... 164 Figure 3.32: Concentration of MgO dissolved in slag with 5 and 10% Fe 2O3 at 90 rpm in air at 1430 C for 1 hour............................................................................ 165 Figure 3.33: Concentration of MgO dissolved in slag with additives at 90 rpm and in air at 1430 C for 1 hour ................................................................................ 167 Figure 3.34: Variation of bulk slag composition (wt%) measured by SEM-EDS with the reaction time at 1430 C in air.................................................................... 169 Figure 3.35: Interfacial region of CaO in contact with slag at 1430 C for the reaction time of 2 hours ......................................................................................... 170

xiii

Figure 3.36: Interfacial region of CaO in contact with slag containing 5 wt% CaF2 at 1430 C for the reaction time of 3 hours ..................................................... 172 Figure 3.37: Interfacial region of CaO in contact with slag containing 5 wt% Fe 2O3 at 1430 C for the reaction time of 3 hours ..................................................... 173 Figure 3.38: Interfacial region of CaO in contact with slag containing 5 wt% TiO 2 at 1430 C for the reaction time of 3 hours ..................................................... 175 Figure 3.39: Interfacial region of CaO in contact with slag containing 5 wt% ilmenite at 1430 C for the reaction time of 3 hours ..................................................... 176 Figure 3.40: Interfacial region of CaO in contact with slag containing 5 wt% Mn3 O4 at 1430 C for the reaction time of 3 hours ..................................................... 177 Figure 3.41: Interfacial region of CaO in contact with slag containing 5 wt% SiO 2 at 1430 C for the reaction time of 3 hours ..................................................... 178 Figure 3.42: SEM micrograph of the CaO and slag interface for lime reacting 30 minutes with slag in air at 1430 C ......................................................................... 180 Figure 3.43: SEM micrograph of the CaO and slag interface for lime reacting 1 hour with slag in air at 1430C .......................................................................... 181 Figure 3.44: SEM micrograph of the CaO and slag interface for lime reacting 2 hours with slag in air at 1430C .......................................................................... 181 Figure 3.45: SEM micrograph of the CaO and slag interface for lime reacting 4 hours with slag in air at 1430C .......................................................................... 182 Figure 3.46: SEM micrograph of the CaO and slag interface for lime reacting 6 hours with slag in air at 1430C .......................................................................... 182 Figure 3.47: SEM micrograph of the CaO and slag interface for lime reacting 12 hours with slag in air at 1430C .......................................................................... 183 Figure 3.48: SEM micrograph of the CaO and slag interface for lime reacting 24 hours with slag in air at 1430C .......................................................................... 183 Figure 3.49: Thickness of solid layer as a function of square root of time in slag in air at 1430C ..................................................................................................... 185 Figure 3.50: Interfacial region of CaO in contact with slag (basicity = 0.9) at 1500 C for the reaction time of 1 hour ......................................................................... 186 Figure 3.51: Interfacial region of CaO in contact with slag (basicity = 0.9) at 1500 C for the reaction time of 2 hours ....................................................................... 187 Figure 3.52: Interfacial region of CaO in contact with slag (basicity = 0.9) at 1500 C for the reaction time of 4 hours ....................................................................... 187 Figure 3.53: Interfacial region of CaO in contact with slag (basicity = 0.9) at 1500 C for the reaction time of 6 hours ....................................................................... 188 Figure 3.54: Interfacial region of CaO in contact with slag (basicity = 0.9) at 1500 C for the reaction time of 11 hours ..................................................................... 188 Figure 3.55: Interfacial region of CaO in contact with slag (basicity = 0.9) at 1500 C for the reaction time of 24 hours ..................................................................... 189 Figure 3.56: SEM micrograph of the magnesia / slag interface from the samples left from the rotation experiments at 90 rpm and 1430C ................................... 190

xiv

Figure 4.1: Diffusivity of CaO in CaO 42 wt% Al2 O3 8 SiO2 slag with 5 wt% addition of CaF2 , MnOx, FeOx , TiO2 , SiO 2 and ilmenite. The activation energy of diffusion calculated on the basis of the slope of these graphs are compared for the base slag (44 kcal/mol) versus the slag with addition of 5 wt% CaF2 (15 kcal/mol). ........................................................................... 204 Figure 4.2: Diffusiv ity of calcium according to the published data and the deduced diffusivity in the present work for base slag (B: basicity, C: CaO, Al: Al2 O3 , Fe: FeO, M:Mg) ............................................................................. 209 Figure 4.3: Influence of addition of FeOx and MnO x on the apparent diffusivity of alumina at 1560-1590C according to Lee et al. (2001)................................ 214 Figure 4.4: CaO-Al2 O3 -SiO2 phase diagram................................................................... 217 Figure 4.5: Estimation of CaO diffusion through a solid layer......................................... 219 Figure 4.6: variation (Thickness2 ) of magnesiowustite layer as a function of time on the basis of work done by Zhang et al. (1994) .................................................. 221 Figure 4.7: Variation of (thickness2 ) of wustite layer with time deduced from data according to Allen et al. (1995) .................................................................. 222 Figure 4.8: Variation of the (thickness2 ) of the spinel layer with time deduced from data according to Allen et al. (1995) .................................................................. 222 Figure 4.9: The CaO concentration predicted by mix-controlled model and the experimental data at30 rpm & 1430 C ....................................................... 227 Figure 4.10: The CaO concentration predicted by mix-controlled model and the experimental data at 60 rpm & 1430 C ...................................................... 228 Figure 4.11: The CaO concentration predicted by mix-controlled model and the experimental data at 90 rpm & 1430 C ...................................................... 228 Figure 4.12: The CaO concentration predicted by mix-controlled model and the experimental data at 120 rpm & 1430 C .................................................... 229 Figure 4.13: The CaO concentration predicted by mix-controlled model and the experimental data at150 rpm & 1430 C ..................................................... 229 Figure 4.14: Arrhenius plots for calculation of the activation energy for diffusion of CaO in the master slag and slags with additives.................................................. 231 Figure 4.15: Arrhenius plot for the diffusion of Ca2+ in the CaO 20 wt% Al2 O3 42% SiO2 , used in the calculation of activation energy on the basis of data from Johnston et al. (1974) ................................................................................ 233 Figure 4.16: Arrehnius plot for the diffusion of Ca2+, F-1 and Fe 2+ in the CaO 20 wt% Al2 O3 42% SiO 2 slag, used in the calculation of activation energy according to data from Johnston et al. (1974) .............................................. 234 Figure 4.17: Arrhenius plot for diffusion of Ca in the CaO SiO2 slags according to diffusivity data from Keller et al. (1979b) ................................................... 236 Figure 4.18: Investigation of applying Eyring theory in diffusion of CaO in the slag ........ 239 Figure 4.19: Electrical conductivity of CaO SiO2 slag, measured experimentally calculated as a function of mole fraction of silica at 1600 C after Keller et al. (1979b) ................................................................................................ 242 Figure A. 1: Concentration of CaO dissolved in slag at 30 rpm and at 1430C ................. 264

xv

Figure A. 2: Concentration of CaO dissolved in slag at 60 rpm and at 1430C ................. 265 Figure A. 3: Concentration of CaO dissolved in slag at 90 rpm and at 1430C ................. 266 Figure A. 4: Concentration of CaO dissolved in slag at 120 rpm and at 1430C ............... 267 Figure A. 5: Concentration of CaO dissolved in slag at 150 rpm and in air at 1430C....... 268 Figure A. 6: Concentration of CaO dissolved in slag with 5 wt% CaF2 at 90 rpm and in air at 1430C ............................................................................................ 270 Figure A. 7: Concentration of CaO dissolved in slag with 5 wt% CaF2 at 90 rpm and in air at 1500 C ........................................................................................... 271 Figure A. 8: Concentration of CaO dissolved in slag with 5 wt% CaF2 at 90 rpm and in air at 1550 C ........................................................................................... 272 Figure A. 9: Concentration of CaO dissolved in slag with 5 wt% CaF2 at 90 rpm and in air at 1600 C for 1 hour............................................................................ 273 Figure A. 10: Concentration of CaO dissolved in slag with 5 wt% Fe 2O3 90 rpm and in air at 1430C for 1 hour ............................................................................ 275 Figure A. 11: Concentration of CaO dissolved in slag with 5 wt% Fe 2O3 at 90 rpm and in air at 1500 C for 1 hour............................................................................ 276 Figure A. 12: Concentration of CaO dissolved in slag with 5 wt% Fe 2O3 at 90 rpm and in air at 1550 C for 1 hour............................................................................ 277 Figure A. 13: Concentration of CaO dissolved in slag with 5 wt% Fe 2O3 at 90 rpm and in air at 1600 C for 1 hour............................................................................ 278 Figure A. 14: Concentration of CaO dissolved in slag with 5 wt% TiO 2 at 90 rpm and in air at 1430C for 1 hour ............................................................................ 280 Figure A. 15: Concentration of CaO dissolved in slag with 5 wt% TiO 2 at 90 rpm and in air at 1500 C for 1 hour............................................................................ 281 Figure A. 16: Concentration of CaO dissolved in slag with 5 wt% TiO 2 at 90 rpm and in air at 1550 C for 1 hour............................................................................ 282 Figure A. 17: Concentration of CaO dissolved in slag with 5 wt% TiO 2 at 90 rpm and in air at 1570C for 1 hour ............................................................................ 283 Figure A. 18: Concentration of CaO dissolved in slag with 5 wt% ilmenite at 90 rpm and in air at 1500 C ....................................................................................... 285 Figure A. 19: Concentration of CaO dissolved in slag with 5 wt% ilmenite at 90 rpm and in air at 1550 C ....................................................................................... 286 Figure A. 20:Concentration of CaO dissolved in slag with 5 wt% ilmenite at 90 rpm and in air at 1570C ........................................................................................ 287 Figure A. 21: Concentration of CaO dissolved in slag with 5 wt% ilmenite at 90 rpm and in air at 1600 C ....................................................................................... 288 Figure A. 22: Concentration of CaO dissolved in slag with 5 wt% Mn3 O4 at 90 rpm and in air at 1430C ........................................................................................ 290 Figure A. 23: Concentration of CaO dissolved in slag with 5 wt% Mn3 O4 at 90 rpm and in air at 1500 C ....................................................................................... 291 Figure A. 24: Concentration of CaO dissolved in slag with 5 wt% Mn3 O4 at 90 rpm and in air at 1550 C ....................................................................................... 292

xvi

Figure A. 25: Concentration of CaO dissolved in slag with 5 wt% Mn3 O4 at 90 rpm and in air at 1600 C ....................................................................................... 293 Figure A. 26: Concentration of CaO dissolved in slag with additional 5 wt% SiO 2 at 90 rpm and in air at 1500 C for 1 hour ........................................................... 295 Figure A. 27: Concentration of CaO dissolved in slag with additional 5 wt% SiO 2 at 90 rpm and in air at 1550 C for 1 hour ........................................................... 296 Figure A. 28: Concentration of CaO dissolved in slag with additional 5 wt% SiO 2 at 90 rpm and in air at 1600 C for 1 hour ........................................................... 297 Figure A. 29: Concentration of MgO dissolved in slag with 5 wt% Fe 2O3 at 90 rpm in air at 1430C for 1 hour ................................................................................. 299 Figure A. 30: Concentration of MgO dissolved in slag with 10 wt% Fe 2 O3 at 90 rpm in air at 1430C for 1 hour ............................................................................ 300 Figure A. 31: Concentration of MgO dissolved in slag with addition of 5% CaF2 &5% Fe 2 O3 at 90 rpm in air at 1430C for 1 hour ................................................ 302 Figure A. 32: Concentration of MgO dissolved in slag with addition of 5% CaF2 &10% Fe 2 O3 at 90 rpm in air and at 1430C for 1 hour .......................................... 303 Figure E. 1: SEM of base slag at 1500 C for time=0 with 1000 magnification. ............... 325 Figure E. 2: SEM of base slag at 1600 C for time=0 with 1000 magnification. ............... 325 Figure E. 3: Identified Ca2 SiO4 phase by MATLAB program for base slag at 1500 C for time=0 in a 512 by 512 microns area and 256 by 256 image resolution. ........ 325 Figure E. 4: Identified Ca2 SiO4 phase by MATLAB program for base slag at 1600 C for time=0 in a 512 by 512 microns area and 256 by 256 image resolution. ........ 325 Figure E. 5: The composition of bulk slag at various reaction times in phase diagram for basic slag at 1500 C................................................................................. 326 Figure E. 6: The composition of bulk slag at various reaction times in phase diagram for basic slag at 1600 C................................................................................. 327 Figure E. 7: Formation of Ca2 SiO4 layer on reaction of lime with master sla g at 1600 C in the platinum capsule ............................................................................. 328

xvii

LIST OF TABLES
Table 1.1: Slag compositions used for the measurements of diffusivity on liquid slags by Johnston et al. (1974) ..................................................................................38 Table 1.2: Diffusion coefficient of Ca45 in slags studied by Saito et al. (1958) ....................40 Table 1.3: Activation energy of diffusion in the CaO SiO2 melt as a function of SiO2 content of slag............................................................................................45 Table 1.4: Mass transfer of alumina in the CaO-Al2 O3 -SiO2 (Al2 O3 = 10 wt%) after Taira et al. (1993) and the deduced diffusivity.......................................................65 Table 1.5: Values of mass transfer coefficient after Matsushima et al. (1977) and deduced diffusivity of lime in the slag..........................................................69 Table 1.6: Mass transfer coefficient of dolomite from the Umakoshi et al. (1984b) and deduced diffusivity data for CaO and MgO in the present work.....................72 Table 1.7: The correlations developed previously for mass transfer from rotating solute cylinder to the solvents.............................................................................. 104 Table 2.1: The source and purity of the chemical composition used in the experiment ...... 111 Table 2.2: Chemical composition of ilmenite ................................................................. 112 Table 2.3: XRF analysis of master slag, wt%................................................................. 113 Table 2.4: Chemical composition of various slags for lime dissolution study, wt%........... 114 Table 2.5: composition of slag with additives for magnesia dissolution study, wt% .......... 114 Table 2.6: Standards used in the calibration of Philips XL 30 SEM ................................. 122 Table 2.7: Standards used in the calibration of SX-50 Micro probe ................................. 124 Table 3.1: The rate of lime dissolution (gr.cm-2 .s-1 ) in the slag at 1430 C in air ............... 130 Table 3.2: The rate of dissolution of lime (g/cm2 .s) in the slag at various temperatures and with additives..................................................................................... 136 Table 3.3: The variation of CaO concentration in base slag by measurements from XRF analysis at 90 rpm and in air at 1430 C for 1 hour ...................................... 137 Table 3.4: The variation of CaO concentration in base slag by measurements from XRF analysis at 90 rpm and in air at 1500 C ..................................................... 140 Table 3.5: The variation of CaO concentration in base slag by measurements from XRF analysis at 90 rpm and in air at 1550 C ..................................................... 141 Table 3.6: The variation of CaO concentration in base slag by measurements from XRF analysis at 90 rpm and in air at 1600 C ..................................................... 142 Table 3.7: The rate of dissolution of MgO in the slag at 1430 C and with various additives .................................................................................................. 166 Table 3.8: SEM EDS analysis of the bulk slag at1430 C in air ................................... 169 Table 3.9: EPMA analysis of the bulk slag close to the lime/master slag interface in air at different temperatures............................................................................ 171 Table 3.10: EPMA analysis of the bulk slag close to the interface of lime/ slag containing 5 wt% CaF2 at various temperatures in air .................................. 171

xviii

Table 3.11: EPMA analysis of the bulk slag close to the interface of lime/ slag containing 5 wt% Fe 2O3 at different temperatures in air ............................... 173 Table 3.12: EPMA analysis of the bulk slag close to the interface of lime/ slag containing 5 wt% TiO 2 at different temperatures in air................................. 174 Table 3.13: EPMA analysis of the bulk slag close to the interface of lime/ slag containing 5 wt% ilmenite at different times in air ...................................... 175 Table 3.14: EPMA analysis of the bulk slag close to the interface of lime/ slag containing 5 wt% MnO x at various temperatures in air ................................ 176 Table 3.15: EPMA analysis of the bulk slag close to the interface of lime/ slag containing additional 5% SiO 2 at various temperatures in air ....................... 178 Table 3.16: The solubility of lime in various slags at different temperatures by FactSage (Bale et al. (2003)) modelling.................................................................... 179 Table 3.17: SEM EDS analysis of the bulk slag at1430 C in air................................... 190 Table 3.18: SEM The solubility of magnesia in various slags ....................................... 191 Table 4.1: The diffusivity of MgO in the CaO 56 wt% Al2 O3 at 1430C in air with additives (wt%)........................................................................................ 201 Table 4.2:Results for the measured diffusivity of CaO in the slag and the calculated slag viscosity at various temperatures ............................................................... 205 Table 4.3: Values for mass transfer coefficient, thickness of boundary layer and deduced effective diffusivity of lime in the slag according to Matsushima et al. (1977) ...................................................................................................... 212 Table 4.4: Activation energy for master slag and slag with additives ............................... 231 Table 4.5: The activation energy for binary and ternary slags according to Saito et al. (1958) ...................................................................................................... 235 Table 4.6: Activation energy for diffusion of various oxides in CaO- 40 wt% SiO 2 - 20 % Al2 O3 slag according to Ukyo et al. (1982) ................................................. 236 Table 4.7: Activation energy from diffusivity data of various ions in liquid CaO- 40 wt% SiO2 - 20 % Al2 O3 slag according to Nagata et al. (1982) ............................. 237 Table 4.8: Estimated Ionic conductivity ( 1cm 1 ) of CaO-Al2 O3 -SiO2 slag and slags with 5 wt% additives at various temperatures ............................................. 241 Table 4.9: Estimated activation energy of conductivity for master slag and slags with 5 wt% additives........................................................................................... 244 Table A. 1: XRF analysis of the bulk slag when lime dissolves in slag in air at 30 rpm and 1430C .............................................................................................. 264 Table A. 2: XRF analysis of the bulk slag when lime dissolves in slag in air at 60 rpm and 1430C .............................................................................................. 265 Table A. 3: XRF analysis of the bulk slag when lime dissolves in slag in air at 90 rpm and 1430C .............................................................................................. 266 Table A. 4: XRF analysis of the bulk slag when lime dissolves in slag in air at 120 rpm and 1430C .............................................................................................. 267 Table A. 5: XRF analysis of the bulk slag when lime dissolves in slag in air at 150 rpm and 1430C .............................................................................................. 268

xix

Table A. 6: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% CaF2 in air at 90 rpm and 1430C....................................................... 270 Table A. 7: XRF analysis of the bulk slag when lime dissolves in slag with addit ion of 5 wt% CaF2 in air at 90 rpm and 1500 C...................................................... 271 Table A. 8: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% CaF2 in air at 90 rpm and 1550 C...................................................... 272 Table A. 9: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% CaF2 in air at 90 rpm and 1600 C...................................................... 273 Table A. 10: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% Fe 2 O3 in air at 90 rpm and 1430C................................................... 275 Table A. 11: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% Fe 2 O3 in air at 90 rpm and 1500 C.................................................. 276 Table A. 12: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% Fe 2 O3 in air at 90 rpm and 1550 C.................................................. 277 Table A. 13: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% Fe 2 O3 in air at 90 rpm and 1600 C.................................................. 278 Table A. 14: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% TiO 2 in air at 90 rpm and 1430C .................................................... 280 Table A. 15: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% TiO 2 in air at 90 rpm and 1500 C ................................................... 281 Table A. 16: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% TiO 2 in air at 90 rpm and 1550 C ................................................... 282 Table A. 17: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% TiO 2 in air at 90 rpm and 1570C .................................................... 283 Table A. 18: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% ilmenite in air at 90 rpm and 1500 C .............................................. 285 Table A. 19: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% ilmenite in air at 90 rpm and 1550 C .............................................. 286 Table A. 20: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% ilmenite in air at 90 rpm and 1570C ............................................... 287 Table A. 21: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% ilmenite in air at 90 rpm and 1600 C .............................................. 288 Table A. 22: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% Mn3O4 in air at 90 rpm and 1430C ................................................. 290 Table A. 23: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% Mn3O4 in air at 90 rpm and 1500 C ................................................ 291 Table A. 24: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% Mn3O4 in air at 90 rpm and 1550 C ................................................ 292 Table A. 25: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% Mn3O4 in air at 90 rpm and 1600 C ................................................ 293 Table A. 26: XRF analysis of the bulk slag when lime dissolves in slag with additional 5 wt% SiO 2 in air at 90 rpm and 1500 C ...................................................... 295

xx

Table A. 27: XRF analysis of the bulk slag when lime dissolves in slag with additional 5 wt% SiO 2 in air at 90 rpm and 1550 C ...................................................... 296 Table A. 28: XRF analysis of the bulk slag when lime dissolves in slag with additional 5 wt% SiO 2 in air at 90 rpm and 1600 C ...................................................... 297 Table A. 29: XRF analysis of the bulk slag when MgO dissolves in slag with addition of 5 wt% Fe 2 O3 in air at 90 rpm and 1430C................................................... 299 Table A. 30: XRF analysis of the bulk slag when MgO dissolves in slag with addition of 10 wt% Fe2 O3 in air at 90 rpm and 1430C................................................. 300 Table A. 31: XRF analysis of the bulk slag when MgO dissolves in slag with addition of 5% CaF2 & 5% Fe 2O3 in air at 90 rpm and 1430C for 1hour ....................... 302 Table A. 32: XRF analysis of the bulk slag when MgO dissolves in slag with addition of 5% CaF2 & 10% Fe 2 O3 in air and at 90 rpm and 1430C ............................. 303 Table B. 1: Equations for B-parameters in Urbain model for viscosity............................. 305

Table C. 1: Recommended values for partial molar volume V of various slag constituents at 1500 C .............................................................................. 308 Table E. 1: Chemical compositions of Nepheline Syenite ............................................... 329 Table E. 2: Chemical compositions of ilmenite .............................................................. 329 Table E. 3: Growth of Ca2 siO4 layer at 1500 C............................................................. 330 Table E. 4: Mass (grams) of CaO dissolved in the slags (per 100 grams of slag) at 1600 C............................................................................................................ 330

xxi

CHAPTER 1.
1.1 Introduction

Literature review

In secondary steelmaking, synthetic slag is used to influence the final chemistry and residual oxide inclusions in the steel. The most important functions of the ladle slag are to: prevent direct contact between the liquid steel and oxidizing atmosphere, desulphurise the liquid steel and absorb oxide inclusions formed as the result of deoxidation reactions. Lime based slags are used for aluminum killed steels, to absorb the alumina and silicate inclusions and form a calcium alumino-silicate slag with low level of other oxides. Fluorspar (CaF2 ) is commonly used as an additive to increase the dissolution rate of calcined lime into this ladle slag as well as reducing the slag viscosity. While fluorspar is known to be an effective fluxing agent, there are some drawbacks associated with its use. These include volatilization of fluorine containing species into the atmosphere, leaching of the residual fluorine from the discard slag, higher refractory wear rate and relatively high cost of fluorspar. Therefore, an alternative flux or slag practice, which will promote rapid and complete dissolution of lime into slag without adverse technical and environmental effects, is highly desirable.

1.2

Secondary steelmaking

The purpose of secondary steelmaking (also referred to as ladle metallurgy) is to produce clean steel, which satisfies stringent requirements of surface, internal 1

and micro-cleanliness quality and of mechanical properties. Ladle metallurgy is a secondary step of the steelmaking processes often performed in a ladle after the initial refining process in a primary furnace is completed.

Increasingly industrial and economic developments in many countries have increased the demand for high quality steels, such as alloy steels and steels used in arctic line-pipe and jet aircraft parts. To fulfil their functions these steels must meet the more stringent requirements that necessitate the use of ladle metallurgy. With increasing demand for such high quality steels, ladle metallurgy has became a routine step in the production of steel in the plant.

Although satisfactory for making steels for most applications, conventional steelmaking and refining practices such as BOP, Q-BOP, open hearth and electric furnaces could not consistently achieve the high specifications the special steels had to meet. To remain competitive and maintain production, steelmakers have accepted the secondary steel refining processes as a crucial part of steelmaking where it is also the last chance the steelmaker has to improve the quality of the steel significantly before casting.

Secondary steelmaking processes are adopted primarily to achieve various objectives. These objectives include:

Control of gases: degassing (decreasing the concentration of oxygen, nitrogen and hydrogen in steel);

Low sulphur contents (normally less than 0.01 wt% and to as low as 0.002 wt%); 2

Micro-cleanliness (removal of undesired non- metallic inclusions, primarily oxides and sulphides);

Inclusion morphology (since steelmakers can not remove undesired oxides completely, this step allows steelmakers to change the composition and/or shape of the undesired matter left in the steel to make it compatible with the mechanical properties of the finished steel).

Although secondary steelmaking processes extend the refining capabilities of modern steel-producing facilities, various prerequisites must be met for effective utilization of these processes.

Temperature and chemical composition of the raw steel must meet the specification in the primary furnace and must be maintained through tap time into the ladle or secondary vessel to produce quality steel. Accurate assessment of temperature, chemical composition, and quantity of steel in the ladle are important. Precise chemical composition control is also dependant on accurate charge-control measurements and good tapping practices, as is the provisioning for further processing.

Efficient ladle desulphurization of steel and ladle refining to produce ultraclean steels are attained only when the steel is treated under a basic, nonoxidizing slag. Ladle-refining methods were also developed whereby the addition of a nonoxidizing slag to the ladle as a supplement to low-cost argon gas stirring treatments produced cleaner steel. Synthetic slags must meet the general requirements such as: low oxygen potential, low melting point, moderate fluidity

and large solubility for alumina and sulphur. The slags having these characteristics are generally found in the CaO SiO 2 Al2 O3 system. Synthetic slags of this type are added to the ladle during or after tapping to provide refining of the steel.

The chemistry of synthetic slag plays a major role in the function of ladle steelmaking and will be presented in the next section.

1.3

Ladle slag and dissolution of lime in the slag

The recovery and impurity content of liquid steel in steelmaking processes is influenced by the physico-chemical properties of the molten slag and metal as well as physical processes such as mixing. Synthetic slag is being used in secondary steelmaking to:

Remove impurities such as sulphur from the steel. Absorb non- metallic inclusions such as Al2 O3 or MnO-SiO 2 which are formed as products during of steel.

Insulate the steel from the atmosphere and reduce heat losses while suppressing oxygen pick- up from the air, etc.

Lime is one of the ingredients used in making ladle slag, especially when a calcium aluminate based slag is employed for refining the steel. Usually argon is used to stir and homogenise the bath, by injection into the steel ladle.

When the principal slag component is lime, its main purpose is to de-sulphurise the steel according to the following reaction:

(CaO) + [S ] = (CaS ) + [O]

(1.1)

Where ( ) indicates the species is in the slag phase and [ ] the metal phase.

It is well known that the chemistry of this slag and the progress of slag- metal reactions in the steelmaking process, is largely influenced by the lime content, and accordingly by the dissolution rate of solid lime into the slag. The effective dissolution of solid lime into slag plays an important role in steelmaking practices. However, lime has a very high melting point of 2570 C. As, maintaining high basicity slags for the desulphurisation and dephosphorization of steel requires close control of lime-silica ratio; therefore, silica, as a flux cannot be used to aid slag formation.

1.4

Fluospar as flux to aid lime dissolution

With reference to reaction in Equation (1.1), a decrease in slag viscosity promotes desulphurisation through an increase in reaction kinetics. However, an increased level of fluidity promotes corrosion/erosion between slag and refractory. Therefore, a balance must be obtained between a fluid slag, which promotes adequate refining, and a viscous slag, which provides adequate refractory protection. Generally, many grades of steel require lime-based slag with addition of the fluxing agent to increase the rate of lime dissolution.

Thus in order to promote favourable chemical reactions during the process while having a fluid slag, there is a need for an axillary flux. Fluxes are required in steelmaking basically:

to lower the melting point of slags so that slags of higher basicity can be used to reduce the levels of residuals in steel.

to decrease the viscosity of slags and thus speed up refining reactions.

Calcium fluoride is known to be an effective flux, which accelerates the dissolution of CaO in the slag by lowering the liquidus temperature and viscosity of slags (Tribe, Kingston, MacDonald and Caley (1994)). Fluorspar, which consists of calcium fluoride, is most widely used flux in the steelmaking industry. Presently, the replacement of fluorspar has become of interest, because of the associated environmental and economical concerns such as:

Emission of hazardous fluoride gaseous species, such as SiF 4 , HF and NaF, including the vapour species CaF2 (Turkdogan (1985)).

Leaching of fluorine from generated slags. The relatively high cost of fluorspar (Kor (1977)). Dwindling accessible supplies. In USA, the industries depend on foreign supplies of fluorspar from Mexico and other source and in Australia the supply is partly met by imports from China.

For these reasons, there has been effort in developing alternatives to partially or fully replace fluorspar.

1.4.1

Use of fluorspar in Australian steel industry

Australia has three major steel producers, Bluescope Steel, OneSteel and Smorgon. Both Bluescope and OneSteel operate integrated steel works, in which iron ore is reduced in blast furnaces to produce liquid iron, which is then converted to steel. Smorgon on the other hand operates only Electric Arc Furnaces (EAF). The process uses mainly steel scraps as feed material. OneSteel also has a mini- mill in Sydney, which has no primary iron making and operates EAF using scraps for steel making only.

Bluescope uses about 1 kg fluorspar per tonne of steel in the BOS furnace. In the ladle, they use much less, about 0.1 kg per tonne and are trying to eliminate the use of fluorspar. Apparently they do not measure the off gases from fluorspar addition; however, it is noted that there is a mass imbalance (loss) when they measure what ends up in a known mass of slag from a given amount added, indicating that the fluoride partially reports to the fume/dust. Leachate is measured at various sites surrounding the slag processing area. Further information on these data has been requested but is not available at the time of this report.

In Onesteel, fluorspar is currently used in the ladle metallurgy furnaces (LMF). They used fluorspar in the past in the BOF practice but it showed drawbacks, so the practice was stopped. (There is no detail publicised.) For the LMF typically

40 kg of spar is added in a total addition of fluxes around 2 tonne. It was reported that no environmental work has been done in measuring the off- gases or on the measurement of leachate from the slag to the surroundings.

A rough estimate indicates that the tonnage of fluorspar utilized is in the order of 5 - 10 kt/a between BlueScope and OneSteel.

1.5

Factors affecting dissolution

Generally, the rate of dissolution of solid oxides such as lime into a molten slag is affected by:

Solubility of solid oxide in the slag; Rate of flow of the molten slag past the solid oxide; Physical properties of the solid oxide; Possible formation of a solid phase at the interface between solid oxide and slag;

Viscosity of slag; Diffusion of solid oxide in the liquid slag.

1.5.1

A guide for the literature review

During the next section of this chapter, the available published data on the factors affecting the dissolution of lime/magnesia will be investigated and the tools for

prediction of these effects will be studied. As CaF2 is currently being used in ladle slag to promote the dissolution of lime, the published data on the effect of this flux and potential substitute oxides on the factors affecting the rate of lime dissolution in slag will be investigated.

1.5.2

Rate of flow of the molten slag past the solid oxide

Many of the secondary steelmaking processes make use of argon bubbling treatment to stirr the steel bath, to promote bulk movement of the liquid steel for chemical and thermal homogeneity and to promote intimate slag and metal mixing for refining operations. The injection of an inert gas into steel offers a simple and inexpensive method to decrease the thickness of the stagnant slag boundary layer by convective mixing in the bulk slag, and thus improves the rate of mass transfer.

1.5.3

Solubility of solid oxide in the slag

The dissolution of solid oxides in molten slag is directly related to the saturation level of solid oxides in the slag. The higher the solubility of solids in the slag the larger the driving force for the solid oxide to dissolve in the slag. For the purpose of the present work, the solubility of lime in the CaO SiO 2 Al2 O3 slag and magnesia in CaO Al2 O3 slag at various temperatures can be found by from ternary phase diagram according to Figure 1.1 and Figure 1.2.

The solubility of lime in the low silica ladle slag could be determined from the intersection of the line connecting the composition of bulk slag to the corner of CaO and the liquidus lines for various temperatures in the phase diagram. The 9

dashed lines show the estimated liquidus boundaries in the phase diagram. The effect of additives investigated in the present work, on the liquidus temperature might be studied in the quaternary phase diagrams of CaO SiO 2 Al2 O3 X, where X = CaF2 , FeO x , TiO 2 , ilmenite, MnO x , etc.

Figure 1.1: CaO-Al2 O3 -SiO2 phase diagram in Slag Atlas (Eisenhuttenleute (1995))

10

Figure 1.2: CaO Al2 O3 MgO phase diagram in Slag Atlas (Eisenhuttenleute (1995)) The tentative liquidus surfaces in the system Al2 O3 CaO SiO 2 - FeOx was measured by Kalmanovitch and Williamson (1984) & (1986) but they did not consider slags with low concentration of silica.

Baisanov, Takenov, Gabdullin and Buketov (1983) and Takenov (1987) investigated liquidus surface in Al2 O3 CaO MnOx SiO 2 where the focus was mainly on the coexisting phases in the melt.

11

The liquidus surface in the system of Al2 O3 CaO SiO 2 TiO 2 was measured by Pierre (1954) & (1956) but they just considered the slags with 10 and 20 wt% Al2 O3 .

Nadyrbekov, Akberdin, Kulikov and Kim (1980) studied the crystallisation temperatures in the Al2 O3 CaF2 CaO SiO 2 system with 2, 4, 6, 8, 10 mass % CaF2 and the deduced the information from viscosity versus temperature curves. But they did not consider slag with low silica content.

In the case of MgO, one can find the solubility of MgO in the established CaO Al2 O3 MgO phase diagram, but for slag with addition of FeO x and CaF2 , no data was found in the published literature.

The solubility of solid oxides in the slag can also be determined from the thermodynamic modelling, although its accuracy greatly depends on the verification of the results by the experimental data.

Therefore, there are gaps in the published data on the measured solubility of lime in ladle slag with the above mentioned additives and the solubility of magnesia in the calcium aluminate slags with the mentioned additives.

1.5.4

Physical properties of solid oxides

Natalie and Evans (1979) studied the relationship between lime properties and the rate of dissolution in molten slags. They subjected lime samples to a series of tests such as; water reactivity test, mercury penetration porosimetry and surface area measurement aimed at the characterizing their structure. They applied the 12

water ASTM test and the rotating lime in slag test to estimate the reactivity of different lime samples. In these water ASTM tests, the reactivity is assessed from the temperature rise in water. The authors found that the pre-treatment of the lime in the kiln had significant effect on the reactivity result, with hard burnt lime the least reactive and soft burnt lime the most reactive. According to their mercury penetration porosimetry results, the soft burnt lime had the highest porosity and also the soft-burnt lime had larger pores compared with other limes. The lime samples with higher reactivity had higher porosity and total pore volumes. Then the rate of dissolution of lime samples was examined by rotating lime specimen in the CaO FeO x SiO 2 slag at 1350C. The rate of dissolution was measured on the basis of the reduction in diameter of the cylindrical lime samples. It appeared that soft burnt limes dissolved more rapidly in the slag compared with other samples and the rate of dissolution was found to be directly dependent on the pore surface area. They postulated that the dissolution of lime takes place within pores by penetration of slag into the lime, this dissolution process would thus result in faster dissolution of porous, soft burnt lime. They concluded that more reactive porous limes dissolve more rapidly in the slag.

Umakoshi, Mori and Kawai (1984b) studied the dissolution of burnt dolomite with the apparent porosities of 20 to 35% into the molten FeO x CaO SiO 2 in the temperature range of 1350 to 1425C. They found that the dissolution rate was increased by about 20% when the apparent porosity was increased, although the mass flux into the slag was scarcely changed. It was postulated that the penetration of molten slag into burnt dolomite pores increased with increasing apparent porosity.

13

1.5.5

Formation of a product layer at the solid oxises/slag interface

According to phase equilibria data and observations made by researchers, (Slag Atlas (Eisenhuttenleute (1995)) formation of a solid layer of oxide e.g. 2CaO.SiO 2 takes place at solid oxide/ slag interface, which may prevent the direct contact of solute with the slag. Formation of such a layer lowers the rate of dissolution because of the lower mass transfer rate through the solid.

Natalie et al. (1979) performed experiments in which the lime samples with various porosities were dissolved in two types of slag, i.e. (FeO x 12 wt% SiO 2 10 % CaO) at 1350C and (FeO x 18 wt% SiO 2 27 % CaO 10 wt% CaF2 ) at 1400 C. They reported that a layer of dicalcium silicate was formed a few micrometers away from the lime. It is postulated that this observed layer might have been anchored to its position by chemical or electrochemical bonding phenimena. This layer was reported to be discontinuous in the presence of FeO x . The formation of this layer appeared to have a substantial effect on the concentration profile. However they did not report the effect of slag composition on the morphology of this layer.

Matsushima, Yadoomaru, Mori and Kawai (1977) studied the mechanism of dissolution of a static solid lime in a slag, by dipping a single crystal of lime into slag bath of CaO 40 wt% SiO 2 20 % Al2 O3 and CaO 40 wt% SiO 2 20 % FeO x at 1400 1500 C. They used the crystal of CaO to avoid the complexity due to the porosity of the sintered lime. It was observed that a layer of 2CaO.SiO 2 formed at places slightly apart from the interface, which retarded the dissolution. The slag near the interface was enriched in Al2 O3 or FeO as shown 14

in Figure 1.3. It was also observed that the lime dissolution rate into slags containing high FeO and low SiO 2 was greater than rates for slags with low FeO and high SiO 2 . This observation was explained on the basis that formation of 2CaO.SiO 2 and 3CaO.SiO 2 layers in slag of high FeO was discontinuous and thus had small retarding effect on the dissolution rate of lime, whereas in slags of high SiO 2 and low FeO, the layer was dense and continuous hence retarded the rate of dissolution markedly.

Figure 1.3: Schematic diagram of distribution of slag components near the interface according to Matsushima et al. (1977)

Noguchi, Ueda and Yanagase (1976) studied the rate of dissolution CaO crystals into the CaO 45 wt% SiO 2 melt at 1500 C using a hot stage microscope. They observed a layer of 2CaO.SiO 2 formed around the CaO, grew with time. They also studied the effect of addition of FeO (2 to 40 wt%) on the morphology of the reaction layer. It was shown that by increasing FeO content of slag, some cracks formed on the 2CaO.SiO 2 layer and the formation of 3CaO. SiO 2 layer and (Ca, Fe) O solution were also observed between lime and 2CaO.SiO 2 layer.

15

Williams, Sunderland and Briggs (1982) investigated the dissolution of lime and dolomite in iron-silicate melts at 1300 C and reported the existence of four distinct zones. The iron-silicate melt became separated from the lime specimen by a zone of 2CaO.SiO 2 and FeO rich region. Initially a granular form of 2CaO.SiO 2 was precipitated which gradually sintered and formed a continuous barrier around the lime sample. In case of a soft burnt lime an FeO rich, two phase region between 2CaO.SiO 2 and lime sample was formed. After 120 seconds, the 2CaO.SiO 2 layer lost its appearance and produced a solid irregular shaped layer. Soft burnt lime produced greater proportion of FeO-rich liquid between 2CaO.SiO 2 /lime interface than hard burnt lime. Addition of 5 wt % MgO to iron-silicate melt produced an extensive and dispersed zone of 2CaO.SiO 2 which was discontinuous and no longer formed a barrier to melt penetration.

Umakoshi et al. (1984b) reported that when burnt dolomite was immersed and rotated into stagnant molten slag of (15 40 wt%) CaO - (15 50 %)SiO 2 - (20 70 %) FeO x , a solid solution film of 2CaO.SiO 2 and (Ca, Fe)O was formed at a short distance from the surface of dolomite and at the same time CaO and MgO began to dissolve individually. But due to the difference in crystal structure of CaO and MgO, 2CaO.SiO 2 detached easily and dispersed into bulk slag phase under forced convection. The molten slag rich in FeO penetrated into gaps and formed dense (Fe, Mg)O film, hence dissolution of MgO proceeded through the dissolution of (Fe, Mg)O solid solution formed on the surface.

16

Satyoko and Lee (1999) studied the dissolution of doloma and dolomite at 1350C in the stagnant melt of CaO 4 wt% MgO 30 % SiO 2 30 % FeO x 6 % MnO x . Their results showed that molten slag penetrated into the doloma through pores and/or cracks, then reacted with MgO and CaO forming low melting phases of ((Fe, Mg)O) and dicalcium ferrite (2CaO.Fe2 O3 ). The formation of these phases indicated the breakdown of the 2CaO.SiO 2 layer formed at the lime/slag interface, making it easier for slag to penetrate through. In the dissolution of dolomite, the formation of CO2 claimed to act as a barrier for further dissolution. Also the formation of 2CaO.SiO 2 was shown to be accelerated by localized cooling due to CO2 gas evolution. On the basis of this work; Satyoko, Lee, Parry, Richards and Houldsworth (2003) suggested iron oxide enrichment of doloma using mill scale as a substitute material for normal doloma in a BOF flux charge.

Kor, Martonik and Miller (1986) investigated the effect of temperature and degree of calcination of lime on its dissolution rates in the slag. They found that lime calcined at 900C dissolves about twice as fast in a FeO 30 wt% SiO 2 slag as lime calcined at 1200C. However, the calcination temperature hardly affected the dissolution rate in FeO 25 wt% SiO 2 15 % CaO slags. They explained their observations by making use of the lime dissolution mechanism in FeO SiO 2 CaO slags as proposed by Oeters and Scheel (1971). This mechanism is based on the counter diffusion of CaO and SiO 2 and relates the dissolution rates of lime to the formation of a dicalcium silicate phase within the slag phase, a certain distance away from the lime-slag interface. The magnitude of this distance, according to the proposed mechanism, is determined by the relative

17

magnitude of the diffusive fluxes of CaO and SiO 2 . The lower the diffusive flux of CaO with respect to that of SiO 2 , the further away from the lime-slag interface the formation of the Ca2 SiO 4 phase takes place. For slags containing 70 wt% FeO, the thickness of the 2CaO.SiO 2 phase was small and was expected to be less coherent for lime calcined at 900 C. This was explained as small particles would detach early from the lime spheroid calcined at 900 C and this would form an incoherent 2CaO.SiO 2 phase, causing the CaO to more rapidly diffuse into the slag phase. This mechanism was claimed not to occur in lime calcined at 1200 C and explained why the dissolution of calcined at 1200 C was slower than that for lime calcined at 900 C. The authors also show that the degree of calcination of the lime (amount of CO2 remaining) has a significant effect on the lime dissolution rate. For lime containing residual quantities of CO2 (degree of calcination <100%), the faster dissolution rate observed which was claimed to be attributed to CO2 bubbles disturbed the 2CaO.SiO 2 layer and generally increased the mass transfer coefficient through agitation of the melt.

According to the analysis by Turkdogan (1983), the penetration of iron oxide rich slag through the cracks in the solid layer to the surface of lime particles provides a passage for calcium and oxygen ions to diffuse into the slag bulk, thus facilitating the rate of dissolution. With porous lime the depth of slag penetration into the reacted zone is greater, and the rate of dissolution is much faster, as would be expected from an increase in the solid/liquid contact area. Slag penetration into the pores and between the lime and dicalcium silicate also contributes to the disintegration of the particle and thus accelerates the rate of lime dissolution.

18

Finn, Cripps and McCarthy (1973) investigated the dissolution of burnt lime in the basic oxyge n steelmaking slags and proposed the concept of coating lime with iron ore during calcination. They explained that the structure of burnt lime is dependant on the raw limestone and more importantly, on the temperature reached during calcination. Hard burning, at temperatures of order of 1400C, produces a relatively coarse- grained, chemically unreactive, sintered structure with large pores and fissures. Soft burning, at about 1000C, yields a finer, more porous structure, with a larger, more reactive, surface area. However, during the steelmaking process, as the temperature rises from 1400 to 1600 C, solid lime remaining in the slag tends to become hard burnt. The authors mentioned that at the lime/slag interface, there are two competing reactions possible:

Absorption of iron oxide on solid lime to form low melting point calcium ferrite.

Reaction with silica in the slag to form solid silicates, mainly dicalcium silicate, 2CaO.SiO 2 .

