THE MECHANISM OF ALUMINOSILICATE

Transcription

1 THE MECHANISM OF ALUMINOSILICATE FORMATION IN ALUMINA REFINING A thesis Submitted for admission to the degree of DOCTOR OF PHILOSOPHY at the University of South Australia KALI ZHENG BSc. (Hons), MSc. Ian Wark Research Institute University of South Australia March, 1997

2 OF SOUTH AUSTRALIA LIBRARY

3 TABLE OF CONTENTS LIST OF FIGURES LIST OF TABLES PUBLICATIONS PRIZES ABSTRACT AUTHOR'S STATEMENT ACKNOWLEDGMENTS ii vii xv xviii xviii xix xxii xxiii

4 TABLE OF CONTENTS CHAPTER 1. INTRODUCTION The Aluminium Industry The Bayer Process Scaling Problems Synthetic Spent Bayer Liquor Aims of this Research 7 CHAPTER 2. LITERATURE REVIEW Introduction Crystallisation of Sodium Aluminosilicate Phases Solution Species in Spent Bayer Liquor Structural Characteristics of Sodium Alurninosilicate Phases Crystalline Structure Vibrational Spectroscopy Nuclear Magnetic Resonance Crystallisation of Sodium Aluminosilicate Phases in Synthetic Solutions Preparation of Zeolite Preparation of Sodalite Preparation of Cancrinite Comparison of Preparation of Zeolite, Sodalite and Cancrinite Characterisation of Plant Desilication Products Crystalline Phase Transformations of Sodium Aluminosilicate Phase Changes Unit Cell Changes The Effect of Salts or Impurities in the Liquors Other Factors Silica Equilibrium Concentrations in Bayer Liquors The Effect of Solid Phases The Effect of Temperature The Effect of NaOH Concentration The Effect of Al(OH)3INaOH Ratio The Effect of Impurities on Solubility 44 11

5 Empirical Correlations for the Estimation of the Silica Solubility Desilication of Bayer Liquors Methods of Desilication Pre-digestion Desilication Pre-digestion-desilication with the Aid of CaO Post-digestion desilication Other Methods for Desilication Factors that Influence Desilication Kinetics of Sodium Aluminosilicate Crystallisation and Dissolution Kinetics of Crystallisation Kinetics of Dissolution Mechanism of Sodium Aluminosilicate Formation Summary 59 CHAPTER 3. EXPERIMENTAL METHOD Introduction Crystallisers Atmospheric Pressure Reactor for under 100 C Parr Bombs Experiments Autoclave Experiments at High Pressures Preparation of Synthetic Spent Bayer Liquor Preparation of the Liquor under Atmospheric Conditions Preparation of the Liquor for Autoclave Experiments Preparation of the Liquors for Parr Bomb Experiments Seed Preparation Sodalite Seed Cancrinite Seed Plant Cancrinite Seed Crystallisation Experiments Formation of Sodium Aluminosilicate as a Function of Time Solubility Measurements Crystallisation and Dissolution of Sodium Aluminosilicate Liquor Composition Analysis Silica Concentration Analysis Molybdate Method ICP Method

6 Comparison of the Two Methods Sodium Hydroxide, Sodium Carbonate and Aluminium Hydroxide Concentrations Crystal Characterisation Techniques 74 CHAPTER X-ray Diffraction Fourier Transform Infrared Scanning Electronic Microscopy and Energy Dispersion Spectroscopy X-ray Photoelectron Spectroscopy Particle Sizer 77 CHARACTERISATION OF SODIUM ALUMINOSILICATE NUCLEATED BETWEEN C Introduction Sodium Aluminosilicate Synthesis between C Effect of Temperature Effect of Sodium Carbonate Concentration Effect of Silica Supersaturation Conclusion 92 CHAPTER 5. INVESTIGATION OF SODIUM ALUMINOSILICATE PHASE TRANSFORMATION Introduction Phase Changes of Nucleated Sodium Aluminosilicate Effectof Aging Effect of Sodium Carbonate at Low Temperature Effect of Sodium Carbonate at High Temperature Effect of Silica Supersaturation Phase Changes of Sodalite Seed Phase Changes of Sodalite Seed at 90 C Phase Changes of Sodalite Seed at 160 C Cancrinite Seed Conclusion 116 CHAPTER 6. SILICA EQUILIBRIUM SOLUBILITY OF DIFFERENT PHASES IN SYNTHETIC SPENT LIQUOR Introduction 118 iv

7 6.2. Silica Equilibrium Solubility of Synthetic Cancrinite The Effect of Sodium Carbonate Concentrations The Effect of Temperature Silica Equilibrium Solubility of Plant Cancrinite The Effect of Sodium Concentrations The Effect of Temperature Silica Equilibrium Solubility of Sodalite The Effect of Sodium Carbonate Concentrations The Effect of Temperature: Crystalline Phase Transformation Comparison of the Si02 Solubility among Different Seeds Conclusion 132 CHAPTER 7. CRYSTALLISATION OF SYNTHETIC SPENT LIQUOR Introduction Unseeded Experiments The Effect of Supersaturation The Effect of Sodium Carbonate Concentration and Temperature Sodalite Seeded Crystallisation The Effect of Temperature The Effect of Sodium Carbonate Concentration The Effect of Seeding The Effect of Supersaturation Secondary Nucleation Plant Cancrinite Seeded Crystailisation The Effect of Temperature and Sodium Carbonate Concentration The Effect of Supersaturation The Effect of Seeding Seed Growth Comparison of Crystallisation Behaviour Due To.Different Seeding Types Kinetics of Desilication Conclusion 152 V

8 CHAPTER 8. DISSOLUTION OF SODALITE AND CANCRINITE IN SYNTHETIC SPENT LIQUORS Introduction Dissolution of Sodalite Dissolution at 90 C Dissolution at 160 C Characterisation of the Sodalite Seed after Dissolution Dissolution of Plant Cancrinite Comparison of Crystallisation and Dissolution Conclusion 161 CHAPTER 9. CONCLUSIONS AND RECOMMENDATIONS Conclusions Recommendations 165 REFERENCE 166 APPENDICES 186 Appendix A Si02 Concentration Analysis 186 A. 1 Worksheet of Si02 Analysis Using Molybdate Method 186 A.2 Operation Conditions of ICP Equipment 187 Appendix B Solubility of Sodium Aluminosilicate in Synthetic Spent Liquor Computation 188 B.1 Experimental and Calculated Solubility for Sodalite Seeded Experiments 188 B.2 Experimental and Calculated Solubility for Synthesised Cancrinite Seeded Experiments 189 B.3 Experimental and Calculated Solubility for Plant Appendix C Cancrinite Seeded Experiments 190 Number Frequency Density of Sodalite and Cancrinite Seed and the Products from the Seeded Experiments 191 vi

9 LIST OF FIGURES Figure 1.1 Simplified Bayer process flow chart. 4 Figure 2.1 Aluminosilicate frameworks of (a) sodalite and (b) cancrinite. 15 Figure 2.2 Hermeler et. at.' XRD patterns of (a) cancrinite, (b) and (c) intermediate phase between hydroxy sodalite and cancrinite, (d) hydroxy sodalite1 with small amounts of hydroxy sodalite2 and (e) hydroxy sodalite2. 17 Figure 2.3 Figure 2.4 Illustration of (a) 4/0 and (b) 3/1 ordering in sodalite and cancrinite. 22 The solubility of sodium aluminosilicate in Bayer liquor as a function of temperature with A1(OH)3/NaOH (molar) =0.3 1 and CS/TS (CS = weight of NaOH expressed as Na2CO3, TS = weight of NaOH expressed as Na2CO3 + weight of Na2CO3) = 0.83 at caustic concentrations of NaOH = 4.72 M, (.)NaOH = 3.77 M, (o) NaOH = 2.83 M measured by Breuer et at., (A) at NaOH = 3.88 M and AI(OH)3/NaOH (molar) = 0.69 by Arlyuk etal., (D) at NaOH = 3.35 M and Al(OH)3/NaOH (molar) = 0.56 by Eremin et at., (.) at NaOH = 4.19 M and A1(OH)3/NaOH (molar) = 0.94, and (.) at NaOH = 4.84 M and A1(OH)3INaOH (molar) = 0.81 by Shvartsman et at.. 41 Figure 2.5 Figure 2.6 Figure 2.7 The solubility of silica in sodium aluminate solution as a function of A1(OH)3INaOH molar ratio at 90 C with (o) 4 M and (.) 5.9 M NaOH determined by Ni et at., at 120 C with 3.22 and (A) 6.45 M NaOH by Kraus et at., at 150 C with 3.78 and (.) 4.72 M NaOH and at 250 C with (*) 3.78 and (.) 4.72 M NaOH by Breuer et at. 43 Desilication rates of high and low silica bauxite during Bayer digestion. 51 Si02 concentration in aluminate solutions as a function of time during the leaching of (ta) hydrargillite bauxite, (.) finely ground bauxite sinter and (o) finely ground nepheline sinter. 55 vii

10 Figure 2.8 Figure 3.1 The chemical model for the formation of solid sodium aluminosilicate. X2 = C032, S042, 20H, etc. 57 Illustration of the stainless steel reactor vessel (all dimensions in mm). 61 Figure 3.2 Comparison of Si02 concentration measured by the molybdate method and ICP. 71 Figure 4.1 XRD patterns of nucleated sodium aluminosilicate at (a) 60 C in a liquor containing M Na2SO4.10H20, (b) 70, (c) 80, (d) 90, (e) 100, (f) 150, (g) 160 and (h) 200 C all in liquors containing 0.38 M Na2CO3. 80 Figure 4.2 Figure 4.3 Figure 4.4 FTIR spectra of the sodium aluminosilicate nucleated from solution for the 0.38 M Na2CO3 solutions at (a) 70, and (b) 80 (zeolite), and at (c) 90, (d) 100, (e)150, (f) 160 and (g) 200 C (sodalite). 82 SEM micrographs of zeolite nucleated at (a) 70 and (b) 80, and sodalite nucleated at (c) 90 and (d) 100 C from liquors containing 0.1 M Si02 and M Na2CO3. 83 XRD patterns of nucleated product obtained at 80 C from a liquor containing (a) 0.38 M and (b) M Na2CO3 and at 90 C from a liquor containing (c) 0.38 M and (d) M Na2CO3. 85 Figure 4.5 FTIR spectra of the sodium aluminosilicate obtained at 90 C from the solution containing (a) and (b) 0.38 M Na2CO3. 86 Figure 4.6 Figure 4.7 XPS results of carbon is binding energy of the products obtained at 90 C before ion etching (a) from a liquor containing M Na2CO3 and (b) from a liquor containing 0.38 M Na2CO3, and after ion etching to 2.5 nm from the surface (c) from a liquor containing M Na2CO3 and (d) from a liquor containing 0.38 M Na2CO3. 87 XRD patterns of nucleated sodium aluminosilicate obtained at 160 C from a liquor containing (a) M and (b) 0.38 M Na2CO3. 89 viii

11 Figure 4.8 FTIR spectra of nucleated sodium aluminosilicate at 160 C from solutions with Si M and Na2CO3(a) and (b) 0.38M. 90 Figure 4.9 SEM micrographs of sodium aluminosilicate nucleated at 160 C from a liquor containing 0.02 M Si02 and M Na2CO3 of (a) high and (b) low magnification and a liquor containing 0.02 M Si02 and 0.38 M Na2CO3 of (c) high and (d) low magnification. 91 Figure 4.10 Figure 5.1 Figure 5.2 XRD patterns of sodium aluminosilicate obtained from a (a) low Si02 and (b) high Si02 liquors containing 0.38 M Na2CO3 at 95 and 100 C respectively. 93 XRD patterns of sodium aluminosilicate products from a synthetic spent liquor containing NaOH 4.52 M, Na2CO M, A1(OH) M and 0.1 M Si02 at 100 C. 95 Photographs of (a) the stainless steel vessel and (b) the scales formed after 360 hours of reaction time at 100 C: sodalite2 (with trigonal morphology on the wall) and Na2CO3 (white layer scale at the bottom). 97 Figure 5.3 XRD patterns of the solid products obtained at 100 C after 360 hours from (a) the wall (sodalite) and (b) the bottom of the reactor vessel (Na2CO3). 98 Figure 5.4 Figure 5.5 SEM micrographs of (a) sodalite obtained from solution and (b) Na2CO3 scale obtained from the bottom of the vessel. 99 FTIR spectra of sodium aluminosilicate products from 0.38 M Na2CO3 synthetic spent liquor at 100 C after (a) 0.5 h, (b) 22 h, (c) 64 h, (d) 118 h, (e) h, (f) 213 h, (g) h and (h) 359.5h. 100 Figure 5.6 SEM micrographs of sodium aluminosilicate obtained at 100 C ma liquor containing 0.1 M Si02 and 0.38 M Na2CO3 after (a) 3and (b) hours. 101 Figure 5.7 XRD patterns of the solid products (sodalite1 + Na2CO3 hydrates represented by peaks marked 'c' and cancrinite by ix

12 "can") obtained (a) at 100 C after 30 days in a liquor containing 0.38 M Na2CO3 103 Figure 5.8 XRD patterns of sodium aluminosilicate obtained at 200 C after (a) 6.5 hours (sodalite1) and (b) lodays (cancrinite). 104 Figure 5.9 XRD patterns of sodium aluminosilicate obtained at 150 C in a liquor containing M Na2CO3 after (a) 3, (b) 6 and (c) 10 days and in a liquor containing 0.38 M Na2CO3 after (d) 3, (e) 6 and (f) 10 days. 109 Figure 5.10 XRD patterns of sodium aluminosilicates obtained from 0.01 M Si02 solutions in a liquor containing 0.38 M Na2CO3 at 95 C after (a) 5 (sodalite1), (b) 10 and (c) 15 days (sodalite1 + cancrinite). 111 Figure 5.11 Figure 5.12 Figure 6.1 Figure 6.2 XRD patterns of the transformation of sodalite seed in synthetic spent liquor with 0.01 M Si02 and 0.38 M Na2CO3 at 160 C. 114 FTIR spectraof (a) sodalite seed and (b) the cancrinite product obtained after the experiment at 160 C. 115 Si02 equilibrium solubility of synthesised cancrinite seeded solutions as a function of solution Na2CO3 concentrations (o) at 90 C and (.) at 160 C and ( ) predicted result by Equation 6.1, parameters from Table FTIR spectra of synthetic cancrinite seed before the solubility experiment (a) and after the experiments at (b) 90 C, with MNa2CO3 121 Figure 6.3 Figure 6.4 XRD patterns of (a) synthesised cancrinite seed and cancrinite obtained after solubility experiments at 90 C in a liquor containing (b) and (c) 0.38 M Na2CO3 and at 160 C from a liquor containing (d) and (e) 0.38 M Na2CO Si02 equilibrium solubility of plant cancrinite in synthetic spent liquor as a function of Na2CO3 concentrations (o) at 90 C from above (initial Si02 concentration 0.01 M), (.) at 160 C from. above (initial Si02 concentration 0.01 M) and at 160 C from below (initial Si02 concentration 0 M) and ( ) predicted by Equation 6.1, parameters from Table x

13 Figure 6.5 Figure 6.6 Figure 6.7 FTIR spectra of plant cancrinite after being in liquors containing 0.38 M Na2CO3 and, (a) 0.0 and (b) 0.O1M Si02 initial concentrations for 14 days at 160 C 127 Silica equilibrium solubility of sodalite seeded solutions as a function of solution Na2CO3 concentrations (o) at 90 C from above Si02 equilibrium solubility, (.) at 160 C from above 5i02 equilibrium solubility, at 160 C from below Si02 equilibrium solubility, Breuer et al. at 90 C, Breuer et al. at 160 C and (A) Ni et al. at 90 C and ( ) predicted by Equation 6.1, parameters from Table The solubility of (o) sodalite seed, (A) synthesised cancrinite seed and (.) plant cancrinite seed in synthetic spent liquor at (a) 90 and (b) 160 C as a function of Na2CO3 concentration. 131 Figure 7.1 Solution 5i02 concentration in unseeded desilication experiments as a function of time (with 0.01 M Si02 initial concentration) (o) at 90 C and M Na2CO3, (.) at 90 C and 0.38 M Na2CO3, (A) at 160 C and M Na2CO3, (A) at 160 C and 0.38 M Na2CO3, (o) Si02 solubility at 90 C and M Na2CO3, (.) solubility at 90 C and 0.38 M Na2CO3, (.) solubility at 160 C and M Na2CO3, and (.) Si02 solubility at 160 C and 0.38 M Na2CO Figure 7.2 Figure 7.3 Figure 7.4 Solution Si02 supersaturation in unseeded desilication experiments as a function of time (with 0.02 M Si02 as initial concentration) (o) at 90 C and M Na2CO3, (.) at 90 C and 0.38 M Na2CO3, (A) at 160 C and M Na2CO3 and (A) at 160 C and 0.38 M Na2CO Solution Si02 supersaturation in precipitation experiments in the presence of 5 g dm3 sodalite seed with an initial concentration of 0.01 M Si02 as a function of time (o) at 90 C and M Na2CO3, (.) at 90 C and 0.38 M Na2CO3, (A) at 160 C and M Na2CO3 and (A) at 160 C and 0.38 M Na2CO FTIR spectra of (a) sodalite seed, and the crystalline product obtained from (b) a liquor containing M Na2CO3 at 90 C, (c) a liquor containing 0.38 M Na2CO3 at 90 C, (d) a liquor xi

14 containing M Na2CO3 at 160 C and (e) a liquor containing 0.38 M Na2CO3 at 160 C. 138 Figure 7.5 Figure 7.6 Figure 7.7 SEM micrographs of sodalite seed (a) and sodalite obtained after 4 hours from a sodalite seedeçl crystallisation experiment at 160 C from a liquor containing (b) and (c) 0.38 M Na2CO Solution 5i02 supersaturation of sodalite seeded experiments at 90 C as a function of time for a liquor containing (o) 0.01 M Si02 and M Na2CO3, (.) 0.01 M Si02 and 0.38 M Na2CO3, 0.02 M Si02 and M Na2CO3 and (A) 0.02 M Si02 and 0.38 M Na2CO Particle volume distribution of (o) sodalite seed, sodalite obtained from a liquor containing 0.01 M Si02 and (.) M Na2CO3 at 90 C, 0.38 M Na2CO3 at 90 C, (A) M Na2CO3 at 160 C and (D) 0.38 M Na2CO3 at 160 C. 141 Figure 7.8 Volume cumulative size distribution of (o) sodalite seed and (.) crystals obtained from sodalite seeded experiment in a liquor containing 0.38 M Na2CO3 at 160 C. 142 Figure 7.9 Figure 7.10 Figure 7.11 Number frequency density function versus particle size of (o) sodalite seed, (.) sodalite after 4 hours in a liquor containing 0.01 M Si02 and M Na2CO3 at 90 C with stirring of 400 rpm and sodalite after 4 hours in water at 90 C with stirring 400rpm. 143 Si02 supersaturation of precipitation experiments in the presence of plant cancrinite seed as a function of time in a liquor containing (o) M Na2CO3 at 90 C, (.) 0.38 M Na2CO3 at 90 C, M Na2CO3 at 160 and (A) 0.38 M Na2CO3 at 160 C. 144 Si02 supersaturation in plant cancrinite seeded synthetic spent liquor at 160 C as a function of reaction time (o) and (.) in a liquor containing 0.01 M Si02 and 0.38 M Na2CO3 and in a liquor containing 0.02 M Si02 and 0.38 M Na2CO xii

15 Figure 7.12 Figure 7.13 Figure 7.14 Figure 7.15 Figure 7.16 Figure 8. 1 Figure 8.2 Figure 8.3 Si02 supersaturation as a function of time in a liquor containing 0.38 M Na2CO3 at 160 C with cancrinite seed added after 40 minutes. 146 Particle size distribution of (o), cancrinite seed, cancrinite obtained from a liquor containing 0.01 M Si02 and (.) M Na2CO3 at 90 C, 0.38 M Na2CO3 at 90 C, (A) M Na2CO3 at 160 C and 0.38 M Na2CO3 at 160 C. 147 Number frequency density function of (o) cancrinite seed, and the products obtained from cancrinite seeded experiments with a liquor containing 0.01 M Si02 and (.) M Na2CO3 at 90 C, (A) 0.38 M Na2CO3 at 90 C, (A) M Na2CO3 at 160 C and 0.38 M Na2CO3 at 160 C. 148 Si02 supersaturations as a function of time at 160 C in liquors containing 0.01 M Si02 and 0.38 M Na2CO3 with (o) no seeding, (.) sodalite seeding, (surface area, 3.2 m2 g1) and (A) plant cancrinite seeding, (surface area, 2.3 m2 g1). 149 Comparison of solution Si02 concentration (o) determined by experiment, predicted by n=1 and n=2 (lined curves) in (1) nucleation, (2) sodalite seeded and (3) cancrinite seeded experiments at 160 C in liquors containing 0.38 M Na2CO Solution Si02 concentration as a function of time at 90 C in sodalite seeded dissolution experiments in a liquor containing (o) M Na2CO3, (.) 0.38 M Na2CO3, Si02 equilibrium solubility of sodalite determined by approach from above after 14 days in a liquor containing (1) and (2) 0.38 M Na2CO Solution Si02 concentration in sodalite dissolution experiments at 160 C in a liquor containing (o) M Na2CO3, (.) 0.38 M Na2CO3, Si02 solubility of sodalite transformed to cancrinite determined by approach from above after 14 days in a liquor containing (1) and (2) 0.38 M Na2CO SEM micrographs of sodalite dissolution products obtained at 90 C from a liquor containing (a) and (b) 0.38 M Na2CO3 and at 160 C from a liquor containing (c) and (d) 0.38 M Na2CO xlii

16 Figure 8.4 Particle size distribution of (o) sodalite seed and crystalline products obtained from sodalite dissolution experiments (.) at 90 C from the liquor containing M Na2CO3, at 90 C from the liquor containing 0.38 M Na2CO3, (A) at 160 C from the liquor containing M and (D) at 160 C from the liquor containing 0.38 M Na2CO Figure 8.5 Figure 8.6 Figure 8.7 Si02 concentration in plant cancrinite dissolution experiments in a liquor containing (o) M and (.) 0.38 M Na2CO3 at 90 C, Si02 solubility of plant cancrinite measured from precipitation after 14 days in a liquor containing (1) M and (2) 0.38 M Na2CO3 at 90 C. 159 Solution Si02 concentration in plant cancrinite dissolution experiments at 160 C in a liquor containing (o) M Na2CO3 and (.) 0.38 M Na2CO3 and Si02 solubility of plant cancrinite determined from above after 14 days in a liquor containing: (1) M Na2CO3 and (2) 0.38 M Na2CO Approach of 5i02 concentration to equilibrium during (o) crystallisation and (.) dissolution at 160 C from liquors containing 0.38 M Na2CO xiv

17 LIST OF TABLES Table 1.1 Major minerals present in some bauxites (Wt%). 2 Table 1.2 Table 2.1 Concentration of solutions used in the work herein, described in M" and in industrial terminology. 7 Bond distances (A) and bond angles (degrees) of geometryoptimised (OH)3Si-(OH)-Al(OH)3 dimer 11 Table A1 and 29Si chemical shifts of the aluminosilicate solutions 13 Table 2.3 Four typical minerals of the sodalite group. 15 Table 2.4 Infrared spectra assignment of sodalite (or nosean) and cancrinite frameworks. 18 Table 2.5 FTIR spectra of water and anions in sodium aluminosilicate (cmt). 19 Table 2.6 Table Si MAS NMR chemical shifts d, unit cell sizes a and Si-O-Al bond angles a of some sodalites. 21 Comparison of solid products from solution and aluminosilicate gel at 80 and 100 C. 23 Table 2.8 Unit cell sizes of four cancrinites. 25 Table 2.9 Weight percentage from XRD analysis of some plant scales. 28 Table 2.10 The relationship between temperature, reaction time and crystallised products. 29 Table 2.11 Table 2.12 Table 2.13 Table 2.14 Unit cell sizes of the sodalite phase with different anions before and after equilibration. 32 Products obtained in synthesis with NaNO3 addition to the solutions at 80 C. 34 Thermal stability of some sodalites and cancrinites with imbibed salts. 35 Influence of anions on products of aluminosilicate synthesis at Si/Al=l. 36 xv

18 Table 2.15 DSP phases in QAL as a function of temperature. 37 Table 2.16 The experimental conditions of some of the studies published. 53 Table 2.17 Kinetic data from published results. 54 Table 3.1 The experimental conditions of the unseeded desilication experiments. 67 Table 3.2 Binding energies of the elements in sodium aluminosilicate 77 Table 4.1 Table 4.2 Table 4.3 Table 5.1 The experimental conditions and the products of sodium aluminosilicate synthesis in liquors containing M Na2CO3. 79 The experimental conditions and the products of sodium aluminosilicate synthesis in liquors containing 0.38 M Na2CO3. 79 The percentage of C032 in samples shown in Figure 4.6 before and after ion etching. 88 The comparison of the XRD peak intensities and unit cell size of the sodalite obtained from low and high C032 solutions. 105 Table 5.2 The phases and unit cells obtained at 150 C from Parr Bombs. 108 Table 5.3 Table 5.4 Unit cells of sodium aluminosilicate obtained from a solution containing 0.01 M Si02 and 0.38 M Na2CO3 at 95 C in an atmospheric reactor. 110 The experimental conditions and relative peak intensities of sodalite seeded products at 90 C. 112 Table 5.5 Unit cell size change of sodalite seeded experiments at 160 C. 115 Table 6.1 Table 6.2 Table 6.3 Table 6.4 Parameters estimated for Equation 6.1 Si02 equilibrium solubility of synthesised cancrinite in synthetic spent Bayer liquor. 122 The unit cell and volume of the cancrinite phase before and after the solubility experiments. 122 Parameters estimated for the model of Si02 equilibrium solubility (Equation 6.1) of plant cancrinite in synthetic spent Bayer liquor. 125 The unit cell and volume of plant cancrinite seed crystals before and after the solubility experiments. 126 xvi

19 Table 6.5 Parameters estimated for Equation 6.1 of Si02 equilibrium solubility in sodalite seeded synthetic spent Bayer liquor. 129 Table 6.6 The difference of Si02 equilibrium solubility between 160 and 90 C in synthetic spent liquor for 3 types of seeds under the same Na2CO3 concentration. 131 Table 7. 1 Table 7.2 Table 7.3 Unit cells and the average linear size increase of the product crystals in the sodalite seeded experiments. 140 Unit cell sizes and the average linear size increase of the product crystals for the cancrinite seeded experiments. 146 Comparison of sodalite and plant cancrinite desilication at 90 and 160 C in a liquor containing 0.01 M Si02 and 0.38 M Na2CO Table 7.4 The results of kinetic regression. 150 Table 8.1 Table 8.2 Unit cells of the crystalline products obtained from the sodalite dissolution experiments. 154 Unit cells of crystalline phase products obtained from plant cancrinite dissolution experiments. 160 Table A.1 Worksheet for Si02 concentration calculation. 187 Table B.1 Table B.2 Table B.3 Table C.1 Table C.2 The experimental and calculated Si02 solubility data of sodalite seeded system. 188 The experimental and calculated Si02 solubility data of synthesised cancrinite seeded system. 189 The experimental and calculated Si02 solubility data of plant cancrinite seeded system. 190 The number frequency density function of sodalite seed and the products from sodalite seeded experiments with 0.01 M Si The pumber frequency density function of cancrinite seed and the products from cancrinite seeded experiments with 0.01 M Si xvii

20 PUBLICATIONS "The influence of sodium carbonate on the formation of aluminosilicate and its solubilities" (Journal of Crystal Growth, 171 (1997) ) Zheng, K.; Gerson, A. R.; Addai-Mensah J. and Smart, R. St.C. "Bayer process plant scale: transformation of sodalite to cancrinite" (Journal of Crystal Growth, 171 (1997) ) Gerson, A. R. and Zheng, K. "The precipitation mechanism of sodium aluminosilicate scale in Bayer plants" (Light Metals, (1997) 23-28) Addai-Mensah, J.; Gerson, A. R.; Zheng, K.; O'Dea, A. and Smart, R. St.C. "The solubility of sodium aluminosilicates in synthetic Bayer liquor" (submitted to Journal of Crystal Growth) Zheng, K.; Gerson A.; R. Addai-Mensah, J. and Smart, R. St.C. "Kinetic studies of nucleation and crystal growth of aluminosilicate in synthetic spent Bayer liquor" (in preparation) Zheng, K.; Gerson, A. R.; Addai-Mensah, J. and Smart, R. St.C. "Sodium Aluminosilicate scale Control in the Bayer Process" (AUSTRALIA Patent Act 1990) Ian Wark Research Institute, University of South Australia (Zheng, K as one of the inventors) CONFERENCE PAPERS "The influence of sodium carbonate on aluminosilicate scale formation and silica solubility in sodium aluminate solutions" (The 18th Australian Colloid and Surface Chemistry Student Conference, February, 1995) Zheng, K.; Gerson, A. R.; Addai-Mensah J.; Ralston, J. and Smart, R. St.C. "The solubilities of aluminosilicate in synthetic spent Bayer liquor" (RACI 10th National Convention, September, 1995) Zheng, K.; Gerson A. R.; Addai-Mensah, J. and Smart, R. St.C. "The Mechanism of Sodium Aluminosilicate Scale Formation in Alumina Refineries" (The 4th International Alumina Quality Workshop, July 1996, Darwin, Australia) Gerson, A. R.; Addai-Mensah, J.; Zheng, K.; O'Dea, A. and Smart, R. St.C. PRIZES Postgraduate Research Student Prize 1996 (University Foundation, University of South Australia) xviii

21 ABSTRACT The Bayer process uses hot caustic soda solution to extract aluminium hydroxide from bauxite leaving impurities such as hematite, rutile and ilmenite undissolved. Silicon containing minerals such as kaolinite and quartz also dissolve to some extent in hot caustic solution. Much of the dissolved silica reacts with sodium aluminate to precipitate as sodium aluminosilicate. The time needed for DSP precipitation to reach equilibrium, however, is much longer than that required for bauxite dissolution, hence the remaining dissolved silica passes into the Bayer cycle. After gibbsite precipitation the concentration of aluminium hydroxide is reduced to half. Silica solubility is also reduced when the A1(OH)3/NaOH molar ratio of the liquor decreases from 0.72 to The aluminate solution after gibbsite precipitation (spent liquor) still contains a significant concentration of aluminate and is recycled to the digester through a series of heat exchangers. Sodium aluminosilicate scale forms on the surfaces of the heat exchanger pipes, tanks and throughout the plant. The formation of sodium aluminosilicate scale causes severe difficulties in alumina refinery plants, especially the fouling and blockage of the heat exchange pipes, resulting in low heat transfer efficiency. Valuable amounts of alumina and caustic soda are also lost with the silica which precipitates as sodium aluminosilicate. An extensive chemical cleaning with process shutdown is needed every fortnight. The acid solution used for cleaning, attacks the heat exchanger tubes, hence frequent replacement is needed. The cost of scaling to Queensland Alumina Limited is estimated to be in excess of $10 million per annum. This project focuses on the fundamental aspects of the mechanism of sodium aluminosilicate scale formation in low and high pressure heat exchangers of a Bayer plant. It has been reported that zeolite, sodalite and cancrinite are the phases that form in Bayer liquor, but very little is known about the mechanism of crystallisation of these aluminosilicate phases under conditions prevailing in an alumina plant. This project investigates scale formation in pure, synthetic spent liquor. In order to mimic plant spent liquor concentrations, the experimental conditions have been chosen to be 3.77 to 4.52 M NaOH, Al(OH)3INaOH molar ratio 0.36, or 0.38 M Na2CO3 with a Si02 concentration of 0.01 to 0.1 M. Seeded and unseeded isothermal, batch crystallisations have been carried,..out at temperatures from 60 to 200 C. Synthesised sodalite and cancrinite, and plant cancrinite have been used as seed. Experiments have been carried out in Parr Bombs, an autoclave and a 316 stainless steel reactor. Characterisation of the solid products has been carried out by X-ray diffraction (XRD), Fourier transform infrared spectroscopy (FTIR), scanning electron microscopy (SEM) with energy dispersive spectroscopy (EDS) and X-ray photoelectron spectroscopy (XPS). xix

22 Between 60 to 80 C the nucleated sodium aluminosilicate scale was a zeolite phase and between 90 to 200 C sodalite phases. Two phases of sodalite may be observed. Both are cubic, however, sodalite2 corresponds to a low CU32- concentration in the lattice and has a smaller unit cell than sodalite1 which corresponds to high CU32- concentration in the lattice. The presence of a high C032- concentration in the solution, 0.38 M Na2CO3, prevents the formation of sodalite2 to some extent. Sodalite1 may, partly or totally, transform to sodalite2 by the loss of C032. Sodalite1 and sodalite2 may, on the other hand, transform to cancrinite. The phase transformation of sodalite to cancrinite is promoted by high Na2CO3 concentrations and long reaction times and high temperatures. The crystalline phase changes are non-reversible in synthetic spent liquor between 90 to 200 C. The effect of solution C032 concentration on crystalline phase change is complicated: at high temperatures (> 150 C), high C032 concentrations promote cancrinite formation; at low temperatures (< 100 C), low solution C032 promotes cancrinite formation. A certain degree of Si02 supersaturation is needed for the nucleation to occur. At low Si02 concentrations ( M) metal substrate-influenced heterogeneous nucleation of aluminosilicate scale particles occurs at the surfaces of metal substrates. Thereafter, the Si02 supersaturation is further decreased by the growth of the scale layer. At high Si02 concentration (0.1 M) the crystallisation of aluminosilicate scale at the metal substrate surface and colloidal crystals in bulk solution both occur. Seeding the liquor with either sodalite or cancrinite prevents heterogeneous nucleation of scale at metal surfaces as Si02 desupersaturation occurs through seed crystal growth. The order of solution desilication rate with different seeds under similar conditions is: plant cancrinite> synthetic sodalite > unseeded. influences the equilibrium 5i02 concentration in the liquor. The presence of different crystalline phases The rate constants estimated for the crystallisation of sodalite and cancrinite showed that the concentration of Na2CO3 in solution had a significant effect on the rate. It appears that increasing the Na2CO3 increased the rate constant. The solubility of the phases in order is: synthetic sodalite> plant cancrinite> synthesised cancrinite. Systematic increases of Na2CO3 concentration in synthetic liquor (0.043 to 0.38 M) cause the Si02 equilibrium concentration to decrease by about 30%. Cancrinite solubility in the liquor increases with increase of temperature. As the sodalite to cancrinite phase transformation occurs faster at 160 C than at 90 C, sodalite seeded liquor at 160 C tends to give rise to a Si02 solution concentration similar to that at 90 C. xx

