Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Kinetics of carbothermic reduction of natural aluminosilicates Wong, Siu Chung 1995

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

Item Metadata

Download

Media
831-ubc_1995-0652.pdf [ 6.91MB ]
Metadata
JSON: 831-1.0078529.json
JSON-LD: 831-1.0078529-ld.json
RDF/XML (Pretty): 831-1.0078529-rdf.xml
RDF/JSON: 831-1.0078529-rdf.json
Turtle: 831-1.0078529-turtle.txt
N-Triples: 831-1.0078529-rdf-ntriples.txt
Original Record: 831-1.0078529-source.json
Full Text
831-1.0078529-fulltext.txt
Citation
831-1.0078529.ris

Full Text

KINETICS OF CARBOTHERMIC REDUCTION OF NATURAL ALUMINOSILICATES by SIU CHUNG WONG B.A.Sc . ( M M A T ) , The University of British Columbia, 1990 A T H E S I S S U B M I T T E D IN P A R T I A L F U L F I L L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F M A S T E R O F A P P L I E D S C I E N C E in T H E F A C U L T Y O F G R A D U A T E S T U D I E S (Department of Metals and Materials Engineering) We accept this thesis as conforming to the required standard T H E U N I V E R S I T Y O F B R I T I S H C O L U M B I A October 1995 © Siu Chung Wong, 1995 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of (NPUS *«J * ^"f"1 *J, The University of British Columbia Vancouver, Canada Da,e /j ()cf /??<; DE-6 (2/88) Abstract Thermogravimetric analysis was used to investigate the kinetics of the carbothermic reduction of aluminosilicates between 1450 and 1550°C. The majority of the experiments involved the reduction of meta-kaolin, mullite or silica with lamp black within an argon atmosphere This dissertation focused on the mechanism and the rate limiting step involved in the reduction process. Also investigated were various aluminosilicates, reductants and reactor atmosphere to identify less expensive sources of raw materials. The carbothermic reduction of aluminosilicates is believed to be simply the reduction of mullite and silica after their formation at the reaction temperatures. Thermodynamics show that only the silica phase will be reduced within the temperature range of this study. The reduction mechanism of aluminosilicates is therefore based on the reduction mechanism of silica. This mechanism can be described as follows: S i 0 2 + C = SiO ( g ) + CO Step (1) SiO (g) + 2C = SiC + CO Step (2) From examination of the mechanism, the two possibilities for the rate limiting step are the SiO formation step and the SiC formation step. Experiments involving temperature effect, particle size effect, Si and SiC seeding effect, reactor atmosphere effect (CO(g> and N2(g)) and various combinations of precursor powders were carried out. The collected weight loss data are presented as conversion, X, versus time, t, and conversion rate, , versus time, t. The activation energies for the different reaction mixtures dt have been determined and compared with values from literature. Product powders were n characterized by X R D (X-ray diffraction), particle size analysis and S E M (scanning electron microscopy). The two step mechanism is supported by experimental evidence. However, the results of this study are not conclusive in defining the rate limiting step. The rate data are fitted to various rate models, the Contracting Volume Model (phase boundary control) and Jander's equation (diffusion control through a product layer) provide the best overall fit. The activation energies have been determined to be between 260 and 290 kJ/mole for aluminosilicates (meta-kaolin and mullite) reduction and approximately 360 to 420 kJ/mole for silica reduction. The activation energy values for silica reduction agree with published data. The activation energies for mullite and meta-kaolin reduction are approximately 100 kJ/mole lower than those for silica reduction. The alumina phase of the aluminosilicates may play an important role in this difference. A study of alkali aluminosilicates reduction has been carried out to examine the possibility of using these materials as precursors for low technology applications. The reduction rates of bentonite, illite, mica and syenite are similar to meta-kaolin. The study of various carbon reductants shows that graphite and coke may be used in place of lamp black. The use of nitrogen as a purge gas has been examined and the results show that although SiC is retained in the product powders, its yield is low. iii TABLE OF CONTENTS Abstract » Table of Contents : iv List of Figures v i i List of Tables x Table of Nomenclature x i Acknowledgement x i i 1.0 A D V A N C E D C E R A M I C S 1 1.1 CERAMIC COMPOSITES 2 1.2 COMMERCIAL REALIZATION 5 1.3 PREVIOUS RESEARCH ON POWDER PRODUCTION METHODS 7 1.3.1 Aluminosilicates 8 1.3.2 Impurities 9 2.0 C A R B O T H E R M I C R E D U C T I O N O F A L U M I N O S I L I C A T E S 11 2.1 THERMOCHEMISTRY 12 2.2 KINETICS 18 2.2.1 Mechanism of Aluminosilicate Reduction 19 2.2.1.1 Mechanism of Mullite Formation 20 2.2.1.2 Mechanism of Silica Reduction 22 2.2.1.3 Carbon Source 26 2.3.2.4 Role of CO® : • 27 2.2.1.5 Role of Si : 29 2.2.1.6 Role of SiC 29 222 Determination of the Rate Limiting Step 30 2.3 OBJECTIVES 35 iv 3.0 E X P E R I M E N T A L 3.1 APPARATUS 36 3.1.1 Crucible 39 3.1.2 Reaction Atmosphere 39 3.2 RAW MATERIALS 4? 3.3 EXPERIMENTAL PROCEDURE 3.3.1 Powder Preparation.... 3.3.2 Procedure for a Reduction Experiment Run 43 4.3.2 Powder Characterization 45 3.4 D A T A PROCESSING 46 3.5 PRELIMINARY STUDIES 48 3.5.1 Reproducibility Study 48 3.5.2 Removal of the Water Loss Factor 49 4.0 R E S U L T S 51 4.1 EXPERIMENTAL CONFIGURATION 51 4.1.1 Batch Size Effect 51 4.1.2 Stoichiometry Effect 52 4.1.3 Purge Gas Flow Rate Effect 53 4.2 PRODUCTPOWDER 54 4.2.1 XRD Results 54 4.2.2 Morphology and Particle Size 57 4.2.3 Impurities 64 4.3 ALUMINOSILICATE VERSUS SILICA REDUCTION 65 4.4 REACTION PARAMETERS 67 4.4.1 Direct Reduction with CO 67 4.4.2 The Effect of Particle Size 67 4.4.3 Si and SiC Seeding 69 4.5 REDUCTION OF VARIOUS ALUMINOSILICATES 71 4.6 REDUCTION USING VARIOUS CARBON SOURCES 76 4.7 N 2 REACTION ATMOSPHERE 81 5.0 D I S C U S S I O N 83 5.1 ANALYSIS OF THE REDUCTION MECHANISM 83 5.1.1 Mullite Reduction Mechanism 83 5.1.2 Silica Reduction Mechanism 84 5.2 DETERMINATION OF THE RATE LIMITING STEP 85 5.3 DETERMINATION OF ACTIVATION ENERGY 87 5.4 ALTERNATIVE RESOURCES FOR A L 2 0 3 - S I C POWDER PRODUCTION 89 5.4.1 Various Aluminosilicate Clays and Minerals 89 5.4.2 Purge Gases 90 6.0 S U M M A R Y A N D C O N C L U S I O N 91 7.0 F U R T H E R S T U D I E S 94 B I B L I O G R A P H Y 95 E X P E R I M E N T A L S U M M A R Y 100 vi LIST OF FIGURES Figure 1: Levels of Ceramic Composites [47] . 3 Figure 2: Fracture Strength versus Temperature [34] i 4 Figure 3: Rational Analysis of Clays [48] 10 Figure 4: Equilibrium pressures of C O for silica reduction and alumina reduction [6] 13 Figure 5: Gibbs Free Energy of Reactions 1-19 [27] 16 Figure 6: Mull i te Formation from Kaolinite [14] 22 Figure 7: Two Step Reduction Mechanism [65] 24 Figure 8: Reactivity of Activated Carbon, Graphite and Carbon Black in the Reduction of Silica [55] 27 Figure 9: Silica Reduction using C O ( g ) (pCo = 1.5 atm) [6] 28 Figure 10: Thermogravimctric Analysis Apparatus 37 Figure 11: Gas System 38 Figure 12: Crucible Design 39 Figure 13: Data Representation: (a) Raw Data, (b) Converted Data, (c) Converted and Interpolated Data, (d) Interpolated Data 48 Figure 14: Reproducibility Runs for Meta-Kaolin Reduction at 1530°C 49 Figure 15: Weight Loss for Kaolinite Reduction at 1500°C 50 Figure 16: Batch Size Effect for Kaolinite Reduction at 1450°C (a) Conversion versus Time and (b) Conversion Rate versus Time 52 Figure 17: Stoichiometry Effect for Silica Reduction at 1450°C (a) Conversion version Time (b) Conversion Rate versus Time 53 Figure 18: Conversion versus Time for Silica Reduction at 1530°C at Argon Flow Rates of 11, 22, 46 and 94 ml/s 54 Figure 19: X R D of Meta-Kaolin/Carbon Reduction Products 55 Figure 20: X R D of Silica/Carbon Reduction Products 56 Figure 21: S E M photomicrograph (5000X) of Meta-Kaolin 57 Figure 22: Particle Size Analysis for Meta-Kaolin 58 vii Figure 23: S E M Photomicrograph (5000X) of Silica Flour. 59 Figure 24: Particle Size Analysis of Silica Flour. 59 Figure 25: S E M Photomicrograph (2000X) of Lamp Black 60 Figure 26: S E M Photomicrographs of Meta-kaolin/Carbon Powders (a) Precursor and (b) Product (AI 2 0 3 -S iC) 61 Figure 27: Particle Size Distribution of the A l 2 0 3 - S i C Composite Powders from Meta-Kaolin Reduction 62 Figure 28: S E M Photomicrographs of Silica/Carbon Powders, (a) Precursor and (b) Product (SiC) 63 Figure 29: Meta-Kaolin, Mullite and Silica Reduction at 1450°C (a) Conversion versus Time (b) Conversion Rate versus Time 65 Figure 30: Meta-Kaolin, Mulli te and Silica Reduction at 1550°C (a) Conversion versus Time (b) Conversion Rate versus Time 66 Figure 31: Temperature Effect for Meta-Kaolin Reduction (a) Conversion versus Time (b) Conversion Rate versus Time 66 Figure 32: The Effect of Silica Particle Size on Conversion Rate (Reacted at 1450°C) (a) Conversion versus Time (b) Conversion Rate versus Time 68 Figure 33: The Effect of Kyanite Particle Size on Conversion Rate (Reacted at 1530°C) (a) Conversion versus Time (b) Conversion Rate versus Time 69 Figure 34: Conversion versus Time with or without Si Seeding (Reacted at 1530°C in A r Atmosphere).... 70 Figure 35: S i ( s ) Reaction with C O ^ at 1380°C 70 Figure 36: SiC Seeding at 1530°C (a) Conversion versus Time (b) Conversion Rate versus Time 71 Figure 37: Reduction of Various Aluminosilicates at 1530°C (a) and (b) Conversion versus Time (c) and (d) Conversion Rate versus Time 73 Figure 38: S E M Photomicrograph (6000X) of Product Powder from Bcntonite Reduction 74 Figure 39: S E M Photomicrograph (6000X) of Product Powder from Feldspar Reduction 74 Figure 40: S E M Photomicrograph (6000X) of Product Powder from Illite Reduction 75 Figure 41: S E M Photomicrograph (6000X) of Product Powder from M i c a Reduction 75 Figure 42: S E M Photomicrograph (6000X) of Product Powder from Syenite Reduction 76 Figure 43: Silica Reduction with Graphite and Lamp Black at 1530°C (a) Conversion versus Time (b) Conversion Rate versus Time 77 Figure 44: Meta-Kaolin Reduction with Coke and Lamp Black at 1530°C (a) Conversion versus Time (b) Conversion Rate versus Time 77 Figure 45: Meta-Kaolin Reduction with Coke Particles Sizes of 250 and 50 um at 1530°C (a) Conversion versus Time (b) Conversion Rate versus Time 78 Figure 46: S E M Photomicrographs of the 50 jim Coke Particles (a) 2000X and (b) 4000X 79 Figure 47: S E M Photomicrographs of the 250 um Coke Particles (a) 200X and (b) 4000X 80 Figure 48: X R D Plot of the Silica Reduction Product Powder Reacted in Nitrogen Atmosphere at 1430°C 81 Figure 49: X R D Plot of the Silica Reduction Powder Reacted in Nitrogen Atmosphere at 1530°C 82 Figure 50: Conversion versus Time for Meta-Kaolin/Carbon Reduction at 1510°C 84 Figure 51: Reduction Models for (a) Silica Reduction (b) Meta-Kaolin Reduction both at 1530°C 85 Figure 52: Meta-Kaolin Reduction at 1550, 1510,1470 and 1450°C are plotted using the Contracting Volume Model 86 Figure 53: Reaction Rate versus Inverse of Average Particle Radius of Powder (a) Silica Reduction and (b) Kyanite Reduction 87 Figure 54: In k versus 1/T plots for (a) Meta-Kaolin (b) Mulli te and (c) Silica Reduction 88 ix LIST OF TABLES Table 1: Comparison of Fracture Toughness Values [41] 4 Table 2: Relative Cost of the Manufacturing Process (1986) [44] 5 Table 3: Ceramic Powder costs (1993) 6 Table 4: Potential Raw Materials for A l 2 0 3 - S i C Composite Synthesis 9 Table 5: Relative Costs of Various Purge Gases (1992) 40 Table 6: Supplier and Particle Size of Raw Powders 42 Table 7: Weight Loss at 100% Conversion for Various Precursor Powders 47 Table 8: Actual versus Theoretical Weight Loss in Reduction Experiments 56 Table 9: Impurity Concentrations of Precursor and Product Powders [12] 64 Table 10: Comparison of Activation Energies for Silica, Mullite and Meta-Kaolin Reduction 89 X TABLE OF NOMENCLATURE PCO Partial pressure of carbon monoxide Pco2 Partial pressure of carbon dioxide pSiO Partial pressure of silicon monoxide AG Gibbs free energy T Temperature P Pressure (P Number of phases (Phase Rule) 3 Degrees of freedom (Phase Rule) C Number of components (Phase Rule) A Constant/Surface area of a powder compact / Time AE Activation energy Q Experimental activation energy R Gas constant C Concentration k Experimental reaction rate r Average particle radius P Number of steps involved in nucleus formation X Number of dimension in which nuclei grow p Density X Conversion dX r~ Conversion rate dt ffl0 Initial sample weight 171 Sample weight at time t W Weight at 100% conversion/Weight of powder compact xi A C K N O W L E D G E M E N T The author acknowledges the guidance and encouragement provided by his supervisor, Dr. A . C . D . Chaklader (Professor Emeritus) throughout the Master Degree program. The directions and insights given by Dr. G . G . Richards of Cominco Research are also gratefully acknowledged. Special thanks goes to Mr. Edmond Y . C . Lin, a former researcher with the Ceramics Group, for his technical assistance and support. The help provided by other members of the Department of Metals and Materials Engineering as well as fellow graduate students is also greatly appreciated. Financial assistance for this thesis work comes from the Natural Sciences and Engineering Research Council of Canada and the University Graduate Fellowship. xii 1.0 Advanced Ceramics Advanced ceramics are a class of inorganic, non-metallic materials which are processed or consolidated at high temperatures [44]. This group of materials consists mainly of oxides, carbides, nitrides and borides. Advanced ceramics have the potential to lead the way in technological progress by making products and processes more competitive, more reliable and less expensive. In many structural applications, especially in hostile environments, advanced ceramics possess unique properties which can outperform metals and polymers. The inherent properties which make advanced ceramics highly desirable are as follows: • High compression strength and high stiffness. • Retained high strength at elevated temperatures due to high melting point. • High hardness for wear resistance. • Corrosion resistance and oxidation resistance (oxide ceramics). • Low density (high strength-to-weight ratio). Unfortunately, there are difficult fundamental obstacles to overcome before advanced structural ceramics are commercially available. The key mechanical deficiencies of ceramics are: • Susceptibility to thermal and mechanical shock (brittleness) leading to catastrophic failures. • Negligible ductility, therefore poor resistance to growth of small flaws (ie. poor damage tolerance). A number of solutions to these material limitations of ceramics are under development: • Improved design practice to minimize the limitations imposed by the use of ceramics. For instance, use ceramics under compressive loading. • Processing of monolithic ceramics with controlled microstructure from highly pure, nanosized and uniform particle size ceramic powders. A submicron 1 microstructure in the ceramic product lessens the impact of flaws on its properties, as the grain size is reduced to the size of the flaws. • Development of new ceramic compositions. For example, the stress induced microstructural transformation of PSZ (partially stabilized zirconia). Under high stresses which exist at a crack tip, the P S Z undergoes a martensitic transformation which involves a volume change leading to the blunting of the crack tip and the slowing of crack propagation. • Multiphase ceramic composites. The last solution holds the most promise in manufacturing materials of adequate mechanical properties. This dissertation by the author is meant to be a small element in the current ceramic research field which will eventually bring advance ceramics to commercial realization. 7.7 Ceramic Composites Ceramic composites are defined as a class of materials whereby one or more ceramic materials are added to another in order to enhance or provide some property not possessed by the original material [17]. The second phase reinforcing elements are usually in the form of particulates, whiskers or fibers. The technical difference between whiskers and fibers is one of scale; the diameter of whiskers are in the micrometer range while that of fibers is significantly larger. The most common particulate reinforcement materials are Z r 0 2 and SiC. A broad range of ceramic materials are used in whisker and fiber reinforcements, with graphite (carbon-carbon composites) and SiC being the two most common [17]. There are three levels which define ceramic composites (Figure 1) 1. Macrocomposites are commonly laminated structures similar in construction to polymeric composites. 2 2. Microcomposites are composites where the reinforcement phase is in the micrometer size range. The reinforcement phase is mostly located at the grain boundaries of the matrix phase. 