Lithium Production from Spodumene by Ernest Mast A Thesis submitted to the Faculty of Graduate Studies in partial fulfillment of the requirements of the degree o f Master of Engineering. I Department of Mining and Metallurgical Engineering McGill University, Montreal, Canada © Ernest Mast May 1989 rtf Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. To the incomprehensible forces which started it all. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ABSTRACT A new metallo-thermic reduction process designated "Melt - Leach Evaporation " is under development for the extraction o f lithium and other Group IA and IIA metals. The metal to be recovered is leached out of the ore or concentrate by an excess o f liquid metal reductant which is then vacuum refined to recover the metal as a solid or liquid condensate. The experiments performed contacted approximately one mole of Beta spodumene, LiAlSi^O^, with molten aluminum-magnesium. The molar ratio of the aluminum to spodumene was approximately 86 to 1 and the molar ratio of magnesium to spodumene was increased from zero to ten. The experiments were performed in alumina crucibles, under an argon atmosphere, at 900°C for ninety minutes. Mixing was provided by mechanical stirring. Assays of lithium and silicon in the cast ingot confirmed that lithia and silica contained in the spodumene were reduced and dissolved into the excess molten aluminum reductant. As much as 60 % o f the lithium in the spodumene was recovered to the aluminum ingot cast after the experiment. Increasing the magnesium to spodumene ratio improved the recovery of lithium to the ingot. This was as predicted by a thermodynamic model. X-Ray diffraction results detected spinel, MgAl20 4 and silicon in the powder residue for all experiments. These compounds were formed by the following two-step reaction process, Li20 Al20 34 S i0 2 + 11 A1 Li20 A l 20 3 + Mg = = L i20 A120 3 + 4Si + i A l ^ 2Li + M g O -A l^ [i] [2] which produced the lithium and silicon dissolved m the cast ingot. A one-step reaction may have occurred simultaneously at magnesium to spodumene ratios greater than or equal to eight to one, Li20 A l 20 3-4Si02 + 9Mg = 2Li + M g O A l ^ + 4Si + 8Mg0. I i Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. m RESUME Un nouveau procedd de reduction thermo-m^taJlique commundment appeie "Fusion - Lixiviation - Evaporation" est sous ddveloppement pour 1'extraction du lithium et d ’autres metaux du groupe IA et IIA. Le metal k extraire est lixivie a partir du mineral ou du concentre par un exc&s de r^ducteur metalhque liquide Ce dernier est ensuite raffine sous vide pour recouvrer le metal d ’interet sous la forme d ’un solide oud ’un liquide condense. Les essais expenmentaux imphquaient une reaction avec pr6s d ’une mole de spodumene Beta, LiAlSi.,0 , et un alliage d'aluminium -magnesium liquide. Le rapport tnolaire aluminium/spodumene etait environ de 86 pour 1, tandis que le rapport magnesium/spodumene variait suivant un accroissement de zero h dix. Les experiences furent condmtes en tenant compte de certains para- metres ltnportants tels que- des creusets en alumine, une atmosphere d ’argon et une temperature de 900°C duiant 90 minutes avec agitation mdcanique. L’analvse chimique des lingots d ’alumimum a confirmee que 1’oxide de lithium et la silice contenus dans la spodumene ont dtd rdduit et dissou,-. dans l’exces d’aluminium liquide. Jusqu’a 60% du lithium present dans la spodumene a ete recouvre dans le lingot d ’alumminm aprbs l’expdnence. L ’augmentation du rapport magnesium/spodumene a demontrde une meilleure extraction du lithium dans le lingot. De plus, ce resultat a et6 prddit ti l ’aide d ’un modele thermodynamique. Les resultats obtenus par diffraction-X ont ddmontrds que le rdsidu poudreux obtenu lors des experiences etait forme de spinel, (MgAl20 4) et de silicium. Ces composes ont ete formes en deux etapes considerant les reactions suivantes, Li20-A l20 3 4S i02 + Li20-A l20 3 + Mg A1 = = L i 20-A120 3 + 4Si + i A120 3 2Li + M gOA l2C>3 m [2] lesquelles ont produites le lithium et le silicium dissous dans le lingot. Par contre, une reaction considerant une seule etape peut avoir eu lieu pour un rapport magnesium/spodumene plus grand ou dgal k huit. Li O A1 0 4SiO + 9Mg 2 2 3 2 = 2Lt + MgO A1 O + 4Si + 8MgO. 2 3 11 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. pi ACKNOWLEDGEMENTS I would first like to express gratitude to my supervisor, Dr. Ralph Harris, who was most responsible for my decision to undertake this endeavour. His foresight, patience, encouragement and special talents were much appreciated. The author is indebted to members of Dr. H am s’ group: Robert Selby, Zhou Wang, Liu Jin and especially Birendra Jena for their helpful assistance in all aspects of the project. I would like to acknowledge the helpful suggestions from Professors James Toguri and Bert Wraith of the Universities of Toronto and Newcastle, respectively. Thanks to all of the students, technicians and professors in the Department for making the author’s stay at McGill a most rewarding one. The author wishes to acknowledge The Tantalum Mining Company of Canada and especially Mr. John Fleming, for expressing an interest in this project, H providing matenals and performing chemical analyses. Thanks to my office mates, Murray Brown, Weixing Wang and Sugundo for their companionship and acceptance o f my territorial expansion within the office. I would like to thank my family for their support. A special thank you to Cannen, my girlfriend, for being herself. i 11 I Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TABLE OF CONTENTS 1 Page No. A BSTRA CT RESU M E ii ACKNOW LEDGEM ENTS iii T A B L E O F C O N TEN TS iv L IS T O F TA B LES viii L IS T O F FIG U R ES x C H A P T E R 1: IN T R O D U C T IO N 1. 1. Melt-Leach-Evaporation 1 I 2.1. Characteristics and Uses of Lithium Metal 1 1.2.2. Lithium Occurrence 4 1.2.3. Lithium Industry 5 1.2.4. Lithium Extraction Technology 6 1.3. Alpha to Beta Spodumene Transition 8 C H A P T E R 2: L IT E R A T U R E SU R V E Y 11.1. Metallo-Thermic Reduction Processes 12 11.2. Lithium Production by Metallo-Thermic Reduction 12 11.3. Lime-Spodumene Reactions 13 11.4. Interfacial Phenomena 14 I IV Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. CHAPTER 3 :THERMODYNAMIC ANALYSIS OF SPOD U M EN E R ED U C TIO N III. 1. Reducing Agent 17 C H A P T E R 4: L IM E -SP O D U M E N E TESTS IV. 1. Introduction 22 IV.2. Apparatus and Reagents 22 IV.2. Procedure 23 IV.3. Results of Lime-Spodumene Tests 24 IV.4. Discussion 26 IV.5. Conclusions, Lime-Spodumene Tests 28 C H A P T E R 5: E X P E R IM E N T A L < V .l. Objective 29 V.2. Experimental Variables 29 V.3. Materials 29 V.4. Apparatus 30 V.5. Experimental Procedure for Spodumene Reduction Tests 34 V.5.1. Experimental Preparation 34 V.5.2. Experimental Procedure 34 V.6. Experimental Program 35 C H A P T E R 6: RESULTS VI. 1. Spodumene Transition 37 VI.2. Visual Observations of Reduction Products 39 VI.3. XRD Analysis of Reduction Products 47 VIA. Atomic Absorption Analysis 49 VI.4.1. Sampling For Atomic Absorption Analysis 50 VI.4.2. Atomic Absorption Analytical Methods 50 VIA.3. Atomic Absorption Results 54 VI.5. Scanning Electron Microscope Analysis 58 C Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. C H A PT E R 7: EXAMPLE CALCULATIONS VII.I. VII.2. Determining Sample Concentrations From Absorbance Readings 60 Sample Calculations o f Analytical Error 61 C H A P T E R 8: D ISC U SS IO N VIII 1. Introducr.on VIII.2 1. Discussion of Present Experimental Program 63 VIII.2 2. Discussion of Experimental Procedure 63 VIII.2.3.1. Relative Errors of Sample Assays 64 VIII.2.3.2. Comparison of Assays with an Outside Analysis VIII.3. Experimental Modeling 67 VIII.3.1. Determination of Gas Purging Rate 67 VIII.3.2. Formation of Solid Species 69 VIII.4. Mass Balances 70 VIII.5. The Effect of Magnesium Addition on the Reaction Products Assays 73 VIII.6. Kinetics of the Spodumene Reduction System 86 VIII.7. Discussion of XRD Results 88 V III.8. Wetting Effect of Magnesium 88 VIII.9. Discussion of Reaction Mechanisms 89 VIII. 10. Discussion of Reaction Pathways 90 VIII. 10.1. Two-Step Reaction Pathway 90 VIII. 10.2. One-Step Reaction Pathway 91 V III.11. Vacuum Refining of Lithium from Aluminum 91 VIII. 12. Comparison of Experimental Results with F*A*C*T Calculations 93 VIII. 12.1. Introduction 93 VIII. 12.2. Comparison o f Thermodynamic Model with Experimental Results V VI Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 93 C H A P T E R 9: C ONCLUSIONS A N D FUTURE WORK IX. 1. Conclusions 101 IX.2. Future Work 102 C H A P T E R 10: C O N T R IB U T IO N S 104 REFEREN CES 105 A P P E N D IC E S A: B: F*A*C*T Simulation o f Alumino-Thermic Reduction of Spodumene 109 Phase Diagrams 116 i VI I Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. LIST OF TABLES Table 1.1. Physical Properties of Lithium 2 Table 1.2. Lithium Bearing Minerals 4 Table 1.3 Uses of Lithium Metal in the U.S. (Short Table 1 4 Operating Conditions for Li Electrolysis Cell Table 1 5. Crystal Structural Data for Alpha and Beta Tons) 5 8 Spodumene 10 Table 2.1. Surface Tensions for Selected Liquid Metals 15 Table 3.1. Characteristics of Possible Reductants 18 Table 3.2. Standard Free Energy of Solution o f Elements in Liquid A1 and their Absolute Entropies Table 3 3. F*A*C*T Output o f Spodumene Reduction with Al-Ca A l'oy at 1173 K and 1 Atm. Table 3.4. 19 20 F*A*C*T Output of Spodumene Reduction with Al-Mg Alloy at 1173 K and 1 Atm. Table 4.1. Spodumene Assays Table 4.2. XRD Results of Lime Spodumene Sinters and 20 22 Compounds Predicted by F*A*C*T 26 Table 5.1. Specifications of Aluminum (Weight%) 29 Table 5.2 Spodumene Reduction Experiments Performed 36 Table 6.1. Products o f Spodumene Reduction Expenments 40 Table 6.2. Settings for Li A.A. Analysis 51 Table 6.3 Settings for Mg A.A. Analysis 52 Table 6.4. Settings for Si A.A. Analysis 53 Table 6.5. A.A. Raw Data for a Set of Li Standards and Ingot Assays 55 Table 6.6. Assays of Condensate (Weight%) 55 Table 6.7. Assays of Flue Powder (Weight%) 56 Table 6.8. Assays of Residue (Weight%) 56 Table 6.9. Assays of Ingot (Weight%) 57 Table 6 10 Assays of Dross (Weight%) 57 Table 6.11. Lithium Assays from Kinetic Samples 57 Table 6.12. SEM Assays of Particle in Figuie 6.12, (Weight%) 59 Table 7.1. Results of Linear Regression 60 VII) permission of the copyright owner. Further reproduction prohibited without permission. Table 8.1. Average Relative Errors in Sample Concentrations 65 Table 8.2. Comparison o f Analytical Results 66 Table 8.3. Mass Balance for Lithium 71 Table 8.4. Mass Balance for Magnesium 72 Table 8.5. Mass Balance for Silicon 72 Table 8.6. Volatility Coefficients of Various Solutes in Molten Aluminum € < Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 92 LIST OF FIGURES Page No. Figure 1.1. The Acid Process for Lithium Carbonate Production from Spodumene 7 Figure 1.2. The Silica Tetrahedral 9 Figure 1.3. Correlation Between Crystal Structure and the Strength of Tetrahedral Frameworks 10 Figure 2.1. The Contact Angle Between a Liquid and Solid 15 Figure 2.2. Surface Tensions for Aluminum Alloys at 50°C 16 to 80° C Above Liquidus Temperature 16 Figure 3.1. Free Energy o f Formation for Lithium and Potential Reductants Figure 3.2. 17 Thermodynamic Simulation of the Lithium Extraction from Spodumene by Reduction with 21 Al-Mg and Al-Ca Alloys Figure 4.1. Apparatus used to Perform the Lime-Spodumene Experiments Figure 4.2. 23 Reaction Product from Lime-Spodumene Experiments at Temperatures of 1050°C and Below Figure 4.3. Reaction Product from Lime-Spodumene Experiments at 1100°C Figure 4.4. 24 25 Reaction Product from Lime-Spodumene Experiments at 1150°C 25 Figure 5.1. Experimental C ap Assembly. 30 Figure 5.2. Experimental Apparatus used to Perform the Spodumene Reduction Experiments Figure 6.1. XRD Patterns o f the Raw Materials in this Study and Various Lithium Aluminum Silicates Figure 6.2. 37 Cumulative Size Distribution of Alpha Spodumene 38 and Transformed Material Figure 6.3 31 Microphotograph of the Received and Transformed Spodumene 39 Figure 6.4. Condensate 41 Figure 6.5. Fine Black Powder 42 Figure 6.6. Fine Black Powder Lost During an Experiment 43 x Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. £ Figure 6.6. Fine Black Powder Lost During an ExpeHment 43 Figure 6.7. Powder Residue 44 Figure 6.8. Dross 45 Figure 6.9. Metal Ingot 46 Figure 6.10 Material S'uck to Inside of the Crucible After an Experiment Figure 6.11. 47 SEM Microphotograph o f Particle in the Powder Residue Figure 8.1. 59 The Analysis o f Solid Formation in the Reduction of Spodumene for Two Reaction Paths Figure 8.2. 70 Lithium Weight% in the Ingot and Dross vs. the Magnesium to Spodumene Ratio Figure 8.3. 74 Lithium Concentration in the Residue and Very Fine Black Powder vs. Magnesium to Spodumene Molar Ratio Figure 8.4. 75 The Masses of the Powder Residue and [3 Spodumene Charge vs. the Magnesium to to Spodumene Ratio Figure 8.5. 75 Lithium Weight% in the Condensate vs. the Magnesium to Spodumene Ratio Figure 8.6. Lithium Weight% in the Very Fine Black Powder vs. the Magnesium to Spodumene Ratio Figure 8.7. 79 Magnesium Concentrations in the Ingot and Dross vs. the Magnesium to Spodumene Ratio Figure 8.10. 78 The Silicon Weight% in the Powder Residuevs. the Magnesium to Spodumene Ratio Figure 8.9. 77 The Silicon Weight% in the Dross and Ingot vs. the Magnesium to Spodumene Ratio Figure 8.8. 76 80 Magnesium Weight% in the Powder Residueand Very Fine Black Powder vs. the Magnesium to Spodumene Ratio Figure 8.11. 81 Lithium Extraction from the Powder Residue vs. the Magnesium to Spodumene Ratio Figure 8.12. 82 Lithium Recovery to the Metal Ingot vs. the Magnesium to Spodumene Ratio i XI Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 83 Figure 8.13. Lithium Distribution in the Reaction Products for each Experiment Figure 8.14. 84 Silicon Distribution in the Reaction Products for each Experiment Figure 8.15. 85 Magnesium Distribution in the Reaction Products for each Experiment Figure 8.16. 86 Rate of Lithium Recovery to the Melt vs. Time for Various Experiments Performed in the Study Figure 8.17. 87 Lithium Extraction from Spodumene from Experimental Results and F*A*C*T Calculations vs. the Magnesium to Spodumene Molar Ratio 94 Figure 8.18. Lithium Extraction from Spodumene vs. Temperature 96 Figure 8.19. Lithium Extraction from Spodumene vs. the amount of excess liquid aluminum Figure 8.20. Lithium Extraction from Spodumene vs. Pressure at 900°C Figure 8.21. 98 Pressure of 100% Lithium Extraction vs. Temperature Figure 8.22. 1 97 99 Magnesium’s Effect on the Lithium Extraction from Spodumene as Temperature Changes XI I Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 100 I. INTRODUCTION 1.1. MELT-LEACH-EVAPORATION In the Melt-Leach-Evaporation (MLE)(1) process for extracting Group IA a n d Group IIA metals, solid particulate ores or concentrates are contacted w ith an excess o f molten metal. The excess molten metal extracts volatile, metal species into solution by reducing compounds o f the species present in the ores or concentrates. The dissolved species are then vacuum refined from the excess molten metal. An advantage of MLE is that it improves the materials handling and production kinetics as compared to a powder process, such as Pidgeon type vacuum retorts. By using an excess liquid metal reductant, the activity o f the recovered species would be lowered by dilution in the excess molten metal and thus more o f the species would enter the solution. The extraction of lithium from spodumene (Li2O-Al20 3-4Si02) b y molten metallo-thermic reduction w as the subject of this thesis and is a representative example of metal extraction by MLE. The evaporation of the lithium from the excess molten metal was not examined experimentally in this thesis. 1.2. 1.2.1. LITHIUM CHARACTERISTICS AND USES O F LITHIUM METAL Lithium is the third element in the Periodic Table and it is the first of the alkali metals. Lithium was first discovered in 1817 by Arfwedson in Sweden(2) and the metal was first isolated in 1855. The first commercial use occurred when the German firm, Metallgesellschaft A.G., used lithium as a hardener in a lead alloy(2). Since then, scientists and engineers have envisioned many new uses for the metal, due to litnium ’s unique physical properties shown in Table 1.1. 1 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 1 .1 . (3) (4) PHYSICAL PROPERTIES OF LITHIUM PROPERTY DENSI TY, MP, °C BP, °C VALUE gem '3 REF. 0 . 534 3 1 8 0 .5 3 134 2 3 F I R S T I O NI Z A T I O N POTENTIAL k j / g m o l e 519 3 ELECTRON A F F I N I T Y k j / g m o l e 5 2 .3 3 CRYSTAL STRUCTURE BCC 3 3 .5 5 3 4 .3 9 3 -3 .0 5 4 3 .8 6 4 S P E C I F I C HEAT AT 25 ° C , S P E C I F I C HEAT AT MP, kj/k g STANDARD POTENTIAL AT ELECTRO. CHEM. kj/kg °C °C 0 °C, V EQUIVALENCE A - h o u r s / g Lithium’s density makes it the lightest o f all metals. Alone it can not be used as an structural engineering component due to its reactivity with air and water. However, it is a very effective alloying element, especially with aluminum. One weight percent lithium added to aluminum decreases the density by three percent and increases the elastic modulus by six percent.(5) Aircraft manufacturers and users aie excited about using lithium-aluminum alloys to increase the payload capacity, fuel economy, flight distance, and overall performance of their aircraft. These advantages will also be available for other users of lithium-aluminum alloys. Lithium’s density combined with its very high electrochemical standard potential and electrochemical equivalent make lithium an exceptional anode material for energy cells. Many variations o f lithium cells exist depending on the cathode material and electrolyte used. Characteristics o f lithium cells are: 1) Flat discharge. Constant current production with respect to voltage and impedance. 2) Longer shelf life. The lithium-manganese oxide battery available commercially can be stored for ten years at room temperature and retain 85% of its capacity. A zinc alkaline battery can be stored 2 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. for only 2-3 years(6). Storage for one year at 170°C is also possible for lithium cells. 3) Wider operating temperatures. Lithium batteries with organic electrolytes can be operated at temperatures as low as -155°C. 4) Higher voltage. Higher cell output reduces the number of cells in a battery pack by a factor of two when compared with regular zinc anode batteries 5) High energy density. The energy density of lithium batteries is 2-4 times higher than zinc alkaline batteries. 