The first reaction was claimed to bring the lime into solution while the second tends to form a solid layer around the lime inhibiting the solution of lime in the slag. The difference in the rate of solution of hard and soft burnt limes was related to the first reaction. They argued that the denser structure of hard burnt lime inhibits the absorption of FeO and formation of liquid ferrite; instead, it provides a continuous foundation for an adherent 2CaO.SiO 2 . Conversely, FeO is readily absorbed in the porous structure of soft burnt lime, dissolves it, dilutes 19

the CaO and inhibits the formation of a continuous layer of 2CaO.SiO 2 . This layer decreases the rate of further reaction until it is dissolved or mechanically broken or removed. The authors mentioned that the dissolution of this protective layer could be promoted by addition of fluxes to the slag, which lowers the liquidus temperature at the 2CaO.SiO 2 end of the pseudo binary mixture with that flux. High FeO slags could promote the dissolution of lime directly and also the dissolution of 2CaO.SiO 2 , but in practice, FeO concentrations are not sufficient unless the lime is burnt at low temperature. It was explained that Fluorspar was a standard flux for promotion of lime dissolution but environmental concerns associated with the use of this flux, increasing demands and decreasing availability of metallurgical grade fluorspar have led the steel companies to an intensive search for suitable alternatives. The authors proposed that an alternative approach to the problem was to prevent the formation of dicalcium silicate layer r ather than attempt to dissolve it once formed. From consideration of the mechanism of lime dissolution, it was postulated that, if each lime particle were surface impregnated with iron oxide prior to contact with the BOS slag, this layer, at temperatures, would form an intermediate liquid phase between the lime and the slag preventing the formation of an adherent layer of dicalcium silicate. It was also postulated that this phase would advance into the lime particle ahead of its external surface as solution progressed. The authors carried out laboratory tests, during which the coated and uncoated lime samples were immersed for the required time in the slag at 1400C and then rapidly removed from the crucible. The samples were mounted and sectioned for XRD and microscope examination to identify the various phases. For the coated lime, as it absorbed sufficient iron oxide ahead of its contact with silicates, the 20

particles assumed a structure of FeO in solid solution in lime (C (W)) grains in a largely continuo us matrix of low melting point dicalcium ferrite, which rendered it mechanically weak and therefore readily disintegrated in the slag. Also SiO 2 present in the solution, or as 2CaO.SiO 2 , formed 3CaO.SiO 2 , which precipitated as separate hexagonal prisms and these provided less resistance to further absorption of liquid slag. However, in the uncoated hard burnt lime, a layer of C (W) formed on the surface with insufficient dicalcium ferrite being produced to weaken the particle mechanically or react the 2CaO.SiO 2 layer with lime. Also the FeO in the solid solution layer was a firm basis on which 2CaO.SiO 2 formed a continuous layer, which prevented further reaction. Their results confirmed the postulated mechanisms and indicated that a preliminary coating with iron oxide can provide sufficient calcium ferrite to alter the mechanism radically and thereby obtain rapid solution of the lime into the slag. Finally, the authors indicated the feasibility of manufacturing coated lime with iron oxide in a rotary kiln providing proportion and size and kiln temperature are controlled.

Zhang and Seetharaman (1994) studied the dissolution of MgO in CaO FeO CaF2 SiO 2 slags in the temperature range 1573-1673 K under static conditions. They identified formation of a magnesiowustite solid solution layer at the MgO/slag interface. They measured the thickness of the product layer, change in the dimension of MgO sample and the concentration profile of Mg in the product layer and in the melt. It was shown that the thickness of the product layer w as changing linearly with time. The effect of CaF2 on the dissolution of MgO in the slag was also studied. The dissolution of MgO increased with CaF2 content in the slag. The thickness of the solid-solution layer also increased with the

21

concentration of CaF2 below 15 wt% and decreased substantially at higher levels of CaF2 . Based on the experimental observation, the authors proposed that the dissolution of the magnesiowustite layer in the slag was faster than its growth rate in the early stages of the dissolution. In later stages, it was shown that the dissolution of solid solution in the slag was slower than its growth. It was concluded that the mechanism of MgO dissolution is consist of two parallel steps:(a) the formation of solid solution Mg1-x Fex O and (b) the dissolution of solid solution in the slag and proposed that the rate of growth of the magnesiowustite layer could be the rate-controlling step in the dissolution of MgO in slags.

Bygden, DebRoy and Seetharaman (1994) studied the dissolution of dense MgO in (10 25 wt%) CaO (45 60 %) FeO (30 45 %) SiO 2 slag in the temperature range of 1473-1673 K. They found that a layer of Wustite, (Fe, Mg)O, and a silicate layer formed surrounding the MgO. They also suggested that slag attack of MgO was in three steps; first the formation of the magnesiowustite solid solution, followed by diffusion of the core MgO into the (Fe.Mg)O layer and then chemical dissolution of this layer that increased linearly with squa re root of time , suggesting that it was solid diffusion controlled. The dissolution of MgO increased with the increase of FeO, and decreased with the increase of CaO content in slag.

Sandhage and Yurek (1988) & (1990) investigated the dissolution of sapphire in slag of CaO 20 wt% Al2 O3 40 % SiO 2 5 % MgO with variable amount of MgO. It was found that at lowest amount of MgO (5 wt%) and at 1450 C and

22

1550 C, sapphire dissolved directly and without formation of any layer in the slag, as it was observed that only isolated particles of spinel was formed in the sapphire/melt interface. It was found by the authors that the dissolution was slowed down by formation of a layer of spinel at 1450 C when the magnesia content of the slag was 10 15 wt% and also at 1550 C when the magnesia content was more that 5 wt%, so the dissolution was continued in an indirect way. They argued that the process of steady-state, indirect sapphire dissolution into the melt consisted of two components, which operated in parallel: the rate of formation of the spinel reaction product, and the rate of dissolution of the spinel reaction product. It was postulated that the solid-state diffusion of a reactant or product species was the slowest step in formation of the spinel and that the liquid-phase diffusion of a reactant or product species was the slowest step in the dissolution of the spinel.

It can be concluded that solid oxide dissolution often involves the formation of an intermediate solid oxide reaction product which itself dissolves into the melt. Formation and stability of these solid phases are in accord with established phase diagrams for slag systems, thus in general the formation of any product layer at solid oxide/slag interface for a given slag composition and temperature can be predicted by studying the phase diagram and application of the thermodynamic models.

1.5.6

Viscosity of slag

Viscosity is one of the key properties, which can influence the dissolution of lime in slag. Viscosity affects the mass transfer of ions through the liquid slag, to and 23

from the solid oxides and slag interface. Another aspect of furnace performance, which is influenced by slag viscosity, is refractory life, since the rate of attack by a potentially corrosive slag is reduced if that slag is very viscous. Therefore requirements dictate that slags should achieve a rheological balance between being adequately fluid, to have the rate of reaction high, and not being too fluid to cause excessive corrosion/erosion of the refractory.

The viscosity of slag depends on composition and temperature, and since the viscosities of metallurgical slags have an important influence on furnace operation, then accurate experimental data and predictive models of the viscosity of slags have always been desirable. There are many published data on the viscosity of CaO Al2 O3 SiO 2 which has been measured experimentally. Methods used to determine viscosity of slags include: Capillary method, Falling body method, Rotating cylinder method, Oscillating method. It was shown by various experimental data that the prime source of experimental uncertainties were:

Changes in the composition of the melt due to the reaction between the melt and graphite, where graphite crucibles are used.

To lesser extend, errors in temperatures of the melt.

According to the published data, most viscosity measurements were subject to experimental uncertainties of 25%, where in some cases experimental uncertainties could be > 50% , although experimental uncertainties of 10% could be achieved by careful calibration of viscosimeters with high and low

24

temperature reference materials. Viscosity of some slags are available in the literature, such as in Slag Atlas (Eisenhuttenleute (1995)).

It is often the case, however, that experimental information on the slag viscosities is not available for the particular composition and conditions of interest to a particular practice. In this event, mathematical models can be used to predict the trends in viscosity as a function of the key variables, and so assist in the selection of process conditions and the optimization of the performance of the system of interest. The viscosity models normally utilize the temperature and composition dependency of the viscosity in various forms.

A number of viscosity models make use of Arrhenius Equation (1.2) to describe the temperature dependence of silicate viscosity;

E = AA exp A RT

(1.2)

Where is the viscosity, E is the activation energy, R is the gas constant, and T is the temperature in K. Viscosity-temperature data are, on the basis of the above equations, usually presented in the form of ln as a function of reciprocal temperature (T-1 ). It is, however important to point out that the viscositytemperature plots for silicates usually show a slight curvature.

Many viscosity models are based on the Weymann Frenkel (WF) kinetic theory of liquids. In many cases, these models have been found to give better agreement with experimental data than can be achieved using the Arrhenius expression: 25

 E = Aw T exp RT

(1.3)

The modified WF equation for slag viscosity developed by Urbain (1987) & Urbain and Boiret (1990) is given by the following expression:

10 3 B = AT exp T

(1.4)

Where A and B are compositionally dependant parameters. These classes of oxides are introduced in the Urbain model as: glass formers, modifiers, and amphoterics. Silicon dioxide (SiO 2 ) is an example of a glass forming component, which is characterized by a net structure and very high viscosity. Modifier oxides containing e.g., Na+, K+, Ca2+, Fe2+ and Mg2+ ions modify or break the net structure and lower the viscosity. Oxides such as Al2 O3 or Cr2 O3 can behave either as modifier or glass former, depending on the concentrations of other constituents of slag. From the analysis of experimental data, Urbain postulated that the parameter B increases proportionally to the third power of XG; (X G, XM and XA are the corresponding molar fractions of glass forming, modifier, and amphoteric components). At a given XG, the parameter B has a maximum value at the intermediate ratio of modifier to amphoteric fraction, XM/( XM + XA ), this latter compositional dependence can be described by the second power parabola. The parameter A is linked to B by the compensation law:

ln A = mB + n , where m and n are constants.

26

Riboud, Roux, Lucas and Gaye (1981) using Urbain formalism which corrected compositional dependence of A and B, successfully applied the model to describe the viscosities of some industrial mold fluxes. It was reported that between these two models, the model of Urbain gives a slightly better fit than the Riboud model according to Slag Atlas (Eisenhuttenleute (1995)).

Hu and Reddy (1988) used a Weyman type Equation (1.2) to describe the viscosities of slags (Reddy model). The compositional dependence was estimated by hole theory and atomic pair model of the slag structure. The Reddy model has been applied to some binary systems and some ternary borosilicates.

Seetharaman and Sichen (1994) & Sichen, Bygden and Seetharaman (1994) developed a KTH model, for estimation of viscosities of multi-component slags. This model adopts the Arrhenius Equation (1.2) and Eyring equation for the description of viscosities by estimation of the Gibbs energies of activation for viscosities of pure components and in the case of multi-component solutions, the non- linear variation of the activation Gibbs energies are included. The activation energy for viscous flow was modelled in analogy with the modelling of Gibbs energy.

The quasi structural models have been developed to take into account the complex internal structures of molten slags. Zhang and Jahanshahi (1998b) & (1998a) developed a structurally related viscosity model, which, was originally applied to three binary and two ternary melts. The temperature dependence of viscosities was described by the Weyman Equation (1.3) but linked the parameters to the concentrations of bridging, non bridging, and free oxygens, 27

which were calculated using a thermodynamic cell model of slag structure. The composition effect in the slag was also modelled by expressing the activation energy of the slag as a function of structural parameters, namely, the fraction of bridging and free oxygen ions, which relate directly to the bonding environment in the silicate melts. The results from this model showed that it can be used to calculate the viscosity of homogeneous multi-component silicate melts over the broad temperature and composition ranges according to Sun, Zhang and Jahanshahi (2003).

Mills and Sridhar (1999) developed the NPL model, which uses the Arrhenius equation for temperature dependence of slag viscosity, but links the compositio nal dependence to the corrected optical basicity of the slag, which in turn can be obtained from experimental data or estimated. This model was successfully applied to a range of metallurgical slag compositions as well as mold fluxes, although it can be used only in limited compositional ranges.

Iida, Sakai, Kita and Shigeno (2000) & Iida and Kita (2002) used a quasistructural approach towards the modelling of slag viscosities as functions of composition. Their model was based on an Arrhenius-type equation, wherein a modified basicity index is used to link so-called structurally related network parameters to the viscosity. This model was used to describe experimental viscosity measurements on mold fluxes.

Kondratiev and Jak (2001b) & (2001a) have revised and expanded the Urbain formalisation so that separate model parameters can be included for the various chemical components. The major advantages of this modified Urbain model is 28

that it enables the difference in chemistry of individual components to be taken into account while retaining the strength of the Urbain assumptions, i.e. the silicate slag viscosity increases with third power of the glass former concentration and exhibits parabolic behaviour with varying proportions of amphoterics and network modifiers

In the recent work by Seetharaman, Mukai and Sichen (2004) a set of compositions for binary, ternary, quaternary, and multicomponent slag systems was distributed to the authors of various models. The authors obtained the corresponding experimental results by the rotating cylinder method. The experimental results were compared with the results of the viscosity data, which was carried out by the modellers themselves. It was shown that the various models were able to approximately predict the order of magnitude of viscosities for the various systems. In most cases, Idias model was able to predict the viscosities close to the experimentation while the KTH model also seemed to predict the viscosities of the binary, ternary and quaternary slags reasonably well. The other models did not seem to deviate drastically from the experimental values.

The discrepancies between the experimental values and the predicted values are of the order of 25-30%, which are similar to the experimental uncertainties for the viscosity measurements. Therefore the viscosity of molten slag can be predicted by using one of the established models.

29

1.5.6.1

Effect of additives on the viscosity of slag

The effect of additives like, CaF2 , FeO and TiO 2 on the viscosity of CaO Al2 O3 SiO 2 slags has been reported widely in the literature on the basis of the experimental work. Mills (1977) and Slag Atlas (Eisenhuttenleute (1995)), have reviewed and complied the published data, all the mentioned additives decreased the viscosity of slag but the effect of CaF2 is much more pronounced compared to FeO x and TiO 2 .

While there are data on effects of additives on viscosity of slags, they do not always cover the composition range of interest. Thus, one needs to use models for predicting the values in composition range that data is lacking.

The viscosity of slags with addition of various additives can be predicted by using the viscosity models. For example the model developed by Zhang et al. (1998b) & (1998a) predicted that addition of metal-oxide components, tends to reduce viscosity, so the developers of this model proposed that the effect of cations on viscosity may be attributed to the strength of interaction between the M2+ cation and oxygen ion. It was claimed by the authors that, the stronger interaction between M O atoms in the melt would cause the more difficult movement of atoms, which consequently increases the viscosity.

Calcium fluoride as the main component of fluorspar is being used currently in steelmaking to dissolve the lime in slag efficiently, through lowering the melting point of the slag and decreasing the viscosity of the slag. The addition of calcium fluoride to a silicate melt disrupts the polymeric constitution of the melt more effectively than basic oxide additions according to Baak (1958) & Turkdogan 30

and Bills (1960) & Bills (1963). This is attributed to the fluorine anion replacing the oxygen anion in the matrix owing to the F- ion having a greater affinity for the Si4+ cation according to Gammal and Stracke (1988). The liberated oxygen anion may then re-enter the structure, breaking yet another bond. With depolymerisation of the slag, melt fluidity increases. Through the breakdown of silicate networks, metal cations are produced, and their subsequent behaviour depends on the system into which they are incorporated. If the refractory material containing the slag is not compatible, these cations may react with the container lining.

1.5.6.1.1

Effect of CaF2 Substitutes on the viscosity of slag

CaF2 is proved to decrease the viscosity of slag effectively, however there are certain drawbacks associated with using CaF2 . The research in the past by Shimizu, Suzuki, Jimbo and Cramb (1996) & Shimizu and Cramb (2002) has indicated that a significant amount of fluorine may evaporate from slags in the form of toxic fluoride vapours such as SiF 4 and HF. Fluoride evaporation is not only concerned with the environment but is also associated with interfacial phenomena such as surface tension and adsorption. Also the steel industry has been facing the dwindling supplies and increase in prices of fluorspar due to use of more costly mineral processing methods, such as floatation cells.

The effect of additives on the viscosity of slag, and in particular, replacement of CaF2 with other minerals has been evaluated by a number of researchers.

Tribe et al. (1994) & Tribe, Kingston and Caley (1997) performed a rheological study of the CaO SiO 2 MgO CaF2 system to determine the influence of the 31

various mineralogical species on slag rheology and to access the possible replacement of fluorspar in this system with an alkali- aluminosilicate tailing material, Nepheline Syenite Tailing (NST) with the composition of 50 wt% SiO 2 - 20 wt% Al2 O3 - 15 wt% (Na2 O and K2 O) - 10 wt% FeO - 3 wt% CaO. They reported that basic slags with fluorspar exhibited both high fluidity levels, and well-defined solidification temperature. In contrast, the acidic slags were less fluid and, as might be expected, solidified over a wide temperature range. It was also shown that the freezing point of slags in which Nepheline Syenite Tailing replaced fluorspar was less-defined than for the basic fluorspar slags. This was attributed to the higher concentration of acidic components, such as SiO 2 , in the NST slags. With respect to the replacement of fluorspar with NST, it was shown that NST can effectively replace fluorspar, thereby sacrificing a lower overall level of slag fluidity while gaining a slag with a broader solidification temperature range. Whereas the fluidity was somewhat reduced, no problems with respect to either slag retention or reaction kinetics were evident during industrial testing.

MacLean, Kingston, MacDonald and Caley (1997) investigated the possibility of replacement of fluorspar with Nepheline Syenite in the CaO-MgO-SiO 2 system at ladle furnace practice. The Nepheline Syenite used came from a rock, which was consisted of three minerals, nepheline (NaAlSiO 4 ), microline (KAlSi3 O8 ), and albite (NaAlSi3 O8 ). In the slag rheology tests, nepheline syenite proved to be very effective. Although slags containing this mineral did not achieve the same degree of fluidity as a fluorspar containing slag, the difference was not considered to be significant at steelmaking temperature. The presence of Na2 O 32

decreased the melting point of the slag considerably, a result of formation of lower melting point minerals in the NaO-CaO-SiO 2 system. Investigation on the corrosive effect of adding the various minerals to the slag showed that fluorspar to be the most aggressive towards the refractory and Nepheline Syenite proved to be slightly less corrosive than the fluorspar. However, no experiment was conducted to investigate the volatisation of species from the slag due to the use of Nepheline Syenite.

Singh, Ravat, Chatterjee and Chakravarty (1977) conducted extensive laboratory and industrial trials and showed that ilmenite, can be successfully used instead of fluorspar as a fluxing agent in open- hearth steelmaking. As ilmenite is a compound of FeO and TiO 2 (FeTiO 3 ), so they initially showed the effect of FeO and TiO 2 on slag liquidus temperature and compared it to CaF2 . It was claimed that as the concentration of the fluxing agent increases, there is a decrease in the liquidus temperature of the slag. However, it was notable that there is no marked difference between CaF2 and TiO 2 at lower concentrations. They proposed that if ilmenite with much lower melting point than TiO 2 (1470 C compared with 1830 C), is used, it can be expected to be more effective than TiO 2 alone. They also compared effect of concentration of various fluxes on lowering melting point of dicalcium silicate as it is shown in Figure 1.4 and mentioned that the curve corresponding to ilmenite was expected to lie between the curves for TiO 2 and CaF2 . Therefore, it was suggested that there should be no appreciable difference between ilmenite and fluorspar as far as the fluidity action of the slag is concerned, if the relative quantities of these fluxes are not very large. They measured the relative fluidity of slags with ilmenite and fluorspar, and showed 33

that there was no difference between the fluidity of ilmenite and fluorspar. However, the action of ilmenite was found to be slower than that of fluorspar. This problem was overcome by adding ilmenite at the earlier stage of steelmaking than fluorspar. This sluggishness in the action of ilmenite was reported to be due to its higher meting point (1470 C), compared with that of fluorspar (1400 C). One of the advantages of using ilmenite instead of fluorspar is that, while fluorspar liberates harmful fluorine gas, ilmenite does not contribute to atmospheric pollution. As ilmenite was replaced fluorspar completely with a replacement of about 2:1, so the authors concluded that if ilmenite would be cheaper than fluorspar then ilmenite would be a more attractive flux.

Poggi and Lee (1974) measured the viscosity of ilmenite and fluorspar and found out the high grade ilmenite (with low amount of SiO 2 ) has a similar viscosity and melting point as the fluorspar. They also showed that the dissolution rate of lime in liquid ilmenite is about 2.5 times faster than in fluorspar, however it was found that to dissolve the same amount of lime, 30% more ilmenite than fluorspar must be used as a flux.

34

Figure 1.4: Effect of concentration of various fluxes on lowering of melting point of dicalcium silicate Singh et al. (1977)

Rheological investigations into the use of Nepheline Syenite and ilmenite as suitable fluxing additions for secondary steelmaking slags have shown promising characteristics when compared to fluorspar. In testing the corrosive properties of the fluxing agents, Nepheline Syenite showed less corrosive behaviour than fluorspar, and no published literature was found on the effect of ilmenite on the refractory. Although Nepheline Syenite could be a source Na emission due to the presence of Na2 O in this flux, there is no published study to consider this hazard. However, a previous investigation showed that there is no environmental concern with respect to utilizing ilmenite instead of Fluorspar in steelmaking.

35

1.6

Diffusivity in molten slag

The mass transfer of any solute in the solvent can be described by Ficks first law of diffusion as in Equation (1.5):

ji = D

dCi dx

(1.5)

The proportionality constant D is called the diffusion coefficient or diffusivity of i in the solvent. With the diffusion flux of species i, j i is in mol.m-2 .s-1 and the concentration gradient Ci in mol.m-3 , the diffusion coefficient has the unit of m2 .s-1 . Therefore the dissolution of lime on the slag is greatly dependant on the diffusion coefficient of CaO in the liquid melt.

The diffusion coefficient of a solute in a melt is a fundamental quantity required to characterize mass-transport rates. Diffusion is the movement of the components or constituent species of melt from point to point within melt.

In order to present the results in the literature and assess their practical significance, it is necessary to define various diffusivities based on the possible driving force for diffusion.

Generally, a thermodynamic gradient acts as a force for diffusion causing a net flux of the diffusing species in a direction such as to diminish that gradient. Chemical diffusion coefficients describe diffusion taking place under a gradient of chemical composition. The limiting case of chemical diffusion, where the thermodynamic gradient has been reduced to zero is self diffusion in which 36

diffusion is the continuous interchange that takes place between species of the same type in a medium of uniform composition. It is measured using stable or radioactive isotopes(tracers) so that ideally, diffusion occurs down a gradient of only isotope composition.

Based on these types of diffusion, the diffusion coefficient can be measured by various methods such as: instantaneous plane source method, capillary reservoir method, diffusion couple method and rotating disk technique (Kubicek and Peprica (1983)). These methods are briefly explained in this section with the relevant diffusivity data. The detailed description of each method and the criteria for selection of the technique used in the present study will be explained in the next section.

In the instantaneous source method, a small amount of radioactive tracer is deposited on the surface of slag capillary specimen. The diffusion coefficient is obtained by plotting the concentration of radioactive atoms with time.

In the capillary method, the diffusing species is held in a capillary, which is immersed in the melt and diffusion occurs between the contents of the capillary and liquid. The capillary is moved after a set of time and the distribution of the species determined.

The diffusion couple consisted of two melts of identical composition except that one contains the tracer element. The diffusion coefficient can be derived by a linear regression of concentration versus distance.

37

The rotating disk method will be explained after presenting the published data on the above- mentioned methods.

Johnson (1970) & Johnston, Stark and Taylor (1974) used radioactive tracers in their experiments and measured the diffusion coefficients of calcium in the slag under varying conditions of temperature and composition. They adopted two methods; instantaneous plane source method where a small amount of radioactive tracer was deposited on to surface of the slag capillary specimen (called by the authors; chemical diffusivity) and capillary reservoir technique where a small amount of labelled finely powdered slag of the same composition as the slag in capillary was placed on the capillary surface (self diffusivity). The calcium allowed diffusing under conditions of chemical diffusion, or more strictly through a thermodynamic gradient, the gradient was so small that it could in most cases probably be ignored.

They measured the chemical diffusivity of Ca45 in three slags with compositions tabulated in Table 1.1. Conditions of self-diffusion, where the composition was uniform throughout the sample, were established on slag A2.

Table 1.1: Slag compositions used for the measurements of diffusivity on liquid slags by Johnston et al. (1974)

Slag A1 A2 A3

Chemical composition (wt%) CaO 50.1 38 23.5 Al2 O3 49.9 20 14.5 SiO 2 42 62

38

Inspections of their results as shown in Figure 1.5 revealed that the diffusion coefficient of Ca45 is higher in slag A1, compared to slags with higher silica contents of A2 and A3. The values of diffusivity for slags with and without SiO 2 at 1450 C was reported to be 4.8 10 6 cm2 s-1 in slag A2 and was 1.9 10 5 cm2 .s-1 in the slag of A1.

Figure 1.5: Effect of temperature and slag composition on the chemical diffusivity of Ca2+ introduced as Ca45 O into slags of A1, A2 and A3 after Johnston et al. (1974) The authors also compared the self diffusivity and chemical diffusivity of Ca45 in slag A2 and found that the results are identical within the limits of experimental error. The fluorine ion F18 was introduced with Ca45 as Ca45 F2 18 into the slag of CaO 42 wt% SiO 2 20% Al2 O3 and the simultaneous diffusion of Ca45 and F18 was measured. It was shown that the introduction of fluorine ions into the melt had a marked effect on the diffusivity of Ca45 and increased it by a factor of two

39

or more, however the activation energy for diffusion remained constant. Johnston et al. (1974) also measured the chemical diffusion coefficient of iron as Fe59 O in slags containing 10 wt% and 1 wt% calcium fluoride. The results indicated that the presence of the high-diffusivity fluorine ion in the melt considerably increased the iron diffusivity. Their results also showed that different levels of CaF2 content in the slag did not affect the rate of diffusion iron in the slag.

Saito and Maruya (1958) measured self-diffusion of calcium in molten slags of CaO-SiO 2 , CaO-Al2 O3 , CaO-SiO 2-Al2 O3 and CaO-SiO 2-Al2 O3 -MgO systems. They performed the experiments by method of semi- infinite medium where two specimens, one with Ca45O and the other without, were brought into contact at the melting point of the slags. The results are shown in Table 1.2.

Table 1.2: Diffusion coefficient of Ca45 in slags studied by Saito et al. (1958)
Melting Diffusion coefficient *10-7(cm2/s) of Ca45 ion in molten slag at point various temperatures ( C ) (C) CaO SiO2 Al2O3 MgO CaO/SiO 2 1350 1395 1420 1440 1485 1510 1530 1540 1565 1575 Slag composition wt% 55.2 44.8 48.7 51.3 19 5.4 1.24 0.97 1.15 1.24 1.24 1.33 1.55 1.36 1460 1415 1300 1400 1390 1430 1440 1530 1480 3.9 6.9 3.3 3.8 5 10 8 11.5 7.1 7.8 7.1 8.4 8.7 19 17 8.5 8.4 9.8 11 9.9 10.3 11.7 5.5 8.1 13 -

39.8 41.2 43

37.2 20.2

48.9 39.4 11.7 45.2 36.3 18.6 45.6 34.1 20.3 48.4 31.3 20.3 43.4 31.8 19.5

The authors also compared the trend of diffusivity data with published data on the viscosity and specific electrical conductivity. As it was found that there is no large difference between the diffusivity, viscosity and electrical conductivity 40

between CaO-SiO 2 and CaO-Al2 O3 slags, the authors proposed that in the neighbourhood of the composition CaO:Al2 O3 =1:1, the greater part of Al2 O3 exists as the tetrahedra structure of aluminate ion AlO 4 5- similar to that of SiO 4 4in basic CaO-SiO 2 slags. Diffusivity, conductivity and their activation energies in aluminate melts were larger than in silicate melts. The viscosity of aluminate melts was about the same as or slightly smaller than corresponding silicate melts. Thus there was a good relationship between diffusivity and conductivity and reciprocal of viscosity. For the purpose of studying the behaviour of Al2 O3 in liquid slags, diffusivity of CaO SiO 2 slags (CaO/SiO 2 =1.24) were compared with those of CaO-SiO 2 -Al2 O3 (CaO/SiO 2 =1.33, Al2 O3 =20.3%). The authors mentioned that according to the previously published data, it had been considered that Al2 O3 had an intermediate properties between acidic and basic elements, and at low Al2 O3 concentrations in acid slags, Al2 O3 might dissociate into Al3+ and O2- ions and showed basic properties (network modifier), but in strong basic slags, Al2 O3 became aluminate ion such as AlO 3 3- or AlO 4 5-, etc. and showed a trend of acidity (network former) regardless of the concentration of Al2 O3 . As the addition of 20% Al2 O3 scarcely changed the values of diffusivity, it was postulated that the trend and the degree of formation of network structure of aluminate ions were not so strong as those of silicate ions. It was also found that addition of 5 wt% MgO, increased the diffusion coefficient of Ca2+ by about 10~15% in the slags with similar basicity, so the authors postulated that Mg2+ ion dissociated from MgO might behave as a networkmodifier, increasing the activity or mobility of Ca2+ ion under the reducing condition in their experiment. They supported their theory by looking at the published viscosity data, as viscosity decreased 15% due to addition of 5% MgO at the composition 41

CaO/SiO 2 = 1.2. Generally the magnitude of self-diffusion coefficient in liquid slags was about 10-6 ~10-7 cm2 /s and it was shown that the data had a good trend with changes of electric conductivity and the reciprocal of viscosity coefficient based on the compositions of slag.

Towers and Chipman (1953) & (1957) measured the diffusion coefficient of calcium and silicon ions in the slag consisting of CaO 40.5 wt% SiO 2 20.9% Al2 O3 by tracer technique. They used two techniques for measuring diffusivities, one contacting two slags of same composition with one of them containing the radioactive tracer and in the other technique, slag capillaries to which tracer had been added, were brought into contact with a relatively large volume of the master slag. After the diffusion, concentration of tracer along the capillary was measured, which resulted in the calculation of diffusion coefficient. Their results are shown in Figure 1.6, where diffusivity is plotted against 1/temperature. The values of silicon diffusivity were about one tenth that of calcium diffusivity. The average values of Ca45 diffusion coefficient at 1350, 1400 and 1450 C, were
3.5 10 7 , 2 .1 10 7 and 3.4 10 6 cm2 .s-1 respectively, with the activation

energy of 70 kcal/mole. The average diffusivities of Si31 ion was calculated at 1365 and 1430C to be 4.7 10 8 and 1.05 10 7 cm2 .sec-1 , respectively.

42

Figure 1.6: Diffusion coefficient of calcium (upper line) and silicon (lower line) after Towers et al. (1957).

Keller, Schwerdtfeger and Hennesen (1979b) measured the tracer diffusivity and electrical conductivity of Ca45 in the CaO-SiO 2 melts, where SiO 2 content of slag varied between 0.448 to 0.634 (mole fraction) in the temperature range of 1500 to 1700 C, using capillary technique. The average diffusivity values were in of the order of magnitude of 10-6 cm2 /s. Their results showed that the diffusivity was decreased by increasing the silica content of the slag as it is illustrated in Figure 1.7.

43

Figure 1.7: Tracer diffusivity of Ca45 in CaO SiO2 melts a function of mole fraction of silica and temperature after Keller et al. (1979b)

Their diffusivity values agreed best at the same basicity with those measured by Towers et al. (1953) & (1957) in slag with the composition of (CaO - 40 wt% SiO 2 20 % Al2 O3 ) but higher than results from Saito et al. (1958) and lower than those obtained from Johnston et al. (1974), although the diffusivity data show the order of magnitude of 10-6 . The authors measured the activation energy of diffusion and it was shown that the activation energy was increased by addition of silica content of the slag, as it is tabulated in Table 1.3. The authors also measured the electrical conductivity of Ca with AC current of variable frequency using the four-pole method. The results are shown in Figure 1.8.

44

Table 1.3: Activation energy of diffusion in the CaO SiO2 melt as a function of SiO2 content of slag Composition of melt Mole (SiO 2 ) 0.448 0.488 0.530 0.587 0.634 Activation energy (kcal g atom-1 ) 29 34 34 30 35

Figure 1.8: Electrical conductivity ( 1cm 1 ) of CaO-SiO2 melts as a function of mole fraction of silica and temperature after Keller et al. (1979b) As it can be seen, the electrical conductivity was decreased by increasing the mole fraction of SiO 2 . The activation energies for diffusion of Ca45 (average 32 kcal g-atom-1 ) and for electrical conduction (average 26 kcal g-atom-1 ) as 45

determined from the slopes of the straight lines in Figure 1.7 and Figure 1.8. This indicates the same mechanism for diffusion and conduction. The authors also computed the electrical conductivity from the Nerst Einstein Equation (1.6) on the basis of the assumption that conduction was solely due motion of Ca2+ ion and also validity of Nerst Einstein equation as:

k=

z 2 C Ca DCa F 2 RT

(1.6)

Where z is the charge of the Ca2+ ion, Cca the concentration in moles.cm-3 of the Ca2+ ions, and F the Faraday constant. A comparison of the computed values of electrical conductivity with the measured values at 1600 C as presented in Figure 1.9 shows that the calculated values of conductivity are lower than the measured data. The difference is small at low SiO 2 content, but it increases with increasing SiO 2 content. Hence, it appeared that in the CaO rich slag, most of the current was transported by Ca2+ ions, whic h is in agreement with the published transference numbers of cations by Bockris, Kitchener and Davis (1952).

46

Figure 1.9: Tracer conductivity and computed conductivity of Ca45 in the CaO SiO2 melt as function of SiO2 at 1600 C after Keller et al. (1979b)

Keller and Schwerdtfeger (1986) also measured the tracer diffusivities of Ca45 and Fe59 in silica saturated (0 15.9 wt%)FeO (33.8 23.2 wt%) CaO (66.2 60.9 wt%) SiO 2 melts at 1600 C. They deployed the porous frit technique, where the frits consisted of a packing of silica powder, which was sintered into vitreous silica tubing and the pores of the first were filled with pre- melted silica saturated slag. Diffusion into the slag was from a thin layer of the same chemical composition but doped with Fe59 or Ca45 . It was shown that the diffusion of iron was much faster than that of calcium as the average value of diffusivity obtained were about 2 10 6 cm2 .s-1 for Ca45 and 1 10 5 cm2 .s-1 for Fe59 as it is also illustrated in Figure 1.10.

47

Figure 1.10: Diffusivities of iron and calcium in silica saturated CaO FeO SiO2 melts at 1600 C after Keller et al. (1986)

The authors compared their results with electrochemically determined Fe diffusivity data by Nowak and Schwerdtfeger (1975) at most compositions, the data (Keeler) were smaller about a factor of two. It was motioned that the electrochemical method yields self diffusivities which were related directly to the mobility of the iron and the tracer diffusivity must be smaller than the selfdiffusivity due to the correlation factor, which in silicate melts is about 0.5. The authors also claimed that the difference between data could be due to difficulty in the diffusivity measurement in liquid slags and deviation by a factor of two between data obtained with different techniques is about normal. Their investigation showed that the frit technique could be used to produce diffusivity data at high temperatures, however it was mentioned that its application was

48

limited to the slag compositions, which are in thermodynamic equilibrium with the frit material.

Hara, Akao and Ogino (1989) measured the self diffusivity of Ca45 in FeO- 33 wt% SiO 2 - 7% CaO slag and mutual diffusivity of Ca45 in FeO- 32 wt% SiO 2 melt equilibrated with solid iron in the temperature range 1270 to 1450C. They used the capillary reservoir me thod with Ca45 as the radioactive tracer. Their results are shown in Figure 1.11.

Figure 1.11: Diffusivity of Ca45 in melts as a function of temperature after Hara et al. (1989)

Goto, Kurahashi and Sasabe (1977) measured the tracer diffusivities of calcium and iron in a steelmaking slag of CaO 27 wt% SiO 2 40% Fe2 O3 under varying oxygen pressure in the gas phase, by the instantaneous plane source method. The 49

tracer diffusivities were measured at 1360, 1410, and 1460C with an oxygen activity of 10-1 to 10-8 atm. The results are shown in Figure 1.12. Their results indicated that the diffusivity of iron was about 2 to 3 times larger than that of calcium. The ratio of Fe+3 /(Fe+3 +Fe+2 ) in their slag was measured to be in the range of 0.1 to 0.6. It was shown than the diffusivity had a direct relationship with this ratio.

Figure 1.12: Relationship between logarithm of tracer diffusivities of calcium and iron and reciprocal temperature after Goto et al. (1977)

Keller and Schwerdtfeger (1979a) measured the tracer diffusivity of Si31 ion in CaO-SiO 2 melts with capillary technique, where the silica content of slag varied (mole fraction of silica NSiO2 = 0.484-0.634) at 1600 C. They reported that diffusivity of Si31 decreases with increasing silica content of the slag, as it is

50

shown in Figure 1.13. The authors also compared the silica diffusivity with their previous publication on Ca45 diffusivity Kelle r et al. (1979b) and showed that the values of Si31 diffusivity was lower than that of Ca45 by approximately one order of magnitude in the silica rich melt. The difference decreased with decreasing silica content. The results measured by these authors agreed with the results published by Towers et al. (1957) at lower temperatures in the slag with composition of CaO 40 wt% SiO 21 % Al2 O3 slag.

Figure 1.13: Diffusion coefficients of Ca45 and Si31 as a function of melt composition at 1600 C after Keller et al. (1979a)

Ukyo and Goto (1982) measured the Quasi- inter-diffusivities of several solute oxides in liquid CaO-SiO 2 -Al2 O3 slag of 4:4:2 of charge weight ratio and liquid FeO x-CaO-SiO 2 slag of 2.5:3:4.5 of charge weight ratio. The slags were equilibrated with air and the measurements were done using the diffusion couple

51

method at 1350 to 1450C. After diffusion runs of 20 or 40 min, the sample was quenched to glassy state and analysed by a X ray micro-analyser. According to Figure 1.14, their results revealed that, MgO in the slag of CaO SiO 2 Al2 O3 had the highest diffusivity with an order of magnitude of 10-5 and P2 O5 and TiO 2 had the lowest diffusivity with an order of magnitude of 10-7 . Also the diffusivity data for MnO showed an average of 10-6 while diffusivity of FeO x was proved to be about 10-5 . For the slag containing FeO x their result is illustrated in Figure 1.15, where MgO still was the fastest oxide and the other oxides had a similar order of magnitude to slags without FeO x .

Figure 1.14: Diffusivities of oxides in CaO-40 wt% SiO2 -20 % Al2 O3 slag after Ukyo et al. (1982)

52

Figure 1.15: Diffusivities of oxides in FeOx - 30 wt% CaO - 45 % SiO2 slag after Ukyo et al. (1982)

Agarwal and Gaskell (1975) measured the self diffusion of iron in Fe2 SiO 4 and CaFeSiO 4 melts in the temperature range of 1250 to 1540C using Fe59 as the radio tracer and the capillary liquid reservoir method of diffusion measurement. Their results are plotted as diffusivity of Fe versus 1/temperature for the two compositions and are shown in Figure 1.16 and Figure 1.17. Based on their data, the activation energy for diffusion was 17.4 2.3 kcal/mole for Fe2 SiO 4 and was

24.9 2.8 kcal/mole for CaFeSiO 4 slag.

53

Figure 1.16: The variation of DFe with T in Fe2 SiO4 after Agarwal et al. (1975)

Figure 1.17: The variation of DFe with 1/T in CaFeSiO4 after Agarwal et al. (1975)

54

The self diffusivity of Fe in 61 wt% FeO SiO 2 melt was measured by Yang, Chien and Derge (1959) at12 50, 1270 and 1304 C using the capillary reservoir method with Fe55,59 as the radioactive tracer. The iron diffusivity values obtained were ( 7.9 0.3) 10 5 cm2 /s at 1250 C, (9.6 0.2) 10 5 cm2 /s at 1275 C and (1.2 0.1) 10 4 cm2 /s at 1304 C. Their results are presented in Figure 1.16, which are more than an order of magnitude higher than the results by Agarwal et al. (1975).The activation energy for diffusion, evaluated from the iron diffusivity was about 40 kcal/mole, while the activation energy for electric conduction, calculated from Wejnarth (1934) data for melt of similar composition was about 16 kcal/mole. The authors concluded that the diffusion and electric conduction were operating on different mechanism in that melt.

Simnad, Yang and Derge (1956) developed an electrochemical radioactive tracer method for using for the direct determination of ionic mobility in a molten electrolyte. They immersed a radioactive iron electrode and supplied a current. The distribution of the radioactivity in the quenched slag was measured in the vicinity of the electrode. The ionic mobility of Fe2+ in silica saturated iron silicate is found to be 09 10 4 cm2 /volts.s, which corresponds to a diffusion coefficient of 5.8 10 5 cm2 /s at 1250 C for the melt of Fex O 34 wt% SiO 2 in equilibrium with solid iron and silica. Their results agreed well with the data from Agarwal et al. (1975) in which the diffusion coefficient was 7.9 10 5 cm2 /s at the same temperature.

Mori and Suzuki (1969) measured the inter-diffusivities of iron in iron oxide melts by the capillary method over the temperature from 1430 to 1550 C and the 55

composition range from r (= Fe 3+ / Fe) = 0.12 to 0.42. Agarwal et al. (1975) applied Darkens diffusion equation to the their inter-diffusivity results using the thermodynamic data of Darken and Gurry (1946) for iron oxide melts, and the deduced diffusivity data is presented in Figure 1.16. Their values were in excellent agreement with the work done by Agarwal et al. (1975), which is in shown in Figure 1.16.

Nowak et al. (1975) used a galvanostatic technique to determine the mobility of Fe2+, Co2+, Ni2+, and Ca2+ ions in silicate melts at 1600 C. The cell used was made of quartz glass. Hence, the liquid silicates obtained the composition of the silica saturation isotherm in the systems MeO CaO SiO 2 (Me = Fe, Co, Ni). Three electrodes (anode, cathode, reference electrode) were located at the bottom of the cell. They consisted of the liquid metal Me contained in a capillary. When current was passed through the cell, the concentration gradient caused an increase of the voltage measured between anode (or cathode) and reference electrode. The mobility of the Me2+ and Ca2+ ions were calculated from an analysis of the voltage current relationship. The results are shown in Figure 1.18, where the obtained diffusion coefficients are plotted versus the MeO content of the silicate melt. It can be seen that the diffusivity for the Ca2+ - ion is about one order of magnitude smaller than those of the Me2+ ions.

56

Figure 1.18: Diffusivities of Fe2+, Ni2+, Co2+ and Ca2+ in silica saturated MeO CaO SiO2 melts at 1600 C after Nowak et al. (1975)

Nagata, Sata and Goto (1982) reviewed and compared the diffusivities of various elements in molten slag for blast furnace and molten slag for steelmaking. The self diffusivities of elements in molten slag of CaO 40 wt% SiO 2 2 % Al2 O3 are summarized in Figure 1.19.

The Mg, Fe and Mn were shown to the fastest elements in the melt compare to the rest of elements. It is evident from Figure 1.19 that of the elements studied, Mg, Fe and Mn have the highest self-diffusivities. This could result from their lower charge and hence weaker interaction with the oxygen anions and also the size of ions.

57

Figure 1.19: Self diffusivities of elements in molten slag for blast furnace CaO-40 wt% SiO2 -20 % Al2 O after Nagata et al. (1982)

The diffusivity of Ca in the temperature range of 1388-1533C was in an order of magnitude of 10-7 to 10-6 cm2 /s. Also the self-diffusivity of elements in steelmaking slag with composition of (25-40 wt%) Fe2 O3 - (30-40 wt%) CaOSiO 2 was reviewed by the same authors according to Figure 1.20. Again Fe was shown to have the highest diffusivity, while Ti was the slowest diffusing species. The calcium diffusivity in the temperature range of 1300-1550 C was in an order of magnitude of 10-5 .

58

Figure 1.20: Self diffusivities of elements in molten slag for steelmaking (25-40) wt% Fe2 O3 -(30-40)% CaO-SiO2 after Nagata et al. (1982)

The diffusivity of lime in slags with various chemistries and at different temperatures is summarized in Figure 1.21. These results are from published data by Towers et al. (1957) & Saito et al. (1958) & Johnston et al. (1974) & Goto et al. (1977) & Keller et al. (1979a) & Keller et al. (1979b) & Keller et al. (1986). All results in this figure are on the basis of diffusivity measurements with different techniques such as diffusion couple, capillary and instantaneous plane technique. It can be seen that by increasing the temperature, the diffusivity increases. Apart from the results from Saito et al. (1958), which are lower than the rest of published data, the basicity of slag affects the diffusivity of lime. By increasing the basicity, the diffusivity of lime in the slag increases. The effect of additives such as FeOx and CaF2 shows an increase in the diffusivity of lime in 59

the slag. It can be seen that the there is no published data on the diffusivity of lime in the ladle type slag where the basicity is around 5~6, furthermore, the effect of other oxides on the diffusivity of lime in the slag has not been quantified in the past. The poor reproducibility of diffusion results with the methods mentioned above at high temperatures required that a large number of experimental measurements had to be carried out in order to obtain statistically significant data. Also, the experimental difficulties and hence the inaccuracies recorded in results increased with increasing temperature. Moreover, the diffusivity measurements in liquids using the above-mentioned method could easily be in error because of convection in the melt.

