UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Reverse flotation as a method of coal cleaning for preparation of coal-water slurries Pawlik, Marek 2002

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

Item Metadata


831-ubc_2002-732339.pdf [ 11.57MB ]
JSON: 831-1.0081120.json
JSON-LD: 831-1.0081120-ld.json
RDF/XML (Pretty): 831-1.0081120-rdf.xml
RDF/JSON: 831-1.0081120-rdf.json
Turtle: 831-1.0081120-turtle.txt
N-Triples: 831-1.0081120-rdf-ntriples.txt
Original Record: 831-1.0081120-source.json
Full Text

Full Text

REVERSE FLOTATION AS A METHOD OF C O A L CLEANING FOR PREPARATION OF COAL-WATER SLURRIES  by MAREK P AWLIK B . S c , Maria Curie-Sklodowska University, Lublin, Poland, 1994  A THESIS SUBMITTED IN P A R T I A L F U L F I L M E N T OF THE REQUIREMENTS FOR THE D E G R E E OF DOCTOR OF PHILOSOPHY in THE F A C U L T Y OF G R A D U A T E STUDIES (Department of Mining and Mineral Process Engineering)  We accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH C O L U M B I A April 2002  © Marek Pawlik, 2002  In  presenting  degree  this  at the  thesis  in  partial fulfilment  of  University of  British Columbia,  I agree  freely available for reference copying  of  department publication  this or  and study.  of this  his  or  her  representatives.  Department of  Mining Engineering  The University of British Columbia Vancouver, Canada  Date  DE-6 (2/88)  April  19,  2002  that the  may be It  thesis for financial gain shall not  permission.  requirements  I further agree  thesis for scholarly purposes by  the  is  an  advanced  Library shall make it  that permission for extensive granted by the understood  be  for  that  allowed without  head  of  my  copying  or  my written  ABSTRACT  Due to strongly attractive hydrophobic forces that operate between hydrophobic particles in water, aqueous suspensions of bituminous coals are characterized by high yield stresses and behave as non-Newtonian systems. Since coal must be cleaned before it can be used to prepare coal-water slurries (CWS), flotation is commonly applied to clean fine coal. However, coal flotation has never been optimized with regard to C W S and the use of flotation reagents in coal flotation renders the final clean coal product even more hydrophobic making CWS preparation even more difficult. The objectives of this thesis include the following: (i) to study reverse coal flotation as a cleaning method for the preparation of coal-water slurries, (ii) to study the applicability of amines in reverse coal flotation, and (iii) to elucidate the mode of action of various polymers used as either coal flotation depressants or CWS dispersants. In order to study the effect of coal surface properties on coal reverse flotation, a very hydrophobic bituminous coal, oxidized bituminous coal and sub-bituminous coal were utilized. As flotation and rheological experiments showed, anionic (polystyrene  sulfonate,  carboxymethyl cellulose, humic acids) and nonionic (dextrin, hydroxyethyl cellulose) polymers of low molecular weight were capable of acting as both flotation depressants and CWS dispersants for bituminous coals. Using humic acids as a model additive, this dual action of polymers was attributed primarily to increased coal hydrophilicity and to a higher coal surface charge in the presence of these additives.  ii  Contrary to literature reports, dodecyltrimethyl ammonium bromide (DTAB) was shown not to be able to depress the flotation of high rank coals below the critical micelle concentration, as determined through contact angle and flotation studies. It was concluded from adsorption and zeta potential measurements that D T A B molecules assume a flat orientation on a hydrophobic coal surface. D T A B behaved as a weak flotation collector for low rank/oxidized coals. It was shown that the quaternary amine did not increase water contact angles on these coals despite a very high amine adsorption density. This behavior was related to a highly chaotic orientation of the adsorbed amine molecules at the "gel-like" coal/water interface. The adsorption mechanism involved electrostatic interactions between the cationic surfactant and the negatively charged coal surface. Reverse flotation experiments on artificial coal/silica mixtures revealed that while humic acids were necessary in the reverse flotation of a hydrophobic bituminous coal, the reverse flotation of a sub-bituminous coal did not require any depressant. The separation of silica from coal was a strongly kinetic process taking place in a very narrow range of relatively high D T A B dosages. Contact angle and adsorption studies revealed that, despite a low adsorption density, D T A B could strongly increase the hydrophobicity of silica. This collecting action of D T A B in the flotation of silica was associated with the deposition of D T A B molecules at the air/silica interface upon collisions with air bubbles. The observed beneficial effect of a short conditioning time with D T A B on the selectivity of coal reverse flotation supports such a mechanism.  TABLE OF CONTENTS ABSTRACT  ii  T A B L E OF CONTENTS  iv  LIST OF T A B L E S  vii  LIST OF FIGURES  viii  ACKNOWLEDGEMENTS  xii  1  INTRODUCTION  1  2  R E S E A R C H OBJECTIVES  3  3  LITERATURE REVIEW  4  3.1  4  Mineral Suspensions 3.1.1 3.1.2 3.1.3  3.2  4  4 6 16 18 21 26 29  Fine Coal Cleaning and Utilization  36  3.2.1  36 36 40 42 49 56 59 63 66  3.2.2 3.2.3 3.2.4 3.2.5 3.3  Concentrated Suspensions Versus Dilute Ones Fine Particle Interactions and the D L V O Theory of Colloid Stability Rheology of Concentrated Suspensions High Deformation (Steady-State) Measurements Effect of Solid Surface Properties on Rheology Effect of Particle Size and Size Distribution on Rheology Effect of Adsorbed Surfactants and Polymers on Rheology  Coal Surface Properties Organic Matter of Coal Mineral Matter of Coal Coal Surface Wettability Electrokinetics and Electrical Charge at Coal/Water Interface... Traditional Fine Coal Processing Seam-to-Steam Strategy and Coal-Water Fuels (CWF) Effect of Coal Cleaning on Properties of CWF Coal Reverse Flotation  Chemistry and Mode of Action of Amines  72  3.3.1 3.3.2 3.3.3  72 75 87  Classification of Amines Chemistry of Amine Solutions Adsorption of Amines at Mineral/Water Interface  EXPERIMENTAL PROGRAM  96  iv  4.1  4.2  Reagents  96  4.1.1 4.1.2 4.1.3  96 96 97  Coal and Silica Samples 4.2.1 4.2.2 4.2.3  4.3  Dodecyl-Trimethyl Arnrnonium Bromide (DTAB) Humic Acids Polymers and Other Reagents  Silica Coals Coal Surface Characterization - FT-IR Study  98 98 101  Equipment and Methods  108  4.3.1 4.3.2 4.3.3 4.3.4 4.3.5 4.3.6 4.3.7  108 109 Ill 115 119 121 123 124 125 126 128 129 135  Proximate Analysis - Ash Content Determination Alkali Extraction Tests and Coal Oxidation B E T Specific Surface Area Determination Humic Acids (HA) Adsorption D T A B Adsorption Electrokinetic Studies Contact Angle Measurements Coal Surface Preparation 4.3.8 Batch Flotation Tests Flotation with D T A B and Polymeric Depressants Reverse Flotation 4.3.9 Rheological Tests 4.3.10 Surface Tension Measurements  5  98  RESULTS  136  5.1  Rheology of Coal-Water Slurries (CWS)  136  5.1.1 5.1.2 5.1.3  136 145 149  5.2  5.3  Effect of Humic Acids and Coal Surface Properties Effect of Low Molecular Weight Polymers Effect of Polymers on Natural Floatability of F4 Coal  Batch Flotation  151  5.2.1 5.2.2 5.2.3 5.2.4  151 152 153 154  Flotation of LS43 and F4 coals with M I B C Flotation of Coals and Silica with D T A B Effect ofpH on Flotation of F4 and F4-Oxidized Coals Flotation of Coal/Silica Mixtures with D T A B  Adsorption Studies  160  5.3.1  Adsorption of Humic Acids on Coals and Silica  160  5.3.2  Adsorption of D T A B on Coals and Silica  162  5.4  Zeta Potential Measurements  166  5.5  Wettability (Contact Angle) Studies  169  5.6  6  Coal Oxidation  170  5.6.1  170  Alkali Extraction Tests  DISCUSSION  173  6.1  Dispersing/Depressing Action of Polymers  173  6.2  Interaction of D T A B with Coals and Silica  176  6.2.1  Adsorption of D T A B and Its Effect on Silica Surface Properties Flotation of Silica with D T A B 6.2.2 Adsorption of D T A B on F4 Bituminous Coal 6.2.3 Action of D T A B in Flotation of Low Rank/Oxidized Coals Surface Composition of Coals and Coal Oxidation Adsorption of Humic Acids Adsorption of D T A B on LS43 and F4-Oxidized Coals 6.2.4 D T A B versus Humic Acids - Their Behavior at Different Interfaces...  