23 Both crystallisation and dissolution of sodalite in synthetic liquor may be accompanied by crystalline surface phase transformations. Crystallisation and dissolution of sodalite and cancrinite are also strongly influenced by temperature. The higher the temperature, the faster the crystallisation and dissolution of the seed crystals. The rates of desilication of unseeded liquor and concomitant scale/crystal formation increase with increasing temperature. The mechanism of formation of sodium aluminosilicate crystalline products in synthetic spent Bayer liquor and transformations underlying their formation were found to be similar to those occurring in a Bayer alumina plant. xxi

24 AUTHOR' S STATEMENT I declare that this thesis does not incorporate without acknowledgment any material previously submitted for a degree or diploma in any university; and that to the best of my knowledge it does not contain any materials previously published or written by another person except where due reference is made in the text. Kali Zheng xxii

25 ACKNOWLEDGMENTS I would like to thank Prof. John Ralston for offering me a scholarship to study in the Ian Wark Research Institute, providing excellent working conditions, including modern equipment, as well as academic support in the first two years and financial assistance for me to attend two national conferences. I would like to thank Prof. Roger Smart for academic support and encouragement, for his thoughtful care of the progress of the project including borrowing Parr Bombs from interstate at the beginning of the study, for many fruitful discussions on FTIR and XPS analysis and other aspects of this thesis; to thank Dr. Andrea Gerson for extensive discussions on crystalline phase characterisation, on XRD, and on many other fundamental studies, for teaching me the operation of XRD and the calculation of unit cells; to thank Dr. Jonas Addai-Mensah for helpful support and discussions on nucleation and crystal growth kinetics; and also to thank Dr. Donald Mair for some help during the first year of my study; I would like to thank Dr. Benoit Crystal, Dr. Jurg Wehrli, Dr. Tony Piccaro, Dr. Devlet Sizget and Dr. Anne Duncan in Queensland Alumina Limited and Dr. Janine Lay, Dr. Ray Shaw, Dr. Darren Rodda, Dr. Gerry Roe and Mr. Warren Staker in Comalco Research Centre for organising my trips to QAL and CRC and useful discussions of practical problems in plant process, I would also like to thank Sherron Hunter and all other library staff members for assistance with literature searching; Len Green in Metallurgy for undertaking SEM and EDS measurements; Queensland Alumina Limited for carrying out chemical XRF analysis for some crystal samples; Comalco Research Centre for having me there to use their computer programme to do kinetic regression of my experimental data; Pawittar Arora for performing XPS analysis; Andrew Robinson for the BET surface area measurements; Philip Souter for help in making some experimental apparatus and all other staff members in the School of Chemical Technology for their assistance. I would also like to thank Queensland Alumina Limited and Comalco Research Centre for financially sponsoring my Ph. D. project. Finally, I would like to thank my husband Minghua for his constant help with his computer knowledge and the support and love from Minghua and my son David; both were much needed to complete this work. xxiii

26 CHAPTER 1. INTRODUCTION 1.1. The Aluminium Industry The aluminium industry was founded in 1854 by In the early days aluminium chloride was reduced by sodium to obtain substantially pure aluminium metal. industry was launched in 1855 when ingots of aluminium were exhibited at the Paris Exhibition. The rapid development of the industry is due not only to the substitution of aluminium for other scarcer or more costly metals, but also to the exceptional combination of properties possessed by aluminium It is light, soft and flexible and has high thermal and electrical conductivities, high reflectivity and good resistance to a variety of corrosive environments, including industrial and marine atmospheres. The It is subject to electrolytic polishing and to chemical and electrolytic oxidation, with the formation of hard or soft, coherent, protective, electrically-insulating oxide films. This passivation film may also be dyed to produce decorative finishes such as in paint. Its application is greatly extended by alloying it with such elements as magnesium, silicon, iron, copper, manganese and zinc to give commercially suited products of higher strength than the pure metal. The properties of these alloy products are especially suited to particular fabricating processes, including casting, rolling, fabricating, extrusion, twisting, deep drawing and machining. The aluminium industry is one of the newest and fastest growing industries in Aluminium was first produced in Australia in 1955 when operations started at Bell Bay in Tasmania. Production was based on imported bauxite and partly on imported alumina. During the l950s huge deposits of bauxite were discovered in Australia. Since 1976 Australia has been the worlds largest alumina producer. In 1979 Australia produced 7.4 million tons of alumina or about one-quarter of world Currently, the alumina industry is placed third after gold and coal, as Australia's leading export earners. The production of aluminium metal involves two distinct In the first process, hydrated alumina (gibbsite) and boehmite are extracted by hydrometallurgical means from bauxite and converted into purified, calcined alumina. This is the Bayer process. Its detailed description is given below. In the second process, the pure calcined alumina is dissolved in molten çryolite in a carbon lined reduction cell and converted into the primary metal by an electrolytic 1

27 1.2. The Bayer Process Almost all of the pure alumina used by the aluminium industry is manufactured by the Bayer process invented and patented by Karl Josef Bayer in Austria in 1 (the seeded crystallisation of gibbsite) and 1892[81 (pressure extraction). Since then gibbsite has been extracted industrially from aluminium bearing bauxite Following continuous innovation by European companies, the Bayer process established itself as the most economic alumina extraction process. However it took 50 years of laborious developments for the Bayer process to become dominant in the manufacture of alumina. Since 1940, the Bayer process has been developed into the modern process which is known today. The Bayer process takes advantage of the fact that aluminium, present in bauxite ore as gibbsite or boehmite, is soluble in hot sodium hydroxide solutions while the iron and titanium bearing minerals do not react with caustic. Silica containing minerals dissolve or partly dissolve in the liquor (depending on the digestion conditions) which has a limited silica capacity. The bauxite ore has a complex mineralogy. The major minerals present in some bauxite ores are shown in Table i.i[101. Apart from those listed in Table 1.1, carbonate, oxalate and other organic impurities are also present in bauxite. Table 1.1 Major minerals present in some bauxites (Wt%)1101. Mineral Weipa Gove ALCOA WA Greece Jamaica Guinea Boke Brazil Trom -betas Gibbsite, Al(OH) Boehmite, AlO(OH) Diaspore, A11O(OH) Kaolinite, Al203.2SiO2.2H Quartz, 5i Hematite, Fe Goethite, FeO(OH) Anatase, rutile, Ti Total The conditions under which the bauxite is digested, i.e. temperature, time, pressure and caustic concentration, depend upon the types of aluminium containing minerals in the Gibbsitic (Al(OH)3) bauxites are normally digested at temperatures between 2

28 130 to 150 C, although a temperature of atmospheric digestion may be used. Boehmite (A100H) bearing bauxites have to be digested at C. Normally some lime (CaO) is added along with the bauxite to the digesters to aid the precipitation of impurities. After the aluminium containing minerals have been extracted, the digested slurry is flashed down to atmospheric pressure ( 100 C) to cool. The flashed off steam is used to preheat the spent liquor that is returned to the digesters to digest more bauxite. This preheating is done in shell and tube heat exchangers, where severe scaling takes place on the tube side (refer to Section 1.3). The cooled slurry is first decanted, and the solids, commonly called "red mud" because of their red colour, are separated from the supersaturated liquor by The filtered liquor is then further cooled to 70 C to supersaturate as pregnant liquor. The liquor is fed to large precipitation tanks, charged with fine gibbsite seeds and agitated for periods of hours to cause gibbsite precipitation. About half of the dissolved alumina in the liquor precipitates. The precipitation slurry is then sent to classifiers where the solids are classified into coarse and fine fractions. The coarsest fraction is washed and dewatered well to remove all liquor and is then calcined at C to give anhydrous alumina (A1203). The fine fraction is returned to the precipitation circuit to act as seed for the next cycle. The spent liquor is strengthened by caustic soda addition to the desired level to make up for caustic losses and returned to the digester to digest more bauxite after The overall process may thus be represented by two 111: Extraction Al(OH)3 ore + NaOH A1(OH)4 (1.1) Crystallisation 2Al(OH)3 + 3H20 (1.2) Calcination Figure 1.1 is a schematic illustration of the Bayer process [51 Normally the Bayer process is suitable for high grade bauxite in which the aluminium occurs predominantly as gibbsite or a mixture of gibbsite and diaspore (A100H) with low silica When the silica consent is higher than 7%, the Bayer process becomes uneconomic, due to caustic and aluminium losses associated with high silica concentration due to the crystallisation of sodium aluminosilicate. A combination process of lime (CaO) soda (Na2CO3) and lime has been developed for high Si02 bauxites. The red mud resulting from the digestion step containing the valuable aluminium containing compounds and soda, is sintered at around C with requisite amounts of soda ash and 3

29 limestone (CaCO3) to convert all the aluminium content to sodium aluminate (NaA1(OH)4) and silica to dicalcium silicate (2CaO.Si02) respectively. The latter is only slightly soluble. The sinter is then leached with caustic soda solution. The leached liquor contains the alumina and soda and the slightly soluble dicalcium silicate is left as solid residue. The liquor is returned to the digesters for the next cycle of digestion. Bauxite 1 - E11 : Mud Wash and Disposal Pregnant Mud Separator Liquor Fine Fraction (Seed) Coarse Fraction Kiln End Product Calcinated Alumina Figure 1.1 Simplified Bayer process flow 516] Alumina can also be extracted from clay by the lime sinter process The clay is roasted between 1200 and 1350 C with the requisite amount of limestone to convert all the Al content to an alkali soluble 5CaO.3A1203 and the Si02 to 2CaO.5i02 The aluminium containing material is subsequently dissolved in dilute sodium carbonate solution to produce sodium aluminate solution as: 4

30 5CaO.3A Na2CO3 + 12H20 * 5CaCO3(s) + 6NaA1(OH)4(aq)+4NaOH(aq) (1.3) Due to the amphoteric nature of aluminium, acid processes, acid salt processes and Vereinigte-Aluminium-Werks-European Bayer Process (VAW-EBP) (calcining boehmite containing 12-18% silica at C so as to decompose the kaolin to amorphous silica and alumina) have also been developed for alumina 1.3. Scaling Problems Silica in bauxite can occur in many forms, including kaolinite (A1203.2SiO2.2H20), halloysite (Al203.2SiO2.4H20), quartz (Si02), kyanite (A1203. S i02), the montmorillonite group, feldspathic silicates and alumina silica The aluminium-containing minerals in bauxite are dissolved in hot caustic solution, as are quartz and kaolinite, the two major silica In the Bayer process, the behaviour of the silica is dependent on the nature of the compounds and the extraction temperature. The kaolinite minerals dissolve rapidly, even at moderately elevated temperatures, and are usually completely dissolved during the first few minutes of digestion. Quartz is only attacked at the high temperatures used to treat boehmite ( OC)[5 The dissolved silica precipitates as sodium aluminosilicate scale throughout alumina The exact nature of the scale depends on the conditions of formation Some of the dissolved silica is re-precipitated as and/or cancrinite, both Na8[A1SiO4]6X2.nH2O, or as tricalcium aluminate silicate in the presence of lime under certain In sodalite and cancrinite the 'X' can be l/2s042, Cl-, 0H, or a mixture of these and possibly other anions. The process of formation can be written as: [A1SiO4]6X2.nH2O(s)+(6-n)H ff Aluminosilicate scale (1.4) There are many formulas for sodalite and cancrinite in the literature with different proportions of sodium, aluminium, silicon, water and anions. The formulas are presented in many different ways. To make the discussion easier, the formulas of both sodalite and cancrinite in later chapters will be expressed in the form Na8(A1SiO4)6X2.nH2O. Pregnant Bayer liquor i.e. prior to gibbsite precipitation, contains 0.01 M Si02 and has the equilibrium concentration of Si02 of approximately 4.2x103 to 5.0xl03 As the desilication reaction required to achieve near equilibrium Si02 concentration takes 4-5 times as long as bauxite dissolution, SiO2 equilibrium is not achieved. It has been shown that the equilibrium solubility of Si02 in Bayer liquors decreases with decreasing alumina 5

31 Hence after gibbsite precipitation, the resulting spent liquor is further supersaturated with Si02. Consequently desilication occurs during spent liquor recycle in the Bayer circuit. Aluminosilicate scale forms on the surfaces of heat exchanger tubes used to preheat the spent liquor being recycled for further bauxite digestion. examined the distribution of Si02 impurities in Al203 with respect to its origin within the Bayer plant. About 65-70% of the Si02 originated from the Bayer process plant, and 30-35% from the lining material of the rotary-kiln used for calcination. This thesis mainly deals with the scale formed in heat exchangers from spent Bayer liquors. The formation of sodium aluminosilicate scale causes severe difficulties in alumina refinery plants. Longer digestion times are required to achieve more complete Loss of soda and alumina occur during DSP scale formation results in reduced heat transfer in the heat exchangers due to increased fouling (heat transfer coefficients may fall by as much as 75%[26 281), 19; 29-30], contamination of the gibbsite product by surface reduced residence times through decreased flow volumes, the need to construct spare equipment to 27] ensure the continuity of the production limitations from reduced flows, blockages from detached scale and the need for scale removal (chemical or Depending upon temperatures and the silica concentration of the liquor, high temperature heater operation times can be as low as five days and, in often less than bi-weekly intervals, chemical cleaning of the heat exchanger tubes is required126 Scaling is more severe in the high temperature heat exchangers of plants due to desilication rates that increase exponentially with 7; 28] The scale must be removed mechanically or by cleaning with strong acids at which times individual heater units have to be shut down (spares are generally available). Chemical cleaning (sulphuric acid 10% followed by 8%) takes approximately 24 hours. The acid solutions attack the heater tubes. The duration of mild steel pipe at Queensland Alumina Limited (QAL) is only one year. Digesters at QAL have a 6-12 month 'operating life before descaling. The volume reduction at that time is approximately 30 to 40 %. The descaling in this case is Scale removal incurs great expense and impacts directly on the cost of the product. The cost of scaling to QAL is estimated to be in excess of $10 million per annum. In order to alleviate this problem, it is important to understand the factors controlling scale formation Synthetic Spent Bayer Liquor The composition of Bayer liquor may vary from plant to plant. The spent Bayer liquor that is referred to in this text has a composition as shown in Table 1.2. The terminologies used in this thesis are different from those used in the industry. The equivalent 6

32 concentrations of species in the simulated, synthetic, spent Bayer liquor are shown in Table 1.2. Table 1.2 Concentration of solutions used in the work herein, described in "M't and in industrial terminology. Caustic (C) Alumina (A) 1 A/C Si02 Na2CO3 *Na2CO3 Industry 240 g dm3 84 g dm (by This thesis (as Na2CO3) (as A1203) weight) g dm-3 g dnr3 g dm M NaOH 1.65 M Al(OH)3 *Due to 2.5% Na2CO3 impurity in NaOH 1.5. Aims of this Research 0.36 (by mole) 0.01 M 0.38 M M The aim of this project was to determine the fundamental mechanism of the formation of sodium aluminosilicate scale under conditions close to those prevailing in Bayer plant heat exchangers. As discussed in the next chapter, copious literature is available regarding the formation of sodium aluminosilicate solid phases by hydrothermal or gel methods, however the reaction conditions, such as temperature, caustic concentration, aluminium concentration, silica concentration, Si/Al ratio and impurities are different from those in a spent Bayer liquor. In determining the direction of the research undertaken, a series of questions were formulated. (1) Why does sodium aluminosilicate form in highly alkaline sodium aluminate solutions with relatively low silica concentrations? What is the driving force for such scale formation and how high is the solubility of sodium aluminosilicate in spentbayer liquor? What factors affect the solubility? (2) Most literature suggests that sodalite is the only scale phase formed in Bayer liquors. Is there any precursor to sodalite? Are there any polymorphs or solid phase transformations? At what temperatures do zeolite, sodalite and cancrinite form? Can we obtain any evidence about solid phase transformations? (3) What role does carbonate play during the formation of sodium aluminosilicate scale? (4) What factors affect the rate of crystallisation of sodium aluminosilicate in spent Bayer liquor? Can we decrease or increase the scaling rates? 7

33 Understanding the mechanism of scale formation is a necessary precursor to reducing the scale problem, both by indicating how scale growth occurs, and suggesting alternative descaling Investigations were carried out isothermally in batch, unseeded and seeded crystallisation experiments from 60 to 200 C with a liquor concentration range: 0.0 to 0.1 M Si02, 3.78 to 4.52 M NaOH, 1.65 to 1.84 M Al(OH)3 and to 0.38 M Na2CO3. In seeded crystallisation experiments synthesised sodalite and cancrinite, and plant cancrinite crystals were used. 8

34 CHAPTER 2. LITERATURE REVIEW 2.1. Introduction Much literature is available that examines the crystallisation of aluminosilicates. However, the crystallisation conditions used seldom approach those in the Bayer process. The literature that does exist mainly deals with zeolite synthesis and covers aspects of crystallisation including crystal structure, chemical composition, solution speciation and standard characterisation procedures. Some of the findings may be relevant to aluminosilicate crystallisation from Bayer liquors. However, dramatic differences in crystallisation conditions renders assumptions concerning crystallisation in Bayer liquors Crystallisation of Sodium Aluminosilicate Phases Aluminosilicate crystallisation is, indeed, a very difficult research area due to the unobservable phenomenon of the nucleation and the subsequent very complex crystallisation The scale forming species in Bayer liquors and the reactions which transform them into crystalline product are not clearly defined Solution Species in Spent Bayer Liquor A number of Al containing species have been reported in the literature. These citings have been obtained under a wide variety of conditions and are not all in agreement. gave the structures of the aluminium ion and the aluminate ion as and Al(OH)4(H20)2 respectively. Both ions can be envisaged as a small central in an octahedral co-ordination with the ligands. and Dubrovinskii, Iskhakova and studied absorption spectra of sodium aluminate solutions (Al 0.39 to 1.76 M, Al(OH)3/NaOH molar ratio 0.55 at about ph 14) and found two characteristic absorption bands at and nm. It was presumed that the absorption bands of Xmax. at 230 and nm were due to the formation of the species and Al(OH)4(H20)2, respectively. In apparent contradiction to the observation of Al(OH)4(H20)2 it was believed that above the ph of 8-9 the aluminate ion was mononuclear and of the form Al(OH)4 when the concentration of aluminium is well below Na and 27A1 NMR spectra of sodium aluminate dissolved in H20 have been studied by Moolenaar, Evans and McKeever in the concentration range 0.5 to 6 M At aluminium concentrations below 1.5 M and above ph 13 tetrahedral Al(OH)4 was the predominant aluminium-bearing species in solution with a Raman peak at 623 cm1[431 and combination-scattering spectra (CSS) lines at 630 cm1[441. 9

35 At higher aluminium concentrations and a ph range from slightly alkaline to acidic, condensation reactions occur, and polyaluminate ions are Plumb and presumed that the aluminate ion was a polymeric anion of composition in which the Al(OH)4 units were linked together by two hydroxide bridges and each aluminium ion was surrounded by six hydroxide ions. observed the molecular weight of the polymeric aluminate ion to coincide with either A12(OH)104- or A1202(OH)86-. Infrared and Raman spectra of sodium aluminate dissolved in H20 at aluminium concentrations below 1.5 M showed four vibrational bands, two of them were infrared active (at 950 and 725 cm-1) and three of them were Raman active at 725, 625 and 325 cm-1. New vibration bands appeared at 900 (IR), 705, and 540 cm-' (Raman) for aluminium concentrations of about 1.5 M and The 27A1 NMR resonance of sodium aluminate solution at above 1.5 M Al was considerably broadened, but no change in chemical shift was noted. These observations were interpreted in terms of the condensation of A1(OH)4 with increasing concentration to form A120(OH)62 such that, in 6 M solution the two forms coexisted. thought this new species was probably a dimer as [(OH)3A1-O-Al(OH)3]2. The concentration of Al in spent Bayer liquor is approximately 1.65 M. The iiteraturj471 indicates the solution aluminium species to be predominantly A1(OH)4 with some of the dimer A120(OH)62-. Na2SiO3.5H20 was the main silica source used in this project. The anions in its crystals are chain With increasing alkali concentration silicon species dehydration 491 occurs, changing from SiO(OH)3 to Si02(OH)22 and even to studied infrared spectra of aqueous silicate solutions with different silica sources and bases, and observed that the FTIR frequency increased with increasing molecular weight of the silicate species. found that all silicates gave a strong band near 1000 cm1. in his Raman spectrum study of aqueous ions, observed four distinct lines at 448, 607, 777, and 935 cm1 attributed to the Si032 ion. McNicol, Pott and also observed Raman bands of Si02(OH)22 at 772 and 915 cm1. Si032 ions have also been attributed to Raman bands at 775 and 275 cm1[431. A fifth line at 1040 cm1 was observed in solutions containing little or no NaOH with an increase in the sharpness and intensity of the 777 and 935 lines on increasing NaOH concentration. Two species which may give rise to the 1040 cm1 line are hydrolysed silicate ion, SiO(OH)3, or a dimer, H4Si2O72. concluded that the 1040 cm1 line was generated by the dimeric species of the silicate ion. Considering silicon species are easily conclusion of the presence of both monomer and dimer is quite acceptable. Kramer, de Man and published bond distances and bond angles of the dimer (Table 2.1). These data will be useful to understand how the framework was formed. in

36 Table 2.1 (OH)3Si-(OH)-A1(OH)3 Bond distances (A) and bond angles (degrees) of geometry-optimised Bond Distance (A) Angle ( ) SiO SiOH AlO A1OH OHH SiOHAI OSjOH OA1OH 99.4 Guth, Caullet and speculated that ions in solutions containing both silicate and aluminate were mostly complexes. They carried out calculations on the relative occurrence of the species in aluminosilicate solutions and found that the ratio of monomers, dimers, trimers and tetramers changed with the Si/Al ratio at ph 13. When Si/Al increased, the percentage of monomers decreased and the dimers, trimers and tetramers increased. Thilo and added increasing amounts of Aid3 to M Na-silicate solutions. The silicate anions observed were monomeric. A "saturated" compound, [(OH)2AlO]3SiOH(aq), was formed when the Al/Si ratio increased to 3:1. It is possible that the hydroxy aluminosilicate, in a solution is fully dissociated to Na+ cations and complex aluminosilico-oxygen hydrated anions Porotnikova, Tsvetkova, Slepukhin and found a new absorption at 280 nm when silicon additives were added to an aluminate solution. Guth, Caullet and saw a decrease in intensity of the A1(OH)4 Raman band at 625 cm1 in the presence of silicate. Sizyakov, Volokhov and investigation showed a formation of aluminosilicate complex in aluminate solutions. (OH)3A1OA1(OH)32 + SiO(OH)3 (OH)3AlOSiO(OH)22 + A1(OH)4 (2.1) However, at an Al(OH)3 concentration up to 1.96 M, monomer ions of aluminium and silicon predominate. Sizyakov et aij601 did not state their NaOH concentration in the paper. Both Dent Glasser and and Guth, Caullet, Jacques and gave the same conclusion, that aluminosilicate complexes could not survive at high ph and that the concentration of the complexes increased with the silica and alumina concentration. ETIR results indicate that only the Al(OH)4 monomer (950 and 736 cmt) plus some C032 and S042 were present in plant spent liquor i.e. the concentrations of aluminosilicate species were below the detection limit. 11

37 Aluminosilicate anions may be formed by the condensation of Al(OH)4 anions and silicate anions in Si02 gel in zeolite preparation1641. If in very basic solutions most of the silicon is assumed to be present in the form of the formation of aluminosilicate anions can be represented by the following reaction: A1(OH)4 + +yoff (2.2) where "n' and "y" are the molar ratio of Si/Al and the number of Off released by Reaction (2.2) respectively. It is not surprising that the ratio of aluminium to silicon is found to be great than 1 in sodium aluminosilicate as Al/Si in spent Bayer liquor is approximately 160. Dent Glasser and Harvey165' examined 27A1 and 29Si NMR spectroscopy of aluminosilicates in solution. Four groups of chemical shifts were assigned. The n in the number of silicate groups attached to the central aluminium atom, e.g. for 27A1 NMR, as shown below. is I I I Si Si Si Si ii 0 Si Si0 0 Si -r -r Q Q Q 27Al chemical shifts were also observed by Harvey and Dent Glasser1661 and Mueller et aij671. The n in Qfl for 29Si NMR peaks represented the number of aluminate groups attached to a silicon centre1681. A set of 29Si chemical shifts has been observed by Barrer et al. These are all listed in Table 2.2. The Al/Si ratio is an important factor affecting which species are formed in the solution. Kinrade and obtained five bands of 29Si NMR spectra which were attributable to dissolved species containing Si-O-Al links in solutions containing 1 M Si02 and 0.07 M Al in 2.4 M aqueous NaOH at 5 C. The major aluminosilicate species in solution appeared to be Al-substituted derivatives of the silicate dimer cyclic trimer 12

38 cyclic tetramer (U) and linear trimer (t\)[691. Little or no free Al(OH)4 was present in the solution mentioned above. On the other hand, Al(OH)4 species predominated in a solution of 0.01 M of each Si and Al in 0.03 M NaOH. In the former case the concentration of Si species was over 10 times that 9f Al so that almost no Al(OH)4 was left free of Si coordination. Low Al/Si ratios lead to the formation of almost all Al into aluminosilicate complexes in solution. In spent Bayer liquor the Al/Si ratio is very high. Therefore almost all Si is likely to be coordinated with Al. Table Al and 29Si chemical shifts of the aluminosilicate Cation] Q0 Q1 Q2 Q3 Q4 295j* (-103.1) - (-96.7) - (-92.5) - (-86.4) - (-80.1) - (-110) (-105.3) (-99.4) (-94.0) (-86.5) All values in ppm measured relative to Al A A Q0 is defined as the monomer of the silicon or aluminium species. * 29Si chemical shifts ppm relative to that of ** from [67] Al(OH)4 appears to complex preferentially with any larger silicate species These complexes, once formed only slowly polymerise further. This may be due to the charge density of the silicate anions. A monomeric (Si044, Q ) ion could theoretically carry a charge of up to -4. The dimer (O3SiOSiO36) and other Q1 units, can only carry a maximum of three negative charges per silicon, but Q2 and Q3 silicons (O3Si(Si02)Si034 and O3Si(Si02)2SiO34- respectively) can carry at the most two and one negative charges respectively. Thus the larger the silicate species, the smaller its negative charge per silicon, and the easier the approach of a Al(OH)4 group will be. Crystal growth in the Linde A, sodalite and faujasite systems has been studied. Laser Raman spectra of the silicate starting solution gave rise to absorption bands at 772 and 925 cm1 associated with monomeric Si02(OH)22 Aluminate solutions showed one band at 618 cm1 due to A1(OH) Only monomeric Si02(OH)2 and Al(OH)4 anions were present initially in aluminosilicate In McNicol et al.s study, silicic acid was used, and again no evidence was found of the existence of any soluble aluminosilicate anions. Angell and Roozeboom and fl

39 and Dutta and found that the Raman spectrum band at 620 cm-1 in liquid phasedisappeared (assigned as aluminate ion) and a new band at cm-1 appeared in the solid phase (assigned as zeolite A-Y spectrum) after 2-9 hours of digestion. Angel! et al. [711, Roozeboom et al. [72] and Dutta et at. [731, however, also did not observe the formation of aluminosilicate complexes in solution. Summary: The solution speciation is dependent on the relative concentrations of Al and Si and the solution ph. It appears that the predominant Al species is likely to be Al(OH)4. Due to the high Al/Si ratio in Bayer liquors almost all the Si is likely to exist in oxygen bridged aluminium containing species Structural Characteristics of Sodium Aluminosilicate Phases Sodium hydroxy-aluminosilicate belongs to the type of minerals (zeolite, sodalite and permutite) which are used technologically as sorbents ("molecular screens") for drying and separating different chemical substances (gas, Sodalite and cancrinite are hydrated feldspathoids and are classified to belong to chabasite group of if the following definition of is accepted: "A zeolite is an aluminosilicate with a (tetrahedral) framework structure enclosing cavities occupied by large cations and water molecules, both of which have considerable freedom of movement, permitting cation exchange and reversible dehydration". As naturally occurring minerals, sodalites are the chloride containing minerals, noseans are sulphate containing minerals, and cancrinites are carbonate containing For the interpretation of experimental data presented in this text sodalite refers generally to all cubic sodium aluminosilicate structures observed, cancrinite refers generally to all hexagonal sodium aluminosilicate phases Crystalline Structure The aluminosilicate frameworks of both sodalite and cancrinite have been well but relative peak intensities and unit cell dimensions in the XRD patterns depend not only on the structure of the framework but also on the degree of water of crystallisation and the amount and type of ionic presence in the channels. Illustrations of sodalite and cancrinite frameworks are shown in Figure 2.1 (a) and (b) respectively. Sodalite's structure was first investigated by Pauling[8H. It consists of an octahedral framework of Al and Si tetrahedra which form a "cage" around the centrally located anion (C1) and the cations (Na+) and has a space P43n. Its chemical formula was given as Na8(AlSiO4)6Cl2.nH2O. Sodalite is composed of a network of large cages 79; 8283] (Figure 2.1) and has an ABC layer A sodalite cage (also known as is built from 24 corner-sharing Si044 and A

40 Figure 2.1(a) is a line drawing of the sodalite structure as determined by refined by Lons and Si and Al atoms are located at vertices of polyhedra. For clarity, oxygen atoms, through which Si and Al are joined, and the cations are not Access to and egress from these cages is controlled by six rings of (Al, Si)04 tetrahedra, so that small atoms such as Na and Ar and small molecules such as water and ammonia can pass through by slow activated diffusion. Ions such as Cl-, Br, and 1, and a variety of oxygenated anions such as C104, P043, and W042 are firmly trapped once inside [85] and (a) (b) Figure 2.1 Aluminosilicate frameworks of (a) and (b) A phase having the framework structure of sodalite but containing no Cl and synthesised in the Na20-A1203-Si02-H20 system has been given several basic sodalite hydrated hydroxysodalite (zeolite HS)[921. For consistency it will be called hydroxy sodalite in this thesis. The crystallographic distinction between nosean and sodalite is unclear, except that nosean has a slightly larger unit cell. The typical structure of sodalite may be described by four minerals recognised as belonging to the sodalite (P43n) shown in Table 2.3: Table 2.3 Typical minerals of the sodalite group. Name of the mineral Formula Unit cell a (A) Sodalite Na8(AlSiO4)62C Hydroxy sodalite Na8(AISiO4)6. 20H 8.90* * Natural sodalite 8.89***; # Nosean Na8(A1SiO4)6 SO # Natural nosean Hauynite Na6(AISiO4)6CaSO Lazurite## Na8(A1SiO4)6S 9.08 * from [95]; from [19]; from 176j l0#

41 Apart from the salts listed in Table 2.3, Na2CO3 and CaCO3 may also be present in sodalite-nosean Barrer and characterised sodalite, nosean and cancrinite. The sodalite crystals obtained between 150 to 450 C in the presence of NaCl having the same unit as that listed in Table 2.3. The intensities the X-ray powder diffraction peaks of hydroxy sodalite were extremely similar to those of natural type product obtained at 280 C after 6 hours of Sazhin and Pankeeva studied a sodalite Their XRD result indicated that the product had the diffraction pattern of sodalite, but had a hexagonal appearance. Nosean has been defined by Leiteizen et a!. [991 as having the same framework as sodalite but differing in space group P43m due to a changed arrangement of species in the cages. However, in most of the published literature such as Winchell and books and Barrer et and paper nosean has the same space group as sodalite i.e. P43n. In fact the definition of nosean remains unclear and is generally based on unit cell size (Table 2.3). It is replaced by the term sodalite herein unless defined in the text of the reference. Figure 2.2 shows the XRD (Cu-Kai) patterns taken from Hermeler et of (a) cancrinite; (b), (c) intermediate phases between cancrinite and hydroxy sodalite, (d) hydroxy sodalite1 (a= 8.965(3) A) plus a small amount of hydroxy sodalite2 (a= 8.854(3) A) and (e) hydroxy sodalite2. The most obvious difference of the intermediate phase to that of sodalite and cancrinite is the absence of the peaks at 20 and in Figure 2.2b and c, and the difference between sodalite and cancrinite is that cancrinite has an intensive diffraction peak at about 19 degrees 20 (101). The infrared spectra confirmed that the intermediate phase had a spectrum between sodalite and cancrinite. Whether the intermediate phase had either a cubic or hexagonal crystal system was not discussed. Cancrinite is of the hexagonal crystal system with space group P63 with ABAB layer type 75; 78; 80; 82-83] A wide variation in degree of water of crystallisation has been observed in cancrinite structures, e.g. 1.0(Na20).(A1203). 1.68(Si02). 1.73(H20) [1011 Na8(A16Si6O24)S (H20) [102] and Non-aluminosilicate framework crystallographic sites in these structures do not necessarily have 100% occupancy. A large channel and also a series of small cages run parallel to the z-axis [791 (Figure 2.1). A list of cancrinite formula is shown in Table 2.8. Although the frameworks of sodalite and cancrinite groups of minerals are well defined and essentially of a 1:1 ratio of Al04 and Si04 tetrahedra, the inter framework ions are chemically diverse and typical of zeolite The large channels in cancrinite were found to contain OH-, CO32, S042 or H20 and the small cages, 16

42 Nat, OH-, SO42-, C1 or H2O. 0H, H20 or C1; the sodalite-group cavity can contain Nat, (a) (b) LML (e) Figure 2.2 Hermeler et. XRD patterns of (a) cancrinite, (b) and (c) intermediate phase between hydroxy sodalite and cancrinite, (d) hydroxy sodalite 1 with small amounts of hydroxy sodalite2 and (e) hydroxy sodalite Vibrational Spectroscopy of infrared spectra of sodalite and cancrinite have been published by Flanigen and Hermeler, Buhi and Porotnikova and Avdeeva and Leiteizen, Pashkevich, Firfarova and and Derevyankin and Table 2.4 shows the infrared spectra data assigned to the sodium aluminosilicate framework. Raman band assignment 107], 108] have been made for the aluminate silica and zeolite Details of these assignments are reported in the solution species section (2.2.1). It has been reported that nosean and sodalite can be differentiated by infra-red As shown in Table 2.4, the three finger-print peaks for nosean are at the same position as 17