3. Nanocomposites are composites where the reinforcement phase is in the nanometer size range. The reinforcement phase is evenly dispersed within the matrix grains. Nanocomposites are the limit to which different materials can be intermixed without the formation of a new composition. Researchers have found that the nanocomposites possessed greatly improved mechanical properties when the reinforcement phase has dimensions of 100 nm or less [52]. O f the three levels of composites, the fabrication of components using nanocomposites is the easiest to form and exhibits the least shrinkage during sintering. Lamella Microcomposite Nanocomposite Figure 1: Levels of Ceramic Composites. [47] Ceramic composites possess definite mechanical advantages over monolithic ceramics. • Increased reliability and durability. • Reduced notch and flaw sensitivity. • Increased toughness and improved damage tolerance due to the reinforcement phase (Table 1) • Increased high temperature strength. Al 2 03-SiC composites possess the highest strength in the temperature range up to 1300°C (Figure 2). • Non-catastrophic (benign) failure. 3 Material Fracture Toughness (MParn"2) Pyrex glass 0.75 Common wood parallel to grains 0.5-1 Alumina 2.7-4.2 S i 3 N 4 4-6 P S Z 8-9 SiC whiskers/AI2O3 composites 9-10 Common wood perpendicular to 11-13 grains High strength steel 120-153 Table 1: Comparison of Fracture Toughness Values. [41] 0 700 1300 1500 Temperature ("C) Figure 2: Fracture Strength versus Temperature. [34] The reasons for the development of ceramic-ceramic composites are numerous. The usefulness of ceramic materials' unique properties often outweigh their deficiencies. A n important first step is to manufacture ceramic composites economically and in industrial quantities. 4 1.2 Commercial Realization There are a number of production costs which have prevented ceramic composites from becoming widely available. • Cost of powders. • Labour costs. • Fabrication costs. • Extensive and difficult finishing methods. • Inspection and testing methods. The relative cost of each of the above steps is given in Table 2. Step Relative Cost Powder 1 Processing 2-3 Finishing 7-10 Testing 1-2 Table 2: Relative Cost of the Manufacturing Process (1986). [44] Many of the processing and manufacturing costs can be reduced once experience and understanding have been gained on how to handle these materials. The per unit cost will decrease when economy of scale from mass production is put into practice. Near net shape fabrication methods will significantly reduce the finishing cost which is currently the most expensive manufacturing step. Two significant problems in the production of advanced ceramics are: • Inconsistent quality and poor reproducibility during manufacturing. • High rejection rate and poor reliability in service due to difficulty in producing a flaw-free product. 5 The above problems are directly related to precursor powder quality. The outcome of the ceramic processing and fabrication is influenced to a great extent by the nature and condition of the precursor powders. Almost all ceramic components are fabricated from powders. Due to the need for favourable microstructure in the final products, tight control must be placed over the powder chemistry and processing. This means that the following powder criteria must be maintained: • High purity. • Consistent and stable chemical composition. • Free of hard agglomerates which leads to nonuniform microstructures. • Uniform particle size and shape. • Narrow size distribution for even sintering/consolidation characteristics. • Particle sizes < l u m for high reactivity. The presently available commercial powders which meet these specifications are limited in supply and are very expensive. The costs of some of these powders are listed in Table 3. Ceramic Powder $US/kg SiC 30 Si3N4 30 A1203 26 Zr0 2 -Y 2 0 3 58 SiC whiskers from rice hulls 100 SiC whiskers from CVD 150-200 Table 3: Ceramic Powder costs (1993). Costs from The Electrofuel Manufacturing Company, Toronto, Canada. Presently, the two most common methods of producing ceramic composite powders are through powder mixing or by in situ reaction. To obtain an even distribution of the second phase through powder mixing, it requires complex processing techniques such as fluid phase mixing and ultrasonic dispersion. It is difficult to mix very fine nanosized particles and most techniques do not result in a consistently homogeneous powder. The second route is widely 6 studied and includes techniques such as C V D (chemical vapour deposition), sol-gel methods, polymer precursors and other chemical routes. In the case where the second phase is whiskers, there is the added health hazard associated with these materials [12]. When whiskers are grown in situ, this problem is removed. None of the methods mentioned are commercially successful due to poor production rates and high costs. 1.3 Previous Research on Powder Production Methods Low cost synthesis methods for ceramic powders have been tried and successfully used. These production methods involved the use of common, less expensive, natural raw materials as the raw ingredients for the product powders. Some examples are: • SiC powder synthesis by carbothermic reduction of rice hulls (ie. husks) [37]. Rice hulls, which are waste products of the agricultural industry, are composed of 15 to 20% silica with the remaining content consisting of cellulose. The cellulose thermally decomposes to carbon during the pre-reduction step. This carbon then reduces the silica content to form SiC. • S i 3 N 4 powder synthesis by the carbothermic reduction and nitridation of pure silica sand [2, 20, 25]. • Ceramic composite powders using natural minerals. Recent attempts were made to synthesize ceramic composite powders by the carbothermic reduction of natural minerals such as kaolinite, kyanite, etc. In a study by Bechtold and Cutler [4], an A k C V S i C mixture was produced by the carbothermic reduction of kaolinite with carbon. Their purpose was to extract the alumina phase from the product mixture. In their system the clay/carbon mixture was pelletized and surrounded by carbon in the reaction vessel. The silica volatilized as SiO(g) into the surrounding carbon where it reduced to SiC. The pellets were therefore purified of the silica, leaving a porous alumina structure. 7 The above processes all followed the same basic carbothermic reduction steps in deriving their products. Previous studies have focused on the material properties of the products rather than investigating the steps involved in the synthesis process. The conclusions drawn by researchers describing the mechanism of the process and the rate determining step were often contradictory. The use of aluminosilicates to produce an A l 2 0 3 - S i C composite powder was never seriously considered by researchers. Bechtold and Cutler [3] used the carbothermic reduction process to extract alumina from kaolinite. Their goal was to secure a cost-effective production method for highly pure alumina which did not depend on the supply of bauxite ore. They did not look into the possible use of an Al 203-SiC composite powder because they believed the product powders contained too high an impurity content for practical ceramic applications. Kimura et al. [33] carried out limited studies of the carbothermic reduction of sericite, an alkali aluminosilicate. There have been no further reports on their research. Studies by Chaklader et al. [12, 13] on the carbothermic reduction of natural aluminosilicates have only explored the characteristics of the product powders plus the effect of clay impurities on powder quality. A broad kinetic study of the carbothermic reduction reaction based on clay aluminosilicates has not been done at the time of the writing of this dissertation. 1.3.1 Aluminosilicates The present knowledge of aluminosilicate clays and minerals is based on the long history of their use in traditional ceramics. Aluminosilicates are clays or minerals structurally composed of silica and alumina phases. The clays are hydrous formations made up of sheet-like layers separated by water molecules. Other aluminosilicate minerals are non-hydrous, three-dimensional structures. Table 4 is a list of some of the possible aluminosilicates which can be used in the carbothermic reduction process. Kaolinite is an appropriate choice for the mass 8 production of ceramic composite powders. Because of its widespread use in the traditional ceramic and pulp and paper industries, kaolinite is readily available in large quantities and of high quality (ie. low impurity content). C H E M I C A L COMPOSITION ALUMINOSILICATES 1. Hydrates • Kaolinite *, Nacrite, Dickite Al 2 Si 2 0 5 (OH) 4 • Halloysite Al 2 Si 2 0 5 (OH) 4 ;xH 2 0 • Pyrophyllite Al 2 Si 4 O 1 0 (OH) 2 2. Non-Hydrates • Kyanite *, Sillimanite, A l 2 S i 0 5 Andulusite • Mullite * A l 6 Si 2 0 i 3 A L K A L I ALUMINOSILICATES • Bentonite Al,.67(Nao.33 orMg 0.33)(Si 2O 5) 2(OH) 2 • Feldspar K A l S i 3 0 8 • Illite A l 2 . x Mg* KUX.Y (Siu., Al,o.5^ ) 2(OH) 2 • Mica Al 2K(Si,.5A,o.jO s) 2(OH)2 • Syenite K 2 O.Al 2 0 3 . (5-6)Si0 2 Table 4: Potential Raw Materials for A l 2 0 3 - S i C Composite Synthesis * Aluminosilicates used for this study. 1.3.2 Impurities The impurity content of clays and minerals is important in determining the quality of the ceramic composite powders produced by the reduction process. The kaolinite used by the traditional ceramic industries undergoes basic screening and centrifuging to remove organics and dense non-aluminosilicate particles. Common oxide impurities associated with aluminosilicates are Fe203, T i 0 2 , MgO, CaO and the alkalies, K 2 0 and Na 2 0. The major mineral impurities (Figure 3) of aluminosilicates are quartz, anatase, rutile, pyrite, limonite, feldspar, mica and montmorillonite. The effect of impurities on material properties of the composite powders produced by carbothermic reduction has been the main argument against their use in high technology applications. Kaolin A Florida Unidentified Kaolin B North Carolina Ball Clay C Tennessee Ball Clay 0 Tennessee Kaolinite plus — _ Halloysite Loosely Comb. Water Mica Organic Quartz Gibbsite Unidentified Kaolinite -Loosely Comb. Water Montmorillon'rte; Organic pJu Sulfide Unidentifie Kaolinite -Loosely Comb. Water Montmorillonrtel Organic plus' Sulfide Quar t i—— Unidentified-'3 Figure 3: Rational Analysis of Clays [48] 10 2.0 Carbothermic Reduction of Aluminosilicates The overall reaction for the carbothermic reduction of kaolinite is as follows: Al2O3.2SiO2.2H2O + 6C = AI2O3 + 2SiC + 4 C O + 2 H 2 0 or with excess alumina: Al2O3.2SiO2.2H2O + x A l 2 0 3 + 6C = ( l+x)Al 2 0 3 + 2SiC + 4 C O + 2 H 2 0 or with excess silica: Al2O3.2SiO2.2H2O + v S i 0 2 + (6+3v)C = AI2O3 + (2+y)SiC + (4+2.y)CO + 2 H 2 0 The concentration of SiC in the product powder can be controlled by the addition of Si02 and AI2O3 to the precursor powder. Control of the phases in the product powder allows the manufacturer to produce consistent powders with the optimum ratio of the two phases while using kaolinite from various sources. It is believed that the carbothermic reduction of kaolinite can be divided into two general stages: 1. Mullite formation with the release of H 20( g): 3 [Al 2 0 3 . 2S i0 2 . 2H 2 0] = 3 A l 2 0 3 . 2 S i 0 2 + 4 S i 0 2 + 6 H 2 0 2. Reduction of mullite and silica 3 A l 2 0 3 . 2 S i 0 2 + x S i 0 2 + (2+x)C = 3 A1 2 0 3 + (2+x)SiO(g) + (2+x)CO ( g ) (2+x)SiO ( g ) + (4+2x)C = (2+x)SiC + (2+x)CO ( g ) 11 Mullite formation occurs first, followed by the reduction of the free silica and the silica associated with mullite. The theoretical weight loss for the complete reduction of kaolinite is 44.8%. In the case of kyanite (A l 2 0 3 .S i0 2 ) , carbothermic reduction follows a similar path, with mullite formation being the first step: 3 (Al 2 0 3 .S i0 2 ) = 3 A l 2 0 3 . 2 S i 0 2 + S i 0 2 then, the reduction of the free silica and silica within the mullite phase: 3 A l 2 0 3 . 2 S i 0 2 + S i 0 2 + 3C = 3 A l 2 0 3 . 2 S i 0 2 + SiC + 2 C O 3 A l 2 0 3 . 2 S i 0 2 + 6C = 3A1 2 0 3 + 2SiC + 4 C O The theoretical weight loss for the complete reduction of kyanite is 28.3%. 2.1 Thermochemistry In the carbothermic reduction of aluminosilicates, there are two phases which may be reduced; the alumina phase or the silica phase. It has been shown by Bentsen et al. [6] (Figure 4) that the reduction of silica or mullite below 1600°C is more favoured than the reduction of alumina at those temperatures. From Figure 4, the higher value of pco indicates a greater instability of the phase and the more likelihood of it reacting at the specified temperature. Notice that the instability of mullite is slightly lower than silica due to the fact that the activity of silica within mullite is lower because of the bonding with the alumina phase. 12 - 4 I i i I 1400 1500 1600 1700 T e m p e r a t u r e ( C e l s i u s ) Figure 4: Equilibrium pressures of CO for silica reduction and alumina reduction. [6] The overall reaction for silica reduction is as follows: Si02(s) + 3C(S)= SiC(S) + 2CO(g) The above reaction is highly endothermic as one would expect for the reduction reaction of a highly stable oxide material [64]. The reaction's standard free energy value is determined by the following equation: AG° = 609.023-0.35 ir(kJ/mole) The extent of reaction is dependent on both temperature and pco- At reaction temperatures, the equilibrium partial pressures of O2, Si(g), Q g), C0 2 ( g ) and SiC(g) are orders of magnitude lower than the equilibrium partial pressures of SiO(g) and CO ( g ) . The value of pco^ was determined by Biernacki and Wotzak [7] to be 2 orders of magnitude lower than pCo and ps,o- The value of psio was 3 to 4 orders of magnitude greater that psi indicating that SiO ( g ) was more likely to exist 13 than Si(g). The species which are most likely to be involved in the reduction mechanism are Si(s>i), C(S), SiQS), SiO^i), SiO(g) and C O ( g ) . Listed below are the reactions from the Si-O-C system which researchers [5, 6, 50, 65] believe to be significant in the carbothermic process: 1. Basic Gas Reactions: C 0 2 ( g ) + C ( s ) = 2CO ( g ) (1) C O ( g ) + ^ 0 2 ( g ) = C 0 2 ( g ) (2) Q s ) + ~ 02(g) = CO(g) (3) 2. Dissociation Reactions: 3. Formation of SiO 4. Reduction of SiO SiC ( s ) - Si(Sji) + C(S) (4) SiO(g) = S i W ) + ! 02(g) (5) Si0 2 ( S j l ) = S i O ( g ) + ^ 0 2 ( g ) (6) Si02(s>,) + C ( l ) = SiO(g) + C O ( g ) (7) Si0 2 ( s > 1 ) + C 0 ( g ) = SiO { g ) + C 0 2 ( g ) (8) Si02 ( s,D + S i W ) - 2SiO ( g ) (9) SiO(g) + 2C(.) = S i C w + C 0 ( g ) (10) SiO ( g ) + 3CO ( g ) = S i C w + 2C0 2 ( g ) (11) SiO ( g ) + C ( s ) = Si ( S ; l ) + C O ( g ) (12) 14 5. Reduction of S i 0 2 SiOaw) + 3Q.) = S i C w + 2 C O ( g ) (13) Si0 2 ( S ; 1 ) + 4 C O ( g ) = S i C ( s ) + 3 C 0 2 ( g ) (14) SiO^i) + 2C ( S ) = Siw) + 2 C O ( g ) (15) 6. Reduction with SiC SiC ( s) + 2Si02(s,i> = 3 SiO(g) + C O ( g ) (16) 2SiC ( s ) + Si0 2 ( sj) = 3 Si(s,,) + 2 C O ( g ) (17) SiQ.) + Si0 2 ( S ; 1 ) = Si ( s > 1 ) + S i O ( g ) + C O ( g ) (18) S i C ( l ) + SiO(g) = 2 S i W ) + C O ( g ) (19) Free energy values versus temperature for these reactions are plotted in Figure 5. In the free energy plots of the basic gas reactions (Figure 5a), all three reactions have negative free energy values within the reduction temperature range for silica. This indicates that they are all thermodynamically favourable. The plots show that any free oxygen is highly unlikely to exist due to its instability in a reducing environment where C O and C exist at relatively high activities. Also any C 0 2 that exists will react with Q S) to form CO(g). The dissociation reactions (Figure 5b) have positive free energy values within the silica reduction temperature range. This indicates that the reactions have low equilibrium constant values. Likewise for the formation of S i O ( g ) (Figure 5c) the free energy values are also positive. It is important to point out that even though the thermodynamics indicate that the equilibrium constant values are low for S i O ( g ) formation, researchers [3, 6, 21, 49] believed that Si02(S) must be reduced by either C ( s ) or C O ( g ) to form SiO ( g) during the reduction process. In contrast, the reduction of SiO(g) (Figure 5d) in the formation reaction of S i C ( s ) (Equation 10) shows a 15 relatively higher equilibrium constant value compared to the less favourable SiO^) formation reactions (Equations 7 or 8). o £ 03 - 1 0 0 -150 -200 -250 - 3 0 0 1 6 0 0 1700 1800 1900 Temperature (Kelvin) (a) o E 2 5 0 2 0 0 1 5 0 > -D ) l_ <D • C LU a; 1 0 0 1 6 0 0 1 7 0 0 1 8 0 0 1 9 0 0 Temperature (Kelvin) (c) o £ <D C LU 0) 5 0 0 4 0 0 3 0 0 2 0 0 1 0 0 0 - 1 0 0 Eqn 14 Eqn 15 Eqn 13 1 6 0 0 1 7 0 0 1 8 0 0 1 9 0 0 Temperature (Kelvin) o E CD C 4 0 0 3 0 0 2 0 0 h 1 0 0 03 05 1 6 0 0 1 7 0 0 1 8 0 0 1 9 0 0 Temperature (Kelvin) (b) 2 5 0 r o 2 0 0 -E \ 1 5 0 ; ' 1 0 0 ->-D > 5 0 -<D C LJ 0 -CD CD - 5 0 • 1_ Li. - 1 0 0 Eqn 11 Eqn 12 Eqn 10 1 6 0 0 1 7 0 0 1 8 0 0 1 9 0 0 Temperature (Kelvin) (d) 6 0 0 o E D ) 1_ CD c LiJ CD CD 5 0 0 -4 0 0 -3 0 0 2 0 0 1 0 0 0 Eqn 16 Eqn 17 1 6 0 0 1 7 0 0 1 8 0 0 1 9 0 0 Temperature (Kelvin) (c) ( f ) Figure 5: Gibbs Free Energy of Reactions 1-19. [27] 16 The reduction of silica (Figure 5e) is similar to the plot for the reduction of S iO ( g ) . The plot shows that reduction by C ( s ) has the largest equilibrium constant values. Equation 13, the overall silica reduction reaction, has a negative free energy value above. 