178 181 185 188 188 189 191 199  Coal Reverse Flotation  201  6.3 7  CONCLUSIONS  204  8  R E C O M M E N D A T I O N S FOR F U T U R E W O R K  207  9  REFERENCES  209  10 APPENDICES  237  APPENDIX 1  237  APPENDIX 2  238  APPENDIX 3  239  APPENDIX 4  240  APPENDIX 5  241  APPENDIX 6  242  APPENDIX 7  243  APPENDIX 8  244  APPENDIX 9  245  APPENDIX 10  246  A P P E N D I X 11  247  A P P E N D LX 12  248  vi  LIST O F T A B L E S  Table l.p.38  Classification of coals by rank (according to A S T M )  Table 2. p.75  Ionization constants for selected primary amines (Cn^n+iNFfi)  Table 3. p.76  Solubilities of selected primary amines (Apian and Fuerstenau 1962).  Table 4. p.83  Critical micelle concentrations for selected primary and quaternary amines.  Table 5. p.86  Solubilites of alkyl-trimethyl ammonium chlorides in water (g of amine per 100 g of water).  Table 6. p.97  Molecular weight and electric charge of the tested polymers.  Table 7. p.99  Proximate analyses of the tested coals.  Table 8. p. 102  Band assignments for the infrared spectra of coals.  Table 9. p. 114  BET specific surface areas of the tested coal and silica samples.  Table 10. p. 153  Chemical composition of mineral matter from LS43 coal (percent of coal weight).  Table 11. p. 195  Flotation of LS43 coal in the presence of D T A B and M I B C .  Vll  LIST O F F I G U R E S  Figure 1. p.21  Various types of flow behaviors of solid-liquid suspensions.  Figure 2. p.52  Relationship between coal rank and p H of i.e.p. (Figure 2A, Laskowski 2001) and the generalized zeta potential-pH diagram for coals varying in rank (Figure 2B, Laskowski and Parfitt 1989).  Figure 3. p.65  A flowsheet of a CWF preparation plant at the Oristano harbor on Sardinia, Italy (Bozano et al., 2000).  Figure 4. p.73  Geometric representation of ammonia molecule.  Figure 5. p.74  Molecular structure of Dodecyl-Trimethyl Ammonium Bromide, [Ci H25N(CH )3] Br" (DTAB) with the characteristic tetrahedral arrangement of the three methyl groups around the nitrogen atom. +  2  Figure 6. p.78  3  Species distribution diagram for a total dodecylamine concentration of 10- mol/dm at 25°C. 4  3  Figure 7. p.79  Domain diagram for dodecylamine.  Figure 8. p.81  Typical experimental relationship between the solution surface tension and the concentration of a surfactant.  Figure 9. p.85  A general phase diagram for a surfactant near its Krafft point.  Figure 10. p.86  Experimental cmc vs pH data superimposed on the domain diagram.  Figure 11. p.89  Correlation diagram of contact angle, adsorption density, flotation response, and zeta potential for quartz as a function of dodecylammonium acetate concentration at pH 6-7, 20-25°C (Fuerstenau et al. 1964) Schematic adsorption isotherm of an ionic surfactant on a charged homogeneous surface.  Figure 12. p.92  Figure 13. p.94  Effect of quaternary amines (HTAB and D T A B ) on wettability of hydrophobic (graphite and octadecane) and hydrophilic, charged solids (mica, silica). Plot A is taken from Parachuri et al. (2001) while plot B is based on tabulated results of Elton (1957). Vertical, dashed lines indicate the critical micelle concentrations for the two surfactants.  viii  Figure 14. p.100  Particle size distributions of the tested coals (wet screening) and silica (Elzone apparatus)  Figure 15. p. 104  Infrared spectra of F4 and F4-oxidized coals. Band assignments were made based on data in Table 8.  Figure 16. p. 105  Infrared spectra of LS43 coal and humic acids (Na salt). Band assignments were made based on data in Table 8.  Figure 17. p. 116  Absorption spectra of a 150 mg/dm H A solution before and after adsorption on F4 and LS43 coals.  Figure 18. p. 117  Normalized absorption spectra of humic acids solutions before and after adsorption on F4 and LS43 coals.  Figure 19. p. 131  The elongated fixture for testing settling suspensions (original design by Klein 1992).  Figure 20. p. 136  Flow curves (ramping parts) for suspensions of F4 coal in water at various coal contents (weight % solids in suspension).  Figure 21. p. 137  Flow curves (ramping parts) for suspensions of F13 coal in water at various coal contents.  Figure 22. p.138  Flow curves (ramping parts) for suspensions of F4 coal in the presence of 0.8% of H A (per coal weight) at various coal contents.  Figure 23. p.139  Flow curves (ramping parts) for suspensions of F13 coal in the presence of 0.8% of H A (per coal weight) at various coal contents.  Figure 24. p.140  Full flow curves for suspensions of F4 coal in water and in the presence of 0.8% of H A (per coal weight) at the coal content of 55.2% (wt). Arrows indicate shearing history.  Figure 25. p. 141  Full flow curves for suspensions of F13 coal in water and in the presence of 0.8% of H A (per coal weight) at the coal content of 60.1% (wt).  Figure 26. p.142  Effect of humic acids (HA) on the yield stress and apparent viscosity of F4 and F13 coal-water suspensions at varying solids content. Results for two different, double-gap, measuring geometries are shown.  Figure 27. p. 144  Effect of humic acids on surface properties (contact angle and zeta potential) of F4 and F13 coals. Effect of H A on the zeta potential of LS43 coal particles is also shown.  3  ix  Figure 28. p. 146  Examples of flow curves obtained in the presence of carboxymethyl cellulose (MW = 80,000). F4 coal content: 55% (wt).  Figure 29. p. 147  Examples of flow curves obtained in the presence of polystyrene sulfonate (MW = 14,000). F4 coal content: 55% (wt).  Figure 30. p. 148  Examples of flow curves obtained in the presence of humic acids (low MW). F4 coal content: 55% (wt).  Figure 31. p. 150  Natural floatability of F4 coal in the presence of various polymers. Corresponding changes in the yield stress values of 55% (wt) F4 coalwater suspensions are also shown.  Figure 32. p. 151  Flotation of F4 and LS43 coals with the use of MIJ3C as a ffother.  Figure 33. p.152  Flotation of F4 coal, LS43 coal and silica with the use of D T A B . Flotation of F4-oxidized coal, and F4 coal in the presence of 400 g/t of humic acids is also shown.  Figure 34. p. 154  Effect of pH on flotation of F4 and F4-oxidized coals in the presence of 1500 g/t of D T A B .  Figure 35. p.156  Flotation kinetics of F4/Silica mixture (40% of ash in feed) at various dosages of D T A B , pH = 7.0-7.2.  Figure 36. p. 157  Flotation kinetics of F4(400 g/t of humic acids)/Silica mixture (40% of ash in feed) at varying dosages of D T A B , pH = 7.1-7.3.  Figure 37. p. 158  Flotation kinetics of LS43/Silica mixture (40% of ash in feed) at different dosages of D T A B , pH = 8.3-8.6.  Figure 38. p. 159  Yields and ash contents of concentrates after 2 min of flotation for the tested coal/silica mixtures at varying D T A B dosages.  Figure 39. p. 160  Adsorption kinetics of humic acids onto LS43 and F4 coals. Temperature: 23°C  Figure 40. p. 161  Adsorption isotherms of humic acids onto silica, LS43 and F4 coals, T = 23°C  Figure 41. p. 162  Adsorption kinetics of D T A B onto coals and silica. Initial D T A B concentration was 154 mg/dm (0.5 mmol/dm ), room temperature (23°C). 3  Figure 42. p. 163  3  Adsorption isotherms of D T A B onto coals, silica and at the air/water interface, T = 23°C.  x  Figure 43. p.165  Effect of pH on D T A B adsorption for F4 and F4-oxidized coals, T = 23°C.  Figure 44. p. 167  Zeta potential as a function of pH for the tested coals and silica. Numbers above the dashed line indicate estimated values of the i.e.p.'s.  Figure 45. p.168  Effect of D T A B on the zeta potentials of coals and silica.  Figure 46. p.169  Contact angles on coals and quartz plate in the presence of D T A B .  Figure 47. p. 171  Humic acids concentration and transmittance of alkaline extracts at different stages of F4 coal oxidation. Extraction of humic acids from LS43 and F4-oxidized coals at room temperature at varying pH.  