In general, the data are incomplete even for the most important systems because only selected compositions and only some of the transport properties have been investigated. Further, the results obtained by different authors often disagree considerably. Thus, it is desirable to extend some of the experimental studies to cover broader compsotion range while determining effects of chemistry on diffusivity and its activation energy.

60

61

Figure 1.21: Tracer diffusivities of CaO in slags with various chemistry on the basis of previous publication (B: basicity , C: CaO, A: Al2 O3 , Fe: FeO, M: MO)

Some of the researchers measured the dissolution rate and mass transfer of solid oxides in the slag under forced convection using a rotating disc/cylinder technique, where a solid sample is rotated in a liquid slag capable of dissolving the solid under reproducible conditions of fluid mechanics. A well-described flow that affects dissolution in a known way is produced by rotating a disk/cylinder in the melt. By applying the non-dimensional correlations of the mass transfer, the diffusivity of oxides in the slag could be evaluated. The mass transfer coefficient is essentially cons tant across the disk/cylinder surface, simplifying the subsequent mass transfer analysis. A detailed description of this method will be explained in this chapter.

Cooper and Kingery (1964) were the first who applied the rotating disk method to measure the rate of dissolution and diffusion of sapphire in molten slag of 20 wt% CaO- 40 wt%Al2 O3 - 40 % SiO 2 under forced convection at a temperatures range of 1345-1550 C. They calculated the rate of dissolution by measuring the diameter change of the sapphire samples in the melt and found that the dissolution of sapphire was controlled by mass transport in the molten liquid. They also estimated the diffusivity of alumina in the slag, using the solutions for mass transport from rotating disk into the melt. The estimated diffusion coefficients were in the range of 0.6 48 10 8 cm2 /s in which the diffusion coefficient increased with temperature. These results are compared to the results by Henderson, Yang and Derge (1961), who measured the selfdiffusion of aluminium by capillary reservoir technique in two types of CaO SiO 2 Al2 O3 slag with basicity = 1 and Al2 O3 = 6 and 12.5 mole %, in the temperature range of 1400 to 1520C. They found that by decreasing the alumina amount of slag, 62

the diffusion coefficient was increased. The comparison with extrapolated data from Henderson et al. (1961) to higher temperatures shows that both set of data have an order of magnitude of 10-7 and are in a very good agreement (Figure 1.22).

Lee, Sun, Wright and Jahanshahi (2001) studied dissolution of dense alumina discs in slags with composition of (28 48 wt%) CaO (16 25 %) Al2 O3 (7 32 %) SiO 2 MgO (5 %) at 1575 C, by rotating disk method. The dissolution rate was determined by sampling the melt at regular time intervals and measuring the amount of solute dissolved in the slag. It was shown that the dissolution was controlled by the mass transfer in the liquid phase. They also looked at the effect of addition of FeO x and MnO x on the dissolution rate of alumina in the slag. It was shown that addition of mentioned oxides had little effect on activity of Al2 O3 and the driving force of alumina did not change across the liquid boundary layer. They considered the total alumina mass transfer as the combination of the mass transfer from the disk and cylinder side of the immersed sample and it was found that reduction of viscosity due to the addition of transition metal oxides had little effect on the increase of total mass transfer coefficient. However, the authors showed that the increase in the rate of dissolution and mass transfer coefficient were likely to be due to the increase in the diffusion of alumina in the slag phase. So the apparent diffusivity of alumina was obtained (a magnitude of 10-6 cm2 /s) and it was found that addition of transition metal oxides had a considerable effect on increasing the diffusivity of alumina in the slag. (a factor of two for addition of 5 wt% FeO x and a factor of about four for 5 wt% MnO x addition). These data are compared to the extrapolated results from Henderson et al. (1961) on self 63

diffusivity of alumina, where the diffusion coefficient for slag of 12 wt% Al2 O3 and basicity = 1. It appears that for slag without addition of FeO x and MnO x , the diffusivity is slightly lower than Henderson et al. (1961) data but with addition of transition metals, the diffusivity values have the same order of magnitude. The comparison of data is shown in Figure 1.22.

Taira, Nakashima and Mori (1993) used the rotating cylinder technique to investigate the kinetic behaviour of dissolution of sintered alumina into CaOSiO 2 -Al2 O3 slags (basicity from 0.64 to 1.25, Al2 O3 10 wt%) in the temperature range from 1500 to 1580 C. They also looked at the effect of addition of NaF and CaF2 on the dissolution behaviour. They examined the effect of revolution speed, temperature and slag composition on the dissolution rate of alumina into the molten slag. The rate of dissolution was obtained from the reduction in diameter of the alumina specimen in the melt. The dissolution rate increased with increasing revolution speed, temperature and CaO/SiO 2 ratio as well as by addition of NaF and CaF2 . It was concluded that the rate controlling- step during the dissolution process of alumina into molten CaO-SiO 2 -Al2 O3 slag is the diffusion of solute in the slag boundary layer. The dissolution rate of alumina in slags with 15 wt% NaF or CaF2 were 2 to 6 times higher than those for CaOSiO 2 -Al2 O3 slags with the same ratio of CaO/SiO 2 . They evaluated the mass transfer coefficient on the basis of dissolution rate, however they did not measure the mass transfer coefficient when the additives were added to the slag, as the effect of additives on solubility of alumina in their slag was not investigated. The diffusivity of alumina in the slag is deduced by the present author on the basis of the mass transfer dimensionless correlation for rotating cylinder derived by

64

Kosaka and Minowa (1966). The results of mass transfer coefficient and diffusivity are listed in Table 1.4. It appears that there is no trend in change of diffusivity with basicity of slag, but the order of magnitude of diffusion is about 10-7 for temperature of 1500 C and 10-6 for the higher temperatures. These results are in good agreement with results from Henderson et al. (1961) shown in Figure 1.22, where the diffusion coefficient at 1485 C in slag of CaO/SiO 2 = 1, Al2 O3 = 12 wt%, was measured to be 6.1 10 7 cm2 /s and the deduced diffusivity value on the basis of rotating cylinder experiments and for slag of similar composition (slag D) at 1500 C is 6.1 10 7 cm2 /s.

Table 1.4: Mass transfer of alumina in the CaO-Al2 O3 -SiO2 (Al2 O3 = 10 wt%) after Taira et al. (1993) and the deduced diffusivity
Slag A CaO/SiO 2 0.64 temperature (K) 1823 1773 B 0.8 rpm 200 200 100 200 1823 1853 C 0.9 1823 1773 D 1 1823 1853 E 1.25 1823 *Calculated by the present writer 400 600 200 200 200 100 200 400 200 200 mass tranafer (cm/s) 3.00E-05 1.52E-05 2.86E-05 3.49E-05 8.41E-05 1.11E-04 4.16E-05 5.62E-05 3.09E-05 3.86E-05 5.71E-05 1.23E-04 8.31E-05 6.12E-05 diffusion* (cm /s) 1.1E-06 3.1E-07 1.8E-06 1.1E-06 1.9E-06 1.8E-06 1.4E-06 1.8E-06 6.1E-07 1.8E-06 1.5E-06 2.2E-06 2.7E-06 1.5E-06
2

65

Yu, Pomfret and Coley (1997) investigated the dissolution of alumina in the slag of CaO 16.1 wt% Al2 O3 47 % SiO 2 2.9 % Na2 O 3.7 % CaF2 system at 1530C using the rotating disk method. The rate of dissolution of alumina was calculated from the change in the alumina concentration of slag. They applied the boundary layer correlations in rotating disk and estimated the effective diffusivity of alumina in the slag. The diffusivity data were in the range of
7.2 10 8 to 6.8 10 7 cm2 /s changing directly with the amount of Na2 O in the

slag, which was in the range of 2.9 to 11 wt%. The comparison of these results with the data from Henderson et al. (1961) in Figure 1.22 shows the same order of magnitude, where the Na2 O content of slag is greater than 4 wt%.

66

67

Figure 1.22: Comparison of alumina diffusivity data according to Henderson et al. (1961) & Cooper et al. (1964) & Taira et al. (1993) & Lee et al. (2001) (B: basicity, A: Al2 O3 ) B is the basicity, A is Aluminium concentration.

Matsushima et al. (1977) applied the rotating cylinder technique and determined the dissolution rate and mass transfer coefficient of rotating lime into CaO SiO 2 Al2 O3 and FeO CaO SiO 2 slags. They found that the dissolution rate was increased with revolution speed, temperature and reaction time. The authors calculated the boundary layer thickness ( ) form the mass transfer coefficient

(k ) ,

D where the diffusion coefficient k

( D)

was used from the previously

published data by Johnston et al. (1974). It should be noted that Johnstons data was on self-diffusivity but here we are looking at chemical diffusivity, which may be much greater than self-diffusivity. Boundary layer thickness was estimated from the equation derived by Kosaka et al. (1966), which was obtained by the method of dimensional analysis. A comparison of boundary layer thickness calculated from the two methods, showed a good agreement, supporting the postulation that the rate-determining step was the mass transfer through a slag phase boundary layer.

The dissolution rate into slags containing FeO was several times greater than that into slags without FeO. Since the authors calculated the mass transfer of CaO in the slag from the dissolution rate data, the diffusivity of CaO in two types of slag is deduced in the present work from their mass transfer data and on the basis of Kosakas mass transfer correlation. The results are tabulated in Table 1.5.

68

Table 1.5: Values of mass transfer coefficient after Matsushima et al. (1977) and deduced diffusivity of lime in the slag Slag CaO 40 wt% SiO 2 20% Al2 O3 CaO 40 wt% SiO 2 20% FeO Temperature Revolution ( C ) speed (rpm) 1500 1500 1400 1400 200 400 200 400 K (cm/s) 2.9010-4 5.3010-4 9.7010-4 1.7110-3 D (cm2 /s) 6.6710-6 1.0110-5 2.6210-5 3.0810-5

These diffusivity results for slag without iron oxide are compared with the data by Johnston et al. (1974) on chemical diffusivity of Ca45 O in the CaO 62 wt% SiO 2 14.5 % Al2 O3 where the diffusivity of lime was about 5.5 10 6 cm2 /s at 1500 C. The CaO diffusivity results in the slag containing iron oxide are also compared with the data from Hara et al. (1989), where the self diffusivity of Ca45 in the slag of FeO 33 wt% SiO2 7% CaO at 1400C, is 2.7 10 5 cm2 /s. This comparison proves a very good agreement from the lime diffusivity results by rotating disk technique and previously published data.

Umakoshi et al. (1984b) studied the dissolution rate of burnt dolomite in CaOFeO-SiO 2 slag (CaO/SiO 2 =1, FeO = 20 to 70 wt%) at 1350 to 1425C using the rotating cylinder technique. The dissolution rate increased exponentially with the increased stirring rate of the refractory cylinder suggesting that the dissolution rate was controlled by the mass transport. With mass transport in the boundary layer of liquid as the rate controlling step for the dissolution of solid in liquid, the mass flux for the dissolution, J was expressed as:

69

J = k ( n s nb )

(1.7)

Where,

k = mass transfer coefficient (cm/s)

n s , nb = contents of solute at the interface and in the bulk of molten slag (g/cm3 )

The mass flux J (g/cm2 .s) and the dissolution rate dr / dt were defined according to:

dr J = c . dt

(1.8)

Where c is the bulk density in g/cm3 .

They substituted Equation (1.7) into Equation (1.8), which leads to the following rate equation:

k dr k = (c s s cb b ) = b (cs c b ) dt 100 s 100 c

(1.9)

Where;

c = Concentration of solute (%)

s , b = Densities of molten slag (g/cm3 ) at the interface and the bulks
70

It was impossible to determine the mass transfer coefficient of burnt dolomite by Equation (1.9), since CaO and MgO might individually dissolved from the surface of burnt dolomite. Therefore the authors proposed that if the dissolution of CaO in burnt dolomite was much slower than that of MgO, the dissolution rate was controlled by the mass transfer of CaO in the boundary layer of molten slag and the mass flux would be expressed as:

M MgO J = 1 + M .J CaO CaO

(1.10)

On the contrary, if the dissolution rate was controlled by the mass transfer of MgO, the mass flux would be expressed as:

M J = 1 + CaO . J MgO M MgO

(1.11)

Where M CaO , M MgO are the molar weights of CaO and MgO, respectively and
J CaO , J MgO are the mass fluxes for the dissolution of sintered CaO and MgO

cylinders determined from the previously publications by Matsushima et al. (1977) & Umakoshi, Mori and Kawai (1981). By establishing the plots of Equation (1.10) and (1.11) and also checking the linearity of the variables in the two sides of the equations, the authors concluded that Equation (1.10) was valid for the slag with 20 wt% FeO (CaO/SiO 2 = 1) and Equation (1.11) for slags of higher FeO. Therefore, the mass transfer coefficient of dolomite in low FeO slag at 1400C was calculated from Equation (1.12): 71

 M MgO k b J = 1 + M . 100 .(%CaO ) CaO

(1.12)

While the mass transfer coefficient for the other slags from the Equation (1.13).

k M J = 1 + CaO . b .(% MgO ) M MgO 100

(1.13)

The authors calculated the driving force of CaO and MgO according to the relevant phase diagrams. The results of mass transfer coefficients are tabulated in Table 1.6. The diffusivity of CaO and MgO on the basis of the mass transfer coefficient are deduced in the present work by utilizing the dimensionless mass transfer correlation for rotating cylinder developed by Kosaka et al. (1966). The estimated diffusivity results are also listed in Table 1.6.

Table 1.6: Mass transfer coefficient of dolomite from the Umakoshi et al. (1984b) and deduced diffusivity data for CaO and MgO in the present work
Slag Basicity=1 FeO wt(%) viscosity (poise) density (g/cm 3) Mass tranafer (cm/s) 9.15E-04 A 20 1.6 3.1 9.77E-04 9.15E-04 6.81E-04 B C D 30 40 50 1 0.5 0.4 3.2 3.4 3.5 6.45E-04 8.53E-04 7.89E-04 7.43E-04

72

The diffusion coefficient of CaO in slag with 20 wt% FeO was compared to the interdiffuion coefficient of CaO, ( 2.7 10 5 cm2 /s) in the FeO SiO 2 slag measured by Hara et al. (1989) and the diffusion coefficient of MgO in other slags, compared to the MgO diffusivity data of ( 1.5 to 1.8 10 5 cm2 /s), according to Umakoshi et al. (1981).

Effect of additives on the dissolution rate of lime has been investigated in the past by Hamano, Horibe and Ito (2004).They studied the dissolution rate of lime in the slag of FeO 30 wt% CaO 40% SiO 2 at 1573C. They investigated the effect of 10 wt% addition of CaCl2 , Al2 O3 and B2 O3 on the dissolution rate of lime. The dissolution rate of lime was estimated from the reduction in the diameter of the rod used as the solute in their experiments. It was shown that at constant temperature and a given slag composition, the dissolution rate of CaO was governed by the mass transfer in the bulk liquid phase. The dissolution rate of CaO increased with increasing FeO concentration and with basicity of the melt. The effect of additives to the melt was an increase in the dissolution rate of CaO in the order of: CaF2 >CaCl2 >B2 O3 >Al2 O3 .

Umakoshi et al. (1981) applied rotating cylinder method and measured the dissolution rate of sintered MgO into molten Fet O-CaO-SiO2 slags at temperatures from 1350 to 1425 C. It was shown that the dissolution rate of MgO increased with the rotating speed of cylinder and with temperature, and was found to be controlled by mass transport in the boundary layer of molten slag. The authors measured the mass transfer coefficients and utilized the nondimensional correlation in order to deduce the values of diffusion coefficient of

73

MgO into molten slags. The diffusivity of MgO in the molten slag was estimated to be 1.35 10 5 to 2.75 10 5 cm2 /s at 1400 C. Their findings agree well with MgO diffusivity data by Ukyo et al. (1982), where the diffusivity of MgO in FeOx CaO SiO 2 slag between 1350 to 1450 C varies between 1.97 10 5 to
3.05 10 5 cm2 /s.

Xie and Belton (1999) measured the chemical diffusivity of iron oxide in CaO 38 wt% SiO 2 21 % Al2 O3 slag at 1360 C by a rotating disc of solid iron. According to the authors, there is good agreement with the value at near to iron saturation from the galvanostatic studies of Nagata et al. (1982), however, the chemical diffusivity from work of Johnston et al. (1974) and the quasi binary diffusivity determined by Ukyo et al. (1982) are a factor of 5 to 6 higher (Figure 1.23).

It can be seen that the deduced lime diffusivity data obtained from this method agree well with the data from the other techniques.

74

Figure 1.23: Chemical diffusivity of iron oxide in CaO 38 wt% SiO2 21 % Al2 O3 melts in comparison with the results of other studies at 1300 to 1360 C and approximately the same base melt composition as a function of the average iron concentration.

1.6.1

Liquid state diffusion models

The understanding of diffusion phenomena in liquids is inferior in most respects to that of either gaseous or solid state diffusion (Walls and Upthegrove (1964)). The lack of accurate comparison of these and other related transport phenomena in liquids is a consequence of a less complete knowledge of the liquid state. The understanding of diffusion processes in liquids is further complicated by the experimental difficulties encountered in attemp ting to test the various theories proposed to describe diffusion and to predict the effect of relevant system variables upon these phenomena.

75

Researchers have proposed mechanistic models for diffusion in liquids as an attempt to enhance the understanding of liquid diffusion phenomena. These models are reviewed in the following part:

1.6.1.1

Hydrodynamic theory

One of the best known equations relating diffusion and viscosity is that of Einstein (1905). In this theory (Poirier and Geiger (1998)), the diffusing species are non-reacting spherical particles of radius R moving through a continuos medium of viscosity with a steady-state velocity V . The development of this theory is based on Stokes law, which predicts the terminal velocity of the relatively large un-attracting, hard spheres through a liquid. Therefore, the force on a sphere moving at steady state in laminar flow is:

F = 6RV

(1.14)

And upon Einsteins equation, expressing the self-diffusion coefficient as a function of the mobility as:

D = kTM

(1.15)

Where the mobility, M, is the average velocity of the diffusing particle per unit force acting on that particle. To obtain the Stokes-Einstein equation, the mobility is determined from Stokes law as into Equation (1.15) to give:

V and the resulting expression is introduced F

76

D=

kT 6r

(1.16)

Although the Stokes-Einstein equation is derived on the assumption of large solute particles diffusing a continuos medium, the r adii of some liquid metal atoms calculated from this equation show comparatively close agreement with the values of crystallographic ionic radii. In view of the inconsistency between the Stokes-Einstein model and the supposed structure of liquid metals, this agreement is frequently described as merely fortuitous. Indeed, the agreement between crystallographic radii and those calculated using the Stokes-Einstein equation should not be accepted as an unequivocal verification of the validity of the equation to describe diffusion except in an empirical manner. There are several reasons for this conclusion. First, the crystallographic radii are very much dependant upon the rather arbitrary assignment of a radius to one element either from experimental data where the calculated radii are based on the assumed additive nature of ionic radii to give interionic distance in halide and oxide compounds. Second, examination of the radial probability functions from quantum mechanics indicates that no absolute significance should be given to the concept of radii, since these probability functions tail off zero for an infinite radial distance. Third, the reported agreement between Stokes-Einstein radii and crystallographic radii are achieved by judicious selection of the solid state radius value and by subsequently assuming that a given ionic radius is identical in both the liquid and solid state. The fact that the Stokes-Einstein equation does predict approximately the diffusion behaviour in liquid metals implies that Stokes law does provide a good estimate of the mobility. Since the mobility is a ratio of the 77

average velocity (taken to be the terminal value) to the force acting on the particle, both the velocity and the force could be subject to compensating errors and give an approximate value for the mobility.

1.6.1.2

Hole theory

The oldest structural picture of a liquid is the hole theory, which presumes the existence of holes or vacancies randomly distributed throughout the liquid and providing ready diffusion paths for atoms or o i ns. The concentration of these holes would have to be very great in order to account for the volume increase upon melting, thus resulting in much higher diffusion rates in liquids than in solids just below the melting point. The hole theory has been used to estimate the activation energy for self-diffusion in a liquid, by assuming that this energy is equal to that required to form a hollow sphere (hole) of a diameter on the order of a fraction of a nanometre.

1.6.1.3

Eyring theory

Eyring used his activated state theory (Glasstone, Laidlev and Eyring (1941)) to explain the mechanism of diffusion, which worked reasonably well for diffusion in solids and liquids. According to Eyring if the mechanism of activation in diffusion can be assumed identical with that of viscous flow, the relation between the self-diffusion coefficient and viscosity in liquids is given by:

78

 D = 1 kT 2 3

(1.17)

Where 1 is the distance between two adjacent layers and 2 and 3 are the distances between two neighbouring molecules in the moving layer perpendicular to and in the direction of moving, respectively. Since it is reasonable to put the s as the average intermolecular distances, Equation (1.17) reduces to:

D 1 = kT

(1.18)

Based on the Eyrings theory, when a large molecule or ion diffuses or migrates under the influence of an electric field, in a solvent consisting of relatively small molecules, it is unlikely that the rate-determining step will be the jump of the solute molecule from one equilibrium position to the next, since the work required to produce the necessary space would be very large. It is much more probable, that the jump of the solute in one direction is the rate-determining process; the large molecules of solute then moves in the opposite direction into the space left vacant as a result of the motion of the solvent molecule.

Thus the rate-determining mechanism for large molecules diffusing through a liquid composed of small molecules is the diffusion of the smaller molecules around the oncoming large ones by the same mechanism, which these small molecules use in diffusing around other smaller molecules. In Figure 1.24, a

79

small molecule is indicated by A and the large molecule it must diffuse around, is indicated by B. Also, B is for the alternative case of a molecule of the same kind as A. Now B or B will each be advanced the same distance when A in one case flows around B and in the other around B. But the ratio of the distances which A must travel in the two cases is

, where the undetermined number, a, a 2r

will vary with the path followed by A in passing around B or B. We thus expect the ratio of the diffusion coefficients to be:

D = D a (2 r )

(1.19)

Where the diffusion of small molecule around other small mole cule is defined from Equation (1.18), we obtain the diffusion of molecule as:

Dl =

kT a ( 2r )

(1.20)

80

2r

Figure 1.24: Diffusion of large molecule (B) due to the movement of small solvent molecule (A)

Fluctuation theory: Cohen and Turnbull (1959) have proposed an alternative theory for liquid state diffusion based upon the free volume, hard sphere model. This theory, assumes that hard sphere atoms moves randomly within a free volume cell until a fluctuation opens up a path, which permits diffusive displacement of the contained atom. This theory can be regarded as an activation volume analog of the more conventional Botlzmann activation energy concept. This model has been tested by Cohen et al. (1959) using available self-diffusion data to evaluate the radii for the critical free volume per cell required for diffusion. These radii have been shown to be comparable to solid state ionic radii and this result has been taken to support the contention of Glasstone et al. (1941) that the ionic cores are the diffusing particle in liquid melts. The Eyrings equation has been used successfully by several researchers (Eisenhuttenleute 81

(1995)) but values of in Equation (1.18) slightly greater than 2r were needed to obtain good agreement with experimental data.

1.7

General discussion

The above review of the literature revealed that the rate of dissolution of a solid oxide into a molten slag has often been observed to be affected by solubility of solute in the slag, viscosity of the melt, the agitation of bath, formation of a reaction layer of the solute/solvent interface and diffusivity of the solute in the solvent. The formation of the solid product layer is related to thermodynamic and kinetic factors involved in the dissolution process, which can be predicted by thermodynamic modelling or studying the releva nt phase diagrams. The viscosity data are available form the published data or can be predicted by viscosity models. Given the fact the diffusivity of lime plays a significant role in the dissolution of lime in the ladle slag, there is no published data on the lime diffusivity in low silica slag system as all the previous work focused on the high silica slag systems. Therefore, in order to fully understand the mechanism of lime dissolution, there is a need to determine the data for diffusivity of lime in such slags. The understanding of lime diffusivity could also shed lights on the impact of various additives on the diffusivity and possibility of utilizing environmental friendly fluxing agents for enhancing the process of lime dissolution.

In reviewing the diffusivity data by various experimental methods, it has become apparent that the results obtained by different authors often disagree considerably. This is due to inherent experimental difficulties at elevated temperature. In measuring the tracer diffusivity, mass transport by diffusion 82

processes occurs at a very much slower rate than by gross convection processes, which may interfere with measurements of self-diffusivity. Accordingly, experimental methods are required to either minimize convective transport processes or incorporate these effects as a part of the experimental technique under controlled condition.

1.7.1

Questions arising from the literature on diffusivity

What are the diffusivity and dissolution rate of lime and magnesia in the ladle type slag?

What are the effects of additives (like, CaF2 , FeOx, TiO 2 , ilmenite and MnOx ) and temperature on the diffusivity of lime in the ladle type slag?

What are the criteria for selection of experimental technique for measuring diffusivity and which method is the appropriate for the present study?

What are the conditions governing the formation and stability of a solid phase layer at the lime/ladle slag interface and how to monitor the impact of this phase formation on the dissolution rate of lime and diffusivity?

1.8

Methods for measurement of diffusivity

Various experimental techniques for measuring the diffusivity are based on the two classes of diffusion phenomena: self-diffusion (tracer diffusion) and chemical diffusion.

83

Self diffusion involves movement of the various species present in the melt due to the random motions. Self diffusion occurs continuously in any melt and can not be measured because there are no physical manifestations of the process.

Tracer diffusion is the same process as self diffusion, except that a fraction of one or more of the species in the melt is isotopically labelled to establish tracers. The diffusive process can then be monitored by observing the movement of the labelled species. In tracer diffusion both net fluxes and concentration gradients are present, but only for the tracer component.

Chemical diffusion is the movement of species in response to chemical potential gradient in the melt. Such gradients could be created by a number processes including dissolution. Therefore, the diffusivity can be measured experimentally by a tracer or creating a chemical potential in the melt.

The theoretical and mathematical description of the diffusion measurement methods is based on Ficks first and second laws for linear diffusion (in the x direction), whereby the equation of diffusion can be expressed as:

C( x, t ) C ( x , t ) = D[C ( x , t )] x x t

(1.21)

Where C ( x , t ) is the concentration of diffusing species, x is the special coordinate along the path of diffusion, t is time and D is the diffusion coefficient. Unlike diffusion in solid phases, when diffusion in melts is considered, it is usually

84

assumed that the diffusion coefficient is independent of concentration. The equation of diffusion can then be written as:

C ( x , t ) 2C ( x , t ) =D t x 2

(1.22)

The main methods used in the past in measuring diffusivity in melts are outlined below.

1.8.1

Instantaneous plane source method

In this method a tracer (usually radioactive) is deposited as a thin later at one end of long column, and the diffusion coefficient can be derived by the use f the appropriate solution of Ficks law as:

x2 C= . exp 4 Dt DT Q

(1.23)

Where;

C = Concentration of radioactive atoms at a distance x from the deposit

Q = number of radioactive atoms deposited on the surface
D = diffusion coefficient of radioactive species

t = time for diffusion The diffusion coefficient may therefore be obtained by plotting ln C versus x2 . 85

1.8.2

Capillary - reservoir method

In this method, a long capillary tube containing slag is immersed in a large reservoir of the same molten slag with a known content of the element whose diffusion rate is to be measured. The temperature of the system is selected in advance and must be kept constant. After a certain time, during which diffusion takes place, the capillary tube is removed and cooled down (alternatively the whole system may be cooled) and the sample is analysed along its length. The inside diameter of the capillary tube is normally chosen so as to eliminate convection, and is usually 1 3 mm. The capillary is tens of millimetres long, i.e. longer than the distance to which the diffusing species has penetrated. The time allowed for diffusion is usually 103 104 s. A schematic set-up for this technique is illustrated in Figure 1.25.

86

Water-cooled vacuum seal Mullite reaction tube chamotte tube

furnace

Heating elements Graphite crucible Reservoir melt Graphite capillary sample holder

crucible support thermocouple Water cooled vacuum sealed

Figure 1.25: Apparatus for measuring diffusivity of elements dissolved in molten slag by capillary reservoir technique

1.8.2.1

Semi infinite capillary

In this method, the initial and boundary conditions for diffusion are, from Equation (1.22):

87

C ( x, 0 ) = 0

0< x<
(1.24)

C (0, t ) = C 0

t >0

Thus, the concentration of the diffusing species is assumed to be constant in the mouth of the capillary tube. For a semi infinite capillary tube the solution of Equation (1.22) using the conditions given in Equation (1.24), can be written as:

x C ( x, t ) = C 0erf 1/2 2( Dt )

(1.25)

1.8.2.2

Finite capillary

For a finite length l of the column of melt in the capillary tube, comparable with the distance over which the species in the capillary tube has diffused in the course of the experiment, the initial and boundary conditions for Equation (1.22) are:

C ( x, 0 ) = 0

0< xl
C (l , t ) =0 x

(1.26)

C (0, t ) = C 0

t >0

The solution of Equation (1.22), using the condition given in Equation (1.26), is to be found in Crank (1975). The mean concentration C (t ) of a diffusing species may be written as:

88

11 C (t ) = C ( x, t )dx l0

(1.27)

When inserting the solution for C ( x , t ) into Equation (1.27), we again obtain the infinite series that, provided experimental conditions are suitably chosen, converges rapidly. Therefore only the first term of the series need to be considered when calculating the diffusion coefficient, and under such conditions one can obtain

4l 2 8C0 D = ln 2 t 2 1 C( t )

(1.28)

Equation (1.28) is specially suitable for calculating the diffusion coefficient D since it is sufficient to know the mean concentration of the diffusing component
C (t ) in the capillary tube; here, it is not necessary to know the entire

concentration curve as it could be altered and/or displaced, for example by processes taking place during solidification. However, Capillary reservoir methods suffer from a number of shortcomings:

Convection: This increases the apparent value of the diffusion coefficient. Convection may occur:

During immersion of the capillary tube. Because of differences in density, unless the species with the higher density is placed in the bottom part of the system. 89

Because of mechanical effects such as vibration. Because of unfavourable temperature gradients.

To prevent convection it is necessary to use capillary tubes of small diameter, less than 1 3 mm. On the other hand, with such small diameters wall effects may become significant. For example, the diffusing species may attack the wall and so leave the melt, as occurs when oxygen diffusing in molten slag in a capillary with alumina walls reacts with alumina to form phases.

Boundary conditions: The boundary conditions for diffusion as given by Equations (1.24) and (1.26) are not satisfied exactly. At the boundary of the diffusion system the concentration C0 is not constant because of the diffusion in the static melt. Moreover, it is not possible to determine exactly the position

x = 0.

Changes on solidification: The samples are analysed in the as-cooled state, so errors arise from the redistribution of solute during solidification and solid state transformations.

1.8.2.3

Diffusion couple method

This method uses the combination of semi- infinite and finite capillary tubes. It is based on bringing into contact two capillary tubes of the same size filled with slag having different concentrations of the diffusing species in the form of radionuclides (Figure 1.26). To ensure good contact, the two surfaces of the open ends of the capillary tubes are polished.

90

Capillary tube

Melt containing different concentrations of diffusing species

Capillary tube

Figure 1.26: Diffusion couple, two capillaries

An alternative experimental arrangement is shown in Figure (1.23). A capillary tube of 100 200 mm long is filled under vacuum with melt to half its height. After solidification, the remaining half of the column is filled with melt containing the species whose diffusion is to be investigated (the second column having been prepared in another capillary tube). The tube is heated to allow diffusion to occur, and after solidification the sample is analysed along its length. In the finite - source method, the second column of the melt containing the diffusing species is substantially shorter (Figure 1.28).

91

capillary tube

melt containing different diffusing species

melt

Figure 1.27: Diffusion couple, two capillaries

In evaluating the diffusion methods that make use of a combination of semi infinite capillary tubes, it is necessary to take into account the following factors:

The boundary conditions for Equation (1.22) are more closely approached than in the capillary reservoir method in that the undesirable phenomena associated with immersion of the capillary tubes are eliminated.

Convection, and hence mixing, may occur after melting down the samples.

Wall effects are still present.

92

The distribution of concentration of the diffusing species can change during solidification; in some cases rapid quenching can be applied to minimize problems arising from segregation on freezing.

capillary tube

melt containing different diffusing species

melt

Figure 1.28: Diffusion couple, one capillary

1.8.3

Electrochemical method

The electrochemical methods most widely used for studying diffusion in melts are chronopotentiometry, linear voltametry, chronoamperometry, voltametry with rotating-disk electrode (RDE), and polarography. All these processes are controlled by the depolarizer (i.e. the ions of the diffusing species) that has to reach the electrode by either diffusion (as in voltametry, chronopotentiometry, and chronoamperometry) or forced convection (as in RDE and polarography). Transport by migration, i.e. the motion of ions under the electric field in solution, is not usually considered in view of the high electrical conductivity of the melts. 93

The electrochemical method is especially suitable for measuring diffusion in melts as convection is either eliminated or controlled. The experiments last only a few seconds, in contrast to the several hours needed for the capillary methods, and wall effects are not encountered. Of these methods, chronopotentiometry and RDE are most important in the study of diffusion in melt.

Chronopotentiometry is a galvanostatic method in which pulse of constant current density is used to perturb a system from its equilibrium. The response of the system is measured in terms of the dependence of the potential of the investigated electrode on time. The accuracy of the chronopotentiometric method in determining diffusivity varies with experimental conditions. At medium temperature it may be 2 30 %, while at high temperature the accuracy is unknown.

An electrochemical cell is established which contains a metal electrode containing the diffusing species and the molten slag composition of interest and electrolysis is carried out at uniform current (I). If the electrode reaction is controlled by diffusion, Equation (1.29) can be used to derive the diffusion coefficient as;

dC I = ZFD dy =0 y

(1.29)

Where, Z is the valence of the diffusing species, F is the Faraday constant; D is dC the diffusion coefficient and dy is the concentration profile. 94

The Rotating disk electrode (RDE) is the electrochemical variant of the rotating method which will be described in the next section, and is known as voltametry with RDE. An increasing voltage is applied to the RDE and the current is measured. The increase in voltage is very slow owing to the diffusion processes and thus the phenomenon is quasistationary. Voltametry with RDE has been applied more in aqueous solutions than in melts.

Generally, serious difficulties have been faced in the mentioned methods to accurately measure diffusion coefficients in liquid slags at elevated temperatures. The main difficulty is to avoid mass transport by bulk motion of the fluid caused by natural convection, unless the fluid flow is a part of the measuring method. This fluid flow is driven by the buoyancy force produced by any temperature and concentration gradients, which leads to a decrease in liquid density with depth.

1.8.4

Controlled forced convection method

In these methods the errors due to convections are minimised by imposing controlled forced convection where the hydrodynamic conditions are well defined. A sample of solid oxide is rotated in the molten slag and the diffusion coefficient is deduced from the measured rate of dissolution with knowledge of chemical driving force and boundary layer thickness. Since the hydrodynamics of rotating disk is well established, a rotating disk or cylinder can be used. Advantages of this method include:

The experimental conditions, including slag dynamics are exactly reproducible.

95

The slag is agitated to a sufficient degree to render a comparison with the conditions in the real industrial practice.

The mass transfer from the solid oxides to the slag under forced-convection has been modelled successfully in the past by many researchers. The rotating sample of solid oxide used in this method could be in the forms of a disk or a cylinder. The following part will explains the hydrodynamic conditions involved in each type.

1.8.4.1

Rotating disk method

In this method dissolution of a rotating solid disc in a static melt is measured; the rate of rotatio n is chosen so that laminar flow is achieved. The diffusion coefficient is determined from the dissolution rate, angular velocity, and from the time of dissolution. The dissolution rate is obtained from either the loss of the weight of the disk or the reduction in diameter of the rotating disk or the increase in the concentration of dissolving solid oxide in the melt. The diffusion boundary layer is right next to the surface of rotating disk; the matter is considered to be transferred by molecular diffusion. Equations for the tangential, radial, and axial contributions to fluid flow near the surface of a rotating disk have been derived by Cochran (1934). With the use of Cochrans equations, Levich (1962) obtained the equations for boundary layer thickness of a solute species dissolving from a rotating disk. The diffusion boundary layer is not sharply delineated, and its thickness varies with the hydrodynamic conditions, i.e. with the thickness 0 of the hydrodynamic (Prandtl) layer and with the value of the diffusion coefficient: 96

D = 0.5 0

1/ 3

0
(1.30)

0 = 3 .6

1/2

Where is the kinematic viscosity and is the angular velocity of rotation of the disk. Under common experimental conditions, and 0 have values of ~1 and ~0.01 mm, respectively.

To solve the diffusion equation, it is necessary to know the velocity of the vertical ( y ) direction:

3 1/ 2 2 ( y ) 0.51( ) y

y <<
y
y

1/ 2

(1.31)

( y ) 0.89 1 / 2

(1.32)

Transport due to simultaneous convection and diffusion can be determined when the convective flow in the melt is known. The equation of convective diffusion (i.e. mass transfer due to diffusion and convection) in one dimension is:

97

C C 2C +y =D 2 t y y

(1.33)

Where, y is the flow rate in the y-direction and C is the concentration.

The equation for convection-disc method is a stationary example of the masstransport equation for convection and diffusion, i.e.
C = 0 in Equation (1.33): t

dC ( y ) d 2 C( y ) =D ( y) dy dy 2

(1.34)

The boundary conditions are:

C ( y = 0) = C sat

lim C ( y ) = 0
y

(1.35)

Where C sat is the equilibrium concentration at the disc/melt boundary, or the concentration of saturated solution at the disk surface. Equations (1.31) to (1.35) are fundamental formulas for the rotating-disk method.

By solving Equation (1.34) under the boundary conditions given by Equation (1.35), and with the velocity ( y) given in Equation (1.31), it can be shown that:

98

y f( ) C ( y ) = Csat 1 y ) f(

(1.36)

Where;

y f( )=

y /

exp(
0

) d
(1.37)

y f ( ) = 0.8934

To determine the diffusion coefficient by the rotating disk method, it is necessary to know the flux density j of material from the disc surface dissolving into the melt:

dC ( y ) j = D dy y =0

(1.38)

Substituting of Equations (1.30), (1.36) and (1.37) into Equation (1.38) gives:

j = 0.62 D 2 / 3 1 / 6 1/ 2C sat

(1.39)

The Levich correlation (1.39) was used by many to prove that the dissolution is controlled by liquid-phase diffusion. When that is the case, the dissolution rate should be proportiona l to the square root of the angular velocity of the disk. The 99

diffusion coefficient D is determined from experimentally derived values of j , i.e. from the rate of dissolution.

Sandhage et al. (1990) studied the dissolution of sapphire in calcia- magnesiaalumina-silica melts at 1450 and 1550 C. Prior to performing the sapphire dissolution experiments in melts, they conducted a low-temperature modelling study, involving glycerol as the model fluid and aluminium cylinders as specimens, in order to determine the proper specimen and crucible dimensions. The most ideal flow pattern was observed for a 1.3 cm diameter cylinder stirred at less than 2000 rpm in glycerol contained in a 7.6 cm diameter crucible. For the mentioned dimension, when alumina cylinder rotated in the glycerol, the glycerol rose toward the cylinder along an inner, helical path about the rotation axis. When the liquid was near the bottom of the rotating cylinder, it was thrown radially outward. The liquid then sank along an outer, helical path. Thus, the flow pattern near the bottom of aluminium cylinder was consistent with the velocity distribution predicted by Cochran (1934).

The key benefit of the rotating disk is that the mass flux in the axial direction from the disk is not a function of radial position, and is therefore, constant over the entire disk face. This uniformity of the mass flux across the disk surface, simplifying the subsequent analysis and makes the rotating disk ideal for masstransfer experiments. Therefore, The rotating disk is employed largely in electrochemistry and is convenient for studying chemical kinetics under laboratory conditions.

100

1.8.4.2

Rotating cylinder method

Generally, the mass-transfer coefficient is a function of geometry and fluid properties of flow through the use of dimensionless numbers such as the Reynolds number (Re) and Schmidt number (Sc). These dimensionless numbers are defined as Re =

Vd , where is the density, V is the linear and Sc = D

velocity, d is the characteristic length, is the viscosity and D is the diffusivity of species in the fluid.

Chilton and Colburn (1934) by analogy with heat transfer, developed an approximation relationship between these variables in terms of a mass-transfer
( j ) and dimensionless numbers as:

jD =

k 2/ 3 Sc = b Re n V

(1.40)

Where

j D = mass-transfer j factor from the cylinder side of the crucible

V = linear (peripheral) velocity of rotating disk Sc = Schmidt number k = mass transfer coefficient of the cylinder Re = Reynolds number based on the peripheral velocity of the cylinder

101

The constants, b and n, can be determined from experimental data. Generalized relationships of mass transfer coefficient have been obtained experimentally for many specific geometries. Eisenberg and Tobias (1955) measured rates of mass transfer at circular cylinders rotating about their axes in the centre of stationary cylinders by means of solid dissolution and electrolytic reactions. Benzoic and cinnamic acids cast into cylinders were dissolved into water and water- glycerol solutions.
Re =

The characteristic length dimension for the Reynolds number,

dul , was found to be the diameter of the rotating cylinder, instead of the

gap between the concentric cylinders. Their study involved a large variation of cylinder diameters (1.94 to 5.02 cm). For the diameters studied, the magnitude of the gap was found not to affect the rates of mass transfer even under turbulent flow conditions. They also covered a range of Schmidt numbers from 835 to 11490 and of Reynolds numbers from 112 to 241000. The functional dependence of the mass transfer coefficient on physical properties of the system was found to be represented by finding factors b and n of Equation (1.40). Therefore, Eisenberg et al. (1955) obtained a general correlation of the mass transfer coefficient as Equation (1.41), which is a function of Reynolds number based on the rotor diameter.

j cylinder =

k cylinder V

Sc 0.644 = 0.0791 Re

0 .3

(1.41)

Where;

j cylinder = mass-transfer (j factor) from the cylinder side of the crucible

102

V = peripheral velocity of rotating disk Sc = Schmidt number

k cylinder = mass transfer coefficient of the cylinder

Re = Reynolds number based on the peripheral velocity of the cylinder

Kosaka et al. (1966) developed a correlation for the mass transfer from a rotating metal cylinder into liquid metal at about 1400 C (Equation (1.42)). In their research, they employed Steel- Al, Steel- Zn, Cu-Pb, Zn-Hg and Sn-Hg as the combination of solid metal cylinder-liquid metal bath.

j cylinder =

k cylinder V

Sc 0.644 = 0.065 Re 0. 25

(1.42)

A number of researchers measured experimentally the relationship between the mass transfer (J- factor) from the rotating cylinder with the Reynolds number. The solute and the properties of the solvent vary in different experiments as shown in Table 1.7.

103

Table 1.7: The correlations developed previously for mass transfer from rotating solute cylinder to the solvents. Solute Salt Metals Dolomite MgO CaO Al2 O3 Solvent Water organic liquid Liquid metal FeO x-CaO-SiO 2 CaO-FeO x-SiO 2 CaO-SiO 2-Al2 O3 CaO-SiO 2-Al2 O3 Mass transfer coefficient & Reynolds relationship j = 0.664 Re 0. 5 j = 0.065 Re 0.25 j = 0.152 Re 0.36 j = 0.126 Re 0.30 j = 0.495 Re 0.31 j = 0.048 Re 0.19 Researcher Eisenberg et al. (1955) Kosaka et al. (1966) Umakoshi et al (1984a) Umakoshi et al. (1981) Matsushima et al. (1977) Taira et al. (1993)

The logarithmic relation of J- factor and Reynolds number is also shown in Figure 1.29. Umakoshi, Mori and Kawai (1984a) explained the reason for larger values of J-factor compare to the data obtained by Eisenberg et al. (1955) and Kosaka et al. (1966). They claimed that such differences may be caused by the underestimation of the net area for the dissolution process which may be larger then the geometric one because the rotating cylinder is porous. They also claimed that high J - factor of lime dissolution obtained by Matsushima et al. (1977) compared to the other solid oxide dissolutions could be explained by spalling, i.e., the mechanical separation of CaO particles from sintered lime during dissolution. Taira et al. (1993) measured the dissolution of alumina into molten CaO-SiO 2-Al2 O3 slags. They claimed that relationship between J- factor and Reynolds number shown in Figure 1.29, is close to extension of line for Kosaka relation.