6) Virtually no self-discharge. This is due to the hermetic seal o f the battery case. These characteristics translate to the specialized use of lithium cells in cold climates as portable energy sources and in components which are activated occasionally but perform vital functions, such as back-up power for an important installation. Lithium use in batteries for radios, toys, calculators, etc, etc.. is increasing and a number of companies are 1 ( ) developing this technology . Lithium has a very large temperature range between its normal melting ^ and boiling points which, combined with its large heat capacity, make lithium a potential high temperature heat transfer medium. The use of lithium in fusion energy reactors as a blanket material is also anticipated. Lithium reacts with the neutrons produced in the fusion plasma in the following manner: Li + Neutron = He + Tritium [ii] The large amounts of energy carried by the neutrons would be transferred to the excess lithium and then removed by a heat exchanger. Tritium is a fuel for the fusion process, so the lithium blanket material has the benefit of regenerating a fuel. The future of lithium is very promising All reports are optimistic on the metal’s increased use and importance as the twenty first century approaches. < 3 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. f 1.2.2. LITHIUM OCCURRENCE Lithium is widely distributed across the earth. clays, rocks and water bodies. It occurs in soils, The average lithium content of the earth’s (OS. crust is 20 ppm ppm(8). The average concentration of lithium in seawater is 18 Presently two types of lithium deposits are economical as an ore; brines and pegmatite rocks. Eighty percent of the world’s lithium reserves are present as brines. The major lithium-bearing bnne deposits are sub-surface and include Salar de Atacama(Chile), Salar de Uyuni(Bolivia) and in Silver Peak(Nevada). Pegmatites are coarse grained igneous rocks. The lithium-bearing minerals are shown in Table 1.2. Table NAME SPODUMENE EUCRYPTITE PETALITE L EPI DOLI TE 1.2. L i t h iu m B e a r i n g M i n e r a l s FORMULA L i 0 • AL 0 • 4 S i O , 2 2 3 2 L i O -A l 0 •2 S iO 2 2 3 2, L i 0 • A1 0 • 8S i O 2 2 3 2, COMPLEX L i MICA LiALSi 0 2 6 LiAlSiO 4 L iA lSi O 4 10 O f these minerals spodumene is the most important. It hasa theoretical lithia and lithium content of 8.03% and 3.75% respectively. Major spodumene deposits are King’s Mountain(North Carolina), The Greenbush property(Australia) Bemic Lake(Mamtoba), The Bitka Pegmatite(Zimbabwe) and The Kitotolo deposit(Zaire), which is the world’s largest. The mined ore contains usually 2.5-3.5% lithia and is upgraded by froth flotation to a maximum of 7.5% lithium oxide. Other possible sources which are not economically feasible at present, /Q\ are geothermal bnnes and hectonte clays of the Western United States . Information on lithium reserves of the U.S.S.R. and China is not available.(9) 4 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.2.3. LITHIUM INDUSTRY I The largest producing and consuming nation of lithium is the United States. Most of the production is from North Carolina pegmatite ores by the Lithium Company of America and the Foote Mineral Company. These companies have agreements with South American governments to co-develop the Chilean and Bolivian brine resources. Despite prior beliefs that a lithium shortage is imminent(10)<11) the present established reserves are estimated to be 7.6 million tons for pegmatites and 14.0 million tons for brines(9? The probable cumulative demand for the period of 1983-2000 is 179,000 tons(12) o f contained lithium. The applications of lithium are changing with time. The uses of lithium during the period between 1978 and 2000 are shown in Table 1.3. percentage of the total lithium used is in metallic form. A greater This trend is expected to continue as more lithium is used in energy cells and alloys. T ab le 1.2.4. 1 .3 . USES OF LITHIUM METAL IN THE U . S . (SHORT TONS) 1 9 7 8 (12) 1 9 8 3 (13) 2 0 0 0 U3) T o t a l U se 3400 22 0 0 5600 U se in B a tte rie s 0 50 150 U se in A l-L i A llo y s 0 0 750 % of M e ta l U se 0 2 .2 7 1 5 .1 (%) LITH IU M EXTRACTION TECHNOLOGY The only commercial method of producing lithium metal today is molten salt electrolysis of lithium chloride. This process is applied to all lithium bearing raw materials and, therefore, lithium chloride is an essential intermediate product in lithium metal production. Three methods exist to treat spodumene ore for the purpose o f lithium extraction. They are 1) The Acid Process, 2) The Alkaline Process and 3) Ion Exchange. The Acid Process is the only method used commercially and a flow sheet « 5 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. is shown in Figure 1.1. The spodumene concentrate is first converted from its a phase to its more reactive p phase by heating to approximately 1050°C. The ore is then leached with hot sulfuric acid at 250°C. Hydrogen ions from the H2S 0 4 replace the lithium ions in the spodumene, thereby forming a soluble lithium sulfate and an insoluble gangue. During purification, the acid solution is neutralized with ground limestone and filtered to eliminate iron and aluminum impurities. After a preliminary purification step, hydrated lime is used to precipitate magnesium and soda ash is used to precipitate excess calcium. The solution is then adjusted to a pH of 7 - 8 and concentrated by evaporation to 200 - 250 g/litre of lithium sulfate prior, to formation of lithium carbonate via Na2C 0 3 addition. The spent solution contains about 15% of the original lithium and must be recycled. Prior to it rejoining the process at the purification stage, it is cooled to 0°C where sodium sulfate, which precipitates on cooling, is separated and sold as a by-product. I ? % Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. t SPODUMENE CONCENTRATE 1 a TO 0 CONVERSION 1 A CID ROASTING 1 LEACHING -» S I L I C A AND TAILS ! -» P U R IF IC A T IO N -» Mg (OH) 2 , C a C 0 3 I CARBONATION <- Na CO 2 3 1 GLAUBER RECOVERY <------ SEPARATION i N a SO i PRODUCT 2 4 L i CO 2 PRODUCT 3j CONVERSION TO L i C l I 1 MOLTEN SALT ELECTROLYSIS Figure 1.1. — L i METAL The acid process for producing lithium carbonate which is converted to lithium chloride for molten salt electrolysis to produce lithium metal. An interesting feature o f the flow sheet shown in Figure 1.1 is that lithium carbonate is made pnor to the production of lithium chloride. Lithium carbonate is the most widely used lithium chemical, used in aluminum production, and it serves as the raw material for the production of all other lithium cherricals. If lithium metal were to approach lithium carbonate in usage, the process could be easily modified so that the stream o f lithium bearing liquor was transformed directly into LiCl. Some o f the parameters o f the molten salt production of lithium are listed in Table 1.4. C 7 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. t T ab le 1.4. OPERATING CONDITIONS FOR I d ELECTROLYSIS CELL L iC l M aterial E le c tro ly te Tem perature (15) ( 1 6 ) L ic l % 48-50 KC1 % 6 5 .8 -5 0 °C 450-460 C e l l C u r r e n t (A) 3500 V o lta g e 4 .8 6 -8 .1 9 C urrent (V) G raphite Anode Pow er A nalysis I T heoretical, (kWh/kg L i ) 3 1 .5 -3 9 .2 T o ta l Power, (kWh/kg L i ) 1 4 0 (16) 9 9 .0 Li% The result of this process is a lithium product with a market price of $CDN 65.45 / kg(16) for a standard lithium ingot of commercial purity. tonne of spodumene concentrate costs $Cdn 450. One The concentrate has a contained lithium value of, 1000 kg/tonne x 0.0325 Li x $ 65.45/kg Li = $Cdn 2127/tonne [1.2 ] The cost analysis shows that an effective extraction method could be profitable. 1.3. ALPHA TO BETA SPODUMENE TRANSITION The reactivity of the lithia species in the spodumene can be improved by its polymorphic transformation from the stable alpha phase to the more open beta phase. The alteration of the crystal structure is not unique to the mineral spodumene and occurs for many silicates, the most well known are the five polymorphs of quartz: high and low quartz, tridymite, and high and low I 8 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. cristabolite. Polymorphism occurs when the ambient crystal structure is not the most stable phase at other temperatures and pressures. In practice, the transformation is found to occur at temperatures greater than 900°C. Above 930°C, the mechanism is by almost instantaneous nucleation with umdimensional growth of the nuclei while below 930°C only nucleation occurs (17) . The higher temperature is necessary to overcome the activation energy of the transformation, equal to 289 kj/mole above 930°C and 753 kJ/mole below 930°C(18). To better understand the effect of the a to p transformation an examination of silicate chemistry is useful. The building block of silicate chemistry is the Si04 tetrahedral, displayed in Figure 1.2. I Figure 1.2. The silica tetrahedral. Open circles represent oxygen atoms and full circle represents a silicon atom. The silicon atom in the center is covalently bonded to four oxygen atoms and is known as a tetrahedral site (T). Large crystal networks are formed when tetrahedra share a common oxygen atom. Some of the ideal structural requirements of silica are : 1) ideal bond length d(Si-O) = 1.605A 2) strain-free Si-O-Si bond angles of 140° 3) good space filling with as many bonds per unit volume (in order to attain maximum free energy per unit volume): ideal V = 15 A3. The correlation between the stability of the tetrahedral frameworks and crystal data is shown in Figure 1.3(19). i 9 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 t K 10 .PARACELSIAN 8 0 i .K EA TITE HEXACLLSIAN h .QUARTZ U 2 PYROXENES 0 -15 -10 5 0 5 15 10 20 25 A T -0 -T I °l Figure 1.3. Correlation between the difference in oxygen density (AV^), Si-O-Si bond angles (A T-O-T) and relative stability of the tetrahedral frameworks and chains of silica and A1 containing alkali and alkaline earth silicates. The frameworks and chains are less stable, the larger the distance from the point AV = 0 and A T-O-T = 0. Table 1.5(20) shows structural data for a and (3 spodumene. T a b le 1 .5 . CRYSTAL STRUCTURAL DATA FOR ALPHA AND BETA SPODUMENE T -O -T V ox a SPODUMENE 1 3 9 .0 1 6 .2 0 SPODUMENE 1 4 9 .4 2 1 .7 « 10 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A lpha spodumene’s crystal structure is much closer to the ideal than that of (3 spodumene: 1° for bond angle difference for a versus 9.4° for (3 and AV = 1.2A3 for a versus 6.7A3 for (3. ox The AT-G-T and AV ox for a spodumene corresponds to the stable pyroxene structure which is a three dimensional framework. The data for (3 spodumene give it a keatite structure, a two dimensional chain structure which may be easier to penetrate in the metallo-thermic reduction process. Another difference is that the aluminum cations in the (3 spodumene occupy T sites while alpha spodumene does not have its aluminum cations in T sites. T h e importance of this difference in the location of the aluminum atoms is that in the a phase, where less interstitial volume exists, more non-tetrahedral cations must be accommodated. In 3 spodumene, the lithium atoms have larger interstitial volumes to fit into and no interstitial aluminum cations are present. The difference in the V and the crystal structure leads to a density difference between the polymorphs. Alpha spodumene has a specific gravity of 3.150 g/cm 3 and (3 spodumene has a specific gravity of 2.400 g/cm3(3) As a result (3 spodumene is 23% less dense than a spodumene and a corresponding volume expansion occurs during the transition. The conversion of spodumene to the [3 phase from the a phase creates a more reactive raw material by: 1) Increasing the volume o f the material and thus increasing its surface area 2) Giving the lithia species more mobility in the material. 3) Weakening the crystal structure of the material. Beta spodumene appears to be more amenable to metallo-thermic reduction than a spodumene and the spodumene transition would be an effective pretreatment of spodumene for metallo-thermic reduction. < 11 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. II. LITERATURE SURVEY 1 H .l. M ETALLO-THERM IC REDUCTION PROCESSES The m ost widely known metallo-thermic reduction process is the Pidgeon Process for the production of magnesium(2l). In this process, magnesium is produced from calcined dolomite by reduction with ferrosilicon in a vacuum retort. The reaction is described by the equation: 2[CaO-MgO](s) + Si(s) = 2Mg(g) + 2 C a 0 S iO 2(s) [2.1] This process is one o f the methods used to produce magnesium commercially. Alumino-thermic reduction processes are used for the production of manganese, chromium(22)(23) and calcium(24). II.2. LITHIUM PRODUCTION BY M ETALLO-THERM IC REDUCTION Kroll and Schlechten(25) were able to reduce lithium oxide, in the presence of lime, with silicon, aluminum and magnesium as shown below: 2Li20(s) + 2CaO(s) + Si(s) = S i0 2-2CaO(s) + 4Li(g) [2.2] 3Li20(s> + CaO(s) + 2A1(S) = Al2O3'Ca0(s) + 6Li(g) [2 3] Li 0(s) + CaO(s) + xMg(s) = [2 .4] CaO + MgO + Li + (x-l)M g Their experiments had recoveries of the order of 90%. They stated that lime was essential in all the systems, except in the study using magnesium where the lime "had no chemical function" and the condensate contained magnesium impurities. Stauffer /nf.\ earned this process one step further by reducing spodumene directly with ferrosilicon (FeSi ). 90% were attained. Once again recoveries of the order of It was noted that lime was an absolutely essential reagent for the process because of the way it "tied up" the alumina and silica species in spodumene and without lime, poor recoveries were obtained. 12 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In both sets o f experiments a combination of reducing agent, lime and lithium raw material were combined in varying concentrations into a ^ briquette. The briquettes were then placed in a vacuum retort and the condensate product was analyzed for lithium. Pidgeon and Morris(27:i measured the vapor pressure of lithium in the lithium oxide-calcium oxide-silicon system They found that the vapour pressure increased with temperature between 972°C and 1025°C and was sufficient to commence commercial production(28). Fedorov and Sharmai examined the kinetics o f the reduction o f lithium aluminate by aluminum in vacuum; 3 L i 20 A l 2O 3(s) + 2 A l(s ) = 6 L i(g ) + 4 A 1 20 3(s> by utilizing a method of continuous weighing. [2.5] They concluded that a period of constant reaction rate existed and that the kinetics varied considerably with temperature. All the above studies and processes utilized powder reactants in a vacuum retort. The previous studies on lithium production were successful yet the ^ process was never successfully industrialized. Reasons why the commercial process did not succeed were; - slow production rates due to the fact that only small briquettes could be used for fear that the Li vapours would not be released. - a small demand for lithium metal at the time. - poor metal vapour condensation technology. With a MLE process, large reactors could be constructed and the process could be developed into a high yield, continuous one. The lithium may be able to be condensed as a liquid. II.3. LIME - SPODUM ENE REACTIONS In the previous studies o f lithium production by metallo-thermic reduction of lithium oxide and spodumene, calcium oxide played an important role. In view of lime’s importance to a proposed process, this section examines the literature o f spodumene-lime reactions. Various work has been done on the reaction of lime with lithium I 13 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. materials (29 30 31) ' . In the most comprehensive study of lime-spodumene sinters, Lainer and Nazirov(32), stated that the first stage o f the reaction might proceed according to the equation: Li &O A1LOj -4SiOL + 8CaO = LiI O-Al 1O5 + 4(2CaO S iO J [2 .6j and in a second stage excess lime would break up the lithium aluminate to form lithium oxide. In Chapter IV it w as investigated whether a lime pretreatment would be used in the MLE process. II.4. INTERFACIAL PHENOM ENA Solid oxides and molten metals are not materials which readily mix and are often described as non-wetting. Wetting phenomena were important in this study of the MLE process because the reactions involved were between solid oxides and molten metal and if no wetting occurred, no reaction would occur. A measure o f wetting between a liquid and solid can be described by the contact angle, 0, shown in Figure 2.1. The contact angle can be expressed theoretically by Young’s equation; ( Ysv ' YSI) COS0 = ---------------- \v where y , y o V and y o L represent the surface tension between solid and L .Y vapour, solid and liquid and liquid and vapour respectively. I 14 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. [2.71 VAPOUR VAPOUR LIQUID LIQUID Figure 2.1. The angle between the tangent of the liquid-gas interface and the solid plane is the contact angle, 0. In the above Figure, 0, is greater than 02 and as a result greater wetting will occur at surface 1. Table 2.1 shows y for selected molten metals and their solid phases compared to ylv of the same metals with their vapours. The table shows that if yLV for metal A is much larger than yLV f° r metal B then yS L for metal A will be larger that ysL for metal B. Table 2.1. SURFACE TENSIONS FOR SELECTED LIQUID METALS<34) <35> Metal Li Na A1 Ag L v *SL 398 191 914 903 30 20 122 126 Using Equation 2.7, the contact angle, 0 for a spodumene particle in contact with either aluminum or an aluminum-magnesium alloy can be compared. For both liquid metals, y would be equal. According to the relationship 15 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. between ylv and ysl established from Table 2.1, Equation 2.7 can be rewritten, n CO S0 ( ?sv ' = — — --------------- — 128) Ylv thus the contact angle would increase as yiv decreases. Figure 2.2 shows that the surface tension, yLv, of aluminum decreases with magnesium addition. As a result the contact angle between an oxide particle and molten aluminum would decrease with magnesium addition, improving the wetting. 800 i 6 0c) >* T3 700 Sn c o (A c 41 0> 600 o O 3 CO 500 400. Figure 2.2. Surface tensions for aluminum alloys at 50°C to 80°C above hquidus temperature. (35) I 16 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. « IH . THERMODYNAMIC ANALYSIS OF SPODUMENE REDUCTION ff l . l . REDUCING AGENT The Ellingham Diagram in Figure 3.1 and the thermodynamic computer program F*A*C*T [Facility * for the Analysis * of Chemical * Thermodynamics](36) developed by Bale, Thompson and Pelton were used to choose the molten metal reductant. A list of the possible reductants and their characteristics is shown in Table 3.1. The characteristics of lithium are also given. -200 u ► j o x: ►-Jin MO ru >-W 03 KO bW » w w IX -600 -800 -1 0 0 0 -12 0 00 1000 1500 2000 2500 TEHPERATURE ( K ) Figure 3.1. Free Energy of Formation for lithium and potential reductants. 17 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. T a b le 3.1. CHARACTERISTICS OF LITHIUM AND POSSIBLE REDUCTANTS PROPERTY MATERIAL Li A1 Ca Mg Si MELTING P OI NT ° C 180 660 839 648 1410 BOI LI NG P OI NT° C 1342 2467 1484 1090 2355 C O S T ( $ C / k g ) <17) 65. 45 2 9 7.0 3.6 1.3 The Ellingham diagram shows that aluminum and silicon are not suitable reducing agents at atmospheric pressure. Only calcium and magnesium could reduce lithium oxide to form lithium. However, neither is suitable as a bulk metal due to cost, reactivity or vapour pressure. Aluminum is an ideal bulk metal because of its low cost, suitable melting point and very high boiling point, 1 e , a low vapour pressure, iO '10Atm, at the anticipated operating temperature of approximately 1000°C. Since calcium and magnesium have the oxygen affinity necessary to reduce lithium oxide the reducing agent chosen was a molten mixture of aluminum with calcium or magnesium. Magnesium is superior to calcium in terms of economics. It has a lower cost $CDN 3.55/kg vs. $CDN 7.00/kg. Magnesium was also found to be thermodynamically superior. The F*A*C*T program, EQUILIB, was used to simulate the reduction o f spodumene by an aluminum alloy containing magnesium or calcium. EQUILIB predicts the most thermodynamic stable compounds that should be produced from the given amounts of reactants at a certain temperature and pressure. Thermodynamic data was obtained for dissolved magnesium, silicon and lithium in liquid aluminum and this data was added to the F*A*C*T data base and included in the simulation. Thermodynamic data was obtained for the / 07s 1) free energy of dissolution in liquid aluminum entropies(38) of the dissolved species. and 2) the absolute The data used is shown m Table 3.2. The data for calcium was from estimates by Richardson(39). The procedure outlined in the F*A*C*T DATAENTRY program was followed to create the dissolved species. Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission. Table 3.2. STANDARD FREE ENRERGY OF SOLUTION OF ELEMENTS I N AND T H EIR ABSOLUTE ENTROPIES METAL Si Mg Li Ca AG° ( k J / MOL E ) 40,145.0 -14,551.9 -24,267.2 -43,513.6 LIQUID A1 S° ( J / MOLE) - 47.241T 1 .255T + 13.440T 18.82 32.51 28.03 41.63 For the calculation, Inputs 3.1 and 3.2 below were used. approxiamated the experimental trials described later. These roughly The <A> value was varied from 0 to 10 in increments o f 1. L i20 'A l2O 3-(Si02)4 + 85.5 A1 + <A> Mg [3.1] L i20 A l20 3 (Si02)4 + 85.5 A1 + <A> Ca [3.2] The complete F*A*C*T output for the magnesium simulation is given in Appendix A. The results of the simulation are summarized in Tables 3.3 and 3.4 and Figure 3.2. When the <A> value w as zero. F*A*C*T predicted that the molten aluminum would fully reduce the silica species to form silicon. six percent of the lithium oxide in the spodumene Ten point wasreduced tolithium and the remainder was converted to lithium aluminate. Thus the task of a reducing agent would be to decrease the lithium aluminate predicted and form m ore pure lithium. Table 3.3 shows that the simulation predicted that the total number of moles of free lithium decreases, with calcium addition. Instead of reducing the lithium aluminate, the calcium was predicted to form the compound CaAl4 to occur. When magnesium is added, more free lithium is predicted This is due to magnesium reduction of lithium aluminate to form spinel and lithium dissolved m the molten metal. 70% recovery of the lithium was predicted. A t ten moles of magnesium If the amount o f magnesium was increased above ten moles more lithium would be produced. t 19 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 3.3. F*A*C*T OUTPUT OF SPODUMENE REDUCTION WITH A l- C * ALLOT AT 1 1 7 3 K AND 1 ATM. 1 _____________________________ MOLES PRODUCTS <A> A1 Li ( 1 ) L i A l O *3 I s) CaAl 4 ( 1) - A1 O ( s ) 2 3 S i (a) 0 1 0.140 0.134 0.079 0.075 1 . 781 1 .790 1.0 2.81 2.81 3.99 3.99 2 3 0.128 0.122 0.072 0.071 1 .797 1 .809 2.0 3.0 2.81 2.79 3.99 3.99 4 5 0.117 0.111 0.066 0.062 1.818 1 . 827 4.0 5.0 2.79 2.78 3.99 3.99 6 7 0.105 0.099 0.059 0.056 1 . 836 1 . 845 6.0 7.0 2 .78 2 .77 3 . 99 3.99 8 9 10 0.093 0.087 0 .081 0.052 0.049 0.048 1 . 854 1.864 1 . 873 8.0 9.0 10 . 0 2.76 2.76 2 .75 3.99 3.99 3.99 Table 3.4. F*A*C*T OUTPUT OF SPODUMENE REDUCTION WITH A l-M g ALLOT AT 1 1 7 3 K AND 1 ATM. MOLES PRODUCTS <A> [ L i ] Al Lid) L lA lO ^(s ) MqAl O ( s ) 2 4 A1 O ( s ) 2 3 Si (s ) 0 1 0.140 0 .141 0.079 0.079 1 .781 1.780 0.79 2.81 1.75 3.99 3.99 2 3 0 .142 0.278 0.080 0.156 1.777 1.565 1.79 2.21 0.42 0.00 3.99 3.99 4 5 0 .410 0 .514 0.231 0.289 1.356 1.197 2.32 2.40 0.00 0.00 3.99 3.99 6 7 0.603 0 .682 0.339 0.383 1.058 0 . 934 2.47 2.53 0.00 0.00 3.99 3.99 8 9 10 0.755 0 .823 0.888 0.425 0.463 0.499 0.820 0 .713 0 . 612 2.59 2.64 2.69 0.00 0.00 0 .00 3.99 3.99 3.99 - 1 20 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Mg Ca z o ho< er x U j MOLES REDUCTANT Figure 3.2. Thermodynamic simulation of the lithium extraction from spodumene by reduction with Al-Mg and Al-Ca alloys at 1173 K and 1 atm. i The lithium extracted from the spodumene is plotted against the respective magnesium and calcium to spodumene molar ratios., Figure 3.2 shows that magnesium is the superior reductant in this study. An aluminum-magnesium alloy was thus chosen as the reducing agent in the experimental program in this study. The drawback with magnesium as the alloying element is its vapour pressure, which is similar to lithium ’s and which would pose a contamination problem during lithium distillation and recovery. The similarity in vapour pressure may be resolved by fractional condensation o r distillation using lithium ’s much lower melting point as compared with magnesium’s, 180°C versus 648 °C. C 21 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IV. LIME - SPODUMENE TESTS IV .l. INTRODUCTION Kroll and Schlechten(26 ^stated that lime was chemically inactive in the reduction of lithium oxide by magnesium. However, Stauffer(28) found that lime was essential in the reduction of spodumene by ferrosilicon. From this information a decision whether to use a lime pretreatment in this study can not be made. A senes of lime-spodumene tests were thus performed to determine if the lithium oxide portion of the lime treated spodumene was significantly more amenable for reduction than that in p spodumene. The results o f the tests, thermodynamic analysis and common sense were used to decide whether a lime pretreatment of the spodumene was necessary in this study. IV.2. APPARATUS & REAGENTS A muffle furnace was used to heat the fire clay crucibles used to contain the reactant mixtures in the lime-spodumene tests. Temperature measurements were obtained via type K thermocouples, placed in alumina sheaths and inserted into the center of the mixtures. The apparatus is shown in Figure 4.1. The spodumene was a high grade concentrate obtained from The Tantalum Mining Co. of Canada (Tanco), Bernic Lake, Manitoba with the Tanco assays shown in Table 4.1. Table Li 0 Na 0 7.0-7.5 0 . 16 2 (t) 4.1. SPODUMENE ASSAYS 2 K 0 2 0.06-0.15 ( % ) (t> Fe O 2 0 .05 3 P 0 2 5 0.15-0.2 Assay performed by Tanco. The lime was commercial grade obtained from Jolichaud in Montreal, and was calcined at 600°C for one hour, a recommended calcining treatment*40*. 9 22 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. TEMPERATURE RECORDER i 1 • - • 1000 ALUM IN A SHEATH THERM O CO UPLE « MUFFLE FURNACE CRUCIBLE & MIXTURE Figure 4.1. Apparatus used to perform the lime-spodumene experiments. IV.2. PR O C ED U R E Two arbitrarily chosen mixtures o f lime and spodumene containing 38.4 wt% and 68.7 wt% spodumene, respectively, were heated at temperatures o f 900°C, 950°C, 1000°C, 1050°C, 1100°C and 1150°C for four hours in fire clay crucibles in a muffle furnace. cooled. The crucibles and contents were slowly The lime to spodumene molar ratios of the mixes were 4:1 and 10.5:1, respectively. The products of reaction were analyzed by X-Ray Diffraction (XRD). raw data was obtained by a 37 minute scan between 10° and 100° and then analyzed using the Phillips APD 1700(41) software analysis system. 23 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The IV.3. RESULTS OF LIME-SPODUMENE TESTS Three distinct reaction products were apparent after experiments at different temperatures. Photographs are shown in Figures 4 2 - 4.4. At 1050()C and below, a grey powder, slightly affected by the a -> (3 expansion of the spodumene, occurred for both lime-spodumene mixes. a green, friable, sinter that had expanded was obtained. For trials at 1100°C, At 1I50°C, a white, fused product was obtained. Figure 4.2. Reaction product from lime - spodumene experiments at temperatures of 1050°C and below I 24 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 4.3. Figure 4.4. Reaction product from lime - spodumene experiments at 1100°C Reaction product from lime - spodumene experiments at 1150°C. 25 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 4.2. XRD RESULTS OF LIME SPODUMENE SINTERS AND COMPOUNDS PREDICTED BY F*A *C *T T °C RATIO IDENTIFIED PHASES 950 10.5:1 SPODUMENE 10 5 0 1100 115 0 IV.4. LiAlO , C a 2 II 4:1 100 0 F *A *C *T LiAlO 2 2 Si O , CaO 4 , CaSiO 3 L i A l O , C a Si O , CaO 10 . 5 : 1 2 4 :1 If 10 . 5 : 1 II LiAlO LiAlO 2 2 4 , CaSiO 3 , C a S i O , CaO 2 ' 2 4 4:1 LiAlSi 10 . 5 : 1 Ca A1 S i O 4:1 Ca A1 S i O 10 . 5 : 1 NO S P E C I E S DETECTED L i A l O , C a S i O , CaO 4:1 Ca S i O 2 2 3 O L i A 1 0 2 , CaSi O^ 8 2 2 7 C a Si O 2 4 LiAlO , C a 2' 2 Si O , CaO LiAlO , CaSiO 7 2 ? 2 3 2 LiAlO , CaSiO 3 4 4 3 DISCUSSION The goal of the lime - spodumene tests was to examine the results from the XRD and thermodynamic analysis and using those results, to determine the necessity of a lime pretreatment of the spodumene. At 1050°C or below, spodumene was detected for all products except for the 4:1 molar ratio at 1050°C, where the lithium alumino silicate, LiAlSi 3 0 o , was detected by XRD. At temperatures at and greater than 1100°C, products containing calcium silicates and calcium alumino silicates were identified, indicating reaction between the lime and spodumene. At 1100°C these products were gehlenite (Ca2Si2A107) and lamite (Ca2Si04). Both are greenish in colour and were the cause of the green colour of the products. For experiments run at 1150°C, the white mineral, wollastomte (C aSi03) was detected by XRD. The EQUILIB program from F*A*C*T was used to predict the most thermodynamically stable products o f the lime-spodumene pretreatment. 26 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The lamite (Ca2S i0 4) and lime for the 10.5:1 molar mix at all temperatures. For the 4:1 mix, lithium aluminate (LiA102) and wollastonite (CaSiOp were f predicted. In addition F*A*C*T predicted that gehlenite (Ca A1 SiO ) would Z Z / have a high activity in the system, but not high enough for it to form. The XRD analysis verified the F*A*C*T calculations by detecting, gehlenite, larnite and wollastonite. with work by Nazirov The F*A*C*T prediction of lithium aluminate agreed nil The reaction products predicted by F*A*C*T were the same for all temperatures. However, as mentioned above, three different experimental products were observed. This was due to the very slow kinetics at lower temperatures and the activation energies necessary to commence reaction ( 1 ON . No simple method was available to separate the lithium containing species from the calcium containing species in the products. Thus it threw into question the desirability o f the lime-spodumene sinter product over (5 spodumene as a feed material for molten metallo-thermic reduction. The salient information is as follows: 1) The pretreatment did indeed break up the spodumene structure and free lithium oxide species to either form the aluminate or remain as an | oxide. Neither of these species were identified in the XRD study. 2) The lime pretreatment diluted the lithium content in the material. 3) As well, the addition of lime introduced calcium into the system which was shown in Chapter III to have a negative effect on the thermodynamics of the system. 4) The thermodynamic study in Chapter III shows that the reduction of the spodumene with an aluminum-magnesium alloys is feasible without a lime pretreatment. It was concluded that a lime pretreatment was not nescessary and furthermore that lime’s importance in the powder experiments performed by Stauffer was due to the passive oxide layer on the surface o f the powder reductants. In those experiments, lime was necessary as a flux to decompose the oxide layer and expose free metal for reaction. Since the reactions were performed in vacuum, no further oxidation, aside from the formation of reaction products, occurred. With the MLE process, molten metals, free of oxide surfaces, would be the reducing I 27 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. agent, so the fluxing properties of lime would not be necessary. IV.5. CONCLUSIONS; LIME - SPODUMENE TESTS. The lime pretreatment of spodumene was not considered advantageous for the MLE process for the following reasons; 1) Dilution of lithium in feed material for reduction. 2) Thermodynamic feasibility of system without lime. 3) No distinct disassociation of the lithia species from alumina and silica. As a result, the reduction of spodumene was attempted without lime pretreatment in this study. T i 28 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V. EXPERIMENTAL V .l. OBJECTIVE The aim of the experimental program was to determine the effect of changing the magnesium concentration of an aluminum-magnesium molten reductant, on the recovery of lithium to the molten phase, via spodumene reduction at 1 Atm. and 1173 K. V.2. EXPERIMENTAL VARIABLES The process variables which were controlled during the experiments were: a) melt temperature b) A r gas flush rate c) stirring mechanism and rate d) reaction time e) amount o f excess molten aluminum The process variable which w as altered was the amount of magnesium added to the system, calculated as the molar ratio of magnesium to spodumene. V .3. M ATERIALS Three charge materials were used; a) high grade spodumene concentrate supplied by Tanco and converted in house to (3 spodumene. b) aluminum, commercial grade supplied by ALCOA, Pittsburgh, Pennsylvania with the specifications shown in Table 5.1. Table 5.1. S P E C I F I C A T I O N S OF ALUMINUM( + 1WEIGHT% Si Fe C Mn Mg Zn Ti V 0.033 0.029 0.004 0 .001 <0 .001 0.005 0.004 0.012 ( t) Assay performed by emission spectrometry at McGill University. 29 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. c) magnesium, commercial grade supplied by Timminco, Haley, Ontario. V.4. APPARATUS The experimental apparatus used to perform the reduction experiments is shown in Figures 5 .l.and 5.2. I Figure 5.1. Experimental cap assembly. V ------------------30 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. DRIVE CHAW STIRRING DRIVE MOTOR SHAFT SUPPORT IMPELLER SHAFT ARGON INLET EXHAUST GAS CRUCBLE CAP INDUCTION FURNACE IMPELLER ALUMNA CRUCBLE IMPELLER DRIVE & SUPPORT VEHICLE Figure 5.2. Schematic of experimental apparatus used to perform the spodumene reduction experiments. C -------------------------------31 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Below is a description of the equipment used. m Induction Furnace: 100 kW, Bradley Controls: tilting capabilities, removable crucibles, gas ventilation. Crucible: Bonded alumina, 28.8 cm high , 12.8 cm in diameter, supplied by Engineering Ceramics, Gilberts, Illinois. Impeller: For Experiments 1-3: stainless steel blades, 11.8 cm blade to blade diameter, 1.0 cm blade thickness, 11 cm blade height, welded onto 1.90 cm diameter shaft. For experiments 4-8: high temperature refractory cement bonded onto stainless steel mesh reinforced blades, 11 8 cm blade to blade diameter, 1 cm blade thickness, 11 cm blade height with a 2.54 cm shaft diameter ] Impeller shaft: Machined to 1.58 cm diameter at top to fit into brass connector. Experiments 1-3: 1.90 cm diameter 36 cm height. Experiments 4-8: 2.54 diameter, 36 cm height. Brass Connector: Hexagonal shaped, 2.54 cm from edge to edge, 1.58 cm hole drilled in underside to accommodate impeller shaft, three 0.635 cm threaded holes for screws to secure impeller shaft, threaded at top to accommodate drive shaft. Impeller Drive and Support Vehicle: Movable in any direction, raising and lowering of steel angles attached to shaft support, ball bearing attachment of drive shaft to shaft support, two levels of storage. I 32 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Thermocouple: Handheld Omega type K thermocouple probe with 30.48 cm long, with 0.3175 cm tip diameter coated in clay protection. Motor: Tigear two speed motor, sprocket and chain system used to drive shaft at 36 or 60 rpm. Gas: Liquid Air prepurified 99.998% A r gas. Less than 10 ppm water vapour, less than 10 ppm oxygen. Flowmeter: Flowrate controlled by Gilmont flowmeter, size no. four, with a ball diameter of 0.9525 cm, ball mass of 1.142 g and ball density of 2.53 g cm"3 Sampler: ^ Steel crucible, 1.4 cm outer diameter, 1 cm inner diameter, Steel wire, 0.1 cm diameter, spot welded onto crucible side for lowering into the melt through viewport. Crucible Cap: Mild steel, 17.1 cm in diameter, 1.2 cm thick, a groove machined into underside provided a fit between cap and crucible, two 1.59 cm holes drilled through cap, one for gas exhaust and other for viewport / temperature measurement / sampling, one 1.59 cm threaded hole for Ar gas intake, 1.905 cm copper tubing leading from gas exhaust to ventilation system, 0.635 cm copper tube laid over and soldered to surface with connectors leading to PVC hoses for water intake and outlet. Experiments 1-3: 1.90 cm hole drilled in center, Experiments 4-8: 2.54 cm hole drilled in center. 33 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V.5. EXPERIM ENTAL PROCEDURE FOR SPODUMENE REDUCTION TESTS f V.5.1. PREPARATION O F EXPERIMENTS Prior to the reduction experiments the a -> P treatment of the spodumene was performed. The as received a spodumene was heated at 1050°C for three hours in the muffle furnace in fireclay crucibles similar to the ones used in the lime-spodumene tests. When a new crucible was used in the reduction experiments, the crucible was installed so that its nm was parallel to the ground. This was done to ensure that the impeller entered the crucible at an angle of 90°. The crucible was surrounded by coarse silica particles which provided packing and insulation. The silica was covered with cement to contain it during the casting of the reduction experiment products cement to A spout wasmolded from the provide a channel to improve pouring. Prior to a reduction experiment, the impeller was coated with an alumina cement and allowed to dry for 24 hours. After which, it was heated to 900°C for 5 hours to eliminate moisture in the alumina paint and to strengthen the J bonding o f the paint. V.5.2. EXPERIM ENTAL PROCEDURE Charging consisted of placing pieces o f pure aluminum into the crucible. The power was set at 35 kW. and melting, which varied with the shape of the aluminum pieces, due to different efficiencies in energy utilization, took about ninety minutes. Once molten, the temperature of the aluminum rose very quickly. At a temperature of 900°C, magnesium was added in cut blocks by dunking with tongs. Black oxide believed to magnesium oxide was visible on the surface of the aluminum as dissolution was occumng. This oxide formation decreased the efficiency of the experiments as it consumed magnesium. Following magnesium addition, (3 spodumene was added to the top of the melt. The impeller and cap weie then attached to the cart and moved into place and the water cooling and argon gas flow w'ere The impeller was lowered slowly into prevent thermal shock connected the melt, by the manual winch, to As the impeller was lowered, it was turned slowly by 34 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. hand to ensure that no contact of impeller and the crucible wall occurred (i.e., that the impeller was centered properly). The immersion of the cold impeller extracted a great deal of heat from the liquid metal. Once the temperature again reached 900°C, stirring and argon gas flow began. This marked the starting point of the reduction experiments. Temperature measurements sampling were taken every 20 minutes for one hour. The temperature was held at 900 ± 20°C. During a temperature measurement, the impeller was stopped, the temperature measurement hole uncovered and the thermocouple probe inserted into the melt. Sampling was performed every twenty minutes or at every second temperature measurement. The sampler was lowered into the melt through the temperature measurement/sampling hole to obtain the samples. Ninety minutes after the beginning of the experiments stirring was stopped and the impeller and cart apparatus were removed. The melt was allowed to cool to approximately 750°C pnor to pouring the entire contents of the crucible into a mold. Some molten, viscous material remained inside the crucible and was scraped out of the crucible into the mold. A thin metal skin was removed from the impeller after each experiment and added to the products. The removal of this metal stripped the alumina paint and some cement from the impeller. As a result the impeller was touched up with cement and repainted prior to an experiment. i 35 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. V.6. EXPERIM ENTAL PROGRAM The experimental program consisted of eight experiments listed in Table 5.2. Table 5.2. EXP. 1 2 3 4 5 6 7 8 SPODUMENE REDUCTION EXPERIMENTS PERFORMED REACTANTS <g) MOLES REACTANTS A1 Mg SPOD AL Mg SPOD 2340.5 2291.0 2355.8 2410.3 2439.8 2220.0 2280.5 2269.8 97.6 0.0 104.0 47.6 143.3 191.9 220.6 224.8 365.7 343.8 369.0 368.6 370.2 361.8 342.1 348.4 86.7 84.8 87.2 89.3 90.3 82.2 84.5 84.1 4.1 0.0 4.3 2.0 6.0 8.0 9.2 9.4 0 .98 0 .92 0 .99 0.99 1.00 0.97 0 .92 0.94 Mg/ SPOD mole ratio 4.2 0.0 4.3 2.0 6.0 8.2 10.0 10.0 1 1 36 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. VI. RESULTS i VI.1. SPODUMENE TRANSITION The transformation of alpha to beta spodumene during the treatment resulted in a volume expansion. The crucible contents were loosely packed prior to the treatment but afterwards they were densely packed together. In some cases the pressure from the expansion was so great that the crucible cracked. The XRD patterns of the a spodumene and of the transformed material are compared with the patterns o f a and (3 spodumene in Figure 6.1. - io 1 5 .0 0 4 .0 0 3 .0 0 A £-00 1 .0 0 11 -i—4-J- lL.i. J l . l J 1 1- . .J i 4 0 .0 6 0 .0 8 0 .0 100.0 J_L L 4^.0 jJ U . 