Figure 48. p. 172  XI  ACKNOWLEDGEMENTS I am deeply indebted to Professor Janusz S. Laskowski, my research supervisor, for his expert opinion, guidance and assistance in the course of my work. I wish to extend my thanks to Mrs. Sally Finora for her technical assistance in the surface chemistry lab, and for keeping computers with my experimental data updated and virus-free. I also wish to thank our "coal lab" technicians: Mr. Pius Lo for knowing the function of every single piece of machinery in the C M P building; Mr. Frank Schmidiger for constructing so many small devices that made my work so much easier; Mr. Larry Wong for giving me free access to the analytical lab, and for carrying out the majority of ash determinations. Many thanks go to my soccer teammates from our grad team Panethnikos, and to my "Friday bridge" friends for allowing me to realize that free time really exists. Finally, I would like to express my gratitude and appreciation to my parents in Poland for their continuous support and encouragement, and to my dear wife, Joanna, whose smile, understanding and love made it all possible.  xii  1. I N T R O D U C T I O N  Coal's long term future as a leading energy source relies on the development of new utilization technologies. The increasingly rigorous environmental restrictions require new processes for coal's use as a fuel in power generation and for metallurgical industries. In the "seam-to-steam" strategy (Marnell et al. 1983), the product of fine coal cleaning is converted into a Coal-Water Fuel (CWF) at the coal preparation plant, and then the CWF is pipelined (or transported in any other way) to customers where it can be directly burned as a fuel oil. So far, only bituminous coals have been utilized in the form of Coal-Water Fuels on an industrial scale (Usui et al. 1997b, Hashimoto 1999). One of the reasons behind this is the fact that bituminous fines are easy to clean by means of traditional methods, primarily by froth flotation. However, since Coal-Water Fuels are supposed to replace fuel oils for power generation, it is understandable that C W F technology would primarily utilize thermal, difficult-to-float coals. At the present time, though, the thermal coal fines are often rejected without any beneficiation. Therefore, as will be discussed in this thesis, there seems to be a basic conflict between the objectives of coal cleaning by flotation and the objectives of coal utilization in the form of CWF. Also, traditional flotation reagents are not compatible with the additives required in CWF preparation. In this thesis, an alternative fine coal cleaning technique is studied in which the mineral components of coal are floated instead of the hydrophobic carbonaceous parts. The technique is referred to as coal reverse flotation (Eveson 1961, Stonestreet and Franzidis  1  1988, 1989, 1992) and requires a simultaneous depression of coal and an activation of mineral matter. It should be pointed out that coal reverse flotation is aimed at the recovery of mineral matter whose composition and properties are independent of coal rank and consequently some of the difficulties with cleaning low rank/oxidized coals can be entirely avoided. The attractiveness of coal reverse flotation lies not only in its potential superiority as a coal cleaning method for C W F preparation, but also in the fact that the technique can be used for other important applications. For example, the process can be employed in the recovery of coal from tailing ponds that normally contain about 30-40% of potentially recoverable but oxidized coal. Today's flotation techniques are widely applied to clean bituminous, metallurgical coals. Reverse flotation, i f successful, may solve the problem of cleaning fine low rank/oxidized coals, and at the same time such a process would better fit the seam-to-steam strategy.  2  2. R E S E A R C H O B J E C T I V E S 1) To study the applicability of coal reverse flotation as a fine coal cleaning method for the preparation of coal-water slurries. 2) To investigate the effect of coal surface properties on the rheology of coal-water slurries and on the reverse flotation process. 3) To examine quarternary amines (alkyl-trimethyl ammonium salts) as collectors for coal mineral matter in coal reverse flotation. 4) To study the mechanism of interaction of quarternary amines with the carbonaceous and mineral components of coal. 5) To elucidate the mechanism of action of various polymers as both coal depressants and C W F dispersants. 6) To investigate the conditions under which mineral matter selectively floats in reverse flotation mode.  3  3. L I T E R A T U R E R E V I E W 3.1  MINERAL SUSPENSIONS In modern mineral processing plants, mineral fines are exclusively beneficiated by  means of wet methods. These methods primarily include flotation and gravity separation. Wet grinding, filtration, thickening and tailings disposal all involve transport, handling and storage of mineral slurries. Solid-liquid dispersions also find applications in many other areas: paints, dyestuffs,  pigments,  paper  coatings,  cosmetics,  ceramics,  pharmaceuticals,  and  agrochemicals, just to mention the main ones. The control of the properties of these suspensions is critical in the stages involved in their preparation, storage, handling, and in their subsequent utilization. Some of the parameters that control the flow characteristics (rheology) of the solid/liquid systems are: particle size and size distribution, interparticle forces and the volume fraction of the solids.  3.1.1  Concentrated Suspensions Versus Dilute Ones  A distinction between the two types of systems can be made i f one considers the role/contribution of hydrodynamic and interparticle interactions as the concentration of the suspension increases (Tadros 1986, 1996). At one extreme, a suspension may be considered dilute i f the thermal (Brownian) motion of the solid particles predominates over the effect of hydrodynamic and interparticle forces. In such systems, the distance between the particle surfaces is large compared to the range of interaction forces, whether hydrodynamic or surface. In this case, the particles move independently as their translational motion is large and only occasional contact occurs between them. In other words, the particle interactions can  4  be represented by two-body collisions. Provided no settling occurs (the gravitational force can be neglected), the properties of the dilute systems are essentially time-independent. Timeaverage quantities, such as light scattering or viscosity, may be extrapolated to infinite dilutions to obtain fundamental properties of the system, e.g. particle or hydrodynamic radius. Such dilute systems are rarely encountered in industrial applications. As the solids content increases, the probability of particle-particle interaction increases as well. Gradually, a situation is reached where the interparticle distances become relatively small compared to the particle radius. In this case any particle in the system interacts with many neighbors and the interparticle interactions produce a highly ordered structure. At a further increase in solids content, spacing between the particles becomes very small and they can only undergo vibrations with amplitudes smaller than the particle size. The system will behave like a solid showing elastic response and no time-dependence of its properties. Such systems are often referred to as "solid suspensions" or "pastes". In between the two extremes, one may define concentrated suspensions. In this case, the volume fraction of the solids is sufficiently high for many-body collisions to occur. Both hydrodynamic and surface interactions play a role in determining the properties of the system. However, the interparticle distance is comparable to the particle size and, therefore, the particles can still slowly move/diffuse. Such concentrated systems show a time-dependence of their properties. In order to fully understand the macroscopic properties of concentrated suspensions, one must investigate interparticle forces. These strongly influence the rheological properties of suspensions and, therefore, rheological methods can be used to research such interactions.  