1800 degrees Kelvin. This shows that it would be the most likely reduction reaction to obtain SiC ( s). Researchers [23, 24] have also believed in the involvement of SiC(S) in the reduction process but the free energy plots (Figure 5f) show the reactions have low equilibrium constant values within the temperature range of this study. From analysis of the free energy plots, the following reactions are the ones with relatively large equilibrium constant values: Si02 ( s,i) + C w = S i O ^ + C O ( g ) (7) SiO(g) + 2C ( S ) = S i C ( s ) + C O ( g ) (10) The plots also show the following: 1. Reduction of silica by carbon is more favourable that reduction by C O ( g ) . For example, Equation 7 compared with Equation 8 of the S i O ( g ) formation reactions, Equation 10 compared with Equation 11 of the SiO( g ) reduction reactions and Equation 13 compared with Equation 14 of the silica reduction reactions. 2. If Si(Sji) is involved in the reduction mechanism it may provide an alternative route for the transport of the silica component to the carbon source in forming SiC. Most researchers believed SiO( g ) to be the transport species in the reduction mechanism. The free energy plots show that the equations which involve Si(S;i) have lower equilibrium constant values than those that involve C( S ) but are noticeably better than the reactions which involve C O ( g ) . For example, Equation 9 compared with Equation 7 of the S i O ( g ) formation reactions, Equation 12 compared with Equation 10 of the S i O ( g ) reduction reactions and Equation 15 compared with Equation 13 of the silica reduction reactions. 17 It is difficult and unnecessary to investigate each of the 19 reactions. In a study by Paull [50], he declared that only three components existed for the S i -O-C system within the temperature range of 1000 to 3000 K . Using the phase rule, (P+J = C + 2,- Paull determined the variables to be C = 3 and 3 - 2 (fix T, P = 1 atm) and therefore the maximum number of coexisting phases was three. The system comprised of a gas phase and two condensed phases. The gas phase was a mixture of SiO(g) and CO(g). Four condensed phases were believed to exist within the defined temperature range: S i0 2 , C , Si and SiC. Six equations were needed to define all possible combinations of two condensed phases plus a gas phase. Si02(S,i) + C( S ) = SiO ( g) + CO ( g ) (7) Si0 2 ( s,i) + Si ( sj) = 2SiO ( g ) (9) SiO(g) + 2C ( S ) = SiC(S ) + C O ( g ) (10) S i O ( g ) + C ( s ) = Si(s,i) + C O ( g ) (12) SiC ( s) + 2Si02 ( s,,) = 3SiO ( g ) + C O ( g ) (16) S i C ( s ) + S iO ( g ) = 2Si(s,i) + C O ( g ) (19) These equations will be the main focus of the kinetic study. 2.2 Kinetics Reaction products are quite often determined by kinetics rather than thermodynamics. Kinetics is the study of the rate and mechanism by which one species is converted to another [42]. The rate is the mass of reactant consumed or a product produced per unit time. The mechanism is the sequence of individual events whose result produces the observed overall reaction. 18 2.2.1 Mechanism of Aluminosilicate Reduction The question of whether the kaolinite structure or the natural structure of the aluminosilicate affected the reduction rate of the silica phase was examined by Khalafalla and Haas [31]. They found that complex silicates behaved like weakly associated aggregates of simpler oxide groups (ie. SiC^). They believed the energy required to dissociate these complex silicates into the simpler oxide groups was low compared to the energy required for the reduction of the oxygen elements within the simple oxide groups. The carbothermic reducibility of silicate minerals was primarily dependent on their chemical composition and was practically independent of their structure [4]. In the carbothermic reduction of aluminosilicates, it was believed that the alumina phase played an important role in the reduction mechanism. The alumina acted as a mass transfer barrier to the movement of SiO and CO during the reduction of meta-kaolin or mullite. Studies by Vodop'yanov et al. [58] indicated that movement of gases through the porous alumina phase was rate determining. The experimental weight loss curves observed by Vodop'yanov et al. were linear which led to their assumption that the reduction of silica in mullite by CO was determined by the movement of SiO and CO through the porous alumina structure. Vodop'yanov did not consider the effect of the silicon carbide formation step. Investigations by Shimoo et al. [54] found that the presence of alumina actually enhanced SiC formation up to a critical point after which it inhibited the reaction. The dispersed alumina phase separated the silica particles providing easier access to the silica phase for reduction. Bentsen et al. [6] discovered during the reduction of mullite with carbon that partially reacted mullite particles showed an outer layer of porous alumina devoid of silica with a sharp reaction interface between the outer layer and the inner unreacted core. This provided strong 19 evidence that SiO production was occurring. In further experiments, when pCo was increased there was no associated increase in the reduction rate of silica to SiO ( g). On this basis, Bentsen et al. concluded that SiO formation was not rate determining since the effect of increasing pCo , which they believed was necessary for the reduction of SiO ( g), did not affect the overall reduction rate of the process. There has not yet been a comprehensive investigation into the carbothermic reduction mechanism of aluminosilicates. It has been acknowledged that the reduction process starts off with the formation of free silica and mullite upon heating the natural aluminosilicates to the reaction temperature. The free silica would theoretically reduce first in the presence of C ( s) or CO(g). Silica within the mullite phase would face less favourable reduction conditions due to its lower activity. 2.2.1.1 Mechanism of Mullite Formation Natural mullite (3Al203.2Si02) is very rare. The mullite used in industries is produced by heating materials containing alumina and silica to temperatures above 1000°C. Wahl et al. [60] stated that as long as the proper components were present in a mixture, the components would combine to form mullite when heated to the reaction temperature. The relative molar ratio of the alumina and silica components had no effect on mullite formation. Depending upon the stoichiometry, mullite would form with either an excess of alumina or silica remaining in the system. Within the context of the carbothermic reduction of clays, the presence of carbon is believed to have no effect on the formation of mullite and free silica. During the mullite formation process, kaolinite undergoes various stages of transformation (Figure 6). A thorough understanding of this process is lacking. The mullite formation reaction of kaolinite does not occur at a heterogeneous reaction interface [14] and is 20 therefore difficult to monitor. Upon heating (at 500°C) , the crystal structure is altered by the loss of O H groups in the form of water vapour, leading to the formation of meta-kaolin (Al203.2Si02). The dehydroxylation of kaolinite to meta-kaolin results in an intimate mixture of amorphous alumina and silica with increased porosity of the lattice. This pseudomorphic structure is a matrix of the original crystal structure with large concentrations of vacant anion sites. At 9 8 0 ° C an endothermic reaction occurs which produced a cubic phase and amorphous silica. The cubic phase may be a type of spinel with a general formula of A [ B 2 ] X 4 , where A refers to cations (Si 4 +) in tetrahedral sites, B to cations (Al 3 + ) in octahedral sites, and X to anions (O2-) that were cubic-close-packed. The spinel irreversibly transforms into crystalline mullite and silica with the release of heat. As the temperature exceeds 1100°C, the free silica converts to its high temperature polymorphic form, cristobalite. Brown et al. [10] found that heating kaolinite to 1000°C resulted in extensive separation of silica from the aluminosilicate. While silica was separating out, crystalline mullite was also forming. This free silica was initially amorphous which later converted to cristobalite. Studies [9, 14] on the formation of mullite from kaolinite indicated that kaolinite transformed into mullite by a different path than that of a prepared mixture of alumina and silica with the same 1:2 molar ratio as kaolinite. The S i -O-Al linkage of kaolinite was maintained during the mullitization process. A form of chemical or structural continuity existed up to the reaction temperature. In the case of the mixture of alumina and silica, the Si -O-Al linkage must be formed first before mullite formation occurred. The high temperature polymorphs of silica and alumina which are P-cristobalite and corundum were formed first upon heating the mixture, followed by the formation of an amorphous aluminosilicate structure from which the mullite phase would crystallize out. This required both time and energy. 21 METAKAOUMTE KAOUNTTE Removal at traces of -OH groups Intermeotate Alurreno-SMcate complex formed just before 980"C reaction. & Amorphous SiO, 980-C iioo-i6ocrc Muaitization taking place by two parallel paths. • —» Si, Al spinel • Mullite - —• Amorphous Si0 2 /A lurnino Silicate phases. - Interwoven mullite needles in the residual matrix of Alumino-Silicate and some cristobalite. Figure 6: Mullite Formation from Kaolinite [14] 2.2.1.2 Mechanism of Silica Reduction It is believed that the carbothermic reduction of kaolinite leading to an Al 2C>3-SiC composite powder is analogous to the carbothermic reduction of silica into silicon carbide. Only the silica component in the aluminosilicate is reduced. Silicon carbide is a material with a lengthy industrial history. The very first process to produce large quantities of SiC was developed by Acheson in 1891 [28]. In the process, a 22 strong electric current from carbon electrodes was passed through a mixture of clay and coke. The products of the Acheson process were rough SiC crystals which required extensive milling and purification to produce useable powders. This basic process is still used today [64]. Studies of the Acheson process provide the basic understanding of the carbothermic reduction of silica into silicon carbide. This process is represented by the following reactions: S\02+y » S i O ( g ) + ^ 0 ( g ) SiO( g ) + V E n e ^ > S i C + ^ 0 ( g ) The elements of the carbothermic reduction process consist of the reactant, which in this case is silica (Si02), a reducing agent y which may be C ( S) or CO ( g ) and energy (in the form of heat). This study will only use a solid carbon reductant. The silica source in the reaction may come from the decomposition of clays (ie. kaolinite) into silica and mullite. In the discussion of the thermochemistry, it was noted that the solid species involved in the reduction mechanism were Si0 2 , SiC and C while the gaseous species were Si, SiO, C O , C 0 2 and 0 2 . From previous studies it was pointed out that SiO(g> was the key element in the reduction of silica. The oxygen and carbon dioxide species were not likely to exist in significant concentrations because of the high C O partial pressures and high carbon activity. The movement of product gases away from the reaction interface and into the bulk atmosphere may be important in the reaction mechanism. The densities of the individual gaseous species may be a factor in this mass transfer process: Si (28 g/mol), SiO (44 g/mol) and C O (28 g/mol). The nature of the purge gas and its flow rate would play an important role in this process as well. The goal in controlling the reaction atmosphere is to minimize the loss of SiO (which improved the SiC yield) while removing the product gases at a rate which would drive the reduction reaction to completion. 23 The question of whether a liquid phase exists in the system as part of the mechanism is not known. Silicon dioxide particles readily sinter at 1400 to 1600°C so it may be possible that a thin,layer of viscous liquid SiO^i) may exists on the outer surface of the particles. As well, the existence of Si<i) is equally possible above 1430°C. The mechanism for the reduction of silica has been explored by previous researchers but their findings have not been conclusive. One version that is well supported [54, 65] is based on a two-step mechanism (Figure 7): CO? SiO ' C O I j •CO C + Si0 2 — • SiO + C O C O + Si0 2 — • SiO + C0 2 C + C O z — • 2 C O 2 C + SiO — • S iC + C O SiC Figure 7: Two Step Reduction Mechanism. [65] The first step involves the formation of the SiO(g> transport species: Si0 2 ( s,i) + C ( s ) = SiOfc) + C O ( g ) (Step 1 A ) It is believed that the solid-solid reaction with the formation of SiO( g ) is the initiator in the production of CO(g) but this reaction can not be sustained due to loss of solid-solid contact points as the reaction progresses. The measured rate of reduction for silica conversion to silicon carbide indicates that a solid-solid reduction mechanism is highly unlikely. Therefore a second reduction reaction involving gaseous intermediates is needed: SiOzw) + CO ( g ) = SiO(g) + C 0 2 ( g ) (Step IB) 24 The C0 2 ( g ) formed would not be stable at the reaction temperatures. Due to the presence of C ( 8 ) it would revert to CO(g) according to the Boudouard reaction. The formation of the suboxide SiO(g) is fundamental in this reduction mechanism. Silicon monoxide is stable in its gaseous form within the temperature range of 1400 to 1600°C [29]. The S i O ( g ) phase is the transport species that moves silicon from the silica to the carbon where it reacts to form SiC in the second step of this two step mechanism: SiO(g) + 2C ( S ) = S i C ( s ) + CO ( g ) (Step 2) The SiC(S) produced will conform to the morphology of the carbon particles. This is an important consideration since the production of SiC fibres can be accomplished by using carbon fibres as the precursor material. A n alternate mechanism proposed by other researchers [37] was a four step process, starting with the silica dissociation reaction: S i0 2 ( S j l ) = S iO { g ) + |o 2 ( g ) (Stepl) The above reaction is feasible above 1300°C [15] in the presence of a reductant such as H 2 ( g \ or C(S). The oxygen that is produced would react with any carbon that is present in the system to form the reducing gas, C O : ^ 0 2 ( g ) + C ( s ) = C O ( g ) (Step 2) The silicon monoxide is further reduced to gaseous silicon: S i O ( g ) + C ( s ) - S i ( g ) + C O ( g ) (Step 3) And finally the gaseous silicon combines with carbon to form silicon carbide: S i ( g ) + C ( s ) = S i C ( s ) (Step 4) 25 Reactions 2 to 4 are all thermodynamically favourable. The reduction or dissociation of silica to form silicon monoxide did not depend on the nature of the reducing agent being used but its presence is necessary for the reduction process to occur. As well, the formation of SiC is believed to occur through the combination of Si(g> with C ( s ) . The reduction mechanisms described thus far involved the reactant Si02(S> and the transport species SiO ( g) and possibly Si(g>. A closer look at the role of the solid reductant would be beneficial. 2.2.1.3 Carbon Source It has been indicated by the mechanism that the silicon species is transported to the carbon source where it reacts to form SiC. Therefore the carbon source's morphology would determine the SiC product's morphology. Various researchers [5, 12] developed their reduction mechanism model based on their experimental evidence which showed the SiC product having the same morphology as the precursor carbon phase. The type of carbon reductant used in the reaction mixture had a profound effect on the reduction rate (Figure 8). In this plot of % Conversion versus time, the activated carbon had the highest reduction rate. Shimoo [55] found that when the reducing species was activated carbon, a large volume of SiO( g ) was generated. This high production of SiO ( g) was due to the high porosity of the activated carbon which provided large number of reduction sites. Small particle size would also provide the same result due to the large number of contact points between the silica particles and the carbon particles. 