43 those for cancrinite 690, 630 and 560 cm-1. However the nosean absorptions are more The finger print peaks for sodalite are at 737, 713 and 668 cm-1. These are symmetric framework vibrations of A1-O-Si and the shifts of these three absorptions may Table 2.4 Infrared spectra assignment of sodalite (or nosean) and cancrinite Phase Asymmetric stretch Symmetric Hydroxy sodalite Hydroxy sodalite*** Hydroxy sodalite**** Hydroxy sodalite***** Aluminate nosean***** Carbonate nosean** Sulphate sodalite** Sulphate nosean* * * Hydroxy cancrinite vwsh mw msh s msh s stretch m mw m Double rings0 T-000 bend sh sh w m ms ms m m mw ms ms Cancrinite**** Cancrinite ***** Carbonate cancrinite** Sulphate cancrinite* * sh * s = strong; ms =.medium strong; m = medium; mw = medium weak; w = weak; vw = very weak; sh = shoulder; b = broad. ** from from A1-04 from [106] 0 parallel 4 or 6 member rings in framework. 00 tetrahedron Si-04 or sh

44 be a function of the interaction of the species in the cages and the aluminosilicate framework. Why cubic nosean should have similar symmetric Al-O-Si vibrational frequencies to hexagonal cancrinite (three peaks at about 690, 630 and 560 cm-1) and asymmetric A1-O-Si vibrations similar to cubic sodalite (one peak at about 1000 cm-1) is unclear. Hence the interpretation of these data is dubious. Vibrations of the Si-O-Al bond have the most intense band in vibration region for the mixed silicon-aluminium-oxygen The maximum in the spectra for zeolite and sodalite corresponded to only one peak at about 1000 cm-1 for asymmetric stretch of Si-O-Al framework. The cancrinite spectra show characteristic infrared absorbances at 1095, 1035 and 1000 cm-1 for asymmetric stretch of val-o of the Si-O-Al a splitting and shift in the band toward higher frequencies at about 1000 cm-1, and the appearance of a characteristic additional narrow band at 1120 cm-1. The FTIR spectra of water and anions that are trapped in the crystalline lattice are shown in Table ; 109] Table 2.5 FTIR spectra of water and anions in sodium (cm-i). OHstretching H20 bending CO32 out of plane bending*** S042 asymmetric stretching Hydroxy sodalite Hydroxy sodalite* Hydroxy sodalite** Hydroxy sodalite* * * * * Aluminate nosean***** Carbonate nosean Sulphate sodalite Nitrite sodalite**** (N02) Sulphate nosean Cancrinite* Cancrinite***** Carbonate cancrinitê l400sh Sulphate cancrinite Sulphate cancrinite* * References: * [104] ** [99] [110] [1091 ***** [106] 19

45 The most intense OH- stretching band for sodalite is at cm1['o5]. Cancrinite has two intense OH- vibrational absorptions at 3620 and 3530 All hydroxy aluminosilicates have one band at cm-1, located in the region characteristic for water molecules, corresponding to the deformation of Aluminosilicate obtained from the solution with no addition of Na2CO3 has a peak at cm-1. Avdeeva and Vorsina1105' speculated that this band was due to hydroxy deformation. This was in contrast to Leiteizen et al. [99] who suggested that this should be a C032 The source of C032- could be absorbed CO2 from the air. Gaseous CO2 linearly absorbed on a cation of zeolite surface gave a band at 2349 Maxwell and Baks1113' studied the influence of exchangeable cations on zeolite framework vibrations and found an approximately linear relationship between the frequency of some absorption bands and the inverse of the sum of the cation and framework ionic radii. It was proposed that the shift in framework vibration was largely caused by those cations which were strongly interacting with the zeolite framework Nuclear Magnetic Resonance Characterisation of some sodalites was performed using 29Si and 23Na MAS 116] A large variety of 29Si MAS NMR chemical shifts, unit cell sizes and Si-O-Al bond angles of the sodalites, published by Engelhardt, Luger, Buhl and Jacobsen, Norby, Bildsøe and Jacobsen1115' and Kampa, Engelhardt, Buhi and Fe1sche11141, are listed in Table 2.6. Table 2.6 shows that larger cations and anions caused larger 29Si chemical shifts, larger unit cell sizes and larger Si-O-Al bond angles. The order of these values of different cationised sodalites is: K+ > Na+ > Li+. Anions in sodalite caused the 29Si NMR chemical shift to become more negative, the order of the chemical shift that anions of sodium salts caused is: C1O4> no anions > SCN-> 1> B(OH)4> Br> C032.4H20 > NO2> CN-> Cl- = OH.4H2O > > The authors of reference11 16] regressed the following equations: a = a (2.3) = a (2.4) = la (2.5) for sodalites with different cations and anions in Table 2.6 and = a (2.6) for different types of sodium aluminosilicates (sodalites plus zeolites). It was found that both Equations (2.4) and (2.6) had very good agreement with experimental results. It is 20

46 not difficult to see from Table 2.6 that the value of 29Si NMR chemical shift, unit cell size and Si-U-Al bond angle increase or decrease coherently. This indicates that there is an intrinsic relationship among these three factors. The unit cell size determination from the synthetic aluminosilicate showed that smaller cell sizes resulted from lower Table 2.6 angles a of some 29Si MAS NMR chemical shifts 6, unit cell sizes a and Si-U-Al bond No. Sodalites } 6 (ppm) a (A) a ( ) 1 Na6[AISiU4]6.8H2U Na6[A1SiU4] Na8[A1SiU4]6(UH)2.4H Na8[A1SiO4]6(UH)2.3H2U Na8[A1SiU4J6(UH) K6[A1SiU4]6.8H K6[A1SiU4] Li8[A1SiU4]6C LiC1-SUD LiBr-SUD Na8[A1SiU4]6C NaCI-SUD NaBr-SUD Nal-SUD K8[A1SiO4]6Cl Na8[A1SiU4J6{B(UH)4] Na8[AJSiU4]6CU34H Na8[A1SiO4]6(N02) (139.2*) Na8[AISiU4]6(CN) * 21 Na8[A1SiO4]6(SCN) Thomas and used magic-angle-spinning 29Si solid phase NMR spectroscopy to characterise short-range Si and Al ordering in aluminosilicates. Both sodalite and cancrinite were found to have 4:0 (four A1045 surrounding each Si044 tetrahedron) and 3:1 (three Al045 and one SiU44 surrounding each Si044 tetrahedron) (Figure 2.3). The 29Si chemical shift was ppm for 4:0 of sodalite and ± 0.3 ppm for 3:1 of 21

47 sodalite; in the case of cancrinite the value was ± 0.3 ppm and ± 0.3 ppm respectively. Al _Al I I I I Al 0 Si 0 Al A1 0 Si 0 Al I I 0 I I 0I I Al Si (a) (b) Figure 2.3 Illustration of (a) 4/0 and (b) 3/1 ordering in sodalite and XPS analysis as well as other surface studies of zeolites has indicated that bulk Si/Al ratios are similar to surface Si/Al The application of this technique looks promising for elucidating structural, surface and chemical information about sodium aluminosilicate. Summary: Crystalline sodium aluminosilicate (sodalite, nosean and cancrinite) and the influence of impurities in the solid phase can be well characterised by a combination of techniques such as XRD (101 peak for cancrinite only), FTIR (asymmetric (three absorptions for cancrinite and one for sodalite) and symmetric Al-O-Si vibrations (three absorptions for cancrinite and three others for sodalite)), NMR etc Crystallisation of Sodium Aluminosilicate Phases lii Synthetic Solutions Preparation of Zeolite Zeolite has a similar structure to that of sodalite and cancrinite and has been more extensively studied. The formation of zeolite from amorphous gel has been discussed extensively by Zhaanov The main differences between zeolite preparation and aluminosilicate formation in Bayer liquors are: (1) A high degree of solution Si02 supersaturation is always used in zeolite preparation. Aluminosilicate gels appears to be an intermediate product. Silica concentration in Bayer liquors is only 0.01 M and the 2.7.

48 supersaturation is low. (2) There is always either a reaction mixture aging process at room temperature or a heating step (to atmospheric boiling point) in zeolite Formation of sodium aluminosilicate in spent Bayer liquor occurs in conditions of high alkalinity, low silica concentration and high temperature. Therefore it is unlikely that scale formation has the same mechanism as that of most zeolites. At low silica concentrations the time needed for the solutions to gel increases as more alkali is Pang, Ueda and results show high caustic concentrations promote crystallisation from solution and not via a gel. Table 2.7 shows a comparison of the solid products obtained from homogeneous solution nucleation and aluminosilicate gel formation. Table 2.7 Comparison of solid products from solution and aluminosilicate gel at 80 and 100 OC[12l] Temperature ( C) ] Time (hours) Homogeneous solution* Aluminosilicate gel** 80 1 NaA*** 80 4 NaA NaA NaA NaA + NaX + Sod NaA * from homogeneous solution with a composition: 10(Na20).0.2(A1203).(Si02 (high caustic) ** from aluminosilicate gel with a composition: 3(Na20).(A1203).2(Si02). 185(H20) (low caustic) (NaA and NaX are two types of zeolite and Sod is sodalite.) Schwochow and and Pang, Ueda and results showed that, at high caustic concentration (molar A1(OH)3INaOH ratio 0.02), sodalite crystallised and at medium or low caustic concentration (molar Al(OH)3/NaOH ratio 1/3) zeolite crystallised after 2 hours at 100 C (Table 2.7). This may be due to high sodium hydroxide concentration destroying the gel, and thus promoting sodalite formation. Twu, Dutta and studied the effect of two common commercial silica sources, Ludox (manufactured by DuPont and containing 30.0% Si02 and 70% water) and N- Brand (PQ Corp. 29.5% Si02, 8.8% Na20 dissolved in water), on the formation of faujasitic zeolites through a gel process. Ludox gave rise to a more rapid crystallisation rate of zeolite X than N-Brand perhaps due to less caustic in the source. Herreros and research, however, indicated that the source of silicon and aluminium did not have much influence on the phase formed.

49 Preparation of Sodalite Two major preparation techniques may be used in synthesising crystalline aluminosilicate products. One involves the use of kaolinite (Al203.2SiO2.2H20) in alkaline solution, while the other involves crystallisation from aging aluminosilicate gels. Sodium aluminosilicate gels were widely used for the synthesis before the use of kaolinite became Barrer, Baynham, Bultitude and synthesised basic sodalite from very concentrated Na2O.A12O3.2SiO2 gels at 100 C by evaporation. Barrer and obtained a series of gels of Na2OAl2O3nSiO2mH2O, in which 'n" (Si/Al) varied from 1 to 12. Experiments were also performed in which excess sodium chloride, sulphate, or carbonate was added to the reaction mixtures. It was found that the products, over almost the whole range of the conditions of composition, ph, and temperature recorded, were sodalite and nosean in the presence of chloride and sulphate respectively. The addition of sodium carbonate did not cause a marked alteration of the usual products, but extended the conditions of formation of cancrinite. With high silica mixtures some albite and analcite were observed, but the only possible 'n' for spent Bayer liquor scale observed is either i[171 Veit, Buhi and synthesised chlorate and perchlorate sodalite using kaolinite as the raw material at 160 C in 8 M NaOH solutions with the addition of salts and a reaction time of 30 hours. The use of silica and alumina as the raw materials at above 300 C for more than 48 hours gave similar products as above. When kaolinite was used at 300 to 500 C, hydroxy sodalite was produced. The chlorate and perchiorate sodalites were not stable in caustic solutions and tended to decompose to the more stable hydroxy sodalite. It is surprising that at 500 C for 96 hours only hydroxy sodalite (characterised by XRD and IR) was the product with no cancrinite. Normal nosean (XRD) formed in the presence of excess Na2SO4 over the range oc[ % yields of hydroxy nosean (a = 9.10 A) were obtained at 450 C in the presence of a great excess of NaOH ( %), and in the complete absence of Nosean was also obtained in varying yields over a wide temperature and considerable ph Early work on the synthesis of sodium aluminosilicate was carried out by Barrer and co- 127] showed that with 300% excess of NaOH and Al/Si molar ratio 1.5, hydrwçy sodalite (Na8(A1SiO4)6.20H, a = formed between 90 to 140 C and hydroxy nosean (Na8(A1SiO4)620H, a = 9.10 A) formed between 140 to 200 'C i.e. hydroxy nosean formed at higher temperatures than hydroxy sodalite. Buhi, Engelhardt and synthesised tetrahydroxoborate sodalite using kaolinite preheated at 1500 C for 2 hours and boric acid in 8 M NaOH solution at 200, 300,

50 and 500 C for a period of 5 days. By adjusting the ratio of B203/A1203 from 10 to 1 to 0, they obtained tetrahydroxoborate sodalite (B/Al = 10), a solid solution of tetrahydroxoborate sodalite and hydroxy sodalite (B/Al = 1), and hydroxy sodalite (B/Al = 0) respectively. B/Al ratio is zero in Bayer liquor, thus hydroxy sodalite should be the nucleation product of sodium aluminosilicate. Bühl, Engelhardt and also obtained a proportion of cancrinite at 300 and 400 C but not at 200 and 500 C. Edgar and reported a polymorphic transition from a hexagonal cancrinite structure to a cubic nosean structure in the composition 3NaA1SiO4.Na2CO3 at above 600 C Preparation of Cancrinite Four cancrinites (space group P63) were synthesised in a solution of sodium aluminate to which excess sodium nitrate, sodium thiosuiphate, sodium sulphate or sodium sulphide was added. After heating, a solution of sodium silicate was added to the above solutions, and the suspension was held at 110 Oc[102] the crystallisation products are shown in Table 2.8. The formula and unit cells of the products of Table 2.8 Unit cell sizes of c (A) Na8[A16Si6O24](N03)2.4H ± ± 0.01 Na8[A16Si6O24J(S203).3H ± ± 0.03 Na8[A16Si6O24](S04).3H ± ± Na8[Al6Si6O24]S.4H ± ± Na2OA12O32SiO2.Na2CO3* Na8[A16Si6O24]20H.nH2O* * from {89] When the gel Na2OA12O32SiO2 was treated with an excess of aqueous Na2CO3 in the temperature interval C, cancrinite formed as thin The unit cell size is also shown in Table 2.8. Hydroxy cancrinites were grown in the complete absence of Na2CO3, but in the presence of excess of NaOH ( %), at 390 The X-ray powder photographs were almost identical corresponding to expanded and contracted forms; variations in the unit cell sizes of these forms were listed in Table 2.8. Barrer, Cole and found that sodium nitrate, chromate or molybdate could promote the formation of cancrinite in kaolinite and caustic solutions Comparison of Preparation of Zeolite, Sodalite and Cancrinite 7S

51 conflict Avdeeva and synthesised sodium aluminosilicate in a solution of M Si02, 1.7 M Al(OH)3 and 0.8 M NaOH. Na2CO3 or Na2SO4 was added to the solution. It was found that at 105 C after 3 hours, a zeolite-type sodium aluminosilicate formed. At 220 C, a sodalite-type sodium aluminosilicate and a cancrinite-type sodium aluminosilicate formed after 3 and 36 hours respectively. This indicated that with higher temperatures and the longer reaction times, the crystallisation of cancrinite was favoured over sodalite and zeolite. synthesised large monocrystals of sodium hydroxy aluminosilicate in alkali solutions containing M NaOH and M Si02 using metallic aluminium (concentration not stated) at 100 C. He found the product had zeolite characteristics: when heated to C the water in the lattice was fully removed without the destruction of the crystals. The dehydrated product (350 C) exhibited a capability of reverse absorption for water. Cardile, Mullett, Bussell, Graham and thought that the aluminosilicate material initially precipitated as dispersed balls of gel-like material using kaolin, gibbsite, caustic and lime as raw materials in Parr Bombs between 50 to 80 C. As the number of these "gel-balls" increased they were attracted to each other and agglomerated into small loosely packed groups in 2 hours. As the temperature increased the gel-balls in the loosely agglomerated groups began to dehydrate and formed into larger particles to form aggregates of aluminosilicate material. Thus the mean size of the aluminosilicate material did not increase due to the dehydration of the larger groups of particles. SEM analysis of the washed aluminosilicate seed revealed a "knitting-ball" structure. The balls were approximately 1-5 mm in The "knitting-balls" existed as separate particles but did cluster to some degree. Cousineau and suggested that aluminosilicate seed might increase in average size with very low silica supersaturation and decrease in average size with very high supersaturations. Summary: At higher temperatures and longer reaction times, the crystallisation of cancrinite is favoured over sodalite and zeolite. Zeolite is generally synthesised at high Si02 supersaturation and passes through a gel-like intermediate stage. There is some in the literature at to whether this is true for sodalite and cancrinite synthesis which occurs at Si02 supersaturation in Bayer plant conditions.

52 Characterisation of Plant Desilication Products Red mud is the undissolved product of digestion. It contains minerals originating from bauxite, such as undissolved aluminium oxide minerals (boehmite, diaspore), iron minerals (hematite, geothite, lepidocrocite, limonite), and unchanged accompanying minerals (rutile, anatase, pyrite, calcite, dolomite, etc.) and new phases formed during digestion, such as sodium aluminosilicates (sodalite, cancrinite) and boehmite formed by reversion. Sodium, calcium or magnesium titanates are also formed on the partial leaching of anatase. Also calcium and magnesium silicates are formed from calcite and dolomite. The structures of sodium aluminosilicate hydrates in red mud obtained by extracting bauxite with caustic soda solution containing inorganic salts were those of sodalite compounds, containing the anion of the salt Ion exchange of cations and/or anions may take place by the addition of salts. Jonas, Solymar and carried out research on red mud phases analysis by infrared spectrophotometry and X-ray diffraction. Table 2.9 gives the XRD results of some plant scale analyses. The majority of scale in the flash tanks at QAL and Eurallumina (EA) was reported as boehmite and cancrinite (Table 2.9). Boehmite was expected to precipitate from liquors above 100 C. Digester scale at QAL and Euralluniina (EA) was reported as mainly Some of these analyses have subsequently been found to be incorrect. The scale that crystallises during bauxite digestion is mainly cafetite (Na, Ca, Mg)2 2(Fe, Al)2 5(Ti, Si) H20 and also contains small amounts of The unit cell refined from the biggest scale sample XRD was orthorhombic with a = 9.098(1) A, b = (2) A and c = 5.250(1) A as compared to the cancrinite values of a = 8.99(3) A, b = 9.55(3) A and c = (1) A. Sodalite and cancrinite are the main phases formed between 150 and 225 C in Ho, Robertson, Roach and studied Bayer process desilication products morphologically and found that iron oxide coated alurninosilicate particles. This made SEM characterisation difficult. The authors removed the iron oxide from the solid samples without affecting the aluminosilicate crystals by reducing the iron to more soluble iron (II). A typical aluminosilicate morphology was of many curved rods of crystalline material wrapped around each other, giving the appearance of a ball of knitting wool. The formation of boehmite at low temperatures is another undesired scale. Under low Al(OH)3/NaOH molar ratio conditions of there is no boehrnite formation and the 31-32] scale formed is hydroxy Lowering the Al(OH)3/NaOH molar ratio from 0.73 to 0.62 under pseudo-digestion conditions prevents boehmite precipitation'1 171 Boehmite scale can not be removed by acid cleaning alone. An acid/caustic/acid wash cycle is required for complete removal of aluminosilicate-boehmite layered scaie[3h. 7.7

53 . Summary. There are two kinds of sodium aluminosilicate scale found in Bayer plant: sodalite and cancrinite Crystalline Phase Transformations of Sodium Aluminosilicate Generally, in accordance with Ostwald's law of successive transformation147' and with the simplicity principle of Goldsmith, the first type of sodium aluminosilicate to be formed tends to be the least stable thermodynamically and this is replaced by a more stable form, and so on, until a final, most stable product The crystalline phase transformations of sodium aluminosilicates in Bayer liquor have not been reported as extensively as the phase transformation of zeolites. Table 2.9 Weight percentage from XRD analysis of some plant scales [32] Scale name Phase present QAL EA3 QAL 25- Flash tank 24 vapour line Heat exchanger shell side Boehmite 46.0(1.5)* 69.8(0.2) Gibbsite 15.0(0.2) Diaspore 5 Digester vessel EA1 QAL 25-8 EA2 Digester vessel Digester pipe Cancrinite 24.6(1.9) 2.7(0.3) 65(2) 70(2) 16(2) Digester Haematite 18.9(0.1) 7.4(0.1) 8.2(0.1) 9.0(0.1) (0.1) Rutile 5.1(0.1) 2.4(0.1) 7.5(0.2) (0.2) Ilmenite 0-42 Quartz 0.3(0.1) Calcite 1.1(0.1) 25(0.3) Zircon trace Perovskite 14.6(0.1) 6.1(0.1) 29.8(0.4) 38.1(0.1) Magnesite 1.3(0.1) Sodalite Residual Na20 1.7(0.8) 2.5(1.0) pipe could not be determined TOTAL 97.3(2.7) 98.7(0.5) 96.6(2) 95.1(2.3) 93.2(2.4) 76.9(0.2) * the number in bracket is the possible error of the measurement

54 Phase Changes Crystalline phase transformations of sodalite or cancrinite in Bayer liquor has not been reported. Some phase transformation from zeolite to sodalite or cancrinite under other conditions, however, has been published. Seimiya11311 synthesised sodium aluminosilicate by treating the solution (NaOH 2.7 M, Al(OH) M, Si M containing Na2SO4, NaCl or Na2CO3) at room temperature to get a gel and then transferred the mixture to an autoclave to react at 106 C. Seimiya obtained zeolite at 106 C after 3 hours after which the zeolite progressively changed to hydroxy sodalite with increased digestion time. Pang, Ueda and studied NaA type zeolite from homogeneous solutions at 100, 90, 80, 60 C and room temperature. The results (Table 2.10) showed that phase transformation to hydroxy sodalite took place after crystallisation proceeded for 1 hour at 100 C, and 3 hours at 90 C, 6 hours at 80 C and at 60 C, i.e. the higher the temperature, the faster the transformation. Breuer, and also found that Zeolite A was metastable and converted to hydroxy within 90 hours at 100 C. This conversion was much faster in concentrated liquors. aluminosilicate precipitate produced at 25 C from a silicate-aluminate solution remained amorphous even after aging for 4 years, as indicated by X-ray Silica plays an important role in determining the phase precipitated; if too low, gibbsite or kaolin may form in the solution Table 2.10 The relationship between temperature, time and crystallised Temperature ( C) NaA* NaX Sod R.T. 168hr hr hr hr hr hr hr hr hr mm hr hr. * NaA and NaX are types of zeolites, and Sod sodium is hydroxy sodalite. following mechanism for his synthesis: fast amorphous soluble species + nuclei (or zeolite crystals) zeolite A (2.7) slow 7.9

55 The amorphous solid dissolved rapidly in the alkaline solution to form a soluble active species. The rate determining step was the reaction of the soluble species with nuclei or zeolite crystal to yield zeolitic product. Ni, Khalyapina and thought the system Na20-A1203-Si02-H20 gradually underwent structural changes. In the case of a solid phase in contact with an aluminate solution containing about 3.9 M NaOH, regardless of A1(OH)3/NaOH ratio (the actual Al(OH)3/NaOH molar ratios were 0.01, and 0.15, while in spent Bayer liquor this ratio is 0.36), it was assumed that these changes occurred in the following sequence: Amorphous hydroxy aluminosilicate type hydroxy aluminosilicate of the zeolite A hydroxy aluminosilicate of the sodalite type + metasodalite * spherulitic pseudo-sodalite (2.8) Amorphous hydroxy aluminosilicate formed from aluminate solutions with higher caustic concentrations (from 8.05 to 14.5 M NaOH) transformed immediately into sodalite and then into "pseudo-sodalite" (cancrinite?). In some zeolite synthesis, sodium silicate and sodium aluminate water solutions were used and no caustic was involved at Wehrli and synthesis of aluminosilicate supported Ni et al. assumption. Their process went from an amorphous phase (after 5 and 15 minutes) to a zeolite (after 60 minutes) which then slowly transformed to a sodalite at 100 C after about 3 to 4 hours. They examined the stability of zeolite by exposing it to the air at 100 C. The zeolite phase remained unchanged for 4 hours. They tested both zeolite and sodalite in air at 250 C for both 10 and 30 minutes, and found the phases partly changed to cancrinite. Higher concentrations of NaOH promoted faster zeolite crystallisation from amorphous (maximum 1.5 M) and faster transformation of zeolite to (from 8.05 to 14.5 M NaOH). The caustic concentration used by Ni et a1j139' was much larger than that in spent Bayer liquor (4.52 M) which was in turn much higher than that used by It has not been shown whether the formation of sodalite in spent Bayer liquor goes through amorphous and/or zeolite stages. Apart from the ex-situ thermal analysis, Ni et al. did not provide any experimental evidence to support Equation (2.8) regarding phase transformation. Yuhas, Orbanne and systematically investigated structural varieties of sodium aluminosilicates with NaOH = 3.23, 4.84, 6.45 and 8.06 M; Al(OH)3/NaOH molar ratio = 0.25 or 0.5; initial Si02 = 0.4 M at 60, 80, 120, 160 and 200 C for 5 hours of crystallisation. They obtained amorphous sodium aluminosilicate and zeolite at 60 C; zeolite at 80 C when the NaOH concentration was lower than 6.45 M, sodalitei when Al(OH)3/NaOH molar ratio = 0.25 and NaOH = 8.06 M and sodalite2 when

56 Al(OH)3/NaOH molar ratio = 0.5 and NaOH = 8.06 where sodalitei and sodalite2 are both cubic with the former having a larger unit cell; and sodalitei when Al(OH)3INaOH molar ratio = 0.25 and sodalite2 when Al(OH)3INaOH molar ratio = 0.5 at between 120 to 200 C. This indicates that high Al(OH)3INaOH molar ratios promote sodalite2 formation. They also found a transition of zeolite to sodalite2. Conversely, when sodalitei and sodalite2 were placed under the conditions where zeolite appeared to be the equilibrium phase, they do not change their composition and structure. Edgar and reported a polymorphic transition from a hexagonal cancrinite structure to a cubic nosean structure of composition 3NaA1SiO4Na2CO3 at above 600 C in water and with a pressure range 10,000-30,000 psi. A similar transition took place for a solution containing an unspecified ratio of NaAISiO4 and NaOH at a temperature of 570 C. Phase transformations from cubic to hexagonal below 300 C have been 117; 143]; cancrinite was thought to be the most stable phase in this temperature range. Katovic, Subotic, Smit and study of crystallisation of zeolites from freshly prepared aluminosilicate gels showed that the tetragonal form of zeolite P (designated as zeolite B8) appeared as the first crystalline phase and transformed completely into the cubic form of zeolite P (designated as or NaP1) at extended reaction times. Summary: The stability of sodium aluminosilicate in aqueous solution increases as: gel (amorphous) > zeolite > sodalite > cancrinite between room temperature and 250 C. Longer reaction times, higher solution temperatures and higher NaOH concentrations promote phase changes from: amorphous zeolite sodalite cancrinite Unit Cell Changes The basic noseans, sodalites and cancrinites described by Barrer and were made from similar gels (nna2o.a1203.2sio2 aq) by varying only the amount of excess NaOH. Compoundsintermediate between a = 8.90 to 9.10 A were obtained. Thus a continuous range of unit cell sizes is possible. The formation of such a range was thought to be directly dependent upon the amount of NaOH present in the reacting mixture. Synthetic aluminosilicate of smaller unit cell size resulted at lower Variation of unit cell dimensions upon dehydration or destruction of hydrogen bonding at higher temperatures was observed in hydroxy Avdeeva and results showed there was a unit cell size reduction in their sodium aluminosilicate phase at 175 C from 3 to 72 hours of reaction on addition of sulphate, carbonate or chloride (Table 2.11). The densest atom packing in the structure occurs in the chloride sodalite. The structure of the sulphate sodalite was less stable; its

57 rearrangement to cancrinite was completed long before the start of similar changes in the carbonate sodalite which in turn transformed prior to the hydroxy sodalite. In this study it was observed that the unit cell of hydroxy sodalite was smaller than for carbonate in apparent concentration to the results shown in Table Table 2.11 Unit cell sizes of the sodalite phase with different anions before and after equilibration1145' Anion OH- S042 C032 C1 a (A) I: non-equilibrium a (A) II: equilibrium In contradiction to Avdeeva et al.' results others studies found that with increased digestion time the hydroxy sodalite phase expanded toward a = 9.10 A, and then apparently recrystallized to hydroxy cancrinite can be observed by monitoring the growth of the peaks at d = 4.7 (101) and 3.2 A (211); these become the major peaks in well-crystallised hydroxy cancrinite. Summary: There is a unit cell size reduction in crystalline phases with reaction time The Effect of Salts or Impurities in the Liquors The principal components of Bayer process liquor are sodium hydroxide (NaOH) and sodium aluminate (NaA1(OH)4) together with a number of impurities, the most significant of which are organic materials in the bauxite and carbonate, silicate and chloride, Bauxite, such as that mined at Weipa, is a mixture of gibbsite (Al(OH)3), boehmite (AlO(OH)), kaolinite (Al203.2SiO2.2H20), quartz (Si02), anatase (TiO2) and haematite (Fe20 3)[221. Na2CO3 has a quite high concentration in Bayer liquor (0.38 M). The solution initially used to digest the bauxite is a sodium hydroxide However, since only half of the dissolved alumina in the liquor is recovered in the precipitation stage, and after prolonged recirculation, part of the hydroxide is converted to Na2CO3 through the action of atmospheric CO2 and organic materials in the bauxite, the digester liquor is converted to a NaOH-NaAIO2-Na2CO3 solution in which Na2CO3 represents an undesirable contaminant. Carl onate sodalite was obtained in red from a liquor where only sodium hydroxide and no sodium carbonate were added. It is often the case with crystallisation processes that the presence of impurities can have a significant effect on the nucleation, agglomeration and growth of the precipitating Hermeler, Buhi and Hoffmann1100' reported the influence of carbonate on

58 the synthesis of an intermediate phase between sodalite and cancrinite. The presence of "excess' carbonate may promote a phase intermediate to sodalite (Figure 2.2) and cancrinite from (a = 8.965(3) A, NaOH, 2 and 4 M) and plus sodalite2 (a= 8.854(3) A, 8 M NaOH) for 5 days of reaction at 157 C but the same solid phase sodalite2 (a= 8.854(3) A) was obtained when 16 MNaOH was used. These observations are in contrast to the trend in Table 2.11 where it may be expected that on increasing Off concentration the unit cell should increase in size. This indicates a competition of Off and CO32 in solid phase. At low NaOH solution concentration (2-4 M) only sodalitei (carbonate sodalite) was obtained, at medium NaOH concentration 8 M sodalitei and sodalite2 (hydroxy sodalite) were obtained and at high NaOH concentration (16 M) only sodalite2 was obtained. The higher the NaOH concentration of the solution, the more likely sodalite2 (hydroxy sodalite) is obtained. Needle-like crystals of cancrinite were mainly obtained at low concentrations of NaOH in the presence of excess Na2CO3 at 497 OC. Uptake of Na2CO3 by crystals increases with increasing carbonate concentration in solution at 80 oc[851 Sodalite was observed to be exceptionally selective towards H20, C1, Br, Cl03 and C104 at the expense of OH-. Salt occupancy of the sodalite cages reached a maximum on addition of 4 M of NaNO3 in Barrer, Cole and used salts in their solutions to see which salts enhanced the formation of sodalite and which enhanced the formation of cancrinite. The sodalite promoters were found to be: chloride, bromide, perchlorate, tungstate, carbonate and chlorate. Barrer, Cole and found that salts like sodium nitrate, chromate or molybdate could promote the formation of cancrinite in kaolinite and caustic solutions which normally give rise to sodalite crystallisation. The proportion of the components were: 2 grams of kaolinite with 200 cm3 of NaOH (4 M) plus 10 grams of the salts. The reactions were performed at 80 C for 5 days and cancrinite was Apart from the above three salts, Na2SO4, Na2SeO4 and Na2VO4 can also be used as cancrinite promoters. NaMnO4 can facilitate cancrinite formation but is required in a large excess. Another salt Cu(NH3)4SO4 can promote cancrinite in the presence of NH3 aqueous solution. This promotion of the formation of cancrinite was ascribed to association of anions with certain aluminosilicate species in solution, which yielded cancrinite nuclei. Barrer, Cole and used both sodalite and cancrinite promoters in their solution to see if they obtain both phases. The result indicated that the concentration of the cancrinite promoters did have an effect on which solid phase formed. Table 2.12 shows Barrer et results when they used different amounts of sodium nitrate in their solutions.