6.0 .0 8 0 .0 100.0 2 0 .0 4 0 .0 00 .0 8 0 .0 100.0 2O.0 4 0 .0 tO . 0 8 0 .0 100.0 2 0 .0 B uu_ 1O0.0 3 0 .0 6 0 .0 4 0 .0 £0 . 0 II l i 1.1 ?0 . 0 .10J 1 .00 0 .3 0 0 .6 0 0 .4 0 0 .2 0 - D 144- i l-i. 1 00.0 3 0 .0 6 0 .0 4 0 .0 20.0 Figure 6.1. XRD patterns of the material received from Tanco(A), a spodumene(B), the transformed material(C) and P spodumene(D). < 37 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The difference in the patterns between the a spodumene and the transformed material and the resemblance between the patterns of the transformed material and P spodumene indicated a successful pretreatment. The cumulative size distribution of the feed material was altered by the treatment as shown in Figure 6.2 and microphotographs o f a and p spodumene are shown in Figure 6.3. 100 ALPHA BETA CO 3CL 5s IU > 53 2 3 O -38 +58-75 +106-150 +212-300 +38-58 +75-106 +150-212 +300 SIZE FRACTION (MICROMETERS) Figure 6.2. The cumulative size distribution of the received alpha spodumene and of the transformed material. Although the spodumene expanded during the treatment, the fifty percent passing size of the transformed material was finer due to cracking and subsequent breaking of the particles during the a to P conversion. Sixteen point two percent of the p spodumene compared to only 10.8% of the a spodumene was finer than the smallest size fraction, -38 microns. f 38 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 6.3. Microphotographs of the as received (above left) and transformed spodumene. The a to {J conversion altered the crystal structure of the material and the forces involved were great enough to cause cracks in the particles. VI.2. VISUAL OBSERVATIONS OF REDUCTION PRODUCTS 4 Five different materials were identified at the end of an experiment. They were: -a white condensate -a very fine black dust -a powder residue -a dross -a metal ingot. The material scraped out of the crucible at the end of an experiment and the metal skin on the impeller were included in the dross. Photographs o f the products are shown in Figures 6.4 to 6.10. The masses o f each of the products and the total mass recovered compared to the initial mass o f reactants is presented in Table 6.1. i 39 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. T ab le 6 . 1 . PRODUCTS OF SPODUMENE REDUCTION EXPERIMENTS 1 EXP 1 2 3 4 5 6 7 8 REACTANT MASS( g ) 2803 . 8 2634 . 8 2828 . 8 2826 . 5 2953 .3 2782 .7 2843 .2 2843 .0 PRODUCT MASS( g) METAL INGOT DROSS 881.0 2062.5 2099.1 2098.1 2110.1 2056.1 2299.7 2201 0 133 6.3 90.4 157.9 133.7 225.0 178.3 117.1 160.9 POWDER RESI DUE 415 8 415 . 9 477 . 1 433 . 8 455 .7 389.3 361.6 467 .2 FINE POWDER COND. 1 .12 0 .00 1 .03 0 . 00 0.35 0 . 47 1 .35 0 .08 3.00 0.00 1.05 11.80 5.93 20.22 4.82 8.28 TOTAL %DI FF 2264 .2 2568 .8 2735 .1 2665 . 9 2791.1 2624 .2 2779.8 2828 .2 -19.0 -2 . 5 -3.3 -5.7 -5.5 -5.4 -2.2 -0.5 The products were complex and the mass balance was complicated by material losses such as particle emissions from the crucible and adherence to the crucible walls. The % difference of input versus output masses, calculated as, M OUT - M 16. 1] IN 100% M. IN had an average value o f -5.5% that was considered acceptable and indicates that by and large most of the reaction products were collected. During the experiments, a white plume was evident emerging from the exhaust tube o f the reactor. Numerous attempts were made to photograph this vapour, but were unsuccessful. Some vapour did condense on the underside of the cap and in the exhaust tubes, as shown in Figure 6.4. fineness, -75 microns, gave it a soft texture. The condensate’s Condensate was collected for each experiment by scraping the underside of the cap and the inside of the tubes. No condensate was observed for Experiment 2, in which no magnesium was used. 40 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 Figure 6.4. Condensate 41 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 6.5. Fine Black Powder A fine black powder, -75 microns, was produced during the reaction and portions were carried into the exhaust tubes by the argon flushing gas where it settled in the horizontal portion of the tube. During temperature measurements and visual observations of the melt, gas escaped the reactor from the temperature measurement port and carried some fine powder with it. The fits between the cap and crucible and between the impeller shaft and cap were not hermetic, thus permitting the escape of the flue powder. illustrates flue powder losses around the reactor. Figure 6.6 Varying amounts o f flue powder were collected from each experiment but the quantities lost, prevented accurate determination of the total amount produced. 42 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 6.6a, 6.6b. Fine black powder lost during an experiment The amount of fine black powder lost varied among tests. out o f the reactor and onto the cap or furnace. around the crucible. It was carried Figure 6.6a shows the losses Figure 6.6b shows the losses on the cap. During temperature measurements particles would spout out of the reactor and onto the cap surface. These particles consisted mostly of fine powders but occasionally some small droplets of molten metal would emerge. The above examples represented experiments with extreme loses o f fine black powder. i 43 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 6.7. Powder Residue The powder residue contained unreacted spodumene, solid reaction products of the reduction, bits o f impeller and minute pieces of aluminum separated from the main body o f liquid metal floating/resting on the melt surface. After the experiments, it was It was separated from the other products after the entire contents of the reactor were poured out into a mold. In the above Figure of the powder residue from Experiment 3, one can see white, unreacted spodumene particles, powder agglomerates, brown powder assumed to be a reaction product, and pieces of metal. The powder residue was a melange of matenals of many different sizes. ! 44 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1 Figure 6.8. Dross T he dross recovered was present on the melt surface and possibly along the crucible walls. It was composed of irregularly shaped metal clumps with highly oxidized surfaces. Some dross had cavities containing powder residue. All the dross had some powder attached to its rough surface. Any powder that was shaken off the dross was added to the powder residue but some remained in cavities, as inclusions, or embedded in the surface. As shown in Table 6.1, the masses of the dross varied only slightly, except for Experim ent 1, where 1338.3 g was formed as a result o f impeller problems in that experiment. < 45 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 6.9. Metal Ingot The metal ingot was the largest contributor to the total mass of the products, except in Experiment 1 A small amount of powder residue and silica induction furnace packing, which spilled during pouring, was embedded into the ingot’s surface. The ingot from Experiment 2, in which no magnesium was added, was shiny. For experiments where more magnesium was added, the surface of the ingot was dulled due to increased oxidation The above ingots were from Experiment 3 on the left and Experiment 2 on the right. The ingots were 28 cm long and 8 cm wide. V 46 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 6 10. Material stuck to inside of the crucible after an experiment At the end of each experiment, a layer of material was attached to the •J inner surface of the crucible as shown m Figure 6.10. rem ove this layer It was impossible to Since a new crucible was not used for each experiment the buildup present after an experiment varied. Figure 6.10 shows the buildup after Experiments 7 and 8 VI.3. XRD ANALYSIS O F REDUCTION PRODUCTS X Ray Diffraction (XRD) analysis of the non-metallic samples was performed to determine the materials present in these samples. The XRD analysis system was the same as was used for the the lime - spodumene experiments All the species which were matched by the Phillips system are listed by name, compound formula and identification number. Those that "scored high enough" to be positively identified by the analyzing software * are noted by an asterisk . The powder residue was analyzed and the following species were identified: C 47 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. E xperim ent 1: synthetic spinel , M gAl.,0 (21-1152), silicon (27-1042), aluminum (4-787), (5-565), silicon low quartz (33-1161), and synthetic iron (6-696). Experim ent 2: spodumene , LiAlSi 2 O ti (35-797), aluminum (4-787), lithium ------------------------------------- alumino silicate, LiAl(SiO )., (31-706), and synthetic iron (6-696). Experim ent 3 : silicon * (5-565), silicon (27-1042), aluminum (4-787), and synthetic spinel , MgAl 2O4 (21-1152). Experim ent 4 ; spodumene * (35-797), lithium alumino silicate (35-794), LiAlSi3Og, FeAl20 4 (35-794), Mg„Al20 4 (33-853), aluminum (4-787), synthetic spinel , MgAl.,0 (21-1152), lithium alumino silicate, LiAl(SiCy2 (31-706), synthetic iron (6-696). * (4-787), synthetic penclase , MgO(4-829), silicon * (5-565), silicon (27-1032), synthetic iron (6-696), lithium Experim ent 6: aluminum * alumino silicate, LiAl(SiO ),, (31-706), synthetic spinel, MgAl20 4 (21-1152), spodumene (35-797) % Experim ent 7 : spodumene (35-797), silicon (5-565), silicon (27-1042), aluminum (4-787), synthetic penclase, MgO (4-829), synthedc iron, (6-696), lithium alumino silicate, LiA l(Si02)3 (31-706). Experim ent 8 : The powder residue greater than 20 Mesh; silicon (5-565), silicon (27-1042), synthetic spinel , MgAl O (21-1152), ^ 2 4 synthetic penclase , M gO (4-829) and aluminum (4-787). Powder residue less than 65 Mesh; silicon (5-565), silicon (27-1042), The black powder was analyzed for cenain expenments and the following species were identified. Experim ent 4 ; aluminum * (4-787), silicon ★ (5-565), silicon (27-1042), synthetic iron (6-696). 48 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. E xperim 5: aluminum (4-787), silicon ( 5-565), silicon (27-1042), — ——— ent _____ ^ synthetic spinel , MgAl20 4 (21-1152) and synthetic iron (6-696) E xperim ent 6: synthetic periclase , MgO(4-829), hercynite ,FeAl2 0 4 (34-192). --------------------------- The condensate was analyzed for Experiment 7 and the following species were identified: * synthetic periclase , MgO(4-829), aluminum (4-787), synthetic iron, (6-696), The detection of the synthetic iron in many o f the samples was caused by the iron backed sample holder which was utilized in the XRD apparatus. The detection of hercynite in the Expenment 6 black powder was due to it being an iron spinel and having a similar pattern to other species present. The XRD analysis showed that the powder residue was a complex material. It was composed of aluminum, spinel, silicon, spodumene, lithium alumino silicates and penclase The flue powder was composed of aluminum, silicon and spinel. All these constituents were present in the powder residue and may have been a part of the fine sized portion of the powder residue, which was physically transported from the reactor by gases The condensate was formed by the condensation and subsequent oxidation of magnesium. Some aluminum was detected but due to aluminum’s low vapour pressure at 1173 K, 10‘10 Atm, it was unlikely that aluminum vapourized and condensed. VI.4. A TO M IC ABSORPTION ANALYSIS Atomic Absorption (A.A.) was the method o f analysis used to perform numerous chemical assays tor silicon, magnesium, and lithium. common and versatile of the analytical methods for metals. It is the most I* is a sensitive analytical method and elements whichhave enhancement or depression effects must be known Other factors which may cause spunous results are differences in the composition between effects. standards and samples, known as matrix As well, the viscosity o f the solution will affect the reading as it affects the aspiration rate of the flame. With the knowledge of the A.A. characteristics for each element, one can decide how to prepare standard and 49 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. sample solutions to obtain the best results. A technique used to counter interferences from the elements in solution is masking the sample solutions by the standard solutions. masking is as follows An example of A solution being analyzed for element A contained b amounts of element B known to affect the absorbance of element A. To ensure that accurate results were obtained, the standards were made so that they contained b amounts of B as well. This would "mask" the presence of element B in the solution and minimize its effect on the calculated concentration of A All reaction products of the reduction experiments were analyzed for lithium. Analysis for magnesium and silicon was done whenever it was thought necessary. VI.4.1. SAMPLING FOR ATOM IC A BSORPTION ANALYSIS The following sampling method was used to obtain ingot samples: Drillings were made at various locations on the ingot at various depths. The shavings were mixed together to accumulate a one gram sample. This was done to avoid segregation. For the dross samples, small bits were cut off various dross clumps to form a one gram sample. A representative sample of the powder residue was obtained by separating the powder into observable groups by hand such as fused or sintered powders, impeller bits and the rest, massing each part and then adding each component according to its mass fraction. The sampling method for the flue powder and condensate was collection by scraping or brushing from their respective sources. VI.4.2. A TOM IC ABSORPTION ANALYTICAL METHODS Dross and Ingot Assays: One gram samples of ingot and dross samples were dissolved in a 40 ml aqua regia solution containing 10 ml nitric acid and 30 ml hydrochloric acid. Ten milliliters of water was added to the beaker prior to contacting the samples with the acid in order lessen the violence of the reaction between samples and acid The reaction between magnesium and hydrochloric 50 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. acid was extremely explosive. precautionary procedure. The acids were also added in stages as a The liquor was diluted to 250 ml with distilled water. For lithium analysis, the solutions were further diluted four times so that the solutions contained, lg / 250 ml x 0.25 = O.lg / 100 ml [62 ] of sample, which can be converted to ppm by the following equation; 1 ppm = 0.001 g / 1000 ml [6.3] Therefore the solutions contained 1000 ppm of sample; the majority being aluminum. For lithium analysis, aluminum causes depression of the absorbance(42). Standards were made to mask the solutions so that they contained 1000 ppm aluminum as well. Another element present in large quantities was magnesium, which has no effect on lithium readings(43). Silicon was another element present in the ingot and it too has no effect on lithium readings(43). S % A.A. machine settings for lithium are shown in Table 6.2. The These settings were also used for the lithium analysis of the other experimental products. Table 6.2. SETTINGS FOR L i A . A . PARAMETER S ETTI NG CURRENT VOLTAGE 8 mA 460 V BANDWIDTH 0.5 WAVELENGTH BURNER 6 7 0 . 8 nm air-acetylene ANALYSIS nm For magnesium analysis, the liquors produced for the lithium analysis were used as the stock solution and diluted 100 times, which resulted in a background concentration of 10 ppm aluminum. I Magnesium readings are sensitive to depression effects by lithium, silicon and aluminum. As well, 51 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the acids used for dissolution have a large effect on the A.A. readings. The concentrations of the other elements were estimated and masking of the solution by standards was attempted. The absorbance readings of the masked magnesium standards were erratic and a proper calibration curve could not be obtained. Thus masking was unsuccessful. A different technique was utilized. A chemical agent, strontium nitrate, was added at concentrations o f 2000 ppm to the solutions and standards for the purpose o f canceling the effects of other the dissolved elements(4J) This was found to be successful. The analytical parameters for the magnesium analysis are shown m Table 6.3 Table 6.3. SETTINGS POR Mg A . A . ANALYSIS PARAMETER VALUE CURRENT VOLTAGE 4 mA 530 V BANDWIDTH 1 nm WAVELENGTH BURNER 2 8 5 . 2 nm air-acetylene I Silicon does not dissolve in aqua regia, thus new ingot and dross samples were dissolved in the following manner: A one-tenth of a gram samples from drill shavings were added to polyethylene beakers containing 10 ml of water and 15 ml of hydrochloric acid. Fifteen milliliters of hydrogen peroxide were then added in small portions. After cooling, two milliliters of hydrofluoric acid were added. The liquor was filtered and diluted to 100 ml with distilled water. The background concentration was; 0 .lg / 100 ml = 1000 ppm (64] Aluminum depresses the silicon readings as it does lithium readings and the same masking strategy used for the lithium analysis was employed for the silicon analysis. The A.A. settings for the silicon analysis are shown in Table 6.4. i 52 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. T able i SETTINGS FOR S i A . A . 6.4. PARAMETER VALUE CURRENT VOLTAGE 12 mA 5 30 V BANDWIDTH 0 . 