26 Ti m e , f / k s Figure 8: Reactivity of Activated Carbon, Graphite and Carbon Black in the Reduction of Silica. [55] 2.3.2.4 Role of CO(s) The mediating phase during the reduction reaction is most likely SiO(g), C02(g>, Si(g,i) and CO(g). The importance of CO(g> as a reductant has been emphasized by previous researchers [4, 7, 23, 39]. It would seem that if pco was increased then the reaction would proceed at a higher rate according to Step IB of the two step reduction mechanism: Si02(s,i) + C O ^ = SiO(g) + C 0 2 ( g ) (Step IB) It has been proven by Bentsen et. al. that silica can be directly reduced by C O gas. Figure 9 is the plot of conversion, X, versus time (hours) for two powder samples. One was a silica/carbon mixture while the second sample was strictly silica. For reduction to occur a C O partial pressure of 1.5 atm was used. 27 0 0 50 100 150 hours Figure 9 : Silica Reduction using CO ( g ) (pCo = 1.5 atm). [6] According to Bechtold [4], the relationship that existed between SiC yield andp c o was: where A is a constant dependent on whether iron was present as a catalyst, / is time and A E is the activation energy. The equation indicated that lower pCo values would increased the yield of SiC. This suggested that increased pCo would inhibit the second step of the two step reduction mechanism according to Le Chatelier's Principle: Lee and Cutler's study [37] supported the fact that C O suppressed SiC formation. Khalafalla and Haas [32] also determined that by increasing the C O partial pressure it reduced the extent of reaction for a clay/carbon mixture up to the point where it inhibited the reduction reaction altogether. Higgins and Hendry [25] found that once the reaction atmosphere contained 20% C O then the stable phase was mullite. [SiC], \%YieU ~ S i O ( g ) + 2C ( S ) = S i C ( s ) + CO, '(g) (Step 2) 28 The results of Bentsen's research seem contradictory to those of Bechtold, Khalafalla and Haas, etc. In Bentsen's case, the C O partial pressure must exceed 1 atm before reduction was possible and when reduction occurred, it was at a much lower rate than experiments carried out by Bechtold, etc. A different reaction mechanism was probably in effect for the reduction of silica to silicon carbide at very high pco values. 2.2.1.5 Role of Si Kennedy and North [29] found that the addition of Si to their silica/carbon reaction mixtures increased the reduction rates. Their study may indicate the importance of the four step reaction mechanism in which Si is one of the key components of the reduction mechanism. S i O ( g ) + C( s ) = S i ( g ) + C O ( g ) (Step 3) Si(g) + C ( s ) = SiC(s) (Step 4) However, the silicon addition may have increased the yield of SiC due to its reaction with solid carbon that was part of the reaction mixture. 2.2.1.6 Role of SiC It is believed that silicon carbide may increase the reduction rate of silica by the following reaction: S i C ( s ) + 2Si0 2 ( s,i) = 3SiO ( g ) + C O ( g ) (16) with SiO(g) subsequently reacting with C(S> to form SiC. Biernacki and Wotzak [7] discovered that the SiO to C O ratio to be 2.6:1 in their experiments whereas 3:1 is the stoichiometric ratio as shown in Equation 16. They believed that the following reaction must also be occurring: 2SiC ( s ) + Si02 ( s,,) = 3 Si(s,,) + 2 C O ( g ) (17) 29 which would explain the decrease in the SiO to C O ratio. Silicon would only exists in this environment if C O partial pressure and carbon activity were low, otherwise silicon would immediately react with those phases to form SiC. 2.2.2 Determination of the Rate Limiting Step A far more contentious issue is the determination of the rate limiting step in the reduction mechanism. Studies by other researchers have suggested these following possibilities for the rate controlling step: 1. Heat transfer. 2. Solid-state diffusion. 3. Chemical reaction rate. 4. Mass transfer rate. • Diffusion of gaseous reactants and gaseous products from the bulk of the gas phase to the internal surface of the reacting solid particle (external mass transfer). • Diffusion of gaseous reactants or gaseous products through the pores of a solid reaction product or partially reacted solid (pore diffusion) 5. Nucleation and growth rate. Assuming that the two step mechanism to be the correct model for the reduction process, then there would be only two steps to investigate: the SiO formation step and the SiC formation step. It was shown by Weimer et al. [65] that heat transfer was not a factor for small particle sizes. The experimental apparatus used in their research utilized a very high heating rate where complete reaction occurred in less than 10 seconds. Weimer et al.'s ultrahigh heating rate experiments led them to believe that heat transfer was not the rate limiting step. 30 Weimer et al. also stated that solid-state diffusion mechanisms were too slow for the experimental reduction rates observed [65]. Khalafalla and Haas [31] discounted surface diffusion as the rate limiting step as well. Their experiments showed that silicates which melted at reaction temperatures had lower reaction rates than silicates that remained solids. The fact that diffusion is orders of magnitude greater through a liquid phase indicated that this was not the rate limiting step. The liquid silicates' lower rate of reduction may be due to: • significant reduction of the surface area for reaction, or • surface tension which reduced the contact area between the liquid surface and the carbon particles, or • high viscosity of the molten silicates due to retention of its original structure as it converted to liquid state; the high viscosity reduced wetting with the reductant particles and thus reduced the contact area. The effect of alkali oxides as catalysts may be due to their lowering of the molten silicates' viscosity by breaking down the complex silicate structures into more basic forms. If chemical reaction was rate determining, then the following conditions should be examined: • Batch size effect. Small batch sizes would have higher reaction rates compared to large batch sizes (ie. bringing to boil a small volume of water compared to a large volume) • SiO formation is rate determining. • SiC formation is rate determining. The thermodynamic data indicated that the SiO formation step (Equations 7 and 8) had positive free energy values within the experimental temperature range of 1400 to 1600°C. This made it less favourable compared to the SiC formation step (Equation 10) which had negative free 31 energy values in the same temperature range. Therefore if the rate limiting step was governed by chemical reaction rate then the SiO formation step would likely be rate limiting. Earlier studies by Dosaj [21], Filsinger and Bourrie [23], Kennedy and North [29] determined the reaction rate to be first order. Kennedy and North, Kevorkijan et al. [30] believed that the rate limiting step was SiO formation. Lee and Cutler [37] found a definite relationship between silica particle size and reaction rate which led them to conclude that SiO formation was rate limiting as well. Lee and Cutler speculated that the increased reaction rates as silica particle size decreased was due to the increase in Si02 surface area. Chrysanthou et al. [15] also believed that the reaction rate depended on the effective surface area of the silica particles. Their kinetic plots were linearized by applying the first order kinetics equation: -4c = kdt c where c is the concentration (ie. moles/cm3) of the reactant species (Si02), k is the experimental reaction rate and t is time. In studies by Weimer et al. [65] they found no silica particle size effect. This may be due to their experimental setup in which particles were encapsulated in gas bubbles and were not likely to have come into physical contact with one another. The experiments by Weimer et al. did find a carbon particle size effect which led to their conclusion that SiC formation was the rate limiting step. Once this conclusion was made, Weimer et al. applied their mass transfer controlled model to the reduction process. A look at mass transfer as the rate limiting step should begin with an examination of gas diffusion to the bulk atmosphere. The specific points to investigate would be: 32 • Effect of flow rate of the purge gas which removes the product gases from the reaction region creating a driving force for the reaction to progress. • Linear relationship between the extent of reaction and time. If neither of these parameters were met then the other possibility was that the mass transfer rate limiting step may be due to diffusion of reactant and product gases through the SiC layer. Ono and Kurachi [49] believed that the rate limiting step to be diffusion through the P-SiC product layer. Two pertinent equations used to describe this process were: 1. Contracting Volume Model (Phase boundary control) 1 - (1 - X)m = kt 2. Jander's equation (Diffusion control) [ 1 - ( 1 - X ) 1 / 3 ] 2 =kt Both models are based on particles having a spherical geometry. Shimco et al. have shown with coarse graphite particles, that the reduction rate followed the Janders' equation model. Weimer et al. [65] indicated that the second step of the two step reduction mechanism: SiO(g) + 2C(S) = SiC(s) + CO(g) was governed by mass transfer based on a contracting volume model of the carbon particle as it converted into SiC. In both studies, Weimer et al. and Shimoo et al. found no silica particle size effect which led them to believe that SiO formation was not the rate limiting step. Another option that was explored by researchers was nucleation and growth rate control for the rate limiting step. When a reaction is accompanied by the nucleation and growth of the reaction product in the solid state, then the following Avrami-Erofe'ev (AE) equation [49] was frequently obeyed: [-\n(\-X)fn =kt 33 The value of n is equal to: • P + X for phase boundary control X • B + — for diffusion control 2 where B is the number of steps involved in nucleus formation and 1 is the number of dimensions in which the nuclei grow. Generally P was 0 or 1 and X was 3 for spherical particles. When P was 0, this indicated that the SiC crystals grew from the originally existing nuclei in the system. By plotting ln[-ln(l-^f)] versus logarithmic time, the value of n could be derived. Ono and Kurachi [49] determined this value to be 1.5. Khalafalla and Haas [32] also found in their experiments that silica reduction followed a nucleation-growth mechanism. In summary the four most likely models for the rate limiting step are as follows: 1. First Order Reaction Kinetics 2. Contracting Volume Model 3. Jander's Equation 4. Avrami-Erofe'ev Equation O f these four, the strongest support [49, 65] has been for the Contracting Volume Model (phase boundary control). According to these researchers, the growing layer of SiC on the outside of the carbon particles acted as a barrier to the transport of the gaseous species to and from the reduction interface. 34 2.3 Objectives So far as is known, the kinetics of the carbothermic reduction of natural aluminosilicates have not been investigated in any detail. The overall objective of this dissertation is to study the reduction kinetics with particular attention to the following points: 1. The relationship between the kinetics of the carbothermic reduction of aluminosilicates and the reduction of silica. 2. The effect of particle size of the silica source (ie. silica or aluminosilicate) and particle size of the carbon source on the conversion rates. 3. The effect of Si and SiC seeding on the conversion rate. 4. The role that C O ( g ) plays in the mechanism. Through the exploration of these points, a better understanding of the reaction mechanism and the rate limiting step can be derived. With the aim of lowering the production costs of the carbothermic reduction process, alternative materials for the keys components are investigated: silica sources (ie. various aluminosilicates), carbon sources (ie. coke, graphite) and reaction atmosphere (ie. N 2 as an alternative to Ar). 35 3.0 Experimental 3.1 Apparatus It was apparent from the reduction reaction that there would be weight loss due to the formation of C O ( g ) . Additional weight loss may also occur due to elimination of SiO ( g ) . The weight loss could be used to indirectly monitor the kinetics of S i 0 2 to SiC conversion. Therefore, thermogravimetric analysis would be the appropriate tool for this study. A n alternative route which some researchers have used was to measure the CO(g) evolved from the reaction as an indication of the rate of the overall reaction [1, 25, 36]. The involvement of CO( g ) in reverse reactions made it difficult to pin-point the rates of the individual steps involved in the reduction mechanism. As well, to follow C O ( g ) evolution required knowledge of the specific reactions that were occurring. The measurement of CO(g) was difficult and required precise equipment (with gas chromatography being the preferred method). It was therefore preferable to use weight loss as a measure of the reaction rate. The apparatus (Figure 10) was based on an alumina tube furnace (length =114 cm, I.D. = 3.9 cm and O.D. =4.5 cm) heated by Super Kanthal (MoSi2) elements. The maximum temperature of the furnace was 1600°C. A Pt/Pt-10%Rh thermocouple was connected to a temperature controller which maintained the reactor temperature. Location of the hot zone in the tube furnace was determined by lowering a Pt/Pt-10%Rh thermocouple to the hot zone region and measuring the distance from the top of the tube furnace. This also provided a means of calibrating the temperature controller reading to the actual temperature within the hot zone. The equipment used in these experiments did not incorporate a stable cold junction, therefore variations in room temperature may affect the 36 reactor temperature and could be a source of error. Experiments, which required comparison of their kinetic data, were performed consecutively thereby reducing the effect of errors caused by fluctuating room temperature. Super Kanthal Elements Pt/Pt10%RhT/C Alumina crucible Gas inlet Figure 10: Thermogravimetric Analysis Apparatus. The tube furnace was not a closed system. Reactor gases were introduced through the sealed bottom chamber of the tube furnace. These gases travelled upward through the hot zone and passed around the crucible containing the powder sample. The gases would then escape to the atmosphere through a one centimeter diameter opening in the furnace tube cap. This opening allowed the hanging rod to pass through, linking the crucible to the balance. Because 37 this was an open system, it was assumed that the pressure in the reaction zone was 1 atm. The gas system (Figure 11) allowed for two gases to be mixed and introduced into the tube furnace. The purification columns contained silica gel to remove moisture from the gases and maintain a consistent humidity level. Matheson Model 604 flowmeters were used to measure gas flow rates. Gas to furnace Gas 1 Gas 2 Purification Purification Mixing column 1 column 2 column Flowmeter 1 Flowmeter 2 Figure 11: Gas System Samples were loaded into alumina crucibles which were then inserted into the furnace from the bottom. A n alumina hanging rod with an alumina hook held the crucible by its cross bar as it was raised into the hot zone. The length of the alumina hanging rod extended through the length of the furnace tube so that the loading of the sample from the bottom of the furnace was possible. The hanging rod was attached to a steel cable which was connected to the balance. The weighing balance was a Mettler AE240 electronic analytical balance with a RS232C interface port which transferred its digital weight readings directly to an I B M PC. The Mettler AE240 has a measuring range of 0 to 205 grams, a standard error of 0.1 milligram in its readings and a stabilization time of 5 seconds. Data from the thermocouple and the balance were collected on the I B M P C using the data acquisition software Labtech Notebook. 38 3.L1 Crucible The crucible used in the experiments (Figure 12) reflected the need to reduce the effect of certain experimental parameters which influenced data acquisition. Oscillations of the weighing apparatus caused by turbulent gas flow was reduced by choosing an aerodynamically-shaped crucible (ie. small diameter and cylindrical). To reduce mass transfer effects and increase diffusivity of the gases, the depth of the powder sample was relatively low with a large exposed top surface area. The surface of the sample lay just below the lip of the crucible. During the testing of various crucible designs, it was found that with deep crucibles the reduction process was governed by mass transfer of the product gases from the reactant interface to the bulk atmosphere. To improve purge gas flow, part of the sidewalls of the crucible was cut away leaving two sections for the spanning crossbar. By removing the sidewalls from the crucible, it meant a possible increase in SiO(g) loss but it also allowed for easier purging of the product gases from the reaction powder. 15 mm 15 mm 20 mm ' Figure 12: Crucible Design 3.1.2 Reaction Atmosphere The reactor gas atmosphere played two important roles. It acted as a purge gas to remove the gaseous products of reaction (ie. C O ( g ) ) and it could act as a reductant if it contained 39 gases such as H 2 or CFL,. According to Le Chatelier's principle, the removal of gaseous products improved kinetics by driving the reaction towards the product side: Si0 2 ( S) + C(S) = SiO(g) + C O ( g ) SiO(g) + C(s) = SiQs) + CO(g) if either of the equations were the rate limiting reaction. Note that the loss of SiO was also possible. The difference in gas densities between SiO and C O (ie. SiO = 44 g/mol versus C O = 28 g/mol) may play a role in limiting SiO loss. The flow rate was an important factor to consider. Too low a flow rate would cause a stagnant atmosphere. Too high a flow rate would increase the loss of S iO ( g ) and reduce the yield of SiC in the product powders. Table 5 lists the various gases which were employed as a purge gas. The costs of the various gases themselves are important if the overall process is to be made as economical as possible. Argon was the preferred purge gas used in other studies. Gas $/cylinder C 0 2 40 N 2 16 Ar 70 Table 5: Relative Costs of Various Purge Gases (1992) Previous researchers [1, 59] found that the use of nitrogen in the reactor atmosphere led to the formation of: • Silicon nitride (SisN4) or silicon oxynitride (S i 2 N 2 0) during silica reduction, and • Sialons (Si6-xAlxNg.xOx) during aluminosilicates reduction within the temperature range 1400 to 1600°C. The intent in this study is to use N 2 as a purge gas only and to limit its involvement in the carbothermic reactions. 