59 Table 2.12 at 80 C. Products obtained in synthesis with NaNO3 addition to the solutions CNaNO3 (M) Product Can Can + Can + Can -i- Can + Can + Sod trace Sod trace Sod Sod Sod Sod Conditions: 2 g of kaolinite, 200 cm3 of NaOH solution (4 cancrinite and "Sod" refers to sodalite. where "Can" refers to Cations play a prominent structure-directing role in zeolite A good fit was observed for Na+ and (CH3)4N+ ions in gmelinite and sodalite cages, and K+, Ba2+, and Rb+ ions in the cancrinite cage. Ryavkina, Grachev and found that an increase in concentration of suiphide, thiosulphate, sulphate or sulphite in aluminate solution promoted crystallisation of cancrinite-type sodium hydroxy aluminosilicate. Increasing the concentration of these anions from 0.03 to 2.00 M lead to sodium hydroxy aluminosilicate restructuring in the order: nosean nosean-cancrinite 3 cancrinite-nosean cancrinite (2.9) Breuer et al)'19' predicted that the rate that the solid phase changed from sodalite to cancrinite at high temperature was: sulphate sodalite > carbonate sodalite > hydroxy sodalite. In their tests at 200 and 250 C with liquor containing Na2CO3, the presence of NaCl tended to slow or prevent the conversion from sodalite to cancrinite. Na2SO4, unlike Na2CO3, contributed to the formation of a stable and therefore a less soluble form of cancrinite, which determined the difference in the desilication effect of these two salts. Under the most rigorous desilication conditions (200 C, 36 hours), the carbonate sodalite would convert to carbonate cancrinite; the Cl-sodalite did not change under these Breuer et al)119' also found that the aluminosilicate made using chloridecontaining liquor was in some cases also converted to the cancrinite, but in general was hydroxy sodalite, nosean, or a mixture. A number of salts, including sulphate, phosphate, nitrite, and suiphide caused rapid nucleation of nosean (identified by With nitrate or oxalate, hydroxy cancrinite formed in synthetic Bayer liquor in as little as 4 hours, and in a longer time when sulphate was present. It has been indicated that large anions (e.g. C032, S042, N03, and can cause expansion and sometimes distortion of the cubic lattice. However Avdeeva et reported that hydroxy sodalite had a larger unit cell than carbonate sodalite (Table 2.11). Embedding a hydroxide ion instead of a chloride ion yielded hydroxy sodalite without noticeably changing the crystallographic It was found that

60 sulphate increased the rate of conversion of zeolite A to sodalite at 70 C but chloride had no Barrer and gave the thermal stabilities of some sodalites and cancrinites with imbibed salts (Table 2.13). Cancrinite was found in Bayer plant scale at less than 100 oc[151j Ni, Khalyaoina, Bunchuk and did laboratory based research on the crystallisation of cancrinite under low temperature conditions similar to the Bayer process. Their experiments were carried out at 100 C for 2 hours to 3 days with NaOH 7.72 M and A1(OH)3/NaOH (by moles) = 0.17 or NaOH 4.5 M and Al(OH)3INaOH = addition of carbonate did not promote the formation of cancrinite at 100 C; this was the same as Hermeler et al.'s Organic substances, on the other hand, did promote the formation of cancrinite at this temperature. XRD patterns of sodium aluminosilicate obtained from industrial and synthetic K20 solutions showed that the crystalline phase was cancrinite and sodalite respectively, indicating that it was other impurities but not K20 that promoted the formation of The Table 2.13 Thermal stability of some sodalites and cancrinites with imbibed Designation Sodalites [ Cancrinites Class I (Very stable)* Na2SO3 (Na2SO4) Na2SO4 Class II (Stable: may be heated up to approximately. 850 C without lattice destruction) Class ifi (Unstable: lattice destroyed or substantially) decomposed by heating to approximately 850 C. NaC1 Na2SeO4 NaBr NaT NaOH NaOH NaCH3CO2 Na2CrO4 Na2CO3 NaNO3 NaHCO2 Na2MoO4 Na2C2O4 NaFeO4(?) Na2 W04 NaMnO4(?) Na3PO4 Na3 VO4 NaC1O3 Na2TeO4 NaC1O4 Na2S NaN3 * may be heated above 850 C without lattice destruction Vaughan, Strohmaier, Pickering and tested the influence of anions on the phase of aluminosilicate synthesised at Si/Al = (Table 2.14). They did not mention how 1 much of the salts were added (they only stated that sodium salts were dissolved in about

61 0.2 moles of water before mixed with the reaction gel). According to the results in Table 2.14, the amount of the salts must be less than that Barrer et a1j82' used, because the product from sodium nitrate solution was mainly sodalite at 100 C after 4 days of reaction. As shown in Table 2.12, only a large enough amount of salt can promote the formation of cancrinite. Not only the anions in aluminosilicate can be substituted by other anions but also cations can be replaced by other cations. Vaughan et aij1521 substituted aluminium by iron in sodalite. The success of the substitution was confirmed by Mossbauer spectroscopy, as Fe3+ contains an odd number of electrons, and thus was sensitive to magnetic sweeps. Al3+ contains an even number of electrons and is thus not sensitive. Ryavkina, Grachev and investigated the effect of sulphur compounds upon the composition and structure of crystallising sodium hydroxy aluminosilicates in aluminate solutions. It was found that the order of the influence of the sulphur compounds on restructuring and desilication at high temperature (235 C) was: Na2S2O3 > Na2SO4> Na2S > Na2SO3. Na2SO3 was found to be a sodalite promoter. When there was a substantial increase in the solution sulphur concentration, cancrinite containing S042, S2032 and S2 anions formed in solution. Table 2.14 i[152]. Influence of anions on products of aluminosilicate synthesis at Si/Al = Salt, X 100 C (4 days) 150 C (3 days) 220 C (3 days) [ NaOH Lta a) Sod Sod NaC1 Sod b) Sod Sod NaBr Lta >Sod Sod>Nep e) Sod NaT Lta = Sod = Gis Sod = nep Sod>nep Na2SO4 Sod>Can c) Can Can NaNO3 Sod>>Can Can Can Na2CO3 Sod Sod Sod>>Can Na2S Sod>>Can Nep >Sod = Can Nep CH3COONa Lta >Sod>Gis d) Nep Nep NaCrO4 Sod>Lta >Can Can Can a) Lta = Linde A; Sod = sodalite; c) Can = cancrinite; d) Gis = gismondine; e) Nep = nepheline. When nitrite sodalite was heated to C in an air stream, nitrate sodalite was obtained by oxygen uptake from the air during the This was clearly shown in the IR spectra; the concentration of nitrate (1385 cmt) in the solid phase

62 increased and that of nitrite ( cm-1) decreased with time. After being heated for 1 hour at 827 C the sodalite framework remained unchanged. Summary: Liquor impurities and the concentration thereof can have an effect on the rate of transformation from sodalite to cancrinite. The cations found to promote cancrinite formation are: K+, Ba2+ and Rb+ and the anions found to promote cancrinite formation are: S042, N03, C103, Mo70246t Se , S2 and V042. The ions in synthetic Bayer liquors are Other Factors Oil and C032-. Cardile et reported that temperature is a major factor controlling hydroxy sodalite/cancrinite formation. For a given NaOH concentration, high temperature and high 26; 88; 100] pressure resulted in cancrinite Table 2.15 shows aluminosilicate phase formation in Bayer plant heat exchangers as a function of temperature. A number of intermediate phases such as sodalite-nosean, cancrinite-nosean have been reported to exist in heater Table 2.15 Scale phases at QAL as a function of temperature ( C) nosean Aluminosilicate phase nosean-cancrinite cancrinite natrodavyne* * a mineral with a composition of 3Na Al reported that sodalite formed in the early stages of desilication 4 to 5 hours after the addition of Si02 at 140 and 160 C. With pure aluminate solutions (carbonate free), at 200 C the product was at first a mixture of hydroxy sodalite and nosean, changing only to hydroxy nosean after 90 Hydroxy sodalite formed initially at 250 0C[l 17] At 70 C, with Al(OH)3INaOH = 0.31, NaOH = 4.75 M and Si02 initial = M, zeolite A was found to be the main hours to hydroxy sodalite. Zeolite A was metastable and converted over many All solid desilication products were identified by Breuer, Barsotti and as being varieties of the sodalite family for short digestion times (0.5 to 20 hours). The phases formed in long-term digestion (46-96 hours) were invariably found to be hydroxy cancrinite for liquor containing either sulphate, or no salt except carbonate.

63 Polzer, Hem and and Cardile, Mullett, Bussell, Graham and works indicated that high silica levels tended to favour cancrinite formation. Cardile et al.'s experiments were carried out over 1, 2 and 5 days with 0.17 M Si02 and for 5 days with Si M. XRD analysis showed that at 0.17 M Si02, time had no effect and hydroxy sodalite always formed. However at higher silica levels ( M) with longer times (5 days) cancrinite formed. Si02 concentration in Bayer spent liquor is 0.01 M and is constant thus the above conclusion may not be true for scales formed in Bayer liquor. Ryavkina et a!. [149] also reported that increasing the concentration of sulphur containing salts promoted formation of sodium hydroxy aluminosilicate with increased Si02/A1203 and Na20/A1203 ratios. Roach and pointed out that the ratios of caustic, aluminosilicate and water had a significant effect on the formation of the aluminosilicate gel prior to zeolite crystallisation. Sodium sulphide, thiosuiphate or sulphate in aluminate caused the transformation of nosean to cancrinite at high temperature; but at 100 C they caused a shrinkage of the sodalite unit cell. Pang, Ueda and Koizumi11211 reported that seeding of zeolite NaA did not have any effect on phase transformation kinetics from zeolite to hydroxy sodalite. The transformation would occur at the same reaction rate as with no seeding, under similar reaction conditions. It appears that the phase transformation did not occur on the zeolite seed surface. Summary: Temperature, reaction time, caustic concentration, silica concentration and the ratios of Al/Si or Al(OH)3/NaOH of the liquor all have an influence on the phase of sodium aluminosilicate formed and the subsequent phase changes. High temperatures ( 125 C), long reaction times, high caustic concentrations (ph 14) and high silica concentrations promote the phase transformation from sodalite to cancrinite. Seeding does not appear to have an effect on phase transformation but does affect the rate of desilication. 2,4 Silica Equilibrium Concentrations in Bayer Liquors The equilibrium solubility of sodium aluminosilicate in aluminate solutions has been the subject of a large number of The main factors affecting Si02 solubility are: (1) the solid 1571; 1571; 157]; (2) (3) Al(OH)3/NaOH (4) alkali (5) impurities e.g. C032 and and (6) very slow approach to equilibrium, particularly at low temperatures

64 The determination of equilibrium silica concentration is complicated by a very slow approach to equilibrium, particularly at lower temperatures, and therefore this is usually estimated from analysis of desilication Ueda, Murata and showed the crystallisation of mordenite and analcime (two types of zeolite) at 100 C, that had not reached equilibrium after 20 days. pointed out that sodalite had a limited solubility in pregnant (high alumina) Bayer solutions amounting to not more than 4.2x103-5x103 M Si02 in 4 M NaOH solutions and that sodalite solubility in spent (low A/C, < 0.4) Bayer liquor is half of that in pregnant liquor at the same caustic concentration. The solubility of Si02 in spent Bayer liquor was reported to be about 3.3x103 M at 110 C with 3.4 M Si02 equilibrium solubility in spent plant liquor (Alcan International Limited, Canada) was 3.0x103 however, the actual Si02 concentration in the spent liquor was 8.3x103 M. Cousineau and gave the Si02 equilibrium concentration in a North American plant ( 3.77 M NaOH) as 6.7x103 M in the digester and half that much in heat exchangers The Effect of Solid Phases There is evidence to show that the solubility of silica is a function of the phase of the crystalline material presentt541. However in no studies has the possibility of a seed phase change been taken into account. Adu-Wusu and found that sodium aluminosilicate phases formed at high initial silica concentrations dissolve more rapidly than those formed at low initial silica concentrations. An amorphous sodium hydroxy aluminosilicate formed immediately after the start of mixing silica gel in solution, had comparatively high solubility (1.8x102 M as Si02 at 230 oc)[57] Prevzner, Eremin, Rozen, Kolobov and Mironovt1571 reported that the solubility of amorphous sodium aluminosilicate hydrate in 5.0 M solutions of NaC1 + NaOH + NaAlO2 (keeping constant ionic strength) in 1.57 M of Al(OH)3 at 25 C was M as Si02. Breuer, Barsotti and Kellyt191 found that under similar conditions zeolite had higher solubilities than sodalite. Caullet, Guth, Hurtrez and observed the solubility of zeolite A > zeolite X > Losod (a type of zeolite) > hydroxy sodalite at 60 C in 2 and 3 M NaOH solutions. Although their solutions were not aluminate ones, the order was consistent with that of Breuer et ai.s. Yuhas, Orbanne and showed that amorphous sodium aluminosilicate had the highest solubility (9xl02 M as Si02) in aluminate solution (NaOH 9.6 M and Al(OH)3 4.8M) at 75 C followed by zeolite (5.7x102 M) and sodalite (2x1ft2 M) in liquors containing 7.9 M NaOH and 3.9 M A1(OH)3. Cancrinite solubility is not reported in the literature.

65 Summary: The solid phase present in the caustic silicate aluminate solutions has an influence on Si02 equilibrium solubilily. The order of the solubilily is sodalite < zeolite < amorphous The Effect of Temperature There is some disagreement as to the effect of temperature on silica solubility. and Breuer et aij19' reported increasing silica solubility with increasing temperature for synthetic Bayer solution for both zeolite and sodalite as the solid phases, however and Oku and reported no temperature dependence of aluminosilicate solubility up to 150 C. Arlyuk, Anan'eva and showed that at 3.87 M NaOH and A1(OH)3/NaOH = 0.69, the solubility of sodium aluminosilicate decreased from 5x103 M as Si02 at 90 C to 2x103 M at 190 C. The solubility of sodium aluminosilicate in Bayer liquor as a function of temperature is shown in Figure 2.4[ ] Breuer et solubility increases with increase of temperature between 70 to 250 C and increase of caustic concentration between 2.83 and 4.72 M NaOH (Figure 2.4). Yuhas, Orbanne and proposed a so-called metastable solubility and true solubility. The former is the solubility of the starting phase and the latter is the solubility of the equilibrium phase at that temperature. In Breuer et al. 's case only the metastable solubility was measured (reaction time 60 hours at 150 C and above) because sodalite would not be the equilibrium phase at higher temperatures. Cournoyer, Kranich and reported that at 123 C (400 K) hydroxy sodalite eventually reacted to form hydroxy cancrinite. Eremin, Mel'nikova and 166] and Shvartsman and observed a decrease in Si02 concentration between 90 and 125 C followed by an increase between 125 to 175 C at Al(OH)3INaOH molar ratio between 0.56 and 0.94 (Figure 2.4). This was explained as being due to the presence of "truly soluble Si02 in a colloidal state" although thi,s statement was not clarified further. Our observations suggest that solid phase transformation from sodalite to cancrinite may have occurred at 125 to 175 Summary: There are conflicting results regarding the effect of temperature on sodium aluminosilicate solubility in aluminate solutions (including: (1) increase, (2) decrease, (3) independent and decrease then increase with the increase of temperature). 40

66 E T( C) 300 Figure 2.4 The solubility of sodium aluminosilicate in Bayer liquor as a function of temperature with Al(OH)3/NaOH (molar) = 0.31 and CS/TS (CS = weight of NaOH expressed as Na2CO3, TS = weight of NaOH expressed as Na2CO3 + weight of Na2CO3) = 0.83 at caustic concentrations of NaOH = 4.72 M, (.) NaOH = 3.77 M, (o) NaOH = 2.83 M measured by Breuer et au'91, (A) at NaOH = 3.88 M and Al(OH)3/NaOH (molar) 0.69 by Arlyuk et aij1631, (D) at NaOH = 3.35 M and Al(OH)3/NaOH (molar) = 0.56 by Eremin et al)ilmj, at NaOH = 4.19 M and A1(OH)3INaOH (molar) = 0.94, and (o) at NaOH = 4.84 M and A1(OH)3INaOH (molar) = 0.81 by Shvartsman et The Effect of NaOH Concentration The solubility of sodium hydroxy aluminosilicate (no information about the phase) has been measured in pure alkali (NaOH) solutions not in Bayer With alkali concentration below 6.4 M NaOH the temperature had a comparatively small influence on the solubility. With M of NaOH the solubility of the sodium hydroxy aluminosilicate increases sharply under heating. Breuer et Shvartsman et 2.4). solubility also increases with increasing caustic 156; 1691 (Figure Summary: It has been established that the equilibrium concentration of silica in alumninate solution (the solubility of sodium alum inosilicate), at constant alumina concentration, is determined by the caustic concentration. The equilibrium silica concentration increases with increase in the alkali concentration. 41

67 The Effect of A1(OH)3INaOH Ratio Silica solubility decreases with decreasing aluminium hydroxide content. Therefore the solubility of sodalite in spent liquor is only approximately half of that in pregnant liquor at the same caustic and Prevzner, Eremin, Rozen, Kolobov and examined the solubility of sodium hydroxy aluminosilicate in aluminate solutions of varying concentrations and concluded that the solubility of sodium hydroxy aluminosilicate is influenced, first of all, by the Al(OH)3 concentration in the solution. Ni, Perekhrest and studied the solubility of sodium aluminosilicate in aluminate solution over a wide range of aluminate and caustic concentrations with NaOH = M, Al(OH)31NaOH (molar) at 90 C for 3-5 days for solutions with 2.63 to 9.97 M NaOH solutions and 10 to 12 days for solutions with 12.7 MNaOH. Breuer et showed that at low A1(OH)3/NaOH and NaOH M, silica solubility was high (Figure 2.5) and gradually decreased with the increase of Al(OH)3/NaOH. The solubility then increased with increasing Al(OH)3INaOH, particularly at high caustic concentrations. The curve of Si02 solubility versus Al(OH)3/NaOH has a "U" shape from 0.1 to 0.72, i.e. 5i02 solubility decreased from Al(OH)3INaOH = 0.1 to 0.3 or 0.35 and increased from Al(OH)3INaOH = 0.35 to 0.7. Figure 2.5 shows silica equilibrium concentration as a function of A1(OH)3/NaOH molar ratio. No indication of the resulting phases was given in either of these publications ] Similar curves were obtained by Kraus et al. at 120 and 280 and Prevzner et al. at 25, 45 and 65 Kraus, Derevyankin and determined sodium aluminosilicate solubility in aluminate solutions. The Al(OH)3INaOH ratio used was between , which is much smaller than typical Bayer plant A1(OH)3INaOH molar ratios of NaOH was M. Si02 equilibrium concentrations were found to be between lx to M. 4.

68 E A1(OH)3INaOH Figure 2.5 The solubility of silica in sodium aluminate solution as a function of Al(OH)3INaOH molar ratio at 90 with (o) 4 M and (.) 5.9 M NaOH determined by Ni et at 120 C with 3.22 and (A) 6.45 M NaOH by Kraus et aij571, at 150 C with 3.78 and (.)4.72 M NaOH and at 250 C with 3.78 and (.)4.72 M NaOH by Breuer etaij191. It can be seen from Figure 2.5 that there are three regions of Al(OH)3/NaOH molar ratio which have different effects on silica solubility. Range I: the range of high Al(OH)3INaOH solutions where the alumina concentration has a significant influence on the solubility of sodium aluminosilicate at low temperatures only. A decrease in the alumina concentration within this range leads to a sharp decrease in the solubility of aluminosilicate. Range II: the solubility of sodium aluminosilicate is almost constant. Range III: the range of low Al(OH)3INaOH solutions where a decrease in the alumina concentration leads to an appreciable increase in the solubility of sodium aluminosilicate. The Al(OH)3INaOH ratio used in this thesis is fixed at about 0.36 where the solubility is the lowest (starting from 0.01 M or 0.6 g dm3 Si02) comparing to those at other Al(OH)3INaOH ratios. Eremin, Mel'nikova and concentrations examined the influence of Al(OH)3 and NaOH on the solubility of sodium aluminosilicate with Al(OH)3 = M, NaOH = M, (much lower than that in spent Bayer liquor where NaOH = 4.52 M), initial Si02 = 0.1 M, at C and with a seed charge of 20 g dm3. In this range of (A1(OH)3/NaOH molar ratio = oo) conditions, Si02 concentration shows an increase with increasing Al(OH)3 concentration and a decrease with increasing

69 caustic concentration (NaOH = M). This result lies in Region Tin Figure 2.5. This was again similar to Breuer et observation that silica solubility decreased as Al(OH)3/NaOH decreased from 0.72 to Summary: Sodium aluminosilicate solubility in Bayer liquor initially decreases, then is almost constant and finally increases with increasing Al(OH)3/NaOH ratio between 0.0 and The Effect of Impurities on Solubility The solubility of Si02 in Bayer liquor is influenced by liquor addition of 0.40 M sulphate reduced the Si02 solubility by more than 30% at 200 and 250 oc[l9] Volkova, Tsekhovol'skaya and Breuer et and Avdeeva and The found that the presence of Na2CO3 and Na2SO4 reduced Si02 solubility more rapidly than NaC1. Na2SO4 is characterised by a greater desilication action: the residual amount of Si02 in the SO42- containing solution is about one-half the amount in the C032- containing solution of the same salt Rozen, Pevzner and reported that the addition of SO42- to a caustic aluminate solution resulted in a lower Si02 solubility than the addition of Mo042' C032, or Cr042-. This effect was ascribed to the lower degree of hydration of SO42- than the other bivalent ions examined. The authors also reported that SO42- decreased Si02 solubility less than Cl04 and NO Ryavkina et alj149' found that the presence of sodium thiosulphate in substituted sodium hydroxy aluminosilicates, gave rise to the lowest solubility compared to the presence of Na2SO4, Na2S, Na2SO3 and Na2S2O3. Addition of Na2SO3 to the aluminate solution lead to the formation of sodalite with considerable solubility. The presence of salts reduce the solubility of sodium hydroxy Flint, Clark, Newman, Shartsis, Bishop and investigated the extraction of A1(OH)3 from high-silica bauxite with sodium hydroxide-salt solution. The salts used were: sodium carbonate, sodium sulphate, sodium bromide, sodium chloride and sodium nitrate. The presence of salts, chloride, and particularly carbonate and sulphate, stabilised the sodalite structure and thus lowered the equilibrium solubility of silica in the liquor. In the presence of salts (Na2SO4, NaCl and Na2CO3) sodium aluminate solution took longer to reach 175 C when there were no salts in the system, equilibrium was reached in 18 hours; when there were salts in the solution equilibrium was reached in 32 hours. This trend is unusual. Several desilication tests were made at 150 and 200 C using actual plant liquor, which had more impurities than synthetic liquor. The SiO2 solubility was reduced by about 20% 44

70 compared to that of synthetic Another set of tests were made at 200 C with carbonate-free synthetic liquors to compare with a synthetic liquor with Na2CO3 (approximately 0.38M). The equilibrium solubility in carbonate free solution was found to be about 75 and 90% higher than that in the liquor with Na2CO3. Na2CO3 has been found to reduce the equilibrium concentration of silica and as a result increases the driving force for sodium aluminosilicate scale precipitation ; No systematic study, however, has been carried out on the solubility of aluminosilicate as a function of Na2CO3 concentration. Eremin, Mel'nikova and reported that the presence of 50% of KOH in aluminate solution can double the Si02 equilibrium concentration. This effect is opposite to that of the addition of anions. The addition of 0.34 M NaC1 resulted in no change in silica solubility at 200 and 250 C compared to the solubility of silica in the liquor with no salts In Avdeeva and experiment the addition of 0.81 M NaC1, however, decreased Si02 solubility to a very low level with an initial Si02 concentration of M, at 175 C. A concentration of 100 g dm3 (1.72 M) of NaC1 was recommended to improve Gasteiger, Frederick and measured the solubility of aluminosilicate in alkaline solution over a wide range of Al/Si ratios ( ) and caustic concentrations (0.1-4 M NaOH) with the addition of NaCI with no solid phase initially present. The salts used in these studies were A1CI3.6H20, Na2SiO3.9H20, NaOH, and NaC1. Equilibration times were 26 days or more. The solubility of aluminosilicate decreased with increasing ionic strength at constant hydroxide ion concentration. Summary: The presence of impurities in caustic aluminosilicate liquors decreases the solubility of sodium aluminosilicate Empirical Correlations for the Estimation of the Silica Solubility In Oku and Yamada's Si02 (from quartz or bauxite) solubility in Bayer plant liquor was found to be proportional to the concentrations of NaOH (between 3.2 to 7.1 M) and A1(OH)3 (between 1.38 to 3.23 M) and independent of temperature in the range of C. The Al(OH)3/NaOH range of Oku and Yamadas is thus between 0.19 to The Si02 equilibrium concentrations could be calculated from: SiO2eq = kcna2oxcai2o3 (2.10)

71 where CNa2O and CA1203 were the concentrations of these two components in g dm-3. k = 2.7x105 in Oku = 2.6x105 in Adamson et and is between 3.1 to 3.3x10-4 in Ni et at. work. Sizgek and derived an empirical model to calculate equilibrium solubility of Si02 in QAL plant liquor between 100 to 250 C, caustic range 160 to 260 g dm3 (expressed as Na2CO3) and A1(OH)3INaOH ratio of to 0.759: SiO2eq =d.82x106c x1013C2T x105C2(A/C)3 (2.11) where T= temperature ( C); C= caustic (g dm3 Na2CO3) and A/C= aluminalcaustic ratio (g dm3 dm3 Na2CO3). measured the maximum solubility of silica in sodium aluminate liquors and gave the correlation (Si02)max (g dm3) = (Na20)(A1203) (2.12) where (Na20) is the concentration in g dm3 Na20 of caustic or free soda and (A1203) is the concentration of alumina also in g dm3. The maximum achievable silica concentration in Bayer liquors (Equation 2.12) is approximately twice that of the equilibrium concentration (Equation 2.10). Jamialahmadi and constant (Kr) for the precipitation reaction: set up an equation to calculate equilibrium Na2SiO3 + 2NaAlO2 + H20 1/2(Na20.Al2O32SiO2) + 2NaOH. (2.13) = n(T) T xl05T2 (2.14) where T is temperature (K). (NAO + NBO + + NDO)NDO2 1 NBO CO Ae They defined NAO = initial concentration of Na2SiO3 in the liquor, g dm3, NBO = initial concentration of NaA1O2 in the liquor, g dnr3, = initial concentration of H20 in the liquor, g dnr3, = initial concentration of NaOH in the liquor, g dnr3, NAe = equilibrium Na2SiO3 concentration in the liquor. At the same time NAO, NBO, Nco and NDO were also. defined to mean the moles of the relative component in their Table 1. Using both definitions calculated solubility values are much lower than those reported by Breuer et au'91, Ni et 46

72 al.'1561, Eremin et Kraus et al)'571 and Oku and Yamada'162]. Jamialahmadi et however, reported that they had very good agreement of predicted silica solubility and experimental results. Summary: Only two factors, caustic and alumina concentrations, have been taken into account to predict SiO2 solubility in some of the published empirical equations. However, temperature has also been shown to affect solubility as shown by other publications Desilication of Bayer Liquors There are three basic steps for the crystallisation process: achievement of supersaturation, nucleation and crystal growth'177]. Desilication during bauxite digestion involves two essential Dissolution: 158; 178]: (A12O32SiO2,2H20) + 6(OH) 2SiO A1(OH)4+ H20 (2.16) kaolinite Desilication: 6SiO32 + 6A1(OH)4 + Na8 [AlSiO4 16 (OH)2.nH2O+ 100H + (6-n)H20 (2.17) Dissolution is required to supersaturate the liquor to the point where precipitation takes place. The sodalite formed acts as a seed for the precipitation of more sodalite. The overall desilication process in digestion and pre-digestion (Section ) is feasible and [11; 176] endothermic Traditionally, reactive Si02, in the form of kaolin, is added in conjunction with bauxite to the Bayer process in order to effectively desilicate the Bayer liquor The addition of reactive Si02, although consuming large quantities of soda, is beneficial as the resulting locks up impurities such as sulphate anions into its structure thus immobilising them for removal from the liquor. The formation of sodium aluminosilicate is not always harmful. Bohmer'180] attempted to recover soda from sodium silicate by adding sodium aluminate to a solution of soda ash (containing up soluble sodium metasilicate) to form sodium aluminosilicate (zeolite). 4Na25i03 + 2NaA1O2 + H20 Na20.Al203.4SiO2 + 8NaOH (2.18) As sodium aluminate is generally an expensive chemical, he subsequently seeded sodium silicate with a blend of bauxite from more than one source. A reduction of 90% in the 47

73 Si02 was achieved. Another way of making use of the sodium aluminosilicate produced in Bayer plants is by dissolving it with H2S04, and then adding NaOH to the solution to make an aluminosilicate gel for the manufacture of zeolite. A commercial production of zeolite utilises this Methods of Desilication Pre-digestion Desilication In the low-temperature treatment (110 C) of bauxites having less than 2% Si02, a predesilication step is usually carried out. This step involves the holding of a caustic bauxite slurry with high solids content at approximately C for eight hours or 178; 1811, or leaching low reactive Si02 bauxite at a higher temperature at C in a liquor of NaOH concentration of M for 1.5 Sodalite is the desilication 27; Pre-digestion-desilication with the Aid of CaO One method for predigestion desilication for high silica bauxite and clay was to sinter it 183] using soda and A high-silicon product was obtained, thereby purifying the alumina. Bhoray, Sampath and Bhatnagars showed that after calcining at 950 C for 4 hours 90.08% of the Si02 in high Si02 bauxite was removed Post-digestion Desilication Pre-digestion desilication is quite slow and requires additional vessels. Longer holding times are costly in terms of energy and may lead to a significant reversion of Al(OH)3 to boehmite, which is not desirable. An alternative is a post-digestion desilication step in which polymer coated sodalite particles are used to seed the crystallisation of Bayer Other Methods for Desilication A method used to decrease silicate concentration in sodium aluminate solution to several ppm was to employ Ca(OH)2 suspensions to precipitate slightly soluble calcium 185] It was found that the higher the Ca(OH)2 concentration, the faster the desilication With the addition of Ca(OH)2, Si02 concentration could be reduced from ( )x103 M to 3.3x104 M within 100 minutes in a solution of Al(OH) M, NaOH M, Na2CO3 up to M. Comparing spent Bayer liquor of 1.65 M Al(OH)3, NaOH 4.52 M, Na2CO3 up to 0.38 M and Si M, their concentrations were much lower. A major drawback of this approach is that it requires 4X

74 considerable amounts of lime and involves additional chemically-bonded alumina losses1271. Another method of removing silicates from solutions without addition of chemicals1185' consists of the precipitation of sodium aluminosilicate by increasing the temperature of the solution from 147 to 207 C. This produces a decrease in Si02 concentration in the solution to a level of ( )x103 Similar to the principle above, Arlyuk, Mironov and Eremin, Mel'nikova and a two step desilication process to produce alumina with a given Si02 composition. The first step was conducted at 98 C in the presence of 100 g dm3 recycled sodium hydroxy aluminosilicate seed or at C in an autoclave with no seed (the latter was similar to that of A second, thorough desilication stage was conducted at C in the presence of 1891 or tricalcium hydroxy a desilication agent containing CaO, Al203, Si02 Fe203 and The starting Si02 concentration was either 0.05 or M. After 5 hours of the first stage, Si02 concentration decreased to 7.5x103 M at 98 C in the presence of the seed and to 4.2x103 M at 175 C in an autoclave with no seed added. The starting Si02 concentration at the second stage was 4.2x103 or 6.7x103 M. After 5 hours of desilication in the presence of 0.18 M lime, the Si02 concentration decreased to 1.7x103 addition of lime, the following reactions took place: As the solutions were thoroughly desilicated with the 87-3Ca(OH)2+ 2AlO2+ 0.3SiO CaO.A SiO2.5.6H W (2.19) Ca(OH)2 + C032 CaCO3 + 20ff (2.20) Al02 + 2H20 Al(OH)3 + OW (2.21) where Reaction (2.19) represented the formation of hydrogarnet containing moles of Si02; Reaction (2.20) the causticisation of the solution by lime, and Reaction (2.21) hydrolysis of the According to Reaction (2.19) the thorough desilication was carried out with a sacrifice of a larger amount of alumina than that of silica. Arlyuk et al.'s11187' effort was focused on reducing such an aluminate loss in the process. The stability of hydrogarnet was determined by Rozen and pevzner1t921 and Sizyakov, Vysotskaya, Pavletlkoand It was found that hydrogarnet was stable below 145 C. A process for removing Si02 from aqueous solutions (not spent Bayer liquor) was invented by Browne11941 using the addition of a polyelectrolyte flocculent to the liquor. His idea of removing Si02 by flocculent may be of relevance to precipitate/adsorb Si02 49

75 from Bayer liquor. Flotation has been used to separate quartz from bauxite-containing rocks with a mixture of industrial oil, cationate 2B and ANP-14 or acetate of colophony froth. 90% of the free silica was left in the flotation cell with only 10 % present in A further potential method for descaling is to use magnetic treatment. successfully used this method in water treatment. It was shown that the use of magnetic fields in fluids can result in: (1) changes in particle size; (2) changes in crystallinity; (3) changes in crystal morphology; (4) changes in crystal phase; (5) changes in solubility; and (6) changes in rate of precipitation. This idea may also be useful in aluminosilicate scale control but has not as yet been put into practice. Summary: Pre-digestion-desilication, post-digestion-desilication, desilication with the aid of lime and other chemical and physical methods all can be used to decrease Si02 concentration in Bayer liquors Factors that Influence Desilication The rate of desilication can be influenced by: (1) 17; 22; 27; 83; 130; 171; ; ] affects the rate of desilication. The kinetics of desilication increase with reaction temperature. Nucleation and growth of sodium aluminosilicate may be more rapid at higher temperatures even though the supersaturation may be rapidly although the supersaturation is least in those vessels. has The highest temperature heat exchangers scale most 198] (2) Initial Si02 affects the initial rate of desilication. Silica levels in bauxite also influence desilication rate decreases with decreasing reactive Si02 content due to the low Si02 supersaturations prevailing and hence, the lack of formation of sodalite seed. Figure 2.6 shows desilication rate of high and low Si02 bauxite ores. It can be seen that the solution from high 5i02 bauxite has a faster desilication' rate than low Si02 bauxite. According to the rate law, the higher the supersaturation, AC, the faster the desilication. dac E = (2.22) where supersaturation AC = C - is the difference between the Si02 concentration C and the equilibrium Si02 concentration k0 is the pre-exponential factor; E is the activation energy of desilication; R is ideal gas constant; T is temperature and n is the order of the desilication. 51)

76 Metasta le zone Critical supersaturation for homogeneous nucleation c') ::::::::::::::::::::::. - -:-:: Low silica bauxite hqtiur High silica bauxite Sodalite solubility Digestion Time Figure 2.6 Desilication rates of high and low silica bauxite during Bayer A higher silica supersaturation increases the likelihood of nucleation Local conditions at metal substrate (vessels) solution interfaces may enhance nucleation. For example, pipe and tank walls are often cooler than the liquor hence the local supersaturation at the surface will be higher, and nucleation will be more favourable. Pre-digestion desilication offers the advantage that the resulting slurry can be passed through heat exchangers without causing as much (3) It has been demonstrated that desilication rate is inversely proportional to caustic 17; 27; 83; 130; 177; 199] (4) The rate of desilication decreases with increasing Al(OH)3 concentration or Al(OH)3/NaOH 17; 83; 130; 177] (5) The presence of sodalite seed and seed type affect desilication ra1es120 27; 130; 138; 158; 171; 177; 197; 200] After three uses of sodalite seed for desilication, the seed becomes The reasons for seed deactivation are as yet unknown. Flint et found that improvement can be made by boiling the used seed for 2 hours with a solution saturated with NaC1. 199] (6) The presence of salts including chloride but particularly carbonate and sulphate, stabilise the sodalite structure and thus lower the equilibrium solubility of 22; 145; 171] Si02 in the Avdeeva and investigated the effect of salts such as Na2SO4, NaC1 and Na2CO3 on the desilication of sodium aluminate solutions (Al(OH)3 = M, NaOH = 2.92 M and Si02 initial concentration 0.07 M) at 175 C. The presence of salts caused the Si02 concentration in sodium aluminate solutions to decrease more rapidly than without the addition of salts. Avdeeva et a1j145'.51

77 found that under actual desilication conditions (175 C, after 3 hours) Na2CO3 and especially Na2SO4 reduced the Si02 content. Under more rigorous desilications conditions (with times longer than 3 hours and the temperature greater than 175 C) the equilibrium Si02 observed on the addition of the NaC1 was the same as for the addition of Na2SO4. However S042 gave rise to a more rapid desilication than Cl Added salts (e.g. Na2CO3, Na2SO4, NaNO3, NaC1, and NaBr) have been claimed to improve desilication; however, amounts above 100 g dm-3 were At the 5 g dm-3 level, no significant effect on desilication was noted upon addition of Na103 (sodium iodate), Na2V2O5 (sodium vanadate), NaVF5 (sodium fluorovanadate), NaCN (sodium cyanide), NaBO2 (sodium borate), NaF (sodium fluoride), Nal (sodium iodide), NaOOCH (sodium formate), NaOOCCH3 (sodium acetate), or NaOOC(CH2)2COONa (sodium succinate). Desilication was improved slightly in the presence of NaSCN (sodium thiocyanate), Na2S (sodium sulphide), Na2C2O4 (sodium oxalate), Na2C4H4O6 (sodium tartrate), and Na3C6H5O7 (sodium citrate). Better desilication occurred with Na2CO3, Na3PO4, Na2S2O3, and NaNO3. Na2SO4 gave very much improved desilication 25 g dm3 of Na2C2O4 (sodium oxalate) or Na2CO3 was as effective as 5 g dm3 Na2SO4; 15 g dm-3 Na2SO4 was extremely effective, and equally so in liquors without Na2CO3 as with Na2CO3. In an autoclave synthesis of sodium hydroxy aluminosilicate (100 C) the effect of the salts upon the extent of desilication increased in the following sequence: S2 < S032 < S042 i.e. the larger the anions the better the desilication Possible reasons for the influence of salts producing more rapid desilication may be: a lower solubility of the solid phase (refer to Section 2.4.5) and higher ionic strength which decreases the stability of colloidal suspension, and highly charged anions may coprecipitate onto aluminosilicate framework aiding crystallisation. 130; 158; 197] (7) The properties of the seed such as surface and crystal structure of the can affect desilication behaviour. The kinetics of desilication are proportional' to seed surface area. Nucleation is more rapid on a solid surface of similar structure or chemistry to the crystallising phase. (8) The nature and condition of initial materials [49; 83; 199] e.g. types of bauxite, kaolin or clay or quartz impurities and (9) Flow condition of the liquor or 198; 201] also affects desilication. Summary: Many factors such as temperature, Si02 supersaturation, caustic concentration, aluminium concentration, seeding and surface properties of the seed, i,npurities and flow conditions influence desilication. 57.