32 n m WAVELENGTH BURNER 2 5 1 , 6 nm air-nitrous ANALYSIS oxide Condensate Assays: The condensate was analyzed quantitatively by atomic absorption. An amount of approximately one gram was dissolved in 15 ml o f slightly heated nitric acid. The condensate absorbance was compared to lithium standards. The standards contained no magnesium, since it has a neglible effect on lithium absorption as mentioned earlier. B lack Pow der Assays: I About one gram of black powder was dissolved into an aqua regia solution containing 5 ml HNC>3 and 15 ml HC1. Afterwards it was analyzed for lithium and compared with pure lithium standards. A black residue remained. Pow der Residue Assays: The preferred method of analysis for oxide powders is X-Ray Fluorescence (XRF) This is a x-ray spectrographic method. excitation of the atoms occur The sample is heated so that X-ray photons are emitted with an energy proportional to the atomic number Z of the element, detected and counted to determine the chemical composition of the sample. However, lithium can not be analyzed for because its x-rays are too low in energy to pass through the beryllium window used between the chamber and the x-ray detector. platinum crucible is used to contain the sample. In XRF a The powder residue contained pure silicon and pure aluminum which would react with and destroy a platinum crucible. £ Therefore X RF was eliminated as an analytical method and A.A. was used to assay the powders. 53 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The residue samples were sent to Tanco where they were analyzed for lithium, sodium, potassium, and aluminum. The liquors were sent back to McGill to be analyzed for silicon and magnesium. For magnesium analysis, the liquors were diluted by 100 times and strontium nitrate was added to cancel effects of other elements. Analysis of the Tanco liquor for silicon was impossible due to the method used to the prepare the liquor. Thus, silicon analysis of the powder residue was performed at McGill. A one gram sample of powder residue was added to a polyethylene beaker. One hundred and fifty millilitres of hydrochloric acid and 50 ml of nitric acid were added to it. After the reaction subsided, 30 ml of hydrofloric acid was added to the solution. The beaker was heated to 90°C by placing it a water filled metal tray, heated by a hot plate. After reaction ceased, the solution was cooled to 1000 ml and the volume was made up with the addition of distilled water The samples were compared against standards containing 40 ppm aluminum. VI.4.3. T * ATOMIC ABSORPTION RESULTS An example of data generated from a set of lithium standards with 1000 ppm aluminum is given in Table 6.5. The error was calculated as the standard deviation of the absorbance readings and is also shown in Table 6.5. I (x -x ) 2 n 54 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Table 6.5. A . A . RAW DATA FOR A SET OF L i STANDARDS AND INGOT ASSAYS STANDARD ppm Li 0.5 1.0 2.0 3.0 4.0 ABSORBANCE (x) ERROR( s ) MEAN (x) 0.074 0.146 0.268 0.421 0.559 0.087 0.157 0 .291 0.428 0 .577 0.088 0.162 0.296 0.436 0.083 0 .155 0 . 285 0 . 428 0. 568 0.008 0.008 0 .015 0.008 0 .013 0.167 0.002 0.073 0.217 0.291 0.395 0.403 0.382 0.232 0.003 0.070 0.221 0.302 0.406 0.406 0 .407 0.240 0.003 0.072 0.217 0.311 0.404 0.409 0.418 0 .213 0 . 003 0 .218 0 .072 0 .301 0 .402 0 .406 0 .402 0.040 0.001 0.002 0.002 0.010 0.006 0 .003 0.018 I NGOT SAMPLE 1 2 3 4 5 6 7 8 From the absorbance data the concentration and errors of the solutions and samples were determined utilizing the method outlined in the examplecalculations in Chapter VII. The calculated weight perce ~ and the corresponding relative errors of the products analyzed for are presented in Tables 6.6 - 6.10. The relative errors, AC, are discussed in Section ViII.2 3. Table ASSAYS OF CONDENSATE (WEIGHT%) M g : SPOD RATI O MASS 1 2 4.2 0 .0 1.12 0.00 0.42 - 0 .033 - 3 4 4.3 2.0 1.03 0.00 0.57 0.55 0 .053 0.023 5 6 6.0 8.2 0.35 0 .47 0.94 0 . 68 0.045 0 .056 7 8 10.0 10.0 1.35 0 .08 0.26 0 .20 0 .006 0 .007 EXP « 6.6. (g) C Ll AC Ll (Wt%) 55 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. T a b le 6 . 7 . ASSAYS OF FLUE POWDER (WEIGHT %) EXP M g : SPOD RATIO C AC C Ll (Wt%) u I (g) AC Mq <Wt%) Mg 1 2 4.2 0.0 3.00 0.00 3 4 4.3 2.0 1.05 11.80 1.27 0.050 20.11 0.024 5 6 6.0 8.2 5 . 93 20.22 1.18 1.22 0.056 0 .042 24.40 33.10 0.031 0.026 7 8 10.0 10.0 4.82 8.28 1.22 1 . 12 0 .059 0.034 39.97 35.45 0.028 0.028 MASS ASSAYS OF POWDER RESIDUE C 1 " Mg: SPOD RATIO 6.8. o T able EXP MASS (g) C A_ <20 C Na 2 O C (WEIGHT %) S l AC si (Wt%) C Mg AC Mg (Wt%) 1 2 4.2 0.0 415.8 4 15 . 9 1.56 2.53 33.2 31. 6 0.08 0.11 0.18 0.21 8.48 15.28 1.52 1.69 16.90 - 6.140 - 3 4 4.3 2.0 477 . 1 433.8 1.24 1.89 38.1 33. 6 0.07 0.09 0 .12 0.16 4.73 12.47 0.50 2 .25 13.26 9.08 1.260 6.140 5 6 6.0 8 2 455.7 389.3 1.07 1.06 45.7 37 2 0.10 0.09 0 .13 0.16 4.88 7.28 0.59 0.80 16.84 20.00 1.410 1.410 7 8 10.0 10.0 361. 6 467 . 2 1.08 0.83 32.7 41 2 0.11 0.08 0.30 0.12 9.36 6.41 0.76 0.57 122.20 19.48 - I 56 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 1.260 T able EXP M g : SPOD RATI O 6.9. ASSAYS OF INGOT C MASS (g) Ll c Ll (Wt%) 1 Mg AC C Mg (Wt%) S1 AC , si (Wt%) 1 2 4.2 0.0 881.0 2062.5 0.15 0.02 0.034 1 .39 - 0. 15 9 - 3.31 2 .72 0.59 0.34 3 4 4.3 2.0 2099.1 2098.1 0.15 0.05 0.008 0.008 1 .50 0 .51 0.220 0.178 3.82 1.54 0.60 0.43 5 6 6.0 8.2 2110.1 2056.1 0.21 0.28 0.013 0.011 1 .57 4 .71 0.164 0 . 2 62 2.32 3.74 0.16 0.75 7 8 10.0 10 . 0 2299.7 2201.0 0.29 0.28 0.008 0.019 4 .69 3 .84 0.187 0.201 3 .37 3.03 0 . 47 0.54 Table EXP ? AC (WEIGHT %) M g : SPOD RATI O 6.10. MASS ASSAYS OF DROSS C (g) Ll AC C L1 (Wt%) (WEIGHT %) Mg AC Mg (Wt%) __ C S1 AC Sl (Wt%) 1 2 4.2 0.0 1336.3 90.4 0.12 0.01 0.007 0.001 1.26 - - 2.42 1.63 0.37 0.24 3 4 4.3 2.0 157.9 133.7 0.17 0.19 0.011 0.007 2 . 05 1.11 0.438 0.469 2.36 1.93 0.26 0.33 5 6 6.0 8.2 225.0 178.3 0.24 0.34 0.001 0.027 2.59 4 . 90 0.800 0.706 1.75 2.27 0.37 0.32 7 8 10 . 0 10 . 0 117 . 1 160.9 0.39 0.34 0.037 0.012 6 . 60 6.81 0.541 0.823 2.68 6.39 0.49 0.85 Table 6.11. LITHIUM ASSAYS FROM KINETIC SAMPLES Mg:SPOD Li ASSAY IN SAMPLE RATIO 20 MIN 40 MIN 60 MIN 4 2.0 0.05 0.09 0.04 5 6.0 0.14 - 0.22 6 8.2 0.21 0.29 0.31 8 10.0 0.25 0.26 0 .30 EXP (Wt%) I 57 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. VI.5. SCANNING ELECTRON MICROSCOPE ANALYSIS A JEOL JSM-T300 scanning electron microscope (SEM) was used to analyze the powder residue The goal of the SEM examination of the powder residue was to locate evidence of reaction and the manner by which reaction occurred between the spodumene and molten reductant. Samples for SEM analysis were made by impregnating a resin with the powder residue and then allowing the resin to set The resin was polished on a 0.3 micron alumina wheel with the aim o f exposing a cross-section of a reacted spodumene particle. Prior to the SEM analysis, the sample was carbon coated to avoid electrical charging in the sample chamber The SEM is capable of performing chemical analysis in a manner similar to the XRF unit described in Section VI.4.2. The SEM analysis system has two major drawbacks concerning the analysis of the powder residue: 1) A beryllium window is used between the sample chamber and the x-ray detector and as a result lithium and oxygen can not be analyzed for because their characteristic x-rays do not possess the energy to pass through the window. 2) Without oxygen detection, the analysis could not distinguish between pure metals and their oxides, which would make the results speculative. Therefore, analysis results from the SEM were used qualitatively and not quantitatively. The SEM analysis used was the SQ program of the Tracor Northern(44) TN-5400 software analysis system. Figure 6.11 shows a reduced spodumene particle observed in the residue from Experiment 5 at an accelerating voltage of 10 kilovolts. The particle was characterized by large and small cracks all over its surface and some of these cracks were contained within the outer edges o f the particle. chemical assay is shown in Table 6.12. I 58 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Its Figure 6.11. SEM microphotograph o f particle in the powder residue. Magnification equals 2000x. « ---------------------------------------------------------------------------------------------------------Table 6.12. SEM ASSAY 0 7 PARTICLE I N FIGURE 6 . 1 2 , Si Mg A1 28. 6 11.7 59.8 WEIGHT% spodumene. T he particle showed no evidence o f the existence o f a boundary layer between spodumene and molten reductant or a deposited reaction product. finding raises questions concerning the reaction mechanism which occurred which are dealt with m the discussion. » 59 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This VII. EXAMPLE CALCULATIONS VJI.1. DETERMINING SAMPLE CONCENTRATIONS FROM ABSORBANCE READINGS Lotus 123(45) was used to perform a linear regression between the concentration of the standard sections (X) and the corresponding absorbance reading (Y). The regression output from the data of the standard solutions in Table 6.5 is shown in Table 7.1. Table 7 .1 . RESULTS OF LINEAR REGRESSI ON SLOPE (m) I NTERCEPT (b) R SQUARED STANDARD ERROR OF SLOPE (Am) STANDARD ERROR OF Y ESTI MATE 1 (AY) 0 0 0 0 0 143 0 998 002 009 The regression line was forced through zero, the absorbance o f a distilled water solution. The value for r squared obtained, 0.998, revealed an excellent fit of the data to the linear regression. The standard error in the Y estimate is the standard error of the regression and was used to determine the analytical error. A calibration curve, Equation 7.1, was constructed from the regression results. The purpose of the calibration curve is to define a relationship between the absorbance readings of a species in solution and the concentration of that species in solution. Y = mX + b 1711 where, Y is the absorbance, counts m the slope of the regression line, counts-ppm’1 X the concentration, ppm b the intercept = 0. I 60 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The concentration of the sample solutions was determined by modifying Equation 7.1, so that, X = Y/m [7.2] To calculate the sample assays from the measured A.A. concentrations, a series of calculations were required, as shown below for the lithium concentration in the ingot from Experiment 3. For the ingot in Experiment 3 the absorbance was 0.219, thus the lithium concentration of the solution was equal to, using equation 7.2, 0.219 / 0.143 = 1.53 ppm or 1.53 mg Li / litre. [7 3 ] The solution was diluted by a factor of four, therefore theoriginal concentration of the liquor was, 4 x 1.53 mg Li / 1 = 6.12 mg Li / 1 [74j the original solution’svolume was 250 ml, therefore it contained, i 0.25 1 x 6.12 mg Li / 1 = 1.53 mg Li [7si The solution was obtained by dissolving 1000 mg sample, therefore the lithium concentration of the original sample was, 1.53 mg Li / 1000 mg sample x 100 % = 0.153 wt % Li V n.2. [76i EXAMPLE CALCULATIONS OF ANALYTICAL ERROR Two factors contributed to the error of the sample assays:the error in the regression and the error from the absorbance readings. In the Lotus output, the error in the regression curve is given as the standard error of the Y estimate, which was, for the Lotus output from Table 6.5, 0.009 units of absorbance. This can be converted to an error in concentration by the following equation, 61 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. AC, REG = S. [7.7] m t where, ACreg is the error in concentration due to the error in the Regression curve. is the standard error of the Y estimate from the regression, m is the slope of the calibration curve obtained from the regression. For the present example, i.e., the Li assay of the ingot in Experiment 3, the error in the lithium concentration from the regression curve was 0.0063 ppm. To determine the error due to the sample absorbance readings, the error in the absorbance readings, calculatedusing Equation 6.6, replaces the error in the Y estimate inEquation 7.7. For Experiment 3, the standard deviation of the readings was 0.002. This corresponded to a concentration error of 0.014 ppm. The total error in concentration, AC, for the ingot sample in Experiment 3 would then be, AC = AC KuG + AC A A. = 0.0063 + 0.0014 = 0.0077 ppm Li [7.8] where A C „ is the error in concentration form the regression and AC REG A A is the error in concentration of the absorbance readings. Replacing the concentration of lithium in the onginal liquor, with the concentration error, AC, calculated in Equation 7.4 and following Equations 7.5 and 7.6, the error in weight % was calculated and found to be 0.008 wt %. This represents a relative error of, 0.008 [7.9] x 100 % = 5.28 % 0.153 in the lithium reading in the ingot from Experiment 3. The error incurred by mass and volume measurements was considered to be too small to have an effect on the weight % error. 62 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. VIII. DISCUSSION V in .l. INTRODUCTION The goals of the discussion are: 1) to critique the experimental program, experimental error and analytical errors encountered. 2) to examine the quantitative and qualitative results from the experiments and explain their significance in terms of the physical phenomena occurring in the system. 3) to compare the quantitative results with those predicted thermody n amic ally. VIIL2.1. DISCUSSION O F PRESENT EXPERIM ENTAL PR O G R A M The goal of the experimental program was to determine the effect of magnesium addition on the lithium extraction from spodumene. This goal was achieved satisfactorily but the largest amount o f magnesium added, corresponding to a magnesium to spodumene molar ratio of 10.1, was chosen arbitrarily Thermodynamic analysis showed that increased magnesium addition past a molar ratio of ten further increased lithium extraction. Therefore tests at greater magnesium concentrations could be useful. V in.2.2. DISCUSSION OF EX PER IM EN TA L PRO CED U RE Various expenmental errors were incurred during the study. They are brought to light and their impact on the findings are discussed in this section. The order and method in which the reactants were charged was determined by the experimental apparatus. The impeller could only be inserted into the reactor when the metal reductant was molten and the crucible cap could only be placed after the impeller was inserted. Therefore, the melting of the reductant alloy was performed in air and some oxidation o f magnesium and aluminum occurred. A better method o f commencing the expenment would have been to heat all three charge matenals simultaneously. This could only be done in a reactor with the proper design. Temperature control was an important experimental parameter in this 63 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. study. Later in this chapter, Figure 8.12 predicts the theoretical effect o f temperature on lithium extraction. The precision of ± 20°C would have affected the recovery but the temperature fluctuation above and below 900°C would decrease the net effect on recovery White vapours were observed leaving the apparatus which must have contained some lithium since lithium was detected in the condensate. This lost lithium would not have been accounted for in a mass balance. A condensing apparatus that collected all of the vapours would have improved the investigation on the condensate and retrieved more lithium. The separation of the powder residue from the ingot, would have been improved upon if the powders were skimmed or scraped from the metal surface prior to pouring the ingot This would have resulted in more homogeneous dross, ingot and residue samples. The atmosphere in the reactor was argon. Using a noble gas was essential to the success of the experiments due to the highly oxidizing nature of aluminum, magnesium and lithium The seal of the reactor with the outside atmosphere was not perfect and during temperature measurements, plugs were removed from the crucible cap. The argon flushing gas created a positive pressure inside the reactor preventing atmospheric gases form entering the system and oxidizing the melt during sampling and temperature measurements. In the present study, there is ample evidence that reduction of the spodumene by the molten reductant occurred. A question which then arises is why were the impeller, whose surface was an alumina paint, and the crucible, a ceramic made from a silica - alumina combination, not reduced. An explanation is that oxide formation occ< .red at the interface between the excess metal reductant and these objects. This oxide would achieve a thickness which would prevent attack of the crucible or impeller. VIII.2.3. DISCUSSION OF ANALYTICAL ERROR VIII.2.3.1. RELATIVE ERRORS OF SAM PLE ASSAYS As mentioned earlier, the error in the atomic absorption analysis arose from the errors in the regression curve and absorption readings. In this section, the error from each source is analyzed separately to see if any trends exist. 64 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The average relative errors o f the sample concentrations from the regression and from the absorption readings are listed in Table 8.1. Table 8.1. SAMPLE AVERAGE RELATIVE ERRORS I N SAMPLE CONCENTRATIONS REGRESSI ON ERROR(%) Li ABSORBANCE ERROR(%) Mg Si Li Mg Si DROSS 3.4 7.6 4.5 3.8 5.3 8.2 RESI DUE - 8.0 7.8 - 1.9 4.2 INGOT 4.3 7.2 7.2 4.6 1.7 13.5 BLACK POWDER 2.6 7.4 - 1.2 1.0 - CONDENSATE 3.3 - - 1.5 - - AVERAGE(%) 3.4 7.5 6.5 2.8 2 .5 8.7 Table 8.1 shows that not all elements are similar in their A.A. characteristics in terms of regression and absorbance variance. the best element for regression analysis. 3.4%. Lithium was Its average relative error was Next best was silicon with an average relative error o f 6.5%. Magnesium analysis produced the worst regression having a relative error of 7 5%. The error arising from the variance o f the absorbance readings was affected by the d n ft in the absorbance readings over a period of time for a series of readings. For example, an absorbance reading for a sample solution and was identical to the absorbance for a standard solution containing 1 ppm. After other solutions were analyzed the absorbance was retaken for these two solutions. The readings were identical again, but were 10% less than obtained before. The precision of the machine has not changed but the variance of the solution readings has increased. Thus the errors calculated due to the variance of the absorbance are artificially high, as they do not compensate for the machine drift The amount o f dnfting which occurred for each element varied. had severe drifting, causing an average relative error of 8.7%. Silicon As a result of the large absorbance dnft when analyzing for silicon, it is recommended that the amount samples analyzed for silicon in one sitting be restricted. The relative error from the absorbance drift for lithium and magnesium were 65 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 2.8% and 2.5% respectively. For magnesium, the absorbance error from the dross of Experiment 5 was 15.7%. If it was excluded the average relative absorbance error for the dross would have been 2.7% and the average relative magnesium absorbance error would have been 1.8%. VIII.2.3.2. COMPARISON OF ASSAYS W ITH AN OUTSIDE ANALYSIS The results from an outside analysis performed at Centre Recherches des Minerales (CRM) are shown with the corresponding assays used in this work in Table 8.2. Table COMPARISON OF ANALYTICAL RESULTS ANALYTICAL LAB Wt EXP 5 POWDER RESIDUE CRM 0.84 _ 16.8 EXP 5 POWDER RESIDUE TANCO 1.07 - 45.7 EXP 6 METAL INGOT CRM 0.10 1.54 EXP 6 METAL INGOT MCGILL 0.28 3.74 SAMPLE I 8.2. % Li Large differences exist between the results. Wt % S i Wt % A l w However, the outside data are inconsistent with the mass balance for lithium performed in Section VII.4. The total lithium accounted for at the end of the experiments was less than that contained in the spodumene. The CRM values were less than those used for the mass balance, therefore, if they were to be substituted into the mass balance the lithium deficiency would increase. However, the fact that the difference in some values is so large is disturbing and unaccounted for. For future researchers the following recommendations should be considered: -The size of the sample should be as large as possible but balanced against practicality in terms of manageable solutions and safety, for increasing the amount of sample also increases the violence o f the reaction 66 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. with acid. -Segregation exists within the ingot and therefore the samples must be m ade by collected drillings and pieces from as many different locations as possible. V III.3. EX PER IM EN TA L MODELING This study, although it did not concentrate on the process phenomena occurring in the m :tallo-thermic reduction o f spodumene, did rely on, and was affected by, various process phenomena. In this section, two process phenomena are examined. VIII.3.1. D ETERM IN A TIO N OF GAS PU RGING RATE The inert argon atmosphere above the melt surface was essential in preventing oxidation of the molten metal. The rate of argon purging was determined from the volume above the melt surface and the rate of argon input. The volume o f above the melt surface was calculated by subtracting the volume of the molten aluminum, spodumene and impeller from the crucible volume. The volume of the magnesium added was not included and thus to compensate, the mass of aluminum used in the calculation was increased to 2400 g. The intenor dimensions of the crucibles were 28.8 cm high and 12.8 cm diameter. T he cap overlapped the crucible by one centimeter. Thus the effective volume of the reactor was: (28.8 -1) cm x it 12.82 cm 2 = 3 577.4 cm3 is.i] The volume of the impeller was calculated by adding the volume of a) its blades coated in cement, b) the upper portion o f the shaft and c) the lower part of the shaft, coated in cement. The total impeller volume, assuming it operated two centimeters from the bottom of the crucible was, 67 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. a) blades 4 (4.95 x 11.0 x 1.0) cm3 b) m e ta l shaft (2 7 .8 -2 -1 1 ) x re L90» cm3 4 c) s h a f t coated in cement 11 x k 6 . 