104

Figure 1.29: The relationship between the mass transfer and Reynolds number according to the previous investigations.

Umakoshi et al. (1981) measured the dissolution of MgO into molten FeO x -CaOSiO 2 slags at temperatures from 1350 to 1425 C. They applied the correlation developed by Kosaka in their calculations to measure the mass transfer coefficient. Therefore, the Kosaka correlation might be applicable to the dissolution of a cylindrical solid oxide into the molten slag.

1.8.4.3

Applicability of rotating disk/cylinder technique

A number of researchers utilized the rotating disk/cylinder technique to determine the dissolution rate and mass transfer and diffusivity of solid oxides in the slag. It was shown in the previous section that data on diffusivity of alumina obtained with this method by Cooper et al. (1964) & Taira et al. (1993) & Lee et 105

al. (2001) were in a good accord with the results from Henderson et al. (1961) measured by capillary-reservoir technique in a slag of similar chemistry.

Although there is no published data on the determination of diffusivity of lime by rotating disk/cylinder method, Matsushima et al. (1977) & Umakoshi et al. (1984b) applied this method to measure the dissolution rate and mass transfer coefficient of lime and dolomite in the slag. As it was shown in the previous section, the deduced values of apparent diffusivity on the basis of non-dimension analysis of mass transfer data shows a very good agreement with the data from direct measurement of diffusivity with instantaneous plane source and capillaryreservoir techniques by Johnston et al. (1974) & Hara et al. (1989), respectively, with similar slag chemistries.

The excellent agreement on values of diffusivity of MgO obtained with rotating cylinder technique by Umakoshi et al. (1981) and inter-diffusivities data of MgO with diffusion couple method by Ukyo et al. (1982), is also another proof on the reliability of this technique for measurement of diffusivity in the slag.

The comparison between the diffusivity of iron oxide measured by Xie et al. (1999) with a rotating disc of solid iron in the slag and data on the galvanostatic diffusion studies of Nagata et al. (1982) shows a very good agreement .

1.8.5

Selection of experimental technique for the present work

In general, one can conclude the appropriateness of the rotating disk/cylinder technique for measurement of diffusivity of solid oxides in the slag as: 106

The error caused by convection in measuring tracer diffusivity is minimised by creating a forced convection under well-defined hydrodynamic condition.

The experimental conditions, including slag dynamics are exactly reproducible.

The slag is not static and is agitated to a sufficient degree to render a comparison with conditions in steelmaking process.

Furthermore, according to Levich equation, if dissolution rate of rotating solid oxide is controlled by mass transfer, then varying the rotation speed will vary the thickness of the liquid boundary layer adjacent to the disk and hence change the mass transfer coefficient in the slag phase (Lee et al. (2001)). For a given rotation speed, the effects of slag chemistry (such as addition of CaF2 and transition metals) and temperature on the dissolution rate, mass transfer and hence the diffusivity can also be determined by this technique.

1.9

Objectives of this work

Determine the diffusivity of lime in CaO 42 %Al2 O3 8 % SiO 2 slag by the method of rotating disk/cylinder technique.

Examine the effects of formation of a solid reactio n product on the slag/lime interface on the dissolution rate and develop a model of dissolution in the presence of a solid layer.

107

Quantify the effects of additives such as CaF2 , Fe2O3 , TiO 2 , Mn3 O4 , SiO 2 and ilmenite on the dissolution rate, diffusivity and solubility of lime in the slag.

Quantify the diffusivity of magnesia into the CaO 55 wt% Al2 O3 slag and investigate the impact of additives such as Fe2 O3 and (CaF2 + Fe2 O3 ) on dissolution rate, diffusivity and solubility of magnesia in the slag.

108

CHAPTER 2.

Experimental

This chapter explains the experimental setup and material used in this work. Firstly, the outline of experimental work for measurement of diffusivity is described. Secondly, details of the material and equipment used in this work are presented. Thirdly, the procedures used to perform dynamic and static dissolution experiments are given. Finally, the methods used to analyse the samples are described.

2.1

The outline of the experimental work

The dissolution rate of CaO and MgO in the slag was determined using rotating disk/cylinder technique. The rate of dissolution was obtained by measuring the concentration of dissolving oxide in the melt at different time intervals by sampling of the melt. The diffusivity of CaO/MgO was deduced from the dissolution rate data.

According to the literature review, the rotating sample in this technique is either a cylinder or a disk. In the case of the rotating disk, the effects of the edge of the disk on the hydrodynamic mass transport of material from the disk surface have not been precisely quantified. On the other hand, it is often very difficult to immerse the disk in melt in a way to just have the flat bottom of the disk in the liquid and the effect of walls on the fluid dynamics and mass transfer would cause an error. Also in the case of a rotating cylinder, the researchers normally cover the end part of the cylinder (as a disk) with a cap, so that the dissolution of 109

the solute would proceed only at the side surface of the cylinder; however, the bottom of the cylinder affects the fluid dynamics of dissolution. Thus, it was decided to use a cylindrical crucible and combine the effect of dissolution of disk and cylinder sides of the sample in the process of dissolution, which will be discussed in detail in Chapter 4. The dense lime/magnesia crucible had low porosity in order to minimize the effect of porosity on the diffusivity. The base slag considered in the present study has the composition of CaO 42 wt% SiO 2 8 % Al2 O3 with addition of 5 wt% CaF2 , FeO x , TiO 2 , ilmenite, MnO x and SiO 2 , where we have some knowledge of their phase diagram, viscosity and density over the temperature range that was investigated.

In the experiments for measuring the diffusivity of MgO, a dense magnesia crucible was used and the slag investigated was a CaO 55 wt% Al2 O3 slag.

2.2
2.2.1

Material preparation
Dense CaO / MgO crucible

Dense CaO crucibles, 20 mm ID, 16 mm OD and 30 mm high, were supplied by Rojan Ceramics Pty Ltd. in Australia. The crucibles had low porosity (< 1%) and contained more than 96.3 wt% CaO with 2.9 wt% of MgO as grain bonding phase. These crucibles were used as cylinders in the rotating experiments.

Dense MgO crucibles; with the same dimension as lime crucibles were provided by Fuji Sho Inc. Japan.

110

2.2.2

Chemical reagents

The master slags were prepared from Reagent grade chemicals. Table 2.1 details the chemicals (CaCO3 , SiO 2 , Al2 O3 , CaF2 , Fe2O3 , TiO 2 and CaF2 ) used in this work. The purity and powder size of these chemicals are also included in this table. All reagents were dried at 120C for at least 12 hours before use to remove moisture resulting in accurate weight. All the materials were weighted with an electronic balance to a precision of 0.001 grams.

Table 2.1: The source and purity of the chemical composition used in the experiment Chemical CaCO3 SiO 2 Al2 O3 Fe2 O3 TiO 2 CaF2 MnO2 Supplier Ajax Finechem Company, Inc. Consolidation Chemical, Inc. Ajax Finechem Company, Inc. Alrich Chemical Company, Inc. Alrich Chemical Company, Inc. Ajax Finechem Company, Inc. Alrich Chemical Company, Inc. Comments Purity > 99% Analytical reagent Purity ~ 99.8% Purity > 97% Analytical reagent Purity ~ 99% Analytical reagent Purity > 99.9% Analytical reagent Purity > 97% Laboratory reagent Purity > 99% Analytical reagent

Table 2.2 provides the chemical composition of ilmenite used in this study. Ilmenite was provided by Cable Sands Pty Ltd. Australia.

111

Table 2.2: Chemical composition of ilmenite ilmenite Oxides TiO 2 Fe2 O3 FeO MnO Al2 O3 SiO 2 Cr2 O3 P2 O5 Wt% 55.3 24.1 16 1.48 0.58 0.93 0.045 0.03

2.2.3

Preparation of calcium aluminosilicate master slag

Master slags were prepared in a way to represent the composition of ladle type slag. The calcium aluminosilicate master slag was made in a 15 KW, 450 KHz induction furnace at 1500 C in air. The starting materials for the slags were reagent-grade CaCO3 , SiO 2 , and Al2 O3 that were dried at 120C for 12 hours. The required materials were then weighed and then mixed in a plastic container on a rotating mill for 1 hour. The mixed powder was melted in a graphite crucible (90 mm OD and 300 mm high) contained in a clay bonded graphite susceptor, which was heated to 1500 C with a heating rate of 600 C per hour. The slag was held at 1500 C for half an hour as the graphite crucible was charged with the remaining powder. After all the slag powder was charged into the graphite crucible, nitrogen gas was injected to mix the molten slag, by blowing through an alumina lance into the slag for half an hour. The slag was 112

cooled down in the graphite crucible and then the slag was separated from the crucible by breaking the graphite crucible. The slag was then initially crushed in a jaw crusher and then in a tungsten carbide ring mill. The resultant very fine slag powder was again mixed in a plastic container on a rotating mill for one hour to ensure the homogeneity of slag. As the slag contained some graphite particles, it was de-carbonized in a platinum dish heated in air in a muffle furnace at 800 C for 12 hours. The master slag was analysed by XRF analysis at CSIRO, division of minerals and the results are shown in Table 2.3.

Table 2.3: XRF analysis of master slag, wt%

Sample CaO Al2 O3 SiO2 Fe 2 O3 CuO K2 O MgO Na2O P 2O5 42.2 7.8 0.2 0.01 0.01 0.04 SO3 Sum

Master slag 48.8

0.04 <0.005 <0.005 99.1

The slags with additives were made by pre-melting mixtures of reagent-grade CaF2 , Fe2 O3 , TiO 2 and MnO 2 with the master slags. The additives were dried at 120 C for 12 hours. After weighing and mixing the material, the mixture was melted in a platinum crucible in a muffle furnace, which was programmed to attain 1500 C with a ramping rate of 300C per hour, the atmosphere inside the muffle furnace was air. The melt was then poured onto a cold copper plate to quench. The quenched slag was then pulverized in a tungsten carbide ring mill. The final composition of slags were analysed using the XRF technique, the results are shown in Table 2.4.

113

Table 2.4: Chemical composition of various slags for lime dissolution study, wt%
Sample slag + CaF2 5% slag + FeO 5% slag + TiO 2 5% slag + ilmenite 5% slag + MnO x 5% slag + SiO2 5% CaO 47.79 47.3 47.5 47.3 47.6 47.4 Al2 O3 39.8 40.3 40.3 39.8 40.3 40.2 SiO2 8.09 7.52 7.76 7.94 7.84 12.8 CaF2 4.82 FeOx 5.09 2.13 TiO2 4.93 2.82 MnOx 4.6 -

2.2.4

Preparation of calcium aluminate slag

The calcium aluminate master slag for the study of dissolution of magnesia, was prepared in the same method described above. The composition of base slag and slags with additives are listed in Table 2.5.

Table 2.5: composition of slag with additives for magnesia dissolution study, wt% Sample master slag slag + FeO 5% slag + FeO 10% slag + CaF2 5% +FeO x 5% slag + CaF2 5% +FeO x 10% CaO 45 42 40.2 41.5 39 Al2 O3 55 50.7 48.7 49.0 45.8 FeO x 5.4 9.7 5.3 9.9 3.6 3.7 CaF2 -

114

2.3

Experimental apparatus

The dissolution kinetic experiments were carried out in a vertical tube furnace. A schematic diagram of the furnace arrangement is shown in Figure 2.1. A Pythagoras furnace tube (50mm ID and 800 mm high) was heated by three U shaped MoSi2 heating elements. The temperature was controlled with type R thermocouples (Pt/ Pt-13% Rh) positioned outside the Pythagoras furnace tube and connected to a Eurotherm Controller, which maintained temperatures to 1 C. Both ends of the Pythagoras tube were sealed by water-cooled brass endcaps. A type-R thermocouple was used for measuring the temperature of the crucible. This thermocouple was cemented to the alumina platform that the platinum crucible sat on.

115

Figure 2.1: Schematic of the experimental apparatus used for the rotating cylinder tests

116

2.4
2.4.1

Experimental procedure
Rotating experiments

Dissolution rate of dense CaO/MgO in slags with additives was measured by carrying out experiments using the rotating disk method in a tube furnace. The experimental set- up is shown in Figure 2.1. The lime/magnesia crucible (20 mm OD and 30 mm high) was cemented with Zirconia paste to the end of an alumina tube (8 mm OD), which was then cemented to a stainless steel shaft (Figure 2.2).

Figure 2.2: Photo of the CaO/MgO crucible attached with Zirconia paste to the alumina rod

The steel shaft was driven by an electric stirrer at constant speed. The rotation speed of the motor could be varied between 20 and 900 rpm. The speed of stirrer was checked regularly with a digital tachometer. 117

In each experiment about 60 grams of slag was used. The slag was contained in a Pt 10% Rh crucible (40 mm in diameter and 44 mm in height) placed on a platform that could be manually lowered and raised to adjust the immersion depth of the sample in the slag. An R-type thermocouple positioned beneath the platinum crucible was used to monitor the temperature of the melt.

The tip of the lime crucible was located in the centre of the hot zone and the platinum crucible was positioned 10 mm below the lime crucible during the furnace heat-up, at a rate of 120 C/hour. After reaching the target experimental temperature, the slag was allowed to homogenize for 1~2 hours. Dried air was used to control the atmosphere in the tube furnace.

The platinum crucible was then raised to immerse 15 ~ 20 mm of the lime sample so that there was always 10 mm distance between the bottom of the lime crucible and bottom of platinum crucible; and then the rotation was started. The molten slag was sampled at regular time intervals using a cold 2-mm-diameter platinum rod attached to a stainless steel tube from the top of the furnace. The Pt rod was dipped into the melt, and quickly removed from the furnace. The slag attached to the tip of the Pt wire was rapidly quenched in water. Then the slag sample was crushed and dried for 1 hour to remove any moisture. The slag samples were crushed fine for XRF analysis. The XRF results were then used to measure the concentration of the CaO dissolved in the slag. The dissolution rate of CaO was determined from the variation in composition of slag with time. After the last sampling for the required period, the Pt crucible was lowered and the furnace was cooled down at the rate of 180 C per hour. A similar procedure was used for measuring the dissolution of MgO into slags. 118

2.4.2

Static experiments

The solubility of lime in the slag and also the possible formation and growth rate of solid phases formed on the CaO crucible/slag interface were studied by carrying out the series of static experiments. The experiments were carried out using a muffle furnace. In each experiment a platinum capsule was used to contain the slag and a piece of lime. The small platinum capsules used in the experiments (15 mm ID by 32 mm height) were made by welding two sides of a thin platinum foil. About 0.5 gram of slag and 0.6 ~ 0.7 gram of dense chipped CaO from crucible pieces (which was used for rotating experiments) was contained in the platinum capsule and were placed in the shallow holes drilled in refractory bricks in order to hold the platinum crucibles vertically. After reaching the target temperature and staying at temperature for the required reaction time, the platinum capsule were taken out of the muffle furnace and rapidly quenched on a brass plate, which was also cooled by an air flow on its back surface. The platinum capsule with their contents were mounted in resin and then cut and polished for SEM analysis.

2.5

Analytical techniques

Approximately 0.5 grams of the finely ground, oven dried slag power resulted from sampling was accurately weighed into an 95% Pt 5% Au crucible with approximately 6 grams of 12:22 lithium tetraborate/metaborate flux. The mixture was fused into a homogeneous glass over an oxy-propane flame at a temperature of approximately 1050 C. The molten material was poured into a 32 mm diameter 95% Pt Au mould heated to a similar temperature. The melt was then 119

cooled by air jets for approximately 300 seconds. The resulting glass discs were analysed on a Philips PW2404 XRF system using a control program developed by Philips and algorithm developed at CSIRO Minerals by P.G.Fazey. Oxygen in the XRF analysis was calculated by assuming stoichiometry for the oxides species. The error involved in the XRF analysis in the present work was within 1% for the analysis of the low concentration components.

2.5.1

Scanning Electron Microscopy and Energy Dispersive System analysis (SEM-EDS)

The samples from static experiments were mounted in epoxy resin and cut with a diamond saw, then they were ground initially with the Struers waterproof silicon carbide paper at consecutive grits sizes of 600, 800, and 1200 m using a Struers Labopol-5 grinding machine at 300 rpm. Since the samples were very sensitive to moisture, kerosene was used as the lubricating fluid. Diamond polishing of the specimen was done using polycrystalline diamond paste of 6, 3 and 1 m microns consequently on Chemo-textile Cloth (Leco-PAN-W). A LECO G25 Rotary Polisher was used f or the final stage of polishing. Finally, the samples were coated with carbon layer (thickness:~ 200 A) using a vacuum evaporator to provide a conducting surface.

A Philips XL30 Scanning Electron Microscopy (Figure 2.3) at School of Chemical Engineering, University of Melbourne equipped with an Oxford Link ISIS Energy Dispersive System and an ATW Pentafet SiLi detector was used for the SEM analysis of examination of mounted samples. The electron optical system accelerating voltage was 20 KV. The various phases present in the

120

samples were identified from the different crystal forms and colour intensities displayed in back-scattered electron (BSE) images generated by SEM. The EDS was then used to quantify the chemical composition of the phases. The EDS was calibrated with Bedrock Scientific Ltd standard reference block SB 1/a. The standards were carefully selected to be free of line overlaps and were stable inorganic compounds or single elements. Table 2.6 lists the standards used in the calibration of the EDS in this work.

Figure 2.3: The Philips XL30 used for the SEM analysis

121

Table 2.6: Standards used in the calibration of Philips XL 30 SEM Elements Ca Al Si Mg Fe Ti Mn F O Standards CaSiO 3 Al2 O3 SiO 2 MgO FeS2 Ti MnSiO 3 LiF SiO 2

2.5.2

Microprobe analysis

A CAMECA SX50 electron probe micro-analyser (EPMA) at the Electron Microscopy Unit, The University of New South Wales was used in quantitative analysis of the samples (Figure 2.4). It is a fully automated instrument employing four wavelength dispersive spectrometers in order to analyse various elements. These elements were analysed with the TAP, PCO and PET crystals. All samples were examined using an accelerating voltage of 15 KV, a beam current of 20 nA and a beam size of 1 micron. The instrument was operated with SAMx application software. X-ray intensity distributions were acquired for the main constituents to produce elemental analysis across the area of interest. The calibration was performed by taking peak and background measurements on a 122

standards listed in Table 2.7. All standards are in the Austimex Block, which was polished under perfect sample preparation by Mr. Rad Flossman at UNSW.

Figure 2.4: The CAMECA SX-50 Micro Probe used for the EPMA analysis

123

Table 2.7: Standards used in the calibration of SX-50 Micro probe Elements Ca Al Si Mg Fe Ti Mn F O Standards Diopside Sanidine Diopside Diopside Haemetite Rutile Rhodonite Calcium fluoride Diopside

124

CHAPTER 3.

Experimental results

The results of the present work are detailed in this chapter. The results are presented in two main sections: rotating disk experiments in section 3.1, and static experiments in section 3.2. The results from dissolution of CaO in calcium aluminosilicate based slag under forced convection in air are presented in section 3.1.1. The effects of variables such as rotation speed, temperature and slag chemistry (additives; CaF2 , TiO 2 , Fe2 O3 , Mn2 O3 , ilmenite and SiO 2 ) on the dissolution of lime are also presented in this section. The apparent dissolution rate was deduced from the variation of dissolved lime concentration in the slag with time and the effect of variables on the dissolution rate is presented in this section. Finally, results from dissolution of MgO in calcium aluminate based slag in air, at 1430 C and under forced convection are presented in section 3.1.1.5. The variation of dissolution with rotational speed and effects of additives such as Fe2 O3 and (CaF2 + Fe2 O3 ) on the dissolution rate are summarized in this section.

The results from experiments carried out under static condition are presented in section 3.2, reaction of lime with base slag produced a solid phase at the lime/slag interface and the growth rate of the layer is deduced. The measurements on solubility of lime in the base slag and slags with additives such as (CaF2 , TiO 2 , Fe2 O3 , Mn2 O3 , ilmenite and SiO 2 ) at various temperatures are presented in this section. This section also contains the results from solubility of MgO in calcium aluminate based slag and slags with addition of Fe2 O3 and (CaF2 + Fe2O3 ) at 1430 C.

125

3.1
3.1.1

Rotating experiments
Dissolution of CaO in calcium aluminosilicate slag

The CaO dissolution in the calcium aluminosilicate slag was investigated by rotating a lime sample in molten slag in a platinum crucible. The initial slag composition was CaO 42 wt% Al2 O3 8% SiO 2 as shown in Figure 3.1. Lime crucible was used as the rotating samples in melt with dimensions of 20 mm in diameter and 30 mm in height. The selection of this slag chemistry was based on the consideration of phase diagram and typical ladle slag in steelmaking. The experiments were carried out in air, initially at 1430 C, with rotational speeds ranging from 30 to 150 rpm and with the reaction time of up to 1 hour. These experiments were followed by investigation on the effect of temperature and slag chemistry, i.e. addition of CaF2 , TiO 2 , Fe2O3 , Mn2 O3 , ilmenite and SiO 2 on the dissolution of lime at constant speed. The amount of slag used in each test was about 60 grams. The CaO concentrations in the bulk slag were determined by sampling of molten slag at time intervals. The lime dissolved in the slag was analysed using XRF, with the estimated error being within 0.2 0.3 wt% in the slag phase. All the concentration data were smoothed by Rational curve fitting using Curve Fitting Toolbox in MATLAB (MATLAB (2000)). The raw experimental data are presented in CaO (wt %) versus time graphs in the form of dot points throughout this chapter and also in Appendix A. The curve fittings are also presented in the form of continuous curves going through the experimental data in each graph. Both raw experimental data and corresponding curve- fitted values are tabulated as separate columns in tables relevant to the CaO concentration versus time graphs. 126

Figure 3.1: CaO-Al2 O3 -SiO2 system phase diagram from Slag Atlas (Eisenhuttenleute (1995))

3.1.1.1

Effect of rotating speed on dissolution rate

These experiments were carried out to study the effect of increasing the rotation speed from 30 to 150 rpm on the dissolution/dissolution rate of lime in the master slag in air and at a temperature of 1430 C. It was observed that by increasing the rotation speed from 30 rpm to 150 rpm, the dissolution rate was increased by a factor of 13.5. The fact that the dissolution rate was increased with the rotation speed implies that mass transfer in the slag played a significant role in the dissolution of lime in the slag phase.

127

The curve- fitted results are shown in Figure 3.2. The raw data are detailed in Appendix A.1. The initial lime concentration was a bit different for various speeds, where the slag made for each experiment had slightly different slag chemistry as the target lime content of slag was about 50 wt%.

Figure 3.2: The concentration of CaO (wt%) in the melt with increasing the rotation speed at 1430 C

The initial dissolution rate (g/(cm2 .s)) was derived from the slope of CaO concentration curves with respect to dissolution period and on the basis of Equation (3.1):

128

rate =

slope w 100 A 60

(3.1)

Where; w is the weight (g) of slag and A is the surface area (cm2 ) of the lime crucible in contact with the slag. The slope of the dissolution curves were obtained by fitting a straight line through the initial experimental concentration data using MATLAB. The lime crucible used in the experiments, geometrically consisted of a disk at the bottom of crucible and a cylinder at the wall, therefore the total area of the lime sample was taken as area of disk and area of cylinder in contact with the slag. The calculation of the area is detailed in Chapter 4. In calculation of the area of lime samples, the height of a sample was considered as the depth of lime crucible immersed in the melt which was about 15~20 mm. The radius of the lime sample changed with temperatures and slag chemistry during dissolution. To simplify the calculation, the area of the sample was assumed to be constant. This may cause an error, which is detailed in Appendix D. This is one of the reasons that the initial CaO concentration data only was included for the dissolution rate calculations, as the area of the lime crucible was steadier during the early stages of rotation of lime in the melt. The errors associated in calculation of rate of dissolution is about 15~20%, which is detailed in Appendix D.

The rate of lime dissolution is tabulated in Table 3.1, which shows an overall increase in the rate of dissolution by increasing the rotational speed.

129

Table 3.1: The rate of lime dissolution (gr.cm -2 .s-1 ) in the slag at 1430 C in air Slag Master slag Static rpm = 30 rpm = 60 rpm = 90 rpm = 120 rpm = 150

4.7810-6 3.1810-5 2.2310-5 5.0310-5 4.7810-5 6.3710-5

As mentioned before, the lime sample used in the experiments was in the form of a cylinder and a disk. The mass transfer from the disk side of the sample can be calculated by the following Equation (3.2):

j = 0 .61D 2 / 3 1 / 6 1/ 2 (Cs Cb)

(3.2)

Where, j (g/cm2 .s) is the mass flux, C s and Cb (g/cm3 ) are the saturation and bulk liquid slag concentration, D (cm2 /s) is the diffusion coefficient of lime in the slag, is the kinematic viscosity of the melt (cm2 /s), and (rad/s) is the angular velocity of the disk. The mass transfer coefficient k disk (cm/s) from the disk side of crucible can be written as:

k disk = 0.621D 2 / 3 1 / 2 1/ 6

(3.3)

Equation (3.2) describes the dissolution that is rate limited by mass transfer through a concentration boundary layer in the melt.

The mass transfer from the cylinder side of the sample can be calculated by the following Equation (3.4):

130

j cylinder =

k cylinder V

Sc 0.644 = 0.0791 Re

0 .3

(3.4)

Where j (g/cm2 .s) is the mass flux, k cylinder is the mass transfer coefficient of the cylinder, V is the peripheral velocity of rotating disk, Sc is Schmidt number and Re is the Reynolds number. The mass transfer coefficient k cylinder (cm/s) from the cylinder side of crucible can be written as:

k cylinder

D = 0.065

0. 41

(r ) 0. 75

(3.5)

Where r is the radius of the rotating cylinder (cm).

The total mass transfer from the rotating sample to the slag can be calculated according to Equation (3.6), which will be explained in detail in Section 4.1.3.

K total =

r 2h k disk + k cylinder r + 2h r + 2h

(3.6)

Thus, if mass transfer in liquid phase plays a significant role in the dissolution of lime in the slag phase, then the mass transfer coefficient and consequently the dissolution rate should be proportional n-th power of a stirring speed, where n is 0.5 for the disk and 0.75 for the cylinder as two boundary conditions. Figure 3.3 and Figure 3.4 show a linear dependence of the dissolution rate with the 0.5 and 0.75 th power of rotation speed in the speed range of 30 to 150 rpm. Based on Equation (3.6), and by calculation of kinematic viscosity of slag at 1430 C 131

(according to the models explained in Appendix B and C) and the area of disk and cylinder in contact with slag, the total mass transfer and consequently the dissolution rate is proportional to the rotation speed according Equation (3.7):

Ratetotal 0.095 0. 5 + 0.027 0.75

(3.7)

The result in Figure 3.6 shows a linear relationship between the dissolution rate and the correlation of rotation speed expressed in Equation (3.7). The apparent linear dependence of the initial rate on the ( A 0.5 + B 0.75 ) provides an evidence that liquid phase mass transfer played a significant role in controlling the dissolution of these samples over the rotation speed of 30 to 150 rpm. As it will be explained later in this chapter, in static experiments at 1430 C formation and growth of a solid layer on the surface of lime specimen were observed, while at higher temperature this phase was not stable and did not form. It is thus reasonable to consider that the measured dissolution rates at 1430 C were in a mixed controlled regime of liquid phase mass transfer and diffusion in the solid layer formed.

The variation of dissolution rate with 0.5 and 0.75 th power of rotation speed as two boundary conditions was also investigated at 1600 C at 30, 60 and 90 rpm (Figure 3.3 and Figure 3.4). The total dissolution rate is related to the rotation speed according to Equation (3.8), with calculation of kinetic viscosity of slag at 1600 C.

132

Ratetotal 0.115 0.5 + 0.043 0.75

(3.8)

A linear relationship holds between these two variables (Figure 3.5). The observed effect of temperature on destabilising the solid layer separating the lime and slag at 1600 C from the present work is in agreement with the CaO Al2 O3 SiO 2 phase diagram which does not show the formation of any reaction layer while lime is dissolving in the slag at 1600 C. Therefore, the dissolution rate data for various rotation speed at 1600 C in the present work suggesting that the dissolution is controlled by the mass transferred into the liquid phase and not in the mix-controlled regime due to the formation of a solid layer.

The measured data under static condition, i.e zero rotation speed should correspond to the rate of lime dissolution under any natural convection and agitation caused by sampling the melts. As the rotation speed of 90 rpm was in the middle range of the speeds at 1430 and 1600 C, where there is a linear relationship between the dissolution rate and 0.5 and 0.75 th power of rotation speed and their combination, 90 rpm was chosen as the constant speed for the rest of experiments, where the influence of temperature variation and additives in the slag were investigated

133

Figure 3.3: Variation of the dissolution rate of CaO versus the square root of rotation speed in air at 1430 and 1600 C

Figure 3.4: Variation of the dissolution rate of CaO versus the 0.75-th power of rotation speed in air at 1430 and 1600 C

134

Figure 3.5: Variation of the dissolution rate of CaO versus A 0.5 + B 0.75 of rotation speed in air. A and B are defined at 1430 and 1600 C

3.1.1.2

Variation of CaO dissolution at various temperatures in the master slag

Experiments were carried out to study the dissolution of lime in master slag at temperatures of 1430, 1500, 1550 and 1600 C in air and a rotating speed of 90 rpm. The results show an increase in dissolution rate of lime when the temperature was increased from 1430 to 1600 C as it is shown in Table 3.2. At 1600 C the dissolution rate was increased by a factor of 2.

At 1430 C, the variation of CaO concentrations with reaction time at rotating speed of 90 rpm is illustrated in Figure 3.6 and also presented in Table 3.3. The concentration data have the error of about 0.2 0.3 wt% (absolute) in XRF 135

analysis of the slag samples. The dissolved CaO concentration in slag increased more rapidly in the first 10 minutes and then dissolution of CaO slowed. Although the lime crucible was not fully dissolved in the slag after the completion of the experiment, CaO reached a level of 52 wt%, which is still far from the saturation level of about 59 wt%. This observation suggests the formation of a stable solid layer, which hindered further dissolution of lime under the given experimental conditions.

The formation of a product layer at the lime/slag interface was investigated and confirmed by performing static reaction of lime with slag at 1430 C and the results are presented in Section 3.2. The results for rate of dissolution of lime in the slag for all experiments at different times and various additives are tabulated in Table 3.2.

Table 3.2: The rate of dissolution of lime (g/cm 2 .s) in the slag at various temperatures and with additives
Temperature ( C ) master slag slag + CaF2 slag + Fe 2 O3 slag + TiO 2 slag + ilmenite Slag + Mn3 O4 slag + SiO 2 1430 5.0310-5 1.3410-4 7.9610-5 9.5510-5 1500 6.3710-5 2.3110-4 1.6310-4 1.2110-4 9.5510-5 1.4010-4 6.6110-5 1550 8.8310-5 2.7210-4 1.8510-4 1.3410-4 2.1010-4 1.6610-4 1.0110-4 1600 1.2710-4 3.3010-4 2.2310-4 1.7210-4 2.5510-4 2.1710-4 1.3410-4

- slag likely to contain solid phase

136

Figure 3.6: Concentration of CaO dissolved in base slag at 90 rpm and in air at 1430 C Table 3.3: The variation of CaO concentration in base slag by measurements from XRF analysis at 90 rpm and in air at 1430 C for 1 hour
Reaction time (min) 0 5 10 15 20 25 30 35 40 45 50 55 60 XRF analysis of bulk slag (wt%) SiO2 8.06 7.91 7.83 7.83 7.74 7.77 7.74 7.74 7.71 7.80 7.72 7.70 7.69 Al2 O3 42.0 41.9 41.7 41.6 41.3 41.3 41.0 41.1 40.8 41.2 40.8 40.6 40.7 Fe 2 O3 0.31 0.18 0.20 0.15 0.14 0.15 0.14 0.12 0.13 0.15 0.19 0.13 0.13 MgO 0.09 0.08 0.09 0.10 0.11 0.12 0.11 0.13 0.13 0.14 0.13 0.14 0.15 CaO 49.83 50.50 51.00 51.10 51.07 51.34 51.39 51.60 51.46 51.66 51.82 51.71 51.95 CaO from curve fitting 49.83 50.31 50.65 50.91 51.11 51.27 51.40 51.51 51.60 51.68 51.75 51.81 51.86

137

In experiment at 1500 C, the lime dissolution was increased with reaction time. Results in Figure 3.7 show that after one-hour reaction time, the dissolved lime in the slag had not reached the lime saturation limit of about 59 wt%. At 1500 C and after the completion of the experiment, a substantial reduction in diameter of the lime sample was observed, which was due to faster dissolution at higher experimental temperature. The reduction in surface area of the lime sample appeared to be the cause of slowing down of the of dissolution rate. The results are also tabulated in Table 3.4. Results at 1550 C and 1600 C are tabulated in Table 3.5 and Table 3.6 and illustrated in Figure 3.8 and Figure 3.9, respectively. The time intervals for sampling of slag during the rotating of lime in slag at these two temperatures were every two minutes for the first ten minutes of experiment, where the dissolution rate is faster and the rest of the sampling performed at every 10 minutes. At these two temperatures the curves reached plateau after the first ten minutes of reaction. After 60 minutes and at the end of the experiment, it was observed that the bottom of the crucible was totally dissolved in the slag at 1550 and 1600 C. It is likely that the disappearance of the bottom of CaO crucible happened after 10 minutes of reaction time, when the slope of lime dissolution curve just started to decrease, and then the dissolution continued from the remaining wall of lime crucible with much smaller surface contact with the molten slag. As the lime sample was attached to the alumina tubes with zirconia paste, the complete dissolution of lower part of lime sample exposed the zirconia paste to the melt and resulted in traces of zirconia in the chemical analysis of the slag samples after 10 minutes of reaction time. As the calculation of the mass transfer coefficient and diffusion coefficient was based on the complete geometry of the crucible, i.e. disk and cylinder, so the CaO concentration data up to the 138

reaction time of ten minutes were used in the calculations. It can bee seen that at 1500 C as the crucible was not fully dissolved in the slag, during the reaction time, the dissolution occurred from the whole surface of the sample. This lead to higher concentration of CaO in slag compare to the 1550 C close to the end of reaction time, but the rate of dissolution of lime at 1550 C was still higher than at 1500 C. The dissolution rate at 1600 C was found to be higher than the rate at 1550 C.

139

Figure 3.7: Concentration of CaO dissolved in base slag at 90 rpm and in air at 1500 C Table 3.4: The variation of CaO concentration in base slag by measurements from XRF analysis at 90 rpm and in air at 1500 C
Reaction time (min) 0 2 4 6 8 10 20 30 40 50 60 XRF analysis of bulk slag (wt%) SiO2 8.08 7.91 7.74 7.89 7.88 7.82 7.73 7.74 7.57 7.58 7.48 Al2O3 43.1 42.7 41.9 42.6 42.6 42.4 41.8 41.5 40.9 40.6 40.3 Fe 2O3 0.185 0.160 0.067 0.060 0.062 0.059 0.055 0.079 0.060 0.052 0.062 MgO 0.03 0.03 0.07 0.05 0.07 0.07 0.09 0.14 0.14 0.18 0.16 CaO 50.2 50.4 50.6 50.6 51.1 51.1 51.4 52.2 52.5 52.5 53.0 CaO from curve fitting 50.20 50.40 50.59 50.76 50.91 51.05 51.63 52.05 52.39 52.68 52.93

140

Figure 3.8: Concentration of CaO dissolved in base slag at 90 rpm and in air at 1550 C

Table 3.5: The variation of CaO concentration in base slag by measurements from XRF analysis at 90 rpm and in air at 1550 C
Reaction time (min) 0 2 4 6 8 10 20 30 40 50 60 XRF analysis of bulk slag (wt%) SiO2 8.1 7.93 7.80 7.82 7.77 7.77 7.68 7.65 7.5 7.6 7.54 Al2O3 43.1 42.7 42.0 42.1 41.8 41.6 41.3 41.1 40.3 40.6 40.5 Fe 2O3 0.22 0.143 0.089 0.077 0.077 0.086 0.066 0.063 0.08 0.06 0.066 MgO < DL < DL 0.06 0.08 0.09 0.12 0.11 0.13 0.1 0.1 0.14 CaO 50.2 50.6 50.7 51.0 51.4 51.6 51.8 52.3 53.4 52.6 52.5 CaO from curve fitting 50.19 50.55 50.84 51.08 51.28 51.44 51.98 52.26 52.42 52.51 52.56

141

Figure 3.9: Concentration of CaO dissolved in base slag at 90 rpm and in air at 1600 C Table 3.6: The variation of CaO concentration in base slag by measurements from XRF analysis at 90 rpm and in air at 1600 C
Reaction time (min) 0 2 4 6 8 10 20 30 40 50 60 XRF analysis of bulk slag (wt%) SiO2 8.1 8.1 7.7 7.7 7.8 7.9 7.6 7.4 7.3 7.2 7.2 Al 2O3 42.4 42.3 41.3 41.1 41.1 41.1 40.6 40.1 39.4 39.4 38.9 MgO < DL < DL 0.1 < DL 0.1 0.0 0.1 < DL 0.2 0.2 0.2 CaO 49.4 50.6 51.7 51.8 51.8 51.7 52.2 52.6 52.7 52.8 52.2 CaO from curve fitting 49.37 50.86 51.37 51.64 51.82 51.94 52.29 52.50 52.67 52.83 52.97

142

Figure 3.10: Comparison of CaO Concentrations dissolved in slag at 90 rpm and in air at 1430 1600 C

Figure 3.10 shows the curve- fittde data on dissolution of CaO in calcium aluminosilicate slag at various temperatures. It was found that by increasing the temperature of the experiments, the amount of dissolution lime increased.

3.1.1.3

Effect of additives on the dissolution of CaO in slag

In order to investigate the effect of additives on the lime dissolution rate, 5 wt% of CaF2 , Fe2 O3 , TiO 2 , Mn2 O3 , ilmenite and SiO 2 were added to the master slag. The experiments were conducted at temperatures ranging from 1430 to 1600 C and at constant rotating speed of 90 rpm. Fe2 O3 , TiO 2 , Mn2 O3 and ilmenite were expected to increase the diffusivity and ionic conductivity, while SiO 2 was expected to ha ve the opposite effect. The effect of additives on the concentration of lime in the slag at each temperature is presented in Appendix A, while for 143

each additive the results at various temperatures are compared in the following sections.

3.1.1.3.1

Effect of CaF2 addition on dissolution of CaO in slag at various temperatures

The influence of CaF2 on the dissolution of CaO in calcium aluminosilicate slag was investigated by addition of 5 wt% CaF2 to the slag and carrying out dissolution experiments at 1430, 1500, 1550 and 1600 C. The experiments were performed in air and a rotation speed of 90 rpm and for a reaction time of 20 minutes. The dissolution of lime in the slag was so fast specially at temperatures higher than 1500 C, that the bottom and lower half of crucible were dissolved after 10~20 minutes of running experiments, which is shown in Figure 3.11, on this basis, the reaction time was limited by 20 minutes. The results for experiments with 5 wt% CaF2 content of slag and at 1430 C are presented in Appendix A.2. The dissolution of lime at 1430 C increased dramatically compared to higher temperatures. By examining the lime specimen after completion of experiments at 1430 C (Figure 3.11), deep grooves were found on the surface of the crucible, which indicates that solid particles caused physical erosion of lime from the specimen rather than dissolution of lime into the liquid slag. The phase diagram for CaO-Al2 O3-CaF2 system (Figure 3.12) shows the possibility of formation of a compound 11CaO.7Al2 O3 .CaF2 during the reaction of lime with slag at 1430 C. As the melting point of this compound is 1577 C, it was postulated that the solid phase formed was floating in melt causing the erosion of lime crucible and chipped off lime pieces from the surface of lime. This resulted an excessive CaO content of slag from XRF analysis of sampled

144

slags from the bath. By conducting experiments at 1500, 1550 and 1600 C, the dissolution rate was slower than at 1430 C. By increasing the temperature, the 11CaO.7Al2 O3 .CaF2 solid phase becomes unstable and dissolves in the slag, decreasing the possibility of erosion of lime specimen by solid phase, which is also confirmed by static experiments in section (3.2.1.2). The results are shown in Appendix A.2. The lime dissolution curves at 1500 and 1550 C have a similar pattern, but at 1600 C, the dissolution of lime is higher than at 1500/1550 C. The dissolution rate of lime increased considerably with addition of 5 wt% CaF2 as it was shown in Table 3.2. The curve fitted data on variation of CaO dissolution for the temperature range of 1430, 1500, 1550 and 1600 C for the slag, which contains 5 wt% CaF2 at speed of 90 rpm is shown in Figure 3.13.

1430 C

1500 C

1550 C

1600 C

Figure 3.11: The lime specimen after dissolution in the slag with 5 wt% CaF2 at 90rpm and after reaction time of 20 minutes

145

Figure 3.12: CaO- Al2 O3 -CaF2 phase diagram according to Mills and Keene (1981)

146

Figure 3.13: Comparison of total CaO content in slag with 5 wt% CaF2 at 90 rpm and in air at 1430 1600 C

3.1.1.3.2

Effect of Fe2 O3 addition on dissolution of CaO in slag at various temperatures

Experiments on the effect of addition of 5 wt% Fe2 O3 on dissolution of CaO were carried out in air with a rotation speed of 90 rpm at temperatures 1430, 1500, 1550 and 1600 C for a reaction time of 1 hour. The results are illustrated in Appendix A.3. For all experiments with addition of 5 wt% Fe2 O3 in the slag, the bottom of lime crucible was dissolved after the 1 hour reaction and in experiment at 1600 C, the whole crucib le was dissolved in the slag. The lime specimen after the experiments at 1430 and 1500 C are shown in Figure 3.14. Therefore, the concentration data used in the calculations taken from the data up to the reaction time of 8~10 minutes, when the slope of lime dissolution curve

147

has started to change considerably, reflecting a change in the shape of the lime specimen.

1500 C

1430 C

Figure 3.14: The lime specimen after dissolution in the slag with 5 wt% Fe2 O3 at 90 rpm and after reaction time of 20 minutes

The effect of addition of Fe2 O3 on the dissolution of lime in the slag at various temperatures is illustrated in Figure 3.15 (curve-fitted data). Increasing the temperature increased the rate of lime dissolution in the slag.

148

Figure 3.15: Comparison of CaO concentrations dissolved in slag with 5 wt% Fe2 O3 at 90 rpm and in air at 1430 1600 C

3.1.1.3.3

Effect of TiO2 addition on dissolution of CaO in slag at various temperatures

The effect of 5 wt% TiO 2 content in slag on the dissolution rate of CaO in calcium aluminosilicate slag was studied in air with speed of 90 rpm at temperatures of 1430,1500,1550 and 1570 C. The results are presented in Appendix A.4. After reaction time of 1 hour at 1430 C, there was a substantial reduction in diameter of crucible (Figure 3.16). In experiments at 1500, 1550 and 1600 C, the bottom of the crucible was completely lost to the slag, which caused change of dissolution rate of lime in the slag. Therefore the concentration data up to 10 minutes of reaction time were used to calculate the rate of dissolution. The curve-fitted concentration of CaO in the slag, which contains 5 wt% TiO 2 is presented in Figure 3.17. 149

1500 C

1600 C

Figure 3.16: The lime specimen after dissolution in the slag with 5 wt% TiO2 at 90 rpm and after 60 minutes of reaction.

Figure 3.17: Comparison of concentrations of CaO dissolved in slag with 5 wt% TiO2 at 90 rpm and in air at 1430 1570 C for 1 hour

150

3.1.1.3.4

Effect of ilmenite addition on dissolution of CaO in slag at various temperatures

The influence of addition of 5 wt% ilmenite on the dissolution of CaO in calcium aluminosilicate slag was investigated at 1500, 1550, 1570 and 1600 C. The experiments were performed in air and a rotation speed of 90 rpm and for a reaction time of 10 minutes. The reaction time was chosen on the basis of considering the fact that during the dissolution experiments with addition of TiO 2 and Fe2 O3 , the shape of the cylindrical lime sample was changed after 10~20 minutes of rotation of lime in the melt. According to Figure 3.18, it was observed that at 1550 and 1570 C, there was a substantial reduction in diameter of the lime sample after the total reaction time of 10 minutes and at 1600 C, the bottom of the lime crucible was lost after completion of the experiment.