78 2.6. Kinetics of Sodium Aluminosilicate Crystallisation and Dissolution Kinetics of Crystallisation The kinetics of sodalite and cancrinite crystallisation in Bayer liquor have not been the subject of as many studies as has zeolite crystallisation. Crystallisation kinetics should obey the rate law given in formula (2.22) even if there is more than one solid phase involved in the crystallisation. When this is the case, it is very difficult to characterise the contributions of the individual processes involving different phases. Experimental conditions of some published work are listed in Table 2.16 and the results in Table 2.17[202]. Table 2.16 The experimental conditions of some of the studies published. I Authors T ( C) 0 k u a n d NaOH (M) A1(OH)3 INaOH pregnant liquor Liquor Seed charge (g dm3) Solid phase Max. Si02 (M) plant 0 bauxite synth 0 bauxite 7.5x103 Arlyuk et a1j163' synth 10 red mud Cousineau and synth DSP synth 10 DSP 0.03 The desilication rate was found to be first order with respect to Si02 by Adamson eial)118' and Arlyuk et aij'631. However, 158; 203] Oku and Yamada'11621, Cousineau and Murakami et at. [197] and 0?Neill[2061 proposed a second order desilication rate. even proposed a third order desilication rate when mud was used as the solid phase. The activation energies of the desilication reaction in both seeded and unseeded crystallisations were in the range of kj mol' except for 17 kj mold reported by Murakami et alj'97' (Table 2.17). This high value (> 40 kj mo11) indicates that the desilication reaction is surface reaction controlled, as opposed to The reason for the low activation energy in Murakami et al.'s case might be that their crystallisation is in a diffusion controlled region (seed charge up to 100 g dm3). carried out an

79 . investigation of desilication kinetics at 110 C with no addition of Na2CO3, A1(OH)3/NaOH (molar) = 0.41, initial Si02 concentration M and seed charge of 10 g dm-3. The rate constant was found to be 4.52 x 10-2 M1 min1, a little larger than those obtained by Oku and Yamada (3.22 x 10-2 M1 and O'Neill (2.59 x 10-2 M-1 minl)[206]. No indication of the phase of the seed given. The seed was prepared by boiling aluminosilicate solution at 110 C overnight. It is likely that seed was Table 2.17 Kinetic data from published results. Authors Order n 1 bauxite Solid phase Rate constant Activation 5820 exp( T energy (kj mol1) 103 Adamson et 1 aluminosilicate 38.3 Arlyuk etal)'163' 1 mud (min1) (at 88 C) 25.1 O k u 62] & 2 bauxite exp( x103a \ * T Muller- 2 2 aluminosilicate 2 2 aluminosilicate T T T 2 T et 2 aluminosilicate 2 * where A means A1203 concentration (g dm3). T T 233 Summary: Some investigations have been carried out on Bayer liquor desilication kinetics. Conflicting conclusions regarding reaction order were reported (ii = 1, 2 or 3).

80 Kinetics of Dissolution Smirnov1208' investigated the leaching of gibbsitic bauxite, finely ground bauxite sinter and finely ground nepheline sinter. The Si02 concentration versus time is shown in Figure TJ) Time (hour) Figure 2.7 leaching of nepheline Si02 concentration in aluminate solutions as a function of time during the hydrargillite bauxite, (.) finely ground bauxite sinter and (o) finely ground A similar shape of curve was obtained by on the conversion of bauxite Si02 to sodium aluminosilicate and Roach and on the dissolution kinetics of kaolinite in caustic solutions. Pezvner, Eremin, Rozen, Kolobov and studied Si02 concetitration in aluminate solution in the presence of amorphous sodium aluminosilicate hydrate at 25 to 65 C. Yuhas, Orbanne and observed similar Si02-time curves for amorphous sodium aluminosilicate dissolution in aluminate solutions at 75 and 100 C. Simultaneous kaolinite, bauxite or amorphous sodium aluminosilicate dissolution and solution desilication or solid phase transformation 208] 157; 181; 207- A similar SiO2-time curve was observed with sodalite and cancrinite dissolution in synthetic spent liquor. It is known that, in the initial period of contact between Si02 and a caustic aluminate solution, the rate of dissolution of the Si02 is greater than the rate of formation of the 55

81 sparingly soluble sodium aluminosilicate. In the first 2-3 hours of interaction, the maximum metastable concentration of silica is established in the solution subsequently decreases as the process of formation of the sparingly soluble sodium aluminosilicate predominates. Derevyankin, Kraus and Kuznetsov1209' studied the rate of dissolution of sodium aluminosilicate in caustic solutions. The passage of Al(OH)3 and Si02 into the alkaline solution takes place rapidly during the first minutes; then, the rate of hydroxy aluminosilicate dissolution drops and at a point minutes from the start of the mixing, equilibrium is established. The dissolution activation energy for aluminosilicate in alkali solution of Al203 and Si02 respectively are 16.2 ± 1.38 kj mo!-1 and ± 1.55 kj mol-1. The equilibrium A1(OH)3 concentration in sodium hydroxy aluminosilicate dissolution experiment changes with temperature and, appears not to depend on the alkali concentration when the temperature is Summary: Dissolution of sodium aluminosilicate involved two opposite reactions ((1). dissolution, fast and increases solution Si02 concentrations and (2). phase transformation, slow and decreases solution Si02 concentrations) Mechanism of Sodium Aluminosilicate Formation Muller-Steinhagen, Jamialahmadi and investigated the formation of sodium aluminosilicate between 30 to 105 C on heat transfer surfaces. They found that at low silica concentration ( M) aluminosilicate scale mainly deposited on heat exchanger pipes. Deposition increased with increasing heat exchanger surface temperature and silica concentration. At high silica concentration (>0.027 M) bulk precipitation and particulate deposition occurred. It has been suggested that silica is dissolved in solution in the form of sodium silicate. The latter with sodium aluminate to form a soluble aluminosilicate which readily hydrolyses and is gradually converted to insoluble sodium The mechanism of the precipitation of Si02 was found to be of two There is precipitation with aluminium hydroxide in a fixed ratio of Si to Al for a given Si/Al in the solution. This ratio increases in the solid with increasing 5i02 concentration and decreases with increasing caustic concentration; it is independent of Al(OH)3 concentration, temperature and the amount of 5i02 introduced with the hydrate seed. This type of precipitation is referred to as co-precipitation. The second mechanism is the precipitation of sodalite on sodalite nodules which exist on the seed. S6

82 proposed a mechanism of sodium aluminosilicate formation scale in black liquor in kraft recovery cycles in the paper industry (Figure 2.8). It was shown that at low ph morphologically cubic particles formed mainly in solution and at high ph morphologically hexagonal structure scale formed on the tube wall. The break point between the two types of solid formation is roughly at NaOH 0.8 M in a 4 M NaOOCCH3 medium. Cubic structure NaA1SiO4 1/3Na')X(s) (more or less amorphous) + Low [Off] Liquor H20 Particles scale tube wall High [OH-I Scale Hexagonal structure Figure 2.8 The chemical model for the formation of solid sodium aluminosilicate. X2 = C032, S042, 20H, 2HS-, etc!'401. Roach and postulated two basic mechanisms for scale formation in Bayer process liquors: nucleation and growth of aluminosilicate from solution (growth scale) and the settling of boehmite-aluminosilicate slurry particles followed by cementation (settled scale). Growth scale was normally a hard, brittle, highly crystalline, nonporous material with well developed crystal faces. Settled scales were frequently porous and soft. Sodium oxalate was found to precipitate with the scale. The growth scale was relatively insoluble in caustic wash liquor, hence an alternating cleaning cycle of caustic and water was required. has concluded that aluminosilicate scale deposits formed in Bayer plant heat exchangers are the results of chemical reactions involving both solution ionic species and reaction products on the heated metallic surface. Ueda et re'ported that hydroxy sodalite and zeolite B crystallised from clear 19] aqueous solutions. described the mechanism of zeolite crystallisation in terms of a quasi-equilibrium between the solid and liquid phase in gels and emphasised that the formation and growth of nuclei occurred in the liquid phase. The gel solids served as nutrient and dissolved continuously during crystallisation with bulk transport of the 57

83 dissolved species to the growing nuclei or crystals in the liquid phase. The resulting mechanism was analogous to homogeneous nucleation and growth from solution as initially proposed by and supported by i.e. growth occurring from solution, active aluminosilicate species as well as silicate and aluminate ions. In highly caustic spent Bayer liquor, this latter mechanism is more acceptable. The study of the kinetics of crystallisation of zeolite ZMS-5 from aluminosilicate gel in the presence of organic templates at 150, 160 and 180 C showed that all kinetic curves involved two regions: nucleation and crystal The nucleation region could be subdivided into two main periods: induction period when no crystal phase was found, and transition period involving a slow growth of the crystal phase. A third period of relatively rapid crystal growth followed. Two mechanisms have been proposed for the synthesis of zeolite: the solution phase reaction 72; 73; 88; 119; 121; 138; 140; 200; ] and the solid phase 70; 74; transformation The former authors postulated that crystallisation was achieved by the deposition of structured species formed from previously precipitated amorphous aluminosilicate which dissolved and rearranged The latter, on the other hand, proposed the direct transformation of amorphous solid gel to crystalline phases. 138; 143; 217; 221] The peculiarities of zeolite crystallisation published in several were induction period, the auto catalytic nature of the process, increase of crystallisation rate with increasing alkali concentration, Si02 and A1203 source, mixing technique, aging and effects of seeding on the The cause of an induction period is 221] The induction period decreases with increasing temperature and alkaline concentration. It also depends on the nature of the initial aluminosilicate materials utilised in the synthesis. The induction period may be more obvious in gel synthesis as the caustic concentration is usually low. During the formation of sodalite and cancrinite in Bayer liquor the caustic concentration and temperature are extremely high. The induction period may be trivial in these cases. The formation of primary particles of zeolite P was observed to take place by two nucleation The first was the rapid heterogeneous nucleation in the liquor phase, and the second was the formation of particles of quasi-crystalline phase inside the gel matrix. The became the active nuclei after their release from the dissolved part of the gel and consequent full contact with the liquor phase. Adu-Wusu and and Jepson, Jeffs and the kinetics of the silicate precipitation reaction with gibbsite in sodium aluminate solution. Silicate was believed to have adsorbed and precipitated as sodium aluminosilicate on gibbsite by a

84 three-step reaction: a first fast step, a lag period step, and a slow step. The fast and slow steps were confirmed to be of first-order rate law with respect to silicate concentration. The extent of sorption depended on the silicate concentration, suspension concentration and temperature. Silicate increased the equilibrium solubility of gibbsite in sodium aluminate solution Summary (1) The formation of sodium aluminosilicate scale in Bayer plant may be reduced by predigestion desilication, post-digestion desilication or the addition of CaO to the liquor. The main reason for the scale formation in spent liquor is that, after gibbsite precipitation, Al(OH)3/NaOH is reduced to half, Si02 solubility, consequently is also reduced to half. The liquor thus becomes supersaturated with respect to Si02. (2) The solubility of Si02 in spent liquor is also influenced by the solid phases present, temperature, caustic concentration and impurities in the liquor as well as the A1(OH)3INaOH ratio. (3) The kinetics of sodium aluminosilicate crystallisation are influenced by temperature, caustic concentration, A1(OH)3INaOH ratio of the liquor, seed type and impurities in the liquor. Sodium aluminosilicate precipitation in spent Bayer liquor appears to be surface reaction controlled and not mass transfer limited. (4) The phases of the scale formed are sodalite and cancrinite. Sodalite has a cubic crystal system and belongs to the P43n space group. Cancrinite has a hexagonal crystal system and belongs to the P63 space group. The difference between these two crystalline phases can be characterised by XRD and FTIR analysis. Sodalite may transform to cancrinite at high temperatures and long reaction times. (5) The main impurity present in Bayer liquor is Na2CO3. The presence of impurities may reduce Si02 solubility and facilitate solid phase transformation.

85 CHAPTER 3. EXPERIMENTAL METHOD 3.1. Introduction Scale formation in heat exchangers occurs under very severe conditions: high temperature (from C), high pressure ( kpa), high caustic concentration (4.52 M NaOH) and high flow rate (1400 m3 h-1)[2241. Due to the extreme conditions required for experimental studies, it is difficult to use in situ techniques to monitor the crystallisation process. Ex situ techniques such as XRD (X-ray diffraction), FTIR (Fourier transform infrared spectroscopy), SEM (scanning electronic microscopy) and EDS (energy dispersion spectroscopy), XPS (X-ray photoelectron spectroscopy) and particle sizing were used here to characterise crystalline products and monitor any change in crystalline phases. For solution Si02 concentration analysis, ICP (inductively coupled plasma) and UV-VIS (ultra violet-visible spectroscopy) (colorimetric molybdate method) were used Crystallisers Low and high pressure (for elevated temperatures) and caustic resistant reactor vessels are needed. A batch process was used throughout this thesis as it provides a simple, more flexible and less time consuming method than the classical continuous crystalliser. The solution Si02 concentration is an important factor in the formation of sodium aluminosilicate scale. Si02 glass containers were not used during experiments except for the preparation of solutions containing no caustic or after quenching and acidification of sample solutions. Care should always be taken in handling the highly alkaline experimental solutions. Safety glasses and rubber gloves should be worn during sample Atmospheric Pressure Reactor for under 100 C A 1.2 dm3, 316 stainless steel reactor vessel was used for the crystallisation experiments carried out at under 100 C (refer to Chapters 4 and 5). A central, 4-blade 45 -pitch turbine impeller driven by a 70 W, multi-speed motor (Jank and Gunkel, GMbH, Staufen) provided a constant agitation speed to within ± 2 rpm in the crystalliser. To prevent evaporation, a stainjess steel lid containing ports for the impeller and a thermometer, and for sampling was screw sealed onto the vessel. Mixing was provided at 600 rpm unless otherwise stated. An illustration of the vessel is shown in Figure

86 The crystalliser was submerged in a 15 dm3, thermostatically-controlled oil bath (Thermoline Scientific Equipment, Australia), which maintained a constant temperature to within ± 0.1 C. Impeller shaft Sample port \ A rhefmometer v- / L_ I I _JI. Liquid level Figure 3.1 Illustration of the stainless steel reactor vessel (all dimensions in mm) Parr Bombs Experiments Six 4746 and 4747 High Pressure Parr Bombs (Parr Instrument company, U. S. A.) were used in the high temperature and high pressure non-stirred experiments (refer to Chapters 4, 5 and 6)[2261. They are spring loaded, heavy stainless steel vessels which will withstand pressures up to 5000 psig at temperatures as high as 275 C. A teflon liner of volume 23 or 27 cm3 (A255AC) with a long high strength taper seal is enclosed in each bomb to hold the solutions. A safety rupture disk is built into the bottom of the vessel which will allow, at excess pressure, the contents to blow out through the bottom of the teflon cup. In the experiments the Parr Bombs were kept, without any agitation, in the chamber of a thermostatically controlled oven which provided isothermal heating (± 0.3 C) Autoclave Experiments at High Pressures A high pressure crystalliser was used to simulate the severe plant heat exchanger conditions. An autoclave with a volume of 600 cm3 and a programmable temperature controller (4843, Parr Instrument Company, U. S. A.) was used in precipitation and 61

87 dissolution experiments (refer to Chapters 7 and 8). The crystalliser was mechanically stirred at 400 ± 2 rpm and was externally heated by a heating mantle. The highest temperature used is 200 C although higher temperature such as 300 C may be achieved. Experiments showed that this stirring speed does not cause any gross breakage of seed. A 316 stainless steel liner with a volume of 480 cm3 was used to protect the inner surface of the reactor from corrosion due to caustic attack and to simulate the heat exchanger surface. Slurries or solutions to be added to the pressurised autoclave were placed in a 50 cm3 stainless steel cylinder connected to a nitrogen supply line attached to the autoclave via stainless steel tubing and an one way valve. The cylinder was then flushed with nitrogen at a higher pressure to force its contents into the autoclave. Autoclave solution sampling was carried out in a reverse manner. Nitrogen was flushed into the autoclave to force a representative slurry sample up a the sampling tube through a sampling valve connected to a 50 cm3 steel sample receiver. The sampling valve was then shut and the container detached from the autoclave Preparation of Synthetic Spent Bayer Liquor Bauxite ores have a very complex mineralogy and as a result, Bayer liquors may contain traces of several inorganic compounds. In order to study the scale formation mechanism in a systematic manner with minimum complexity, synthetic spent liquor containing the key solution species of sodium, aluminate, carbonate and silica was used. The concentration of the components is shown in Table Preparation of the Liquor under Atmospheric Conditions The chemicals used to prepare synthetic spent Bayer liquor were: (1) sodium hydroxide, NaOH (Ajax Chemicals, Australia, laboratory reagent, greater than 97.5%, maximum limits of sodium carbonate (Na2CO3) impurity 2.5%), (2) aluminium hydroxide (C-31), Al(OH)3 (ALCOA, Arkansas, USA, 99.4%), (3) sodium metasilicate, Na2SiO3.5H20 (Ajax Chemicals, Australia, technical grade), (4) sodium carbonate, Na2CO3 (BDH Chemicals, Australia, analytical reagent, 99.9%). The minimum concentration of Na2CO3 in any of the synthetic spent liquors was M. This was due to the equilibrium concentration of Na2CO3 in the solid NaOH. Water for solution preparation or equipment washing, except where specified, was Milli-Q water. This was purified through a Millipore Super-Q water purification system, i.e. reverse osmosis, two stages of ion exchange and two stages of activated carbon prior to final filtration through a 0.22 pm filter. At 25 C, the surface tension of Milli-Q water was 72.8 ± 0.2 rnn m1, its conductivity was always less than 0.5 and the ph was 5.8 ± 0.2 (C032 concentration 3.76x M if saturated with 62

88 To prepare a given volume of synthetic spent Bayer the required amount of sodium metasilicate was placed in a stainless reactor with 38-40% of the required amount of Milli-Q water. The reactor was stirred on a hot-plate using a magnetic flea. The required amount of NaOH pellets were slowly added to the reactor. When the caustic and sodium metasilicate have dissolved, the solution temperature was increased to near boiling. The required amount of C-3 1 (Al(OH)3) was then added. A minimum of Milli-Q water was used to wash down the walls of the reactor. The solution was kept near boiling until the C-3 1 had dissolved. Na2CO3 was dissolved in the remainder of the water in a separate stirred and heated stainless steel beaker. When the hydrate had dissolved, the Na2CO3 solution was added to the liquor and mixed well. This method of liquor preparation was used in all the experiment run in the 1.2 dm3 stainless steel reactor vessel and presented in Chapters 4 and Preparation of the Liquor for Autoclave Experiments The chemicals used were the same as described in Section High purity oxygen-free nitrogen (CIG, > 99.99%) was used in the nitrogen line of the autoclave for solution and seed addition, and sample tube clearing prior to sampling. When autoclave experiments were carried out (refer to Chapters 7 and 8), liquors were prepared in the following manner. In the stainless steel liner of the autoclave, 40% of the required amount of Milli-Q water was added. The liner was stirred using a magnetic bar and heated on a hot-plate. The required amount of NaOH was gradually added to the liner. The required amount of C-3 1 gibbsite was then added to the liner after all the caustic had dissolved. The liner was placed in the autoclave. The autoclave was assembled. The mixture was heated up to. 160 C with a 400 rpm agitation rate for at least 20 minutes to dissolve the gibbsite. Samples taken after 20 minutes at 160 C were optically clear, thus indicating complete dissolution of gibbsite and NaOH. After gibbsite dissolution (at 160 C), the solution was cooled to the required temperature before the metasilicate solution was added. The required amount of sodium metasilicate was dissolved in 50 cm3 of Milli-Q water in a glass beaker. The beaker was heated on a hot-plate and stirred with a magnetic bar. The required amount of Na2CO3 was added. Stirring and heating were continued until all the solids dissolved. The sodium metasilicate plus Na2CO3 solution was added to the autoclave through the cylinder on the nitrogen line. The beaker and the cylinder were washed three times with the remaining water. Hence the solution was diluted to the required concentration. 63

89 Preparation of the Liquors for Parr Bomb Experiments The chemicals used were the same as those described in Section Two sets of stock solutions containing Si02 either 0.01 M (for experiments commencing above the Si02 solubility) or 0.0 M (for experiments in which the Si02 solubility was approached from below) and 4.52 M of NaOH were prepared with Na2CO3 concentrations of 0.043, 0.11, 0.18, 0.24, 0.31 and 0.38 M. The required amount of sodium metasilicate was dissolved in half of the required water in a stainless steel beaker. The required amount of Na2CO3 was dissolved in the remaining water in a separate stainless steel beaker. Both solutions were heated and stirred. After the NaOH and Na2CO3 had dissolved, these two solutions were combined and mixed. The stock solutions were stored in polymethylpentene (PMP) bottles Seed Preparation Sodalite Seed Unless otherwise stated, the same chemicals as in Section were used for seed preparation. Aqueous ammonia solution (BDH Chemicals, Australia, analytical reagent, sp. gr. 0.91, about 25% NH3) was also used. Sodalite seed was synthesised at 95 C in the stainless steel reactor vessel described in Section g of NaOH was dissolved in 300 cm3 of Milli-Q water in the reactor which was heated on a hot-plate and stirred by a magnetic stirrer. 112 g of Al(OH)3 was added to the reactor after the NaOH had dissolved. The mixture was heated to near boiling. When the solution was nearly optically clear, the reactor was removed from the hot-plate and placed in an oil bath. The magnetic stirrer was removed and a mechanical stirrer at a speed of 600 rpm was then used. In a stainless steel beaker, 152 g of sodium metasilicate was dissolved in 600 cm3 of Milli- Q water. This solution was added to the reactor in the oil bath through a sample port in the lid. The reactor was kept in the oil bath at 95 C (solution temperature) for 4 days. The product was filtered through 0.20 membrane, washed with 10% aqueous ammonia solution and ethanol three times each, and dried in an oven at 120 C overnight. XRD and FTIR analysis (Figures 5.1 la and 5. 12a) were performed and confirmed that the product was sodalite with d unit cell size of a = ± A. The density of the sodalite seed was 2.14 g cm-3 and the BET surface area of it is 3.2 m2 g1. The crystal size range was between 0.2 to 30 jim (Figure 7.7). 64

90 Cancrinite Seed Initially Barrer et method was used to synthesise cancrinite at low temperatures using both kaolinite and synthetic spent Bayer liquor. The manner of synthetic cancrinite preparation was similar to that of sodalite except that 26 g of Na2SO4.10H20 (BDH Laboratory Chemicals Division Poole England, analytical reagent, 99.0%) was added to the gibbsite solution before the sodium metasilicate solution was poured into the reactor. The product was filtered after 3 days, washed and dried as for sodalite. XRD, FTIR and SEM characterisation (Figures 6.3a and 6.4a) were carried out. These indicated that the products were fibrous of sulphate cancrinite with a unit cell size of a = ± A and c = ± A. Synthetic cancrinite has only been used for solubility measurement not kinetic experiments Plant Cancrinite Seed A plant scale sample obtained from Queensland Alumina Limited was chosen as the plant cancrinite seed. It was collected from their heat exchanger #16C where Al(OH)3/NaOH (molar) = 0.41, NaOH = 4.5 M, T = 180 C and P = 1400 kpa weight % of hematite (Fe203) was shown to be present in the sample by X-ray fluorescence analysis. Although the sample appeared slightly pink the concentration of hematite was insufficient to give rise to a resolvable set of diffraction peaks in the X-ray powder diffraction pattern. The plant cancrinite sample was wet ground in a rod mill, settled and filtered through a 1 glass fibre filter. The density of plant cancrinite seed was 2.19 g cm-3 and the BET surface area is 2.3 m2 g-1. The crystal size range was between 0.1 to 30 jim (Figure 7.12) Crystallisation Experiments Formation of Sodium Aluminosilicate as a Function of Time In order to scale formation as a function of time, aluminosilicate scale precipitation experiments were run at 100 C for 15 days using the 1.2 dm3 stainless steel reactor vessel (described in 3.2.1). The liquor contained 4.52 M NaOH, 0.38 M Na2CO3, Al(OH)3/NaOH (molar) = 0.71 (this is similar to that of industrial spent liquors) and 0. 1 M Si02. Si02 concentration in plant spent liquors is approximately is 0.01 M. When this 5i02 concentratiorf was used no obvious scaling occurred within 5 days in the batch experiments. In order to obtain enough crystal sample, the concentration of 0.1 M of Si02 was chosen. Slurry samples were removed periodically using a 50 cm3 plastic syringe and filtered through 0.2 jim membrane. Filtrate solutions were stored in polymethylpentene (PMP) 65

91 bottles for analysis. Crystals were washed with 10% ammonium solution and ethanol until they were free of alkali and dried at room temperature in a fume hood overnight Solubility Measurements Solubility experiments (discussed in Chapter 6) were carried out in synthetic spent Bayer liquors (with fixed caustic concentration and alumina to caustic ratio) as a function of Na2CO3 concentrations at 90 and 160 C for over 13 days in six stainless steel Parr bombs (Section 3.2.2). Six stock solutions containing 0.01 (approaching Si02 solubility from above) or 0 (from below Si02 solubility) M of Si02 and 4.52 M of NaOH with Na2CO3 concentrations of 0.043, 0.11, 0.18, 0.24, 0.31 and 0.38 M (referred to in Section 3.3.3) were used in the solubility measurement experiments. The required weight of Al(OH)3 and approximately 0.3 g of seed were added to each of the 23 or 27 cm3 teflon insert before the experiments were started. The inserts were then filled with the required amount of the stock solutions and sealed. They were kept, without any agitation, in the chamber of a thermostatically controlled oven which provided isothermal heating at 90 and 160 ± 0.3 C. Samples were taken from the slurry periodically. These were centrifuged in order to monitor the change in Si02 concentration of the liquor. The supernatants were analysed by the ammonium molybdate method (refer to Section ) on a Cary 5 UV-Vis-NIR This sampling was repeated until the last three Si02 concentrations were within 4% of each other. Crystal samples taken from the solutions at the end of the experiments were washed with water and ethanol three times each to cleanse them of alkali, and dried in a fume hood overnight Crystallisation and Dissolution of Sodium Aluminosilicate Nucleation experiments were carried out in a sealed autoclave with a stainless steel insert of 480 cm3 capacity. For safety reasons, only 300 cm3 of synthetic spent liquor was used. No seed was used in the experiments examining homogeneous (solution)fheterogeneous (walls) nucleation (Chapter 7). In these experiments, the concentrations of NaOH and Al(OH)3 were fixed at 4.52 M and 1.65 M respectively; the concentration of the remaining components is in Table 3.1. Samples were taken via a sample valve to a plastic bottle if the experiment temperature was 90 C or to a sealed stainless cylinder if the temperature was 160 C. The samples were centrifuged and the supernatants were diluted and acidified for Si02 ICP analysis (Section ). 66

92 Synthetic sodalite and plant cancrinite were used as seed for the desilication experiments (Chapter 7). The experimental concentrations were the same as those in Table 3.1 and mentioned above. The seed charge was 5 g dm3 of solution The seed was added to the autoclave through the nitrogen line after the Al(OH)3 was dissolved, prior to the addition of the sodium metasilicate solution. A similar sampling procedure to that for the nucleation experiments was used except sample centrifuging was replaced by filtering. Table 3.1 The experimental conditions of the unseeded desilication experiments (the concentrations of NaOH and Al(OH)3 were fixed at 4.52 M and 1.65 M respectively). Exp No Si02 (M) Na2CO3 (M) Temperature ( C) ,'j 2 'I 'I 3 -'I "I 4 -.J 5 6 '1 7 8 'J 'J Dissolution of sodalite and plant cancrinite experiments (Chapter 8) were carried out using the liquor containing 0.0 M Si02 concentration and a 20 g dm-3 seed charge. The remaining components were kept the same as in the desilication experiments Liquor Composition Analysis Silica Concentration Analysis Molybdate Method The chemicals used for the molybdate method of solution Si02 concentration analysis (1) sulphuric acid, H2S04 (BDH Chemicals, Australia, 98%, sp. gr. 1.84), (2) ammonium molybdate, (NH4)6M070244H20 (Standard Laboratories Pty. Ltd., Australia, analytical reagent, 99.9%), (3) sodium metabisulphite Na2S2O5 (May & Baker Ltd., Dagenham England, laboratory chemicals, 93%), (4) hydroxylamine hydrochloride, NH2OHHC1 (May & Baker Ltd. Dagenham England, laboratory chemicals, 97%), 67

93 (5) boric acid, H3B 03 (The British Drug Houses Ltd: B.D.H. Laboratory Chemicals Division Poole England, analytical reagent, 99.5%), (6) tartaric acid, (CH0H.COOH)2 (Ajax Chemicals, Australia, analytical reagent, 99.5%), (7) sodium suiphite anhydrous, Na2SO3 (BDH Chemicals, Australia, analytical reagent, 97%), (8) 1 -amino-2-naphthol-4-sulphonic (ANS) acid, NH2C10H5(OH).SO3H (B.D.H. Laboratory Chemical Division, England, extra pure, 95%) and (9) standard silica, Si02 (British Chemical Standards, 99.6%). Reagents: (1) 1+1 diluted H2S04 solution: 1 volume of the 98% H2S04 diluted by 1 volume of Milli-Q water. (2) 1+3 diluted H2S04 solution: 1 volume of the 98% H2S04 diluted by 3 volumes of Milli-Q water. (3) Si02 standard solution I: 0.5 g of high purity standard Si02 was fused with 4 g of anhydrous Na2CO3 at 1000 C for 20 minutes in a platinum crucible in a muffle furnace. The crucible was removed from the furnace the next day and was placed in a 1000 cm3 beaker. 500 cm3 of hot water was added to the beaker. The beaker with the crucible and hot water was stirred on a magnetic stirrer/hot plate. While stirring, 10.0 cm3 of Reagent (2) was added to the beaker. The solution was quantitatively transferred to a 1000 cm3 volumetric flask when the melt was dissolved. The solution was cooled, diluted with Milli-Q water to the mark and mixed well. The solution was transferred to a polyethylene bottle. This standard was discarded after 6 months. (4) Si02 standard solution II: 20.0 cm3 of Reagent (3) was transferred to a 1000 cm3 volumetric flask, diluted with Milli-Q water to the mark and mixed well. The solution was transferred to a polyethylene bottle and discarded after 6 months. (5) 10% (w/v) hydroxylamine hydrochloride solution. The solution was discarded after two months. (6) 4% boric acid (w/v). (7) ammonium molybdate solution: 37.5 g of (NH4)6Mo7024.4H20 was dissolved in 400 cm3 of Milli-Q wat&. 50 cm3 of Reagent (1) was added. The solution was diluted to 500 cm3 and stored in a polyethylene bottle. It was discarded after four weeks. (8) 8% tartaric acid solution (w/v). It was discarded if mould appeared. 68