352 c m 3 d) t o ta l 610.95 cm 3 [8 .2 ] [8.4] An amount equal to one mole of spodumene was added to the reactor which had a volume of, v = 372 g + 2.6 ggem'3 = 143 cm3 [85] however, in its loose form and resting on top of the melt it was estimated to have a porosity of 50% making the effective volume of the powder 286 cm3. The volume of 2400 g of molten aluminum can be calculated from its density, given by the equation for molten aluminum(46), p = 2385 - (0.28 (T-933)) kg m'3 [8.6] at the operating temperature of 1173 K the specific gravity of the aluminum would be, p = 2385 - (0.28(1173-933) = 2318 kg m3 = 2.32 g cm'3 [8.7] this would make the volume of molten aluminum equal to, 2300 g + 2.32 g cm'3 = 1034.5 cm3 Thus the total volume o f the materials in the reactor was 1 931.4 cm3, leaving 1 646 cm of free space. The average residence time, x, of the argon was, ! 68 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. [8 8] where, v is the volume Q is the flow rate o f gas. The equation for the gas purging of a system is, f(t) dt = -exp [8 10] where, f(t) is the fraction of gas purged form the system t is the time in seconds x is the gas flow rate entering the system in m s '1 Solving Equation 8.10 by integration with respect to time determines that 83.8% of the gas would be purged every minute. 1 VIII.3.2. FORMATION OF SOLID SPECIES A possible, practical difficulty with this process is the formation of solids during reaction. Figure 8.4, presented in Section VIII.5 shows that more solids were present as residue at the end o f an experiment than the amount of spodumene added. In this section, two reaction paths having one mole of spodumene react with an excess (XS) of aluminum are examined: one where two moles of magnesium react with the spodumene and the other without any magnesium, to determine the theoretical amount o f solids produced. each reaction path the spodumene is reduced 100% i 69 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. In REACTION PATH 1: Li20 Al20 3-4Si02 + XSA1 +2Mg=2Li+2MgO-Al O + (XS-6)A1 + 4Si + 3A120 3 MASS POWDER IN : SPO DU M EN E = MASS POWDER OUT: M g O - A l 2 0 372 g 284 g S IL C O N *(112) g 306 gz D IFFER EN C E R E A C T IO N L i O A l 2 2 +218 + jc( 112) g PATH 2: 0 3 M ASS POWDER 4S i 0 2 + XS A 1 = 2Li + (XS-3)A1 + 4Si + 4A1 O. 2 I N : SPODUMENE M ASS POWDER OUT: A 1 0 2 = 372 g 408 g 3 S i * (1 1 2 ) g D I FF E R E N C E Figure 8.1. 3 +36 + jc( 112) g The analysis o f solid formation in the reduction of spodumene for two reaction paths. X is the percentage of silicon that would enter the powder phase. Figure 8.1 shows that as the reaction progresses more solids are formed. This would result in a build up of solids in the reactor which would have adverse effects on the process. For a continuous process, to counter the buildup of solids, occasional mechanical removal of the solids would be necessary. The formation of solids is greater when magnesium is added due to the formation of spinel. VIII.4. MASS BALANCES In this study, three elements were very well suited to mass balance analysis: silicon and lithium in the spodumene and magnesium from 70 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. the additions. The total mass of an element existing from an experiment was calculated by multiplying the assays o f the products from Tables 6.6 - 6.10 by their respective masses from Table 6.1 The mass balances for lithium, magnesium and silicon are shown in Tables 8 3 - 8 5 respectively. The silicon and magnesium impurities in the aluminum reactant metal were negligible, as shown in Table 5.1, and were not incorporated into the mass balance. Table 1 8 3. MASS BALANCE FOR LITHIUM EXP MASS I N ( g) MASS OUT( g) % DIFF 1 12 . 3 6 9.40 -24.0 2 1 1 . 62 10 . 5 9 -8.9 3 12 . 4 7 9 . 41 -24.6 4 12 . 4 6 9 . 51 -23.7 5 12 . 5 1 9 . 93 -20.6 6 12 . 2 3 10.76 - 1 2 .0 7 11.56 11.16 -3.6 8 11.78 10 . 68 -9.4 The average difference in the lithium mass balance is minus 15.8%. The sources of the lithium discrepancy could have been, analytical error and loss of lithium vapour It is possible that lithium diffusion due to a lithium concentration gradient between the molten alloy and the crucible and impeller occurred. Lithium has long been known as a destructive material and it /n\ destioys a solid by imbedding itself into the lattice which could have occurred with the impeller and the crucible. During the experimental program the crucibles were changed occasionally due to cracks and material build-up. However, it was not recorded whether an experiment used a new crucible or not. The lithium mass balances appear to divided into two groups: those with a lithium deficiency between 20.6% and 24.6% and those with a lower lithium deficiency between 3.6% and 12.0%. would be very interesting to determine if a relationship existed between the lithium deficiency and the crucible used. C 71 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. It 8.4. T a b le MASS BALANCE FOR MAGNESIUM i EXP MASS IN (g) MASS OUT( g ) % DIFF 1 97 . 60 100.85 3.3 2 0.0 0.0 3 104.00 99.01 -4.8 4 46.71 53.91 13.2 5 143.20 117.46 -18.0 6 191.90 190.57 -0.7 7 220.60 560.74 154.2 8 224.80 189.48 -15.7 The mean, absolute, difference in the magnesium mass balance excluding Experiment 7, where a spurious reading was obtained for the powder residue, was 9.3%. Magnesium losses occurred through volatilisation and the losses of fine magnesium oxide and spinel in the fine black powder. T a b le 8.5. MASS BALANCE FOR SILICON EXP MASS I N (g) MASS OUT( g ) % DI F F 1 110.44 97.41 -11.8 2 103.83 121.16 16.7 3 111.44 106.48 -4 4 111.33 89.04 -20.0 5 111.80 75.03 -17.5 6 109.90 109.26 -0.0 7 103.31 114 .50 10.8 8 105.22 1 0 6 . 98 1.7 4 The mean, absolute, difference in the silicon mass balance was 8.90%. Sources of the silicon could have been losses to the fine black powder and analytical error. 72 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. VIII.5. THE EFFECT OF MAGNESIUM ADDITION ON THE REACTION PRODUCTS ASSAYS The experimental program varied the magnesium addition to the system and in this section its effect on the lithium, silicon and magnesium assays in the products is analyzed. The data used to plot the graphs in this section was obtained from the assays presented in Section VI 4. Figure 8 2 shows that the lithium concentration in the ingot increased as the MgiSpod molar ratio was increased up to a ratio of 8* 1. where it leveled off. Thus, as more magnesium was added as a reductant, more lithium was released and dissolved into the excess molten metal. When no magnesium was used, very little lithium was determined to be in the ingot, thus no reduction of lithia in P spodumene occurred. The weight % Li in the dross also increased steadily. The lithium concentration was higher in the dross than in the ingot for six ol the eight experiments. This was due to the formation of dross in areas of the reactor where the powder was in good contact with the molten metal, i e., on the surface of the melt and along the walls. « \ i 73 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. N a tio n a l L ibrary o f C an a d a C anadian T h e s e s S ervice B iblioth& que n a t i o n a l e du Canada S erv ice des th& ses can ad ien n es NOTICE AVIS THE QUALITY OF THIS MICROFICHE I S HEAVILY DEPENDENT UPON THE QUALITY OF THE THESIS SUBMITTED FOR MICROFILMING. LA Q U A L I T E DE C E TT E M I C R O F I C H E DEPEND GRANDEMENT DE LA QUALITE DE LA THESE SOUMISE AU MICROFILMAGE. UNFORTUNATELY THE COLOURED ILLUSTRATIONS OF THIS THESIS CAN ONLY YIELD DIFFERENT TONES OF GREY. MALHEUREUSEMENT, LES DI F F E R E N T E S ILLUSTRATIONS EN COULEURS DE CETTE THESE NE PEUVENT DONNER QUE DES TEINTES DE G R I S. Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 0.45 INGOT 0.40 — X- _ 0.35 DROSS Li 0.30 WEIGHT% 0 25 0.20 0.15 0.10 0.05 0 00, Mg SPODUMENE MOUVR RATIO Figure 8.2. Lithium weight % in the ingot and dross vs. the magnesium to spodumene ratio. The lithium concentration in the powder residue, Figure 8.3. decreased curvilinearly at a decreasing rate with the Mg:Spod ratio. The range was from 2.53% Li at a Mg:Spod ratio o f 0:1 to 1.06% Li at a ratio of 10:1. The decrease in the amount of lithium in the residue could have been due to lithium extraction from the fi spodumene or dilution due to the formation of other oxides in the residue. case. Figure 8.4 shows that the latter was not the The powder residue did have a greater mass than the charge mass of P spodumene as a result of inclusion o f metal pamcles, alloy oxidation and the production of reaction products such as, silicon and spinel. However, the difference was roughly constant and not enough to account for the decrease of the lithium in the powder residue. 74 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 3.0 2.5 2.0 z 2 ai 5 0 0, MgSPODUMENE MOLAR RATIO < 400 ---- ............ K - ----------- ------- - 500 X Lithium concentration in the residue vs. Mg:Spod ratio. ...................... Figure 8.3. > - i 4 -+ x - RESIDUE X + ► MASS (g) CHARGE 4 300 200 100 10 Mg SPODUMENE MOLAR RATIO Figure 8.4. The masses of the powder residue and (3 spodumene charge vs. the magnesium to spodumene ratio. 75 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The lithium concentration in the condensate, Figure 8.5, increased to a peak o f 0.94% at a Mg:Spod ratio of six and then decreased as more magnesium was added to the system. The increase of lithium in the condensate was believed to be due to increased liberation of lithium from the spodumene as more magnesium was added to the system. The decrease iuier a ratio o f six was due to the increased vapourization and subsequent condensation of magnesium, thereby diluting the lithium in the condensate. This data indicates that the vapours richest in lithium were produced in Expenment 5 which had a Mg:Spod ratio of six to one. 0.8 0.6 i o ui 5 0.2 Mg Figure 8.5. SPODUMENE MOLAR RATIO Lithium weight % in the condensate vs. the magnesium to spodumene ratio. I 76 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The lithium concentration in the fine black powder, Figure 8.6, remained constant at appoxiamately 1.2% Li, as the Mg:Spod ratio was increased from two to one to ten to one. This suggests that the lithium present in the very fine black powder is unaffected by the amount of magnesium addition. 0.8 0.6 0.4 0.0. Mg SPODUMENE MOLAR RATIO Figure 8.6. Lithium weight % in the very fine black powder vs. the magnesium to spodumene ratio. V 77 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The silicon concentration in the ingot and dross, Figure 8.7, increased slightly with the increase in the Mg:Spod ratio. The presence of silicon in the ingot when no magnesium was used, confirms that aluminum reduces the silica in the (3 spodumene to form silicon. The silicon concentration of the dross sample from Experiment 8 (Mg:Spod ratio of 10:1) was extremely high and attests to the non-homogeneity of the dross. INGOT 6.0 —x— DROSS 5.0 W EIGHT% V) 4.0 3.0 x 2.0 « JT .0 " '0 2 4 6 8 10 MgtSPODUMFNE MOLAR RATIO Figure 8.7. The silicon weight % in the dross and ingot vs. the magnesium to spodumene ratio. C 78 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The silicon concentration in the powder residue, Figure 8.8, decreased with the increase in the Mg:Spod ratio and then leveled off. The range of the silicon concentrations v/as from 15.3% at a Mg:Spod ratio of zero to 4.7% at a Mg:Spod ratio of 4.3. The leveling of silicon concentration can be explained by either 1) no further reduction o f die silica species in the spodumene occurred or 2) further reduction o f the silica with the silicon product remaining undissolved in the molten reductant. The phase diagram of the aluminum-magnesium system shown in Appendix B shows that at 900°C a liquid solution exists until liquid metal is 37 wt% silicon so therefore a saturated silicon solution did not occur in the present study. However, it is possible that the silicon remained undissolved for unknown reasons. 16.0 14.0 WEIGHT% Si 12.0 10.0 8.0 6.0 4.0 2.0 0 . 0, Mg:SPODUMENE MOLAR RATIO Figure 8.8. Silicon weight % in the powder residue vs. the magnesium to spodumene ratio. 1 79 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The magnesium concentration in the ingot and dross increased as more magnesium was added to the system as shown in Figure 8.9. The concentrations are less than the weight % magnesium expected according to the stoichiometric amounts of the reactants. This is due to loss of magnesium from the alloy via oxidation, volatilisation, and reaction. INGOT WEIGHT% Mg DROSS Mg:SPODUMENE MOLAR RATIO Figure 8.9. Magnesium concentrations in the ingot and dross vs. the magnesium to spodumene ratio. f V 80 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The magnesium concentration in the powder residue increased as the Mg:Spod ratio increased as shown in Figure 8.10. The magnesium concentration in the flue powder, also shown in Figure 8.10, increased with Mg:Spod ratios. The flue powders were thought to be composed of the finer sizes o f the residue. At the higher Mg:Spod ratios, m ore fine magnesium particles were produced, such as magnesium oxide, and w ere carried away by the flushing gas. RESIDUE 40 - -x — FLUE POW. 20 M gS PO D U M E N E MOLAR RATIO Figure 8.10. The magnesium weight % in the powder residue and very fine black powder versus the magnesium to spodumene ratio. 81 Reproduced with permission o f the copyright owner. Further reproduction prohibited without permission. The lithium extraction from spodumene shown in Figure 8.11 was calculated as the difference in the mass of lithium between the charge, P spodumene, and the powder residue divided by the charge mass of lithium in the p spodumene times one-hundred percent, ^EXTRACT = <Mi,SS L ’SPO0 ' M i >SS ‘' ‘r I S . d J + M ^ SS L W 100% [8.U1 It increased curvilinearly as the Mg.Spod ratio increased at decreasing rates. The largest lithium extraction was 67.2% and occurred at a MgtSpod ratio of ten to one. LU Z LU 2 3 Q O Q. W I 8l Z O H O X LU Mg:SPODUMENE MOLAR RATIO Figure 8.11. Lithium extraction from the powder residue vs. the magnesium to spodumene ratio. 1 82 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The lithium recovery to the ingot, Figure 8.12, was calculated as the mass o f lithium in the metal ingot divided by the mass of lithium in the spodumene charge times one-hundred percent, U RIZCOVERY = ( W ‘% L i INO OT ' M “ W + M i * SS “ s P O lP ' 100% l!'121 The lithium recovery to the ingot increased as the Mg:Spod ratio increased. The maximum recovery attained was 58.1% The lithium recoveries were less than the lithium extraction from spodumene due to lithium occurrence in dross, flue powder and condensate as well as lithium losses to the vapour phase. T he low recovery value at a Mg:Spod ratio of four to one was for Experiment 1 and was due to the small ingot obtained in that trial. « s z cc UJ > LU cc Mg SPODUMENE MOLAR RATIO Figure 8.12. Lithium recovery to the metal ingot vs. the magnesium to spodumene ratio. i 83 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. A summary of the mass balances and the lithium, silicon and magnesium distribution to the reaction products for each Experiment is shown in Figures 8.13 to 8 16 These Figures show much information that has already been presented earlier, but in a slightly different manner, which may aid the reader in grasping the relationships put forwaid in the previous section. New evidence presented is discussed for each Figure The lithium distribution in the reaction products shows that the percentage of lithium present in the ingot increased as the magnesium to spodumene ratio was increased. Conversely, the distribution shows that the lithium percentage of the powder residue decreased as the magnesium to spodumene ratio increased. The lithium distribution for the experiment with a magnesium to spodumene ratio of zero shows that about ninety percent of the lithium remained in the powder residue, attesting to die small amount of lithium reduction which occurred in this experiment. The amount of lithium unaccounted for decreased as the magnesium to spodumene ratio increased for undetermined reasons. NGOT RES. BP EZlZ Z I LOST 41 20 6.0 44 100 62 M gS P O D tM M MOLAR RATIO OF TEST Figure 8.13. Lithium distribution in the reaction products for each experiment. ! ------------------------------------------------------------------------------------------- 84 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The silicon percentage of the powder residue decreased as the magnesium t to spodumene ratio increased. The silicon distribution in the other reaction products is inconsistent among the experiments. \\\V aQ RES. Q J, l/l sssss Iff BP u zm S ll o B Q_ 1 in II i 1 41 20 44 ao LOST i io o &2 100 MgSPOOLMEFE MOLAR RATO OF TEST Figure 8.14. Silicon distribution in the reaction products for each experiment. 