The results for addition of 5 wt% ilmenite to the slag at 1500-1600 C are presented in Appendix A.5. The dissolution of lime in the slag containing 5 wt% ilmenite at 1500 C was increased linearly with time and generally was lower probably due to formation of a solid layer. At 1550 and 1570 C, the dissolution of lime was faster with a similar dissolution pattern, however at 1600 C, there was a pronounced effect on the dissolution of lime compared to other temperatures. The variation of CaO dissolution for the temperature range of 1500 to 1600 C for the slag, which contains 5 wt% ilmenite at rotational speed of 90 rpm is shown in Figure 3.19 (curve- fitted data).

151

1500C

b 1550C

1570C

1600C

Figure 3.18: The lime specimen after dissolution in the slag with 5 wt% ilmenite at 90 rpm and after reaction time of 10 minutes

Figure 3.19: Comparison of concentrations of CaO dissolved in slag with 5 wt% ilmenite at 90 rpm and in air at 1500 1600 C

152

3.1.1.3.5

Effect of Mn 3 O4 addition on dissolution of CaO in slag at various temperatures

Experiments were carried out with addition of 5 wt% Mn3 O4 to the master slag in air, at rotating speed of 90 rpm at temperatures 1430, 1500, 1550, 1600 C for a reaction of time of 10 minutes. The results are presented in Appendix A.6. The reaction time was selected as explained in previous sections. At temperatures above 1430 C, the bottom of crucible was again dissolved in slag after completion of the experiments (Figure 3.20). At 1600 C, there were substantial grooves on the surface of the lime crucible; therefore the dissolution rate was measured from the data up to 6 minutes where the rate of dissolution could be considered linear and not affected by the complete dissolution of crucible bottom. The curve- fitted concentration of lime in the slag with addition of Mn2 O3 for various temperatures is illustrated in Figure 3.21.

153

1430 C

1500 C

1550 C

1600 C

Figure 3.20: The lime specimen after dissolution in the slag with 5 wt% Mn2 O3 at 90 rpm and after reaction time of 20 minutes

Figure 3.21: Concentration of CaO dissolved in slag with 5 wt% Mn3 O4 at 90 rpm and in air at 1430 1600 C

154

3.1.1.3.6

Effect of SiO2 addition on dissolution of CaO in slag at various temperatures

After consulting CaO-Al2 O3-SiO 2 phase diagram in Slag Atlas (Eisenhuttenleute (1995)), the experimental temperatures were chosen in a way in order to avoid formation of solid phases, which would affect the dissolution behaviour of lime in the slag. The effect of SiO 2 addition on dissolution of lime was investigated by adding 5 wt% SiO 2 to the slag and carrying out dissolution experiments in air at a rotating speed of 90 rpm at temperatures 1500 to 1600 C. The results are shown in Appendix A.7. The variation of lime dissolution with changing temperature is given in Figure 3.22 (curve-fitted data).

Figure 3.22: Comparison of concentrations of CaO dissolved in slag with additional 5 wt% SiO2 at 90 rpm in air at 1500 1600 C for 1 hour

3.1.1.4

Effect of variables on the dissolution rate

The effects of temperature and additives (slag chemistry) on the dissolution rate of lime is summarized in Table 3.2 and illustrated in Figure 3.23. The dissolution 155

rate was calculated according to section 3.1.1.1. The results sho w that addition of 5 wt% CaF2 increased the dissolution rate compared to the master slag by a factor 4 at 1500 C, a factor of 3 at 1550 C and a factor of 2 at 1600 C. The addition of 5 wt% Fe2 O3 to master slag at 1430, 1500, 1550 and 1600 C increased the dissolution rate of lime about a factor of 3.3, 3.6, 2.6 and 1.9, respectively. With addition of 5 wt% TiO 2 , the dissolution rate increased by a factor of 1.5, 2, 1.5 in comparison with the values for master slag at temperatures of 1430, 1500, 1550 C, respectively and at 1570 C the rate was about 1.3 times the dissolution rate for master slag at 1600 C. When 5 wt% ilmenite was added to the master slag, the dissolution rates at 1500, 1550 and 1600 C increased by 1.6, 2.3 and 2 times compared to the rate data for master slag. The dissolution rate of slag containing ilmenite at 1570 C was 1.25 times higher than the rate for slag containing 5 wt% TiO 2 at the same temperature. The addition of SiO 2 increased slightly the dissolution rate compared to the master slag at various temperatures.

Increasing the temperature increased the lime dissolution rate in almost all slags studied. The exception being when 5 wt% CaF2 was used, where an apparent decrease in dissolution was observed at 1430 and 1500 C. This is likely to be due to inhomogenity of slag at 1430 C, where the phase diagram for CaO Al2 O3 CaF2 indicates stability of solid phase at temperatures below 1550 C.

Given the fact that there is an error of about 15% in calculation of dissolution rate, the general trend of effect of additives on the dissolution rate at various

156

temperatures shows that Ilmenite, MnO x and FeO x increased the dissolution rate and have comparable effect in comparison with CaF2 .

Figure 3.23: The rate of dissolution of lime (g/cm 2 .s) in the slag at various temperatures and with additives

3.1.1.5

Effect of basicity on the dissolution of lime at constant temperature

Experiments were carried out to study the effect of basicity (at a constant temperature of 1500 C) on the dissolution rate of lime. Experiments were performed in slag of CaO 20.4 wt% Al2 O3 41.4 % SiO 2 (basicity = 0.9) at various rotating speeds of 40, 60 and 90 to investigate if the dissolution processes of lime into liquid are controlled by mass transfer step. The results for data) variation of concentration of lime with time are shown in Figure 3.24 (curvefitted data). As it would be shown later (3.2.1.9) on the static reaction of this slag

157

with lime, two reaction layers were formed on the lime/slag interface. At 40 rpm, these layers seem to be unstable, floating in the slag, eroding the lime sample which results in excessive dissolution of lime in the slag.

Figure 3.24: The concentration of lime (wt%) in the slag (basicity of 0.9) at various rotation speed at 1500 C

The proportionality of dissolution rate with 0.5 and 0.75 th power of rotation speed are shown in Figure 3.25 and Figure 3.26, where a linear relationship exists. As rotation speed of 90 is within the regime, where the dissolution is controlled by diffusion in liquid slag so the dissolution rate at this speed can be compared to the rest of dissolution rate data. The dissolution rate in slag of 0.9 basicity was found to be ( 2.36 10 5 g CaO/cm2 /s) which is a third of the master slag with basicity of 6 ( 6.37 10 5 g CaO/cm2 /s).

158

Figure 3.25: The dissolution rate of CaO with speed 0.5 in slag with basicity of 0.9

Figure 3.26: The dissolution rate of CaO with speed 07.5 in slag with basicity of 0.9

159

3.1.2

Dissolution of MgO in calcium aluminate slag

Preliminary experiments were carried out to investigate the dissolution of MgO in the CaO Al2 O3 slag in air at 1430 C by rotating MgO samples at speeds in the range of 30 to 150 rpm and reaction time of 1 hour. The initial slag composition was 45 wt % CaO and 55 wt % Al2 O3 as shown in Figure 3.27. The purpose of these experiments was to test the experimental set up and evaluation of the results before carrying out the experiments, which investigated lime dissolution.

Figure 3.27: CaO-Al2 O3 -MgO system phase diagram according to Slag Atlas (Eisenhuttenleute (1995))

The MgO samples were in the form of crucibles, 20 mm in diameter and 30 mm in height. A slag sample of 60 grams was used in each experiment. MgO 160

dissolved in the slag was determined by sampling of molten slag using a platinum wire at the tip of a steel rod. The sampling was done at five minutes intervals. The MgO concentration in the slag was analysed using XRF, the estimated error being within 0.2 0.3 wt% (absolute).

3.1.2.1

Effect of rotation speed on the rate of dissolution

Variation of the dissolution rate of MgO in molten calcium aluminate slag with the rotation speed (60, 90, 120 and 150 rpm) was investigated at 1430 C in air for a reaction time of one hour. The concentration of MgO in the slag increased with increasing rotation speed of the MgO sample in the melt. The experimental results are shown in Figure 3.28.

161

Figure 3.28: Concentration of MgO dissolved in slag at different rotation speed, in air at 1430 C for 1 hour

The proportionality of dissolution rate with 0.5 and 0.75 -th power of rotation speed as two boundary conditions for dissolution of disk and cylinder side of the magnesia specimen in the speed range of 60 to 150 rpm, is presented in Figure 3.29 and Figure 3.30. The total dissolution rate is related to the rotation speed according to Equation (3.9), on the basis of combination effect of cylinder and disk and kinematic viscosity of slag (as explained in Section 3.1.1.1).

Ratetotal 0.093 0.5 + 0.026 0.75

(3.9)

As it can be seen, the dissolution rate changes linearly with 0.5 and 0.75 -th power of rotation speed in the speed range of 60 to 120 rpm. Also Figure 3.31

162

shows that the dissolution rate changes linearly with the correlation of rotation speed expressed in Equation (3.9). The result at the speed of 150 rpm show deviation from linear regression. The apparent sudden increase in dissolution rate may be caused by erosion of polycrystalline manganese sample at such high rotating speed. The measured data under static condition, i.e. zero rotation speed should correspond to the rate of lime dissolution under condition of natural convection as well as agitation caused by sampling the melts. The dependency of rate of magnesia dissolution on the rotation speed of 60 120 rpm, suggests that the measured dissolution rates were most likely controlled by mass transfer in the liquid phase. Rotation speed of 90 rpm was chosen for subsequent experiments to determine the effects of addition of Fe2 O3 and (CaF2 + Fe2 O3 ) to the slag. As it is shown in appendix D, the error in the rate of dissolution is about 15 20%.

Figure 3.29: Dependence of rate of dissolution of MgO with 0.5 -th power of speed

163

Figure 3.30: Rate of dissolution of MgO with 0.75 -th power of speed

Figure 3.31: Variation of the dissolution rate of MgO versus A 0.5 + B 0.75 of rotation speed in air. A and B are defined at 1430 C

164

3.1.2.2

Effect of Fe2O3 addition on dissolution of MgO in slag

The influence of Fe2 O3 on the dissolution of MgO was studied by adding 5 wt% and 10 wt% Fe2 O3 to the calcium aluminate slag. The experiments were carried out in air at a rotation speed of 90 rpm at 1430 C and the total reaction time of 1 hour. The results for these experiments are presented in Appendix A.7. The comparison of magnesia concentration in the alumina silicate slag with addition of 5 and 10% Fe2 O3 at 1430 C is illustrated in Figure 3.32. The addition of 5 wt% Fe2 O3 increased the dissolution rate by a factor of 3 and addition of 10 wt% Fe2 O3 increased the rate of dissolution by a factor of 5. The dissolution rate data are tabulated in Table 3.7.

Figure 3.32: Concentration of MgO dissolved in slag with 5 and 10% Fe2 O3 at 90 rpm in air at 1430 C for 1 hour

165

Table 3.7: The rate of dissolution of MgO in the slag at 1430 C and with various additives
Rate (g/cm2 /s) 2.7E-05 6.5E-05 1.0E-04 2.9E-04 2.1E-04

Slag samples master slag slag + Fe 2 O3 5% slag + Fe 2 O3 10% slag + Fe 2 O3 5% + CaF2 5 % slag + Fe 2 O3 10% + CaF2 5%

3.1.2.2.1

Effect of (Fe2 O3 + CaF2 ) addition on dissolution of MgO in slag

An attempt was made to study the influence of CaF2 on the dissolution rate of MgO in the slag but after addition of 5 wt% CaF2 , the slag was not fully molten at experimental temperature of 1430 C. Therefore, 5 wt% CaF2 was added to two types of slag; slag with 5 wt% Fe2 O3 and slag with 10 wt% Fe2 O3 . The experiments were conducted in air at rotational speed of 90 rpm. The results are presented in Appendix A.8. The dissolution rate was very fast for the first five minutes of rotation and then it became flat as it approaches the saturation limit of magnesia in the slag after 10~20 minutes of rotation. Therefore, the calculations were based on the results for the first 20 minutes of the experiment. The rate of dissolution was increased with a factor of 10.7 for the slag which contained 5 wt% CaF2 + 5 wt% Fe2 O3 and also the rate was increased with a factor of 7.7 for the slag which contained 5 wt% CaF2 & 10 wt% Fe2 O3 . The results on the rate of 166

dissolution with these two types of additives are tabulated in Table 3.7. The variation of MgO concentration in the slag with additives at 1430 C is presented in Figure 3.33.

Figure 3.33: Concentration of MgO dissolved in slag with additives at 90 rpm and in air at 1430 C for 1 hour

3.2
3.2.1

Static experiments
CaO experiments

In static experiments, lime was reacted with various slags at different temperatures. The results on solubility of lime in the slags and formation of a solid layer at lime/slag interface will be presented in the following sections. The concentration of various elements in the slag phase was analysed quantitatively

167

with EPMA. Phase identifications and measurement of the thickness of a reaction layer were done using SEM.

3.2.1.1

Solubility of lime in the master slag under various temperatures

The driving force for dissolution of lime can be defined as the difference between the solute and the saturation concentration of lime in the molten slag. Experiments were performed by placing a piece of dense lime with slag in a platinum capsule, heating them up to a specific temperature in a muffle furnace and holding for the required reaction period. The lime sample and slag were then withdrawn from the furnace and quenched in air. The quenched samples were sectioned vertically, mounted and polished.

Initially experiments were carried out in air at 1430 C with reaction times of 0.5, 1, 2, 4, 6, 12 and 24 hours to determine the time required for the dissolution of CaO in calcium aluminosilicate slag to reach equilibrium. The concentrations of CaO in the quenched bulk slags close to the lime/slag interface were analysed using SEM EDS. Samples from the rest of experiments with additives were analysed with EPMA for higher accuracy. The results of the analysis are presented in Figure 3.34 and tabulated in Table 3.8.

168

Figure 3.34: Variation of bulk slag composition (wt%) measured by SEMEDS with the reaction time at 1430 C in air.

Table 3.8: SEM EDS analysis of the bulk slag at1430 C in air

Slag composition (wt %) Time (hr) Ca 0.5 1 2 4 6 12 24 36.63 36.73 35.34 36.66 35.43 36.24 35.91 Al 22.07 21.66 22.51 21.85 22.06 20.89 21.25 Si 4.02 4.1 3.25 4.1 4.07 4.05 3.89 O 37.7 37.6 39.05 37.66 38.33 38.82 38.67

CaO (wt%) 51.28 51.42 49.48 51.32 49.60 50.74 50.27

These results indicate that, within experimental scatter, a steady value of concentration had been reached after half an hour. The reaction time of 3 hours 169

was sufficient to reach equilibrium, which is enough for the experiments at higher temperatures and with addition of additives, where the dissolution rate is faster compared to the master slag.

In these experiments where the lime sample was reacted with master slag at 1430 C, a reaction layer was formed on the lime/slag interface, which is shown in Figure 3.35. Growth rate of this layer will be measured and used for development of a dissolution model detailed in the next chapter.

Reaction layer

CaO

slag

Figure 3.35: Interfacial region of CaO in contact with slag at 1430 C for the reaction time of 2 hours

Experiments were also carried out at, 1500, 1550 and 1600 C in air to determine the dependency of lime solubility on temperature. The obtained results are tabulated in Table 3.9.

170

Table 3.9: EPMA analysis of the bulk slag close to the lime/master slag interface in air at different temperatures Temperature (C) 1430 1500 1550 1600 Composition of slag (wt%) CaO (wt%) O 35.79 35.82 35.59 35.14 Mg 0.12 0.82 0.81 0.87 Al 16.68 17.66 17.45 16.06 Si 3.16 3.28 3.21 3.16 Ca 43.08 41.59 41.33 42.97 60.66 58.68 58.54 61.07

The measured solubility of lime at 1600 C could be compared with the results from phase diagram (Figure 3.1), which is about 62 wt%. Considering the fact that the error involved in the EPMA analysis for the Ca concentration was within

1 ~ 2% , it shows a very good agreement.

3.2.1.2

Effect of addition of CaF 2 on the solubility of lime in the slag

Experiments were performed to determine the solubility of lime in the slag with addition of 5 wt% CaF2 at temperatures 1430, 1550 and 1600 C. The results are presented in Table 3.10.

Table 3.10: EPMA analysis of the bulk slag close to the interface of lime/ slag containing 5 wt% CaF2 at various temperatures in air Temperature ( C ) 1430 1550 1600 Composition of slag (wt%) O 35.29 35.46 34.11 F 0.98 1.40 1.59 Mg 0.56 1.32 1.06 Al 16.41 15.33 15.65 Si 3.39 3.30 3.21 Ca 41.51 41.69 42.68 CaO (wt%) 57.63 58.47 58.34

171

Although the phase diagram in Figure 3.12, predicts the formation of a solid phase (11 CaO.7Al2 O3 .CaF2 ), no phase was evident in the EDS-SEM analysis of the sample (Figure 3.36). This might be due to the similarity between the composition of glassy phase at 1430 C and the expected phase. Therefore the BSE shows a similar contrast because of the similar density of atoms in liquid slag and the expected phase. At higher temperatures no solid phase was also detected, which is in accord with the expected behaviour from the phase diagram for CaO Al2 O3 CaF2 system.

CaO

slag

Figure 3.36: Interfacial region of CaO in contact with slag containing 5 wt% CaF2 at 1430 C for the reaction time of 3 hours

3.2.1.3

Effect of addition of Fe2O3 on the solubility of lime in the slag

The solubility of CaO in slag with addition of 5 wt% Fe2 O3 was determined in air and at temperatures of 1430, 1500, 1550 and 1600 C. The results are tabulated in Table 3.11. 172

Table 3.11: EPMA analysis of the bulk slag close to the interface of lime/ slag containing 5 wt% Fe2 O3 at different temperatures in air Temperature (C) 1430 1500 1550 1600 Composition of slag (wt%) O 35.17 35.06 34.46 35.44 Fe 2.87 2.41 2.43 2.60 Mg 0.60 1.27 0.82 0.94 Al 16.48 15.87 15.78 15.31 Si 2.90 3.24 2.81 2.89 Ca 39.81 40.77 41.49 41.50 CaO (wt%) 56.44 57.99 58.70 59.37

A solid phase was formed in the slag matrix when the experiment was conducted at 1430 C. The oxide composition of the phase is given in Figure 3.37.

CaO

Phase: CaO: 61.35 Al2 O3 : 35.19 SiO 2 : 1.96 FeO x : 1.125

Figure 3.37: Interfacial region of CaO in contact with slag containing 5 wt% Fe2 O3 at 1430 C for the reaction time of 3 hours

173

3.2.1.4

Effect of addition of TiO2 on the Solubility of lime in the slag

Experiments were carried out to obtain the solubilities of lime in the slag containing 5 wt%, TiO 2 at temperatures 1430, 1500, and 1600 C in air. The results are tabulated in Table 3.12.

Table 3.12: EPMA analysis of the bulk slag close to the interface of lime/ slag containing 5 wt% TiO2 at different temperatures in air Temperature ( C ) 1430 1500 1600 Composition of slag (wt%) O 35.74 36.67 36.04 Ti 2.44 2.12 2.09 Mg 0.62 1.07 1.04 Al 15.96 15.15 14.80 Si 2.97 3.03 2.95 Ca 38.66 39.81 40.40 CaO (wt%) 57.13 59.04 59.95

As it can be seen in Figure 3.38 a solid phase was formed on the CaO/slag interface at 1430 C. The composition of the reaction layer is also given in Figure 3.38.

174

CaO Phase: CaO: 62.66 Al2 O3 : 35.15 SiO 2 : 2.16

Figure 3.38: Interfacial region of CaO in contact with slag containing 5 wt% TiO2 at 1430 C for the reaction time of 3 hours

3.2.1.5

Effect of addition of ilmenite on the solubility of lime in the slag

Experiments on solubilities of lime in the slag with addition of 5 wt% ilmenite were carried out by placing a piece of dense CaO in the molten slag for a period of 3 hours in air. These experiments were performed at temperatures of 1430, 1500, 1550, 1600 C. The results are presented in Table 3.13. A reaction layer was observed during the experiment at 1430 C, which is shown in Figure 3.39.

Table 3.13: EPMA analysis of the bulk slag close to the interface of lime/ slag containing 5 wt% ilmenite at different times in air Temperature (C) 1430 1500 1550 1600 O Composition of slag (wt%) Fe Ti Mg Al Si 0.34 1.26 1.16 1.23 Ca CaO (wt%) 60.27 58.52 59.19 60.30

35.59 0.66 36.69 1.19 36.29 1.11 35.41 1.12

0.27 17.42 2.00 42.72 1.00 15.34 3.05 40.39 1.83 14.46 3.41 40.42 0.89 14.78 2.85 42.09 175

CaO

Phase: CaO: 60.26 Al2 O3 : 33.78 SiO 2 : 4.38

Figure 3.39: Interfacial region of CaO in contact with slag containing 5 wt% ilmenite at 1430 C for the reaction time of 3 hours

3.2.1.6

Effect of addition of Mn3O4 on the solubility of lime in the slag

Experiments aimed at investigating the solubility of lime in the slag containing 5 wt% MnO x were performed at temperatures of 1430, 1550 and 1600 C in air. The results are presented in Table 3.14. At 1430 C, a solid phase in the form of a layer was formed in the slag, which is shown in Figure 3.40.

Table 3.14: EPMA analysis of the bulk slag close to the interface of lime/ slag containing 5 wt% MnOx at various temperatures in air Temperature (C) 1430 1550 1600 Composition of slag (wt%) O 36.04 34.80 36.42 Mn 2.34 2.20 2.31 Mg 0.60 1.11 1.36 Al 15.83 16.44 15.87 176 Si 2.94 3.51 3.12 Ca 41.00 40.07 39.51 CaO (wt%) 59.56 57.53 57.66

CaO

Phase: CaO: 62.23 Al2 O3 : 34.15 SiO 2 : 2.16

Figure 3.40: Interfacial region of CaO in contact with slag containing 5 wt% Mn3 O4 at 1430 C for the reaction time of 3 hours

3.2.1.7

Effect of addition of SiO2 on the solubility of lime in the slag

The experimental data on reaction of lime with base slag with addition of 5 wt% SiO 2 at 1500, 1550 and 1600 C are illustrated in Table 3.15.

As it is mentioned before, the rotating experiments were carried out at temperatures at /above 1500 C as it was expected that at 1430 C a reaction layer would be formed on the lime/slag interface. Indeed, in static experiment at 1430 C, a reaction layer was observed at the lime/slag interface (Figure 3.41), which agrees with the CaO Al2 O3 SiO 2 phase diagram.

177

Table 3.15: EPMA analysis of the bulk slag close to the interface of lime/ slag containing additional 5% SiO2 at various temperatures in air Temperature (C) 1430 1500 1550 1600 Composition of slag (wt%) O 37.71 34.82 35.80 34.88 Mg 0.38 0.46 1.12 0.99 Al 22.20 15.46 14.66 14.35 Si 3.50 4.04 4.53 4.48 Ca 34.82 43.45 42.12 43.50 CaO (wt%) 49.09 60.87 60.43 61.62

CaO

CaO: 65.43 SiO 2 : 33.43 Al2 O3 : 1.13

Figure 3.41: Interfacial region of CaO in contact with slag containing 5 wt% SiO2 at 1430 C for the reaction time of 3 hours It is clear from this table that there is little difference in solubility of CaO in the slags with additives compare to the master slag.

3.2.1.8

FactSage thermodynamic modelling

Attempt was made to do thermodynamic modelling with FactSage developed by Bale et al. (2003) to calculate the solubility of lime in base slag and slag with

178

addition of 5 wt% CaF2 , Fe2 O3 , TiO 2 , Mn3 O4 , ilmenite and SiO 2 . The Equilibrium module of the package was used which identifies phases and their compositions by the Gibbs free energy minimization.

Table 3.16: The solubility of lime in various slags at different temperatures by FactSage (Bale et al. (2003)) modelling Temperature Master (C) slag 1430 1500 1550 1600 57.84 59.29 59.54 59.84 Solubility of lime in slag with addition of 5 wt% additives Fe2 O3 TiO 2 ilmenite Mn3 O4 SiO 2 CaF2 57.69 58.01 58.30 58.64 57.43 58.34 58.65 59.01 57.86 59.29 59.55 59.87 57.84 59.23 59.53 59.78 55.368 56.395 56.717 57.087 59.35 60.134 60.35 60.62

There is a very good agreement between the results from static experiments analysed by EPMA and solubility data from FactSage modelling. The data from thermodynamic modelling confirm that with the exception of Mn3 O4 , the additives have little effect on the equilibrium solubility of lime in the slags.

3.2.1.9

Formation of a reaction layer on the lime/base slag interface

Experiments were conducted to measure the growth rate of a solid layer formed during the reaction of the calcium aluminosilicate slag with lime at 1430 C in air. The lime specimen used in the experiment was taken from the same lime crucibles used in the rotating experiment and it was reacted with slag in a platinum capsule made from thin platinum foil. The reaction was followed by quenching the platinum foil with its content at various time intervals and

179

examining them by the Scanning Electron Microscope. It was found that a solid phase product layer was formed between the lime samples and the slag, which was identified by EDS analysis as very close 3CaO.Al2 O3 . This observation is consistent with the phase diagram.

The thickness of the reaction layer formed at the lime/slag interface was measured from the backscatter image generated by SEM. It was seen that the product layer thickness increased with time. The micrographs of the reaction layer and its growth is illustrated in Figure 3.42 to Figure 3.48.

Figure 3.42: SEM micrograph of the CaO and slag interface for lime reacting 30 minutes with slag in air at 1430 C

180

Figure 3.43: SEM micrograph of the CaO and slag interface for lime reacting 1 hour with slag in air at 1430 C

Figure 3.44: SEM micrograph of the CaO and slag interface for lime reacting 2 hours with slag in air at 1430 C

181

Figure 3.45: SEM micrograph of the CaO and slag interface for lime reacting 4 hours with slag in air at 1430 C

Figure 3.46: SEM micrograph of the CaO and slag interface for lime reacting 6 hours with slag in air at 1430 C

182

Figure 3.47: SEM micrograph of the CaO and slag interface for lime reacting 12 hours with slag in air at 1430 C

Figure 3.48: SEM micrograph of the CaO and slag interface for lime reacting 24 hours with slag in air at 1430 C

183

A plot of square layer thickness as a function of time in Figure 3.49 shows a linear relationship and the equation corresponding to this line was calculated as:

x 2 = 2 10 9 t

(3.10)

Where

x = thickness of the solid layer (mm)

t =

time of reaction (s)

According to Zhang et al. (1994),the growth rate of a solid phase may be expressed as:

dx K f K d = x (t )1 / 2 dt

(3.11)

Where Kf (mm2 /s) is the parabolic rate constant for the formation of solid phase, and Kd (mm/s1/2 ) is the rate constant for the dissolution of solid phase by molecular diffusion. Integration of Equation (3.11) with the initial condition of

x = 0 at t = 0 yields:

( x) 2 = K 2t

(3.12)

Where K = { K d + ( K d ) 2 + 2 K f

1 / 2 1/ 2

, which can be considered as an

effective parabolic rate constant for solid-solution growth. From Equation (3.10) 184

and (3.12), one can obtain experimental condition.

K = 4 10 5 mm/s1/2 under the mentioned

Figure 3.49: Thickness of solid layer as a function of square root of time in slag in air at 1430C

185

3.2.1.9.1

Effect of basicity on the formation of reaction layer on the lime/slag interface

Experiments were performed to investigate the effect of basicity on the formation of the reaction layer and dissolution of lime in the slag. Lime pieces were reacted with sla g of CaO 20.4 wt% Al2 O3 41.4 SiO 2 (basicity = 0.9) for predetermined reaction times. Two incoherent phases of 2CaO. SiO 2 and 3CaO.SiO 2 were formed on the lime/slag interface, where 3CaO.SiO 2 was observed to be between the lime and 2CaO. SiO 2 phase. As the formed phases were not in the form of a continuos layer, the measurement of their thickness was not possible. The results are presented in Figure 3.50 to Figure 3.55:

3CaO.SiO 2

2CaO.SiO 2

CaO

slag Figure 3.50: Interfacial region of CaO in contact with slag (basicity = 0.9) at 1500 C for the reaction time of 1 hour

186

2CaO.SiO 2

3CaO.SiO 2 CaO

Figure 3.51: Interfacial region of CaO in contact with slag (basicity = 0.9) at 1500 C for the reaction time of 2 hours

2CaO.SiO2 CaO 3CaO.SiO 2

Figure 3.52: Interfacial region of CaO in contact with slag (basicity = 0.9) at 1500 C for the reaction time of 4 hours

187

CaO 3CaO.SiO 2

2CaO.SiO 2

Figure 3.53: Interfacial region of CaO in contact with slag (basicity = 0.9) at 1500 C for the reaction time of 6 hours

3CaO.SiO 2

CaO

2CaO.SiO 2

Figure 3.54: Interfacial region of CaO in contact with slag (basicity = 0.9) at 1500 C for the reaction time of 11 hours

188

3CaO.SiO 2 2CaO.SiO 2 CaO

Figure 3.55: Interfacial region of CaO in contact with slag (basicity = 0.9) at 1500 C for the reaction time of 24 hours

3.2.2

MgO experiments

The solubility of magnesia in the rotating experiments was determined by examining a piece of slag attached to magnesia samples left after the completion of the rotating experiments as the magnesia samples did not fully dissolve in the slag after the reaction time. The magnesia sample was mounted and polished for EPMA analysis in an area close to the magnesia/slag interface. Figure 3.56 shows solid oxide/slag interface after on hour of the rotation of magnesia sample at 90 rpm in the master slag. The magnesia concentration in the area close to the magnesia/slag interface was considered as the saturation limit of magnesia in the slag. The reaction time of 1 hour during the rotating experiments was assumed enough for magnesia to reach to equilibrium.

189

Figure 3.56: SEM micrograph of the magnesia / slag interface from the samples left from the rotation experiments at 90 rpm and 1430C

This was confirmed by reacting a piece of dense magnesia with base slags in the platinum capsule in air and at 1430 C with reaction times of 0.5, 1, 2, 4, 8 hours to determine the time required for MgO to reach equilibrium in calciumaluminate base slag. The results are tabulated in Table 3.17. These results show that, after 0.5 hour of reaction time, the slag reached to equilibrium, according to the phase diagram in Figure 3.27.

Table 3.17: SEM EDS analysis of the bulk slag at1430 C in air Time (hour) 0.5 1.0 2.0 4.0 8.0 Composition of Slag (wt%) Al 25.7 25.9 26.3 26.6 25.5 Si 1.3 0.9 0.8 1.1 1.1 Ca 31.2 32.2 32.1 30.5 31.9 Mg 3.2 3.5 3.0 3.6 3.3 MgO (wt%) 5.3 5.8 4.9 6.0 5.5

190

The results of solubility of magnesia in the slags with additives are presented in Table 3.18. It can bee seen that the addition of additives in the slag changes the solubility of MgO within 20 wt%.

Table 3.18: SEM The solubility of magnesia in various slags Slag samples Master slag slag + FeO x 5% slag + FeO x 10% slag + FeO x 5% + CaF2 5 % slag + FeO x 10% + CaF2 5% MgO (wt%) 4.32 4.39 4.50 5.60 5.60

191

CHAPTER 4.

Discussion

This chapter provides discussions of effects of temperature and additives on the diffusivity of lime and magnesia in slags.

The first section covers derivation of diffusivity in slags from results of the experiments using rotating disk/cylinder technique. The diffusivity of MgO in the calcium aluminate slag and the effect of additives on the diffusivity will be explained in second part of this chapter.

The diffusivity of lime in the slag and influence of temperature and additives on the lime diffusivity will be also presented in the third section. The development of a mix-controlled model where the dissolution of lime occurs in the presence of a protective layer is detailed in the fourth section.

The fifth section deals with activation energy of diffusion and effects of various additives on the activation energy. The relation of lime diffusivity with viscosity of slag and validity of Eyrings theory is discussed in the sixth part. The seventh section discusses the relation between diffusivity and estimated ionic conductivity in the slag. The last section summarizes the key findings in this study.

192

4.1

Diffusivity of CaO / MgO in slag and effect of additives on the diffusivity

It was shown in the previous chapter that temperature and additive s have significant effects on the dissolution rate of lime/magnesia in the slags studied. As diffusivity plays a substantial role in the dissolution of solid oxides in the slag, therefore the diffusivity of lime/magnesia in the slag will be calculated using the dimensionless correlation of mass transfer under forced convection and the dissolution rate data from the experimental results. The mass transfer from a rotating crucible in liquid slag, which is used in the present study, consisted of two partial fluxes from the disk surface and the cylindrical side of the specimen. The mass transfer correlation used in each part will be detailed in the following section and the derivation of diffusivity from the combined mass transfer will be explained.

4.1.1

Mass transfer from the rotating disk

The fluid dynamics induced by rotating disk is well established. Cooper et al. (1964) and more recently Sandhage et al. (1990) have shown that when rotating disks are used, mass transfer is best described by Levich-Cochran equation for rotating disk. The dependence of mass transfer coefficient on the angular velocity of the disc and the physical properties of the liquid phase can be calculated using the Levich-Cochran equation. Levich (1962) derived the following equation for mass transfer from a rotating disk on the basis of the derivation of the mass transfer from rotating disk, which is shown in Chapter 1.

193

D = 1.61

1/ 3

1/2

(4.1)

where, (cm) is the liquid boundary layer thickness, D (cm2 /s) is the diffusion coefficient , (rad/s) is the angular velocity of rotating disk and

is the

kinematic viscosity of melt. According to Sandhage et al. (1990), Equation (4.1) is strictly valid for an infinitely large rotating disk in a semi- infinite slag bath. The disk and the slag bath can be considered nearly infinite and semi- infinite, respectively, if the thickness of the liquid boundary layer is several orders of magnitude smaller than the disk radius. The liquid boundary thickness close to the disk can be estimated using Equation (4.1). The estimated diffusivity in the liquid slag on the basis of the results calculated in the next part of this section was found to be in the order of magnitude of 10-5 cm2 /s at 1500 C. The density (~2.8 g/cm3 ) and viscosity (~8 g/cm.s2 ) of slag were estimated using published models, which will be explained later. By inserting these values into Equation (4.1), the thickness of the liquid boundary layer could be derived at the given rotation speed (90 rpm) for slag with and without additives. The thickness of boundary layer 0.01 cm was 100~130 times smaller compared to the diameter of the rotating crucible (2 cm) and depth of the slag bath (about 1cm). Therefore, Equation (4.1) could be used for the present study.

For Equation (4.1) to be valid, the fluid has to be Newtonian and the flow must be laminar. Molten silicates have been found to be Newtonian (viscosity is independent of sheer strain) over a range of applied sheer stress according to Michel and Mitchell (1975). In some cases, the presence of solid particles in the 194

slag could cause an apparent non-Newtonian behaviour as investigated by Wright, Zhang, Sun and Jahanshahi (2000) & (2001). Laminar flow near a rotating disk may be assumed if the Reynolds number is less than ~105 . The Reynolds number ( Re = r 2 / , where r is the disk radius in cm) was calculated and ranged from about 2~7 for the present work. So the flow in this study could be assumed to be laminar. Use of Equation (4.1) also requires that the fluid have a relatively large Schmidt number ( Sc = / D >> 1) . The Schmidt numbers of the melts in this study were calculated to be >104 .

Noyes-Nernst Equation (4.2) was used to calculate the dissolution rate of solid oxide from the disk side of the crucible to the liquid boundary layer, where the mass transfer is diffusion controlled.

j disk = D

CS Cb

(4.2)

Where j disk (g/cm2 .s) is the mass flux, C s and Cb (g/cm3 ) are the saturation and bulk concentratio n, respectively and D (cm2 /s) is the diffusion coefficient. Equation (4.2) can be written as:

j disk = k disk (C S Cb )

(4.3)

Where K disk is the mass transfer coefficient in liquid on the disk side of specimen and is defined as:

195

k disk =

(4.4)

Therefore on the basis of Equation (4.1), the mass transfer from the disk side of crucible can be written as:

k disk = 0.621D 2 / 3 1 / 2 1/ 6

(4.5)

Combining Equations (4.3) and (4.5) leads to the Equation (4.6), which describes direct dissolution under conditions where mass transfer through a concentration boundary layer in the melt is controlling the rate.

j disk = 0.61D 2 / 3 1/ 6 1/ 2 (Cs Cb)

(4.6)

According to the Equation (4.6), if dissolution is controlled by mass transfer through the liquid boundary layer, the dissolution rate should be proportional to the square root of rotation speed. This was investigated in Chapter 3 and it was shown that a linear relationship between the rate of solid oxide dissolution and square root of rotation speed (30 to 120 rpm) exists.

4.1.2

Mass transfer from the rotating cylinder

Generalized relationships of mass-transfer coefficient have been obtained experimentally for many specific geometries. These were shown in Chapter 1. Eisenberg et al. (1955) developed the following correlation for mass transfer

196

from a cylinder rotating in a stationary concentric crucible from studies of the dissolution of benzoic and cinnamic solids into water-glycerol solutions:

j cylinder =

k cylinder V

Sc 0.644 = 0.0791 Re

0 .3

(4.7)

Where;

j = mass-transfer j factor from the cylinder side of the crucible

V = peripheral velocity of rotating disk Sc = Schmidt number

k cylinder = mass transfer coefficient of the cylinder

Re = Reynolds number based on the peripheral velocity of the cylinder

Kosaka, Machida and Hirai (1969) studied the mass transfer from a rotating metal cylinder into liquid metal in a temperature range up to 1400 C. They employed Steel-Al, Steel- Zn, Cu-Pb, Zn-Hg and Sn-Hg as the combination of solid metal cylinder- liquid metal bath. They reported the following relationship:

j cylinder =

k cylinder V

Sc 0.644 = 0.065 Re 0. 25

(4.8)

Comparing the applicability of Equations (4.7) and (4.8), the Eisenberg et al. (1955) correlation was derived from the room temperature data with solute dissolving in water or water based solvent, while in the correlation developed by

197

Kosaka et al. (1966), the experimental temperature was up to 1400 C, which is very close to the operating temperature in the present work and also the solvent was liquid metal with more similarity with behaviour of molten slag compare to the water. Umakoshi et al. (1981) measured the dissolution of MgO into molten FeO x-CaO-SiO 2 slags at temperatures from 1350 to 1425 C. They used the correlation developed by Kosaka in their calculations to measure the mass transfer coefficient. Therefore, the Kosaka et al. (1966) correlation (4.8) has been used in this study for the mass transfer calculation from the cylinder side of the lime crucible. The mass transfer coefficient in the liquid slag was calculated by re-arranging Equation (4.8), which yields:

k cylinder = 0.065 Re 0.25 2 / 3 D 2 / 3V

(4.9)

4.1.3

Total mass transfer from the solid oxide specimen

The total mass transfer from the solid oxide sample to the liquid slag was obtained by considering the combined effect of disk and cylinder part of the lime specimen, which is expressed in Equation (4.10):

J total = j disk + jcylinder

(4.10)

This can be written as:

198

K total Atotal C = k disk Adisk C + k cylinder Acylinder C

(4.11)

Where; C = Cs Cb .

The total area of the solid oxid e crucible is consisted of:

Atotal = Adisk + Acylinder

(4.12)

The area of disk and cylinder can be defined as:

Adisk = r 2

(4.13)

Acylinder = 2 r h

(4.14)

Where, r is the radius of the lime crucible and h is the length of the crucible immersed in the melt. By inserting Equations (4.13) and (4.14) into Equation (4.11), we obtain:

K total =

r 2h k disk + k cylinder r + 2h r + 2h

(4.15)

And by substituting the corresponding values of mass transfer coefficient from

Equations (4.5) and (4.9) into the Equation (4.15), we have:

199

K total =

2h r 0.621D 2 / 3 1 / 2 1 / 6 + 0.065 Re 0.25 2 / 3 D 2 / 3V r + 2h r + 2h

(4.16)

The dissolution rate data were used to calculate the total mass transfer coefficient, as in Equation (4.17):

K total =

rate (Cs Cb ) slag

(4.17)

A model developed by Urbain (1984) was used for estimation the viscosity of slag and various additives in the slag, which is explained in detail in Appendix B. The density of slag was also estimated by a model proposed by Mills and Keene (1987), which is explained in Appendix C. By equating Equation (4.17) and (4.16) and rearranging the resulting equation, the effective diffusivity of solid oxide in the slag was obtained. Two sets of data will be presented in the next section, the results for MgO and CaO diffusivities.

4.2

Diffusivity of MgO in calcium aluminate slags

As it was explained in previous chapter, a number of preliminary experiments were performed with MgO to compare the diffusivity results to the data from the literature in order to test the experimental set- up. The MgO crucibles were rotated in the slag of CaO 55wt% Al2 O3 at 1430C. The diffusion coefficient was obtained on the basis of the rate of dissolution and solubility from the experimental results. The diffusivity of MgO in the master slag and slags with

200

additives is tabulated in Table 4.1. The error associated with calculation of diffusivity is about 30%. (see Appendix D)

Table 4.1: The diffusivity of MgO in the CaO 56 wt% Al2 O3 at 1430C in air with additives (wt%)
Slag samples Master slag slag + FeOx 5% slag + FeOx 10% slag + FeOx 5% + CaF2 5 % slag + FeOx 10% + CaF2 5% D (cm2 /s) 1.4510-5 2.510-5 410-5 210-4 310-4 viscosity (poise) 15.78 13.63 11.82 9.72 8.7

The diffusivity of MgO in slag was increased by addition of FeO x and mixture of FeO x and CaF2 . As the slag is high in alumina, the diffusion of MgO might be governed by the movement of alumina ions in the slag.

It has been investigated before by Zhang et al. (1998b) & (1998a) that bond strength between ionic species should influence the structure and, hence, viscosity of such melts. Thus, apart from the differences in bond strength between cations (Ca2+, Mg2+ and Fe3+) and oxygen ions (O 2-), the influence of various cations on the bonding environment of the aluminate anions is different. At given aluminate content, when small fraction of cations is replaced by another in melts, the changes in the structure of aluminate anions are not expected to be great. Therefore, the variation in viscosity and diffusivity caused by such 201

replacement may be attributed mainly to the difference in the strength of M2+ oxygen ion interaction, i.e., the stronger the interaction between the atoms in the melt, the more difficult the movement of atoms and slower diffusion. Therefore, higher viscosity and lower diffusivity values are expected for melts with stronger M-O bond. One may consider the melting point of oxide of a particular cation as an indication of the strength of the interaction between that cation and oxygen ions. For the metal oxides, investigated in the present work, the order of melting points are, (CaO, 2887C)>( (FeO x , 1565C). As it is seen form the results in Table 4.1, that 5 and 10 wt% FeO x increased the diffusivity of MgO.

The effect of additives on the diffusivity and viscosity of slag could also be explained on the basis of observation by Bills (1963), where the effect of MgO and FeO x on the viscosity of silicate melts was claimed to be the difference in the electrostatic binding forces which bind Fe2+ and Mg2+ cations to silicate anions.

Also in the case of addition of CaF2 to the slag, the F- ions with an ionic radii of 1.33 A replace the oxygen ions with a similar ionic radii of 1.32 A , then the existing Ca2+ cations and the ones introduced in the melt by addition of CaF2 have weaker bonding with F- compared to oxygen which makes the movement of anions easier and resulted to the faster diffusion of MgO.

Ukyo et al. (1982) measured the inter-diffusivity of MgO in the FeO 30 wt% CaO 45 % SiO 2 and CaO- 40 wt% SiO 2 20 % Al2 O3 slag in the temperature range of 1350-1450 C. The order of magnitude of diffusivity was 10-5 cm2 /s for both types of slags, which is in agreement with the result of the present work, where Fe2 O3 was added to slag. 202

Umakoshi et al. (1984a) measured the dissolution rate and mass transfer coefficient of sintered magnesia in the CaO FeO x SiO 2 slags of CaO/SiO 2 = 1 (FeO x = 20 to 65 wt%) in the range of 1350 to 1425 C They applied the rotating cylinder technique, using the dimensionless correlations and estimated the diffusion coefficient of MgO to be 1 10 5 to 3 10 5 cm2 /s at 1400 C. The results of this study are in a very good agreement with these data.

Zhang et al. (1994) measured the diffusivity of MgO in the stagnant CaO-FeOCaF2-SiO 2 slag in the temperature range of 1300 1400 C by reacting magnesia rods with stagnant slag and measuring the concentration profile of Mg in the slag. The diffusion coefficient was found to be lower than the diffusivity of MgO in the present study. This could be explained in terms of different experimental techniq ues in both measurements.