5  3.1.2  Fine Particle Interactions and the DLVO Theory of Colloid Stability  In all industrial applications, solid-liquid dispersions show a wide solid particle size 9  3  *  distribution ranging from nanometers (10" m) to several millimeters (10" m) in diameter. Solid-liquid dispersions are commonly classified as colloidal systems or suspensions. Colloidal systems consist of the finest particles with particle sizes ranging from 1 nm to 1 pm (Tadros 1986, Everett 1988). The lower size limit is set by the size of the smallest molecular aggregate for which it is still meaningful to distinguish between a "surface" and "interior" molecule. At the upper limit, a significant proportion of the atoms or molecules are still surface molecules and have different thermodynamic properties from those in the interior. Such dispersions are also termed "colloidal suspensions". On the other hand, systems in which a significant fraction of particles covers a range greater than the colloidal range (coarser than 1 pm) are referred to as suspensions. The rheological behavior of solid-liquid dispersions is governed by a balance between three factors: Brownian diffusion, hydrodynamic interaction and interparticle forces (Tadros 1996). The interparticle forces include attractive (van der Waals and hydrophobic) and repulsive (electrostatic, steric and hydration) ones. Specific contributions of the three components generally depend on particle size and solids concentration. Brownian diffusion has little effect on coarse particles that contribute to suspension rheology only through hydrodynamic effects, while the interparticle forces primarily affect the rheology of very fine, colloidal dispersions. Since typical industrial slurries are both concentrated and contain some very coarse fractions (100-500 pm), only hydrodynamic and interparticle forces are of any significance in  6  these systems. Therefore, the relative contents of coarse and fine colloidal particles control the flow behavior of such suspensions. Four major types of these concentrated systems may be identified (Tadros 1996): "Hard sphere" suspensions in which both repulsion and attraction are weak, or screened, and the main forces responsible for flow are hydrodynamic ones. The rheology of such systems is governed only by particle size and solids volume fraction. Stable systems with soft (electrostatic) interactions where interparticle forces are dominated by electrical double layer repulsion. Particle size, solids content and electrolyte concentration (ionic strength) affect the rheological properties of these suspensions. At sufficiently high electrolyte concentrations, when the double layers are compressed, the systems behave similarly to hard sphere suspensions. -  Sterically stabilized suspensions. In these, particle repulsion results from interactions between adsorbed or grafted layers of surfactants or macromolecules. The adsorbed macromolecules keep particles apart at distances where the attractive forces are very weak. In the absence of any kind of electrostatic forces, the molecular weight and concentration of the protective macromolecules additionally influence the rheology of such dispersions. Flocculated and coagulated systems. These are found whenever the  attractive  interparticle forces dominate over of the repulsive ones. As a result of coagulation or flocculation, different types of structures may be formed that can extend throughout the suspension to create a flocculated network. As a first approximation, any solid-liquid suspension can be looked at as a concentrated colloidal system so the classic rules of colloid chemistry can be applied. In the D L V O theory of colloid stability, proposed independently by Derjaguin and Landau (1941) 7  and Verwey and Overbeek (1948), the net forces acting between particles are determined by the balance of attractive van der Waals interactions, V , and repulsive electrostatic forces, VR. A  The total interaction, Vr, is taken as a sum:  (1)  This fundamental law of surface chemistry deserves a more detailed discussion. For very short separation distances d between the particles, the attractive energy is calculated from:  AD  AR  24d  I2d  (2)  where A is the Hamaker constant, D and R are the particle diameter and radius, respectively. In his calculations, Hamaker (1937) showed that the van der Waals forces between two particles of the same material embedded in a liquid are always attractive. The second type of forces in Equation 1 results from the presence of electric charges around particles immersed in a liquid (water). Surfaces of such particles may become charged by a variety of mechanisms, the more important of which are the following (Kitchener 1969): 1)  Ionization of surface groups. If the solid surface contains acidic groups, increasing  the pH will give rise to a negatively charged surface. Conversely, a basic surface will become positively charged. In both cases the surface charge can be brought to zero by manipulating the pH of the solution (point of zero charge, or p.z.c. in short). Many metal oxides show  8  amphoteric behavior, i.e. both positively and negatively charged surfaces can be obtained by adjusting the pH. 2)  Differential dissolution of ions from the surface of a sparingly soluble mineral.  For example, when a silver iodide crystal is placed in water, dissolution takes places until the product of ionic concentrations equals the solubility product of the salt. In the case of A g l the solubility product is: [Ag ][f] = 10" (mol/dm ) . It turns out that silver ions preferentially +  16  3  2  dissolve in water leaving a negatively charged surface. The charge can again be reduced to zero by introducing a certain amount of silver ions. Since the solubility product must be satisfied, the excess of A g ions will be adsorbed by the crystal leading to the neutralization +  of the surface charge. The charge is equal to zero at [Ag ] = 10~ ' . +  3)  Isomorphous substitution. Some clays may exchange a structural ion with one of  lower valency, producing a negatively charged surface, e.g. A l 4)  5 5  3 +  may replace S i . 4+  Charged crystal surfaces. When a crystal is broken, surfaces with different  properties may be exposed. In the case of kaolinite, the crystal edges contain A l O H groups which give rise to a positive surface charge. This may coexist with negatively charged basal faces as a result of isomorphous substitution. In such a case, each type of surface will have its own p.z.c. value (e.g. clays). 5)  Specific ion adsorption. Multivalent ions as well as surfactants and polymers may be  specifically (chemically) adsorbed. The sign of the surface charge would depend on the charge of the adsorbing entity. Once a particle becomes electrostatically charged through any of the above mechanisms, an ion concentration profile develops around it as counter-ions from solution are attracted towards the charged surface. This concentration profile is called the electrical double layer (EDL) to reflect the fact that the counter-ions closest to the charged surface are strongly  9  adsorbed at the solid-solution interface. These ions form the first, so called, compact layer (also referred to as the adsorption or Stern layer). Farther away from the surface, the weaker attracted ions form the diffuse layer (Gouy-Chapman layer). The counter- or co-ion concentration profile along the diffuse layer follows the Boltzmann distribution function:  n - n exp(-w - / kT) t  i  (  (3)  where «, is the number of ions of type i per unit volume in the bulk solution far from the particle surface, k is the Boltzmann constant, and T is the temperature (K). The work done in getting an ion to a given point, w„ is assumed to be measured simply by the electrostatic energy the ion acquires: w,- = z,e!