40 The reactions for silica reduction in N 2 atmosphere are as follows: 3 S i 0 2 ( s ) + 6C ( S ) + 2 N 2 ( g ) = Si 3N4( S) + 6 C O ( g ) or 2SiO(g) + Nacg) + C O ® = S i 2 N 2 0 ( s ) + C 0 2 ( g ) S i 2 N 2 0 ( s ) + 3C ( S ) = 2SiC ( s ) + N 2 ( g ) + C O ( g ) The above reaction sequences are thermodynamically feasible above 1350°C but sensitive to the total pressure and the composition of the reaction atmosphere. In studies, Cho et al. [16] found SiC to be stable at temperatures greater than 1500°C; Siddiqi et al. [56] found SiC to be stable in a S i - C - O - N system above 1550°C. Higgins and Hendry [25] investigated the carbothermic reduction of kaolinite in the production of B'-sialon (Si6-ZA1Z0ZN8.Z, where z = 3). The production of sialons from aluminosilicates is analogous to the production of silicon nitride from silica. The overall reaction can be described as follows: 3 (2S i0 2 .A l 2 0 3 . 2H 2 0) + 15C + 5 N 2 = 2 S i 3 A l 3 0 3 N s + 15CO + 2 H 2 0 The mechanism described by Higgins and Hendry is: 1. Mullite and silica formation from kaolinite. 3 (2S i0 2 .A l 2 0 3 . 2H 2 0) = 3 A l 2 0 3 . 2 S i 0 2 + 4 S i 0 2 + 6 H 2 0 2. Silicon carbide formation. 4 S i 0 2 + 12C = 4SiC + 8CO 3. Simultaneous reduction and nitridation of mullite. 3 A l 2 0 3 . 2 S i 0 2 + 4SiC + 3C + 5 N 2 = 2 S i 3 A l 3 0 3 N 5 + 7CO The SiC formed reacted with the nitrogen atmosphere to produce sialon which is the most thermodynamically favourable phase. The goal of this study is to halt the process at step 2. 41 3.2 Raw Materials Table 6 contains a list of the materials used in the various reduction experiments including the supplier or location of its origins and the average particle size of the powder, where available. The particle sizes for some materials are not listed due to the fact that they were delivered in 50 to 100 grams chunks which required extensive milling into micro-sized particles before being used in the experiments. Powder Supplier Particle Size (diameter in um) Bentonite Estrin n/a Charcoal, Lamp Black Anachemia 2 Coke, Metallurgical Dofasco Ltd. n/a Feldspar Estrin n/a Graphite, Superflake Superior Graphite Co. 10-30 ("Micro-Mesh") Illite Fithian, Illinois n/a Kaolinite Georgia 2 Kyanite Virginia n/a Mica India n/a Mullite Synthetic freeze-dried <1 Silica flour Fisher Scientific 2-50 a-Quartz Brazilian, optical grade n/a Silicon, 99.99% Atlantic Equipment Engineers <45 Silicon Carbide Hermann Starck, Berlin 2 Syenite Nepheline Syenite, Ontario 50 Table 6: Supplier and Particle Size of Raw Powders. The standard reduction experiments were carried out using stoichiometric mixtures of silica flour with lamp black or meta-kaolin with lamp black. Meta-kaolin was produced by calcination of Georgian kaolinite at 7 0 0 ° C for 8 hours. The use of meta-kaolin eliminated the water loss factor from the weight loss curve. 42 Various parameters were studied which involved the use of specific materials. Silicon and silicon carbide seeding were carried out to determine their effect on the reduction rate. Particle size experiments involved the use of kyanite, ot-quartz and coke. These materials were milled and separated into various particle size fractions. Experiments studying the reduction kinetics of various aluminosilicates involved the use of bentonite, feldspar, illite, mica and syenite in stoichiometric mixtures with lamp black. The results were compared to that of meta-kaolin/carbon mixtures. The study of alternative carbon sources involved the use of graphite and metallurgical coke as substitutes for lamp black. 3.3 Experimental Procedure 3.3.1 Powder Preparation The powders used in the particle size effect experiments were dry milled in porcelain containers with alumina balls for periods of up to 8 hours in a Schutz-O'Neill vibratory mill. The resulting powders were then sieved to various particle sizes using Endecotts U.S. Standard Sieves Series containing Mesh sizes 325, 200, 170, 140, 100, 70 and 50. Reactant powders were prepared by carefully weighing the correct quantities of precursor materials needed for each particular experiment. This mixture was then dry milled in 60 ml Naglene containers using two 1 cm diameter alumina balls for 1/2 hour on a Spec Mixer/Mill. 3.3.2 Procedure for a Reduction Experiment Run The following preparations were made prior to running an experiment: 1. The furnace was raised to the reaction temperature and allowed to stabilized for 1/2 hour prior to the running of the experiment. 43 2. The Mettler balance was turned on and allowed to stabilize for an hour. Prior to experimentation, the balance was recalibrated. 3. Purge gas (ie. Ar) was allowed to pass through the flowmeters and through a by-pass route where the gases was vented to the outside atmosphere. The gases were checked for correct flow rates as measured by the Matheson flowmeters. 4. Samples of approximately 0.4 to 0.5 grams were prepared. The weight of the empty crucible and then the weight of the powder plus crucible were determined. The exact weight of the sample could then be calculated and recorded. All weights were measured to 3 significant figures. Loading and heating procedure: 1. The crucible containing the powder sample was attached to the alumina hanging rod through the base of the furnace tube. 2. Once the brass cap at the base of the furnace tube was screwed back in place, the reactor gases were diverted from the by-pass route and into the furnace. The sample was raised immediately by 10 cm. 3. The sample was raised in 10 cm increments with a 5 minute pause between each stage. Raising the sample to the hot zone 40 cm up in the furnace tube took 20 minutes. The last stage of raising the crucible into the hot zone involved a jump in temperature of approximately 200 to 3 0 0 ° C , depending on the hot zone temperature. 4. At the end of the last stage prior to attaching the sample to the electronic balance, the purge gas was switched to a reducing gas (ie. C O ) if the experiment required it. If C O was used in the experiment then the exhaust gases were burned at the outlet. 5. The final step after the attachment of the hanging rod to the balance was to initiate the data acquisition software. This was the zero point for the collection of the kinetic data. 44 Unloading and cooling procedure: 1. The data acquisition software was terminated at the end of the experiment and the data saved on disk. 2. The sample and hanging rod were detached from the balance. At this point, if the reactor gas used was not A r then it was to be switched over to A r for the duration of the cooling cycle. Otherwise the flow rate of A r was maintained. 3. Detachment from the balance involved lowering the hanging rod by an initial 10 cm step to the apparatus which was used to lower the assembly. After each lowering step there was a 5 minute pause to allow the system to equilibrate. 4. Once the sample had travelled to the base of the furnace tube and the A r diverted to the by-pass route, the base of the furnace was opened and the sample removed. 5. The weight of the crucible and product powder was measured. The weight of the empty crucible determined earlier was then subtracted from this value. The overall weight loss was calculated and this was used to gauge the extent of reaction. 4.3.2 Powder Characterization Both raw and product powders were subjected to the following analyses: 1. X-ray diffraction (XRD) analysis to determine crystalline phases (Philips). 2. Scanning Electron Microscope (SEM) for particle morphology and size analysis (Hitachi S-2300). 3. Energy-Dispersive Spectroscopy (EDS) for identification of phases using elemental analysis from the S E M if clarification of phases was required. 4. Particle Size Analysis for the determination of the particle size distribution (Horiba). The Horiba Particle Size Analyzer used sedimentation rate and the centrifugal force exerted on the powder to determine particle size distribution. Particle size analysis was performed only on the reactant powders because of the composite nature of the product 45 powders. The S E M was used to analyze product powder morphology, particle size and composition. X-ray diffraction (XRD) analysis was used to identify the major crystalline phases in the product powder. 3.4 Data Processing Data coming from the analytical balance were collected by LabTech Notebook software. The weight was measured every 10 seconds. The weight measurement received from the balance included the weight of the hanging rod assembly and the crucible. The weight of the sample was calculated by subtracting the crucible and hanging rod assembly weights. This weight data were then translated into conversion values. Conversion, X , was calculated as follows: m - m x _ ° m -W o where m0 is the original sample mass, m is the mass at time t and W is the weight at 100% conversion for a clay or mineral. Table 7 shows the theoretical weight losses at 100% conversion for the various precursor powders. The kinetic data were plotted as conversion, X, versus time (in seconds) and as conversion rate (ie. reduction rate), ^f-, versus time. dt The weight loss data obtained from the balance featured oscillations caused by turbulence due to gas flow in the furnace tube. As well, the algorithm used by the balance for determining weight values is based on recording the sample weights at the end of specific time intervals rather than averaging the weights within that interval. This leads to the inherent fluctuations of the weight readings received from the balance. 46 Precursor Powder Weight Loss for 100% Conversion Kaolinite 44.84% Meta-Kaolin 38.1% Silica 58.3% Mullite 22.5% Table 7: Weight Loss at 1 0 0 % Conversion for Various Precursor Powders Figure 13a shows the raw data that were sent from the electronic balance for the reduction of silica by lamp black in an argon atmosphere at 1470°C. The data were then converted from weight versus time to conversion, X, versus time (Figure 13b). The data were then interpolated (Figure 13c) using an interpolation algorithm derived from SigmaPlot software (Version 1.02 for Windows). The final plot (Figure 13d) shows just the interpolated result which remained a valid representation of the reduction reaction. The rate of reduction is the slope of the conversion versus time plot. The initial part of the rate curves is used to determine the experimental rate value of a particular experiment. Environment conditions within the powder sample would change as the reaction progressed and the amount of reactants was depleted. Factors such as mass transfer and involvement of product phases in the reduction mechanism would become significant at the later stages. 47 72.2 1.0 71.8 1000 2000 3000 T i m e ( s e c o n d s ) 0 1000 2000 3000 T i m e ( s e c o n d s ) 1000 2000 3000 T i m e ( s e c o n d s ) (c) 1000 2000 3000 T i m e ( s e c o n d s ) (d) Figure 13: Data Representation: (a) Raw Data, (b) Converted Data, (c) Converted and Interpolated Data, (d) Interpolated Data 3.5 Preliminary Studies 3.5.1 Reproducibility Study All experimental apparatuses have an associated error value dependent on a multitude of variables. Parameters such as ambient temperature, humidity, power fluctuations, etc. could all combine to create slight variations between one experimental run and the next. The 48 experimental apparatus lacked a stable cold junction which may lead to errors. Reproducibility experiments (Figure 14) using meta-kaolin/lamp black mixtures with sample weights of approximately 0.5 grams were performed in an argon environment at 1530°C. N o significant changes in slopes and weight losses between the four runs could be detected up to approximately 90% conversion. 0 500 1000 1500 2000 Time (seconds) Figure 14: Reproducibility Runs for Meta-Kaolin Reduction at 1530°C. 3.5.2 Removal of the Water Loss Factor The choice of using meta-kaolin over kaolinite was made to remove one of the variables which could affect the kinetic curves. The various weight losses for a kaolinite sample reduced by carbon in an argon atmosphere at 1530°C are shown in Figure 15. The different stages of weight loss during heating of a stoichiometric mixture of kaolinite and carbon were calculated as follows: ( a ) l o s s o f H 2 0 3(Al 203.2Si02.2H 20) = 3 A l 2 0 3 . 2 S i 0 2 + 4 S i 0 2 + 6 H 2 0 ( g ) 49 with a weight loss of 10.91% (w.r.t. m0). Water loss only affected the start point of the reduction curve but not the slope (rate of reaction), (b) reduction of free S i 0 2 3 A l 2 0 3 . 2 S i 0 2 + 4 S i 0 2 + 12C = 3 A l 2 0 3 . 2 S i 0 2 + 4SiC + 8CO with a further weight loss of 22.63% (w.r.t. m0) and (c) reduction of mullite 3 A l 2 0 3 . 2 S i 0 2 + 6C + 4SiC = 3A1 2 0 3 + 6SiC + 4 C O with a weight loss of 11.3% (w.r.t. ni0). From Figure 15, it can be seen that most of the water loss occurred during the heating procedure and cannot be recorded by the data acquisition equipment. Water loss occurred below the reduction temperature and before reduction began so its presence or absence did not affect the reduction curve. By using meta-kaolin, the water loss segment was removed from the conversion calculations. Figure 15: Weight Loss for Kaolinite Reduction at 1500°C. 50 4.0 Results By following the procedures described in the previous section, experiments were carried out to investigate the mechanism and rate limiting step for the carbothermic reduction of silica and meta-kaolin. The determination of which experimental parameters might influence kinetics involved the investigation of batch size, stoichiometry and gas flow rate effects. Studies were done on the effect of particle size for the reactant powders (silica and kyanite) and for the solid reductant (coke). Analysis of the Si and SiC seeding effects and the effects of CO(g) were made to gain further insights into the mechanism. Various aluminosilicate and carbon sources and gaseous environments were investigated as alternatives sources of raw materials. 4.1 Experimental Configuration 4.1.1 Batch Size Effect Sample size was determined by the size of the crucible used. Enough product powders must be produced so that sufficient material was available for product characterization. Equipment limitations required a quantity that was large enough to offset the oscillation effects of the weighing apparatus but small enough to minimize mass transfer effects. From Figure 16, there was little effect of batch size once a minimum amount has been exceeded. The 0.13 gram sample has a higher conversion rate value at 500 seconds than either the 0.40 or 0.66 gram sample (Figure 16b) because of its higher rate of SiO(g) loss. In the latter part of the conversion rate plot, the smaller sample's rate value dropped below those of the larger batch sizes due to the lower concentration of reactants remaining in the sample. A choice was made to use batch sizes of 0.4 to 0.5 grams for all experiments. 51 0.2 o 0.500 X o.o 0.000 500 1000 1500 2000 2500 400 t=500s 600 800 Time (seconds) Time (seconds) (a) (b) Figure 16: Batch Size Effect for Kaolinite Reduction at 1 4 5 0 ° C . (a) Conversion versus Time and (b) Conversion Rate versus Time 4.1.2 Stoichiometry Effect According to Paull [50], the molar ratio of silica to carbon was important for silica reduction. If the ratio was less than 1:3 (ie. stoichiometric ratio), then SiC and C remained as products of the reduction reaction. If the molar ratio was greater than 1:3, then it was believed that silicon carbide would reduce silica resulting in the formation of silicon metal. Test runs with silica-to-carbon ratios of 1:1, 1:2, 1:3 (stoichiometric) and 1:4 were carried out to determine the stoichiometric effect on silica reduction (Figure 17a). As reduction progressed, the conversion rates dropped off for the mixtures with the higher silica content due to the lack of reductants in the powder to maintain the reduction process. The conversion rate plots (Figure 17b) showed the initial rate values for the different stoichiometric mixtures were relatively close 52 to one another. Stoichiometry did not appear to play a major role in determining the initial rate values for the reduction reactions. 