4.3

Diffusivity of CaO in calcium aluminosilicate slags

The derived diffusivity values from CaO dissolution experiments are presented in this section. The diffusivities were obtained on the basis of experimental data on the rate of dissolution and solubility of lime in the slags and are presented in Figure 4.1 and Table 4.2. The viscosity and density of the slag were calculated using Urbain model and Mill model, respectively. The error in calculation of the diffusivity data is described in Appendix D, which is about 30%.

203

Figure 4.1: Diffusivity of CaO in CaO 42 wt% Al2 O3 8 SiO2 slag with 5 wt% addition of CaF2 , MnOx, FeOx, TiO2 , SiO2 and ilmenite. The activation energy of diffusion calculated on the basis of the slope of these graphs are compared for the base slag (44 kcal/mol) versus the slag with addition of 5 wt% CaF2 (15 kcal/mol).

204

Table 4.2:Results for the measured diffusivity of CaO in the slag and the calculated slag viscosity at various temperatures Composition Temperature (C) 1430 Master slag 1500 1550 1600 Temperature (C) Slag + CaF2 5% 1500 1550 1600 Temperature (C) 1430 Slag + MnO x 5% 1500 1550 1600 Temperature (C) Slag + Imenite 5% 1550 1570 1600 Temperature (C) 1430 Slag + FeO 5% 1500 1550 1600 Temperature (C) 1430 Slag + TiO 2 5% 1500 1550 1570 Temperature (C) Slag + SiO 2 5% 1500 1550 1600 205 Diffusivity (cm2 /s) 9.2010-6 1.3210-5 1.9310-5 3.0710
-5

Viscosity (poise) 13.24 8.789 5.95 5.8 Viscosity (poise) 6.28 4.69 3.68 Viscosity (poise) 12.67 8.15 5.94 4.35 Viscosity (poise) 5.15 4.65 3.86 Viscosity (poise) 11.19 7.1 5.26 3.96 Viscosity (poise) 10.84 6.72 5.04 4.42

Diffusivity (cm2 /s) 7.2910-5 7.9410-5 1.0310-4 Diffusivity (cm2 /s) 3.5410-5 4.2810-5 5.0110 6.8210
-5 -5

Diffusivity (cm2 /s) 4.7310-5 4.7410-5 6.0110-5 Diffusivity (cm2 /s) 3.8310-5 3.8610-5 4.1310
-5

4.6810-5 Diffusivity (cm2 /s) 1.3810-5 2.3710-5 2.4910-5 3.7210-5 Diffusivity (cm /s) 8.7710 1.5110-5 2.0210-5
-6 2

Viscosity (poise) 9.24 6.61 4.96

The increase in temperature had a pronounced effect on the calculated diffusivity of CaO. Also the addition of CaF2 , Mn3 O4 , Fe2O3 , TiO 2 and ilmenite significantly increased the effective diffusivity of lime in the slag, as the addition of SiO 2 slowed down the diffusivity of CaO in the slag.

As it is seen in Table 4.2, addition of CaF2 had the strongest effect and increased the diffusivity by about 5, 4 and 3 times at 1500, 1550, and 1600 C, respectively. Addition of Fe2 O3 increased the diffusion coefficient of CaO substantially compared to the master slag. The value of diffusion coefficient at 1430, 1500 and 1550 C were about 4, 3, 2 times the corresponding values for master slag, while at 1600 C, the diffusivity was about 1.5 times the value for master slag. By adding Mn2 O3 to the slag, the diffusivity increased about 2 4 times in the temperature range of 1430 to 1600 C compared to the diffusivity data in master slag. The influence of addition of TiO 2 to the melt was to increase the diffusion coefficient about 1.2 1.8 times at 1430, 1500, 1550 and 1600 C compared to master slag. The addition of ilmenite to the slag had stronger effect than Fe2 O3 and TiO 2 in increasing the diffusivity of lime in the slag. The diffusivity results obtained with addition of ilmenite at 1550 C are comparable with the case of addition of CaF2 .

The relationship between the structure of melts and their transport properties is a key to the fundamental understanding the mechanism of diffusion in silicate melts and effect of additives on the diffusivities. It has been investigated before by Zhang et al. (1998b) & (1998a) that bond strength between ionic species should influence the structure and, hence, viscosity of such melts. Thus, apart from the differences in bond strength between cations (Ca2+, Ti2+ and Fe3+) and 206

oxygen ions (O 2-), the influence of various cations on the bonding environment of the silicate anions is different. At given silica content, when a small fraction of cations is replaced by other cations in melts; the changes in the structure of silicate anions are not expected to be great. Therefore, the variation in viscosity and diffusivity caused by such replacement may be attributed mainly to the difference in the strength of M2+ - oxygen ion interaction, i.e., the stronger the interaction between the ions in the melt, the more difficult the movement of ions and slower diffusion. Therefore, higher viscosity and lower diffusivity values are expected for melts with stronger M-O bond. One may cons ider the melting point of oxide of a particular cation as an indication of the strength of the interaction between that cation and oxygen ions. For the metal oxides, investigated in the present work, the order of melting points are, (CaO, 2887C)>( TiO 2 , 1843C)> (MnO x , 1842C)>( (FeO x , 1565C).

As it is seen from the results in Table 4.2, additions of 5 % MnO x , FeOx and 5% TiO 2 increased the diffusivity of CaO. However, the effect of a 5 wt% ilmenite addition was greater than either of the individual oxide additions, particularly at the lower temperatures. This indicates that there is a synergistic effect in the combined addition of FeO x and TiO 2 that is presumably associated with their combined effect on bond strength between the ions and oxygen.

Given the fact that the slag considered in this study is very basic, the SiO 4 4anions are mostly in the form of non-chained anions and the CaF2 addition to the slag does not have network breaking effect. By addition of CaF2 to the slag, the F- ions replace the oxygen ions then the existing Ca2+ cations and the ones introduced in the melt by addition of CaF2 have weaker bonding with F207

compared to oxygen which makes the movement of anions easier and resulted to the faster diffusion of CaO as it is shown in Table 4.2.

The diffusivity of lime in the slag of lower basicity (0.9) at 1500 C was not calculated since two non-coherent reaction layers were formed on the lime/slag interface. The formation of these layers affects the diffusivity of lime in the slag, as it is not clear when the diffusion happens in the liquid or solid phase. The lower dissolution rate of lime in the slag of lower basicity compared to the master slag indicates the effect of higher silicate anions on the dissolution.

4.3.1

Comparison of CaO diffusivity with literature data

In comparing the results of the present work with previously published diffusivity data ( Figure 4.2), it should be remembered that slag composition obviously has a major effect on the diffusivity specially where the silicate content of slag is high. Also the method of measurement will affect the diffusivity data obtained. This is particularly the case where self-diffusion rather than Chemical diffusion has been measured in the past. Generally, the values of chemical diffusivity are higher than self-diffusivity. Self-diffusion is the movement of various species present in the melt by random motions (Poirier and Geiger (1998)). The movement of these species are monitored where a fraction of one (or more) of the species is radioactive, which is called tracer diffusivity. Chemical diffusion is the movement of a species in the melt in response to the establishment of a chemical potential gradient resulting from either concentration or temperature gradients in the melt and therefore has higher values compare to self-diffusivity. 208

209

Figure 4.2: Diffusivity of calcium according to the published data and the deduced diffusivity in the present work for base slag (B: basicity, C: CaO, Al: Al2 O3 , Fe: FeO, M:Mg)

The results of the present work are initially compared with data from Johnston et al. (1974), who measured the self-diffusivity of Ca45 into three types of slags as mentioned in the Chapter 1. Their results revealed that diffusivity of Ca45 is higher in slag with no silica content, compared to slags containing silica. Based on the previous viscosity measurements by Kozakevitch (1951), the authors proposed that in the slag containing silica, there is a greater proportion of highly polymerized aluminosilicate units which make the diffusivity slower. The diffusivity in the slag of CaO 50 wt% Al2 O3 at 1500 C shows 1.9 10 5 , which is in good accord with result of diffusivity in the present study in the master slag at 1500 C (1.3 10 5 ). The presence of about 8 wt% silica in the master slag is expected to affect the diffusivity in the present study compared to the mentioned published data. At the same time, the diffusivity of calcium in the base slag from the present work is higher than the other diffusivity data by Johnston et al. (1974), where the basicity of slags were 0.3 and 0.9 respectively. Again it appears that the slag silica content is a major factor, decreasing the diffusivity, as it would explain later in this section. It was also shown by the same authors that addition of CaF2 to the melt had a substantial effect on the diffusivity of Ca45 , which is in a very good agreement with the result of the present study where CaF2 is added to the slag.

The diffusivity of lime in the iron containing slag can be compared to the data by Hara et al. (1989) as they measured the self-diffusivity of Ca45 in FeO SiO 2 CaO slag at 1270 to 1450 C. The average diffusivity was reported to be in an order of magnitude of 10-5 , which is a very good agreement with the results in the present work. 210

The data from measurements of Keller et al. (1979b) on the self-diffusivity of Ca45 in the CaO SiO 2 melts in the temperature range of 1500 to 1700 C is compared to the results from the present work. They showed that diffusivity of calcium was decreased by increasing the silica content of the slag from (mole fraction) 0.448 to 0.634. This is in agreement with the results in the present work, where diffusivity decreases with increase in the silica content of the slag. Their measured self-diffusivity data are lower compared to the present result. This difference could be explained by substantial difference in the silica content of slags, and methods of diffusivity measurements. As the slag considered in their work had high silica content compared to the about 8 wt% silica in the present work and also self diffusivity is lower than chemical diffusivity data.

The self-diffusivity of calcium in steelmaking slag of CaO 27 wt% SiO 2 40 % FeO x from the measurements by Goto et al. (1977) at various oxygen pressures and at 1360 to 1460 C compared to the results from the present work. The diffusivity data in the present study, where F2 O3 is added to the slag show the same order of magnitude and is in a good agreement with their data.

The self-diffusivity of Ca45 measured by Towers et al. (1957) and later Saito et al. (1958) show lower values compare to the results from the present work. Apart from slag che mistry where the silica content of slag was higher in the works by these authors, the method of measurements could be another reason for such a difference. The low diffusivity data by Saito et al. (1958) in comparison with other calcium diffusivity measurements have been also addressed by Keller et al. (1979b) and Johnston et al. (1974).

211

The effect of iron oxide on the dissolution rate of lime in the CaO-SiO 2 -Al2 O3 and FeO-CaO-SiO 2 slag was reported by Matsushima et al. (1977) using rotating cylinder method, which has been reviewed in Chapter 1. The authors calculated the mass transfer of CaO in slags from the rate of dissolution data, thus the diffusivity data have been derived in the present work on the basis of their mass transfer data. The deduced diffusivity data as well as the original mass transfer data form their experiments are tabulated in Table 4.3. It is apparent from their results that diffusivity of lime at 1500 C is about twice the diffusivity of lime obtained through the present work. At the same time, addition of FeO x to the slag increased the diffusivity of lime by a factor of two, which is in good agreement with present results.

Table 4.3: Values for mass transfer coefficient, thickness of boundary layer and deduced effective diffusivity of lime in the slag according to Matsushima et al. (1977) Slag CaO - 40 wt% SiO 2 - 20 % Al2 O3 CaO - 40 wt% SiO 2 - 20 % FeO Temperature ( C ) 1500 1500 1400 1400 Speed (rpm) 200 400 200 400 K (cm/s) 2.9010-4 5.3010-4 9.7010-4 1.7110-3 D* (cm2 /s) 2.6310-5 2.9810-5 5.1910-5 5.5710-5

D* : calculated by the present author

The diffusivity of lime in the present work could be compared with the diffusivity data of similar oxides using the same experimental technique. Umakoshi et al. (1981) measured the dissolution rate of MgO into molten slag 212

CaO SiO 2 Fet O slags at temperatures from 1350 to 1425 C using rotating cylinder method. They used non-dimensional correlation and estimated the apparent diffusion coefficients with the order of magnitude of 10 5 cm2 /s at 1400 C. The apparent diffusivity data in the present work, where Fe2 O3 is added to the slag are within the same order of magnitude and are in good accord with these data.

The effect of transition metals on increasing the diffusivity of oxides in the slag in the present work can also be compared to the work done by Lee et al. (2001). They measured the apparent diffusivity of alumina in the calcium aluminosilicate melt of (30 wt% Al2 O3 53 % CaO 5 % MgO 12 % SiO 2 ) using a rotating disk method. As shown in Figure 4.3, addition of FeO x and MnO x increased the effective diffusivity of slag substantially. It was reported that addition of 5 and 10 wt% FeO x increased the diffusivity by about 2 and 2.5 times and addition of 5 and 10 wt % MnO x showed more profound effect and increased the diffusivity by about 4 to 6 times. The results of diffusivity of lime in the present work in the slag with FeO x and MnO x show a very good agreement, with the same order of magnitude. In both results addition of FeO x and MnO x into the slag increased the rate of dissolution and diffusivity of oxides in the slag.

213

Figure 4.3: Influence of addition of FeOx and MnOx on the apparent diffusivity of alumina at 1560-1590C according to Lee et al. (2001)

The effect of addition of iron oxide to the slag in the present work can be compared to the difference in tracer diffusivity of Fe59 and Ca45 in silica saturated melts of the FeO CaO SiO 2 system at 1600 C which was investigated by Keller et al. (1986). The Fe59 showed a higher diffusivity 1 10 5 compared to 2 10 6 cm2 /s for silica and the higher concentration of iron oxide resulted in higher diffusivity of Fe59 in slag. The authors postulated that the higher diffusivity of iron was possibly due to the different bonding by oxygen, which caused a transition state in the jump of Fe2+ ions from one site to the next. The increase in the diffusivity of CaO in the slag in the presence of FeO x in the present work, is in accord with Kellers findings, where the present results are explained by the effect of additives on the bonding environment of silica melts. 214

The tracer diffusivity of Ca45 in their work shows lower values compare to the chemical diffusivity in the present study.

The effect of addition of 5 wt% SiO 2 to the slag in the present work could be compared to the data by Keller et al. (1979a), who measured the tracer diffusivity of Si31 in CaO-SiO 2 melts (Mole fraction of silica NSiO2 = 0.484 0.634) at a1600 C. They reported that diffusivity of Si31 decreases with increasing silica content of the slag and the values of Si31 diffusivity were found to be lower than that of Ca45 by approximately one order of magnitude in the silica rich melt. The difference decreased with decreasing silica content. The authors argued that the Si31 in the CaO SiO 2 melts is normally assumed to bounded to large complex silicate anions, which move slower than the cations but as the diffusivity of oxygen is much higher than silica, they suggested silica ions would rotate during time intervals in the melt and harbour the tracer atoms, which is easier than translation of silicon ions in the molten liquid. The lower values of diffusivity of silica compared to Ca and Fe reflect the bonding between these ions and oxygen in the melt. In the present work, the order of diffusion of Ca in the melt with additives is slag + FeO x > master slag > slag + additional SiO 2 , which follows the same trend as the mentioned published data.

The effect of additives on the diffusivity of CaO in slag in the present work can be compared to results by Ukyo et al. (1982) on quasi- inter-diffusivities of several solute oxides in CaO SiO 2 Al2 O3 and Fe2 O3 CaO SiO 2 slags. Comparison of diffusivity values of FeO x , MnO x , P2O5 , TiO 2 reveals that the oxides diffused in the order of; MgO>FeO x >MnO>P2 O5 >TiO 2 in the CaO SiO 2 Al2 O3 slag. As the order of diffusivity for the oxides in FeO x CaO SiO 2 slag 215

was: MgO>MnO>NiO>P2 O5 >TiO 2 . The substantial increase in the diffusivity of CaO in the slag in the present study by addition of MnO x and FeO x , is in accord with the findings by Ukyo et al. (1982).

4.4

Diffusion in a Mix controlled regime

It was shown in Chapter 3 that when lime sample was rotated in the calcium aluminio-silicate melt at 1430 C, the CaO concentration dissolved in the slag increased approximately linearly with reaction time, but after about 10 minutes of reaction, the CaO dissolution curve reached a plateau at a level of less than 52 wt%, which is far less than the saturation level of 59 wt%, given that the rotating lime sample was not fully dissolved at the end of one hour reaction time. These results can be explained in terms of CaO Al2 O3 SiO 2 phase diagram (Figure 4.4) where formation of a solid layer on the surface of lime occurred and consequently the dissolution of lime was slowed down.

216

Figure 4.4: CaO-Al2 O3 -SiO2 phase diagram

To investigate this further, experiments were carried out in which the lime was reacted with base slag at 1430C under static condition. These experiments were aimed at obtaining evidence of the formation of a solid layer and measuring growth rate of the layer at the lime/slag interface. The SEM examination of solidified samples confirmed the presence of a 3CaO.Al2 O3 layer and the thickness of the layer was observed to increase with increasing reaction time, which was demonstrated in Chapter 3. The growth rate of the 3CaO.Al2 O3 layer

217

was found to be linear and correlated to reaction time according to Equation (4.18):

x 2 = 2 10 9 t

(4.18)

Where x is the change in thickness in millimiters and t is reaction period in seconds.

The diffusion of lime in the solid layer was then estimated according to the following Equation:

(x )2
t

(4.19)

The value of diffusivity in the solid layer was deduced from the slope of the x 2 versus time, where there is a linear relationship between (x )2 and time, as shown in Figure 4.5.

218

Figure 4.5: Estimation of CaO diffusion through a solid layer

According to Figure 4.5, the 3CaO.Al2 O3 (C 3 A) solid layer grows linearly with time. It is postulated that once a complete layer of C3 A covers the surface of a lime specimen, it increases in thickness by formation of C3 A layer at the lime/C 3 A interface and, the rate of formation of C3 A is much faster than the rate of dissociation. The linear behaviour of growth rate suggesting that dissolution process is controlled predominately by the diffsion of lime in the reaction layer. The diffusion coefficient of lime in the 3CaO.Al2 O3 solid layer is estimated on the basis of Equation (4.19) to be 2 10 9 cm2 /s at 1430 C.

Comparison could be made between the diffusion of CaO in the 3CaO. Al2 O3 layer and growth rate of this layer with the previously published data. Zhang et al. (1994) studied the dissolution of MgO in the CaO FeO CaF2 SiO 2 slags

219

in the temperature range of 1573-1673 C. They observed a layer of magnesiowustite at the MgO/slag interface and measured its thickness versus time. The authors also estimated the inter-diffusivity of MgO in the solid layer by applying the cylindrical coordinate diffusion equation and it was found to be
7 10 8 cm2 /s. The average diffusivity was also estimated by applying Appels

equation, which is the solution to Ficks second law with variable diffusivity in multiphase system and the result was 3 10 8 cm2 /s. Attempt was made in the present work to deduce the diffusion coefficient by the slope of the line in the plot of (thickness)2 versus time as it is shown in Figure 4.6. The diffusion coefficient value was found to be 3 10 8 cm2 /s, which shows that the assumption made in Equation (4.19) is in a very good agreement with the result given by analytical solution. The diffusivity of CaO in the present work is an order of magnitude lower than the diffusivity of MgO in the Zhangs work.

220

Figure 4.6: variation (Thickness2 ) of magnesiowustite layer as a function of time on the basis of work done by Zhang et al. (1994)

Allen, Sun and Jahanshahi (1995) measured the thickness of the reaction layers of spinel and wustite when MgO was reacted with slag of 80 wt% FeO x and 20 wt% CaO. Using their results, an attempt was made in this work to estimate the diffusivity of MgO in the wustite layers at various temperatures by plotting square of thickness versus time according to Figure 4.7. The diffusion coefficient for Wustite is between 3 10 9 to 1 10 7 cm2 /s depending on the temperature. In the case of spinel the diffusion coefficient at 1300 C is estimated as 2 10 9 (cm2 /s) according to Figure 4.8. The results show that the estimation of diffusivity in the 3CaO.Al2 O3 layer in the present work is in the range of published data on growth rate and diffusivity of oxides in the solid layer.

221

Figure 4.7: Variation of (thickness2 ) of wustite layer with time deduced from data according to Allen et al. (1995)

Figure 4.8: Variation of the (thickness2 ) of the spinel layer with time deduced from data according to Allen et al. (1995)

222

Blank and Pask (1969) investigated the inter-diffusion in the MgO Fex O system, which exhibits complete solid solution under low oxygen pressure. In the MgO phase the value of Mg diffusion was found to be 1.24 10 9 cm2 /s and the ferrite phase, diffusion for Fe was with an average value of 5.58 10 8 cm2 /s.

Thus, we can conclude that the assumption made in the present work to deduce the value of diffusivity in solid, results in diffusivity data, which is within the values of diffusivity in solid layer and can be accepted as a reasonable assumption.

Therefore, the mass transfer coefficient of lime in the solid layer was estimated according to the following Equation:

K solid =

D x

(4.20)

As it was mentioned earlier, mass transfer of lime in the liquid phase could be estimated by Equation (4.16). The dissolution of lime into the master slag at 1430C initially started from the diffusion of lime in the liquid slag. On formation of the 3CaO.Al2 O3 protective layer on the lime/slag interface, the diffusion of lime occurs through a mix-controlled regime, i.e. diffusion in the solid layer and the diffusion in liquid slag. The newly formed solid layer is very thin at early stages of dissolution, at this stage the mass transfer in the solid layer is higher than the corresponding value for the liquid phase due to the thickness of the product layer so the overall diffusion is predominately controlled by the diffusion in the liquid slag. On increasing the reaction time, the solid layer on the 223

lime surface grew and the mass transfer coefficient in solid became smaller compared to the mass transfer in the liquid phase, and the diffusion becomes predominantly controlled in the solid layer. This would lead to slow diffusion of CaO in the solid layer, which eventually controls and slows down the dissolution process.

A mixed-control model was developed on the basis of diffusion in the solid and liquid phases. The dissolution data obtained from the developed model was compared with experimental results.

As the concentration of CaO in the 3CaO.Al2 O3 layer was constant, Ficks First law, which according to Equation (4.21) states that the magnitude of the mass flux is proportional to the concentration gradient at that point, could not be used;

j i = D.

C x

(4.21)

Where, J i is mass flux of i species (g/cm2 .s), D is the diffusion coefficient (cm2 /s), and concentration C is expressed in g/cm3 .

It was assumed in this work that the diffusion rate is proportional to the gradient of its activity in both solid layer and liquid melt according to the Equation (4.22):

224

j = DC

a y

(4.22)

Where, D is the diffusion coefficient (m2 .s-1 ) and y is the distance in the direction of diffusion. For steady-state transfer, the rate at which CaO diffuses through the 3CaO.Al2 O3 layer is equal to the rate at which it diffuses into the bulk liquid slag. Therefore, if k l and k s are the local mass transfer coefficients in the solid layer and liquid phase respectively, then the flux of CaO could be expressed as;

j = k s C s (as a i ) = k l Cl (ai al ) Where: k s = mass transfer coefficient of CaO in solid phase k l = mass transfer coefficient of CaO in liquid phase C s = concentration of CaO in solid phase Cl = concentration of CaO in liquid phase a s = activity of CaO in solid phase a l = activity of CaO in liquid phase a i = activity of CaO in the product layer/slag interface

(4.23)

225

At equilibrium, a i , activity of CaO is the same at the interface of the solid layer and liquid slag adjacent to the liquid/slag interface ; therefore, it was eliminated from both sides of Equation (4.23), and it becomes in Equation (4.24):

j = Cl C s K total (a s al )

(4.24)

Where the total mass transfer K total according to Equation (4.23) is expressed as:

K total =

kl ks k sC s + k l C l

(4.25)

k s , was found according to Equation (4.20) and k l on the basis of Equation (4.16). The activity of CaO in lime specimen is equal to 1 and the activity of CaO in bulk slag was calculated by the MPE model developed by Zhang, Jahanshahi, Sun, Chen, Borke, Wright and Somerville (2002).

The concentration of CaO dissolved in the slag was estimated using the mixcontrolled model by integrating the rate Equation (4.24) and the results for two lowest and highest rotation speeds were compared with experimental data. The results are shown in Figure 4.9 to Figure 4.13. There is a good agreement between the results from modelling and the experimental data. At a rotation speed of 60 rpm, the data from modelling deviated slightly from the experimental data after about 20 minutes. This could be due to disintegration of the lime crucible after about 20 minutes due to the physical characteristics of the

226

particular lime sample used in that experiment, as the CaO concentration reached to a plateau in compararison with the experiments at other rotation speeds.

Figure 4.9: The CaO concentration predicted by mix-controlled model and the experimental data at30 rpm & 1430 C

227

Figure 4.10: The CaO concentration predicted by mix-controlled model and the experimental data at 60 rpm & 1430 C

Figure 4.11: The CaO concentration predicted by mix-controlled model and the experimental data at 90 rpm & 1430 C

228

Figure 4.12: The CaO concentration predicted by mix-controlled model and the experimental data at 120 rpm & 1430 C

Figure 4.13: The CaO concentration predicted by mix-controlled model and the experimental data at150 rpm & 1430 C

229

4.5

Activation energy

The activation energy for diffusion (Q) in the slags studied has been calculated using the Arrhenius equation:

Q D = D 0 exp RT

(4.26)

Where

D0 = constant for a given solute; independent of temperature

D = diffusivity (cm2 /s)

Q = activation energy for the diffusion process (cal/mole)

R = 1.9872 is the universal gas constant (cal.mol-1 .K-1 )

A plot of Ln (diffusivity) versus 1/temperature was constructed and the slope of the lines were used for calculation of the activation energy which is presented in Figure 4.14 and Table 4.4.

230

Figure 4.14: Arrhenius plots for calculation of the activation energy for diffusion of CaO in the master slag and slags with additives

Table 4.4: Activation energy for master slag and slag with additives

Slag Composition Master slag slag + CaF2 5% slag + FeO x 5% slag + MnO x slag + ilmenite 5% slag + TiO 2 5% slag + SiO 2 5%

Activation Energy (kcal/mole) 43.18 17.24 8.46 23.54 28.38 30.02 79.24

231

The activation energy decreased markedly with addition of 5 wt% of Fe2 O3 , CaF2 , Mn3 O4 , ilmenite and TiO 2 but increased with addition of 5 wt% SiO 2 in the slag. The activation energy ( Q) is generally regarded as the energy required for a species to diffuse in the melt. In the master sla g, the energy barrier for the diffusion of CaO is the movement of silica anions. By addition of metallic oxides and CaF2 , the bonding environment of the melt becomes weaker, the silicate anions can move easier and the energy barrier for diffusion of CaO becomes lower to the master slag. By addition of SiO 2 , the silicate anions become restricted and their movement is slow thus the energy barrier for diffusion would increase.

Attempt was made to deduce the activation energy diffusivity of CaO and other oxides in the slag from the published data using Equation (4.26).

The results form data published by Johnston et al. (1974) show a decrease in the activation energy of Ca diffusion in the CaO 20 wt% Al2 O3 42 % SiO 2 slag when CaF2 was added to the slag which is shown in Figure 4.15. The activation energy reduced from 30 kcal/mole to 16 kcal/mole, which is in accord with the result of the present work.

232

Figure 4.15: Arrhenius plot for the diffusion of Ca2+ in the CaO 20 wt% Al2 O3 42% SiO2 , used in the calculation of activation energy on the basis of data from Johnston et al. (1974)

Also by comparing the activation energy of diffusion for Ca, F and Fe in the same CaO Al2 O3 SiO 2 slag, deduced from the work by Johnston et al. (1974), the activation energy for diffusion of fluorine(16 kcal/mole) shows the lowest value while this value for iron (20.4 kcal/mole) shows less activation energy in comparison with activation energy for diffusion of Ca (30 kcal/mole). These values show the level of bonding of each element with the silica in the slag. This is in good agreement with the present results, where addition of CaF2 and Fe2 O3 lowered the activation energy of diffusion significantly.

233

Figure 4.16: Arrehnius plot for the diffusion of Ca2+, F-1 and Fe2+ in the CaO 20 wt% Al2 O3 42% SiO2 slag, used in the calculation of activation energy according to data from Johnston et al. (1974)

Saito et al. (1958) measured the activation energy of diffusion of calcium in the binary and ternary systems. The measured activation energy by the authors is shown in Table 4.5. The activation energy calculated in the present work is within the same range.

234

Table 4.5: The activation energy for binary and ternary slags according to Saito et al. (1958) Slag system Activation energy (kcal/mole)

CaO SiO 2 CaO Al2 O3 CaO SiO 2 Al2 O3 (CaO/SiO 2 = 0.73), Al2 O3 = 10% CaO SiO 2 Al2 O3 (CaO/SiO 2 = 1.33), Al2 O3 = 20% CaO SiO 2 Al2 O3 MgO (MgO :3~10%)

50 20
60

60 20 50 20 50 20

The effect of addition of silica on the activation energy of diffusion can be shown by studying the diffusion data measured by Keller et al. (1979b). On deducing the activation energy from the diffusivity data of calc ium in the molten CaOSiO 2 over a range of temperatures, it can be seen that by increasing the silica content of slag, the activation energy increased from 28 to 35 kcal/mole. The result of the present work follows the same trend as the addition of SiO 2 increased the activation energy of diffusion.

235

Figure 4.17: Arrhenius plot for diffusion of Ca in the CaO SiO2 slags according to diffusivity data from Keller et al. (1979b)

As it is mentioned before, Ukyo et al. (1982) measured the diffusivity of several solute oxides in liquid CaO 40 wt% SiO 2 20% Al2 O3 in a temperature range of 1350 to 1450 C. The activation energy from their diffusivity data was deduced by the present author and illustrated in Table 4.6.

Table 4.6: Activation energy for diffusion of various oxides in CaO- 40 wt% SiO2 - 20 % Al2 O3 slag according to Ukyo et al. (1982) Oxides Activation energy (kcal/mole) MgO TiO 2 MnO Fe2 O3 P2 O5

22.11

53.46

54.52

63.2

68.11

236

The results show that activation energy for diffusion of MgO is the lowest with TiO 2 and MnO also having lower activation energies than Fe2 O3 and P2 O5 . These results show that the movement of MgO, Fe2 O3 and MnO in the melt are faster than others. The lower activation energy of diffusion of lime in the slag containing MnO x and FeO x in the present study is in accord with these data.

The activation energy of single elements in the slag of CaO 40 wt % SiO 2 20% Al2 O3 were deduced from the diffusivity data reviewed by Nagata et al. (1982). The results are tabulated in Table 4.7, where Fe and Ti show lower activation energies compared to the Ca and Si. This is in accord with the results of the present work where transition metal decreased the activation energy of CaO in the slag.

Table 4.7: Activation energy from diffusivity data of various ions in liquid CaO- 40 wt% SiO2 - 20 % Al2 O3 slag according to Nagata et al. (1982) Oxide Activation energy (kcal/mole) Mg Fe Ti Al Mn Ca Si P

26.1

31.3

50.2

55.2

58.2

67

68.1

76.3

4.6

Relationship of diffusivity with viscosity

The Eyrings theory of diffusion (Glasstone et al. (1941)) has been explained in Chapter 1, where a model was presented for liquid diffusion based upon their theory of absolute reaction rates and using the concept that the liquid structure 237

contains a number of holes or void spaces. This theory was used by Yu et al. (1997) to investigate the mechanism of dissolution in the molten slag . The slope
D of the line in plot of Ln( ) versus Ln( ) was considered as the basis for T

judgment as according to the Eyring relation this slope is expected to be 1, whether the diffusing molecules are big or small. Their diffusivity data matches in this correlation well and the effective diffusivity of alumina in the CaO Al2 O3 SiO 2 melts was shown to be very close to the self-diffusivity of Si4+ measured by Cooper et al. (1964), thus it was proposed that alumina diffusion was controlled by the mobility of silicate ions.

In the present work, attempt was made to apply the Eyrings correlation to the
D master slag and slags with different additives, so the plot of Ln( ) versus T
Ln( ) was established in Figure 4.18. It is shown that master slag and slag with

addition of SiO 2 follow the Eyring relation as the slope of the lines are close to 1. However, additions of CaF2 , Fe2 O3 , Mn3 O4 , TiO 2 and ilmenite have a different effect on the mechanism of diffusion so the results show a deviation from Eyring relation. This could be explained by the argument mentioned above, i.e.; the movement of species in the melt is governed by the movement of silica anions in the melt. In the case whe re SiO 2 was added to the melt, silica anions lock the silica anions together and consequently the diffusion of calcium cations decreases, again the diffusivity data shows agreement with the Eyrings theory. However addition of other additives, like Fe2 O3 , TiO 2 , Mn3 O4 , ilmenite and CaF2 , makes the bonding environment between the cations and silica anions weaker therefore, the movement of silica anions becomes easier where the Ca2+ 238

cations can move and diffuse faster. So there is a deviation from the Eyrings correlation, as the movement of CaO is not controlled by the movement of silica anions. With additions and the consequent weaker bonding environment in the silica melt, there is more deviation from the Eyrings correlation.

Figure 4.18: Investigation of applying Eyring theory in diffusion of CaO in the slag

4.7

Ionic conductivity

The ionic conductivity of the master slag and slags with additives was calculated in order to investigate whether the transport of ionic species in the melt has the same mechanism to electrical conduction. It should be noted that relationship between conductivity and diffusivity is complex as different driving forces (chemical potentials and electrical charge) affect the effective concentration of

239

charge carriers and transfer numbers. At the present work, the ionic conductivity is computed from the diffusivity data assuming that the conduction is solely by motion of Ca2+ ions and that the Nerst-Einstein equation is valid. The following Equation (4.27) applies for estimation of ionic conductance:

DF 2 Z 2 C = RT

(4.27)

Where:

= ionic conductivity ( 1cm 1 )

D = diffusivity (cm2 /s) of Ca2+, which is assumed to be equal to the diffusivity

of CaO in the melt.

2+ Z Ca = 2 is the charge of the Ca ion,

CCa = Ca2+ concentration in moles.cm-3

F = 96485.3415 C.mol-1 is the Faraday constant

T = absolute temperature (K)

R = 8.3144 J.mol-1 .K-1 is the universal gas constant

The ionic conductivity is tabulated in Table 4.8. The results followed the same trend as the variation of diffusivity, which decreased by addition of 5 wt% SiO 2 and increased by temperature and also addition of 5% wt CaF2 , Fe2 O3 , TiO 2 , Mn3 O4 and ilmenite.

240

Table 4.8: Estimated Ionic conductivity ( 1cm 1 ) of CaO-Al2 O3 -SiO2 slag and slags with 5 wt% additives at various temperatures

Temperature (C) 1430 1500 1550 1600

Conductivity in slag with addition of 5 wt% additives Master slag 0.61 0.84 1.08 1.84 CaF2 4.33 4.64 5.33 Mn3O4 Ilmenite 2.24 2.60 2.95 3.92 2.58 2.65 3.08 Fe3O4 2.42 2.35 2.44 2.69 TiO2 1.02 1.37 1.35 2.15 SiO2 0.53 0.88 1.15

According to Richardson (1974), the conductance of silicate melts is always raised by the addition of metal oxides and the conduction mechanism is primarily ionic. In cases where the metal oxide is an electronic conductor such as FeO x , which conduct by virtue of the presence of cations of variable valency, conduction becomes more electronic.

The validity of Equation (4.27) in estimation of conductivity has been investigated by Keller et al. (1979b). They measured the electrical conductivity of Ca45 in CaO-SiO 2 melts, using the four-point method and compared experimental data with calculated data using Equation (4.27). Although the computed values are lower than experimental data (Figure 4.19), they are quite comparable. The difference is low at low SiO 2 content. The experimental conductivity value at 1600 C and for a silica mole fraction of about 0.44 shows a value of 0.5 1cm 1 . The estimated conductivity in the present work for the base slag shows a higher value (1.84 1cm 1 ), which can be explained by slag chemistry where the silica content is much less and also the fact that the conductivity data in the present study in on the basis of chemical diffusivity which is higher than tracer diffusivity. 241

Figure 4.19: Electrical conductivity of CaO SiO2 slag, measured experimentally calculated as a function of mole fraction of silica at 1600 C after Keller et al. (1979b)

The temperature dependence of the electrical conductivity is usually expressed by the Arrhenius relationship in Equation (4.28):

E = A exp RT

(4.28)

Where; A is the constant, E the activation energy, R the gas constant and T the thermodynamic temperature. The activation energy of conductance can be calculated on the basis of plot of Ln (conductivity) versus (1/temperature). Attempt was made in the present work to compare the activation energy of conductivity and activation energy of diffusivity. Firstly the activation of

242

conductivity for various slag compositions are deduced in the present work on the basis of the data in the literature and compared with the published data on activation energy of diffusivity of Ca in the slags of similar composition. This would justify the validity of the comparison between these two quantities. Sarkar and Sen (1978) & Sarkar (1989) measured the conductivity in the slag of CaO 26 wt% Al2 O3 35% SiO 2 4% MgO in the temperature range of 1500 1600 C. The activation energy of conductivity (38 kcal/mole) deduced according to their data shows a good agreement with the activation energy of diffusivity of Ca ( 50 20 kcal/mole) calculated by Saito et al. (1958) in the slag system of CaO SiO 2 Al2 O3 (CaO/SiO 2 = 1.33), Al2 O3 = 20 wt%. Nesterenko and Khomenko (1985) measured the conductivity of slags with composition of CaO (34 49 wt%) SiO 2 5% Al2 O3 at temperatures of 1500 1600 C and the deduced data of activation energy are in the range of 20 48 kcal/mole. These results are compared with the deduced value of activation energy of Ca diffusivity according to Keller et al. (1979b), where they studied the CaO SiO 2 system (SiO 2 = 0.448 0.634 mole fraction). As the activation energy of diffusion is in the range of 28.16 35.19 kcal/mole, a good agreement exists between these two activation energies of conductivity and diffusivity. Winterhager, Greiner and Kammel (1966) measured the electrical conductivity of CaO 19 wt% Al2 O3 40% SiO 2 5% MgO in the temperature range of 1350 1550 C. The activation energy of conductivity of 42.71 kcal/mole is deduced in the present wo rk on the base of their data and is in a good agreement with the activation energy of diffusivity (50 20 kcal/mole) calculated by Saito et al. (1958) for the slag of similar chemistry. Thus suggesting the validity of Equation (4.27) and hence its

243

application to predicting diffusivity of slags from a knowledge of the electrical conductivity of slags.

Therefore, it can be seen that the activation energies of conductivity, which are presented in Table 4.9 can be compared with those for activation energy of diffusion in Table 4.4.

Table 4.9: Estimated activation energy of conductivity for master slag and slags with 5 wt% additives Slag Composition Master slag slag + CaF2 5% slag + TiO 2 5% slag + FeO x 5% slag + ilmenite 5% Slag + MnO x 5% slag + SiO 2 5% Activation Energy (kcal/mole) 39.66 13.68 26.52 3.61 24.70 19.97 51.28

It can be seen that the presence of CaF2 , FeO x , TiO 2 , MnO x and ilmenite in the slag decreased the activation energy and addition of SiO 2 increased the activation energy of conductivity. Although the activation energy of diffusion is higher than conduction, they both follow the same trend, supporting the theory that ionic conductance and chemical diffusion are controlled by the same processes

244

involving the same energy barriers, which is movement of the large anions. Addition of metallic oxides, ilmenite and CaF2 makes the movement of silicate anions easier in the melt due to the weaker bonding environment, which decreases the energy barrier for conduction of Ca2+ cations. Addition of SiO 2 increases the bonding of silica anions, which leads to the higher energy barrier and slower conductivity of Ca2+ cation. It should be mentioned that as the concentration of additives to the slag, (such as FeO x , MnO x and TiO x ) is low, the conduction is predominantly ionic but at high levels of additives it becomes electronic.

4.8

Summary of key findings

The diffusivity of MgO in the CaO - 46 wt % Al2 O3 slag was measured with rotating disk/cylinder technique, at 1430C. It was observed that addition of FeO x and combination of FeO x and CaF2 increased significantly the diffusivity of MgO in the slag.

The diffusivity of CaO in the CaO 42 wt % Al2 O3 8% SiO 2 slag was measured in the temperature range of 1430 1600 C, using rotating disk/cylinder technique. The effect of temperature and addition of additives (CaF2 , FeOx , TiO 2 , ilmenite, MnO x and SiO 2 ) on the diffusivity were investigated. By increasing the temperature and addition of 5 wt% CaF2 , FeOx , TiO 2 , ilmenite and MnO x , the diffusivity of lime in the slag was increased while addition of 5 wt% SiO 2 reduced the diffusivity.

245

The activation energy of diffusion was calculated on the basis of changes of lime diffusivity in various slags with respect to temperature. It was shown that additives could be categorized in two groups, those that increased and those that decreased activation energy. The effect of additives on the activation energy could explain the mechanism of diffusion of lime in the slags with various additives.

The relationship between diffusivity and viscosity was demonstrated with Eyring theory, where the mechanism of diffusion in the melt was expressed according the validity of Eyring correlation for various slags.

The ionic conductivity of lime in the slags was calculated and it was shown that there is a direct relations hip between the diffusivity and ionic conductivity of lime in the slag.

The results show that FeO x , ilmenite and MnO x are comparable with CaF2 with respect to increasing the diffusivity of lime in the slag and can be used instead of CaF2 for effective dis solution of lime in the slag.

246

CHAPTER 5.
5.1

Conclusion

Dissolution rate of MgO in calcium aluminate slag and lime in the calcium aluminosilicate slags

The dissolution rate of MgO in CaO 55 wt% Al2 O3 base slag has been determined at 1430 C and in air, using the rotating disk/cylinder technique. The results showed the increase in the magnesia content of slag as function of period of rotation of magnesia sample in the melt. The effect of varying the rotation speed on the dissolution rate was investigated. It was shown that there is a linear relationship between the rate of dissolution with 1/2 and 3/4 -th power of rotation speed ( ) and A 1 / 2 + B 3 / 4 over the rotation speeds of 60 to 120 rpm. The dependence of the rate of magnesia dissolution on the rotation speed suggests that the measured dissolution of MgO is most likely controlled by diffusion in the liquid boundary layer. The effect of addition of Fe2 O3 and (Fe2 O3 +CaF2 ) on the rate of dissolution was studied at a constant rotation speed of 90 rpm and at a constant temperature of 1430 C. It was shown that while the average dissolution rate of magnesia in the base slag was about 2.7 10 5 g/cm2 .s, additions of 5 and 10 wt% Fe2 O3 increased the dissolution rate by a factor of 2 and 4, respectively. It was also found that addition of a mixture of (CaF2 5 wt% + Fe2O3 5 wt%) and (CaF2 5 wt% + Fe2 O3 10 wt%) increased the dissolution rate considerably by a factor of 11 and 8, respectively.

The dissolution rate of CaO in CaO 42 wt% Al2 O3 8% SiO 2 base slag in a temperature range of 1430 1600 C and in air was studied by using the rotating 247

disk/cylinder technique At a given temperature, the results revealed the increase in the lime concentration versus the period of rotation of lime cylinders in slag. The effect of rotation speed on the rate of dissolution was studied by conducting experiments in a range of rotation speed of 30 to 150 rpm. As there was a linear dependency of dissolution rate with the 1/2 and 3/4 th power of rotation speed ( ) and A 1 / 2 + B 3 / 4 , it was concluded that liquid slag mass transfer played a significant role in controlling the dissolution. The dissolution rate also increased with temperature at constant rotation speed. At constant rotation speed of 90 rpm, the effect of 5 wt% additives, such as CaF2 , FeOx , TiO 2 , MnO x , ilmenite and SiO 2 in the slag were quantified at various temperatures. The average dissolution rate (g/cm2 .s) of lime in the base slag over a temperature range of 1430 1550 C, was of the order of magnitude of 10-5 (g/cm2 .s) and at 1600 C, the dissolution rate was an order of magnitude higher. While CaF2 had the highest effect, increasing the rate of dissolution by about a factor of 3, addition of FeO x , MnOx and ilmenite increased the rate of dissolution (about a factor of 2) and proved to be comparable with CaF2 . The effect of basicity on the dissolution of lime at constant temperature of 1500 C was investigated, where it was shown that dissolution rate was about a third of the rate data for the slag with basicity of 6. During static reaction of lime with slag with basicity of 0.9, the formation of two non-coherent phases of 2CaO.SiO 2 and 3CaO.SiO 2 on the lime/slag interface was observed, where the measurement of thickness of these phases and consequently the diffusivity measurements were not possible.

During the dissolution of lime in calcium aluminosilicate base slag at 1430 C, a layer of 3CaO.Al2 O3 was formed on the lime/slag interface, which slowed down 248

the dissolution rate and subsequently the diffusivity of lime in the slag. This suggested that dissolution of lime in the slag occurred in a mix-controlled regime. By conducting static experiments, it was found that the thickness of the solid layer increased with reaction time and the growth rate the layer was measured to be about ( 4 10 5 mm/s). A Mix-Controlled model has been developed on the basis of the assumption that the dissolution rate is proportional to the gradient of its activity in both solid layer and liquid melt. The model predicts the concentration of lime in the slag during the course of dissolution of lime while the solid reaction layer grows on the lime/slag interface. The result of modelling was compared to the experimental data and very good agreement was found, confirming the dissolution of lime in a mix-controlled regime at 1430 C.