^ with z,- being the valency of the ion, e - the electronic charge, and IF- the electrostatic potential. The ions having the same charge as the surface will be repelled from the solid-solution interface, while the counter-ions will be attracted. The volume density of electric charge, p, at the interface is obtained by adding up all the ions of either sign in a unit volume of the electrolyte solution in the neighborhood of the point in question (Hunter 1993):  (4)  According to the electrostatic theory, the relation between the density of electric charge, the electrostatic potential and the distance from a flat charged surface is given by the Poisson-Boltzmann equation:  10  d *F 2  1  P  dx  2  £  (5)  w  where x is the distance from the surface, e , £Q are the permitivities of water and vacuum, w  respectively, while e is the dielectric constant (ejeo). r  The above second-order differential equation is usually solved using the, so-called, Debye-Hiickel approximation in which it is assumed that the electrical energy is small compared with the thermal energy (ze P< kT). This assumption of small potential facilitates x  integration after which a very useful solution is obtained in the following form:  f=y exp(-/a) o  (6)  where *FQ is the surface potential (the potential at distance x = 0), while the constant K, often referred to as the Debye-Hiickel parameter, is given by:  (7)  After putting the values of the constants into the above expression (at T = 25°C), the constant Kcan be calculated from a simplified formula:  (8)  11  "3  where c is the concentration of i-th ion in mol/dm . t  The term under the square root in Equation 8 is called the ionic strength and plays a very important role in colloid chemistry. The constant /f has a dimension of reciprocal length and hence the parameter \IK is termed the "thickness" of electrical double layer. The "thickness" in this case actually denotes the distance from the surface at which the value of the electrostatic potential declines e times (see Equation 6). It should be also pointed out that, according to Equations 6-8, the "thickness" of electrical double layer strongly depends on the electrolyte concentration (ionic strength). At higher electrolyte concentrations, the "thickness" of E D L decreases, a process often referred to as the compression of the double layer. Equation 6, which gives a relationship between the electrostatic potential and the distance from a charged surface, is fundamental for calculating the interaction energy due to electrostatic repulsion (VR) between two charged colloidal particles. For two identical spherical particles of radius R, separated by a distance d, suspended in a medium of dielectric constant £, and under the assumption of low and constant potential, the repulsive energy is usually given by two equivalent expressions:  V  R  =  2neR f  2 0  ln(l + exp(-/ctf))  (Hunter 1987)  (9)  (Derjaguin 1989)  (10)  and  V =-R R  f  2 0  ln(l + exp(-Kd))  12  Equations 9 and 10 differ by the factor An whose presence in Equation 9 results from "rationalization" of the electrostatic quantities to allow for the use of SI units. A detailed discussion on this modern convention is provided by Hunter (1987). According to Hunter (1987), Equation 9 is applicable only to situations when the electrical double layer is strongly compressed compared to the particle size (the product Kd is larger than 10). When the electrical double layer around the particles is extensive (Kd < 5) an approximate formula, originally proposed by Verwey and Overbeek (1948), can be used:  V = InsR R  Qxp(-Kd)  0!)  A combination of Equations 2 and 9 (or 11) is the essence of the D L V O theory in its original form:  V = V +V =T  A  R  — + 2nsR Yld  ln(l + exp(-/a/))  ( ) 12  When the van der Waals attractive forces dominate over the electrostatic repulsion, the colloidal particles form aggregates, a phenomenon known as coagulation. Under these conditions the dispersion is said to be unstable. Conversely, when the repulsive forces prevail, the particles are separated and well dispersed. Such dispersed systems are considered stable. Experimental evidence, however, suggests that not all systems can be described using the D L V O theory and that some other forces should also be included in the total force balance (Churaev and Derjaguin 1985, Derjaguin and Churaev 1989). The nature of these non-DLVO forces is again a function of solid surface properties. Experiments of Yoon et al. (1997)  13  strongly suggest that for highly hydrophobic surfaces characterized by large water contact angles (>80-90 deg), hydrophobic attractive forces contribute the most to the total force balance. Other researchers claim that hydrophobic forces arise as soon as the contact angle is larger than 64 deg (Pashley and Israelachvili 1981, Israelachvili and Pashley, 1984] or even 40 deg (Churaev 1995). The attractive, non-DLVO forces between such hydrophobic particles (methylated silica and coal) were shown to dominate over the electrostatic repulsion and cause coagulation even at pH values where the zeta potential was as high as -40 mV (Xu and Yoon, 1989,1990). Based on direct force measurements using hydrophobized (CTAB coated) mica sheets, Israelachvili and Pashley (1984) proposed an empirical equation describing the magnitude of hydrophobic interaction, Frf.  ^ - = -Cexp(-D/D ) R  (13)  0  where R is the curvature (radius) of the mica surface, C is a constant equal to 0.14 N/m, Do is the decay length of 1 nm, and D is the distance from the mica surface. Sometimes a double exponential function can be used when the hydrophobic forces are unusually strong (Yoon et al. 1997):  ( \ f D^ — = C, exp + C exp R v Ay V 2J D  (14)  2  D  This time, two different decay lengths D\ and D are introduced to characterize the forces 2  over larger separation distances. The constants Q are thought to be related to the interfacial tension at the solid-liquid interface. 14  Claesson et al. (1986) used a simple power law to quantify the hydrophobic interactions:  F„ ^T = R  K ()  2  15  D  where K is the hydrophobic force constant of the order of 10" to 10" J. 10  20  The significance of this relationship lies in the fact that it has the same form as the expression for van der Waals forces (Equation 2). As a result, the hydrophobic forces can be represented by a single parameter K, analogously to the Hamaker constant A for van der Waals forces. Consequently, Yoon et al. (1997) found an empirical relationship between the contact angle on silica plates (hydrophobized to a different degree) and the hydrophobic force constant K determined from direct force measurements (using atomic force microscopy):  logK = acos(0) + b  (16)  where a and b are constants. Yoon's relationship implies that once the contact angle on a given solid surface is measured, the magnitude of the hydrophobic force acting between two such surfaces can also be determined. In the case of strongly hydrophilic solids characterized by contact angles lower than 20 degrees, hydration repulsive forces start to play the dominant role (Churaev 1995) which may lead to surprisingly stable dispersions, even at the p.z.c, when the system should aggregate. Such an unusual behavior due to the presence of hydration forces was confirmed experimentally by Pashley (1981) for mica suspensions and by Yotsumoto and Yoon in their 15  stability studies of rutile (TiC^) and silica (SiCh) dispersions (Yotsumoto and Yoon 1993a, 1993b). These hydration forces appear when large amounts of hydrated ions (mainly cations) and water molecules adsorb at the solid-water interface. The hydration sheaths, consisting essentially of oriented water dipoles, act as protective layers preventing colloidal particles from coagulation. The magnitude of the hydration forces was experimentally found to follow a double exponential function similar in form to Equation 14 for the hydrophobic forces (Yotsumoto and Yoon, 1993a):  Vhydr. =^[C D,exp(-D/D ) ]  ]  + C D x (-D/D )] 2  2Q  V  2  (17)  where C\ and Ci are constants and D\ and D are decay lengths. 2  The hydrophobic and hydration forces are often called the structural forces as their presence results from a preferential orientation of water molecules at the water-solid interface; hydrophobic surfaces tend to repel the water dipoles, while the hydrophilic, often charged, solids strongly attract them. The discovery of the non-DLVO forces led to the reformulation of the D L V O theory. The terms "extended" or "modified D L V O theory" are frequently used to denote the inclusion of the structural forces in the total force balance acting between colloidal particles.  