85 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. The distribution for magnesium shows that the percentage o f magnesium in the powder residue decreased as the magnesium to spodumene ratio was increased This meant that at the lower Mg:Spod ratios a greater percentage of the magnesium was utilized in the reduction reactions and magnesium oxide products contained in the powder residue weie formed The second bar from the right was excluded from the analysis due to the spurious reading of the magnesium concentration in the powder residue from Experiment 7. NGOT Q uQ 5 2 □ (/) U i z o DROSS u 5 Q. I 0 41 20 60 44 100 S2 100 MgrSPOOLMM MOLAR RATIO Of TEST Figure 8.15. Magnesium distribution in the reaction products for each experiment. VIII.6. KINETICS OF TH E SPODUMENE REDUCTION SYSTEM The rates of spodumene reduction were qualitatively determined from the plot of lithium recovery to the kinetic samples versus the sample time, Figure 8.16 The lithium recovery was calculated as the lithium assays of the kinetic sample divided by the nominal lithium concentration, according to the amount of reactants, of the excess molten reductant, i.e., 86 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. REC Li Li-KIN = ^M U 18.13) K IN SPO D ^ ^ M E L T ^ High rates of lithium recovery occurred for the first twenty minutes of the experiments and then decreased considerably thereafter. Also, the rate of lithium recovery and amount of lithium recovered increased with increased magnesium to spodumene ratios. The latter finding is consistent with the results of the lithium recovery to the ingot samples, shown in Figure 8.12. $ 6 & §l U 10 cc 3 0 10 20 30 40 50 60 TIM E (MINUTES) Figure 8.16. Rate of lithium recovery to the melt versus time for various experiments performed in this study. The decrease in the rate may have been a result of the following, 1) Equilibrium was attained by the system. 2) A rate limiting step has been reached. 3) Depletion of (3 spodumene ore. 4) Depletion of magnesium reductant. However, a discussion of these is beyond the scope of this thesis. 87 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. VIII.7. DISCUSSION OF XRD RESULTS The XRD results gave insight into the reactions which occurred and aided in the determination of the physical mechanisms which took place. The XRD analysis showed that the powder residue was a very complex material. It was composed of aluminum, spinel, silicon, spodumene, lithium aiumino silicates and penda.se The aluminum in the residue was caused by the entrainment of aluminum in the powder phase. The spinel was a reaction product from the reduction of alumina by magnesium The pure silicon detected in the lesidue was the product of the reduction of silica in the spodumene by aluminum or magnesium. The spodumene present in the residue indicated that not all of the spodumene reacted. This may have been because some o f a spodumene feed may not have undergone the a -> (3 transformation and was thus added as relatively inert a spodumene. In fact, the lithium aiumino silicate species, 35-797 and 31-706, detected in the powder residue have XRD patterns similar to a spodumene. In the spodumene conversion performed, the concentrate expanded and packed itself into a very dense material. It is possible that some particles were "crushed" by others and the resulting compressive forces deterred the transformation of certain crystals It is recommended that future researchers perform spodumene conversions in vessels that allow the spodumene to expand freely. The magnesium oxide present in the residue was fro- ihe oxidation of magnesium; during dissolution and by reduction of the spoc ; nene. The flue powder was composed of aluminum, silicon ,tnd spinel. All these constituents were present in the powder residue and were believed to have been a fine sized portion of the powder residue, which was physically transported from the reactor by the argon flushing gas. The condensate was composed of magnesium oxide. It formed via the vapourization of molten magnesium in the liquid alloy and its subsequent condensation and oxidation on the underside of the crucible cap. VIII.8. WETTING EFFECT OF MAGNESIUM It was thought that magnesium dissolved in aluminum would improve the wetting between the molten alloy and the spodumene as explained in Section 88 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. II.4. In Experiment 2, where no magnesium was used, silicon recoveries to the ingot, shown in Figure 8.7, were similar to those encountered in the other tests. This meant that sufficient wetting occurred, without magnesium, to permit the reaction between molten aluminum and spodumene. Thus magnestum’s importance in the other experiments was thermodynamic. The kinetic study o f the lithium recovery to the molten alloy showed that increased magnesium addition increased the rate of lithium extraction. This effect could be due to improved wetting caused by the magnesium but a more plausible explanation is that the lithium extraction was faster due to a higher magnesium activity, though the exact relationship is not known. VIII.9. DISCUSSION O F REACTION M ECHANISM S The reaction mechanism describes the physical nature o f the reaction between the spodumene particles and the molten reductant The reaction mechanism which occurred in the present work is in fact is beyond the scope o f this thesis. However in this section, observations pertinent to the reaction mechanism are mentioned and the questions which they raise are discussed: 1) There was an absence o f a layer of reaction products on the reacted particles’ surface by SEM analysis: This meant that the reaction products dissolved or were released into the molten reductant. Another possibility is that the reaction products formed within the spodumene and not just at the surface. 2) There was positive identification o f spinel and silicon in the powder residue: This result more than any other indicated that reduction o f the spodumene occurred as these are products o f spodumene reduction. However, the presence of silicon in the powder also meant that the silicon did not dissolve in the excess reductant. This is an entirely unexplained observation 3) The kinetic study found that reaction occurred very rapidly in the first twenty minutes and then slowed considerably: A very fast reaction mechanism occurred. This was invariably due to the 89 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. small size o f the spodumene charge » However, for a system dealing with solid crystals and molten metal, the rate was higher than anticipated. 4) There were cracks and holes evident in the powder residue reaction particles: These phenomena may have resulted from in situ reaction by diffusion of reductant atoms and the subsequent diffusion ot the reaction products out of the crystal into the excess molten reductant leaving the holes. would have facilitated the transport of atoms. The cracks A mobile small atom such as lithium would have a very high diffusion rate and thus the greater amount o f it produced would enter the molten reductant. Silicon, which would have a slower diffusion rate would only partly enter the molten phase. An oxide product, such as spinel, would station itself in the crystal lattice and not enter the melt This would explain the: detection o f spinel and silicon in the powder residue and the silicon assay o f the metal ingot. VIII.10. It DISCUSSION O F REACTION PATHW AYS Pinpointing the reactions which occur in a system is an important task which raises the level of understanding of a system. Information obtained from the different modes of analysis were utilized to develop two proposed different reaction pathways in this Section. VIII. 10.1. TW O-STEP R EA C TIO N PATHWAY From the XRD results o f the powder residue and the silicon assay of the ingot from Experiment 2, where no magnesium was used, the reduction of the silica in the spodumene by aluminum must have occurred, i.e., Li^O-Al 0 -4SiO + m 2 3 2 3 A1 = J A1 0 3 2 3 + 4Si + Li O AI O 2 2> 3 i*.i4] This first step is in agreement with the F*A*C*T simulation o f the system and would have to have occurred by a liquid-solid reaction mechanism. In the second step, the magnesium in the molten aluminum reduced the lithium aluminate to form spinel and lithium, l e., f 90 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Mg + Li 0 A1 O 2 2 3 = 2Li + MgO-Al O ° 2 1815] 3 It is not known which of the two reactions was rate limiting, but as shown in the kinetic study in Section VIII.6, the rate at which the majority of the lithium entered the alloy was very fast. VIII. 10.2. ONE-STEP REA CTIO N PATHWAY In the one-step reaction pathway the aluminum does not participate in the reduction process, only the magnesium reacts, L i20-A120 3'4S i 0 2 + 9Mg = 2Li + 4Si + MgO-AI^O + 8MgO [8t6] due to a much higher rate of reaction with the spodumene by a gas-solid mechanism than that for aluminum by a liquid solid mechanism. The one-step reaction pathway requires a magnesium to spodumene ratio of nine to one. Magnesium oxide or penclase(MgO) is a reaction product formed in the one-step pathway not formed in the two-step pathway. Penclase was detected in the powder residue only for Experiments 6,7 and 8 where the magnesium to spodumene ratios of eight to one for Experiment 6 and ten to one for Experiments 7 and 8 were used. The free energy of this equation at 900°C, calculated by F*A*C*T, is -865343 2 J which means that the reaction will precede to the right. At magnesium to spodumene ratios of six to one and less, lithium was produced by the two step reaction pathway and at higher magnesium to spodumene ratios it is believed that both the one and two-step reaction pathways contributed to lithium formation. VIII. 11. VACUUM R EFIN IN G OF LITHIUM FR O M ALUMINUM In this thesis, an assumption has been made that it is possible to vacuum refine lithium from aluminum In this section, a brief explanation of vacuum refining is given and it is investigated whether it is theoretically possible to of vacuum refine lithium from aluminum. Vacuum refining is based on the difference in vapour pressure between two species in a melt When the melt is exposed to vacuum, the species with the higher vapour pressure, will preferentially evaporate from the species 91 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. which has the lower vapour pressure. A description of the triads transfer phenomena and the series of equations necessary to develop the equation which determines whether solute elimination is possible via vacuum refining is beyond the scope of this thesis. Excellent references are Dimayuga148'1 and Harns(49). The volatility coefficient, <f>, defined by Ollette(50) of a volatile sohite,i , in a less volatile solvent, b, is delined as, M. 18.17) M where, 7° is the activity coefficient in the melt P° is the equilibrium vapour pressure of pure species, (Pa) M is the molecular mass, kg kgmole'1 i, b denote the solute i and solvent b, respectively. If a solute has a volatility coefficient greater than one, then vacuum refining of the solute is possible. Dimayuga(48)calculated the volatility coefficients for various elements in liquid aluminum and a few are listed in Table 8.6. Table SOLUTE, 8.6. i VOLATILITY COEFFICIENTS OF VARIOUS SOLUTES IN MOLTEN ALUMINUM 0 yi T = 973 K T = 1173 K T = 1473 K 2.1 x 104 1.2 x 105 -4 10 Li 0 . 40 1.3 x 107 5.1 x 10J Mg 0.7 1.5 x 10® -8 10 4.2 x 106 Si 0 .04 3.6 x - 7.0 x 10 7 1.6 x The results state that lithium and magnesium would both be eliminated by vacuum refining while silicon would not. If spodumene reduction operation 92 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. utilized an aluminum-magnesium alloy then magnesium would volatilize with the lithium and the two would have to be separated. t This could be achieved by: 1) having different condensation characteristics in the reactor for fractional condensation or 2) separating the lithium from the magnesium in a bulk condensate by melting the condensate and utilizing lithium’s lower melting point, 180°C vs. 648°C. VIII. 12. C O M PA R ISO N O F EX PERIM EN TA L RESU LTS W ITH F*A*C*T CA LCU LA TION S VIII. 12.1. IN TRO D U C TIO N The F*A*C*T analysis performed in Section IV. 1 had a great bearing on the design of the experimental program. The experimental results were compared to the F*A*C*T calculations here. T he goal o f the comparison was to deternvne if a similarity existed between the lithium extraction values and if so, then to use F*A*C*T to simulate the alumino-thermic reduction of spodumene while varying other thermodynamic parameters such as temperature, pressure and amount of excess liquid aluminum. VIII.12.2. CO M PARISO N O F TH ERM O D Y N A M IC M OD EL AND EX PER IM EN TA L RESULTS The graphs from Figures 8.11 and 3.2 tire presented here again in Figure 8.17 in order to compare the expenmental results to the thermodynamic predictions. Equation 8.11 was used to determine lithium extraction and Input 3.1, shown here again as Input 8.17, was used for the F*A*C*T calculations. Li20 A l20 3'(SiO2)4 + 85.5 Al + <A> Mg [8.17] Figure 8.17 shows that g. od agreement existed between the lithium extraction obtained experimentally and calculated by F*A*C*T . At a magnesium to spodumene ratio o f two to one, the lithium extraction in the experiment was much larger than the predicted extraction by F*A*C*T. This may have been due to the less than perfect mixing of all the reactants which « would have resulted in localized, large, magnesium concentrations that would 93 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. have improved spodumene reduction. Another discrepancy is the flattening o f the experimental curve at higher MgiSpod ratios. This was due to less than 100% efficiency of reaction, possibly due to the less than 100% transformation o f a to P spodumene 80 ..... ........'"i .. i i 70 50 2 O 40 X • + i ______ ________ + 60 EXP. 44 x FACT + to tr + > x----- UJ zLU 2 3 a O a. ....... K* 30 20 10 V--------- < : 10 Mg:SPODUMENE MOLAR RATIO Figure 8.17 Lithium extraction from experimental work and F*A*C*T calculations vs. the magnesium to spodumene molar ratio. Differences in a thermodynamic simulation and experimental work will always exist because thermodynamic calculations are done for steady state, ideal systems which were not attained in this study due to: a) Argon flushing gas which earned lithium vapour out o f the vessel, and lowered the amount of lithium in solution, offsetting the equilibrium and permitting more lithium to enter the molten metal. b) Imperfect mixing of the reactants. c) Unknown wetting characteristics between the molten reductant and spodumene. 94 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Other reasons for differences between experimental work and thermodynamic simulated results in metallurgical processes are: i d) miscellaneous material losses. e) analytical error. f) The data used to create dissolved species is continually being updated and until data for dissolved species is obtained which is unrefutable, the thermodynamic predictions are not 100% accurate. With all the possibilities for differences and the complexity of the system, the c'ose resemblance of the experimental and computer modeled curves is nevertheless an outstanding achievement of this thesis and suggests that there is great value in using F*A*C*T as a simulation tool for further study of this system. VIII. 12.3. F*A*C*T SIM U LA TIO N OF TH E ALUM INOTH E R M IC RED U C TIO N OF SPODUM ENE The addition o f magnesium is a thermodynamic parameter in the reaction between spodumene and aluminum. Other parameters are temperature, pressure and the amount of excess molten aluminum. Using the F*A*C*T EQUILIB program the alumino-thermic reduction of spodumene '»'as examined while varying the above three parameters. The Input supplied to F*A*C*T for the temperature change simulation is shown in Input 8.18. Li O A1 O -4SiO 2 2 3 2 + 85.5A1 = [s.isj at a pressure of 1 Atm. and at temperatures ranging from 700°C to 1300°C by increments o f 100°C. The input to F*A*C*T for the pressure simulation also uses Input 8.18. The temperature was fixed at 900°C and the pressure was lowered from 1 Atm. to IE -5 Atm by orders of magnitude. The simulation for the excess molten reductant used Equation 8.19 as the Input, Li OA1 O 4S i02 + <A>A1 = [».i9] « 95 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. at a temperature of 900°C and a pressure of 1 Atm while increasing the number of moles of aluminum from 50 to 739. The results of the simulations are shown in Figures 8.18 - 8.22. The F*A*C+T simulation of the lithium extraction from the ore vs. temperature shown in Figure 8.18 increased exponentially as the temperature was increased. At a temperature of 1582°C, 100% recovery was obtained. would not be a reasonable operating temperature. Aluminum fuming would occur, causing aluminum impurities in the condensate and there would be serious attack on the refractories used. 100 >■ cr LLI > 8 LU QC 1 800 1000 1200 1400 1600 TEMPERATUREC Figure 8.18. Lithium extraction from spodumene vs. temperature, Input 8.18. V 96 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. This The lithium extraction linearly increased as the amount of excess aluminum was increased as shown in Figure 8.19. One-hundred percent recovery occurred when the ratio o f aluminum to spodumene was 739:1, corresponding t to an aluminum to spodumene mass ratio of 53.6:1. % Li RECOVERY 100 200 600 400 800 MOLES Al Figure 8.19. Lithium extraction from spodumene vs. the amount of excess liquid aluminum, Input 8.19. 97 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Figure 8.20 shows that the lithium recovery did not increase as the pressure was lowered until a value of 1.2E-03 Atm was reached, whence 100% recovery of the lithium occurred. The recovery in this simulation is to the gas phase as lithium vapour is formed. This critical pressure could be increased by increasing the temperature o f the system as shown in Figure 8.21. The addition of more liquid aluminum was found to not have an effect on the pressure required for 100% recovery. 100 r...... — --------1--------1------- -i—..... - 80 AU3AO03H 60 H % 40 20 -7 -4 -6 -3 -2 -1 LOG PRESSURE (ATM) Figure 8.20. Lithium extraction from the spodumene vs. pressure at 900°C, Input 8.18. 1 98 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 4 0.0 - 1.0 - 2.0 LU QC 3 m in LU EC Q- 8 -3.0 -4.0 •5.0 900 1000 1100 1200 TEMPERATURE°C Figure 8.21. Pressure of 100% lithium extraction vs. temperature, Input 8.18. Magnesium addition to the system for each simulation would have the effect of decreasing the required pressure, temperature or molar amount of aluminum necessary to obtain 100% recovery. 8.22 for temperature. This is illustrated in Figure Input 8.17 was used with a constant pressure and the temperature was increased from 1025°C to 1275°C by increments of 50°C 99 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 0 Mg (%) 2 Mg 4 Mg Li RECOVERY i----6 Mg — 40 --- B---8 Mg --- K---10 Mg TEMPERATURE C Figure 8.22. The effect o f magnesium to spodumene ratio on the lithium extraction from spodumene with temperature change. I -------------------------------------------The F*A*C*T analysis of the reduction of spodumene by aluminum under various thermodynamic conditions showed that 100% extraction of lithium from the spodumene is possible by either increasing the temperature, increasing the amount of excess molten aluminum or by decreasing the pressure of the system. The amount by which one parameter has to be changed depends on the levels of the other two parameters and the magnesium concentration. This section demonstrated that complicated systems can be simulated and the options available can be narrowed down to the best candidates: without having to obtain and construct the different apparatus, perform a series of experiments and analyze the products. As a result enormous amounts of time, money and effort are saved. More importantly, this section also showed a good agreement in the percent lithium extracted from spodumene obtained experimentally and from a thermodynamic simulation. 