5.2

Solubilities of MgO in calcium aluminate slag and CaO in calcium aluminosilicate slags

The solubility of MgO in the various slags was determined by analysing the rotating magnesia samples after the completion of the experiments. The MgO concentration in the slag attached to the magnesia samples close to the interface was measured quantitatively. The solubility of MgO was shown to be about 4.3 wt% in the base slag and introduction of the additives i.e. 5 and 10 wt% FeO, (CaF2 5 wt% + Fe2 O3 5 wt%) and (CaF2 5 wt% + Fe2 O3 10 wt%) to the slag, increased the solubility slightly to about 5 5.6 wt%. It was found that within experimental scatter, the existence of additives in the base slag, does not increase the solubility MgO in the slag.

249

CaO solubility in CaO 42 wt% Al2 O3 8% SiO 2 base slag over a temperature range of 1430 1600 C and slags with additives has been investigated quantitatively by reacting the CaO with the base slag and slags with additives, at various temperatures in air and fast quenching of the reactants. It was shown that within experimental scatter, by increasing the temperature and addition of CaF2 , FeO x , MnO x , TiO 2 , ilmenite and SiO 2 to the slag, the solubility of lime (about 60 wt%) remained almost constant.

5.3

Diffusivity of MgO / CaO in slags

The diffusivity of the studied solid oxides in the slags was quantified using the dimensionless mass transfer correlations for rotating disk/cylinder in the melt. The diffusivity values deduced using the data on dissolution rate and solubility, of solid oxides in the slag. As the developed correlations for calculation of diffusivity depend on the density and viscosity of slag, these quantities have been estimated by Urbains model for viscosity and Mills model for density of melt.

The diffusivity of MgO in the calcium aluminate slag at 1430C was found to be
1.45 10 5 cm2 /s and addition of 5 and 10 wt% FeO x increased the diffusivity by

a factor ~ 2 to 3, respectively. However, with (CaF2 5 wt% + Fe2 O3 5 wt%) and (CaF2 5 wt% + Fe2 O3 10 wt%) in the slag, the diffusivity was increased considerably by a factor of about 14 and 20, respectively.

The diffusivity of CaO in calcium aluminosilicate was measured at a temperature range of 1430 1600 C. The diffusivity was increased by temperature and additives in the slag. At 1430C, the diffusivity was found to be 9.2 10 6 cm2 /s 250

and by increasing the temperature to 1600 C, the diffusivity was increased to

3.7 10 5 cm2 /s. Addition of CaF2 had the strongest effect and increased the

diffusivity by a factor of 3 to 5 in the temperature range of 1500 to 1600 C. MnOx and FeO x had a comparable effect in increasing the diffusivity by a factor of 2~4. Ilmenite and TiO 2 also increased the diffusivity by a factor of about 2. However, addition of SiO 2 to the slag decreased the lime diffusivity. The influence of additives in the lime diffusivity was expressed according to the effect of various cations on the bonding environment of the silicate anions and their movement in the melt. Therefore, the strength between the interaction of cations and oxygen (from the anions) determines how freely the anions can move in the melt, which affects the diffusion of other species like Ca2+ in the melt.

The effect of temperature on the diffusivity of lime in the base slag and slags with additives were examined by calculation of activation energy of diffusion. The activation energy for diffusion of calcium in the base slag was about 43 kcal/mole. Addition of SiO 2 to the melt increased the activation energy to about 80 kcal/mole; however, other additives in the slag decreased the activation energy, with FeO x and CaF2 and MnO x had the strongest effects. This trend confirmed the mechanism of diffusion as the energy barrier for diffusion in the melt is the movement of silicate anions and any change in the activation energy is an indication of bonding environment between the cations and silicate anions.

The validity of Eyring theory, where the rate-determining step in the diffusing species is the movement of large species in the melt, has been investigated for all slags at various temperatures. It was shown that in master slag and the slag with

251

added

silica,

the

Eyrings

correlation

is

valid

(plot

of

Ln(Diffusivity/Temperature) versus. Ln (viscosity) having slope of 1), but when CaF2 and other oxides are added to the slag, the correlation between diffusivity and viscosity deviated from Eyring. These data showed the effect of additives on the interaction between cations and oxygen in the silicate anions.

The ionic conductivity of slag was estimated using Nerst-Einstein correlation. The changes in the ionic conductivity with temperature and additives follow the same trend as diffusivity, confirming that the same mechanism controls two quantities.

5.4

Recommendations for future work

The above findings show that although CaF2 had the strongest effect on the dissolution rate and diffusivity of lime in the slag but MnO x , FeO x and ilmenite in the slag increased the dissolution rate and diffusivity considerably. As these candidates should not cause emission of toxic species to the environment; thus the industry can consider these candidates as a substitute to fluorspar.

The present work recommends to steelmaking industry to reduce/eliminate the use of fluorspar, applying alternative additives such as MnO x , FeO x and ilmenite. It is also recommended to investigate the replacement ratio of ilmenite for the ladle type slag.

In the present work, the diffusivity of lime in the ladle type slag was investigated at various temperatures and with addition of additives. At 1430 C a solid phase

252

was formed as a coherent layer on the lime/slag interface. A mix-controlled model was developed which confirmed the existence of this layer. However, the effect of basicity on the diffusivity of lime in the slag was not studied. It was shown that when lime was reacted with slag of lower basicity, two non-coherent phases formed on the lime/slag interface where the diffusion in liquid and solid was not clear. An experimental technique is desirable to measure the diffusivity of lime in the slag systems with lower basicity.

Given diffusivity measurement is more difficult than electrical conductivity and we already have good models of electrical conductivity, the linkage between conductivity and diffusivity allows us to extend the application of conductivity models to diffusivity.

253

References
Agarwal, D. P. and Gaskell, D. R. (1975). "Self diffusion of Iron in Fe2 SiO 4 and CaFeSiO 4 melts." Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science. 6 (2): 263-267. Allen, N., Sun, S. and Jahanshahi, S. (1995). "Stability of MgO refractory in contact with iron-rich slags." Second Australian Melt Chemsirty Symposium, Melbourne, Australia, CSIRO : 55-58. Baak, T. (1958). "The action of calcium fluoride in slags." Conf. Phys. Chem. Iron and Steelmaking, Dedham, Massachusetts, Wiley: 84-86. Baisano v, S. O., Takenov, T. D., Gabdullin, T. G. and Buketov (1983). "Study of the properties of oxidic melts of the system iron(II), oxide-manganese(II), oxide-calcium, oxide silica alumina." Fizikokhim.Metall.Margantsa. B.N. Laskorin. Moscow. 58-62. Bale, C. W., Chartrand, P., Degterov, S. A., Eriksson, G., Hack, K., Ben Mahfoud, R., Melancon, J., Pelton, A. D. and Petersen, S. (2003). "FactSage." cole Polytechnique CRCT, Canada and GTT-Technologies GmbH, Germany. Bills, P. M. (1963). "Viscosities in silicate slag systems." Journal of The Iron and Steel Institute. 133-140. Blank, S. L. and Pask, J. A. (1969). "Diffusion of iron and nickel in magnesium oxide single crystals." Journal of the american ceramic society. 52(12): 669675. Bockris, J. O. M., Kitchener, J. A. and Davis, A. E. (1952). Trans. Farad. Soc.,. 48: 536-48. Bygden, J., DebRoy, T. and Seetharaman, S. (1994). "Dissolution of MgO in stagnant CaO-FeO-SiO2 slags." Ironmaking and Steelmaking. 21(4): 318323. Chilton, T. H. and Colburn, A. P. (1934). "Mass transfer (Absorption) coefficients - Prediction from data on heat transfer and fluid friction." Industrial and engineering chemsitry. 26(11): 1183-1187. Cochran, W. G. (1934). "The flow due to a rotating disc." Proceedings of the cambridge philosophical society. 30 : 365-74. Cohen, M. H. and Turnbull, D. (1959). J. Chem. Phys. 31. Cooper, A. R. and Kingery, W. D. (1964). "Dissolution in Ceramic Systems: I, molecular diffusion, natural convection, and forced convection studies of 254

sapphire dissolution in calcium silicate." Journal of the American Ceramic Society. 97(1): 37-43. Crank, J. (1975). "The mathematics of diffusion." Oxford University Press. Darken, L. S. and Gurry, R. W. (1946). "The system iron-oxygen. II.Equilibrium and thermodynamics of liquid oxide and other phases." J.Amer.Chem.Soc. 68: 798-815. Einstein, A. (1905). "Uber die von der molecularkinetischen theorie der warme geforderte beweguny von in ruhenden flussigkeiten suspendierten teilchen." Annales des physik . 17. Eisenberg, C. W. and Tobias, W. (1955). "Mass transfer at rotating cylinders." Chemical engineering progress symposium series. 51 (16): 1-16. Eisenhuttenleute, V. D. (1995). "Slag Atlas." 2nd ed. Verlag Stahleisen GmbH. Dusseldorf. Finn, C. W. P., Cripps, C. J. and McCarthy, M. J. (1973). "Developement of coated rotary lime as a substitute for fluorspar in basic steelmaking processes." Second national chemical engineering conference - The process industries in Australia, Qeensland - Australia: 88-98. Gammal, T. E. and Stracke, P. (1988). "Interfacial Phenomena of Molten Slags." 3rd International Conference on Molten Slags and Fluxes, Glascow, UK, The Institute of Metals : 207-214. Glasstone, S., Laidlev, K. J. and Eyring, H. (1941). "The theory of rate processes : the kinetics of che mical reactions, viscosity, diffusion and electrochemical phenomena." New York, McGraw-Hill. Goto, K. S., Kurahashi, T. and Sasabe, M. (1977). "Oygen pressure dependance of tracer diffusivities of Ca and Fe in liquid CaO - SiO 2 - FeO system." Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science. 8B : 523528. Grau, A. E. and Masson, C. R. (1976). Canadian metallurgical quaterly. 15 : 367. Hamano, T., Horibe, M. and Ito, K. (2004). "The Dissolution Rate of Solid Lime into Molten Slag Used for Hot- metal Dephosphorization." ISIJ International. 44(2): 263-267. Hara, S., Akao, K. and Ogino, K. (1989). "Diffusivity of Ca45 in Molten FeOCaO-SiO 2 slag equilibrated with solid iron." Tetsu-To-Hagane/Journal of the Iron and Steel Institute of Japan. 75(10): 1891-1896. Henderson, J., Yang, L. and Derge, G. (1961). "Self-diffusion of aluminium in CaO-SiO 2-Al2 O3 Melts." Transactions of the metallurgical society of AIME . 221 : 56-60.

255

Hu, H. and Reddy, R. G. (1988). "Modeling of viscosities of binary alkaline earth metal oxide and silicate melts." High Temperature Science. 28 : 195-202. Iida, T. and Kita, Y. (2002). JSPS, Rept. no. 11949. Iida, T., Sakai, H., Kita, Y. and Shigeno, K. (2000). "An equation for accurate prediction of the viscosities of blast furnace type slags from chemical composition." ISIJ International. 40 : 110-114. Johnson, R. F. (1970). "Aspects of diffusion in liquid slags." Met. J. No. 20: 337. Johnston, R. F., Stark, R. A. and Taylor, J. (1974). "Diffusion in liquid slags." Ironmaking Steelmaking. 1(4): 220-7. Kalmanovitch, D. P. and Williamson, J. (1984). "Crystallisation of Coal Ash Melts." Preprints of Papers - American Chemical Society, Division of Fuel Chemistry. 29(4): 162-165. Kalmanovitch, D. P. and Williamson, J. (1986). "Crystallization of Coal Ash Melts." ACS Symposium Series. 234-255. Keller, H. and Schwerdtfeger, K. (1979a). "Tracer diffusivity of Si31 in CaO-SiO 2 melts at 1600 C." Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science. 10 (4): 551-554. Keller, H. and Schwerdtfeger, K. (1986). "Measurement of Tracer Diffusivities of Ca45 and Fe59 in Silica Saturated Feo-CaO-SiO 2 Melts with the Porous Frit Technique." Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science. 17 (3): 497-501. Keller, H., Schwerdtfeger, K. and Hennesen, K. (1979b). "Tracer diffusivity of Ca45 and electrical conductivity in CaO - SiO 2 melts." Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science. 10(1): 67-70. Kondratiev, A. and Jak, E. (2001a). "Modeling of viscosities of the partly crystallized slags in the Al2 O3 -CaO-FeO-SiO 2 system." Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science. 32(6): 1027-1032. Kondratiev, A. and Jak, E. (2001b). "Review of experimental data and modeling of the viscosities of fully liquid slags in the Al2 O3 -CaO-FeO-SiO 2 system." Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science. 32 (6): 1015-1025. Kor, G. J. W. (1977). "Effect of Fluorspar and Other Fluxes on Slag-Metal Equilibria Involving Phosphorus and Sulfur." Metall Trans B. 8(1): 107113.

256

Kor, G. J. W., Martonik, L. J. and Miller, R. A. (1986). "Evaluation of lime for the BOF process and effect of degree of calcination on its dissolution rate in slags." Fifth International Iron and Steel Congeress, Washington: 679689. Kosaka, M., Machida, M. and Hirai, Y. (1969). "On the Dissolution Process of Solid Al into Molten Al-Si Alloy." J Jap Inst Metals. 33 (4): 465-70. Kosaka, M. and Minowa, S. (1966). "Mass transfer from solid metal cylinder into liquid metal." Tetsu-To-Hagane/Journal of the Iron and Steel Institute of Japan. 52(12): 22-36. Kozakevitch, P. (1951). Rev. Met. 51 : 569. Kubicek, P. and Peprica, T. (1983). "Diffusion in molten metals and melts: application to diffusion in moltem iron." International metals reviews. 23(3): 131-157. Lee, M., Sun, S., Wright, S. and Jahanshahi, S. (2001). "Effects of transition metals on the kinetics of slag-refractory reactions." Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science. 32B(1): 25-29. Lee, Y. E. and Gaskell, D. R. (1974). "Densities and structure of melts in the system CaO- "FeO"-SiO 2 ." Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science. 5 : 853. Levich, V. G. (1962). "Physiochemical Hydrodynamics." Prentice-Hall. MacLean, J. R., Kingston, P. W., MacDonald, J. B. and Caley, W. F. (1997). "Potential role of feldspar/feldspathoid minerals in secondary steelmaking." Ironmaking Steelmaking. 24(5): 406-411. MATLAB (2000). "The Math Works, Inc." Matsushima, M., Yadoomaru, S., Mori, K. and Kawai, Y. (1977). "A fundament al study on the dissolution rate of solid lime into liquid slag." Trans. Iron Steel Inst. Jpn. 17(8): 442-9. Michel, J. R. and Mitchell, A. (1975). "A study of the rheological behavior of some slags in the system CaO+SiO 2 +Al2 O3 +CaF2 ." Canadian Metallurgical Quarterly. 14 (2): 153-159. Mills, K. C. (1977). "Part (3), Review of viscosities of CaF2 -based slags." The Physiochemical Properties of Slags. National Physical Laboratory: 1-22. Mills, K. C. and Keene, B. J. (1981). "Physiochemical properties of molten CaF2 - based slags." International Metals Reviews. (1): 21-69. Mills, K. C. and Keene, B. J. (1987). "Physical properties of BOS slags." International Materials reviews. 32(1-2): 1-120. 257

Mills, K. C. and Sridhar, S. (1999). "Viscosities of ironmaking and steelmaking slags." Ironmaking and Steelmaking. 26(4): 262-268. Mori, K. and Suzuki, K. (1969). "Diffusion in Iron oxide melts." Tranaction of the iron and steel institute. 9 : 409-12. Nadyrbekov, A. K., Akberdin, I. S., Kulikov, I. S. and Kim, V. A. (1980). Deposted Doc.VINITI. 26. Nagata, K., Sata, N. and Goto, K. (1982). "Diffusivities in molten slag, molten iron, steel and refractories." Tetsu-to-Hagane. 68 : 20-31. Natalie, C. A. and Evans, J. W. (1979). "Influence of lime properties on rate of dissolutio n in CaO-SiO2-FeO slags." Ironmaking and steelmaking. 6(3): 101-109. Nesterenko, S. V. and Khomenko, V. M. (1985). "Study of the effects of alkalis on the surface tension and the electrical conductivity of slags of the CaOMgO-SiO 2 system containing 5% Al2 O3 ." Russian Metallurgy (Metally) (English translation of Izvestiya Akademii Nauk SSR, Metally). (2): 4245. Noguchi, F., Ueda, Y. and Yanagase, T. (1976). "The mechanism of calcium oxide dissolution into slag melts." Jt. MMIJ-AIME Meet. proceedings World Min. Met. Technol. : 685-92. Nowak, R. and Schwerdtfeger, K. (1975). "Determination of mobilities of Fe2+, Co2+, Ni2+, and Ca2+ in silicate melts at 1600 C by a galvanostatic technique." Metal - slag - gas reactions and processes. Z.A. ForoulisW.W. Smeltzer. Princeton, NJ, The Electrochemical Society. 98110. Oeters, F. and Scheel, R. (1971). "Investigation on the process of lime dissolution." Arch Eisenhuettenwes. 42(11): 769-777. Pierre, P. D. S. S. (1954). Journal of American Ceramic Society. 37(6): 243-258. Pierre, P. D. S. S. (1956). Journal of American Ceramic Society. 39(4): 147-150. Poggi, D. and Lee, H. Y. (1974). "Massive ilmenite as a slag thinner in steelmaking." Canadian Metallurgical Quarterly. 13 (3): 529-33. Poirier, D. R. and Geiger, G. H. (1998). "Transport Phenomena in Materials Processing." Minerals, Metals, & Materials Society. Riboud, P. V., Roux, Y., Lucas, D. and Gaye, H. (1981). Fachber. Huttenprax. Metallweiterverarb. 19 : 859-869. Richardson, F. D. (1974). "Physical Chemsitry of Melts in Metallurgy." Academic press.

258

Saito, T. and Maruya, K. (1958). "Diffusion of Calcium in Liquid Slags." Series A, Physics, Chemistry and Metallurgy. Tohoko University Science Report of the Research Institute. Tohoko University: 306-314. Sandhage, K. H. and Yurek, G. J. (1988). "Indirect Dissolution of Sapphire into Silicate Melts." Journal of the American Ceramic Society. 71(6): 478489. Sandhage, K. H. and Yurek, G. J. (1990). "Direct and Indirect Dissolution of Sapphire in Calcia-Magnesia-Alumina-Silica Melts - Dissolution Kinetics." Journal of the American Ceramic Society. 73(12): 3633-3642. Sarkar, S. B. (1989). "Electrical conductivity of molten high-alumina blast furnace slags." ISIJ international. 29(4): 348-351. Sarkar, S. B. and Sen, P. K. (1978). Trans Ind. Inst. Met. 2 : 42-45. Satyoko, Y. and Lee, W. E. (1999). "Dissolution of dolomite and doloma in silicate slag." British Ceramic Transactions. 98(6): 261-265. Satyoko, Y., Lee, W. E., Parry, E., Richards, P. and Houldsworth, I. G. (2003). "Dissolution of iron oxide containing doloma in model basic oxygen furnace slag." Ironmaking and Steelmaking. 30(3): 203-208. Seetharaman, S., Mukai, K. and Sichen, D. (2004). "Viscosities of slags - an overview." 7th International Molten Slags, Fluxes and Salts, Cape Town, South Africa : 31-41. Seetharaman, S. and Sichen, D. (1994). "Estimation of the viscosities of binary metallic melts using Gibbs energies of mixing." Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science. 25(4): 589-595. Shimizu, K. and Cramb, A. W. (2002). "The kinetics of fluoride evaporation from CaF2 -SiO 2 -CaO slags and mold fluxes in dry atmospheres." Iron & Steelmaker. 29(6): 43-53. Shimizu, K., Suzuki, T., Jimbo, I. and Cramb, A. W. (1996). "An inve stigation on the vaporization of fluorides from slag melts." Seventy Ninth Conference of the Steelmaking Division of the Iron and Steel Society, Pittsburgh, Pennsylvania; USA, Iron and Steel Society/AIME: 727-733. Sichen, D., Bygden, J. and Seetharaman, S. (1994). "Model for estimation of viscosities of complex metallic and ionic melts." Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science. 25(4): 519-525. Simnad, M. T., Yang, L. and Derge, G. (1956). "Direct measurement of ferrousion mobility in liquid iron silicate by a radioactive-tracer technique." Journal of Metals, AIME Trans. 206 (8): 690-92. 259

Singh, B. N., Ravat, Y. F., Chatterjee, A. and Chakravarty, P. K. (1977). "Use of ilmenite sand as a substitute for fluorspar in open-hearth steelmaking." Ironmaking Steelmaking. 4(3): 170-5. Sun, S., Zhang, L. and Jahanshahi, S. (2003). "From Viscosity and Surface Tension to Marangoni Flow in Melts." Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science. 34(5): 517-523. Taira, S., Nakashima, K. and Mori, K. (1993). "Kinetic behavior of dissolution of sintered alumina into CaO-SiO 2 -Al2 O3 slags." ISIJ International. 33 (1): 116-123. Takenov, T. D. (1987). Kompleksn. Ispolz. Miner. Syrya. 10 : 55-59. Taylor, J. R. (1982). "An introduction to error analysis." CAlifornia, University Science Books. Towers, H. and Chipman, J. (1953). "Diffusion of calcium ion in liquid slag." Journal of Metals. 1455-1458. Towers, H. and Chipman, J. (1957). "Diffusion of calcium and silicon in a limealumina-silica slag." Journal of Metals. 769-773. Tribe, T. S., Kingston, P. W. and Caley, W. F. (1997). "Rheology and constitution of the CaO-SiO 2 -MgO-CaF2 system." Canadian Metallurgical Quaterly. 36(2): 95-101. Tribe, T. S., Kingston, P. W., MacDonald, J. B. and Caley, W. F. (1994). "Reduction of fluorspar consumption in secondary steelmaking." Ironmaking and Steelmaking. 21(2): 145-9. Turkdogan, E. T. (1983). "Physiochemical properties of molten slags and glasses." The Metals Society. Turkdogan, E. T. (1985). "Slags and fluxes for ferrous ladle metallurgy." Ironmaking and Steelmaking. 12(2): 64-78. Turkdogan, E. T. and Bills, P. M. (1960). "A critical reveiw of viscosity of CaOMgO-Al2 O3 -SiO 2 melts." American Ceramic Society Bulletin. 39 : 682687. Ukyo, Y. and Goto, S. (1982). "Measurement of Quasi-binary interdiffusivities of various oxides in liquid slags." Tetsu-to-Hagane. 68 : 117-122. Umakoshi, M., Mori, K. and Kawai, Y. (1981). "Dissolution rate of sintered MgO into molten Fet O-CaO-SiO 2 slags." Tetsu-To-Hagane/Journal of the Iron and Steel Institute of Japan. 67(10): 1726-1734.

260

Umakoshi, M., Mori, K. and Kawai, Y. (1984a). "Corrosion kinetics of refractory materials in molten CaO-FeO-SiO 2 slags." Second international symposium on metallurgical slags and fluxes, Nevada, USA, Umakoshi, M., Mori, K. and Kawai, Y. (1984b). "Dissolution rate of burnt dolomite in molten iron oxide-calcium oxide-silicon dioxide (Fet O-CaOSiO 2 ) slags." Trans. Iron Steel Inst. Jpn. 24(7): 532-9. Urbain, G. (1984). "Viscosite de liquides de systeme SiO 2 -PbO." Revue Internationale des Hautes Temperatures et des Refractaires. 21 (2): 107111. Urbain, G. (1987). "Viscosity estimation of slags." Steel Research. 58(3): 111116. Urbain, G. and Boiret, M. (1990). "Viscosities of liquid silicates." Ironmaking and Steelmaking. 17(4): 255-260. Urbain, G., Cambier, F., Deletter, M. and Anseau, M. R. (1981). "Viscosity of Silicate Melts." Transactions and Journal of the British Ceramic Society. 80(4): 139-141. Walls, H. A. and Upthegrove, W. R. (1964). "Theory of liquid diffusion phenomena." Acta Metallurgica. 12 : 461-471. Wejnarth, A. (1934). Trans. Amer. Electrochem. Soc. 65 : 1386 - 1390. Williams, P., Sunderland, M. and Briggs, G. (1982). "Interaction of between dolomit lime and iron silicate melts." Ironmaking and steelmaking. 9(4): 150-162. Winterhager, H., Greiner, L. and Kammel, R. (1966). "Untersuchungen uber die dichte und die electrische leitfahigkeit von schmelzen der systeme CaOAl2 O3 -SiO 2 ." Forschungsber Landes Nordrhein-Westfalen. Wright, S., Zhang, L., Sun, S. and Jahanshahi, S. (2000). "Viscosity of a CaOMgO-Al2 O3 -SiO 2 melt containing spinel particles at 1646 K." Metallurgical and Materials Transactions A: Physical Metallurgy and Materials Science. 31B : 97-103. Wright, S., Zhang, L., Sun, S. and Jahanshahi, S. (2001). "Viscosities of calcium ferrite slags and calcium alumino-silicate slags containing spinel particles." Journal of Non-Crystalline Solids. 282 (1): 15-23. Xie, D. and Belton, G. R. (1999). "Rate of reaction of solid iron with oxidized FeO-CaO-SiO 2 -Al2 O3 slags at 1360C - the chemical diffusivity of iron oxide." Metallurgical and Materials Transactions A: Physical Metallurgy and Materials Science. 30(3): 465-472. Yang, L., Chien, C.-Y. and Derge, G. (1959). "Self-diffusion of iron in iron silicate melt." J. Chem. Phys. 30 : 1627. 261

Yu, X., Pomfret, R. J. and Coley, K. S. (1997). "Dissolution of alumina in mold fluxes." Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science. 28 (2): 275-279. Zhang, L. and Jahanshahi, S. (1998a). "Review and modeling of viscosity of silicate melts: Part I. Viscosity of binary and ternary silicates containing CaO, MgO, and MnO." Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science. 29(1): 177-186. Zhang, L. and Jahanshahi, S. (1998b). "Review and modeling of viscosity of silicate melts: Part II. Viscosity of melts containing iron oxide in the CaO-MgO-MnO-FeO-Fe2 O3-SiO 2 system." Metallurgical and Materials Transactions B: Process Metallurgy and Materials Processing Science. 29(1): 187-195. Zhang, L., Jahanshahi, S., Sun, S., Chen, C., Borke, B., Wright, S. and Somerville, A. (2002). "CSIRO's multiphase reaction models and their industrial applications." JOM-Journal of the minerals metals and materials society. 54(11): 51-56. Zhang, P. and Seetharaman, S. (1994). "Dissolution of MgO in CaO-FeO-CaF2SiO 2 slags under static conditions." Journal of American Ceramic Society. 77(4): 970-976.

262

APPENDIX A. Solid oxides dissolution data

A.1 The effect of rotation speed on dissolution of CaO in calcium aluminosilicate slag at 1430 C
The effect of variation of rotation speed on the concentration of lime in the slag with time is presented in this section. The experimental data analysed by XRF along with fitted curve are presented in Figure A. 1 to Figure A. 4 and Table A. 1 to Table A. 4. There is an error of 0.2-0.3 wt% (absolute) from the XRF analysis.

The Curve Fitting Toolbox in MATLAB uses the method of least squares when fitting data. The fitting process requires a model that relates the response data to the predictor data with one or more coefficients. The result of the fitting process is an estimate of the true but unknown coefficients of the model. The concentration versus time data for the first 10 minutes of experiment is given to the Curve Fitting Toolbox, then the linear least squares method is applied to fit a linear model to data.

263

Figure A. 1: Concentration of CaO dissolved in slag at 30 rpm and at 1430 C Table A. 1: XRF analysis of the bulk slag when lime dissolves in slag in air at 30 rpm and 1430 C
Reaction time (min) 0 5 10 15 20 25 30 35 40 45 50 55 60 XRF analysis of bulk slag (wt%) SiO2 8.03 7.90 7.92 7.81 7.89 7.87 7.87 7.84 7.8 7.84 7.8 7.79 7.78 Al2O3 41.9 41.9 42.2 41.6 42.2 41.8 42.0 41.7 41.7 41.7 41.7 41.6 41.5 Fe 2O3 0.315 0.211 0.157 0.293 0.126 0.203 0.137 0.137 0.1 0.131 0.1 0.133 0.136 MgO 0.09 0.08 0.07 0.06 0.06 0.05 0.07 0.06 0.08 0.08 0.09 0.09 0.10 CaO 48.5 49.2 49.2 48.7 49.1 49.6 49.4 49.3 49.5 49.6 49.6 49.5 49.9 CaO from curve fitting 49.60 50.21 50.27 49.78 50.16 50.67 50.45 50.32 50.50 50.69 50.62 50.56 50.99

264

Figure A. 2: Concentration of CaO dissolved in slag at 60 rpm and at 1430 C Table A. 2: XRF analysis of the bulk slag when lime dissolves in slag in air at 60 rpm and 1430 C
Reaction time (min) 0 5 10 15 20 25 30 35 40 45 50 55 60 XRF analysis of bulk slag (wt%) SiO2 7.98 7.90 7.95 7.89 7.82 7.83 7.85 7.82 7.82 7.79 7.77 7.77 7.71 Al2O3 42.1 42.1 42.2 42.1 41.6 41.8 42.0 41.8 41.5 41.4 41.3 41.3 41.2 Fe 2O3 0.240 0.171 0.143 0.186 0.160 0.132 0.133 0.173 0.168 0.129 0.156 0.120 0.130 MgO 0.07 0.06 0.06 0.07 0.07 0.08 0.08 0.08 0.08 0.10 0.10 0.10 0.12 CaO 48.9 48.9 49.2 49.3 49.4 49.5 49.6 49.6 49.6 49.7 49.8 49.9 49.9 CaO from curve fitting 48.90 49.06 49.19 49.31 49.41 49.50 49.58 49.65 49.71 49.77 49.82 49.87 49.91

265

Figure A. 3: Concentration of CaO dissolved in slag at 90 rpm and at 1430 C Table A. 3: XRF analysis of the bulk slag when lime dissolves in slag in air at 90 rpm and 1430 C
Reaction time (min) 0 5 10 15 20 25 30 35 40 45 50 55 60 XRF analysis of bulk slag (wt%) SiO2 8.06 7.91 7.83 7.83 7.74 7.77 7.74 7.74 7.71 7.80 7.72 7.70 7.69 Al2O3 42.0 41.9 41.7 41.6 41.3 41.3 41.0 41.1 40.8 41.2 40.8 40.6 40.7 Fe 2O3 0.310 0.178 0.200 0.152 0.143 0.152 0.136 0.121 0.126 0.149 0.187 0.128 0.130 MgO 0.09 0.08 0.09 0.10 0.11 0.12 0.11 0.10 0.13 0.14 0.13 0.14 0.15 CaO 49.8 50.5 51.0 51.1 51.0 51.3 51.4 51.6 51.4 51.6 51.8 51.7 51.9 CaO from curve fitting 49.83 50.31 50.65 50.91 51.11 51.27 51.40 51.51 51.60 51.68 51.75 51.81 51.86

266

Figure A. 4: Concentration of CaO dissolved in slag at 120 rpm and at 1430 C Table A. 4: XRF analysis of the bulk slag when lime dissolves in slag in air at 120 rpm and 1430 C
Reaction time (min) 0 2 4 6 8 10 20 30 40 50 60 XRF analysis of bulk slag (wt%) SiO2 7.91 7.68 7.75 7.65 7.63 7.66 7.62 7.57 7.55 7.53 7.50 Al2O3 42.8 42.1 42.3 41.9 41.7 42.0 41.7 41.3 41.2 41.2 40.9 Fe 2O3 0.231 0.155 0.152 0.168 0.152 0.143 0.135 0.132 0.139 0.129 0.142 MgO 0.02 0.03 0.02 0.01 < DL 0.04 0.04 0.05 0.07 0.08 0.09 CaO 50.2 50.4 50.5 50.5 50.7 51.0 50.1 51.2 51.5 51.8 51.8 CaO from curve fitting 50.19 50.40 50.55 50.67 50.76 50.84 51.14 51.35 51.53 51.69 51.85

267

Figure A. 5: Concentration of CaO dissolved in slag at 150 rpm and in air at 1430 C Table A. 5: XRF analysis of the bulk slag when lime dissolves in slag in air at 150 rpm and 1430 C
Reaction time (min) 0 2 4 6 8 10 20 30 40 50 60 XRF analysis of bulk slag (wt%) SiO2 7.95 7.81 7.90 7.78 7.84 7.83 7.68 7.58 7.69 7.61 7.55 Al2O3 42.6 42.5 43.0 42.8 42.9 43.0 42.0 41.4 41.6 41.5 41.0 Fe 2O3 0.314 0.201 0.178 0.156 0.170 0.167 0.157 0.172 0.172 0.156 0.154 MgO 0.02 0.01 0.02 0.01 0.01 0.03 0.04 0.05 0.07 0.07 0.10 CaO 50.0 50.4 50.5 50.3 50.8 51.1 50.9 51.3 51.7 52.0 52.0 CaO from curve fitting 50.00 50.37 50.55 50.67 50.77 50.84 51.15 51.40 51.64 51.87 52.10

268

A.2 Effect of CaF 2 addition on dissolution of CaO in calcium aluminosilicate slag at various temperatures
The effect of 5 wt% addition of CaF2 on the variation of concentration of lime in the slag with time is presented in this section. The experimental data obtained by XRF along with the fitted curve are presented in Figure A. 6 to Figure A. 9 and Table A. 6 to Table A. 9. There is an error of 0.2 0.3 wt% (absolute) from the XRF ana lysis. The dissolution rate was obtained from the slope of the dissolution curves, by fitting a straight line through the initial experimental concentration data using MATLAB.

269

Figure A. 6: Concentration of CaO dissolved in slag with 5 wt% CaF2 at 90 rpm and in air at 1430 C Table A. 6: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% CaF2 in air at 90 rpm and 1430 C
Reaction time (min) 0.00 1.33 2.67 4.00 5.33 6.67 8.00 9.33 10.66 20 XRF analysis of bulk slag (wt%) SiO2 8.09 7.76 7.27 7.07 6.91 6.94 7.05 6.97 7.16 6.85 Al2O3 39.8 38.0 37.1 36.8 36.4 36.1 36.2 35.7 36.1 35.2 MgO 0.05 0.06 0.22 0.30 0.49 0.33 0.33 0.36 0.31 0.36 CaF2 4.82 4.13 4.23 4.33 4.31 4.45 4.56 4.45 4.97 4.50 CaO 47.8 50.7 52.1 52.3 53.0 52.9 53.0 53.4 52.6 53.5 CaO from curve fitting 47.78 50.86 51.92 52.46 52.77 52.97 53.11 53.20 53.27 53.43

270

Figure A. 7: Concentration of CaO dissolved in slag with 5 wt% CaF2 at 90 rpm and in air at 1500 C Table A. 7: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% CaF2 in air at 90 rpm and 1500 C
Reaction time (min) 0.00 1.33 2.67 4.00 5.33 6.67 8.00 9.33 10.66 20 30 XRF analysis of bulk slag (wt%) SiO2 7.92 7.70 7.64 7.64 7.63 7.54 7.56 7.54 7.49 7.41 7.40 Al2O3 40.3 39.7 39.7 39.5 39.5 39.1 39.3 39.3 39.2 38.2 38.3 MgO 0.04 0.07 0.08 0.08 0.09 0.10 0.11 0.11 0.13 0.18 0.21 CaF2 5.15 4.74 5.15 4.82 5.01 4.89 5.03 4.84 4.80 4.54 5.09 CaO 47.4 48.2 48.4 48.8 48.9 49.1 49.2 49.3 49.7 50.1 50.2 CaO from curve fitting 47.49 48.06 48.45 48.74 48.97 49.15 49.29 49.42 49.52 49.99 50.24

271

Figure A. 8: Concentration of CaO dissolved in slag with 5 wt% CaF2 at 90 rpm and in air at 1550 C Table A. 8: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% CaF2 in air at 90 rpm and 1550 C
Reaction time (min) 0.00 1.33 2.67 4.00 5.33 6.67 8.00 9.33 10.66 20 XRF analysis of bulk slag (wt%) SiO2 7.97 7.81 7.70 7.58 7.46 7.57 7.56 7.52 7.54 7.50 Al2O3 40.3 39.6 39.7 39.4 38.8 39.1 39.0 39.0 38.9 38.9 MgO 0.06 0.04 0.07 0.04 0.13 0.13 0.11 0.13 0.11 0.12 CaF2 5.07 4.99 4.72 5.23 4.91 4.84 4.91 4.95 4.91 4.80 CaO 47.4 48.1 48.6 48.4 48.9 49.2 49.2 49.2 49.2 49.7 CaO from curve fitting 47.45 48.10 48.53 48.82 49.03 49.18 49.30 49.40 49.47 49.69

272

Figure A. 9: Concentration of CaO dissolved in slag with 5 wt% CaF2 at 90 rpm and in air at 1600 C for 1 hour

Table A. 9: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% CaF 2 in air at 90 rpm and 1600 C
Reaction time (min) 0.00 1.33 2.67 4.00 5.33 6.67 8.00 9.33 10.66 20 XRF analysis of bulk slag (wt%) SiO2 7.83 7.74 7.67 7.74 7.70 7.57 7.52 7.46 7.47 7.39 Al2O3 40.4 40.2 40.0 40.0 40.0 39.5 39.2 38.8 38.8 38.5 MgO 0.05 0.02 0.06 0.04 0.07 0.10 0.11 0.13 0.13 0.19 CaF2 4.58 4.52 4.43 4.82 4.52 4.50 4.08 4.25 4.25 4.35 CaO 47.5 48.1 48.8 48.3 48.9 49.2 49.7 49.9 49.9 50.4 CaO from curve fitting 47.45 48.13 48.63 49.00 49.30 49.53 49.71 49.86 49.99 50.42

273

A.3 Effect of Fe2O3 addition on dissolution of CaO in calcium aluminosilicate slag at various temperatures
The effect of 5 wt% addition of Fe2 O3 on the variation of concentration of lime in the slag with time is presented in this section. The experimental data analysed by XRF along with the fitted curve are presented in Figure A. 10 to Figure A. 13 and Table A. 10 to Table A. 13. There is an error of 0.2 0.3 wt% (absolute) from the XRF analysis.

274

Figure A. 10: Concentration of CaO dissolved in slag with 5 wt% Fe2 O3 90 rpm and in air at 1430 C for 1 hour

Table A. 10: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% Fe2 O3 in air at 90 rpm and 1430C
Reaction time (min) 0 2 4 6 8 10 20 30 40 50 60 XRF analysis of bulk slag (wt%) SiO2 7.52 7.43 7.37 7.39 7.30 7.34 7.22 7.21 7.10 7.10 7.03 Al2O3 40.3 40.3 40.2 40.0 39.8 40.0 39.5 39.0 38.5 38.4 38.0 Fe2O3 5.09 5.01 5.00 5.00 4.91 4.98 4.90 4.86 4.76 4.77 4.72 MgO 0.03 0.06 0.04 0.04 0.06 0.06 0.08 0.12 0.14 0.17 0.18 CaO 47.3 47.9 48.2 48.5 48.3 48.8 49.3 49.8 49.9 50.6 50.9 CaO from curve fitting 47.20 47.90 48.24 48.47 48.63 48.77 49.29 49.72 50.12 50.50 50.88

275

Figure A. 11: Concentration of CaO dissolved in slag with 5 wt% Fe2 O3 at 90 rpm and in air at 1500 C for 1 hour Table A. 11: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% Fe2 O3 in air at 90 rpm and 1500 C
Reaction time (min) 0 2 4 6 8 10 20 30 40 50 60 XRF analysis of bulk slag (wt%) SiO2 7.66 7.60 7.40 7.40 7.31 7.39 7.17 7.21 7.19 7.15 7.28 Al2O3 40.4 40.1 39.8 39.6 39.2 39.7 38.5 38.5 38.3 38.1 37.3 Fe2O3 5.81 5.14 5.04 4.95 4.88 4.93 4.79 4.99 4.92 4.71 4.62 MgO 0.05 0.03 0.10 0.14 0.14 0.11 0.19 0.15 0.20 0.22 0.29 CaO 46.8 48.0 47.1 48.3 49.2 48.8 50.0 49.9 50.4 50.6 51.1 CaO from curve fitting 46.83 47.86 48.39 48.72 48.96 49.14 49.72 50.10 50.42 50.71 50.98

276

Figure A. 12: Concentration of CaO dissolved in slag with 5 wt% Fe2 O3 at 90 rpm and in air at 1550 C for 1 hour Table A. 12: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% Fe2 O3 in air at 90 rpm and 1550 C
Reaction time (min) 0 2 4 6 8 10 20 30 40 50 60 XRF analysis of bulk slag (wt%) SiO2 7.61 7.39 7.42 7.46 7.39 7.34 7.23 7.31 7.07 7.00 6.98 Al2O3 40.65 39.47 39.65 40.27 39.48 39.33 38.83 39.05 37.68 37.15 36.93 Fe2O3 5.13 4.95 4.92 5.05 4.94 4.92 4.83 4.84 4.73 4.57 4.54 MgO 0.05 0.14 0.12 0.03 0.11 0.15 0.15 0.15 0.16 0.25 0.28 CaO 47.2 49.0 49.0 48.5 48.9 49.6 50.3 49.9 50.7 50.8 50.7 CaO from curve fitting 47.26 48.09 48.65 49.04 49.34 49.57 50.21 50.50 50.65 50.74 50.79

277

Figure A. 13: Concentration of CaO dissolved in slag with 5 wt% Fe2 O3 at 90 rpm and in air at 1600 C for 1 hour Table A. 13: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% Fe2 O3 in air at 90 rpm and 1600 C
Reaction time (min) 0 2 4 6 8 10 20 30 40 50 60 XRF analysis of bulk slag (wt%) SiO2 7.82 7.53 7.43 7.33 7.24 7.21 7.10 6.92 6.87 6.73 6.73 Al2O3 40.7 40.1 39.8 39.4 39.0 38.7 38.0 36.8 36.5 35.7 35.7 Fe2O3 5.35 5.21 5.02 4.97 4.94 4.89 4.95 4.66 4.66 4.51 4.51 MgO 0.02 0.01 0.10 0.13 0.15 0.19 0.22 0.29 0.26 0.35 0.36 CaO 47.2 48.3 48.6 49.2 49.5 50.4 50.7 51.9 52.8 53.2 53.4 CaO from curve fitting 47.33 48.08 48.66 49.13 49.52 49.85 51.04 51.85 52.51 53.08 53.61

278

A.4 Effect of TiO2 addition on dissolution of CaO in calcium aluminosilicate slag at various temperatures
The effect of 5 wt% addition of TiO 2 on the variation of concentration of lime in the slag with time is presented in this section. The experimental data analysed by XRF along with the fitted curve are presented in Figure A. 14 to Figure A. 17 and Table A. 14 to Table A. 17. There is an error of 0.2 0.3 wt% (absolute) from the XRF analysis.