3.1.3  Rheology of Concentrated Suspensions  When a material of any type is subjected to a stress it will be to some degree deformed. The relation between the applied stress and the resulting deformation (strain) is  16  studied by rheology. The stress can be applied in various ways: as a tension, as a compression, or as a shearing force, or as some combination of the three. Once the applied stress is removed three types of behavior may be observed (Hunter 1993): 1) the material returns to its original shape/position (purely elastic response), 2) the material remains in its new position - flow has occurred (purely viscous response), 3) some partial recovery takes place, which is an intermediate type of response between the first two (viscoelastic response). These three behavior patterns are characteristic for solids, liquids and plastics, respectively. However, most materials can exhibit any or all of them depending on the time scale involved in the application of the stress and the measurement of the resulting strain. Two main types of experiments may be carried out depending on the time scale (Tadros 1996). In the first type, referred to as transient measurements, a constant stress or strain is applied and the relaxation of stress (or strain) is followed as a function of time. In the second type, referred to as dynamic measurements, a stress (or strain) is applied within a well-defined frequency regime (usually sinusoidal) and the resulting strain (or stress) is compared with the applied stress (or strain). These two types of experiments are also called low deformation tests as they can be made before the "structure" of the suspension is broken down. However, Theological measurements can also be carried out under conditions where the "structure" of the suspension is deformed or permanently broken during the measurement. This is the case when a suspension is subjected to continuous shear, while the stress in the sample is measured. Such measurements are referred to as steady-state (high deformation) measurements.  17 High Deformation (Steady-State) Measurements  When a suspension is placed between two moving parallel surfaces, a certain strain rate is applied to it and the resulting stress can be measured as the torque exerted by the suspension on the moving (rotating) surface. This type of motion is called simple shear, the strain rate is referred to as the shear rate, while the stress is termed the shear stress. In this type of experiment the suspension is deformed irreversibly (is forced to flow) and a basic relationship between the shear stress and the shear rate is obtained in the form of the flow curve:  r=  f(dyldt)  (18)  where dyl dt is the shear (strain) rate often denoted by D. In order to make use of the flow data, it is important to describe the flow curves in a mathematical manner. Several flow curve models have been developed to quantify the variety of responses observed for typical industrial slurries. The simplest flow behavior obeys Newton's law that assumes the following form:  T  = rjD  (19)  where 77 is the viscosity, that is constant at a given temperature and pressure. This response is often called the ideal fluid-like behavior as the equation was originally derived by Newton for ideal viscous fluids. The model implies that the liquid (or suspension) flows under any applied stress. 18  Any suspension giving a linear plot between the shear stress and the shear rate running through the origin of the shear stress/shear rate coordinate system is termed Newtonian. Any departure from the above is an indication of a non-Newtonian flow behavior and the majority of the models describing non-Newtonian systems originate from empirical observations. Some systems behave like solids until a certain critical stress is applied to make the material flow. Once this yield stress is exceeded the material flows with a constant differential viscosity, rjpi. Such suspensions are often described in terms of the Bingham plastic model:  = T  B  +  T]  PL  •D  (20)  where T is the Bingham yield stress and T]PL is the plastic viscosity. The Bingham model is B  often used to estimate the yield stress. As pointed out by Nguyen (1983) and Nguyen and Boger (1983), the lack of data in the low shear rate range may lead to a significant overestimation of the yield point. In an extreme case, the model may predict a yield stress where none actually exists. It is more usual to observe a non-linear relation between the shear stress and shear rate for stresses higher than the yield value. The simplest model used to fit such systems is the Hershel-Bulkley (H-B) model:  T =  Tr,+K-D,n  (21)  19  where To is referred to as the primary yield value and K and n coefficients are called the fluid consistency index and the flow behavior index, respectively (Nguyen 1983). When n - 1, the model assumes the form of the Bingham plastic equation. A simple derivative of the Hershel-Bulkley model is often applied to pseudoplastic materials with negligible yield values (To = 0). The modified H-B equation is termed the Ostwald-De Waele model, or the power law model:  T  = K-D',n  (22)  The model parameters are the same as in the H-B model. When n = \ the model becomes the Newton's model and any deviation of n from unity is a measure of non-Newtonian character of the system. Because of that similarity, a high K value indicates a liquid with a high viscosity. In 1959 Casson suggested a two-parameter equation based on some physical arguments for aggregating systems. In the model, it is proposed that particles may form aggregates whose shape and dimensions control the viscosity. Also, under flow, disruptive stresses may break up the aggregates so that for a given shear rate there is a mean aggregate size. This limiting size would also depend on the interparticle interactions. Casson's model has the following form (Casson 1959):  (23)  20  where r CY is the Casson yield stress and 7] c is the limiting viscosity (slope) at high shear rates. This model is also frequently used to estimate the yield stress. It also better than the Bingham equation describes the flow behavior of many materials at very low shear rates. The above models are most frequently used to characterize mineral suspensions and coal-water slurries in particular. Figure 1 shows examples of flow curves encountered in rheological measurements on suspensions.  Shear Rate Figure 1. Various types offlow behaviors of solid-liquid suspensions. Effect of Solid Surface Properties on Rheology  As the modified D L V O theory suggests, the dispersion/aggregation equilibria in suspensions depend on the balance between the attractive and repulsive forces. Whenever the attractive forces dominate over the repulsive ones (regardless of the nature of either) the solid particles form aggregates. In concentrated solid-liquid suspensions this aggregation brings about the formation of spatial structures within the liquid phase. This structuring exhibits a  21  certain mechanical strength that manifests itself by the presence of a yield stress (Rehbinder 1964, 1965). This fundamental observation immediately suggests that a non-Newtonian behavior of suspensions is a result of strong, attractive forces acting between the particles. Because of that, the yield stress, as a macroscopic evidence of particle aggregation, is often used as a convenient measure of the interparticle interactions. In a pioneering work on the relationship between the solid surface properties and the rheological response of suspensions, Firth, Hunter and Frayne (Firth and Hunter 1976a, 1976b, Firth 1976, Hunter and Frayne 1979, Hunter 1982) developed the elastic floe model to describe the rheological behavior of coagulated colloidal dispersions. In this model, it is assumed that the interparticle bonds within the aggregates are strong enough to withstand the shearing and hydrodynamic forces. Obviously, the "floe" resistance must be limited but the bonds are allowed to stretch elastically up to a breaking point under the action of these external forces. Using the D L V O theory and the Bingham plastic model, Firth and Hunter (1976a) showed that for a coagulated system the Bingham yield stress, T B, is determined by the van der Waals attractive force:  AR lid  2n R^  (  2  30 A 2  *B  (24)  The term on the left-hand side is the van der Waals force (Equation 2), <j> is the volume fraction of solids. Next, Firth (1976) made use of the fact that the magnitude of electrostatic repulsion changes with the square of the surface potential (Equation 9). Assuming the zeta potential being equal to the surface potential (valid for small surface potentials and low ionic  22  strengths), Firth measured the zeta (surface) potential, £ for Ti02 suspensions and plotted the corresponding Bingham yield stresses as a function of the square of £ He found the following linear relationship for different Ti02 suspensions varying in solid content:  T = T (£ = 0) + a£  (25)  2  B  where  TB(£=  B  0) is the intercept of the graph where the zeta potential is equal to zero, a is a  constant with a negative value ranging from -4.3 • 10" to -3.4 • 10" Pa/mV . 