1.0 0.8 c O 0.6 > 0.4 C o o 0.2 0.0 Silica + 4C ^ Silica + 3 C / V ^ /j^-^ \ Silica + 2C /OC^ Silica + C 0.350 1000 2000 3000 Time (seconds) 0.200 400 600 800 1000 Time (seconds) (a) (b) Figure 17: Stoichiometry Effect for Silica Reduction at 1 4 5 0 ° C . (a) Conversion version Time (b) Conversion Rate versus Time 4.1.3 Purge Gas Flow Rate Effect Experiments were performed at 4 different flow rates of Ar: 11, 22, 46 and 94 ml/s. A plot of conversion versus time for the various flow rates (Figure 18) shows that A r flow rate did not affect the reduction reaction. It was only when the flow rate dropped to 11 ml/s that a slight decrease in conversion appeared in the later stages of reduction. At this low flow rate, insufficient purging of the product gases probably led to this decrease. The Ar flowrate used in subsequent experiments was 22 ml/s. 53 0.0 1 ' 1 1 0 1000 2000 3000 Time (seconds) Figure 18: Conversion versus Time for Silica Reduction at 1 5 3 0 ° C at Argon Flow Rates of 11, 22, 46 and 94 ml/s. 4.2 Product Powder 4.2.1 X R D Results The product of the reduction of silica is either a-SiC which has a hexagonal structure or P-SiC which is cubic. In both these structures, every Si atom is tetrahedrally surrounded by four carbon atoms forming strong near-covalent bonds. In the Acheson process, crystals of a-SiC (the high temperature phase) were formed. In this study only P-SiC was found in the product powders. P-SiC is the preferred product since it is suitable for use in pressureless sintering. Figure 19 and Figure 20 are the XRD results of the reduction products from aluminosilicate and silica reduction. Meta-kaolin reduction (Figure 19) produced a product powder containing 54 mainly (X-AI2O3 and P-SiC. Silica reduction (Figure 20) produced a product powder consisting of only P-SiC with a small quantity of carbon left unreacted. Experimental weight losses were close to the theoretical values (Table 8). The difference between the actual and the theoretical weight losses were within 6 to 8% for silica reduction and 1 to 3% for meta-kaolin reduction. This provide support for the use of thermogravimetric analysis to measure reaction kinetics. Almost all weight loss was due to CO(g) formation as expressed in the overall reduction reaction (Eqn 13 on page 16). SiO(g) loss did not appear to be excessive. a-AI 2 0 3 P-SiC a-AljOj a-AI 2 0 3 a-AI 2 0 3 a-AI 2 0 3 a-AI 2 0 3 a-A ! 2 0 3 70 60 50 40 30 20 10 29 Figure 19: XRD of Meta-Kaolin/Carbon Reduction Products 55 C 20 Figure 20: XRD of Silica/Carbon Reduction Products Experiment Actual Weight Loss Theoretical Weight Loss Silica reduction at 1470°C 53.4% 58.3% Silica reduction at 1530°C 55.7% 58.3% Meta-Kaolin reduction at 1470°C 36.4% 38.1% Meta-Kaolin reduction at 1530°C 40.8% 38.1% Table 8: Actual versus Theoretical Weight Loss in Reduction Experiments 56 4.2.2 Morphology and Particle Size The terms Kevorkijan et al. [30] used to describe the morphology of powders are as follows: 1. Crystallite: a coherently diffracting region of a lattice. 2. Aggregate: a non-porous particle composed of crystallites. 3. Agglomerate: a porous cluster of crystallites or aggregates. 4. Whiskers: fibers within the micrometer/nanometer size range. 5. Fibers: millimeter or larger in size. The reactant and product powders were categorized using these terms. Meta-kaolin used in the reduction experiments was described as an aggregate of flake-like crystallites. The S E M photomicrograph (Figure 21) shows the typically layered structure of the aluminosilicate clays. Particle size analysis (Figure 22) shows a narrow particle size distribution with a median particle size of 2 u.m. Figure 21: S E M photomicrograph (5000X) of Meta-Kaol in . 57 40 i-30 -<u 20 -J3 0 2 4 6 8 10 Diameter (u.m) Figure 22: Particle Size Analysis for Meta-Kaol in . Silica flour (Figure 23) appears as a collection of angular crystallites with a wide particle size distribution (Figure 24). The larger particles of silica flour may account for its lower reduction rate when compared to the aluminosilicates. Figure 24 is the result of gradient sampling by the Horiba Particle Size Analyzer therefore it did not acquire the number percent values between 10 and 20 jam. Gradient sampling allowed a broader range of particle sizes to be acquired by the Horiba. The carbon reductant used in the reduction experiments was lamp black (Figure 25) which is an agglomerate of nanosized particles of carbon. This carbon source was highly reactive due to its small particle size (ie. high surface area). 58 Figure 23: S E M Photomicrograph (5000X) of Silica Flour. U u XI E 3 SK 5 10 15 Diameter (u,m) Figure 24: Particle Size Analysis of Silica Flour 59 Figure 25: S E M Photomicrograph (2000X) of L a m p Black. The product of carbothermic reduction of aluminosilicates was a composite of AI2O3 and SiC particles with a submicron particle size and uniform particle shape. The product powders had very low impurities content. Figure 26a shows the morphology of the precursor powders. The light coloured aggregates are the meta-kaolin while the darker phase is the lamp black. Figure 26b is the product powders resulting from reduction of the precursor powders at 1530°C in an argon atmosphere. There are significant amount of SiC whiskers distributed throughout the powders. From examination of the S E M photomicrograph the particle size of the product powder is estimated to be in the nanometer size range. 60 Figure 27 shows the particle size analysis for the meta-kaolin reduction product powders. A n average density between alumina and silicon carbide proportional to the expected ratio within the products was used for particle size determination. The particle size values from the plot should not be used quantitatively. The plot is a qualitative look at the spread in particle size distribution of the product powder. The result shows a narrow distribution range. 0.0 0.5 1.0 1.5 2.0 Diameter (fim) Figure 27: Particle Size Distribution of the AhOvSiC Composite Powders from Meta-Kaolin Reduction Figure 28a shows a S E M photomicrograph of the precursor silica/carbon powder. Silica crystallites are clearly seen embedded in the lamp black. The reduction products (Figure 28b) appear as very fine submicron particles with little whisker formation owing to the starting powders' high purity. Again from examination of the photomicrograph, the particle size is in the nanometer range. The photomicrographs illustrate how the SiC conformed to the morphology of the carbon reductant. 62 4.2.3 Impurities Chemical analyses of the precursor and product powders in the carbothermic reduction of meta-kaolin and kyanite were carried out by Min-En Labs, North Vancouver, B . C . (Table 9). Kaolinite (Georgia) Kyanite (Virginia) Phase Precursor Product Precursor Product A 1 2 0 3 43!5 54.12 56.14 61.18 Si phase (S i0 2 or SiC) 51.58 42.8 38.46 33.15 CaO 0.33 0.12 0.05 0.07 F e 2 0 3 0.51 0.63 1.65 2.95 M g O 0.34 0.02 0.02 0.01 K 2 0 0.41 0.073 0.25 0.35 N a 2 0 1.66 0.10 1.24 0.01 P 2 0 5 0.09 0.01 0.12 0.05 T i 0 2 1.55 2.00 1.93 2.10 Total 99.87 99.78 99.86 99.87 Table 9: Impurity Concentrations of Precursor and Product Powders. [12] Volatilization of impurities such as K 2 0 , P 2Os and N a 2 0 during carbothermic reduction greatly reduced their concentrations in the product powders. Kimura et al. [33] found that for sericite, an alkali aluminosilicate, much of the alkali components volatilized during the reduction process. The apparent K 2 0 increase for kyanite reduction is misleading. The physical amount of this alkali has been reduced but because of the reduction of other phases involved in the concentration calculation, the concentration of K 2 0 has been inflated relative to the phases that made up the impurity content of the product powder. The impurities CaO and F e 2 0 3 , which may serve as catalysts, remained in the product powder. Again, the concentration of these impurities appeared to have increased owing to the volatilization of other phases. The main impurity in the product powders was F e 2 0 3 . In the case of kyanite reduction, F e 2 0 3 made up 64 nearly 3% of the product powder. T i 0 2 , which made up 2% of the product, was not considered to be detrimental to the properties of the composite powder since it transformed into T i C . 4.3 Aluminosilicate versus Silica Reduction Figure 29 shows the plots of conversion versus time and conversion rate versus time for meta-kaolin/carbon, mullite/carbon and silica/carbon mixtures reacted at 1450°C. Figure 30 shows the conversion plots for the same mixtures reacted at 1550°C. The reduction rate values for meta-kaolin and mullite are closer to one another than to silica. At the lower temperature the difference in rate values between aluminosilicates and silica reduction is more pronounced (Figure 29b). Meta-kaolin transforms to mullite and free silica at the reaction temperatures and should have similar reduction rates as mullite. The lower silica reduction rate may be an effect of particle size. 0 1000 2000 3000 400 600 Time (seconds) Time (seconds) (a) (b) Figure 29: Meta-Kaolin, Mullite and Silica Reduction at 1450°C. (a) Conversion versus Time (b) Conversion Rate versus Time 65 Figure 30: Meta-Kaolin, Mullite and Silica Reduction at 1550°C. (a) Conversion versus Time (b) Conversion Rate versus Time Figure 31 shows the effect of temperature on meta-kaolin reduction. At 1350°C the conversion versus time plot is linear. This is likely due to a different rate limiting reaction mechanism than that at higher temperatures. 1.0 0.8 \-c O 0.6 tn i_ <D > 0.4 c o o 0.2 0.0 i 1 ] 1550*C / / / \ 1 510°C / U 5 0 ' C 1 5S0'C^_— ' 1000 2000 T ime ( s e c o n d s ) (a) 1.500 3000 co O 1.000 X p ?^ 0.500 00 0.000 1350"C 200 400 600 Time (seconds) (b) Figure 31: Temperature Effect for Meta-Kaolin Reduction, (a) Conversion versus Time (b) Conversion Rate versus Time 66 4.4 Reaction Parameters The following parameters were studied in order to gain more insight into the mechanism of the carbothermic reduction of aluminosilicates. 4.4.1 Direct Reduction with CO Bentsen et al. [6] showed that direct reduction of silica using CO( g ) was possible. In this study, a silica sample placed in a 100% C O atmosphere at 1530°C showed a linear plot with 5 to 8% weight loss after 2 hours. A baseline experiment with an identical sample at 1530°C but in argon atmosphere showed 0 to 1% weight loss. This proved that the weight loss was not due to volatilization of the silica. Silica powder easily sinters at these reaction temperatures, therefore a pore-free structure may have formed. C O ( g ) would be unable to penetrate the surface layer to reduce the underlying bulk mass. Only the surface layer would be reduced. Experiments with direct reduction of meta-kaolin with C O showed no weight loss. The reactant powder converted to a hard aggregate mass consisting of mullite and silica. Likewise, a clay/carbon mixture reacted at 1530°C in 100% C O atmosphere resulted in no SiC formation. This was consistent with results from early attempts to increase reduction rates by using enclosed crucibles with carbon or alumina lids. The resultant powders contained little or no SiC. The C O product gas must be purged from the system in order for reduction to occur. 4.4.2 The Effect of Particle Size Reaction mixtures were prepared using silica particles sizes of 2, 6, 35, 60 and 65 jam diameter mixed with a stoichiometric amount of lamp black. Figure 32 shows the resulting conversion and conversion rate plots. There appears to be a definite particle size effect. 67 0.8 0 .400 2 micron CO co 0 .300 o X 0.200 £ 0 .100 03 65 micron 0.0 0 .000 0 1000 2000 3000 400 600 800 Time ( s e c o n d s ) Time (seconds) (a) (b) Figure 32: The Effect of Silica Particle Size on Conversion Rate (Reacted at 1450°C). (a) Conversion versus Time (b) Conversion Rate versus Time Kyanite was milled and sieved into particle sizes of 35, 60, 80 and 135 um diameter. Reaction mixtures were made by mixing the kyanite powders with a stoichiometric amount of lamp black. Figure 33 shows the resulting conversion and conversion rate plots and there appears to be a definite particle size effect as well. The reason for using kyanite instead of meta-kaolin was because natural kaolinite only exists as very fine particles (2 Lim) with a narrow particle size range. Kyanite is a rock mineral which can be readily milled and separated into various particle size fractions. 68 0 1000 2000 400 600 800 T i m e ( s e c o n d s ) Time (seconds) (a) (b) Figure 33: The Effect of Kyanite Particle Size on Conversion Rate (Reacted at 1 5 3 0 ° C ) (a) Conversion versus Time (b) Conversion Rate versus Time 4.4.3 Si and S i C Seeding From the previous discussion on silica reduction, it was believed that Si<g> and SiC(S) may play important roles in the mechanism. The goal of this experiment was to compare a stoichiometric silica/carbon mixture seeded with 10 wt.% Si ( s ) to that of a non-seeded stoichiometric silica/carbon mixture. Figure 34 shows the conversion plot for both these powders reacted at 1530°C. The product powder from the seeded mixture was a loose agglomerate which showed no evidence of liquid phase formation even though the melting point of silicon metal is 1410°C. In a separate set of experiments Si(S) was found to react with C O at 1380°C (Figure 35) through the following reaction: Si ( s ) + CO(g) = SiC ( s) + — 02(g) 69 In one experiment the reaction occurred in a 100% C O environment and in the second experiment a 30:1 mixture of argon to C O was used. 1.0 > 0.4 0.2 0.0 1 1 Silico + C with ^ Si seeding -0 1000 2000 3000 T i m e ( s e c o n d s ) Figure 34: Conversion versus Time with or without Si Seeding. (Reacted at 1530°C in Ar Atmosphere) 0.4 0.0 0 500 1000 1500 Time (seconds) Figure 35: Si ( s ) Reaction with CO ( g ) at 1380°C. 70 The investigation of the effect of SiC seeding was to determine if the following sequence of reactions were occurring: SiC + 2 S i 0 2 = 3SiO(g) + C O ( g ) 3SiO(g) + 6C = 3SiC + 3 C O ( g ) Reduction of a stoichiometric mixture containing 17.26 wt.% SiC, 51.72 wt% S i 0 2 and 31.02 wt% carbon was compared with a standard stoichiometric silica/carbon mixture. The resulting conversion plot (Figure 36 a) at 1530°C showed a noticeable increase in the conversion rate due to the presence of SiC. 0 1000 2000 400 600 800 Time ( s e c o n d s ) Time (seconds) (a) (b) Figure 36: SiC Seeding at 1530°C. (a) Conversion versus Time (b) Conversion Rate versus Time 4.5 Reduction of Various Aluminosilicates Conversion and conversion rate plots (Figure 37) for bentonite, feldspar, illite, mica and syenite reduction at 1530°C are compared with the conversion and conversion rate plots for 71 meta-kaolin reduction. The purpose was to explore the possible use of less pure minerals and clays as precursor materials for the production of ceramic composite powders. The reducibility of these aluminosilicates was dependent on the formation of a liquid intermediate phase and the actions of the alkali impurities as catalysts. Studies by Khalafalla and Haas [31], found that mica had a reaction rate equivalent to quartz and meta-kaolin. From Figure 37c, it was observed that meta-kaolin and mica reduction do have similar reaction rates in the initial stage of the reduction process. The conversion rates of these aluminosilicates were similar to meta-kaolin reduction (at t = 200 seconds, values of the conversion rates for various aluminosilicates range between 0.9 and 1.3xl0"3 s"1). Only feldspar possessed a significantly lower reduction rate (Figure 37d). Figures 38-42 show the product powders' morphology. Bentonite (Figure 38) reduction produced a powder with a morphology similar to that of silica reduction. The SiC particles were in the nano-size range and had a narrow particle size distribution. There was some evidence of a liquid phase due to the convoluted shapes of some of the product phase. Feldspar reduction (Figure 39) produced a product with a large number of whiskers. Its appearance was similar to meta-kaolin product powders. Illite (Figure 40) and mica (Figure 41) reduction powders did not appear to contain many whiskers. Illite showed evidence of an intermediate liquid phase due to the stringy nature of its product powders. Syenite reduction (Figure 42) produced a highly inhomogeneous powder containing short whiskers, large angular crystallites and aggregates. 72 0.000 1 1 200 400 600 Time (seconds) (c) 0.0 1 ' 1 0 1000 2000 Time (seconds) (b) 1.500 | , r I 0.000 1 ' t— 200 400 600 Time (seconds) (d) Figure 37: Reduction of Various Aluminosilicates at 1530°C. (a) and (b) Conversion versus Time (c) and (d) Conversion Rate versus Time Figure 38: S E M Photomicrograph (6000X) of Product Powder from Bentonite Reduction. Figure 39: S E M Photomicrograph (6000X) of Product Powder from Feldspar Reduction Figure 40: S E M Photomicrograph (6000X) of Product Powder from Illite Reduction. Figure 41: S E M Photomicrograph (6000X) of Product Powder from M i c a Reduction. 75 Figure 42: S E M Photomicrograph (6000X) of Product Powder from Syenite Reduction 4.6 Reduction Using Various Carbon Sources Lamp black was the standard reductant used in the carbothermic reduction experiments. Several trial runs using metallurgical coke or graphite as alternatives to lamp black were performed. Figure 43 shows silica reduction with graphite compared with silica reduction with lamp black at 1530°C. Likewise, Figure 44 shows meta-kaolin reduction with coke compared with meta-kaolin reduction with lamp black at 1530°C. Lamp black produced the highest reduction rates in all cases. Lamp black has a small particle size that presumably accounts for its high reactivity. The difference in conversion rates between meta-kaolin reduction using coke or lamp black was not great. Coke may serve as a possible reductant for the carbothermic reduction of aluminosilicates if it was not critical to obtain highly pure product powders. 