94 (9) ANS (1 -amino-2-naphthol-4-sulphonic acid) reducing solution: 1.75 g of anhydrous sodium suiphite (Na2SO3) was dissolved in 50 cm3 of Milli-Q water g of 1-amino- 2-naphthol-4-sulphonic acid (NH2C10H5(OH).SO3H) was added, and stirred until dissolution was complete. In a separate beaker, 22.5 g of anhydrous sodium metabisulphite was dissolved in 150 cm3 of Milli-Q water. The two solutions were combined and diluted to 250 cm3. The final solution was stored in polyethylene bottle and discarded after two weeks. Procedure: (1) 1 cm3 of Bayer liquor and 30 cm3 of Reagent (6) (to complex possible fluorides) were diluted with Milli-Q water to 150 cm3 in a plastic beaker. The solution ph was adjusted to 1.3 (± 0.1) by adding Reagent (2). The solution was transferred into a 250 cm3 volumetric flask quantitatively, diluted to the mark and mixed well. Blank and standard Si02 solution (using Reagent (3) as the source) were prepared the same way at the same time. (2) A 25 cm3 aliquot of each assay solution mentioned above (sample, blank and standard), was taken with a pipette and delivered into 100 cm3 volumetric flasks. In addition, two extra aliquots of one sample solution were provided for the use as standard additions, and 5 cm3 and 10 cm3 of Reagent (3) were added to each of them. (3) 3 cm3 of Reagent (5) (to reduce possible vanadium (V+) to (IV +)) was added to each of the aliquots prepared in the 100 cm3 volumetric flasks. The flasks was immersed in a water bath at 90 C, for a period of five minutes. (4) The flasks were removed from the bath. 30 cm3 of Reagent (6) (to further complex possible fluorides) was added. The flasks were cooled to room temperature in a cool water bath. (5) 3 cm3 of Reagent (7) (to form molybdenum silica complex) was added and mixed. The solution was let stand for 15 ± 3 minutes. (6) 8 cm3 of Reagent (8) (to destroy possible molybdophosphoric and molybdoarsenic acid) was added and mixed. (7) 2 cm3 of Reagent (9) (to produce molybdate blue complex) was added, The flasks were swirled to mix, diluted to the mark, mixed well, and stood for at least 20 minutes. (8) The absorbance of the solution due to molybdenum silica complexes was measured on a Cary 5 UV-Vis-Nir Spectrophotometer, zeroed on water, using wavelength of 810 nm (to minimise organic interferences) and 1 cm cells, and a slit width of 0.2 nm. 69

95 The worksheet for Si02 in liquor analysis is shown in Appendix A. 1. The standard error of this measurement is 4% ICP Method The chemicals used in the standard solution were: (1) aluminium ribbon Al (Merck, Germany, 99.9%); (2) sodium hydroxide, NaOH (Ajax Chemicals, Australia, laboratory reagent, greater than 97.5%, maximum limits of sodium carbonate (Na2CO3) impurity 2.5%), (3) Na2CO3 (BDH Chemicals, Australia, analytical reagent, 99.9%) and (4) hydrochloric acid, HCI (BDH Chemicals, Australia, 36%). The Si02 concentrations in spent Bayer liquors are relatively small compared to that of sodium and aluminium. The standard solutions have to have the same background matrix as for a similarly prepared Bayer liquor. Aluminium metal ribbon was cut and dissolved in 4.52 M NaOH solution. The liquor was stored in a PMP bottle in a refrigerator. Two sets of standard solutions were prepared with the extremes of the Na2CO3 concentrations used, and 0.38 M. A solution of 0.33 M Na2CO3 was prepared (because the other 0.05M Na2CO3 was already present due to the Na2CO3 in NaOH). 1.0 cm3 of the solution was transferred by a pipette to a 250 cm3 volumetric flask cm3 of Milli-Q water was added. 10 cm3 of 50% HC1 aqueous solution was added and mixed. The required amount of Reagent (3) or (4) (described in Section ) was taken to make the final Si02 concentrations of 0.0, 0.4, 1.0, 2.0, 4.0 and 10.0 ppm. The solutions were diluted to the mark, mixed well and transferred to polyethylene bottles. A similar preparation was performed to match the liquor with 0.38 M Na2CO3 by adding 1 cm3 of the 0.33 M Na2CO3 solution to the flask before the dilution. For preparation of the liquor samples cm3 of Milli-Q water was added to a 250 cm3 volumetric flask. A 1.0 cm3 aliquot of the liquor was transferred to the flask and mixed. 10 cm3 50% HC1 aqueous solution was added and mixed. The solution was diluted to 250 cm3. The samples were analysed by ICP for 5i02 concentration analysis. Operation conditions are listed in Appendix A.2. The standard error for this measurement is ±5% Comparison of the Two Methods The molybdate method for determining SiO2 concentration was relatively more accurate than the ICP method and had very good reproducibility but was very time consuming. The molybdate method was used in Si02 solubility measurement and the ICP method was used to determine SiO2 concentration later in crystallisation and dissolution experiments. Figure 3.2 shows the comparison of Si02 concentrations measured by these two methods for the same set of solutions. The concentrations obtained from the molybdate method 70

96 were always higher than those obtained by the ICP method. The relationship of the concentrations measured for these two methods was Cm (3.1) where Cm and refer to Si02 concentration measured by the molybdate and the ICP methods respectively E CID Si02(g dm1 ICP Figure 3.2 ICP. Comparison of Si02 concentration measured by the molybdate method and It is therefore possible to convert the results from one method to those of another according to Equation (3.1). Most of the kinetic results were measured by However, the solubility measurements of sodalite and cancrinite were completed with the Si02 concentration being determined by the molybdate method. As shown in Figure 3.2, for the same set of solutions the values determined by the two methods are different, however the deviation is systematic. When the driving force AC = C - C0 was calculated, C0 has to be chosen to be based on the same measurement technique as that of C. The reason for the systematic deviation between the two techniques is not known. All Si02 concentrations measured by moly6date method were converted to those as by the ICP method so that comparisons are based on the same technique. 71

97 Sodium Hydroxide, Sodium Carbonate and Aluminium Hydroxide Concentrations The sodium gluconate method was used for NaOH, Na2CO3 and Al(OH)3 concentration titrations In the first part of the reaction, the sample containing sodium aluminate, sodium carbonate and free sodium hydroxide is acidified with excess hydrochloric acid. The carbon dioxide is driven off by nitrogen purging. NaA1(OH)4 + Na2CO3 + NaOH + 7HC1 A1CI3 + co2'i' +4NaCI + 6H20 (3.2) Sodium gluconate is then added. Free acid and combined acid are released by complexing the aluminium with gluconate. The solution is titrated to ph 8.3. A1C13 + HC1 + ngl + 4NaOH + 4NaC1 + H20 (3.3) The net hydrochloric acid titration from Equations (3.2) and (3.3) represents total soda. Titration of carbonate to bicarbonate can be performed in an alkaline aluminate solution that is complexed with sodium gluconate. Sodium gluconate is added to the sample to complex aluminium. The sample is titrated to ph 8.10 (inflection midpoint in the absence of aluminium) with standard hydrochloric acid. The amount of acid used represents hydroxide plus one-half carbonate. NaA1(OH)4+Na2CO3+NaOH+nGl+3HC1 Al(OH)3 (3.4) After titration of total soda or hydroxide-bicarbonate the aluminium concentration can be determined. After any of the two titrations, aluminium is present as the gluconate complex and the solution has been titrated to either ph 8.3 or 8.1. Either ph can serve as the starting ph of aluminium titration, provide the same ph is used for the final end point. Potassium fluoride is added to replace OH- from aluminium gluconate complex, three equivalents of hydroxide per equivalent of aluminium are released. Hydrochloride acid is added to the solution to increase the ph back to 8.3 or or 10 ml excess of acid is added and the solution is back titrated by NaOH to ph 8.3 or HC1 + 6KF K3AIF6 + 3KCI + 3H20 + ngl (3.5) Chemicals: (1) NaOH (Ajax Chemicals, Australia, laboratory reagent, greater than 97.5%, maximum limits of sodium carbonate (Na2CO3) impurity 2.5%), (2) hydrochloric acid, HC1 (BDH Chemicals, Australia, 36%), 72

98 (3) sodium gluconate, CH2OH. (CHOH)4.COONa (Aj ax Chemicals, Australia, laboratory chemicals, 97.5%); (4) potassium fluoride anhydrous, KF (BDH Chemicals, Australia, general purpose reagent, 97%); (5) di-sodium tetraborate, Na2B4O7.10H20 (BDH Chemicals Ltd., Poole England, analytical reagent, 99.5%); (6) potassium hydrogen phthalate, KOOC.C6H4.COOH (Ajax Chemicals Ltd. Sydney- Melbourne, analytical reagent, 99.9%); (7) phenolphthalein, (HOC6H4)2.C.C6H4. COO (Aj ax Chemicals Ltd. Sydney-Melbourne, analytical reagent); (8) methyl red (May & Baker Ltd. England) and (9) high purity oxygen-free nitrogen (CIG, greater than 99.99%). Reagents: (1) Sodium gluconate 20% solution, stored in polyethylene bottles, adjusted to phenolphthalein neutrality (ph 8.3). (2) Standard HC1 and NaOH solutions, 0.4 M. Standardisation was performed as described in (3) Potassium fluoride solution, 50% KF.2H2O by weight was filtered, and stored in polyethylene. 25 cm3 of the this solution was sampled to 250 cm3 of water and 50 cm3 of Reagent (1) was added to test if the ph was approximately 8.3 which is also the phenolphthalein end point. If the correct ph was achieved, then the addition of 1 drop of standard HC1 or NaOH would change the ph or the colour of the solution if the indicator was used. (4) Phenolphthalein, 0.1% solution. Procedure: (1) Initial preparation of all samples; 1.0 cm3 of the liquor sample was removed into a 600 cm3 beaker. Several drops of phenolphthalein indicator were added. (2) Titration of total soda (total hydroxide and total carbonate): Standard HC1 (a recorded amount, Reagent (2)) was added to the sample until Al(OH)3 precipitates and is redissolved by The solution was purged with high purity oxygen-free nitrogen for at least five minutes. The nitrogen tip pipette and the side of the beaker were washed down, 50 cm3 of sodium gluconate solution was added and the sample was diluted to about 250 cm3. The solution was titrated to ph 8.3 with standard NaOH. 73

99 Grams of NaOH (as Na2CO3) plus grams of Na2CO3 = (cm3 of HCI - cm3 of NaOH) x normality (equivalent to the molar concentration of or OH-) x (3.6) where is the weight (in grams) of a half millimoles of Na2CO3. (3) Titration of total hydroxide plus one-half carbonate: 50 cm3 of reagent (1) was added to the sample, and the sample was diluted to about 250 cm3. The solution was titrated with standard HCI (Reagent (2)) to ph Grams of NaOH (as Na2CO3) plus 1/2 grams of Na2CO3 = cm3 of HC1 x normality x (3.7) (4) Hydroxide and carbonate calculations: Grams of Na2CO3 2 x (result from (2)-result from (3)) (3.8) Grams of NaOH (as Na2CO3) = result (2) - grams of Na2CO3 (3.9) (5) Titration of aluminium oxide or aluminium: The sample from either (2) or (3) was titrated continuously. 25 cm3 of Reagent (3) was added. The sample was titrated with standard HC1 (Reagent (2)) at the full speed of the burette until it became colourless to phenolphthalein; A 5- to 10- cm3 excess of acid (Reagent (2)) was added. The sample was back-titrated with standard NaOH to the starting ph (8.3 or 8.1). Grams of A1203 = (cm3 of HC1 - cm3 of NaOH) x normality x (3.10) where is the weight (g) of 1/6 millimoles of A12O Crystal Characterisation Techniques X-ray Diffraction All crystals samples were analysed by using X-ray powder diffraction (XRD, Philips PW1 130/90) with a voltage of 30 kv and a current of 10 or 20 ma. X-ray patterns were collected on samples in 0/20 mode using either or CoKa radiation = A and = A respectively). 20 was varied between 5 and 90. The step size was 0.02 degrees/step. The scan rate was 1 min1. A search/match computer program (UPDSM) was installed. The program chooses the five most intense diffragtion peaks and matches them with approximately 70 thousand patterns stored in a data base. The closest matches are displayed. A problem with this procedure is that if the sample is a new phase or a phase whose unit cell is slightly variable, it is unlikely that search/match procedures will give a meaningful answer. If the sample was a 74

100 mixture of several well known minerals the programme can pick them up and display them according to priority. Unit cells were initially estimated manually from the diffraction peak positions and then refined using the computer program The calculations carried out by this program are based on those of Evans, Appleman and D-spacings or 20 degrees of diffraction peaks are used as input data to match those predicted by the programme with a chosen unit cell and space group. The unit cell is then adjusted by the programme to give the least standard error Fourier Transform Infrared Samples were vacuum dried over night in a vacuum desiccator. The samples were then mixed with KBr (Merck, Germany, for spectroscopy) using a mortar and pestle and pressed into a disk by a hydraulic press at a pressure of 10 tons m2. The disk consisted of less than 2% of the crystalline samples (by weight). Fourier transform infrared analysis (FTIR) was carried out in transmission geometry on a Bio-Rad Digilab Division FTS-65 with gratings covering 4000 to 400 cm4, resolution 4 cm-1 and 256 scans Scanning Electron Microscopy and Energy Dispersive Spectroscopy High resolution and field emission scanning electron microscopy (SEM) and Energy Dispersive Spectroscopy (EDS) were performed at 15 kv voltage on a Cam Scan FE. A few crystals of washed and dried sample were mounted on an aluminium stub using double-sided, adhesive tape. The sample was then coated with a thin layer of carbon using a vacuum evaporator X-ray Photoelectron Spectroscopy Samples photoelectron spectroscopy (XPS) were dried in a vacuum furnace at 65 C for 20 hours. Samples were transferred on a double-sided adhesive tape to an ultra vacuum (UHV) experimental chamber of a Perkin-Elmer PHI 5600 ESCA System equipped with a magnesium (hv = hv) X-ray source. The anodes were operated at a power dissipation of 300 W (15 kv, 20 ma). Binding of the elements involved in this are listed in Table 3.2. Pass energy was chosen to be ev for the survey of all the elements in the samples or to be ev for the survey of each individual element. Concentration of each element is calculated by formulae [2361: 75

101 Ii Ci= (3.11) where is the experimentally measured intensity of 'i' element, Si is the sensitivity factor for element "i' and is automatically calculated by soft-ware stored in the XPS equipment, here represents normalised atomic concentration of element 'i". Table 3.2 Binding energies of the elements in sodium Element] Binding energy (ev) (graphite) (Na2CO3) (Na zeolite) (KOH) (Si02 gel) (A1(OH)3) (Na2CO3) (Na zeolite) 73.4 (Na zeolite) 74.6 (A1203) (Na zeolite) (silicates) Particle Sizer The specific range of measurement of the Malvern 2600C particle sizer used was from 0.1 jim to 90 jim. The measurement was carried out in water. Sonication time for each sample was chosen to be 3 minutes in order to break aggregates formed during sample drying into the original, primary particles. The use of sonication was shown not to cause gross breakage to the primary particles. The number frequency density function is a function of the particle number can be calculated from mass (or volume) density: F(L)= ML1 (3.12) M andlj represent mass of the particles with average size of total mass of the particles and average size in 'i" section respectively). The calculated results of the number frequency density function of the crystal products from seeded crystallisation experiments by sizer experimental results are shown in Appendix C. 76

102 CHAPTER 4. CHARACTERISATION OF SODIUM ALUMINOSILICATE NUCLEATED BETWEEN C 4.1. Introduction A wide range of crystalline aluminosilicate products have been reported to form in Bayer plant heat exchangers. In our studies of plant scales the two main phases, cancrinitj2381 were observed. Other phases, including zeolite and have also been reported but were not observed during this study. and phases, Experiments have been carried out to investigate the factors affecting the formation of sodalite and cancrinite including, temperature, sodium hydroxide, silica, alumina and carbonate impurity concentrations. The influence of these variables needs to be examined prior to the definition of a crystallisation mechanism Sodium Aluminosilicate Synthesis between C Fourteen experiments were performed between 60 and 200 C. The experiments were carried out in Parr bombs, an autoclave or an atmospheric stainless steel vessel with stirring speeds of 0, 400 and 600 rpm respectively. The processes of solution preparation, sampling and crystal treatment were as described in Chapter 3. The reaction conditions and the products obtained from the solution containing M Na2CO3 are listed in Table 4.1 and those from solutions containing 0.38 M Na2CO3 are in Table 4.2. An agitation rate of 600 rpm was used in the investigation, unless stated otherwise for the results given in Tables 4.1 and Effect of Temperature The effect of temperature is given by results NC7O-NC100 in Table 4.1 and C70-C 100 in Table 4.2. number in parentheses in the unit cell column represents the standard deviation of the last digit, e.g (2), means the cell size is ± A. Sodalite1 and sodalite2 in Table 4.1 are both cubic with the former having a larger unit cell (possibly due to larger anion substitution, discussed in Chapter 5). Sodalite1 and sodalite2 may be called nosean and hydroxy sodalite respectively in literature. However this terminology is this context is based on unit cell size and does not indicate a distinct crystallographic phase. Sodalite1 and sodalite2 are relative definitions in the presence of two types of cubic structure with the former has a 'larger (comparatively) unit cell. If there was only one type of cubic crystal as long as its unit cell is between 8.5 to 9.9 A it will be called sodalitei. The XRD patterns of the zeolite and sodalite phases obtained 77

103 . from solutions containing 0.38 M Na2CO3 are shown in Figure 4.1. XRD analysis shows that these products are substantially crystalline, with no evidence of an amorphous phase. Patterns a, b and c in Figure 4.1 are due to zeolite and Pattern d to h are due to Table 4.1 The experimental conditions and the products of sodium aluminosilicate synthesis in liquors containing M Na2CO3. Exp Name T NaOH Al(OH)3 Na2SiO3 Time Products** Unit cell ( C) (M) (M).5H20 (days) a (A) (M) NC6O* Zeolite (2) NC7O Zeolite (4) NC8O Zeolite (3) NC9O Sodalite (2) NC Sodalitej 8.980(1) NC Sod1 + Sod (4) 8.906(1) NC /6 Sod1 + Sod (3) (4) * M Na2SO4.10H20 added. ** Sodj and Sod2 represent sodalitei and sodalite2 respectively. 0 at 0 rpm agitation. 00 at 400 rpm agitation Table 4.2 The experimental conditions and the products of sodium aluminosilicate synthesis in liquors containing 0.38 M Na2CO3. Exp Name T NaOH A1(OH)3 Na2SiO3 Time ( C) (M) (M).5H20 (days) (M) Products Unit cell a C Zeolite (6) C Zeolite (4) C (1) C Sodalitej 8.961(1) C Sodalitej 8.974(2) C / (1) C /4 Sodalitei 9.025(1) 0 at 0 rpm agitation. 00 at 400 rpm agitation * (A) 78

104 I Figure 4.1 XRD patterns of nucleated sodium aluminosilicate at (a) 60 C in a liquor containing M Na2SO4.10H20, (b) 70, (c) 80, (d) 90, (e) 100, (f) 150, (g) 160 and (h) 200 C all in liquors containing 0.38 M Na2CO3. It can be seen from Tables 4.1, 4.2 and Figure 4.1 that the minimum temperature for sodalite crystallisation was 90 C. Zeolite was observed to nucleate between 60 and 80 C. Breuer et al.1191 also found zeolite A to be the main product at 70 C, with Al(OH)3/NaOH = 0.31, NaOH = 4.75 M and Si02 initial = M. Zeolite A was found to be 79

105 metastable, converting to hydroxy sodalite over many hours. Barrer et results showed that zeolite appeared to form naturally at temperatures as low as 60 C. These confirm the reproducibility of zeolite formation between 60 and 80 C as shown by the results of the present work. The results of Breuer et al)"91, Yuhas et al)'251 and Cardile et al)'1171 showed that sodalite/hydroxy sodalite was formed in spent Bayer liquors at temperatures ranging from 90 to 250 C within 5 hours. The observation of initial formation of sodalite in the present work in this temperature range is strongly supported by these literature reports. FTIR spectra of the products crystallised from solutions containing 0.38 M C032 are given in Figure 4.2. The characteristic bands for nosean at 690, 622 and 560 (refer to Table 2.4) are observed for the products obtained between 90 and 160 C (c to f). The identification of nosean (possibly an intermediate between sodalite and cancrinite) is based on the FTIR only and there fore is not in agreement with the identification of this phase as sodalitel in Table 4.2 which is based on d-spacing unit cell analysis. The symmetric stretching peak of Al-O-Si of sodalite at 737 cm-1 is absent in all in Figure 4.2 except for the product obtained at 200 C (g). The other sodalite finger print peaks at 712 and 665 cm-1 are also for the sample obtained at 200 C. Zeolite, obtained at 70 and 80 C (Figure 4.2a and b), did not have the same finger print absorptions as for either sodalite or nosean. It is not easy to distinguish between nosean and sodalite by their XRD patterns. They are both cubic and have the same layer packing, ABC ABC but differ in the crystal symmetry of their cage substitutions [80] In some references the only difference is reported to be in their unit cells: a= 9.09 A for nosean and a= 8.90 A for sodalite. The products of Figure 4.2c to f are called sodalitei in Table 4.2 because their unit cell sizes were relatively large compared to those of sodalite2 listed in Table 4.1. Although nosean has similar FTIR finger print peaks as those of cancrinite it is not reported as having the same asymmetric Al-O-Si peaks as those of cancrinite (three peaks at about 1000 Nosean may be the transition state from sodalite to cancrinite. By comparing their Si-O-Al symmetric stretch from 737, 712 and 655 cm' for sodalite to 690, 622 and 560 for nosean, it may be suggested that the Si-O-Al bond in sodalite is stronger than that in nosean by having higher As we defined earlier, any cubic crystalline structure will be defined as sodalite in later chapters to simplify the discussion. Figure 4.3 shows the SEM micrographs indicating the morphology of the crystalline zeolite and sodalite products formed at 70, 80, 90 and 100 C from the solutions with M Na2CO3. The scales of these figures are not the same among the samples, however, the morphology of the crystals can be seen clearly. Zeolite obtained at 70 C (Figure 4.3a) shows a morphology of brick-like form. The product at 80 C appears as a 80

106 mixture of rod-like and brick-like crystals (Figure 4.3b). The morphology of the crystalline products formed at 90 and 100 C (Figure 4.3c and d) were rod-like crystals, agglomerated to form a "knitting-ball" The "knitting-balls" existed mostly as separate particles but also clustered to some degree. It has been previously reported that at C, sodalite precipitated as agglomerated rod-like a) ' p Cu I.5 Q W a v e n u m b e r s Figure 4.2 FTIR spectra of the sodium aluminosilicate nucleated from solution for the 0.38 M Na2CO3 solutions at (a) 70, and (b) 80 (zeolite), and at (c) 90, (d) 100, (e)150, (f) 160 and (g) 200 C (sodalite). It is clear that temperature is a major factor controlling zeolite, hydroxy sodalite and cancrinite phase formation The temperature of spent plant liquor in heat exchangers varies from 105 to 205 It can be seen from Tables 4.1 and 4.2 that nucleation of sodalite occurs between 90 to 200 C and that slight variation of species concentration does not affect the phase formed. Cancrinite nuclei have not been observed under these conditions. The unit cell size determination from the sodalite syntheses showed that generally larger cell sizes resulted from higher temperatures. Zeolite crystallises at 81

107 (a) (b) (c) (d) Figure 4.3 SEM micrographs of zeolite nucleated at (a) 70 and (b) 80 C, and sodalite nucleated at (c) 90 and (d) 100 C from liquors containing 0.1 M Si02 and M Na2CO3. 82

108 temperatures below 90 C, thus, it does not appear that zeolite nucleates in Bayer plant heat exchangers. Conclusion: Temperature does affect the phases observed in similar solutions after 1 day. Between C zeolite is the crystalline product. The change of temperature from 80 to 90 C appears to affect the phase crystallised from zeolite to sodalite. Between 90 to 200 C sodalites form. Cancrinite has not been found to nucleate directly from solution in this temperature range Effect of Sodium Carbonate Concentration Between C, low (0.043 M) and high (0.38 M) C032 concentrations gave sodium aluminosilicate products: zeolite (70-80 C), and ( C) (refer Table 4.2). Figure 4.4 shows the XRD patterns of the crystal products obtained from the solutions with and 0.38 M Na2CO3 at 80 and 90 C. Figure 4.4 indicates that C032 level had no effect on the phase crystallised. The products obtained from low and high carbonate solutions at the same temperature (e.g. Figure 4.4a and b, and Figure 4.4c and d) were similar. XRF results gave a Na: Al: Si ratio of 6.6 : 6.0: 5.5 for the product from M Na2CO3 solution and 6.4 : 6.0: 5.1 for the product from 0.38 M Na2CO3 solution at 90 oc. Figure 4.5 shows the comparison of FTIR spectra of the products obtained from and 0.38 M Na2CO3 solutions at 90 C. It is of interest to look at the peak at 1430 cm1. There are three possible explanations for the presence of this peak: (1) It is a C032 peak and belongs to the V3 10] (2) It is a peak due to adsorbed ammonia The NH stretching mode occurs at about cm1 and the NH deformation mode at about % ammonia solution was used to wash these crystal products. As FTIR is a bulk analysis technique, NH4+ surface concentration would not be expected to give rise to such a strong absorption. (3) It is an absorption band for the deformation of hydroxyl Minerals which contain loosely adsorbed or absorbed water molecules give rise to strong absorption bands at 1630 cm1 (OH bending The most probable explanation is that this band is due to C032 as this species was present in the mother liquors. The proposition of hydroxyl deformation for this wave number is not commonly accepted. The concentration of C032 in the crystals obtained at 90 C from a low Na2CO3 solution (Figure 4.5 a) is higher than that in crystals obtained from a higher Na2CO3 solution (Figure 4.5b). This trend is abnormal and the reason for this is unclear. This is also the case for the samples obtained at 100 C and is very reproducible at low temperatures ( C). 83

109 9000 Figure 4.4 XRD patterns of nucleated products obtained at 80 C from a liquor containing (a) 0.38 M and (b) M Na2CO3 and at 90 C from a liquor containing (c) 0.38 M and (d) M Na2CO3. Figure 4.6 shows XPS binding energy (BE) of sodalite obtained at 90 C from (a) a liquor containing M Na2CO3 and (b) a liquor containing 0.38 M Na2CO3. The C032 peak occurred at ev and, those of charged and uncharged adventitious SO

110 (1430) H ci) o C -Q Wavenumbers 500 Figure 4.5 FTIR spectra of the sodium aluminosilicate obtained at 90 C from the solution containing (a) and (b) 0.38 M Na2CO3. hydrocarbon at ev and ev BE respectively. The percentage ofc032 in each sample shown in Table 4.3 was obtained by curve fitting C032 concentration over total carbon concentration. Thus there was C032 in the products from the liquors containing either or 0.38 M Na2CO3, with the amount of CU32- being similar for both samples. This is consistent with the FTIR result if the peak at about 1450 cm' is assigned to C032. After ion beam etching to about 2.5 nm deep (Figure 4.6c and d), the C032 intensity remained similar to that before etching (Figure 4.6a and b), and the hydrocarbon percentage decreased. The C032 in the crystals, therefore, was not adsorbed directly from CO2 in the after crystallisation but from the solution during the crystallisation. As the experiments were carried out in an atmospheric vessel which was not completely air-tight, the absorption of CO2 from the air by the highly caustic Bayer liquor should be taken into account. The difference the products obtained in low (0.043 M) and high (0.38 M) Na2CO3 solution at 150 C and above (in Tables 4.1 and 4.2) was that the former consists of both sodalitei and sodalite2, and the latter consists of only sodalitej. Sodalitei and sodalite2 have been differentiated in Tables 4.1 and 4.2 on the basis of unit cell size. The unit cell of the sodalitei listed in Tables 4. 1 and 4.2 is larger than that of pure hydroxy 85

111 sodalite (between 8.77 and 8.90 and smaller than that of pure carbonate sodalite (9.03 This intermediate cell size may be due to the presence of both OH- and CO32 in the structure. II II 2 II I' II Figure 4.6 XPS results of carbon is binding energy of the products obtained at 90 AC before ion etching (a) from a liquor containing M Na2CO3 and (b) from a liquor containing 0.38 M Na2CO3, and after ion etching to 2.5 nm from the surface (c) from a liquor containing M Na2CO3 and (d) from a liquor containing 0.38 M Na2CO3. The experiments at 150 C and above showed that two phases of sodalite formed from liquors containing M Na2CO3: sodalitej, possibly a C032 type of sodalite (larger cell) and possibly an OH- sodalite (smaller cell). Sodalite2 only appears to nucleate directly in solutions in which the C032 concentration is limited, i.e. where G032 concentration is high sodalite2 nucleation is negligible or absent. However, sodalite2 does form gradually in high C032 content by hydrothermal transformation from sodalitei subsequently followed by transformation to cancrinite at 100 C. This transformation will be discussed in Chapter 5. Both transformations occur in synthetic and plant 86

112 Table 4.3 ion etching The percentage of C032- in samples shown in Figure 4.6 before and after Sample Total carbon (%) C032/total carbon (%) C032 in sample (%) NC9O before etching NC9O after etching C90 before etching C90 after etching XRD analysis showed that the products nucleated at 160 C were sodalites. The liquor containing M Na2CO3 produced both sodalite1 (cubic with a larger unit cell than sodalite2) and sodalite2 (Figure 4.7a) indicated by double peaks at 310, 222, 300, 510, 400 and 600 etc. The doublet peak at larger 20 in each case represents a lower "d" spacing of the crystal and hence a smaller unit cell. The crystalline product obtained from a liquor containing 0.38 M Na2CO3 (Figure 4.7b) was sodalitei (a = A) with cancrinite indicated by a broad (101) peak at (d = 4.7 A) Sazhin and studied a sodalite-type product obtained at 280 C after 6 hours of synthesis. Their XRD analysis indicated that the product had the diffraction pattern of sodalite, but had a hexagonal appearance. The FTIR spectra of the crystals precipitated from the and 0.38 M Na2CO3 solutions at 160 C are shown in Figure 4.8. Interference fringes in the 3000 to 1500 cm1 range are apparent. The crystals precipitated from the M Na2CO3 solution (Figure 4.8a, XRD pattern Figure 4.7a) were relatively pure sodalite with bands at 737, 713 and 668 cm' and those precipitated from the 0.38 M Na2CO3 solution (Figure 4.8b, XRD pattern Figure 4.7b indicating sodalite) showed mainly sodalite and a small amount of nosean (also cubic) by also having characteristic peaks at 690, 630 and 560 cm1[99]. The latter finger prints are also indicative of cancrinite if they are better It has been suggested that nosean may be a phase that links the transformation of cubic sodalite to hexagonal The above results indicate that the presence of Na2CO3 may promote the formation of cancrinite under the experimental conditions. The absorbance due to C032 at 1450 cm1 indicates that the concentration of C032 in the crystalline phase obtained at 160 C from lower C032 solution was significantly less than that from the higher C032 solution, as expected but the opposite trend to the case of the samples obtained at 90 (Figure 4.5) and 1QO C. Another difference between the scale from low and high C032 concentration solutions was that the crystals from low C032 solution were harder, larger in size, easier to filter and were attached more strongly to the wall of the vessel. Figure 4.9 gives the SEM 87

113 ' 'c iJ a Figure 4.7 XRD patterns of nucleated sodium aluminosilicate obtained at 160 C from a liquor containing (a) M + sodalite2) and (b) 0.38 M Na2CO3 (sodalitei) morphology of scale at 160 C from low C032 solution (Table 4.1) and high C032 solution (Table 4.2). It can be seen that the crystals and the agglomerates of the M Na2CO3 solution at 160 C (Figure 4.9a and b) were larger than those of the 0.38 M Na2CO3 solution (Figure 4.9c and d). Single crystals of both scales were hexagonal 88

114 Al -0- Si 0.80 ci) 0.75 U 0.70 o 0.65 < Waven urn bers Figure 4.8 FTIR spectra of nucleated sodium aluminosilicate at 160 C from solutions with Si M and Na2CO3 (a) and (b) 0.38 M. plates, with the ones from the M Na2CO3 solution 4.5 and the ones from the 0.38 M Na2CO3 solution pm in in and suggested that sodalite seed might increase in average size with very low Si02 supersaturation (mainly growth) and decrease in average size with very high supersaturations (simultaneous nucleation) in multiple-cycle batch experiments. With the same initial Si02 concentration, Si02 supersaturation in M Na2CO3 solution is lower than that in 0.38 M Na2CO3 solution due to a higher solubility in the former (see Chapter 6). Thus the trend of particle size is similar to that observed by Cousineau et al. The two products that were obtained at 150 C from the solutions containing M Na2CO3 (Table 4.1) and 0.38 M Na2CO3 (Table 4.2) were sodalites, very similar to Vaughan products (Table 2.13) in the presence of NaOH and Na2CO3. In both cases sodalite product after 3 days of reaction. Sodalite was shown to be the product at 200 C after 6.5 hours (Figure 4.1). This indicates that sodalite precipitates at 90 C and also at higher temperatures. Conclusion: C032 concentration does affect the character of the phaseformed above 150 In high C032 solutions sodalitej, with a large unit cell and high crystalline C032 C. pm 89

115 (a) (b) (c) (d) Figure 4.9 SEM micrographs of sodium aluminosilicate nucleated at 160 C from a liquor containing 0.02 M Si02 and M Na2CO3 (a) high and (b) low magnification and a liquor containing 0.02 M Si02 and 0.38 M Na2C 03 (c) high and (d) low magnification. 90