3  2  2  This simple expression indicates that as the surface charge increases the yield stress declines very quickly. Since the magnitude of electrostatic repulsion is proportional to the surface charge, it can be concluded that for an electrostatically stabilized (dispersed) system characterized by a high zeta potential, the yield stress is negligible. This conclusion, in turn, suggests that dispersed suspensions should exhibit a behavior close to the Newton's law of viscosity. As already mentioned, the surface charge on many solid particles may be altered by manipulating the pH. Metal oxide suspensions (silica, alumina, titania and zirconia) are such systems and, so far, the effect of interparticle forces on the rheological behavior of solid/liquid suspensions has been investigated mostly for these simple, well-characterized materials (Scott 1982, Evanko et al. 1997, Leong et al. 1991a,b, 1993, 1995, Scales et al. 1999, Johnson et al. 2000, Zhou et al. 2001). When oxide particles are dispersed in a solution at a pH far from the isoelectric point (the pH at which the zeta potential is zero) the oxide surface becomes electrostatically charged; positively below the i.e.p. and negatively above it. Such charged particles will repel one another and will be dispersed. On the other hand, when  23  the p H of solution is set exactly at the i.e.p. the net surface charge will be zero and electrostatic repulsion disappears. Under these conditions the ever-present, attractive, van der Waals forces, as the only forces in the system, will result in particle aggregation. When the yield stress is measured as a function of pH, and hence electrostatic repulsion, the relationship exhibits a clear maximum value of the yield stress at the p H of the isoelectric point. The farther the system is from the i.e.p. the stronger the repulsion and the lower the yield stress will be. Recently, Klein (2001) has demonstrated similar results for nickel laterites which are a mixture of various clays and oxides with goethite being the major constituent. Kapur et al. (1997) proposed a new approach to understanding the yield stress of concentrated particulate suspensions. The model defined the inter-dependence of the yield stress of a flocculated particulate dispersion not only on particle surface properties but also on particle size, size polydispersity and particle concentration. While the model of Kapur et al. correctly predicted the yield stress of an aggregated suspension at the i.e.p., further modification by Scales et al. (1998) added the role of electrostatic repulsive forces to the model for p H values far from the isoelectric point. Although the final model of Scales et al. is quite complex, it also incorporates the ^  2  term, similarly to Firth's expression, and thus  explains the parabolic form of the yield stress data as a function of pH. Muster and Prestidge (1995), Prestidge (1997) and Huynh et al. (1999) studied the rheological behavior of sulfide mineral slurries (sphalerite and galena). They concluded that oxidized sulfides behave in suspension as corresponding oxides with the maximum value of the yield stress appearing around the i.e.p.. Fresh galena suspensions, however, gave nonNewtonian responses, with high yield stresses, that were attributed to the presence of hydrophobic attractive forces acting between naturally hydrophobic galena particles. Using 24  Firth's expression (Equation 25), Prestidge (1997) was able to estimate the non-DLVO stress contribution to the overall yield value. In this way, he showed that the hydrophobic forces overcame the electrostatic repulsive forces even at pH = 6, far from the i.e.p. which is approximately 2.2 for galena. Woskoboenko et al. (1987, 1989) tested the rheology of brown coal slurries as a function of pH. They also found that the Bingham yield stress reached a maximum around the i.e.p. of the coal (~2.5) regardless of the coal content in the slurry. This behavior was also related to the attraction/repulsion balance between the coal particles. Essentially the same conclusions were drawn by Leong and Boger (1990) in their rheological investigation of brown coal suspensions. Using two coals with different surface charge densities, they concluded that the yield stress appears whenever the surface charge density is low or the ionic strength is high. The effect of the ionic strength may be explained in terms of the compression of the electrical double layer and the resulting partial coagulation of the coal particles. Pawlik and Laskowski (1998) showed that for bituminous, highly hydrophobic coal suspensions the yield stress is practically independent of pH (from 3 to 10) suggesting that the hydrophobic forces fully control the response of the system. After coal oxidation, however, the yield stress not only decreased but also showed a dependence on pH. This change in response suggests that for the more hydrophilic material the hydrophobic forces disappear allowing electrostatic repulsion to markedly contribute to the total interaction. Nguyen and Boger (1998) in their paper discussing the Australian practices in solving waste disposal problems in the bauxite and coal industries showed that the characterization and understanding of the rheology of tailings slurries, and the underlying surface chemistry, are fundamental in selection of a proper disposal strategy. 25 Effect of Particle Size and Size Distribution on Rheology  In 1905, Einstein derived a simple formula relating the viscosity of suspension to the solids volume fraction:  (26)  ^ - = 1 + 2.5^ ?7o  where rj and rjo are the viscosities of the suspension and the suspending medium, respectively (the ratio of the two values is often called the relative viscosity), and (p is the solids volume fraction. In Einstein's derivation, it was assumed that particles in suspension were sufficiently far apart to make any interparticle interaction impossible. As discussed in previous chapters, this assumption is consistent with the definition of dilute suspensions. Therefore, it was quickly observed that in order to describe more concentrated systems, higher order polynomial terms should be added to the equation as the relative viscosity increases exponentially with the solids concentration. Several empirical or semi-empirical equations were proposed to characterize the viscosity-solids content relationship for concentrated monodisperse suspensions. The most frequently used expressions are those of Mooney (1951), Krieger and Dougherty (1959) and Chong et al. (1971). These equations are as follows:  Mooney:  (27)  In V <t>mj  Krieger/Dougherty:  JL  1  <P_  (28)  V <l>mj  26  Chongetal.:  — no  (29)  1+0.75!  