7 6 1.0 0.0 silica+graptiite 1000 2000 3000 Time (seconds) (a) 0.800 CO *0 0.600 5 >? 0.400 TO CD CO 0.200 silica+lampblack silica + graphite 400 600 Time (seconds) (b) 800 Figure 43: Silica Reduction with Graphite and Lamp Black at 1530°C. (a) Conversion versus Time (b) Conversion Rate versus Time 1.0 0.0 1.000 1000 2000 3000 Time (seconds) (a) w 0 . 6 0 0 0.500 400 600 Time (seconds) (b) 800 Figure 44: Meta-Kaolin Reduction with Coke and Lamp Black at 1530°C. (a) Conversion versus Time (b) Conversion Rate versus Time 77 The effect of particle size of the reductant on the reaction rate was examined. Metallurgical coke was milled and separated into various particle size fractions. The resulting kinetic plots did not show the expected particle size effect. The results pointed to the importance of the carbon surface area in determining the reduction rate. Figure 45 shows the conversion and conversion rate plots for the two particle sizes of coke. The larger coke particles showed a higher reduction rate. The reason for this can be deduced by examining the S E M photomicrographs (Figure 46 and Figure 47) of the coke particles. The milling and sieving process that created these two particle sizes of coke caused the preferential separation of small, hard, angular, non-carbonaceous particles into the smaller size ranges (Figure 46). The larger coke particles (Figure 47) show a highly porous structure which resulted in high surface area. The high surface area of the larger particle size accounted for the higher reduction rates. Figure 45: Meta-Kaolin Reduction with Coke Particles Sizes of 250 and 50 u,m at 1530°C. (a) Conversion versus Time (b) Conversion Rate versus Time 78 4.7 N2 Reaction Atmosphere The use of nitrogen as a purge gas depended on the ability to suppress the nitridation step so that SiC would be retained in the reduction products. Figures 48 and 49 show the XRD plots for the silica reduction products reacted in nitrogen atmosphere at 1430°C and 1530°C. Reduction at 1430°C showed no SiC in the product powders. Reduction at 1530°C showed some evidence of SiC but not to the same extent as obtained in an argon atmosphere. No silicon nitride or silicon oxynitrides were detected in the product powder, probably a result of the low N 2 flow rate (22 ml/s). This result showed that SiC can be formed and retained in a nitrogen atmosphere. Previous studies [2, 16, 25, 38] which reported the successful formation of S13N4 from pure Si02 and sialon from kaolinite involved optimized conditions using a high N 2 flow rate and higher reaction temperatures. Vi C a-Si0 2 a-SiOi a-SiOj a-Si0 2 a -Si0 2 70 60 50 40 30 20 10 20 Figure 48: XRD Plot of the Silica Reduction Product Powder Reacted in Nitrogen Atmosphere at 1430°C. 81 C 0) -SiO: 70 -Si02 P-SiC/C p-SiC J a-Si02 a-Si02 | «-Si02 a-Si03 60 50 40 30 20 10 20 Figure 49: XRD Plot of the Silica Reduction Powder Reacted in Nitrogen Atmosphere at 1530°C. 82 5.0 Discussion In Chapter 2, the results of studies investigating the reduction mechanism and the rate limiting step of the carbothermic reduction of silica were presented. There was strong support for the two step mechanism but the opinions on the rate limiting step were conflicting. Both SiO formation (Eqns 7 and 8 on page 14) and SiC formation (Eqn 10 on page 14) were favoured by researchers. 5.1 Analysis of the Reduction Mechanism The meta-kaolin reduction mechanism was simply an extension of the silica reduction mechanism. The alumina phase was not involved in the mechanism. Reduction of aluminosilicates began with mullite formation followed by silica and mullite reduction. 5.1.1 Mullite Reduction Mechanism There is general agreement among researchers that mullite and free silica are formed at the reaction temperatures during the reduction process. However, the effect of mullite formation on the overall reduction has not been clarified. By following the slope of a conversion curve (Figure 50), it can be seen that there does not exist any discontinuity when the change from free silica reduction to mullite reduction occurs. It appears that the reduction of the free silica does not occur first and then followed by the reduction of the silica within mullite. This suggests that the reduction rate of the free silica and the silica within the mullite phase are subject to the same rate limiting step. This would be supported by assuming the SiC formation step (Step 2) to be rate limiting. 83 1.0 c o o Free silica reduction ends mullite reduction begins. ~ 0.0 0 1000 2000 3000 Time (seconds) Figure 50: Conversion versus Time for Meta-Kaol in /Carbon Reduction at 1 5 1 0 ° C . 5.1.2 Silica Reduction Mechanism Si seeding did not affect the reduction rate, therefore the four step reduction mechanism in which Si played an important role is not the correct reduction mechanism for the cases studied here. In the experiments performed by Kennedy and North [29], they used high concentrations of Si (equimolar) to achieve their results which showed an influence of Si<S) addition on reaction rates. The results were probably due to the SiC formation from Si and C O . SiC seeding introduced an alternative path to the reduction of silica. This path would occur in the latter half of the reduction process when the concentration of SiC became significant. Since it was the initial part of the kinetic curves which were of concern, the SiC involvement in the latter stages of the reduction mechanism have not been examined in great detail. It cannot be concluded that the addition of SiC increased the reduction rate of silica. The measured increased reaction rate may be due to the combined effect of the reduction of silica as well as the oxidation of the seeded SiC phase. 84 5.2 Determination of the Rate Limiting Step Figure 51 displays the plots of the following rate laws: • In X versus / (1st Order Kinetics), • 1 - (1 - X)m versus t (Contracting Volume Model), • [ 1 - (1 - X)m ]2 versus t (Jander's Equation) and • [-ln( 1 - X)]1/n versus t (Avrami-Erofe'ev Equation), applied to silica reduction and meta-kaolin reduction at 1530°C. The Contracting Volume Model and Jander's Equation displayed the most linear fit of the kinetic data. Meta-kaolin reduction plots (Figure 52) at temperatures 1550, 1510, 1470 and 1450°C show a linear relationship when plotted using the Contracting Volume Model. The best fitting rate equation does not necessarily define the rate limiting mechanism. These rate equations were developed to model complex heterogeneous reactions. 1 —7— 1st Order / -Avrami-Erofe'ev / Contracting Volume ——" Jander's Eqn ' i 0 1000 2000 Time (seconds) Jander's Eqn 500 1000 1500 Time (seconds) (a) (b) Figure 51: Reduction Models for (a) Silica Reduction (b) Meta-Kaolin Reduction both at 1530°C. 85 0.0 I 1 1 1000 2 0 0 0 Time (seconds) Figure 52: Meta-Kaolin Reduction at 1550,1510, 1470 and 1450°C are plotted using the Contracting Volume Model. Figure 53 shows the plots of reaction rate versus the inverse of the particle's radius ( r 1 ) for silica and kyanite reduction. The surface area of a powder compact (A) is inversely 3W proportional to the radius of the particles (ie. A = , where Wis the weight, r is the average rp radius and p is the density of the particle). As can be seen from the plots there is a linear relationship indicating that the surface area of the particle plays a significant role in the rate of reaction for both silica and kyanite reduction. This evidence indicates that the SiO formation step may be the rate limiting step in the reduction mechanism. But as shown earlier, there is also a relationship between carbon particle size and reaction rate which supports the SiC formation step as being rate limiting. This indicates the possibility that the reduction reaction is under mixed control. These apparent contradictory results were the reasons why the investigation of the rate limiting step was not conclusive. Whether it was mass transfer controlled governed by the 86 Jander's Equation or it was chemical reaction controlled determined by the Contracting Volume Model, the experiments from this study have not been able to support either hypotheses. '</) in O 3 X 2 "co cn -2 L i n y ; 35 (im / • 6 urn -• / / > 6 0 Lim • — 65 um 0.0 0.2 0.4 0.6 0.8 1.0 Particle Radius"1 (Lim)" 1 (a) 8 r 7 -'(/) 6 -o 5 -X 0) 4 : 00 cn 3 -2 35 Lim 60 Lim 135 L i m / . 8 0 ^ -• / 1 0.01 0.02 0.03 0.04 0.05 0.06 Particle Size"1 (jim)"1 ( b ) Figure 53: Reaction Rate versus Inverse of Average Particle Radius of Powder. (Surface area of powder « r"1) (a) Silica Reduction and (b) Kyanite Reduction 5.3 Determination of Activation Energy The temperature dependence of the reduction reactions was used to determine the activation energy for silica, mullite and meta-kaolin reduction. The kinetic data were plotted using the best fitting kinetic model. For example, the Contracting Volume Model was applied to meta-kaolin (Figure 52) and mullite reduction while Jander's Equation was applied to silica reduction. From these plots the rate constants (slopes of the linear portion of the plots) were calculated. The Arrhenius equation relates the reaction rate to temperature as follows: k = ,4exp(——) RT 87 where A is a constant, Q is the experimental activation energy, R is the Gas Constant and T is the temperature. Plots of In k versus 1/T (Figure 54) were made to determine activation energies. -7 | , , , -7 6.00 _ 1 3 I ' • 5.40 5.60 5.80 6.00 1/Tx10 4 (c) Figure 54: In k versus 1/T plots for (a) Meta-Kaolin (b) Mullite and (c) Silica Reduction The experimental activation energies were compared with published results (Table 10). The experimental values fell within the range of those found in literature. It could be seen that the aluminosilicate activation energy values were significantly lower than the silica values,' The 88 difference was approximately 100 kJ/mole. This effect may be due to the presence of the porous alumina phase formed during aluminosilicate reduction which facilitated the movement of the product gases away from the reaction interface. The silica particles formed from the decomposition of meta-kaolin and mullite were amorphous and in the nano-size range which may also account for this difference. Reference Precursor Powders Q (kJ/mole) Kennedy and North silica/graphite 466 Khalafalla and Haas silica/graphite 322 (1972) Klingeretal. (1966) silica/graphite 510 Lee and Cutler (1975) silica/charcoal 544 Shimoo [53] silica/carbon black 544 silica/activated carbon 244 Viscomi and Himmel silica coke breeze 552 (1978) Weimer etal. (1993) silica/carbon black 382 This study silica/lamp black 3 9 0 ± 3 0 mullite/lamp black 280+10 meta-kaolin/lamp black 2 6 0 ± 5 Table 10: Comparison of Activation Energies for Silica, Mullite and Meta-Kaolin Reduction 5.4 Alternative Resources for AI203-SiC Powder Production 5.4.1 Various Aluminosilicate Clays and Minerals In exploring the carbothermic reduction of various aluminosilicates it was hoped that an inexpensive source of composite powders for non-critical uses (ie. refractories) may be developed. The results of this study showed products of bentonite reduction as having very similar microstructure to that of the silica reduction products. Also noteworthy was the fact that 89 feldspar produced a reduction product which contained numerous whiskers. The morphology of this powder was similar to the meta-kaolin product powder. The existence of whiskers in the product powders has been linked to the presence of Fe2C>3 which exists as an impurity in the reactant powders. The whiskers were formed via a mechanism that involved a ferrosilicon phase. If Fe20*3 existed as a free oxide in the ceramic composite powder then it would lower the melting point of the composite. The formation of whiskers may in fact be a positive side effect of the reduction process. These whiskers would provide a strengthening mechanism for products manufactured using the composite powders. 5.4.2 Purge Gases The purge gas used in most of the reduction experiments was argon. Purging the CO(g) formed during the process was critical for achieving conversion. Due to the high cost of argon, nitrogen was examined as an alternative purge gas. Results showed that SiC was retained in the product powders during silica reduction at temperatures greater than 1500°C within a N 2 environment. The yield of SiC was poor due to the fact that the experimental conditions required to promote retention of SiC also inhibited its yield. More study is required before this route can be used to produce SiC-Al2C>3 composite powders. 90 6.0 Summary and Conclusion 1. A study was made of the kinetics of the reduction of silica and aluminosilicates in the presence of carbon. 2. For this purpose, isothermal weight loss experiments leading to the formation of SiC and A l 2 0 3 - S i C were carried out in the temperature range of 1450 to 1550°C in an Ar environment. The basic reactions for silica reduction are as follows: S i 0 2 + 3C = SiC + 2 C O and for meta-kaolin reduction. 3(Al 2 0 3 .2Si0 2 ) + 18C = 3A1 2 0 3 + 6SiC + 12CO 3. In order to understand the reaction mechanism and the rate limiting step, the effect of Si and SiC seeding, the effect of silica, aluminosilicate and carbon particle size and the effect of C O atmosphere were investigated. 4. The mechanism by which aluminosilicates are reduced is analogous to the silica reduction mechanism. The analysis of product powders clearly showed that alumina was not reduced at the reaction temperatures. Silica, as a pure phase or in mullite was reduced as follows: S i 0 2 + C = SiO(g) + C O Step (1) SiO ( g) + 2C = SiC + C O Step (2) The Si seeding experiments showed that Si did not influence the reaction rate and therefore the mechanism did not involved a four step mechanism as was 91 proposed by certain researchers [37]. The effect of C O ( g ) in the reaction atmosphere was a lowering of the extent of reaction for SiC formation. This supported the overall mechanism shown above. 5. The effect of kyanite and silica surface area indicated that silicon monoxide formation may be the rate limiting step. The effect of coke particle size indicated that silicon carbide formation may be the rate limiting step. This apparent contradiction shows that a mixed controlled mechanism may be occurring. 6. The experimental data were fitted to various empirical rate models. The best fit to any model does not necessarily mean that the rate limiting step was defined by the mechanism from which the rate model was derived. Jander's equation (diffusion control through a product layer) provided the best fit for the data from the silica reduction experiments. The Contracting Volume model (chemical reaction rate control) provided the best fit for the data from the meta-kaolin reduction experiments. 7. The temperature dependence of the reaction rates was used to determine activation energies for meta-kaolin, mullite and silica reduction. The activation energy for meta-kaolin and mullite reduction is approximately 260 to 290 kJ/mole. The activation energy for silica reduction is approximately 390 kJ/mole, similar to values found in literature. The difference in activation energy between aluminosilicate reduction and silica reduction cannot be explained at present. It may be an effect of particle size or the nature of the silica phase formed from the decomposition of meta-kaolin and mullite. 8. Various alternative sources of reactants were investigated in order to lower the production costs. The reduction of various natural alkali aluminosilicates showed reduction rates that were similar to meta-kaolin reduction but the product powders may only be suitable for low technology applications due to their high impurity content. Various carbon sources were tested and they proved effective in their role as the solid carbon reductant. Use of nitrogen as the purge gas was shown to be compatible with the formation and retention of SiC in the product powders but with poor yield results compared to an argon purged system. 93 7.0 Further Studies This study has further explored the reaction kinetics of the carbothermic reduction of aluminosilicates. There is still considerable work which remains to be done before this process can be used commercially. Some recommended investigations to further this goal are: 1. The effect of particle size as a function of temperature and time requires further study in order to gain more insight into the rate limiting step of the reduction mechanism. 2. The catalytic effect of CaO and Fe203 on the kinetics of the reduction process should be investigated. 3. The whisker formation mechanism should be studied. It is known that iron oxides enhanced the formation of SiC whiskers [15]. It is possible that other oxides may also be participants in this mechanism. 