279

Figure A. 14: Concentration of CaO dissolved in slag with 5 wt% TiO2 at 90 rpm and in air at 1430 C for 1 hour Table A. 14: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% TiO2 in air at 90 rpm and 1430 C
Reaction time (min) 0 2 4 6 8 10 20 30 40 50 XRF analysis of bulk slag (wt%) SiO2 7.76 7.66 7.68 7.63 7.64 7.60 7.61 7.51 7.39 7.28 Al2O3 40.3 39.8 40.0 39.9 39.9 39.7 39.5 39.0 38.4 37.8 TiO2 4.93 4.89 4.92 4.86 4.90 4.86 4.84 4.81 4.75 4.67 MgO 0.05 0.06 0.01 0.07 0.03 0.04 0.10 0.09 0.16 0.20 CaO 47.5 48.0 47.7 48.3 47.9 48.3 48.9 49.4 49.7 50.5 CaO from curve fitting 47.55 47.69 47.83 47.97 48.10 48.23 48.84 49.40 49.91 50.37

280

Figure A. 15: Concentration of CaO dissolved in slag with 5 wt% TiO2 at 90 rpm and in air at 1500 C for 1 hour Table A. 15: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% TiO2 in air at 90 rpm and 1500 C
Reaction time (min) 0 1.5 3 4.5 6 7.5 9 10.5 20 30 40 50 60 XRF analysis of bulk slag (wt%) SiO2 7.81 7.64 7.60 7.51 7.39 7.48 7.45 7.36 7.36 7.34 7.30 7.14 7.14 Al2O3 40.1 39.9 39.6 39.0 38.3 38.7 39.0 38.4 38.1 38.4 37.7 36.7 36.7 TiO2 4.99 5.03 4.90 4.84 4.69 4.80 4.79 4.74 4.70 4.72 4.60 4.51 4.51 MgO < DL 0.08 0.08 0.08 0.14 0.10 0.09 0.16 0.16 0.14 0.21 0.24 0.26 CaO 47.4 48.1 48.5 49.4 49.0 49.3 49.0 49.6 50.0 49.8 50.4 51.2 51.7 CaO from curve fitting 47.39 48.14 48.48 48.70 48.86 48.99 49.11 49.21 49.73 50.26 50.77 51.19 51.68

281

Figure A. 16: Concentration of CaO dissolved in slag with 5 wt% TiO2 at 90 rpm and in air at 1550 C for 1 hour Table A. 16: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% TiO2 in air at 90 rpm and 1550 C
Reaction time (min) 0.0 1.5 3.0 4.5 6.0 7.5 9.0 10.5 20.0 30.0 40.0 50.0 60.0 XRF analysis of bulk slag (wt%) SiO2 7.81 7.68 7.65 7.57 7.55 7.52 7.47 7.45 7.32 7.27 7.13 7.11 7.04 Al2O3 40.2 39.8 39.5 39.2 39.1 39.1 38.9 38.6 37.8 37.4 36.9 36.5 36.1 TiO2 4.97 4.92 4.89 4.83 4.88 4.80 4.82 4.74 4.64 4.64 4.57 4.46 4.43 MgO 0.02 0.06 0.07 0.03 0.08 0.12 0.13 0.09 0.17 0.20 0.21 0.26 0.29 CaO 47.7 48.4 48.7 48.5 48.7 49.2 49.3 49.5 50.3 51.2 51.6 51.9 52.1 CaO from curve fitting 47.94 48.23 48.50 48.74 48.97 49.18 49.37 49.55 50.41 51.10 51.57 51.87 52.14

282

Figure A. 17: Concentration of CaO dissolved in slag with 5 wt% TiO2 at 90 rpm and in air at 1570 C for 1 hour

Table A. 17: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% TiO2 in air at 90 rpm and 1570 C
Reaction time (min) 0 2 4 6 8 10 20 30 40 50 60 XRF analysis of bulk slag (wt%) SiO2 7.80 7.58 7.60 7.45 7.47 7.43 7.39 7.20 7.05 6.93 6.82 Al2O3 40.4 39.4 39.4 38.7 38.8 38.5 38.3 37.1 36.1 35.7 35.1 TiO2 4.93 4.86 4.92 4.76 4.80 4.77 4.69 4.58 4.47 4.36 4.26 MgO 0.06 0.10 0.06 0.12 0.12 0.13 0.16 0.23 0.26 0.34 0.41 CaO 47.4 48.6 48.6 49.3 49.3 49.9 50.0 51.7 52.6 52.8 53.2 CaO from curve fitting 47.52 48.33 48.92 49.37 49.73 50.04 51.08 51.77 52.34 52.83 53.29

283

A.5 Effect of ilmenite addition on dissolution of CaO in calcium auminosilicate slag at various temperatures
The effect of 5 wt% addition of ilmenite on the variation of concentration of lime in the slag versus time is presented in this section. The experimental data analysed by XRF along with the fitted curve are presented in Figure A. 18 to Figure A. 21 and Table A. 18 to Table A. 21. There is an error of 0.2 0.3 wt% (absolute) from the XRF analysis.

284

Figure A. 18: Concentration of CaO dissolved in slag with 5 wt% ilmenite at 90 rpm and in air at 1500 C

Table A. 18: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% ilmenite in air at 90 rpm and 1500 C
Reaction time (min) 0 1.333 2.666 3.999 5.332 6.665 7.998 9.331 XRF analysis of bulk slag (wt%) SiO2 8.01 7.82 7.74 7.75 7.72 7.65 7.67 7.58 Al2 O3 40.0 40.0 40.1 39.8 39.8 39.4 39.6 39.1 Fe 2 O3 2.16 2.13 2.13 2.12 2.10 2.09 2.10 2.07 TiO2 2.82 2.81 2.80 2.79 2.77 2.75 2.76 2.73 MgO < DL 0.04 0.05 0.07 0.08 0.08 0.09 0.12 CaO 47.3 47.3 47.6 47.9 48.1 48.0 48.4 48.9 CaO from curve fitting 47.29 47.48 47.67 47.86 48.05 48.24 48.43 48.62

285

Figure A. 19: Concentration of CaO dissolved in slag with 5 wt% ilmenite at 90 rpm and in air at 1550 C

Table A. 19: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% ilmenite in air at 90 rpm and 1550 C
Reaction time (min) 0 1.333 2.666 3.999 5.332 6.665 7.998 9.331 10.664 XRF analysis of bulk slag (wt%) SiO2 7.94 7.73 7.66 7.56 7.61 7.58 7.56 7.57 7.55 Al2 O3 39.85 39.84 39.31 38.81 39.27 39.00 38.80 39.04 38.81 Fe 2 O3 2.13 2.12 2.09 2.07 2.09 2.08 2.07 2.08 2.08 TiO2 2.82 2.81 2.77 2.73 2.76 2.74 2.75 2.75 2.74 MgO 0.04 0.04 0.08 0.13 0.09 0.1 0.12 0.12 0.1 CaO 47.3 47.5 48.2 48.9 48.4 48.6 49.0 48.9 49.1 CaO from curve fitting 47.18 47.83 48.24 48.52 48.73 48.89 49.02 49.12 49.20

286

Figure A. 20:Concentration of CaO dissolved in slag with 5 wt% ilmenite at 90 rpm and in air at 1570 C

Table A. 20: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% ilmenite in air at 90 rpm and 1570 C
Reaction time (min) 0 1.333 2.666 3.999 5.332 6.665 7.998 9.331 10.664 XRF analysis of bulk slag (wt%) SiO2 7.95 7.84 7.73 7.68 7.60 7.58 7.47 7.63 7.47 Al2 O3 40.0 40.0 39.9 39.6 39.1 38.9 38.4 39.0 38.3 Fe 2 O3 2.22 2.15 2.12 2.11 2.08 2.07 2.05 2.08 2.04 TiO2 2.81 2.81 2.80 2.78 2.77 2.75 2.72 2.75 2.70 MgO 0.04 0.04 0.06 0.08 0.10 0.13 0.15 0.13 0.16 CaO 47.3 47.7 48.2 48.5 48.8 49.2 49.8 49.2 49.8 CaO from curve fitting 47.29 47.74 48.15 48.52 48.85 49.15 49.40 49.62 49.80

287

Figure A. 21: Concentration of CaO dissolved in slag with 5 wt% ilmenite at 90 rpm and in air at 1600 C

Table A. 21: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% ilmenite in air at 90 rpm and 1600 C
Reaction time (min) 0 1.333 2.666 3.999 5.332 6.665 7.998 9.331 10.664 XRF analysis of bulk slag (wt%) SiO2 7.87 7.74 7.68 7.65 7.60 7.54 7.55 7.44 7.35 Al2 O3 40.1 39.4 39.4 39.2 38.8 38.7 38.6 38.2 37.8 Fe 2 O3 2.14 2.14 2.12 2.09 2.07 2.07 2.07 2.04 2.02 TiO2 2.83 2.77 2.76 2.75 2.74 2.72 2.72 2.69 2.66 MgO 0.05 0.12 0.10 0.11 0.14 0.14 0.15 0.18 0.21 CaO 47.4 48.3 48.7 48.9 49.0 49.6 49.7 50.0 50.5 CaO from curve fitting 47.40 48.11 48.59 48.99 49.33 49.62 49.87 50.09 50.28

288

A.6 Effect of Mn3O4 addition on dissolution of CaO in calcium aluminosilicate slag at various temperatures
The effect of 5 wt% addition of Mn3 O4 on the variation of concentration of lime in the slag versus time is presented in this section. The experimental data analysed by XRF along with the fitted curve are presented in Figure A. 22 to Figure A. 25 and Table A. 22 to Table A. 26. There is an error of 0.2 0.3 wt% (absolute) from the XRF analysis.

289

Figure A. 22: Concentration of CaO dissolved in slag with 5 wt% Mn3 O4 at 90 rpm and in air at 1430 C Table A. 22: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% Mn3 O4 in air at 90 rpm and 1430C
Reaction time (min) 0.0 1.3 2.7 4.0 5.3 6.7 8.0 9.3 10.7 XRF analysis of bulk slag (wt%) SiO2 7.76 7.65 7.70 7.60 7.56 7.57 7.60 7.52 7.46 Al2O3 40.4 39.8 40.0 39.6 39.4 39.3 38.5 39.0 38.7 MgO 0.05 0.06 0.08 0.10 0.10 0.10 0.1 0.14 0.16 Mn3O4 4.76 4.81 4.82 4.77 4.74 4.73 4.65 4.70 4.64 CaO 47.6 47.7 48.1 48.4 48.4 48.5 48.4 49.1 49.0 CaO from curve fitting 47.66 47.83 48.00 48.18 48.35 48.53 48.70 48.88 49.05

290

Figure A. 23: Concentration of CaO dissolved in slag with 5 wt% Mn3 O4 at 90 rpm and in air at 1500 C Table A. 23: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% Mn3 O4 in air at 90 rpm and 1500 C
Reaction time (min) 0.0 1.3 2.7 4.0 5.3 6.7 8.0 9.3 10.7 XRF analysis of bulk slag (wt%) SiO2 7.91 7.70 7.68 7.67 7.62 7.67 7.65 7.61 7.50 Al2O3 40.1 40.0 40.0 40.1 39.9 39.9 40.1 39.7 39.1 MgO 0.100 0.080 0.070 0.080 0.080 0.080 0.090 0.110 0.130 Mn3O4 4.46 4.57 4.57 4.59 4.54 4.54 4.57 4.53 4.44 CaO 47.4 47.9 47.9 48.3 48.1 48.3 48.5 48.8 49.2 CaO from curve fitting 47.48 47.67 47.86 48.05 48.24 48.43 48.61 48.80 48.99

291

Figure A. 24: Concentration of CaO dissolved in slag with 5 wt% Mn3 O4 at 90 rpm and in air at 1550 C

Table A. 24: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% Mn3 O4 in air at 90 rpm and 1550 C
Reaction time (min) 0.0 1.3 2.7 4.0 5.3 6.7 8.0 9.3 10.7 XRF analysis of bulk slag (wt%) SiO2 7.84 7.72 7.73 7.59 7.58 7.58 7.58 7.52 7.52 Al2O3 40.3 40.2 40.1 39.6 39.7 39.5 39.4 39.2 39.1 MgO 0.05 0.07 0.08 0.10 0.11 0.11 0.14 0.15 0.14 Mn3O4 4.60 4.59 4.57 4.52 4.51 4.52 4.49 4.46 4.45 CaO 47.6 47.9 48.3 48.7 48.7 48.9 49.3 49.3 49.5 CaO from curve fitting 47.58 47.96 48.28 48.56 48.79 49.00 49.18 49.34 49.48

292

Figure A. 25: Concentration of CaO dissolved in slag with 5 wt% Mn3 O4 at 90 rpm and in air at 1600 C

Table A. 25: XRF analysis of the bulk slag when lime dissolves in slag with addition of 5 wt% Mn3 O4 in air at 90 rpm and 1600 C
Reaction time (min) 0.0 1.3 2.7 4.0 5.3 6.7 8.0 9.3 10.7 XRF analysis of bulk slag (wt%) SiO2 7.93 7.63 7.66 7.57 7.59 7.51 7.53 7.47 7.40 Al2O3 40.3 39.6 39.8 38.4 39.2 38.9 38.8 38.8 38.3 MgO 0.05 0.10 0.12 0.14 0.14 0.13 0.15 0.17 0.20 Mn3O4 4.62 4.56 4.56 4.40 4.47 4.45 4.44 4.47 4.38 CaO 47.6 48.5 49.0 48.3 49.2 49.2 49.7 49.9 50.2 CaO from curve fitting 47.71 48.35 48.79 49.11 49.36 49.56 49.72 49.85 49.96

293

A.7 Effect of SiO 2 addition on dissolution of CaO in calcium aluminosilicate slag at various temperatures
The effect of 5 wt% addition of SiO 2 on the variation of concentration of lime in the slag with time is presented in this section. The experimental data analysed by XRF along with the fitted curve are presented in Figure A. 26 to Figure A. 28 and Table A. 26 to Table A. 28. There is an error of 0.2 0.3 wt% (absolute) from the XRF analysis.

294

Figure A. 26: Concentration of CaO dissolved in slag with additional 5 wt% SiO2 at 90 rpm and in air at 1500 C for 1 hour Table A. 26: XRF analysis of the bulk slag when lime dissolves in slag with additional 5 wt% SiO2 in air at 90 rpm and 1500 C
Reaction time (min) 0 2 4 6 8 10 20 30 40 50 60 XRF analysis of bulk slag (wt%) SiO2 12.8 12.6 12.7 12.6 12.3 12.6 12.3 12.2 12.3 12.2 12.0 Al 2O3 40.2 39.9 40.1 39.9 38.8 39.6 38.9 38.7 38.8 38.3 37.8 MgO < DL 0.03 0.02 0.07 0.12 0.08 0.07 0.14 0.14 0.16 0.19 CaO 47.4 48.3 48.0 48.4 49.3 48.4 49.8 49.5 49.6 50.4 50.5 CaO from curve fitting 47.43 47.74 48.00 48.22 48.41 48.58 49.21 49.65 50.00 50.30 50.57

295

Figure A. 27: Concentration of CaO dissolved in slag with additional 5 wt% SiO2 at 90 rpm and in air at 1550 C for 1 hour Table A. 27: XRF analysis of the bulk slag when lime dissolves in slag with additional 5 wt% SiO2 in air at 90 rpm and 1550 C
Reaction time (min) 0 2 4 6 8 10 20 30 40 50 60 XRF analysis of bulk slag (wt%) SiO2 12.9 12.5 12.4 12.3 12.3 12.3 12.4 12.3 12.1 11.9 11.7 Al 2O3 40.3 39.6 39.3 39.0 38.7 39.0 39.2 38.7 38.4 37.6 36.8 MgO < DL 0.04 0.10 0.10 < DL 0.08 0.04 0.12 0.11 0.22 0.25 CaO 47.7 48.0 48.8 48.7 49.9 49.3 49.4 49.4 50.2 51.0 51.3 CaO from curve fitting 47.64 48.15 48.48 48.73 48.92 49.08 49.66 50.10 50.50 50.87 51.24

296

Figure A. 28: Concentration of CaO dissolved in slag with additional 5 wt% SiO2 at 90 rpm and in air at 1600 C for 1 hour Table A. 28: XRF analysis of the bulk slag when lime dissolves in slag with additional 5 wt% SiO2 in air at 90 rpm and 1600 C
Reaction time (min) 0 1.33 2.66 4 5.33 6.66 8 9.33 10.66 20 30 40 50 XRF analysis of bulk slag (wt%) SiO2 13.1 12.6 12.5 12.3 12.3 12.5 12.4 12.4 12.3 12.4 11.9 11.6 11.5 Al 2O3 41.4 39.8 39.4 39.2 39.1 39.8 39.3 39.0 38.7 39.3 37.6 37.0 36.7 MgO 0.03 < DL 0.15 < DL 0.11 0.07 0.10 0.09 0.12 0.12 0.20 0.21 0.26 CaO 47.4 48.0 48.9 48.9 49.3 48.3 49.1 48.9 49.4 49.4 50.5 51.1 51.2 CaO from curve fitting 47.37 48.20 48.60 48.85 49.03 49.17 49.29 49.39 49.48 49.99 50.48 50.93 51.31

297

A.8 Effect of Fe2O3 addition on dissolution of MgO in calcium aluminate slag

The effect of 5 and 10 wt% addition of Fe2 O3 on the variation of concentration of magnesia in the slag with time is presented in this section. The experimental data analysed by XRF along with the fitted curve are presented in Figure A. 29 to Figure A. 30 and Table A. 29 to Table A. 30. There is an error of 0.2 0.3 wt% (absolute) from the XRF analysis.

298

Figure A. 29: Concentration of MgO dissolved in slag with 5 wt% Fe2 O3 at 90 rpm in air at 1430 C for 1 hour

Table A. 29: XRF analysis of the bulk slag when MgO dissolves in slag with addition of 5 wt% Fe2 O3 in air at 90 rpm and 1430C
XRF analysis of bulk slag (wt%) Reaction time (min) 0 10 20 30 40 50 60 Al 2O3 48.7 47.9 47.3 47.6 46.9 47.3 47.1 CaO 42.0 41.8 41.5 41.4 41.5 41.4 41.4 Fe 2O3 9.65 9.69 9.61 9.37 9.67 9.31 9.39 MgO 0.00 1.34 1.83 2.16 2.47 2.58 2.63

299

Figure A. 30: Concentration of MgO dissolved in slag with 10 wt% Fe2 O3 at 90 rpm in air at 1430 C for 1 hour

Table A. 30: XRF analysis of the bulk slag when MgO dissolves in slag with addition of 10 wt% Fe2 O3 in air at 90 rpm and 1430C
XRF analysis of bulk slag (wt%) Reaction time (min) 0 10 20 30 40 50 60 Al 2O3 50.7 50.3 49.9 50.0 50.0 49.7 49.8 CaO 40.2 39.9 39.5 39.3 39.4 39.1 38.9 Fe 2O3 5.43 5.14 5.26 4.96 5.06 5.02 5.00 MgO 0.00 1.54 2.35 2.78 3.12 3.26 3.41

300

A.9 Effect of (CaF 2 + Fe 2O3) addition on dissolution of MgO in calcium aluminate slag
The effect of (5 wt%CaF2 + 5 wt% Fe2 O3 ) and (5 wt%CaF2 + 10 wt% Fe2 O3 ) addition on the variation of concentration of magnesia in the slag with time is presented in this section. The experimental data analysed by XRF along with the fitted curve are presented in Figure A. 31 to Figure A. 32 and Table A. 31 to Table A. 32. There is an error of 0.2 0.3 wt% (absolute) from the XRF analysis.

301

Figure A. 31: Concentration of MgO dissolved in slag with addition of 5% CaF2 &5% Fe2 O3 at 90 rpm in air at 1430 C for 1 hour

Table A. 31: XRF analysis of the bulk slag when MgO dissolves in slag with addition of 5% CaF2 & 5% Fe2 O3 in air at 90 rpm and 1430 C for 1hour
Reaction time (min) 0 10 20 30 40 50 60 XRF analysis of bulk slag (wt%) Al2O3 49.0 46.4 46.1 46.1 45.4 45.8 46.5 CaO 41.54 38.96 38.84 39.09 38.85 38.64 39.04 Fe 2O3 5.26 6.17 6.04 6.02 6.07 5.51 5.58 CaF2 3.57 2.93 2.83 2.68 2.00 2.93 2.90 MgO 0.00 4.62 5.15 5.39 5.10 5.30 5.27

302

Figure A. 32: Concentration of MgO dissolved in slag with addition of 5% CaF2 &10% Fe2 O3 at 90 rpm in air and at 1430 C for 1 hour

Table A. 32: XRF analysis of the bulk slag when MgO dissolves in slag with addition of 5% CaF2 & 10% Fe2 O3 in air and at 90 rpm and 1430 C
Reaction time (min) 0 10 20 30 40 50 60 XRF analysis of bulk slag (wt%) Al2O3 45.83 44.36 44.05 43.98 43.78 43.67 43.72 CaO 38.98 37.70 37.37 37.19 37.23 37.23 37.29 Fe2O3 9.92 10.01 9.96 9.66 9.60 9.36 9.33 CaF 2 3.68 3.36 3.39 3.40 3.37 3.35 3.25 MgO 0.00 3.20 4.19 4.64 4.88 4.89 5.07

303

Appendix B.

Model for estimating the slag viscosity

A model developed by Urbain, Cambier, Deletter and Anseau (1981) was used in the present study for estimation of viscosity of slags. The model uses the Frenkel equation, as given in Equation (B. 1), where A and B are viscosity parameters, T is the thermodynamic temperature K, and is in poise;

B = AT exp( ) T

(B. 1)

This model is based on the behaviour of CaO Al2 O3 SiO 2 system and the parameters A and B are calculated by dividing the slag constituents into three categories:

Glass formers:

X G = X SiO2 + X P2 O5

Modifiers:
X M = X CaO + X MgO + X Na 2O + X K 2O + 3 X CaF2 + X FeO + X MnO + 2 X TiO2 + 2 X ZrO2

Amphoterics:

X A = X Al2 O3 + X Fe2O3 + X B2 O3

304

In the present work, Fe2 O3 has been classified as a modifier and in the computer program for the calculation of viscosity, where 1.5 x Fe2O 3 has been added to X M
* * * and deduced from X A . Normalized values X G , XM , and X A are obtained by

dividing

the

mole

fractions,

XG ,

XM ,

and

X A by

the

term

(1 + 2 X CaF2 + 0.5 X FeO1.5 + X TiO2 + X ZrO2 ) . Urbain et al. (1981) proposed that the

parameter B was influenced both by the ratio =

* XM * and by X G . The * (X * + X ) M A

parameter B can be expressed in the form of Equation (B. 2), where B1 , B2 , and B3 can be obtained by Equation (B. 3).

* * 2 * 3 B = B0 + B1 X G + B2 ( X G ) + B3 ( X G )

(B. 2)

Bi = i + bi + ci 2

(B. 3)

B0 , B1 , B2 , B3 can be calculated from the equations listed in Table B. 1 and these parameters are then introduced into Equation (B. 2) to calculate B.

Table B. 1: Equations for B-parameters in Urbain model for viscosity

B0 = 13.8 + 39.9355 44 .049 2

B1 = 30.481 117.1505 + 139.9978 2 B2 = 40.9429 + 234.0486 300.04 2

B3 = 60.7619 153.9276 + 211.1616 2

The parameter A can be calculated from B by Equation (B. 4) and the viscosity of the slag (in poise) can then be determined using Equation (B. 5).

305

 LnA = 0.2693B + 11.6725

(B. 4)

10 3 B ) = AT exp( T

(B. 5)

This model has been used to calculate the viscosities of slags with widely varying compositions and it has been found that it gives values, which agree well with experimental data. According to Slag Atlas (Eisenhuttenleute (1995)), the discrepancies between the experimental values and the predicted values are of the order of 25 30%, which are similar to the experimental uncertainties for the viscosity measurements.

306

APPENDIX C. Model for estimating the slag density

An additive method for the estimation of densities in alloys and slags has been widely used for some time. In this method, the molar volume V , can be obtained from Equations (C. 1) and (C. 2) below, where M , x , and V are the molecular weight, mole fraction, and the partial molar volume, respectively, and the subscripts 1, 2, and 3 denote the various oxide constituents of the slag.

V=

M 1 x1 + M 2 x 2 + M 3 x 3 ...

(C. 1)

V = x1V 1 + x 2 V 2 + x3 V 3 + ...

(C. 2)

Partial molar volume is usually assumed to be equal to the molar volume of the pure component V 0 but it has been pointed out by Lee and Gaskell (1974) and Grau and Masson (1976) that the density of slag is also related to the structure.

Slags containing SiO 2 , Al2 O3 , and P2 O5 consist of chains, rings, and complexes, which are independent upon the amount and nature of the cations present according to Slag Atlas (Eisenhuttenleute (1995)). Thus it is necessary to make the partial molar volumes dependant upon composition for oxides of this type. According to Mills et al. (1987), the values for V SiO2 have been derived using the experimental density data for the systems, FeO SiO 2 , CaO SiO 2 , MnO SiO 2 ,

307

Na2 O SiO 2 , K2O SiO 2 , and CaO FeO SiO 2 . From these experimental values the relation V SiO2 = 19.55 + 7.966 xSiO2 (cm3 .mol-1 ) was derived.

Values for V Al2 O3

were also determined by Mills et al. (1987), using

experimental density data for the systems; Al2 O3 CaO, Al2 O3 CaF2 , Al2 O3 SiO 2 , Al2 O3 CaO MgO, and Al2 O3 MnO SiO 2 . The relation V Al2O3 = 28.31 + 32 x Al2 O3 31.45 x 2 Al2 O3 was derived. The recommended values for V for the various oxides at 1500 C by the same authors are given in Table C. 1.

Table C. 1: Recommended values for partial molar volume V of various slag constituents at 1500 C Constituent Al2 O3 CaF2 CaO FeO Fe2 O3 K2O MgO MnO Na2 O P2 O5 SiO 2 TiO 2

V , cm3 mol-1 28.31 + 32 x Al2 O3 31.45 x 2 Al2 O3 31.3 20.7 15.8 38.4 51.8 16.1 15.6 33 65.7 19.55 + 7.966 x SiO2 24

In order to provide a temperature coefficient, the temperature dependencies of the molar volumes ( dV / dT ) of several slag systems were examined by Mills et al. (1987) and a mean value of 0.01% K-1 was adopted.

308

The standard deviation of the data predicted by the model compare to the experimental values

est

exp )

exp

was reported by Mills et al. (1987) to be

between 1 and 2%. The experimental uncertainties associated with density measurements for slags is about 2 3%.

309

APPENDIX D. Error analysis

The errors on dissolution studies come from a number of sources, such as uncertainties in XRF che mical analysis of dissolved solid oxide concentration in the slag, fitting a linear curve through initial concentration data, slag weight and the exact geometrical shape of samples in individual experiments.

In the estimation of errors and the way they propagate through calculations, three general rules have been applied in the present work according to Taylor (1982) as the followings:

Suppose that x,, w are measured with uncertainties x ,..., w and the measured values used to compute is :

q = x + ... + z (u + ... + w )

(D. 1)

If the uncertainties in x, , w are known to be independent and random, then the uncertainty in q is the quadratic sum as in Equation (D. 2):

q = [(x ) 2 + ... + (z ) 2 + (u) 2 + ... + (w) 2 ]1/ 2

(D. 2)

If the same variables are measured, used to compute as :

310

q=

x ... z u ... w

(D. 3)

If the uncertainties in x, , w are independent and random, then the fractional uncertainty in q is the sum in quadrature of the original fractional uncertainties as:

2 2 2 2 q x z u w = + ... + + + ... + q z u w x

1/2

(D. 4)

Also if x is measured with uncertainty of x and is used to calculate the power q = x n (where n is a fixed, known number), then the fractional uncertainty in q is:

q x = n q x

(D. 5)

To estimate the uncertainty of diffusivity measurements, the first step is to estimate the error involved in calculation of initial rate of dissolution, which was calculated from the slope of linear line fitted through the initial concentration data, slag weight ( w ) and immersed area ( A ) of the solid oxide sample in the melt as in Equation (D. 6):

311

Rate =

Slope w 60 A 100

(D. 6)

The error involved is found out on the basis of Equation (D. 7) as:

( rate) ( slope ) ( w) ( A) = + + rate slope w A

(D. 7)

As the rate of dissolution was deduced from the slope of the linear line fitted through the initial data on the concentration of solid oxides in the slag, a Curve Fitting Toolbox in MATLAB software was used to perform the curve fitting. The Curve Fitting Toolbox uses the linear least square method to fit a linear model to data. The result of the fitting process is an estimate of the coefficients of a firstdegree polynomial mode. As these coefficients determine the slope of the fitted line, the Toolbox also generates maximum and minimum prediction bounds for each coefficient, thus the error involved can be calculated. Depending on various experiments, the average uncertainty for the slope of fitted line calculated to be 10%.

( slope ) was slope

The initial interfacial reaction area in the lime/magnesia dissolution experiment was well defined by the immersed area of solid oxide cylinder in the melt. It is difficult to estimate the change in the shape and area of the solid oxide surface during the rotation of lime sample in the melt but based on the observation of dissolution data, it was concluded that the area of solid oxide sample does not change radically during the first 10 minutes of reaction in the melt. This error is 312

calculated according to calculation of area of disk and cylinder immersed in the melt as in Equation (D. 8):

A = r 2 + 2rh

(D. 8)

Where r is the radius of the disk (cm) and h is the height (cm) of the immersed sample. The error is found as:

A = 2rr + 2hr + 2rh

(D. 9)

The change in radius and height of the sample could be found on the basis of the dissolution rate data. The change in weight of lime sample (dw) in the slag can be calculated according to the volume change of the sample dv as in Equation (D. 10):

dw = dv = ( 2rhdr + r 2 dh)

(D. 10)

It is assumed here that the change in diameter (dr) and height (dh ) are approximately the same and it is called characteristic length, therefore when the left side of the equation is divided by (2 rh + r 2 ) , it would be equivalent to the rate of dissolution. As the average rate of dissolution is 2 10 4 g/cm2 .s for various experiments, this value is divided by the density of slag (about 3 g/cm3 ) to find out the changes of the characteristic length, which was calculated to be
6 10 5 cm/s. By considering the changes of this length during 5 minutes of

rotation, the radius change dr is about 2 10 2 cm. In the estimation of the error 313

involved in the hight of the immersed sample, it is also assumed that while the sample is rotating, the melt could wet the sample within 0.1 cm of the assumed height of immersion. This value is added to the characteristics length and results in 0.12 cm error in the estimation immersion height dh . By inserting these values into Equation (D. 9) and dividing it by total area of the sample, the error from the calculation of immersed area in the melt is around 7 %.

The uncertainty of estimated slag weight during experiments is limited. According to the concentration of solid oxide dissolved in the slag, the maximum amount of lime dissolved in the slag during the first 10 minutes is about 1.2 grams. The mass of slag taken during sampling of the slag during 10 minutes, is about 2.5 grams (7 times sampling, each 0.3~0.4 grams). The error due to measurement of slag weight and loss of volatile component of slag at high temperature is also included, where they totally result to 4 % error.

The estimation of error in calculation of rate of dissolution is found by inserting the errors from the initial slope of dissolution curve, immersed area of the sample and weight of molten slag into Equation (D. 7). The error is about 15%.

The next step is estimation of error from calculation of mass transfer coefficient. This parameter was calculated from rate of dissolution, the driving force of the solid oxide and density as in Equation (D. 11):

314

K=

rate 100 C

(D. 11)

The error involved could be found as in Equation (D. 12):

K rate C = + + K rate C

(D. 12)

The error from the chemical analysis of the concentration of solid oxide in the melt is about 0.2 grams. As the driving force between the original solid oxide concentration is about 5-10%, then

C is about 2-4%. It is mentioned in the C

literature that the error from the density calculation is about 2-3%. These errors result in about 22% uncertainty in calculation of mass transfer coefficient.

In the calculation of diffusivity, the mass transfer coefficient calculated in Equation (D. 11), was inserted in Equation (D. 13) in order to deduce the diffusivity ( D) in the right side of the Equation (D. 13):

K=

2h r 0.621 1 / 2 1 / 6 D 2 / 3 + 0.065 Re 0.25 2 / 3VD 2 / 3 r + 2h r + 2h

(D. 13)

Therefore, the error for diffusivity 14):

d is calculated according to Equation (D. d

315

d k ( coefficient ) = 1.5 + 1.5 d k (coefficient )

(D. 14)

Where the coefficient is defined as in Equation (D. 15):

Coefficient =

r 2h 0.065 Re 0.25 2 / 3V 0.621 1/ 2 1/ 6 + r + 2h r + 2h

(D. 15)

In estimation of error;

( coefficient ) r , the random errors resulted from ( coefficient ) r + 2h

and

2h are very small, also the rest of components, due to their magnitudes r + 2h

have very small random errors, therefore the calculation of the error on diffusivity data mainly depends on the error resulted from the mass transfer coefficient in Equations (D. 12) and (D. 14). Thus the total error on diffusivity calculations is about 33%.

316

Appendix E.

Preliminary study of lime dissolution in static slag

Experiments were carried out to study static dissolution of CaO into a CaO 45 wt% SiO 2 10 %Al2 O3 slag and effect of additives (10 wt%: Fluorspar, Nepheline Syenite and ilmenite) on the dissolution of lime. The slag was packed into dense CaO crucibles and heated to 1500 and1600 C. After reaction period, the crucible was air cooled and cross-sectioned. The formation of the Ca2 SiO 4 phase and concentration profiles of various cations across the reaction zone were examined by Electron Probe Micro Analysis. At 1500 C, a Ca2SiO 4 layer was found to form at the CaO/slag interface. No such layer was evident at 1600 C, however the formation of this layer was verified by conducting the experiments with the same slag chemistry in platinum capsule. Ilmenite and nepheline syenite were found to be effective in increasing CaO dissolution.

E.2 Experimental
Static dissolution of CaO into slag was studied by holding slag in a dense lime crucible, varying holding reaction time, temperature and slag chemistry.

E.1.1 Materials
While acknowledging that in industry, the burnt lime is porous (more than 50 % porosity for soft-burnt lime and about 25 % porosity for hard-burnt), it was decided to manufacture and use a low porosity of 8.4 % lime to ensure a sharp lime/slag boundary and consequently better control of experimental conditions,

317

under which the lime dissolution included two major processes, the chemical reaction and CaO mass transfer in the slag phase.

Manufacturing of lime crucibles included the following operations. The limestone powder with 50-70 microns average particle size was milled with alumina balls, which resulted in 0.8-1 micron particles that were then calcined to CaO and pressed in a mould and subsequently fired to provide the necessary density and strength. This technique allowed a close control of bulk density and apparent porosity by precisely varying compacting pressure, sintering temperature, rate of heating and cooling and sintering time.

The experimental slags were prepared in a platinum crucible by melting a mixture of slag components, quenching the melt and then re- melting to ensure slags homogeneity. The master slag was a three component CaO-SiO 2 -Al2 O3 system with 45 wt% CaO, 45 % SiO 2 , and 10 % Al2 O3 . Flux additions were made by adding 10 wt% of the chosen flux to the master slag. The flux compositions used in the present work are presented in Table E. 1 and Table E. 2. In the case of the CaF2 addition, subsequent analysis revealed that the CaF2 content of the resultant slag was only 5.7 wt% indicating a loss of fluorine. The fluorine loss and oxidation of calcium of fluorspar was reported by Shimizu et al. (1996).

E.1.2 Experimental Procedure

The dissolution of dense lime in molten slag was studied at 1500 and 1600 C as a function of time and slag composition. The concentration profiles of different

318

elements in the lime/slag interface were quantitatively measured by EPMA mapping.

The CaO crucible was filled with slag and charged to the muffle furnace. A sample was heated slowly at 100 C/hr up to 1250 C which is below the melting point of slag studied in this work, in order to minimize the risk of thermal shock in the crucible and then heated quickly at 600 C/hr up to 1500 or 1600 C. Reaction time was counted from the moment of reaching to 1500 or 1600 C. After the required reaction time, the crucible and slag were taken out of the furnace and quickly quenched in air. The crucible was cut by a diamond saw parallel to bottom of the crucible. The slice of the crucible with its contents was mounted in resin.

Samples must be provided with a flat and well-polished surface, finished to (0-1) micrometers for electron probe microanalysis (EPMA). Therefore, they were polished initially with the Struers waterproof silicon carbide paper at consecutive grits sizes of 320, 800 and 1200 using a Struers Labopol-5 grinding machine at 300 rpm. Since samples are very sensitive to moisture, Shell Macron oil was used as the lubricating fluid. Diamond polishing of the specimen was done initially with a Chemo-textile Cloth (Leco-PAN-W) using polycrystalline diamond paste of 3 microns. The final mirror finish was obtained by using a Silk type cloth (Kemet-MSF) with 1 micron polycrystalline diamond paste. A LECO G25 Rotary Polisher was used for the final stage of polishing.

A CAMECA SX50 EPMA was used in quantitative analysis of the samples. It is a fully automated instrument employing 319 four wavelength dispersive

spectrometers in order to analyse various elements. These elements were analysed with the TAP, PCO and PET crystals. All samples were examined using an accelerating vo ltage of 15KV, a beam current of 20nA and a beam size of 1 micron. The instrument was operated with SAMx application software. X -ray intensity distributions were acquired for the main constituents to produce elemental analysis across the area of interest. A MATLAB script program was developed to process the X-ray intensity distribution data. A JOEL 840 Scanning Electron Microscope was used to obtain the backscattered electron images of various phases in the samples operated at 20 KV.

Line scan analyses were carried out from the CaO region in the crucible into the slag region. The step size for stage movement was 6 microns. Point analysis was performed for each distinct phase identified using back-scattered electron mode. The MATLAB program converted the concentration profile of each element in the matrix to the mole percentage of various oxides and determined the phase locations in the ternary phase diagram. The phase diagrams at different temperatures were obtained from FactSage package developed by Bale et al. (2003).

The elemental distribution in the slag adjacent to the slag/crucible layer was mapped by 2- micron step size stage movement in an area of 512 by 512 microns size and at a 256 by 256 image resolution. A MATLAB program was also developed to analyse the results from mapping. This program incorporates a Graphics User Interface media, which changes the concentration intensity of the maps to the intensity images in order to illustrate the mole percentage of various

320

phases. The program also enabled the recognition of Ca2 SiO 4 and Ca3 SiO 5 phases in the mapped area.

E.3 Experimental Results and Discussion

Electron Micrographs at 1500 C (Figure E. 1) show formation of Ca2 SiO 4 layer, which was also observed by Matsushima et al. (1977). However at 1600 C (Figure E. 2), no discrete layer of Ca2 SiO 4 was found . The appearance of crystallites indicates that they were not present in the liquid phase but formed in the process of crystallization. The Ca2 SiO 4 phase, identified by MATLAB program on the basis of the EPMA analysis is shown in Figure E. 3 and Figure E. 4. Although a layer of Ca2 SiO 4 was not observed at 1600 o C, it does not mean that it was not formed at this temperature.

Change in the slag chemical and phase composition in process of lime dissolution at 1500 and 1600 o C is shown in the CaO-SiO 2 -Al2 O3 phase diagrams in Figure E. 5and Figure E. 6. After achieving 1500 o C (zero time), CaO concentration in the slag increased from 45 wt% to 48 wt%, while at 1600 o C (zero time) slag CaO content was 53 wt% and very close to the Ca2 SiO 4 line. At 1600 C as reaction time increased to 30 and 60 minutes, change in slag CaO content was small (54-55 wt%) indicating that dissolution of lime had slowed dramatically. At this CaO content, the slag is saturated with di-calcium silicate, however the proportion of the Ca2 SiO 4 phase is negligible.

At 1500 o C, CaO concentration increased to 52 wt% after 30-min reaction and further to 54 wt% after 60 min. In the CaO-SiO 2 -Al2 O3 phase diagram, these 321

compositions are within the liquid Ca2 SiO 4 area (Figure E. 5) reflecting microscopic observations of the Ca2 SiO 4 phase. Total content of CaO in the slag was about the same at 1500 and 1600 C after dissolution for 30 min (52-54 wt%) and 60 min (53-55 wt%). However, at 1600 C all slag was presumably liquid, while at 1500 C, after reaction with lime crucible for 30 and 60 min, it consisted of liquid phase and solid di-calcium silicate.

At 1500 C, the growth rate of di-calcium layer was measured by line scan with EPMA. The thickness of the layer is shown in Table E. 3. These measurement shows that at time = 0, the Ca2 SiO 4 layer has already been formed with a thickness of 164 m, so from the real time that the layer starts to form, the dissolutio n of lime has been hindered. By increasing time to 30 minutes and one hour, it seems that it is the Ca2 SiO 4 formed at the lime/slag interface which dissolves and not the lime itself. Although, the experimental results show a net growth of Ca2 SiO 4 layer but formation of layer seems greater than its dissolution.

At 1600 C, we know that while heating from the melting point to 1600 C, the Ca2 SiO 4 was formed at 1500 C with a thickness of 164 m. It takes 10 minutes from 1500 to reach 1600 C (heating ramp: 600 C/hour). It seems that during this period of time, the net dissolution and formation of Ca2 SiO 4 left a layer, which prevented the complete dissolution of lime. Since no Ca2 SiO 4 was observed in the quenched samples, it is postulated that it was lost during quenching/sample preparation, perhaps as a result of the large volume change associated with the phase transformation of Ca2 SiO 4 at 860-780 C. This behaviour has been observed by other researchers and is referred to as dusting. 322

The formation of the Ca2 SiO 4 layer was later verified by reacting a lime piece and master slag in a platinum capsule and fast quenching in air. The result is shown in Figure E. 7. It seems that the difference in the rate cooling in two techniques would affect the stability and recovery of Ca2 SiO 4 layer.

On the other hand, the presence of the Ca2 SiO 4 layer in the experiments in the lime crucibles is indicated by the fact that the amount of lime dissolved in the slag at 1600 C was not significantly higher than its corresponding values at 1500 C and that dissolution stopped substantially at the Ca2 SiO 4 line at both temperatures. The rate limiting stage could be diffusion of CaO through the Ca2 SiO 4 layer. Future work may be defined to investigate further the formation of Ca2 SiO 4 layer at 1500 1600 C with confocal scanning laser microscopy.

With the addition of the ilmenite and Nepheline Syenite fluxes to the slag, CaO content of slag increased more rapidly with time (Table E. 4). Aga in no Ca2 SiO 4 layer was recovered in the experiments.

E.4 Key findings

The dissolution of dense lime in molten CaO-SiO 2 -Al2 O3 slags was studied at temperatures of 1500 and 1600 C by reaction of slags with a static lime crucible. The effects of different fluxing agents and time of reaction were investigated. Ilmenite was the most effective flux and nepheline syenite was comparable to fluorspar, increasing the lime dissolution. At 1500 C a di-calcium silicate layer was formed between the slag and CaO crucible and the growth rate was measured. However at 1600 C no di-calcium silicate layer was onberved. As the 323

formation of Ca2 SiO 4 layer was verified by a different experimental technique, it would appear that at both temperatures rate of CaO dissolution is limited by the formation of a Ca2 SiO 4 layer.

324

Slag

C2S CaO

Figure E. 1: SEM of base slag at 1500 C for time=0 with 1000 magnification.
Phase: Ca2SiO4 250 1
250

Figure E. 2: SEM of base slag at 1600 C for time=0 with 1000 magnification.
phase: Ca2SiO4 1

0.9

0.9

200

0.8

200

0.8

C2S
150

0.7

0.7

0.6

150

0.6

Slag
100

0.5

0.5

0.4

100

0.4

CaO
50

0.3

0.3

0.2

50

0.2

0.1

0.1

0 50 100 150 200 250

0 50 100 150 200 250

Figure E. 3: Identified Ca2 SiO4 phase by MATLAB program for base slag at 1500 C for time=0 in a 512 by 512 microns area and 256 by 256 image resolution.

Figure E. 4: Identified Ca2 SiO4 phase by MATLAB program for base slag at 1600 C for time=0 in a 512 by 512 microns area and 256 by 256 image resolution.

325

Figure E. 5: The composition of bulk slag at various reaction times in phase diagram for basic slag at 1500 C C2S: Ca2 SiO 4 , C3S: Ca3 SiO 5 , CA: Ca2 Al2 SiO 7 , C: CaO, L: Liquid Slag

326

Figure E. 6: The composition of bulk slag at various reaction times in phase diagram for basic slag at 1600 C C2S: Ca2 SiO 4 , C3S: Ca3 SiO 5 , C: CaO, L: Liquid Slag

327

Ca2 SiO 4

CaO

Slag

Figure E. 7: Formation of Ca2 SiO4 layer on reaction of lime with master slag at 1600 C in the platinum capsule

328

Table E. 1: Chemical compositions of Nepheline Syenite Oxides SiO 2 Al2 O3 Fe2 O3 CaO MgO Na2 O K2O FeO Wt% 59.30 19.28 2.25 0.91 0.09 8.33 5.32 2.27

Table E. 2: Chemical compositions of ilmenite Oxides TiO 2 Fe2 O3 FeO Al2 O3 SiO 2 MnO P2 O5 Cr2 O3 Wt% 55.3 24.1 16 0.58 0.93 1.48 0.03 0.045

329

Table E. 3: Growth of Ca2 siO4 layer at 1500 C Temperature 1500 C time = 0 164 m time = 30 min. 192 m time =1 hr. 321 m

Table E. 4: Mass (grams) of CaO dissolved in the slags (per 100 grams of slag) at 1600 C Slags Master slag Slag + CaF2 Slag + N.S Slag + ilmenite T=0 17 15 18.7 29.4 T=30 minutes 19.5 16 23.3 30.8 T=1 hour 19.9 20 26.3 30

330