100 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. IX. CONCLUSIONS AND FUTURE WORK t IX. 1. CONCLUSIONS The main finding of this study was that magnesium addition to molten aluminum increases the lithium extraction from (3 spodumene in a molten metallo-thermic reduction process. The effect of magnesium addition on the extraction o f lithium from spodumene was similar to calculations made by a computer thermodynamic program, F*A*C*T. The lithium producing reactions at magnesium to spodumene ratios of less than eight to one occurred by a two-step process, Li O A1 0 -4SiO + 2 2 3 2 Li O-Al^O + 9Mg A1 3 = = Li OA1 O 2 2 3 + 4Si + ? A1 O 3 2 [9.1] 3 2Li + MgOAl^O^ [9.2] In the first step the aluminum reduced the silica in the spodumene to form silicon, lithium aluminate, and alumina. In the second step the magnesium reduced the lithium aluminate to form spinel and lithium. At magnesium to spodumene ratios greater than o r equal to eight to one a one-step reaction pathway occurred simultaneously, Li20-A l70 3 4SiO + 9Mg = 2Li + M g0-A l,O + 4Si + 8MgO [9.3] whereby the magnesium reduced the spodumene to form lithium, spinel, silicon, and magnesium oxide. The discovery that pure molten aluminum was able to reduce the silica species in the spodumene indicated that sufficient wetting occurred between the spodumene and the molten aluminum for reaction to occur. Therefore magnesium’s role in the system was thermodynamic in nature. The presence of lithium in the condensed metal vapours indicated that lithium was produced and volatilized. This suggests that at sufficiently low pressures all the lithium present in the molten ingot could be volatilized 101 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. from the ingot. This concurs with the vacuum refining theory o f lithium in aluminum. F*A *C*T predicted that one hundred percent extraction o f lithium from spodumene can be obtained without magnesium addition, by altering the thermodynamic parameters of the system, such as, the amount o f excess aluminum, the pressure or temperature. These parameters could be varied individually or in combination to result in ideal operating conditions. The rate of reaction was high in the first twenty minutes o f the experiments and in that time period the majority of the lithium was extracted from the spodumene Previous claims that lime was an absolutely necessary reactant in the alumino-thermic reduction of spodumene does not apply to a M LE process. The conversion of a to J3 spodumene is a necessity because it produces a more reactive charge material w ith a larger surface area. IX.2. FU TU R E W O RK This first investigation on the metallo-thermic reduction of spodumene was successful in achieving its goals of a reducing spodumene to produce a lithium bearing molten ingot. There is great potential for this process commercially, but more work is necessary for further development. In this section projects and tasks that w ould aid in the further development of this process are proposed below: The theory and the experimentation o f the condensation of pure metal vapours, preferably to the liquid phase. These models would be used in a lithium condensing apparatus in the alumino-thermic reduction of spodumene. T he surface chemistry of (3 spodumene and liquid metals. The vacuum refining of lithium from molten aluminum. The kinetics of the reduction of spodumene by metallo-thermic reduction. 102 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. T he reaction mechanisms between liquid metals and oxide particles and their effect on the kinetics. The process, ideally would not include magnesium as a reactant due to its similar vapour pressure to lithium and its high cost. Therefore an all important task would be the reduction of (3 spodumene by molten aluminum under conditions specified F*A*C*T at which 100% lithium extraction was predicted. A vital task would be to design and construct a reactor suitable for future tests. The reactor would have to: charge all the materials simultaneously in an inert atmosphere or vacuum, have intense mixing between the ore and the molten reductant, control of fine powder emissions, a method o f powder removal from the molten metal, materials unreactive with lithium, and an option for the development of a continuous process. 1 103 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. X. CONTRIBUTIONS This work examined four original topics which were: 1) the production of lithium via the molten metallo-thermic reduction of spodumene. 2) the production of lithium via metallo-thermic reduction at at atmospheric pressure. 3) the reduction of spodumene by molten metallo-thermic reduction. 4) the reduction of an oxide ore by a molten alloy. The findings from this study were extremely useful and provided new information on the chemistry and physical phenomena on MLE processes and more specifically on the spodumene, aluminum and magnesium system. The encouraging results indicate that the development of a process for lithium production from spodumene by molten metal reduction is worth pursuing. 1 I 104 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. REFERENCES R. Harris, A.E. Wraith, and J. Togun, "Producing Volatile Metals", Canadian Patent Application, Ser.No.: 539,058, June 8, 1987, U SA Patent Application, SerNo.. 201,446, June 2, 1988 E.P. Comer " The Lithium Industry Today ”, Energy, 3 (1978) pp.237-240 Ricardo Bach, J.R. Wasson "Lithium and Lithium Compounds",Kirk-Othmer Encyclopedia of Chemical Technology, Vol. 14., New York, Wiley & Sons, 1981, pp.449-476. 4 D. Linden, Battenes and Fuel Cells, McGraw Hill Book Co., Montreal, 1984, p. 11.6. 5 M. Hunt, " New Frontiers in Superlight Alloys ", Materials Eng., Aug (1988), pp 29-32. 6 1 T.R. Crompton, Small Battenes- Primary Cells, The McMillan F*ress Ltd, Vol 2, London, 1986, p. 147 7 David Linden, Handbook of Batteries and Fuel Cells, McGraw H ill Book Co., Montreal, 1984, pp 11.1-11.87. s SR Taylor, "Abundance of Chemical Elements m the Continental Crust: A New Table," Geochimica et Cosmochemica Acta 1^64, Vol. 28, pp.1273-1285. 9 L. Crocker, R H Lien, "Lithium and its Recovery From Low-Grade Nevada Clays", U S Bureau of Mines Bulletin 691, 1987 10 J.D. Vine, "The Lithium Resource Enigma", Lithium Resources and Requirements by the Year 2000, U.S G S. Prof. Paper 1005, 1976. 11 J. Norton, "Lithium Resource Estimates - What Do They Mean?", Lithium Resources and Requirements by the Year 2000, U.S.G.S. Prof. Paper 1005, 1976. 12 J.E. Ferrel, "Lithium", Mineral Facts and Problems, U.S, Bureau o f Mines Bulletin 675, U S. Department o f the Interior, 1985, pp.461-470. 1 105 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 13 J. Searls, "Lithium", Mineral Facts and Problems, U.S. Bureau o f Mines Bulletin 671, U.S. Department o f the Intenor, 1980, pp.521-534. 14 A.T. Kuhn, Industrial Electrochemical Processes, Elsevier, London, U.K., 1971. 15 C.L. Mantell, Electrochemical Engineering, McGraw-Hill, London, U.K., 1960. 16 17 Mining Journal, Vol. 312, No. 8002, A pnl 7, 1989, p.35. I.L. Botto, S Cohen Arazi, T.G. Krenkel, (Fac. Cienc. Exactas. Univ. Nac. La Plata, La Plata, Argentina) Bol Soc Esp. Ceram. Vidrio, 15(1)(1976), p p.5-10. 18 Friednch Liebau, Structural Chemistry o f the Silicates, Springer Verlag, New York, 1985, p.260. 19 Friedrich Liebau, Structural Chemistry o f the Silicates, Springer Verlag, New York, 1985, p.265. 20 Friedrich Liebau, Structural Chemistry o f the Silicates, Springer Verlag, New York, 1985, p.254. 21 L.M. Pidgeon and W A. Alexander, "Thermal Production o f Magnesium: Pilot Plant Studies on the Retort Ferrosilicon Process", Trans. AIME, 159, 1944, pp.315-352. 22 O. Kubaschewski, C.B. Alcock, Metallurgical Thermochemistry, 5th Ed., Pergamon Press, Toronto, pp.212-214. 23 R.H. Parker, An Introduction to Chemical Metallurgy, Pergamon Press, Toronto, 1967, pp 272-273 24 C.J. Kunesh, "Calcium and Calcium Alloys", Kirk Qthmer Encyclopedia o f Chemical Technology, Vol. 14, New York, Wiley & Sons, 1981, pp.414-415. 25 W.J. Kroll, and A W. Schlechten, "Laboratory Preparation o f Lithium Metal by Vacuum Metallurgy", Trans AIME, 182, 1948, pp.266-274. 26 R.A. Stauffer, "Vacuum Process for Preparation o f Lithium Metal from Spodumene", Trans. AIME, 182, 1948, pp 275-285. I 106 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 27 W . Morris and L.M. Pidgeon, "The Vapor Pressure of Lithium in the Reduction of Lithium Oxide by Silicon," Can J. Chem., 36, 1958, pp.910-914. 28 T.F. Fedorov and F.I. Sharmai, "The Physico-Chemical Principles o f the Vacuum Thermite Reduction of Lithium", The Uses of Vacuum in Metallurgy, Oliver & Boyd, Edinburgh, U.K., 1964, pp. 126-132. Mikulinski et al, "Kinetics and Conditions o f Condensation During the Preparation of Alkali Metals by a Vacuum-Thermal Method", Redk. Shchelochnye Elem., Sb D okl Vses Soveechch 2nd. Novosibirsk, 1964, p p .350-360. 30 A.S. Kozhevnikov, " Reduction of Lithium Oxide With Aluminum", Redk. Shchelochnye Elem., Sb D okl. Vses. Soveechch., 2nd. Novosibirsk, 1964, pp 343-349. 31 Mikulinski, A.S. and Efremkin, V.V., "Lithium Ores as Complex Raw Materials", Redk. Shchelochnye Elem , Sb. Dokl. Vses. Soveechch., 2nd. Novosibirsk, 1964, pp.339-342. 32 A.I. Lainer, A. Kh. Nazirov, "Mineral Composition o f Lime Spodumene Sinters", Izv. Akad. Nauk SSSR, Metal. 6(1966), pp.36-39. 33 34 35 B C. Allen, Liquid Metals, Marcel Dekker, New York, 1973, p. 161. R.H. Ewing, Phil Mag. , Vol. 25, 1979. A.M. Korol’kov, "Casting Properties o f Metals and Alloys", Consultants Bureau, 1960, p.37. 36 W .T. Thompson, A.D. Pelton and C.W . Bale, F*A*C*T- Facility for the Analysis of Chemical Thermodynamics, (CRCT) Centre for Research in Computational Thermochemistry, Ecole Polytechnique, Montreal, 1988. 37 G.K. Sigworth, T.A. Engh, "Refining o f Liquid Aluminum - a Review of Important Chemical Factors", Scandinavian Journal o f Metallurgy, 11. 1982, pp. 143-149. 38 CRC Handbook of Chemistry and Physics, 64ih Ed., CRC Press Ltd., Boca Raton, Florida, p.d-46. < 107 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. Richardson, F.D., Physical Chemistry o f Melts in Metallurgy, Vol. 1, Academic Press, New York, 1974, pp. 141-143. R.S. Boy ton, "Chemistry and Technology of Lime and \ imestone", John Wiley & Sons Inc., Toronto, 1980, p.358. Software for Automated Powder Diffraction, APD 1700, Operation Manual, 1st Ed. Philips Corp, The Netherlands, 1984. G.F Bell, "The Analysis of Aluminum Alloys by Means of Atomic Absorption Spectrophotometry", Atomic Absorption Newsletter, Vol. 5, No. 73, 1966, pp.73-76. J. Soters and R. Stux, "Atomic Absorption Methods Manual", Vol. 1, Instrumentation Lab Inc , Wilmington Ma., 1979. Tracor Northern, 2551 West Beltline, Middleton, Wis., 1985. Lotus 123, Release 2, Lotus Development Corporation, SeattleWash., 1985. Peter Hayes, Process Selection in Extractive Metallurgy, Hayes Publishing Co., Brisbane, Aus., 1985, p.276. M.A. Filyand, E.I. Semenova, Handbook of the Rare Elements, Edited by M.E. Alfeneff, Boston Technical Publishers, Cambridge Mass., 1968, p.162. Dimayuga, F., "Vacuum Refining of Molten Aluminum", Ph.D. Thesis, McGill University, 1980, pp. 18-45. H am s, R.L., "Vacuum Refining o f Molten Steel", Ph.D. Thesis, McGill University, 1980, pp. 14-38. Olette, M., "Vacuum Distillation o f Minor Elements from Liquid Ferrous Alloys", Physical Chemistry of Process Metallurgy ; Part 2, Edited by G. St. Pierre, Interscience, N.Y., 1961, pp. 1065-1087. ! 108 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. APPENDIX A The F*A*C*T simulation of the addition of magnesium to the alumino-thermic reduction of spodumene is shown below. The product species identified with ’ were drawn from the private data created for the dissolved species in molten aluminum. The gaseous magnesium identified with a T was extrapolated from its normal temperature range, the value given would be the magnesiums’s partial pressure over the system. T he cutoff concentration was set at 1 x 10'5, species with a concentration or activity less than that were not listed. MgrSpod ratio = 0 (LI20)*(A L203)*(SI102)4 + 85.5 AL + <A> MG = 80.318 ( 0.99720 + 0.17424E-02 + 0.9796 IE-03 ( 1173.0, 1.00 A1 Li Li) ,SOLN 2) + 3.9939 ( 1173.0, 1.00 Si ,S1, 1.0000 ) 2.8124 ( 1173.0, 1.00 A1203 ,S1, 1.0000 ) 1.7814 ( 1173.0, 1.00 LiA102 ,S1, 1.0000 ) + + <— WHERE ’A ’ ON TH E REACTANT SIDE IS 0.000E+00 Mg:Spod ratio = 1 (L I20)*(A L203)*(SI102)4 + 85.5 AL + <A> MG = 81.055 ( 0.99467 + 0.20963E-02 + 0.17409E-02 + 0.97878E-03 + 0.43586E-03 ( 1173.0, 1.00 A1 Mg Li Li Mg) ,SOLN 2) + Si ,S1, 3.9939 ( 1173.0, 1.00 1.0000 < <— ) 109 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. + + 1.7796 ( 1173.0, 1.00 LiA102 ,S1, 1.0000 ) 1.7540 ( 1173.0, 1.00 A1203 ,S1, 1.0000 ) + 0.79476 ( 1173.0, 1.00 (Mg0)(A1203) ,S1, 1.0000 ) + O.OOOOOE+OO ( 1173.0, 1.00 MgO ,S1, 0.22977E-01) CTANT SIDE IS 1.00 Mg:Spod ratio = 2 85.5 AL + <A> MG 81.723 ( 0.99467 + 0.20963E-02 + 0.17409E-02 + 0.97878E-03 + 0.43586E-03 ( 1173.0, 1.00 A1 Mg Li Li Mg) ,SOLN 2) + 3.9938 ( 1173.0, 1.00 Si ,S1, + 1.7931 ( 1173.0, 1.00 (Mg0)(A1203) ,S1, 1.0000 ) + 1.7777 ( 1173.0, 1.00 LiA102 ,S1, 1.0000 ) + 0.42409 ( 1173.0, 1.00 A1203 ,S1, 1.0000 ) + O.OOOOOE+OO ( 1173.0, 1.00 MgO ,S1, 0.22977E-01) WHERE ’A ’ ON THE REACTANT SIDE IS 1.0000 << --- ) 2.00 Mg:Spod ratio = 3 (LI20)*(AL203)*(SI 102)4 + 85.5 AL + <A> MG + O.OOOOOE+OO ( 0.25912E-03 ( 1173.0, 1.00 Mg ,G ,0.284E-03) 82.724 ( 0.98521 + 0.78317E-02 + 0.33649E-02 A1 Mg Li 110 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. T) <— <— « + + 0.18919E-02 0.16284E-02 ( 1173.0, 1.00 Li Mg) ,SOLN 2) + 3.9937 ( 1173.0, 1.00 Si ,S1, + 2.2174 ( 1173.0, 1.00 (Mg0)(A1203) ,S1, 1.0000 ) + 1.5651 ( 1173.0, 1.00 LiA102 ,S1, 1.0000 ) 0.00000E+00 ( 1173.0, 1.00 A1203 ,S1, 0.37036 ) + O.OOOOOE+OO ( 1173.0, 1.00 MgO ,S1, 0.62040E-01) + O.OOOOOE+OO ( 1173.0, 1.00 Mg2Si ,S1, 0.63224E-03) + WHERE ’A’ ON THE REACTANT SIDE IS 1.0000 ) 3.00 Mg:Spod ratio = 4 (LI20)*(A L203)*(SI102)4 + 85.5 AL + <A> MG I + O.OOOOOE+OO ( 83.827 ( 0.97224 + 0.16585E-01 + 0.48968E-02 + 0.34485E-02 + 0.2753 IE-02 ( 1173.0, 1.00 A1 Mg Li Mg Li) ,SOLN 2) + 3.9937 ( 1173.0, 1.00 Si ,S1, + 2.3206 ( 1173.0, 1.00 (Mg0)(A1203) ,S1, 1.0000 ) + 1.3587 ( 1173.0, 1.00 LiA102 ,S1, 1.0000 ) 0.00000E+00 ( 1173.0, 1.00 A1203 ,S1, 0.20958 ) 0.00000E+00 ( 1173.0, 1.00 MgO ,S1, 0.10964 ) + + 0.54874E-03 ( 1173.0, 1.00 Mg ,G ,0.585E-03) 1.0000 T) <-<— ) 111 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. + O.OOOOOE+OO ( 1173.0, 1.00 WHERE ’A’ ON THE REACTANT SIDE IS Mg2Si ,S1, 0.28354E-02) 4.00 Mg:Spod ratio = 5 (LI20)*(AL203)*(SI102)4 + 85.5 AL + <A> MG = + O.OOOOOE+OO ( 84.908 ( 0.95986 + 0.25337E-01 + 0.60523E-02 + 0.5268 IE-02 + 0.34028E-02 ( 1173.0, 1.00 A1 Mg Li Mg Li) ,SOLN 2) + 3.9936 ( 1173.0, 1.00 Si ,S1, + 2.4014 ( 1173.0, 1.00 (Mg0)(A1203) ,S1, 1.0000 ) + 1.1972 ( 1173.0, 1.00 LiA102 ,S1, 1.0000 ) O.OOOOOE+OO ( 1173.0, 1.00 MgO ,S1, 0.15162 ) 0.00000E+00 ( 1173.0, 1.00 A1203 ,S1, 0.15154 ) O.OOOOOE+OO ( 1173.0, 1.00 Mg2Si ,S1, 0.66173E-02) + + + Mg ,G ,0.883E-03) 0.83830E-03 ( 1173.0, 1.00 WHERE ’A’ ON THE REACTANT SIDE IS 1.0000 T) <-<— ) 5.00 Mg:Spod ratio = 6 (LI20)*(AL203)*(SI102)4 + 85.5 AL + <A> MG = + O.OOOOOE+OO ( 85.977 ( 0.94793 + 0.33983E-01 + 0.70658E-02 + 0.70094E-02 + 0.39409E-02. 0.11244E-02 ( 1173.0, 1.00 Mg ,G .0.118E-02) A1 Mg Mg Li Li) 112 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. T) <— ( 1173.0, 1.00 .SOLN 2) 3.9935 ( 1173.0, 1.00 Si ,S1, + 2.4707 ( 1173.0, 1.00 (Mg0)(A1203) ,S1, 1.0000 ) + 1.0585 ( 1173.0, 1.00 LiA102 ,S1, 1.0000 ) + O.OOOOOE+OO ( 1173.0, 1.00 MgO ,S1, 0.19015 ) O.OOOOOE+OO ( 1173.0, 1.00 A1203 ,S1, 0.12083 ) + + WHERE ’A ’ ON THE REACTANT SIDE IS 1.0000 ) 6.00 Mg:Spod ratio = 7 (LI20)*(AL203)*(SI102)4 + 85.5 AL + <A> MG = + 0.14058E-02 ( 1173.0, 1.00 O.OOOOOE+OO ( 87.039 ( 0.93636 + 0.42489E-01 + 0.88344E-02 + 0.78376E-02 + 0.44065E-02 ( 1173.0, 1.00 A1 Mg Mg Li Li) ,SOLN 2) + 3.9934 ( 1173.0, 1.00 Si ,S1, + 2.5329 ( 1173.0, 1.00 (Mg0)(A1203) ,S1, 1.0000 + 0.93427 ( 1173.0, 1.00 LiA102 ,S1, 1.0000 + O.OOOOOE+OO ( 1173.0, 1.00 MgO ,S1, 0.22622 + O.OOOOOE+OO ( 1173.0, 1.00 A1203 ,S1, 0.10157 WHERE ’A’ ON THE REACTANT SIDE IS Mg ,G .0.146E-02) 1.0000 7.00 « 113 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. T) <<— Mg:Spod ratio = 8 (LI20)*(AL203)*(S1102)4 + 85.5 AL + <A> MG = + 0.1682 IE-02 ( 1173.0, 1.00 Mg ,G ,0.174E-02) O.OOOOOE+OO ( 88.097 ( 0.92512 + 0.50840E-01 + 0.10571E-01 + 0.85733E-02 + 0.48202E-02 ( 1173.0, 1.00 A1 Mg Mg Li Li) ,SOLN 2) + 3.9933 ( 1173.0, 1.00 Si ,S1, + 2.5900 ( 1173.0, 1.00 (Mg0)(A1203) ,S1, 1.0000 + 0.82008 ( 1173.0, 1.00 LiA102 ,S1, 1.0000 + O.OOOOOE+OO ( 1173.0, 1.00 MgO ,S1, 0.26037 + O.OOOOOE+OO ( 1173.0, 1.00 A1203 ,S1, 0.88246E-01 + O.OOOOOE+OO ( 1173.0, 1.00 Mg2Si ,S1, 0.26643E-01 WHERE ’A’ ON THE REACTANT SIDE IS T) <— <— 1.0000 8.00 MgrSpod ratio = 9 (LI20)*(AL203)*(SI102)4 + 85.5 AL + <A> MG = + 0.19531E-02 ( 1173.0, 1.00 Mg ,G .0.202E-02) O.OOOOOE+OO ( 89.150 ( 0.91419 + 0.59030E-01 + 0.12274E-01 + 0.9238 IE-02 + 0.51939E-02 ( 1173.0, 1.00 A1 Mg Mg Li Li) ,SOLN 2) + 3.9933 ( 1173.0, 1.00 Si ,S1, 2.6433 (Mg0)(A1203) + 1.0000 T) <— <— ) 114 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. ( 1173.0, 1.00 ,S1, 1.0000 ) 0.71338 ( 1173.0, 1.00 LiA102 ,S1, 1.0000 ) O.OOOOOE+OO ( 1173.0, 1.00 MgO ,S1, 0.29298 ) + O.OOOOOE+OO ( 1173.0, 1.00 A1203 ,S1, 0.78426E-01) + O.OOOOOE+OO ( 1173.0, 1.00 LiAl ,S1, 0.76319E-01) + O.OOOOOE+OO ( 1173.0, 1.00 M g2Si ,S1, 0.35918E-01) + + WHERE ’A’ ON THE REACTANT SIDE IS 9.00 MgrSpod ratio = 10 (LI20)*(AL203)*(SI 102)4 + 85.5 AL + <A> MG + 0.22187E-02 ( 1173.0, 1.00 Mg ,G .0.229E-02) O.OOOOOE+OO ( 90.201 ( 0.90354 + 0.67057E-01 + 0.13943E-01 + 0.98462E-02 + 0.55359E-02 ( 1173.0, 1.00 A1 Mg Mg Li Li) ,SOLN 2) + 3.9932 ( 1173.0, 1.00 Si ,S1, 2.6937 ( 1173.0, 1.00 (Mg0)(A1203) ,S1, 1.0000 ) 0.61253 ( 1173.0, 1.00 LiA102 ,S1, 1.0000 ) O.OOOOOE+OO ( 1173.0, 1.00 MgO ,S1, 0.32427 ) + O.OOOOOE+OO ( 1173.0, 1.00 A1203 ,S1, 0.70858E-01) + O.OOOOOE+OO ( 1173.0, 1.00 Mg2Si ,S1, 0.46352E-01) + + + W H E R E ’A ’ O N THE R E A C T A N T SID E IS 1.0000 10.0 115 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. T) <— <— ) APPENDIX B In this study the reductant was a mixture o f molten aluminum with dissolved magnesium, lithium and silicon. Figures B.l, B.2, and B.3 show the cs phase diagrams for aluminum with magnesium, lithium and silicon respectively, with the intention of showing that at 900°C and the weight percents incurred, that no precipitation of intermetallic species would occur. 20 700 Weight P e r c e n t M a g n e s i u m BO 60 ion 600 0U 500 uv (At) 3 u 400 V a E 41 300 200- 100 20 A1 Figure B.l. 30 40 50 60 7C BO Atomic Percent Magnesium 100 Mg Aluminum-magnesium phase diagram T. Massalski, Ed., "Binary Alloy Phase Diagrams", AS ME, Metals Park, Ohio, 1985. ? 116 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. W eight P ercent L ith iu m ao 800 30 40 too 50 700 - 600 (AI) u O ) 500U 3 C 3 ud> o e£ 400 tu t- 330°C 300 200 (Ll) 100 0 A1 Figure B.l. 10 20 30 40 50 60 70 eo 90 A tom ic Percent L ith iu m Aluminum-Lithium phase diagram 117 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 100 Li W eight P e rc e n t S ilic o n eo 20 1500 i too 1300 0 CD U 3 co 1100 900 0CL> sa> 5 0 o jf < A‘) 300 20 A1 Figure B.3. (S i) ^ 40 50 60 A to m ic Percent Silicon Aluminum-silicon phase diagram. 118 Reproduced with permission of the copyright owner. Further reproduction prohibited without permission. 90 too