4. As mentioned earlier, the need to keep costs down is vital for the commercial viability of the carbothermic reduction process. The use of argon as the purge gas is one of the more expensive components of this process and therefore nitrogen was tested as an alternative. Further testing is required to boost the yield of the product powders. An AT-N2 mixture might lower costs significantly compared to using 100% Ar. 94 Bibliography 1. Alcala, M . D . , Real, C , Criado, J . M . "Application of Constant Rate Thermal Analysis ( C R T A ) to the Synthesis of Silicon Nitride by Carbothermal Reduction of Silica." Journal of Thermal Analysis. Vol . 38 (1992), pp. 313-319. 2. Bandyopadhyay, S., Mukerji, J. "Reaction Sequence in the Synthesis of Silicon Nitride from Quartz." Ceramics International. Vol . 17 [3] (1991), pp. 171-179. 3. Bechtold, Bryant Coffin. "Carbothermal Reduction of Kaolin Clay and Resulting Silicon-Carbide Formation." Diss. Abstr. Int. B 1979, 39 (9), 4516. 4. Bechtold, Bryant C , Cutler, Ivan B . "Reaction of Clay and Carbon to Form and Separate A 1 2 0 3 and SiC." Journal of the American Ceramic Society. 63 [5-6] (1980), pp. 271-275. 5. Bechtold, Bryant Coffin. Carbothermal Reduction of Kaolin Clay and Resulting Silicon Carbide Formation. PhD Thesis, University of Utah, August 1978. 6. Bentsen, S., Jorgense, S., Wiik, K . , Motzfeldt, K. "Reactions between Carbon and the Oxides Mullite and Silica." Materials Science Monographs. Vol . 28A (1985), pp. 621-626. 7. Biernacki, J.J., Wotzak, G.P. "Stoichiometry of the C + S i 0 2 Reaction." Journal of the American Ceramic Society. 73 [1] (1989), pp. 122-129. 8. Blaha, H . , Komarek, K . L . "The Reduction of Silica with Graphite." High Temperature Science. 28 (1989), pp. 87-97. 9. Brindley, G.W. , Nakahira, M . "The Kaolinite-Mullite Reaction Series: I, A Survey of Outstanding Problems." Journal of the American Ceramic Society. 42 [7] (1959), pp. 311-324. 10. Brown, I .W.M. , MacKenzie, K . J . D . , Bowden, M . E . , Meinhold, R . H . "Outstanding Problems in the Kaolinite-Mullite Reaction Sequence Investigated by 2 9 S i and 2 7 A l Solid-State Nuclear Magnetic Resonance: II, High-Temperature Transformations of Metakaolinite." Journal of the American Ceramic Society. 68 [6] (1085), pp. 298-301. 11. Chaklader, A . C . D . "Synthesizing In-Situ Ceramic Composite Powders for Thermal Spraying." Project Report, University of British Columbia, 21 Feb 1991. 12. Chaklader, A . C . D . , Lin, E . C . Y . "Synthesis of Ceramic-Ceramic Composite Powders from Natural Mineral Precursors." Journal of Materials Synthesis and Processing. Vol . 1, No. 3 (1993), pp. 145-152. 13. Chaklader, A . C . D . , Wong, S.C., Lin, E . C . Y . " A l 2 0 3 - S i C Composites from Natural Mineral Precursors." The 3rd I U M R S International Conference on Advanced Materials. Tokyo, Japan, 1993. 95 14. Chakravorty, A . K . , Ghosh, D . K . "Kaolinite-Mullite Reaction Series: The Development -and Significance of a Binary Aluminosilicate Phase." Journal of the American Ceramic Society. 74 [6] (1991), pp. 1401-1406. 15. Chrysanthou, A . , Grieveson, P., Jha, A . "Formation of silicon carbide whiskers and their microstructure." Journal of Materials Science. 26 (1991), pp. 3463-3476. 16. Cho, Y.W., Charles, J.A. "Synthesis of nitrogen ceramic powders by carbothermal reduction and nitridation. Part I: Silicon nitride." Materials Science and Technology. Vol . 7 (April 1991), pp. 289-298. 17. Cranmer, D . C . "Overview of Technical, Engineering and Advanced Ceramics." Ceramics and Glasses. Vo l 4: Engineered Materials Handbook. A S M International, 1991. 18. Cutler, R . A . , Virkar, A . V . , Holt, J.B. "Synthesis and Densification of Oxide-Carbide Composites." Ceramic Engineering and Science Proceedings. Vol . 6, No. 7-8 (Jul-Aug 1985). 19. van Dijen, F . K . , Metselaar, R. "Reaction-Rate-Limiting Steps in Carbothermal Reduction Processes." Journal of the American Ceramic Society. 69 [1] (1985), pp. 16-19. 20. van Dijen, F . K . , Vogt, U . "The Chemistry of the Carbothermal Synthesis of a - S i 3 N 4 : Reaction Mechanism, Reaction Rate and Properties of the Product." Journal of the European Ceramic Society. 10 (1992), pp. 273-282. 21. Dosaj, V . D . , Trunzo, J .E . , Filsinger, D . H . "Addition of Calcium Compounds to the Carbothermic Reduction of Silica." European Patent Application E P 272.777. (29 June 1988). 22. Durham, S.J.P., Shanker, K . , Drew, R . A . L . "Carbothermal Synthesis of Silicon Nitride: Effect of Reaction Conditions." Journal of the American Ceramic Society. 74 [1] (1991), pp. 31-37. 23. Filsinger, D . H . , Bourrie, D . B . "Silica to Silicon: Key Carbothermic Reactions and Kinetics." Journal of the American Ceramic Society. 73 [6] (1990), pp. 1726-32. 24. Henderson, J.B., Tant, M . R . " A Study of the Kinetics of High-Temperature Carbon-Silica Reactions in a Ablative Polymer Composite." Polymer Composites. Vol . 4, No. 4 (1983), pp. 233-237. 25. Higgins, I., Hendry, A. "Production of P'-Sialon by Carbothermal Reduction of Kaolinite." Journal of British Ceramics Transactions. 85 [5] (1986), pp. 161-166. 26. Hutchison, S .G. , Richardson, L .S . , Wai, C M . "Carbothermic Reduction of Silicon Dioxide—A Thermodynamic Investigation." Metallurgical Transactions B . Vol . 19B (Apr 1988), pp. 249-253. 96 27. J A N A F Thermochemical Data. Army-Navy-Air Force Thermochemical Panel, The Dow-Chemical Co. , Midland, Mich. , 1962-63. 28. Janney, M . A . , Wei, G . C . , Kennedy, C.R., Harris, L . A . "Carbothermal Synthesis of Silicon Carbide." Report No. 9514868, Oak Ridge National Lab., T N , U S A , May 1985. 29. Kennedy, P., North, B . "The Production of Fine Silicon Carbide Powder by the Reaction of Gaseous Silicon Monoxide with Particulate Carbon." Proceedings of the British Ceramic Society. No. 33 (May 1983), pp. 1-15. 30. Kevorkijan, V . M . , Komac, M . , Kolar, D . "Low-temperature synthesis of sinterable SiC powders by carbothermic reduction of colloidal Si0 2 ." Journal of Materials Science. 27 [10] (1992), pp. 2705-2712. 31. Khalafalla, S.E., Haas, L . A . "Carbothermal Reduction of Siliceous Minerals in Vacuum." High Temperature Science. 2 (1970), pp. 95-109. 32. Khalafalla, S.E., Haas, L . A . "Kinetics of Carbothermal Reduction of Quartz Under Vacuum." Journal of the American Ceramic Society. 55 [8] (1972), pp. 414-417. 33. Kimura, I., Hotta, N . , Matsuda, G , Saito, N . , Yasukawa, S. "Preparation of SiC-Al203 Composite Powder." Journal of Materials Science Letters. 6 (1987), pp. 1359-1360. 34. Kingery, W . D . , Bowen, H . K . , Uhlmann, D.R. Introduction to Ceramics. 2nd E d . , John Wiley & Sons, Inc., Toronto, 1976. 35. Klein, C , Hurlbut Jr., C.S. Manual of Mineralogy. 20th E d . , John Wiley & Sons, Toronto, 1985. 36. Klinger, N . , Strauss, E . L . , Komarek, K . L . "Reactions Between Silica and Graphite." Journal of the American Ceramic Society. 49 [7] (1966), pp. 369-375. 37. Lee, J . - G . , Cutler, I.B. "Formation of Silicon Carbide from Rice Hulls." Ceramic Bulletin. Vol . 54, No. 2 (1975), pp. 195-198. 38. Lee, J . -G . , Cutler, I.B. "Sinterable Sialon Powder by Reaction of Clay with Carbon and Nitrogen." Ceramic Bulletin. Vol . 58, No. 9 (1979), pp. 869-871. 39. Lee, J . - G . , Miller, P.D., Cutler, I.B. "Carbothermal Reduction of Silica." Reactivity of Solids (Proceedings of the 8th International Symposium). (1976), pp. 707-11. 40. Lee, June-Gunn. Carbide and Nitride Ceramics by Carbothermal Reduction of Silica. PhD Thesis, University of Utah, June 1976. 41. Lehman, R . L . "Primer on Engineering Ceramics." Advanced Materials & Processes. 6/92, pp. 31-41. 42. Mahan, B . H . University Chemistry 3rd E d . . Addison-Wesley Publishing Co. Inc., 1975, pp 365-375. 97 43. Misra, A . K . "Thermochemical Analysis of the Silicon Carbide-Alumina Reaction with Reference to Liquid-Phase Sintering of Silicon Carbide." Journal of the American Ceramic Society. 74 [2] (1991), pp. 345-51. 44. Murthy, M . K . Advanced Ceramics. Report No. 3/86, Office of Industrial Innovation, Dept. of Regional Industrial Expansion, Ottawa, Jan. 1986. 45. Murray, H . H . "Clay minerals for advanced ceramics." Mining Engineering. (Nov 1989), pp. 1123-25. 46. Namba, T. , Koyama, T. , Imada, G . , Kanno, M . "Kinetics of the Carbothermic Reduction of a T h 0 2 + C Mixture to Prepare (Th,U)C." Journal of Nuclear Materials. 150 (1987), pp. 226-232. 47. Niihara, K . "Nanostructure design and mechanical properties of ceramic composites." Journal of the Japan Society of Powder and Powder Metallurgy. 37 [2] (Feb 1990), pp. 348-351. 48. Norton, F . H . Fine Ceramics. Technology and Applications. McGraw-Hill , 1970. 49. Ono, K . , Kurachi, Y . "Kinetic studies on P-SiC formation from homogeneous precursors." Journal of Materials Science. 26 (1991), pp. 388-392. 50. Paull, J . M . The Interaction of Silicon Monoxide Gas with Carbonaceous Reducing Agents. Report No. 1969, National Institute for Metallurgy, Randburg, South Africa, May 1978. 51. Penugonda, M R . "Alumina-Silicon Carbide Composites from Kaolinite-Carbon Precursors by Hot-Pressing." M . A . S c . Thesis, Dept. of Metallurgical Engineering, University of British Columbia, Dec. 1987. 52. Sheppard, L . "Advances in Nanophase Ceramics." Advanced Materials & Processes. 10/94, pp 25-27. 53. Shimoo, T. , Okamura, K . "Kinetics of Synthesis of SiC by Carbothermic Reduction of Silica." Journal of the Japan Society of Powder and Powder Metallurgy. 38 [2] (Feb 1991), pp. 121-126. 54. Shimoo, T. , Inagaki, H . "Effect of A 1 2 0 3 Addition on Rate of Carbothermic Reduction of Si0 2 ." Journal of the Japan Institute of Metals. Vol . 55, No. 2 (1991), pp. 186-193. 55. Shimoo, T. "Effect of Carbonaceous Materials on Kinetics of Carbothermic Reduction of Si0 2 ." Journal of the Japan Institute of Metals. Vol . 54, No. 1 (1990), pp. 41-47. 56. Siddiqi, S.A., Hendry, A . "The influence of iron on the preparation of silicon nitride from silica." Journal of Materials Science. 20 (1985), pp. 3230-3238. 98 57. Srinivasan, N.S. "Theoretical analysis of carbothermic reduction of metal oxides proceeding through gaseous intermediates." Scandinavian Journal of Metallurgy. 19 (1990), pp. 72-78. 58. Vodop'yanov, A . G . , Baranov, S.V., Kozhevnikov, G . N . , Zakharov, R . G . "Reactions between Mullite and Carbon." Russian Metallurgy (Metally). No . 4 (1984), pp. 25-29. 59. Wada, H . , Wang, L . "Effect of Gas Phase Composition of SiC and S i 3 N 4 Formation." Ceramic Engineering Science Proceedings. 11 [9-10] (1990), pp. 1463-1479. 60. Wahl, F . M . , Grim, R . E . , Graf, R .B . "Phase Transformations in Silica-Alumina Mixtures as Examined by Continuous X-Ray Diffraction." The American Mineralogist. Vol . 46, Sept-Oct, 1961. 61. Wang, H . , Fischman, G.S. "In Situ Synthesis of Silicon Carbide Whiskers from Silicon Nitride Powders." Journal of the American Ceramic Society. 74 [7] (1991), pp. 1519-22. 62. Wang, M . - J . , Wada, H . "Synthesis and characterization of silicon nitride whiskers." Journal of Materials Science. 25 (1990), pp. 1690-1698. 63. Wang, L . , Wada, H . , Allard, L . F . "Synthesis and characterization of SiC whiskers." Journal of Material Research. Vol . 7, No. 1 (Jan 1992), pp. 148-163. 64. Wei, G . C . , Morgan, C.S., Kennedy, C.R., Johnson, D.R. Synthesis. Characterization and Fabrication of Silicon Carbide Structural Ceramics. Oak Ridge National Lab, T N , U S A , 1982. 65. Weimer, A . W . , Nilsen, K . J . , Cochran, G . A . , Roach, R.P. "Kinetics of Carbothermal Reduction Synthesis of Beta Silicon Carbide." A I C h E Journal. Vol . 39, No . 3 (Mar 1993), pp. 493-503. 99 Experimental Summary Exp ID Date Powder Temp Atm Comment Initial Wt Final Wt 5KYC15A 02.9.1993 KY+C 1530 AR PARTICLE SIZE 0.407 0.301 6KYC15A 02.9.1993 KY+C 1530 AR PARTICLE SIZE 0.41 0.298 7KYC15A 02.9.1993 KY+C 1530 AR PARTICLE SIZE 0.405 0.28 10SS53C 04.1.1994 SK32+SIC 1530 CO 0.498 0.51 1AS53C 04.1.1994 AL+2S1 1530 CO 0.504 0.5 1K53C 04.1.1994 KAOUNITE 1530 CO 0.502 0.493 SSI C145 04.5.1992 QUART2+C 1450 AR PARTICLE SIZE 0.41 0.299 6SI C145 04.5.1992 QUARTZ+C 1450 AR PARTICLE SIZE 0.402 0.306 7SI C145 04.5.1992 QUARTZ+C 1450 AR PARTICLE SIZE 0.403 0.347 8SI C145 04.5.1992 QUARTZ+C 1450 AR PARTICLE SIZE 0.406 0.368 10SCS53A 06.1.1994 SI02+3C+SI 1530 AR SI SEEDING 0.504 0.26 20SSC53A 06.1.1994 SIC+2SI02+6C 1530 AR SIC REDUCTION 0.509 0.256 30SC53A 06.1.1994 SI02+4C 1530 AR STOICHIOM ETRY 0.505 0.244 31SC53A 06.1.1994 SKD2+3C 1530 AR 0.498 0.238 2KA013 06.3.1992 KAO+C 1350 AR 0.345 0.284 1SC43N 07.1.1994 SKD2+C 1430 N2 0.505 0.43 1SC53N 07.1.1994 SI02+C 1530 N2 0.511 0.333 1KC53A 07.10.1993 MKAO+C 1530 AR REPRODUCIBILITY 0.506 0.299 2KC53A 07.10.1993 MKAO+C 1530 AR REPRODUC1BIUTY 0.503 0.3 3KC53A 07.10.1993 M KAO+C 1530 AR REPRODUCIBILITY 0.503 0.298 4KC53A 07.10.1993 M KAO+C 1530 AR REPRODUCIBILITY 0.501 0.294 1AR0S 08.9.1993 SI02+C 1530 AR AR FLOW RATE 0.409 0.19 1AR30 06.9.1993 SI02+C 1530 AR AR FLOW RATE 0.404 0.169 1AR60 08.9.1993 SI02+C 1530 AR AR FLOW RATE 0.407 0.178 2KA014 09.3.1992 KAO+C 1450 AR 0.394 0.23 3KA014 09.3.1992 KAO+C 1450 AR 0.661 0.381 2SYC53A 1.10.1993 SYENITE+C 1530 AR GRAPHITE CRUC 1.015 0.396 1SCS3A 1.10.1993 SI02+C 1530 AR GRAPHITE ROD 1.568 1.194 2SCS3A 1.10.1993 SI02+C 1530 AR 0.996 0.658 1KA015 10.3.1992 KAO+C 1500 AR 0.433 0.249 2KA015 10.3.1992 KAO+C 1500 AR 0.677 0.385 1KA0135 11.3.1992 KAO+C 1350 AR 0.431 0.326 1MKA01S1 11.6.1992 M KAO+C 1510 AR 0.406 0.243 1MUL151 11.6.1992 MULL+C 1510 AR 0.41 0.309 1SI 151 11.6.1992 SI02+C 1510 AR 0.411 0.182 2SI C147 11.6.1992 SI02+C 1470 AR 0.42 0.224 WFIB C 11.6.1992 PROD FIBERS+C 1450 AR 0.413 0.372 1SI C145 12.3.1992 SK52+C 1450 AR 0.632 0.371 1MKA0155 12.6.1992 M KAO+C 1550 AR 0.407 0.259 1MUL155 12.6.1992 MULL+C 1550 AR 0.416 0.307 1SI C155 12.6.1992 SI02+C 1550 AR 0.411 0.145 2MKA0155 12.6.1992 MKAO+C 1550 AR 0.415 0.243 22SC1SA 13.7.1993 SI02+C 1530 AR CRUCIBLE DESIGN 0.404 0.167 23SC15A 16.7.1993 SK32+C 1530 AR CRUCIBLE DESIGN 0.409 0.171 1MUL145 19.3.1992 MULL+C 1450 AR 0.696 0.548 1SCA15C 20.7.1993 SI02+CAO 1530 CO CAO SEEDING 0.407 0.351 20SC14A 20.8.1993 SK52+C 1450 AR APPARATUS 0.406 0.234 3SG15A 20.9.1993 SI02+GRAPHITE 1530 AR 0.409 0.304 2SF15C 21.7.1993 SK32+FEO 1530 CO FEO SEEDING 1SYC15A 21.9.1993 SYENITE+C 1530 AR 0.408 0.238 2SCA15C 22.7.1993 SI02+CAO 1530 CO 0.407 0.345 1ASIL151 23.6.1992 SI02+C 1510 AR 0.414 0.208 2MKA0151 23.6.1992 MKAO+C 1510 AR 0.418 0.253 2MUL1S1 23.6.1992 MULL+C 1510 AR 0.405 0.315 2SI C151 23.6.1992 QUARTZ+C 1510 AR 0.421 0.21 1IC53A 23.9.1993 ILLITE+C 1530 AR 1.025 0.408 1MC53A 23.9.1993 MICA+C 1530 AR 1.007 0.539 3KA0145 24.4.1992 KAO+C 1450 AR 0.489 0.288 3SI C145 24.4.1992 a-SI02+C 1450 AR 0.502 0.263 1SILCO 24.6.1992 SILICA 1500 AR/CO 0.414 0.414 1 SILICON 24.6.1992 SILICON 1380 AR/CO 0.417 0.431 21SC14A 24.8.1993 SI02+C 1450 AR APPARATUS 0.411 0.272 2MUL145 26.3.1992 MULL+C 1450 AR 0.416 0.327 2SI C145 26.3.1992 SI02+C 1450 AR 0.4 0.229 20SC1SA 26.7.1993 SI02+C 1530 AR APPARATUS 0.408 0.172 21SC15A 26.7.1993 SK52+C 1530 AR APPARATUS 0.408 0.17 1SCSI15A 27.7.1993 SI02+SI+3C 1530 AR SI SEEDING 1SSI15C 27.7.1993 SI02+SI 1530 CO SI SEEDING 0.405 0.46 1SSC53A 27.9.1993 SI02+SIC 1530 AR SIC REDUCTION 0.407 0.4 100 Exp ID Date Powder Temp Atm Comment Initial Wt Final Wt 1BC53A 27.9.1993 BENTONITE+C 1530 AR 1.023 0.402 1FOC53A 27.9.1993 FELDSPAR+C 1530 AR 1.006 0.402 1I53A 27.9.1993 ILLITE 1530 AR 1.019 0.88 1KCK53A 28.1.1994 MKAO+COKE 1530 AR COKE PART SIZE 0.492 0.287 2KCK53A 28.1.1994 MKAO+COKE 1530 AR COKE PART SIZE 0.508 0.304 3KCK53A 28.1.1994 MKAO+COKE 1530 AR COKE PART SIZE 0.51 0.309 4KCK53A 28.1.1994 MKAO+COKE 1530 AR COKE PART SIZE 0.506 0.295 9SI C145 28.7.1992 SI02+1C 1450 AR STOICHIOMETRY 0.405 0.268 10SIC145 28.7.1992 SI02+3C 1450 AR 0.401 0.194 11SIC145 28.7.1992 SI02+3C 1450 AR 0.457 0.213 1SI15C 28.7.1993 SIUCON 1530 CO 0.406 0.41 1KAOC145 29.4.1992 KAO+COKE 1450 AR COKE PART SIZE 0.401 0.275 2KAOC145 29.4.1992 KAO+COKE 1450 AR COKE PART SIZE 0.4 0.28 3KAOC145 29.4.1992 KAO+COKE 1450 AR COKE PART SIZE 0.407 0.259 4KAOC14S 29.4.1992 KAO+COKE 1450 AR COKE PART SIZE 0.408 0.28 4SI C145 29.4.1992 a-SI02+3C 1450 AR 0.401 0.277 1SCS145 29.7.1992 SIC+2SI02+6C 1450 AR SIC REDUCTION 0.403 0.22 1FKA0145 29.7.1992 AL+2SI+6C 1450 AR STOICHMIXMKAO 0.401 0.263 12SIC145 29.7.1992 SI02+C 1450 AR/CO 0.406 0.256 13SIC145 29.7.1992 SI02+C 1450 AR 0.402 0.251 3SCA15C 3.8.1993 SI02+CAO 1530 CO CAO SEEDING 0.406 0.385 3SF15C 3.8.1993 SI02+FEO 1530 CO FEO SEEDING 0.408 0.401 1SG15A 30.8.1993 SI02+GRAPHITE 1530 AR GRAPHITE 0.4 0.209 25SC15A 30.8.1993 SI02+C 1530 AR 0.404 0.173 1KYC15A 31.8.1993 KY+C 1530 AR PARTICLE SIZE 0.405 0.282 2KYC15A 31.8.1993 KY+C 1530 AR PARTICLE SIZE 0.406 0.314 4KYC15A 31.8.1993 KY+C 1530 AR PARTICLE SIZE 0.403 0.285 1KF53C 4.1.1994 KAO+FEO 1530 CO FEO SEEDING 0.503 0.497 2KF53CA 6.1.1994 KAO+FEO 1530 CO/AR FEO SEEDING 0.502 0.494 1S1380C 7.1.1994 SILICON 1380 CO 0.408 0.628 1SF1380C 7.1.1994 SILICON+FEO 1380 CO 0.406 0.604 1MKA0145 8.6.1992 MKAO+C 1450 AR 0.403 0.247 1MKA0147 9.6.1992 MKAO+C 1470 AR 0.406 0.239 1MUL147 9.6.1992 MULL+C 1470 AR 0.42 0.335 1SI C147 9.6.1992 SI02+C 1470 AR 0.413 0.235 1KA0145 KAO+C 1450 AR 0.13 0.07 2SG15A SI02+GRAPHITE 1530 AR 0.405 0.189 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0078529/manifest

Comment

Related Items