116 content is the product. In low C032 solutions sodalitej and sodalite2, the latter having a small unit cell and low crystalline GO32- content, are the products Effect of Silica Supersaturation Experiments were carried out using a solution with a 0.01 M Si02 concentration, which is similar to that of plant liquor, with M Na2CO3 at 95 C. No obvious scale or detectable change in solution Si02 concentration was obtained after one week of reaction. Even when the Si02 concentration was raised to 0.04 M at 95 C, the solution appeared optically clear. With the addition of Na2CO3 or Na2SO4, some crystals were obtained at 95 C but not even enough to be characterised by XRD. Further experiments were carried out with solutions containing 0.38 M Na2CO3. In these cases sufficient scale for characterisation was precipitated. Figure 4.10 shows the comparison of the XRD patterns of the scale at 95 C after 5 days in low (0.01 M) Si02 liquor, (Figure 4.lOa), and the scale obtained at 100 C in high (0.1 M) Si02 liquor after overnight crystallisation, where "c' represents sodium carbonate peak, (Figure 4. loa), with Na2CO3 concentration of 0.38 M in both cases. The XRD patterns appear very similar, however the crystalline product arising from the low Si02 solution has an unidentified diffraction peak at approximately 16 20, to the right of sodalite (110). The reason for the appearance of this peak is unknown. The unit cell derived for the low Si02 crystallisation product is A and from the high Si02 crystallisation is A. The FTIR of the 0.01 and 0.1 M Si02 crystallisation products was similar. High C032 concentrations promote the phase transformation of sodalite to cancrinite as indicated by the (possible) intermediate phase of nosean being identified by FTIR for the 0.38 M C032 solution in Figure 4.8 but not for the M C032 solutions. It appears from the literature and the results presented here that the effect of increased C032 is the enhanced kinetics of transformation between sodalite and cancrinite and not altered Si02 supersaturation. Therefore the effect on additional supersaturation, provided by increased Si02 concentration is not expected to be the same as on the addition of C032. Increased Si02 concentration does not result in a phase change. Conclusion: Si02 concentration within the limits investigated, does not affect the crystalline phase precipitated Conclusion (1) This investigation has demonstrated that sodalite is the first phase to crystallise between 90 and 200 C in synthetic Bayer spent liquor. Zeolite nucleates between 60 and 80 C. The formation of sodalite was not observed to be preceded by the precipitation of amorphous aluminosilicate and/or zeolite phases. 91

117 I b C Ia JL4LkU a SO Figure 4.10 XRD patterns of sodium aluminosilicate obtained from (a) 0.01 M Si02 and (b) 0.1 M Si02 liquors containing 0.38 M Na2CO3 at 95 and 100 C respectively. (2) Two phases df sodalite, a liquor containing M Na2CO3 with an initial Si02 concentration of 0.02 M at 160 C. Sodalitei forms in a liquor containing 0.38 M Na2CO3 under similar conditions. Sodalitet has a larger unit cell than sodalite2 and also contains more C032. High C032 concentrations seem to favour the formation of sodalite i. 92

118 (3) A certain degree of supersaturation is needed for the nucleation and crystallisation of sodium aluminosilicate. The Si02 concentration required for nucleation at 95 C is higher than 0.04 M and at 160 C higher than 0.02 M, if the liquor Na2CO3 concentration is M. No visually detectable crystallisation occurred in a liquor of 0.01 M Si02 containing M Na2CO3 at 95 C within 5 days. The extent of crystallisation was greater in a liquor containing 0.01 M Si02 with 0.38 M Na2CO3 at 95 C after 5 days. 93

119 CHAPTER 5. INVESTIGATION OF SODIUM ALUMINOSILICATE PHASE TRANSFORMATION 5.1. Introduction The results of the previous chapter showed that between 90 and 200 C sodalite was the phase that formed initially in synthetic spent Bayer liquor. However plant scale is mainly Aging of a liquor (NaOH 4.52 M, Na2CO M, Al(OH) M and 0.1 M Si02) showed that after 3 days sodalite was still the crystalline product. This chapter describes the investigation of the stability of sodalite and the transformation of sodalite to cancrinite Phase Changes of Nucleated Sodium Aluminosilicate Effect of Aging In order to investigate aluminosilicate formation as a function of time, aluminosilicate precipitation experiments were run at 100 C for 15 days. The synthetic spent liquor contained NaOH 4.52 M, Na2CO M, Al(OH) M (similar to that of industrial spent liquors) and Si M. The reason that 0.1 M Si02 was chosen was given in Section The experimental and sampling procedure is described in Section X-ray fluorescence analysis of the crystalline phase at 0.5 hours and hours gave Na: Al : Si ratios of 6.9 : 6.0 : 4.9 and 9.4 : 6.0 : 4.8 respectively. Ideally this ratio is 8 : 6 : 6 for both sodalite and cancrinite. The trend of increasing NalAl with time is in agreement with the result of Kraus, Derevyankin and where the ratio of NaOH/Al(OH)3 in the crystalline phase also increased with reaction time. XRD patterns of the crystalline products are shown in Figure 5.1. The crystalline phase of sodalite did not appear to change. However, unit cell analysis of the diffraction peak positions of the phases precipitated after 0.5 and hours showed that a change in unit cell size had occurred. The phase in both cases was sodalite (cubic). After 0.5 hours the unit cell was calculated as a= ± A and after hours as a= ± A. Hence the unit cell decreased in size by a significant margin. By our previous definition, the larger unit cell is due to sodalitei and the smaller unit cell due to sodalite2. Evidence of this decrease in unit cell size can be seen in Figure 5.1. There is a double diffraction intensity maximum (Figure 5.1) for the diffraction peaks 222, 330, 510, 440 and 611 in the diffraction patterns of samples from 0.5 to hours. Double diffraction intensities are in fact present for all diffraction peaks but in some cases cannot be clearly 94

120 identified as either the peaks are too small (i.e. high 20) or the doublet was too close together (i.e. low 20). With increasing time the diffraction peaks due to sodalitei decreased in intensity while those for sodalite2 increased H z 0 U THETA Figure 5.1 XRD patterns of sodium aluminosilicate products from a synthetic spent liquor containing NaOH 4.52 M, Na2CO M, Al(OH) M and 0.1 M Si02 at 100 oc. After hours (15 days) of experiment, a thick layer of scale was found settled at the bottom of the vessel. A much smaller amount of sodium aluminosilicate was attached to the wall of the vessel. Figure 5.2 shows (a) the vessel without scale and (b) the different 95

121 kinds of scale formed: the scale on the wall which appears trigonal was characterised as sodalite2 and that at the bottom of the vessel as Na2CO3. The comparison of the XRD patterns of these two scales are shown in Figure 5.3: (a) sodalite2 attached to the wall of the vessel and (b) Na2CO3 settled at the bottom. The formation of Na2CO3 scale appears to be a slow process as this was not observed in the experiments carried out overnight at the same temperature. SEM photographs of the scale obtained from the wall and the bottom of the vessel are shown in Figure 5.4. Figure 5.4a is an SEM micrograph of sodalite2 obtained after 360 hours (15 days) and Figure 5.4b is that of Na2CO3 from the bottom of the vessel. As this experiment was carried out in a reactor which was not sealed, evaporation of H20 was significant; thus the concentration of all the components increased during the 15 days of reaction time. If, with time, the Na2CO3 concentration exceeded its metastable 242] through the coupled effects of C032- diffusion from the crystal lattice channels back into solution and CO2 absorption from the air, it is likely that Na2CO3 crystals would be observed in solution. The FTIR spectra of the crystalline products are shown in Figure 5.5. During the initial stages of precipitation, C032 species dominated the infrared absorption spectra (1450 cm I (vs), 876 (v2), 712 With time the C032 peaks became relatively weaker; and the main band for Si-O-Si(Al) (aluminosilicate frame structure) moved from 1017 (0.5 hours) to 985 cm1 (359.5 hours). Sodalite characteristic peaks at 739, 711, 669 cm1 were gradually substituted by more nosean-cancrinite-like peaks at 692, 623 and 562 cm 1 CO32 peaks are always present in JR spectra of aluminosilicates, regardless of the amount of Na2CO3 added ; 126; 151] A broad hydrogen bonded hydroxyl stretching peak at about became more intense with The characteristic peak at 1650 cm_i for the water bending vibration was not observed. This implies an increasing concentration of bonded OH- in the crystal structure rather than H20. These observations indicate a decreasing concentration of C032 with time in the crystalline phase and increasing OH-. The diffusion of a large anion such as CO32 out of channels in The crystal structure followed by replacement by smaller OH- groups may be responsible for the change in unit cell size of the crystalline products with time. Meier and found that embedding a OH- instead of a Cl- yielded hydroxy sodalite without changing the crystallographic space group. The trend is the same for the replacement of C032 by OH-. A morphological modification of the crystals is observed from SEM photomicrographs (Figure 5.6 a and b). Figure 5.6a shows the morphology of the sample taken after 3 hours of reaction time, the crystals are spheroidal, relatively smooth and with blocky crystals dimensions from surface. After hours, the structure has been destroyed 96

122 (a) (b) Figure 5.2 Photographs of (a) the stainless steel vessel and (b) the scales in the vessel after 360 hours of reaction time at 100 C: sodalite2 (scale with trigonal morphology on the wall) and Na2CO3 (white layer scale at the bottom)

123 Figure 5.3 XRD patterns of the solid products obtained at 100 C after 360 hours from (a) the wall (sodalite) and (b) the bottom of the reactor vessel (Na2CO3). by framework shrifikage and the crystals are protruding, elongated aggregates with rough surfaces containing projections of many particles (Figure 5.6b). Wehrli and Aguilas synthesis of sodium aluminosilicate in synthetic Bayer liquor at 100 C went from an amorphous phase (after 5 and 15 minutes) to a zeolite (after 60 minutes) 98

124 (a) (b) Figure 5.4 SEM micrographs of (a) sodalite2 on the wall and (b) Na2CO3 from the bottom of the vessel

125 1450 Si-O-AI(Si) 1000 b S 0 r b n C e Figure 5.5 FTIR spectra of sodium aluminosilicate products from 0.38 M Na2CO3 synthetic spent liquor at 100 C after (a) 0.5 h, (b) 22 h, (c) 64 h, (d) 118 h, (e) h, (t) 213 h, (g) h and (h) h. which then slowly transformed to a sodalite after about 3 to 4 This phenomenon was not observed in our experiments. Our 15 day experiments were repeated 3 times. It is surprising that they observed zeolite after 60 minutes of reaction at 100 C while we obtained only sodalite at the same temperature after 0.5 hours of reaction time. In zeolite preparation, high SiO2 concentrations, low caustic and low temperatures may promote the formation of an initial gel and then amorphous But in Bayer processing the conditions are the opposite, i.e. low Si02 and high caustic and high temperatures, and thus no amorphous products were observed. The rate that Wehrli and added to 1.5 dm3 of spent liquor was mol min1(in 1.68 cm3 of volume) and the stirring speed was 300 rpm (half as much as used here). After 60 minutes, moles of Si02 was added and the SiO2 concentration should be M, higher than that used in the present work. 100

126 (a) (b) Figure 5.6 SEM micrographs of sodium aluminosilicate obtained at 100 C in a liquor containing 0.1 M Si02 and 0.38 N Na2CO3 after (a) 3 and (b) hours

127 It is unclear from the literature whether the phase change from sodalite to cancrinite would occur at low temperatures if the reaction time is long enough. To further investigate the effect of aging, an experiment was carried out in a Parr bomb for 30 days under conditions similar to the one run for 15 days discussed at the beginning of this section at 100 C. The XRD pattern (Figure 5.7a) of the crystalline product indicated that (a = ± 0.001) plus sodium hydrogen carbonate hydrate (Na3HCO3.C03.2H20, suggested by search match programme) and a small amount of cancrinite were the products. In Figure 5.7, the peaks topped with letter 'c' belong to sodium hydrogen carbonate hydrate and "can" for cancrinite. This indicated that cancrinite would form at 100 C but would not be the main phase even after 30 days. Figure 5.7b and c show XRD patterns of sodium aluminosilicate plus sodium carbonate obtained at 90 C in Parr Bombs for 3 days and will be discussed in Section The amount of Na2CO3 hydrate is much smaller after 3 days of precipitation at 90 C than that after 30 days at 100 C. Thus nucleated sodium aluminosilicate (sodalite 1) will experience cell shrinkage and leaching of large C032 anions to form another sodalite (sodalite2) accompanying the precipitation of Na2CO3 or Na3HCO3.C03.2H20 at 100 C during long term experiments. Carbonate sodalite may be easier to nucleate but thermodynamically less stable than hydroxy sodalite. The experiment, run at 200 C in a liquor containing 0.02 M Si02, 4.52 M NaOH, 1.65 M Al(OH)3 and 0.38 M Na2CO3 (Chapter 4), showed that sodalite was the crystalline product after 6.5 hours (Figure 5.8a). A parallel experiment was carried out in a Parr Bomb at the same temperature but for a longer time of 10 days. An XRD pattern of the product was indexed as cancrinite (Figure 5.8b). The sodalite to cancrinite phase transformation was promoted by high temperature and longer reaction time. The phase transformation was more complete after 10 days at 200 C than 15 days at 100 C. Similarly, Avdeeva and obtained a sodalite-type product at 220 C after 3 hours, and a cancrinite product at 220 C after 36 hours. This indicates that the higher the temperature the shorter the time required to complete the transformation. Conclusion: The length of in situ aging at 90 and 100 C does have an effect on the crystalline phase present. In solutions containing a high C032 concentration, the crystalline phases and their characteristics follow the following trend: Sodalitej sodalite2 cancrinite (5.1) high C032 low C032 (101 diffraction peak) At 200 C in solution containing high C032 concentration, the crystalline phases may experience the following change: 102

128 Figure 5.7 XRD patterns of the solid products (sodalitei + Na2CO3 hydrates represented by peaks marked "c' and cancrinite by 'can") obtained (a) at 100 C after 30 days in a liquor containing 0.38 M Na2CO3; and at 90 C for 3 days in a liquor containing (b) and (c) 0:38 M Na2CO3. Sodalitej cancrinite (5.2) 103

129 Sodalite2 was not observed at 200 C, however no periodic sampling was carried out between 6.5 and 10 days. At this elevated temperature it may be expected that sodalite2 would be observed after 6 hours compared to 15 hours at 100 C (I) Figure 5.8 XRD patterns of sodium aluminosilicate obtained at 200 C after (a) 6.5 hours (sodalitei) and (b) 10 days (cancrinite). 104

130 Effect of Sodium Carbonate at Low Temperature NC9O (90 C, M Na2CO3, Table 4.1) and C90 (90 C, 0.38 M Na2CO3, Table 4.2) experiments were performed in the atmospheric reactor (which allows the exposure of the solution to air). CO2 absorption from the air may have had an effect on the C032 concentration in the crystals. Both of the experiments were repeated in Parr Bombs for 3 days instead of overnight. The increased experimental time was to compensate for the lack of agitation in the Parr Bombs. The difference between experiments in the atmospheric vessel and in Parr Bombs (NC9Oa and C90a), and the unit cell of the products are shown in Table 5.1. In both cases, a mixture of sodalite, sodium carbonate hydrate (Na2CO3.H20) and a small amount of cancrinite were the products as indicated by XRD (Figure 5.7b and c). The relative amount of cancrinite can be judged by the intensity ratios of cancrinite diffraction peaks to sodalite diffraction peaks (Table 5.1). The formation of cancrinite was more pronounced after 3 days at 90 C in a low carbonate solution than after 30 days at 100 C in a high carbonate solution (Figure 5.7a and b); and was more pronounced in 3 days reaction time in Parr Bombs (NC9Oa and C90a in Table 5.1) than in the atmospheric reactor after 1 day (NC9O and C90 in Table 5.1). Table 5.1 Comparison of the XRD peak intensities and unit cell size of the sodalite obtained from low and high C032 solutions. Samples Time (day) Stirring (rpm) T ( C) Na2CO3 (M) lb/ia Tb/Ic Unit cell size** a (A) NC9O ± C ± NC ± ClOD na na ± NC9Oa ± C90a ±0.002 * Ia, Tb and Ic are the intensities of the peaks at about (a) 14 (d= 6.4 A, sod(110) + can (110)), (b) 19 (d= 4.7 A, can(101)) and (c) 24 (d= 3.7 A, sod(211) + can(300)) respectively, 'na" means not applicable as no sharp peak of can(10l) was observed. * * Unit cell size of the main phase sodalite i. The three XRD patterns in Figure 5.7 appear similar i.e. sodalite is the main product although the Na2CO3 concentration used in Figure 5.7b was much lower than that of the other two experiments. The transformation of sodalite to cancrinite is very slow under 100 C. This agrees with Ni et and Hermeler et reports that the presence of Na2CO3 does not promote cancrinite formation at 100 C. Barrer et aij89' reported that 105

131 the addition of Na2CO3 did not result in marked alteration of the usual products, but extended the conditions of formation of cancrinite after 3 to 4 days to below 150 0C[243] The intensity of the Na2CO3.H20 peaks was the smallest in Figure 5.7b where Na2CO3 concentration was the lowest in solution and thus the least Na2CO3 "saturated'. shows that Na2CO3 in solid products was not only due to CO2 absorption from the air. Na2CO3 solubility in sodium aluminate solution with Al(OH)3/NaOH between 0.28 to 0.42 at 125 C was between 1.3 to 1.4 much higher than the Na2CO3 concentrations in our solutions. Na2CO3.H20 is the crystalline phase expected to precipitate at 100 C in NaOH Perhaps its solubility is lower than that of Na2CO3 under these experimental conditions. Na2CO3.H20 was also found in plant scale This Precipitation of Na2CO3 appears common for longer term experiments. The reason for precipitation of Na2CO3 or its hydrates under Bayer liquor condition is not clear yet. Avdeeva et al.'s suggested that "hydroxy sodalite" (no salts added during preparation) had a larger unit cell (9.04 A) than "carbonate sodalite" (8.98 A) (0.29 M Na2CO3 added during preparation). Table 5.1 shows no distinguishable trend in the unit cell sizes of the sodalites obtained at 90 and 100 C. Care should be taken in interpreting the names Avdeeva et at. used for their sodalites. As shown in Figure 4.5 it is not necessarily true that sodalite obtained from a solution containing a lower C032 concentration gives rise to a lower CO32 concentration in its crystals. It appears that carbonate sodalite precipitation was more favoured kinetically than 0H sodalite, but that OH- sodalite was thermodynamically more stable i.e. OH- will replace C032 in sodalite phase. As shown by the transformation from high C032 sodalitej to low CO32 sodalite2. This may give rise to the initial inclusion of GO32- followed by replacement by OH-. Chlorate and perchlorate sodalite are also unstable in caustic solutions and tend to shrink to more stable hydroxy sodalite as shown by Veit, Buhi and As shown by Barrer and a continuous range of compounds is possible between collapsed sodalite (a < 8.87 A) and expanded nosean (cubic, similar to sodalite with a larger unit cell) structures (a> 9.10 A). Thus a cubic unit cell size change can occur from smaller than 8.87 A, between 8.87 to 9.10 A, to larger than 9.10 A. Conclusion: Below 100 C Na2CO3 concentration of the liquors does not thermodynamically affect the crystallised phase (still mainly cubic). Low C03 2 in solution is more effective for cancrinite formation below 100 C (increasing intensity of (101) peak in XRD patterns). 106

132 Effect of Sodium Carbonate at High Temperature It has been reported that almost all plant scale formed at greater than 120 C is However Hermeler, BuhI and Hoffmann1100' obtained sodalite at 157 C after 5 days and cancrinite at 497 C after 2 days. An experiment was carried out in a Parr bomb with a 27 cm3 teflon insert using a solution containing, 0.01 M Si02, 4.52 M NaOH, M Na2CO3 and 1.64 M Al(OH)3 at 150 C for 5 days. No obvious scale, either heterogeneously (i.e. on the vessel walls) or homogeneously nucleated was detected in such a small amount of solution (18 cm3). Hence in unseeded synthetic spent liquor, the silica level of 0.01 M was too low to form aluminosilicate scale in this time period. Therefore the initial silica concentration was chosen to be 0.1 M to achieve homogeneous nucleation. The other conditions were kept unchanged from those described above. Three Parr Bombs containing this solution were placed in a thermostat controlled oven at 150 C and samples were taken at 3, 6 and 10 days. The sample treatment was the same as that described in Section A second set of experiments was carried out at 150 C in Parr Bombs under similar conditions except that the concentration of Na2CO3 was 0.38 M. One bomb was removed and opened after 3, 6 and 10 days. The crystalline products were filtered and washed as described in Section and dried at room temperature. The detailed difference among the experimental conditions and the unit cell size of the products are shown in Table 5.2. Double diffraction peaks can be seen at 211, 310, 222, 330 and 510 in the XRD patterns of the crystals obtained from low Na2CO3 concentration solution (Figure 5.9a to c) indicating that sodalitei and sodalite2 with a small amount of cancrinite were present all the way through the 10 days of experiment. Not much change in the crystalline phases occurred between 3 and 10 days for the products from the solution with M Na2CO3. Na2CO3 crystals did not precipitate at 150 C in a liquor containing M Na2CO3. The XRD pattern taken of the sample from the solution of 0.38 M Na2CO3 after 3 days (Figure 5.9d) indicated the presence of poorly formed cancrinite. The products from the same solution after 6 and 10 days can be indexed as cancrinite (Figure 5.9e and 1). The peak on the left side of cancrinite 101 peak in Figure 5.9f is a Na2CO3.H20. This is in contrast to the products of the M Na2CO3 solution where mainly sodalitei was present. There was no obvious change among the FTIR spectra of both sets of samples from 3 to 10 days. The spectra are similar even in the finger print peak regions. Three cancrinite characteristic peaks near 1000 cm' could not be clearly seen except for the product after 107

133 10 days for the high C032- concentration solution. In this case a shoulder can be seen at 1050 cm-' (Si-O-Al asymmetric stretch of cancrinite In the finger print area cubic nosean and hexagonal cancrinite have the same peaks except that those for the latter are It may be possible that the products of 6 and 10 days from the solution of 0.38 M Na2CO3 were still transforming from sodalite to cancrinite. Table 5.2 The phases and unit cells obtained at 150 C from Parr Bombs. Experiment Name Na2CO3 (M) Time (days) Product* NC 150-3d Sod1 +Sod2 + trace Can NC15O-6d Sod1 +Sod2 + trace Can NC15O-lOd Sod1 +Sod2 + trace Can C150-3d Sod1 + trace Can lb/ia ** Tb/Ic Unit cell size a (A) ai= 8.935(4); a2=8.906(1) (2); 8.913(2) (4); (2) 8.917(1) C150-6d Sod1 +Can (6) c (A) 12.62(2) 5.182(4) (no solution for cubic unit cell) C150-lOd Sod1 + Can (7) 8.977(1) 5.187(5) 5.156(3) *Sodi and Sod2 are sodalites. The former has a larger unit cell size. **The same as in Table 5.1. The number in brackets after the unit cell dimension is the standard deviation of the last digit. The addition of Na2CO3 to synthetic spent liquor did make a large difference to the product phase. It appears to facilitate the phase transformation from sodalite to cancrinite. Formation of sodalite2 at high temperatures only occurs in solution with lower Na2CO3 concentration where the amount of C032 might not be enough for all the components to precipitate only as sodalite j. Conclusion: The effect of C032- on the soda lite to cancrinite phase transformation is not simple, however, the result at above 100 C supports the general conclusion that high concentrations of Na2CO3 promotes cancrinite formation: at high C032 concentration, sodalitej nucleated and transformed to cancrinite gradually; at low C032 concentration sodalitej and sodalite2 formed with a trace of cancrinite. No co-precipitation of Na2CO3 was observed at high temperature from a liquor containing M Na2CO3 after 10 days. 108

134 Figure 5.9 XRD patterns of sodium aluminosilicate obtained at 150 C in a liquor containing M Na2CO3 after (a) 3, (b) 6 and (c) 10 days and in a liquor containing 0.38 M Na2CO3 after (d) 3, (e) 6 and (f) 10 days, where 'can' and 'c' indicate cancrinite and carbonate peaks respectively. 109

135 5.2.4 Effect of Silica Supersaturation A further experiment was carried out for 15 days in the atmospheric reactor at 95 C with the same NaOH and A1(OH)3 concentrations as for the 15 days experiments described in Section with 0.38 M Na2CO3 but in this case with 0.01 M Si02. Very small amounts of crystalline samples were obtained at 5, 10 and 15 days. Evaporation may have aided nucleation as no scale formed under similar conditions in Parr Bombs. The XRD results (Figure 5.lOa to c) showed that there was a double peak at 110, which is not observed in the product obtained from 0.1 M Si02 solution (Figure 5.1). This diffraction peak is also observed in Figure 4. loa which is also an XRD of the crystalline products from a low Si02 concentration solution. This diffraction peak and the one on the left side of 101 peak appear to be associated with a Na2CO3 phase. The unit cells of the products are shown in Table 5.3. Table 5.3 Unit cells of sodium aluminosilicate obtained from a solution containing 0.01 M Si02 and 0.38 M Na2CO3 at 95 C in an atmospheric reactor. Experiment Name Time (day) lb/ia Tb/Ic Unit cell (A) dsp8a (2) dsp8b (3) 5.073(3) dsp8c (4) 5.185(4) The relative intensity of the cancrinite at about 22 a (101) 20 increased marginally with time. This phenomenon is also observed in the product obtained from the higher Si02 concentration solution (Figure 5.1). Comparing the patterns in Figure 5.1 to those in Figure 5.10, it is possible, that cancrinite formation in a liquor with low Si02 concentration may be more rapid than in a liquor with high Si02 concentration. There is no evidence of sodalite2 formation (Figure 5.10) from the low Si02 liquor in contrast to the XRD results shown in Figure 5.1 for high Si02 (0.1 M) liquor. Conclusion: Si02 concentration affects the rate of formation of sodium alum jnosilicates. Low Si02 concentration appears to inhibit the formation of sodalite Phase Changes of Sodalite Seed Sodalite seeded experiments were carried out in Parr Bombs at 90 and 160 C for 13 or 14 days. The solution used had a concentration of 3.77 or 4.52 M NaOH, 1.64 M Al(OH)3, 0 c 110

136 or 0.01 M Si02 and or 0.38 M Na2CO3. The seed charge was 20 g dm3. Refer to Section for the liquor preparation procedure and sample treatment. Sodalite seed preparation is described in Section m (N 0 0 m 0 (N 0 N 4000 C (N (N 0 m 0 1 N ii (N 0 0 a Figure 5.10 XRD patterns of sodium aluminosilicates obtained from 0.01 M Si02 solutions in a liquor containing 0.38 M Na2CO3 at 95 C after (a) 5 (b) 10 and (c) 15 days (sodalitei + cancrinite), where "c" indicates sodium carbonate peak. 111

137 Phase Changes of Sodalite Seed at 90 C The ratios of XRD intensities of the peaks at about 14 (a, indexed as 110 for both sodalite and cancrinite), 19 (b, indexed as 101 for cancrinite) and 24 (c, indexed as 300 for both sodalite and cancrinite) of the solid products obtained after 14 days of growth at 90 C and the difference in the experimental conditions are listed in Table 5.4. The unit cell of sodalite seed is relatively small indicating sodalite2. The X-ray diffraction patterns of the products in Table 5.4 can be indexed as either cancrinite or a mixture of sodalite and cancrinite. As all the diffraction peaks can be indexed to cancrinite it has been assumed that this is the product. No double diffraction peaks were observed. Table 5.4 The experimental conditions and relative peak intensities of sodalite seeded products at 90 C. Exp. Name NaOH Na2CO3 lb/ia lb/ic Products Unit cell size (A) (M) (M) * * a (cubic) a (hex.) c (hex.) Sodalite seed 0 0 Sod (2) SNC9O-l4dIc Can (4) 5.187(3) SC9O-l4dhc Can (3) 5.181(2) SNC9O-l4dlc Can (2) 5.187(2) SC9O-l4dhc Can (4) 5.181(2) * where Ia, lb and Ic were the intensity of the peaks at about (a) 14 (Sod (110) + Can (110)), (b) 19 (Can (101)) and (c) 24 (Sod (211) + Can (300)) degrees 28 respectively. ** The experiment name refers to seed type, no added Na2CO3 at 90 C for 14 days at high caustic. The conversion to cancrinite can be observed by monitoring the growth of the peaks at 20 = 19 (Can (101)). This becomes a major peak in well-crystallised hydroxy The ratio of the intensities in Table 5.4 roughly represent the extent of cancrinite formation. Lower C032-{ J concentrations favour cancrinite formation at this temperature. XRD and FTIR analysis of the product obtained from the experiments at 90 C showed a trend of phase transformation from sodalite to cancrinite but at a slower rate than at 160 C. After 14 days of reaction, some new diffraction peaks at 18.9 and 27.5 (indexed as cancrinite 101 and 211 which would not exist in (cubic) space group) could be seen by XRD analysis. The FTIR sodalite finger prints at 737, 713 and 668 cm1 had faded and cancrinite ones at 690, 630 and 560 cm' The asymmetric stretch of 112

138 aluminosilicate framework at around 1000 cm-' did not obviously change. FTIR spectra of the four products showed a decrease in C032 concentration in crystals. Conclusion: With sodalite2 seeding at 90 Cfor 14 days, some cancrinite was observed. Cancrinite formation seems to be favoured by low C032- concentration; the same trend as in unseeded experiments Phase Changes of Sodalite Seed at 160 C XRD analysis of the crystalline phase obtained at 160 C showed that the initial seed (sodalite2) had gradually transformed to the final product (cancrinite). The experiment was then repeated in order to monitor this transformation. The sodalite seed was cubic with a = ± A. The final crystalline phase of cancrinite is hexagonal and had a unit cell of a = ± A and c = 5.172± A. The XRD patterns of the products as a function of time are shown in Figure lb/ia and lb/ic are given in table 5.5. Clear cancrinite diffraction peaks can be seen for the products obtained after 71.5 hours of experiments. This indicates that phase transformation from sodalite seed to cancrinite is much faster at 160 C than at 90 C. FTIR spectra of the starting sodalite seed (Figure 5.12a) and the product cancrinite crystalline phase samples (Figure 5.12b) are obviously different. The asymmetric stretch v(al-o) of the Al-O-Si sodalite framework at approximately 1000 cm-', splits into three peaks at 1095, 1035 and 1000 cm-' [100] The symmetric stretch of the sodalite framework at 737, 713 and 668 cm1 transformed into vibrations characteristic of cancrinite at 690, 628 and 567 cm1[991. The absorbance by C032 at 1450 cm1, compared to that of the framework at 1000 cm', decreased with time. The v(oh) absorption peaks near 3500 cm I also decreased. It appears that the crystalline phase became less hydrated and carbonated with time. The unit cell size of the seed and the products are given in Table 5.5, where the ratios of Tb/Ia and lb/ic (14 (Sod (110) + Can (110)), 19 (Can (101)) and 24 (Sod (211) + Can (300) respectivelyj degrees 20 represent the formation of cancrinite. The rate of transformation to cancrinite appears to be largely independent of caustic and C032 concentrations within the concentration ranges observed. Conclusion: No sodalite is observed after 14 days at 160 C in sodalite2 seeded experiments. Thetransforination of sodalite to cancrinite at 160 'C appears to be complete. 113

139 C 'I C 4000 h 3000 I THETA Figure 5.11' XRD patterns showing the transformation of sodalite seed in synthetic spent liquor with 0.01 M Si02 and 0.38 M Na2CO3 at 160 C Cancrinite Seed It has been shown that sodalite transforms to cancrinite at high temperatures, therefore cancrinite is the stable crystalline phase at high temperatures. The aim of these experiments was to determine whether there is any evidence of a reverse transformation of cancrinite to sodalite at low temperatures. 114

140 Si-O-AI(Si) 0 0) H 0 U) (a) -o 0 Ci) -o 1.0 (b) Wavenumbers Figure 5.12 FTIR spectra of (a) sodalite seed and (b) the cancrinite product obtained after the experiment at 160 C. Table 5.5 Unit cell size change of sodalite seeded experiments at 160 C. Experiment Name NaOH (M) Na2CO3 (M) Si02 (M) Tb/Ia * lb/ic Unit cell size a (A) Sodalite2 seed (2) c (A) (1) (6) SC16O-14d (1) (5) SNC16Od-14d (3) (1) SC16Od-14d (8) 5.160(7) SNC16O-14a (2) (5) SC16O-l4da (4) 5.178(1) * where Ia, lb and Ic were the intensity of the peaks at about (a) 14 (Sod (110) + Can (110)), (b) 19 (Can (101)) and (c) 24 (Sod (211) + Can (300)) degrees 20 respectively. ** the name of experiment refers to seed type, temperature and duration of the experiment. 0 d' after 160 means dissolution experiment. Synthesised cancrinite and plant cancrinite seeded experiments were carried out in Parr Bombs at 90 and 160 C for 13 or 14 days. The solutions contained 4.52 M NaOH,

141 M Al(OH)3, 0.01 M Si02 and or 0.38 M Na2CO3. The seed charge was 20 g dm3. Section describes the experimental procedure. It was found that both the synthesised and plant cancrinite phase remained unchanged at both 90 and 160 C. This indicated that the transformation of sodalite to cancrinite is a non-reversible process. Cancrinite has been shown here to be the stable phase in synthetic Bayer spent liquor up to 200 C. Conclusion: Between 90 and 200 C cancrinite is the thermodynamically stable phase Conclusion (1) There is a loss of C032 from nucleated sodalite crystals at 90 to 100 C (initial Si02 concentration 0.1 M), i.e. a crystalline phase change occurs from carbonate sodalite to hydroxy sodalite. (2) Sodalitei may partly or totally change to sodalite2 due to the leaching out of large anions such as C032. The latter has a smaller unit cell size. Lower Na2C 03 concentrations (such as M) favour sodalite2 formation. (3) In unseeded solutions sodalitei is the initial solid product between 90 to 100 C. Sodalitei appears to transform to sodalite2 at 90 to 100 C and then transforms to cancrinite. Sodalitei and sodalite2 nucleate at T 150 C in solutions containing M Na2CO3 due to limited C032 availability. Both sodalites transform to cancrinite. Cancrinite may form only in the presence of relevant seeds or by sodalite transformation; but not by nucleation between 90 to 200 C. The transformation to cancrinite is promoted by longer reaction time and higher temperatures. (4) The effect of solution C032 concentration on crystalline phase change is complicated: at high temperature (> 150 C), high C032 concentration promotes cancrinite formation more (indicated by higher (101) peak intensity in Tables 5.1 and 5.4) than low C032 concentration does; at low temperature (< 100 C), low solution C032 promotes cancrinite formation more (indicated by higher (101) peak intensity in Tables 5.2 and 5.5) than high C032 concentration does. (5) Cancrinite is the stable crystalline phase from 90 to 200 C. The crystalline phase changes are non-reversible in synthetic Bayer spent liquor between 90 to 200 C. (6) Combining the conclusion of Chapter 4 and Chapter 5, A mechanism of scale formation can be illustrated as below: 116

142 (<150 C) >- Sod1 >- Sod2 (High and low C032) (High,> 150 C) >- Sod1. (Low C032, > 150 C) > Sod2 > Can (Sod2 seeded) (Can seeded) Although not shown in the schematic mechanism, it is assumed that all solid products will be in equilibrium with the solution. (7) The transformation from sodalite to cancrinite obeys Ostwalds law of that the first phase formed is the least stable thermodynamically and is replaced by a more stable form, and so on, until a final, most stable product results. 117

143 CHAPTER 6. SILICA EQUILIBRIUM SOLUBILITY OF DIFFERENT PHASES IN SYNTHETIC SPENT LIQUOR 6.1. Introduction The result of previous chapters shows that sodalite is the phase that nucleates in Bayer spent liquor. Sodalite, however is not stable in the liquor and gradually transforms to cancrinite. This transformation is facilitated by higher temperatures i.e. the transformation is more complete at a higher temperature than that at a low temperature for a same period of time. For example, at 160 C after 14 days of reaction sodalite to cancrinite transformation is very obvious; at 90 C for the same period of time the formation of cancrinite can be observed by the presence in the XRD of cancrinite 101 peak but the main sodalite characteristics are still observable by XRD and FTIR analysis. Cancrinite, on the other hand, is very stable and does not change to sodalite at low temperatures such as 90 oc. In Bayer liquor, before and after gibbsite precipitation, Si02 equilibrium concentration has been reported as approximately 0.01 and M The actual Si02 concentration in spent Bayer liquor, however, is approximately 0.01 M. This indicates that thermodynamically the Si02 component in the spent liquor is not stable but the kinetics of its co-precipitation with sodium and aluminium is not fast. It has been reported that the solubility of sodium aluminosilicate in Bayer liquor is subject to many factors such as the crystalline phase, temperature, caustic concentration, Al(OH)3/NaOH ratio and impurities in the In this project, the caustic concentration and the Al(OH)3/NaOH ratio were fixed. The solubility of sodium aluminosilicate (sodalite and cancrinite) in synthetic spent liquor during this research was investigated as a function of crystalline phases, temperature and the concentration of one major impurity, Na2CO3. Several authors have studied Si02 solubility as a function of temperature with conflicting 156; 162; 166] In none of these studies has the sodium aluminosilicate phases of the final products been reported to preclude the possibility of phase changes. For example, our previous results indicated that within the time period of solubility determination, sodalite may transform to cancrinite, however, no cancrinite solubility data have been reported for comparison. 145; 173] Na2CO3 has been found to reduce the equilibrium solubility of however, no systematic study has been carried out on the solubility of aluminosilicate scale as a function of Na2CO3 concentration. Rozen, Pevzner, Tsekhovoskayat173' measured Si02 118

144 solubility at only one Na2CO3 concentration, 2.04 M, which was much more concentrated than in Bayer liquor. Breuer, Barsotti and and Avdeeva and investigated synthetic Bayer liquors with concentrations comparable to those used here. Ni, Prekhrest and measured the solubility of sodium aluminosilicate at 90 C only, over a wide range of NaOH concentrations (3.87 to M) and Al(OH)3INaOH ratios (0.04 to 0.5) with no Na2CO3 added and found that the solubility increases with increasing NaOH concentration. To understand the role that Na2CO3 plays in sodium aluminosilicate solubility in synthetic Bayer liquor, a systematic investigation as a function of Na2CO3 was carried out Silica Equilibrium Solubility of Synthetic Cancrinite Ueda, Murata and showed that for the crystallisation of mordenite and analcime (two types of zeolite) at 100 C, equilibrium had not been reached after 20 days. As shown by Avdeeva et experiments, determinations of equilibrium Si02 concentrations are complicated by the very slow approach to equilibrium, particularly at lower temperatures, and therefore solubilities are usually estimated from analysis of desilication 130] Since cancrinite does not change phase between 90 and 160 C, it is easier to discuss its solubility first. Solubility measurements of synthetic cancrinite in synthetic spent Bayer liquor were carried out in Parr Bombs at 90 and 160 C as a function Na2CO3 concentration. For the experimental conditions refer to Section The error of each measurement was ± 4%, to the point of 95% of confidence. Equilibrium was judged to have been achieved when the variation of Si02 concentrations between at least the last two measurements of Si02 concentration were within the experimental error. Normally this took 13 to 14 days The Effect of Sodium Carbonate Concentrations The solubility of synthesised cancrinite containing S042 in synthetic spent liquor as a function of Na2CO3 concentration is shown in Figure 6.1. The measurement of 5i02 equilibrium solubility of synthesised cancrinite in synthetic spent liquor was approached from above the solubility only. The 5i02 equilibrium solubility decreased from 1.9x103 to 1.7x103 M at 90 C and from 3.0xl03 to 2.5x103 M at 160 C with an increase in the Na2CO3 concentration from to 0.38 M. This indicates that systematically increasing Na2CO3 concentration will cause Si02 equilibrium in the liquor to decrease. The magnitude of the decrease in solubility of synthetic cancrinite caused by the increase in Na2CO3 concentration (AC= 1.4x104 M at 90 C and 4.2x104 M at 160 C) was not large. This may be due to the possible stronger attachment of S042 to the Al-O-Si 119

145 framework than OH- and CO32. This is confirmed by FTIR analysis (Figure 6.2). C032 concentration (1426 cm-1) in crystals decreases with time while that of S042 (1102 cm-1) does not. 3.0k r?r' C C 90 C I I Na2CO3(M) Figure 6.1 Si02 equilibrium solubility of synthesised cancrinite seeded solutions as a function of solution Na2CO3 concentrations (o) at 90 C and (.) at 160 C and ( ) predicted result by Equation 6.1, parameters from Table 6.1. Since no correlation for sodium aluminosilicate solubility in synthetic spent Bayer liquor as a function of Na2CO3 concentration is available in the literature, an empirical equation of the form given below was developed to correlate the equilibrium Si02 solubility data in synthetic cancrinite seeded experiments from above the solubility with temperature (90 and 160 C) and Na2CO3 concentration (0.043 to 0.38 M): CSiO2 eq = a + bt + + dtlogcna2co3 (6.1) Where CSi02 eq represents Si02 solubility in synthetic spent liquor (g dm3), T represents temperature ( C), CNa2CO3 represents Na2CO3 concentration (g dm3), a, b, c and d are regression parameters. The parameters of the above model were obtained by regression analysis and are given in Table 6.1. The upper and lower limits of the parameters within a 95% confidence interval are also listed in Table 6.1. Figure 6. 1 compares the experimental data (indicated by legends) and the results predicted by Equation 6.1 (represented by lines), indicating a good 120

146 agreement between the two. The influence of Na2CO3 concentration on Si02 equilibrium solubility is more pronounced at 160 C than at 90 C. Si) i Wavemumbem (crif1) Figure 6.2 FTIR spectra of synthetic cancrinite seed before the solubility experiment (a) and after the experiments at (b) 90 C, with M Na2CO3; (c) 90 C, with 0.38 M Na2CO3; (d) 160 C, with M Na2CO3 and (e) 160 C, with 0.38 M Na2CO The Effect of Temperature There was a significant differentiation between the Si02 solubility for synthesised cancrinite at 90 and 160 C. The solubility of synthetic cancrinite increases with the increase of temperature (Figure 6.1). The trend of decrease of Si02 equilibrium concentration as a function of increasing Na2CO3 is less at 90 C than that at 160 C. 121

147 Table 6. 1 Parameters estimated for Equation 6. 1 Si02 equilibrium solubility of synthesised cancrinite in synthetic spent Bayer liquor. Parameter Value Lower limit Upper limit a 1.69x x x102 b 1.25x x103 l.26x103 c 2.70x x x102 d -3.94x x x104 XRD diffraction patterns of synthetic cancrinite seed and crystalline products are shown in Figure 6.3. The most intense peak of the cancrinite changed from 300 (d= 3.67 A at ) for the seed to 221 (d = 3.24 A at ) for all the solid products. Leiteizen, Pashkevich, Firfarova and also observed that the most intense cancrinite peak was at d = 3.24 under alumina production conditions. A unit cell size reduction of the crystalline product as compared to the cancrinite seed was observed and is shown in Table 6.2. The ratio of lb/ia and Tb/Ic increased with increasing temperature. Table 6.2 The unit cell solubility experiments. and volume of the cancrinite phase before and after the Experiment Name Synthetic cancrinite T CC) Na2CO3 (M) Tb/Ia * Tb/Ic Unit cell Cell volume (A) (A) (5) 5.221(3) 740(1) SCNC9O** (7) 5.150(5) 712(2) SCC9O (6) 5.104(4) 696(1) SCNC16O (3) 5.152(2) 713.4(7) SCC16O (2) 5.161(2) 714.7(6) * where Ta, Tb and Ic were the intensity of the peaks at about (a) 14 (Sod (110) + Can (110)), (b) 19 (Can (101)) and (c) 24 (Sod (211) + Can (300)) degrees 20 respectively. The experiment name refers to seed type (synthetic cancrinite: SC), no added Na2CO3 (NC) at 90 FTIR results indicated a reduction in the C032 concentration (1426 cnr1) in the crystalline products during the experiments (Figure 6.2). CU32- concentration in the crystals was higher in the products obtained from a liquor containing M Na2CO3 than that from a liquor containing 0.38 M Na2CU3 (Figure 6.2b and c) at 90 C (not 122

148 Figure 6.3 XRD patterns of (a) synthesised cancrinite seed and cancrinite obtained after solubility experiments at 90 C in a liquor containing (b) and (c) 0.38 M Na2CO3 and at 160 C from a liquor containing (d) and (e) 0.38 M Na2CO3. normal but consistent with the result of Figure 4.5 of nucleated sodalite at the same temperature); but the trend was opposite at 160 C (normal and consistent with the result of Figure 4.8 of nucleated sodalite at that temperature). This confirmed that the influence of C032 on sodium aluminosilicate formation is not simple and the observation of this 123

149 kind of complicated phenomena is quite reproducible. The absorbances for the cancrinite frame work (1036, 1006 and 969 were better resolved at 160 C. The relative absorbance of the sulphate band at 1102 had not changed, confirming that the attachment of the sulphate anion to the framework was very stable. This was not influenced by reaction time, temperature or Na2CO3 concentration Silica Equilibrium Solubility of Plant Cancrinite The Effect of Sodium Carbonate Concentrations The solubility of plant cancrinite was measured in a similar manner to that of synthetic cancrinite except that measurement from below the solubility was carried out at 160 C for two Na2CO3 concentrations (0.043 and 0.38 M). The Si02 equilibrium solubility as a function of Na2CO3 concentration at 90 and 160 C is shown in Figure 6.4. It can be seen that Si02 solubility decreased from 2.6x103 to 1.7x103 M at 90 C 0.9x103 M) and from 3.5x103 to 2.6x1ft3 M at 160 C 0.9x1ft3 M) with the increase of Na2CO3 concentration from to 0.38 M. This trend is similar to that of synthetic cancrinite C Na2CO3(M) 90 C Figure 6.4 Si02 equilibrium solubility of plant cancrinite in synthetic spent liquor as a function of Na2CO3 concentrations (o) at 90 C from above (initial Si02 concentration 0.01 M), (.) at 160 C from above (initial Si02 concentration 0.01 M) and at 160 C from below (initial Si02 concentration 0 M) and ( ) predicted by Equation 6.1, parameters from Table

150 A similar regression of Equation 6.1 of Si02 equilibrium solubility of plant cancrinite in synthetic spent Bayer liquor (from above) with temperature (90 and 160 C) and Na2CO3 concentration (0.043 to 0.38 M) was performed. The parameters obtained by regression analysis and the upper and lower limits of them within a 95% confidence interval of the model of Equation 6.1 are given in Table 6.3. Figure 6.4 compares the experimental data (legends) with the results predicted by Equation 6.1 (lines). Table 6.3 Parameters estimated for the model of Si02 equilibrium solubility (Equation 6.1) of plant cancrinite in synthetic spent Bayer liquor. Parameter Value Lower limit Upper limit a b 8.72x x x104 c -4.59x lx x 10-2 d -1.67x1ft4-1.76x1ft4-1.57x The Effect of Temperature As for synthetic cancrinite, plant cancrinite solubility increased with the increase of temperature from 90 C to 160 C. Si02 equilibrium solubility values measured at 160 C from below are larger than those measured from above (Figure 6.4). The reason may be that the solution was equilibrated with phases of slightly different structure such as differing anion content in the channels of the cancrinite. Plant cancrinite did not transform phase to sodalite at 90 C as confirmed by both XRD and FTIR analysis. The unit cells of the crystals were observed to decrease in plant cancrinite seeded experiments (Table 6.4). The crystals obtained at 160 C from a liquor containing Si02 below the solubility (below in later text) had smaller unit cells than those obtained from a liquor containing Si02 above ("above in later text) the solubility. FTIR analysis indicated that the initial C032 level in the plant cancrinite seed was much lower than in the synthetic cancrinite. C032 concentration in the final crystalline products of the plant cancrinite seed was higher at 160 C than that 90 C, indicating a greater extent of anion exchange at 160 Three cancrinite framework asymmetric stretch peaks at about 1000 cm' can be clearly seen but are not as sharp as those in synthetic cancrinite (Figure 6.5). There could be slightly higher C032 concentration in the crystals from the "above" experiment at 160 C (Figure 6.Sb) than in those from the "below" (Figure 6.5a). This may contribute to a lower Si02 equilibrium concentrations measured from "above" than those from "below" at 160 C (Figure 6.4). There was some S042 (1109 plant cancrinite. This may be due to the impurities in plant spent Bayer liquor. in the 125

151 Table 6.4 the solubility experiments. The unit cell and volume of plant cancrinite seed crystals before and after Experiment Name Plant cancrinite Na2CO3 (M) Si02 (M) T CC) lb/ia * lb/ic * Unit cell a (A) c (A) Cell volume (A3) (3) 5.179(2) 724.8(7) PCNC9O** (4) 5.178(6) 718(1) PCC9O (1) 5.189(1) 725.3(2) PCNC16O (2) 5.184(2) 722.3(4) PCC16O (2) 5.182(1) 721.6(4) (3) 5.160(2) 715.3(7) PCC16OdO (9) 5.133(1) 703(1) * where Ta, Tb and Ic were the intensity of the peaks at about (a) 14 (Sod (110) + Can (110)), (b) 19 (Can (101)) and (c) 24 (Sod (211) + Can (300)) degrees 28 respectively. ** The experiment name refers to seed type (plant cancrinite: PC), M Na2CO3 (NC) at 90 C. 0 "d" approach of equilibrium from below Silica Equilibrium Solubility of Sodalite Solubility measurements of sodalite in synthetic spent Bayer liquor were carried out in Parr Bombs at 90 and 160 C as a function of Na2C 03 concentration. For the experimental conditions refer to Section The method of the measurement was similar to that of synthetic cancrinite solubility The Effect of Sodium Carbonate Concentrations Si02 equilibrium solubilities measured from sodalite seeded experiments as a function of Na2CO3 concentration are shown in Figure 6.6. A comparison of the results interpolated from literature is also shown. Si02 equilibrium solubility in liquors containing M Na2CO3 solution was higher than that in liquors containing 0.38 M Na2CO3 solution. The solubility of sodalite (as Si02) decreased from 4.3x103 to 2.6x103 M with the increase of Na2CO3 concentrations from to 0.38 M (AC = 1.7x103 M) at 90 C and from 3.9x103 to 2.8x1ft3 M at 160 C (AC = l.1x103 M). Systematically increasing Na2CO3 concentration in the liquor caused a systematic decrease in Si02 solubility of sodium aluminosilicate. 126

152 o 0 AI-O-Si 1.6 L) 1.4 C C', -o U m 0 0 co'n N (a) Wavenumbers (cm-1) Figure 6.5 FTIR spectra of plant cancrinite after being in liquors containing 0.38 M Na2CO3 and, (a) 0.0 and (b) 0.O1M Si02 initial concentrations for 14 days at 160 C. A similar regression of Equation 6.1 for Si02 equilibrium solubility of sodalite in synthetic spent Bayer liquor (from above the solubility) with temperature (90 and 160 C) and Na2CO3 concentration (0.043 to 0.38 M) was performed. The parameters and the upper and lower limits within a 95% confidence interval for the model of Equation 6.1 obtained by regression analysis are given in Table 6.5. The results predicted by Equation 6.1 are fitted as 2 lines through the experimental data (legends) in Figure 6.6, indicating a very good agreement between the two. 127

153 (a) (b)160 C 3.5 I I Na2 CO3 (M) I Figure 6.6 Silica equilibrium solubility of sodalite seeded solutions as a function of solution Na2CO3 concentrations (o) at 90 C from above Si02 equilibrium solubility, (.) at 160 C from above Si02 equilibrium solubility, at 160 C from below 5i02 equilibrium solubility, Breuer et al. [19] at 90 C, (M) Breuer etal. [19] at 160 C and (A) Ni et al. [156] at 90 C and ( ) predicted by Equation 6.1, parameters from Table 6.5. The solubility curve obtained at 90 C by molybdate method is well matched to those values literature'19' and [156], and that calculated from the empirical formulas of Adamson, Bloore and Carr'181 (5.3x103 M as Si02) and Oku et aij162' (5.1x103 M as 5i02). The published solubilities (all converted as ICP results in Figure 6.6 for comparison) are very close to the Si02 solubility measured at 90 C from a solution containing M Na2CO3 with sodalite seeding. The influence of temperature and Na2CO3 concentration were examined by Adamson et al. and Oku et al. (Equation 2.10). 128

154 Ni et did not indicate that there was Na2CO3 in their solution but at their NaOH concentration, a minimum amount of M was assumed when the comparison was carried out. stated that the sodalite solubility in plant spent Bayer liquor should be less than 2x x103 M as Si02. This value is very close to the result from a liquor containing 0.38 M Na2CO3 in Figure 6.6a and is much smaller than those reported in the There are impurities in plant liquor not present in pure synthetic liquor. They may have had a significant effect on Si02 solubility. Table 6.5 Parameters estimated for Equation 6. 1 of Si02 equilibrium solubility in sodalite seeded synthetic spent Bayer liquor. Parameter Value j Lower limit Upper limit a b -8.68x x x104 c d -6.28x x x The Effect of Temperature: Crystalline Phase Transformation The decrease of Si02 equilibrium solubility caused by the increase of Na2C 03 concentration from to 0.38 M was greater for sodalite at 90 C than for sodalite transformed to cancrinite at 160 C (Figure 6.6). Si02 solubility at 160 C (Figure 6.6) is not significantly greater than at 90 C. Parameter 'b" listed in Table 6.5 has a negative value while those in Tables 6.1 and 6.3 are positive. There was a crystalline phase transformation of the seed from sodalite to cancrinite at 160 C (refer to Section 5.3.2). This led to a lower 5i02 equilibrium concentration at 160 C than at 90 C in a liquor containing M Na2CO3. A similar phenomenon was also observed by Arlyuk et alj'6311. The result obtained at 160 C did not match Breuer et at. result possibly as the crystalline phase transformation from sodalite to cancrinite was much faster at 160 C and the liquor was equilibrating with a different crystalline phase at the end of the experiments. Breuer et at. 's119' experiment lasted for only 60 hours at 150 C. The crystalline phase transformation from sodalite to cancrinite has only just started after this time as shown in Figure The sodalite seed had mostly transformed to cancrinite after 13 days at 160 C. The transformed phase obviously has a lower solubility in the liquor than the starting seed. The solubility Breuer et at. measured was not for a system at Oku and and Adamson, Bloore and stated that 5i02 equilibrium solubility was independent of temperature. This could be explained by a possible crystalline phase transformation in their systems to a less soluble crystalline product at 129

155 higher temperatures. The solubility of the transformed phase at high temperatures may have been similar to that of original seed at low temperatures. Eremin, Mel'nikova and 1661 and Shvartsman and Volkova11651 observed a decrease in Si02 concentration between 90 to 125 C followed by an increase between 125 and 175 C. This was explained as being due to the presence of "truly soluble Si02 in a colloidal state". However the meaning of this statement is not clear. Cournoyer, Kranich and reported that at 123 C in some cases the crystallisation of hydroxy sodalite was followed by a recrystallisation to hydroxy cancrinite. Yuhas, Orbanne and observed that the Si02 concentration does not decrease if the temperature is not high enough for this transformation, e.g. sodalite at 90 C. The results shown in Chapter 5 indicated that sodalite transformation to cancrinite occurred at both 90 (Table 5.4) and 160 C (Table 5.5) but at a much slower rate at 90 C. Yuhas, Orbanne and Matula1251 proposed a so-called metastable solubility and true solubility. The former is the solubility of the starting phase and the latter is the solubility of the equilibrium phase. Thus Breuer et al.'s solubility at high temperature is likely to be a metastable solubility Comparison of the Si02 Solubility among Different Seeds For all three types of seeding examined, the increasing addition of Na2CO3 had the largest effect on sodalite and plant cancrinite seeded solutions and the least effect on synthetic cancrinite seeded solutions. Increasing Na2CO3 concentration has the most effect on sodalite solubility in synthetic spent liquor at 90 C. The decrease in Si02 concentration, z\c, from solutions containing to 0.38 M Na2CO3 at 90 C was, AC(sodalite seed) = 1.7x103 M, AC(synthetic cancrinite seed) = 0.1x103 M, AC(plant cancrinite seed) = 0.9x103 M, and at 160 C AC(sodalite seed) = 1.1x103 M, AC(synthetic cancrinite seed) = 0.4x103 M, AC(plant cancrinite seed) = 0.9x103 M. Figure 6.7 gives the comparison of the solubility of different seeds at (a) 90 and (b) 160 C respectively. For a given Na2CO3 concentration the solubility decreased as: sodalite> plant cancrinite> synthetic cancrinite. When Na2CO3 concentration was greater than M, the solubility curves at 160 C of plant and synthetic cancrinite are extremely similar. The influence of temperature on the solubility of the 3 types of seeds for a given Na2CO3 concentration is shown in Table 6.6 and is the smallest for plant cancrinite. 130

156 I Na2CO3 (M) Figure 6.7 The solubility of (o) sodalite seed, synthesised cancrinite seed and (.) plant cancrinite seed in synthetic spent liquor at (a) 90 and (b) 160 C as a function of Na2CO3 concentration. Table 6.6 The difference of Si02 equilibrium solubility between 160 and 90 C in synthetic spent liquor for 3 types of seeds under the same Na2CO3 concentration. Na2CO3 (M) ACx (M Si02) sodalite -4 2 (M Si02) synthetic cancrinite 11 8 ACx (M Si02) plant cancrinite

157 If spent Bayer liquor was seeded with cancrinite, its Si02 concentration could be reduced to a lower level than that of sodalite solubility. It has previously been demonstrated that sodalite is the phase to nucleate between 90 and 200 C but it transforms to cancrinite. Thus it may be possible to reduce sodium aluminosilicate scaling by cancrinite seeding Conclusion (1) Under the same experimental conditions, the solubility of different crystalline phases decreased in the following order: sodalite > plant cancrinite > synthetic cancrinite in synthetic spent Bayer liquors. (2) Systematically increasing the Na2CO3 concentration in synthetic spent liquor (0.043 to 0.38 M), caused the Si02 equilibrium solubility to decrease systematically. This drop was most pronounced for the sodalite seeded solutions at 90 C. (3) The Si02 solubility of sodalite seeded liquor did not show a significant increase between 90 and 160 C due to the phase change from sodalite to the more stable cancrinite. The crystalline products from all sodalite seeded solutions included some cancrinite. The proportion of cancrinite present was greater for the solutions held at 160 C (Table 5.5) than for the solutions at 90 C (Table 5.4). (4) Where no phase change took place, i.e. with cancrinite seeded solutions, the Si02 solubility was always greater at 160 C than at 90 C. (5) A reduction of C032 concentration in the seed was observed on in situ aging with sodalite and synthesised cancrinite seed but was not observed in the plant cancrinite seed where the initial concentration of C032 was very low. (6) The unit cell size for all synthetic cancrinite seed decreased on in situ aging compared to the seed prior to the experiment. No unit cell size reduction took place for plant cancrinite seed. No phase variation occurred in the products of either type of cancrinite seed regardless of temperature or Na2CO3 concentration. 132

158 CHAPTER 7. CRYSTALLISATION OF SYNTHETIC SPENT LIQUOR 7.1. Introduction Si02 concentration in plant spent liquor is approximately 0.01 M. As this concentration is above the equilibrium solubility, precipitation may occur in plant pipes and heat exchangers. In the 7.3 metre long heat exchanger tubes in which scale occurs in the plant, liquor residence time of about 5 seconds prevails. Experiments were carried out at 90 and 160 C for 4 hours using synthetic spent liquor to investigate scale formation. The aim of these experiments was to determine the nucleation/growth boundary and to investigate the relative desilication and scale formation processes in the absence and presence of seed Unseeded Experiments Unseeded experiments were carried out in liquors containing 4.52 M NaOH, 1.25 M Al(OH)3, or 0.38 M Na2CO3 and 0.01 or 0.02 M Si02 at 90 and 160 C. The details of the experimental process are given in Section The Effect of Supersaturation The observed 5i02 concentration as a function of time in unseeded experiments is shown in Figure 7.1. No significant change in 5i02 concentration and no nucleation (visual) was observed in unseeded synthetic spent liquor with 0.01 M initial Si02 concentration in 4 hours of reaction time at both 90 and 160 C although the concentrations were above the equilibrium solubility. Figure 7.2 shows 5i02 supersaturation (the difference between 5i02 concentration and Si02 equilibrium solubility under similar conditions) versus time for the 0.02 M Si02 experiments. The higher Na2CO3 gave rise to more pronounced desilication. For the time period of 4 hours, therefore, at 160 C 0.02 M Si02 is sufficient to induce crystallisation whereas 0.01 M Si02 is not. When 0.02 M Si02 initial concentration was used, some scale formed at 90 C (Figure 7.2). No observable scale crystals formed in solution but a thin layer of scale formed on the stainless steel surfaces in contact with the liquor. At the end of the experiment the solution appeared cloudy but insufficient precipitated materials had been formed for XRD analysis. A larger amount of sodium aluminosilicate, however, was obtained from 133

159 solutions of both Na2CO3 concentrations at 160 C. The products obtained from the solution containing M Na2CO3 at 160 C were (a = 8.962(3) A) and sodalite2 (a = 8.918(4) A). Only sodalite1 (a = 8.965(1) A) formed from the solution containing 0.38 M Na2CO3 (Figure 4.7). Both crystals in solutions and scale on the wall were observed at 160 C and initial Si02 concentration of 0.02 M. 1OAI 7.5 C = = = = = = = = = = = =& Time (mm) Figure 7.1 Solution Si02 concentration in unseeded desilication experiments as a function of time (with 0.01 M Si02 initial concentration) (o) at 90 C and M Na2CO3, (.) at 90 C and 0.38 M Na2CO3, at 160 C and M Na2CO3, (A) at 160 C and 0.38 M Na2CO3, (a) SiO2 solubility at 90 C and M Na2CO3, (.) solubility at 90 C and 0.38 M Na2CO3, (.) solubility at 160 C and M Na2CO3, and (.) SiO2 solubility at 160 C and 0.38 M Na2CO3. These results are consistent with Muller-Steinhagen, Jamialahmadi and report; deposition increased with increasing SiO2 concentration. They found that at low Si02 concentration ( M) aluminosilicate scale mainly deposited on heat exchanger pipes (30 to 105 C). This showed that only heterogeneous nucleation occurred. At high silica concentration (>0.027 M) bulk nucleation occurred in the same temperature range The Effect of Sodium Carbonate Concentration and Temperature All samples from the solution containing 0.02 M 5i02 and M Na2CO3 at 90 C were optically clear but a thin layer of scale was observed on the stirrer and other surfaces of the 134

160 reactor. The final samples from the solution containing 0.02 M Si02 and 0.38 M Na2CO3 at 90 C appeared cloudy indicating nucleation in the bulk solution had occurred as well as nucleation on substrate surfaces. The amount of crystalline product was too small to be characterised. Neither of these solutions gave rise to a significant solution Si02 concentration decrease over the 4 hour experimental period (Figure 7.2). As Na2CO3 influences solubility, the higher C032 solution had a lower solubility and hence, a higher supersaturation (Figure 7.2) I Figure Time (mm) Solution Si02 supersaturation in unseeded desilication experiments as a function of time (with 0.02 M 5i02 as initial concentration) (o) at 90 C and M Na2CO3, (.)at 90 C and 0.38 M Na2CO3, (A) at 160 C and M Na2CO3 and (A) at 160 C and 0.38 M Na2CO3. Desilication is more rapid at 160 C than at 90 C. This agrees with Muller-Steinhagen et results. It is reasonable that the amount of sodium aluminosilicate formed at 160 C in our experiment is much more than they obtained at and below 105 C. Desilication at 160 C was more rapid in the 0.38 M Na2CO3 solution than in M Na2CO3 solution due to a higher supersaturation. There was an increase in scale formation and bulk nucleation with increasing Na2CO3 concentration for a given Si02 concentration and temperature. This phenomenon was also observed by Breuer et 135

161 7.3. Sodalite Seeded Crystallisation Sodalite seeded experiments were performed with experimental conditions similar to those for the unseeded experiments (Section 7.2). 5 g 3.20 m2 g-1) to each experiment. of seed was added (seed surface area Unlike the unseeded experiments, no observable scale formed on the wall of the reactor at either 90 or 160 C, with either or 0.38 M Na2CO3 and either 0.01 (Figure 7.3) or 0.02 M (Figure 7.6) initial Si The Effect of Temperature Si02 supersaturation in sodalite seeded experiments at 90 and 160 C in liquors containing and 0.38 M Na2CO3 and 0.01 M initial Si02, as a function of time, is shown in Figure 7.3. Si02 supersaturation decreased with time due to seed growth and secondary nucleation. No heterogeneous nucleation on the wall of the vessel was observed. The extent of desilication in 4 hours at 160 C was greater than at 90 C (Figure 7.3). Thus the higher the temperature, the faster the 17; 22; 27; 171; 178; 197] 7 IS 11) Time (mm) Figure 7.3 Solution Si02 supersaturation in precipitation experiments in the presence of 5 g dm3 sodalite seed with an initial concentration of 0.01 M Si02 as a function of time (o) at 90 C and M Na2CO3, (.) at 90 C and 0.38 M Na2CO3, at 160 C and M Na2CO3 and (A) at 160 C and 0.38 M Na2CO3. 136

162 The Effect of Sodium Carbonate Concentration Si02 supersaturations were greater at the beginning of the experiments in liquors containing 0.38 M Na2CO3 (due to a lower equilibrium concentration) than those in liquors containing M Na2CO3 and dropped to less than those in liquors containing M Na2CO3 with time (Figure 7.3). This indicates that crystallisation in solutions containing 0.38 M Na2CO3 was greater than in solutions containing M Na2CO3 at the same temperature or the higher the supersaturation, the faster the crystallisation. FTIR spectra (Figure 7.4) show that the products obtained at 90 C and 160 C from solutions containing M Na2CO3 were still sodalite; a change in the sodalite finger print spectrum was seen in the product obtained at 160 C from solutions containing 0.38 M Na2CO3. The Al-O-Si asymmetric stretch shifted from 979 of the seed to 979, 981, 982 and 993 cm' of the products obtained at 90 C from a liquor containing and 0.38 M Na2CO3 and at 160 C from a liquor containing and 0.38 M Na2CO3 respectively, indicating a tighter bonding with increases of Na2CO3 concentration and temperature. The concentration of C032 in the crystalline phase was largely reduced after the experiments as compared to the original sodalite seed. The crystalline product from the 0.38 M Na2CO3 was finer than that obtained from M Na2CO3 solution as indicated by SEM analysis. Figure 7.5 shows the comparison of the seed and two crystalline products obtained at 160 C from liquors containing and 0.38 M Na2CO3. The seed is stored in ajar and is aggregated. When put in solution with 400 rpm stirring the aggregates were dispersed. As the Si02 supersaturation is higher in a liquor containing 0.38 M Na2CO3 than in a liquor containing M Na2CO3, more rapid nucleation occurred and finer crystals resulted The Effect of Seeding In the presence of sodalite seed, desilication (Figure 7.3) is more rapid than in the absence of seed (refer to Figure 7.1), where no crystallisation occurs from a liquor containing an initial Si02 concentration of 0.01 M. In contrast to the unseeded experiments, no scale formed on the metal surface of the autoclave during the sodalite seeded experiments. This indicates that if seed is used to reduce Si02 levels in plant liquor, scaling may be minimised by desilication through growth on the seed as opposed to the plant metal surfaces. The solid particles could then be filtered and a liquor with a much reduced Si02 supersaturation could be recycled through the circuit. 137

163 993 At Si U Cu.0 I- 0 'I.' e Wave numbers Figure 7.4 FTIR spectra of (a) sodalite seed, and the crystalline product obtained from (b) a liquor containing M Na2CO3 at 90 C, (c) a liquor containing 0.38 M Na2CO3 at 90 C, (d) a liquor containing M Na2CO3 at 160 C and (e) a liquor containing 0.38 M Na2CO3 at 160 C The Effect of Supersaturation Initial Si02 concentrations of 0.01 and 0.02 M were tested under identical conditions (seed charge was the same) for comparison. The results are shown in Figure 7.6. The relative rate of desilication increased with increasing supersaturation (Figure 7.6). This is particularly evident in the liquor containing 0.02 M Si02 and 0.38 M Na2CO3. The decrease of z\c was faster in the presence of 0.38 M Na2CO3 in solution with both initial Si02 concentrations. This is also the case for 0.01 M initial Si02 at 160 C (Figure 7.3) Secondary Nucleation Table 7.1 shows the unit cell sizes of the crystalline products from the solutions containing 0.01 M Si02 after 4 hours of experiment. In contrast to the 14 day solubility experiments (Tables 5.4 and 5.5), the unit cell size of the sodalite seed did not change significantly. After 4 hours the sodalite phase remained unchanged. Only the product obtained at 160 C 138

164 (a) (b) (c) Figure 7.5 SEM micrographs of sodalite seed (a) and sodalite obtained after 4 hours sodalite seeded crystallisation at 160 C from the liquor containing (b) and (c) 0.38 M Na2CO