mj  In the above expressions [77] is the "intrinsic viscosity" and <j)  m  is the maximum packing  fraction. The maximum packing fraction of spherical particles can be calculated for several different geometrical arrangements (Lee 1970). For cubical packing, <f> orthorhombic - <j>  m  = 0.6045, tetragonal - <f>  tetrahedral (hexagonal, close packed) - <p  m  m  m  = 0.5236, for  = 0.6980, pyramidal - <f>  m  = 0.7405 and  = 0.7405. The last two are the closest possible  packings. Krieger and Dougherty (1959) used a range of values for fitting different sets of data from as low as 0.541 up to 0.782, while Chong et al. (1971) estimated </) to be 0.605. m  The above equations also suggest that increasing the maximum packing fraction can significantly reduce the viscosity of the suspension. Farris (1968) theoretically analyzed the rheological response of monodisperse, bimodal and trimodal suspensions of non-interacting particles. Using viscosity data for monodisperse systems, Farris's model predicted the existence of a minimum viscosity for polymodal dispersions at certain blend ratios, and the theoretical results agreed very well with some experimental data from literature. Chong et al. (1971) studied the effect of bidispersity on the viscosity of glass beads suspensions in polyisobutylene with solids volume fractions ranging from 0.54 to 0.74 and the particle size ratios varying between 0.477 and 0.138. Keeping the fines content at 25% of the total solids volume fraction, Chong et al. found that for a given solids concentration, the viscosity decreased markedly as the particle size ratio decreased. In these bimodal systems, the smaller beads could fill up the voids between the larger beads thus increasing the  27  maximum packing. More importantly, Chong et al. also calculated the relative viscosity as a function of the fraction of fines at a given total solids content. The obtained curves showed minima at about 25-30% of fines regardless of the total solids concentration. This bimodality effect was also experimentally found by many other researchers (Rodriguez et al. (1992), Hoffman (1992), Poslinski et al. (1988), Ferrini et al. (1984), Nguyen et al. (1997) and Parkinson et al. (1970)). Parkinson et al. (1970) studied the behavior of trimodal and tetramodal suspensions. Their results indicate that the relative viscosity of such systems is a product of the "partial" relative viscosities of the suspensions of the contributing fractions. For example, for a tetramodal system with equal proportions of the contributing fractions, at a total solids content,  <J>T,  the relative viscosity is given by:  Vrel(<Pr) = Vrell (<Pr ' 4) " Vre,2 i<t>T I 4) " Vrel3^T  ' 4) " VrelA^T  I ) 4  (30)  which is in excellent agreement with the theoretical model of Farris (1968) for multimodal suspensions. Parkinson et al. (1970) also demonstrated that suspensions of monodisperse fine particles have higher viscosities than the dispersions of coarser particles at the same solids content. Several models were proposed to account for the effect of bimodal, or polymodal distributions (Sudduth 1993a,b,c, Probstein and Sengun 1987, Sengun and Probstein 1989a,b, Probstein et al. 1994). In the approach of Sengun and Probstein it is proposed that the fine colloidal fractions primarily affect slurry rheology through their surface properties and interparticle interactions, giving the slurries either Newtonian or non-Newtonian character. The coarse fraction contributes only by increasing the solids volume fraction in the slurry. 28  Based on further theoretical analysis of the experimental data, Probstein et al. (1994) estimated the boundary particle size at 1.5 pm and concluded that the fines in suspension behave as a continuous medium for the coarser fractions. Packing densities for continuous particle size distributions were theoretically analyzed by Zhang et al. (1984) and Wang et al. (1993) in their studies on the rheology of coal-water slurries. Particle size distributions of fine coal are often described using the Rosin-RammlerBennett (RRB) distribution function:  F(d) = 100 1-exp V  ^63.2  J  (31)  where F(d) is the cumulative percent passing size d, d^.2 and m are the size modulus and distribution modulus, respectively. Assuming the top size of 300 pm and allowing the minimum size to vary from 1 to 20 pm, Zhang et al. (1984) and Wang et al. (1993) showed that the range of acceptable m values, for achieving highest possible packing densities at a given coal content, widens with decreasing size of the smallest particle. Effect of Adsorbed Surfactants and Polymers on Rheology  According to the modified D L V O theory, whenever attractive interparticle forces dominate over repulsive ones colloidal particles coagulate. When repulsion prevails the particles remain well dispersed. As already discussed, both coagulation and dispersion have a profound effect on the rheology of solid-liquid suspensions. The modified D L V O theory also suggests that aggregation/coagulation may be prevented by introducing a mechanical barrier which would keep the particles at such distances where the attractive forces become 29  insignificant. Such barriers may be provided by layers of adsorbed polymers or surfactants and this mechanism of dispersion is referred to as steric stabilization (Napper 1977, 1981, Sato and Ruch 1980, Laskowski 1988). The steric effect of adsorbed macromolecules is often divided into two contributions: the mixing (osmotic) interaction and the volume (elastic) restriction (Tadros 1996). The free energies of the two components are given by the following equations:  c  =c steric  (32)  +c mix.  elastic  2kTV}  Mixing:  Elastic  (1 X  ^mix  (33)  K i x W  (34)  G i =2kTv R (h) e astic  2  el  In the above equations k is the Boltzmann constant, T is the temperature, V\ and V are the 2  molar volumes of the macromolecule and solvent, respectively, v is the number of chains per 2  unit area, R ix(h) and R i(h) are geometric functions related to the segment density distribution m  e  of the macromolecule normal to the surface. The % parameter is the macromolecule-solvent interaction parameter usually referred to as the Flory-Huggins parameter (Flory 1953). The elastic interactions result from the loss in configurational freedom of the adsorbed molecules on the approach of a second particle as the adsorbed chains become compressed. The mixing interactions, on the other hand, arise from the interpenetration of the adsorbed chains that, in turn, results in an increase of the chain segment density in the interaction zone. This also leads to a local build-up in the osmotic pressure (Tadros 1982).  30  For an effective steric stabilization,  G ic ster  must be positive. In other words, some  external work must be done to force the polymer-coated particles to aggregate. It is obvious from the above equations that the elastic interaction,  G iastic, e  is always positive (repulsive),  while the sign of the mixing component depends only on the value of the Flory-Huggins (FH) parameter. According to Napper (1981), when ^ i s smaller than 0.5 the polymer is in good solvent conditions and the macromolecules prefer to interact with the solvent rather than with one another. In contrast, when % is larger than 0.5, the polymer chains are in poor solvent conditions and mixing of the chains is now favorable. The condition when %  =  0-5 is called  the theta condition and determines the onset of incipient flocculation. Therefore, it is evident that  G ic ster  will be positive (repulsive) only when the adsorbed  polymer/surfactant is in a good solvent environment, and the solid particles with the adsorbed polymer would thus be sterically stabilized. Some exceptions were reported, though flocculation was observed in worse than "theta" solvents and the phenomenon, termed the "enhanced steric stabilization" (Smitham and Napper 1979), appears to be associated with stabilizing chains that very strongly interact with the surface of the particles. For a sterically stabilized system, according to Kitchener (1972), the repulsion can be considered to arise from a swelling pressure at the point of a sufficiently close approach. Hence, the steric "force" component ( G  iteric  ) is often included in the total force balance to  complete the modified D L V O theory. Water-soluble polymers and surfactants always carry some functional groups in addition to a main hydrocarbon (hydrophobic) skeleton/chain. If the adsorbed polymer molecules are electrostatically charged, repulsion between similarly charged particles also contributes to the total repulsion energy. Leong et al. (1993, 1995) found that the presence of  31  surfactants or polymers (low molecular weight polyacrylic acid) caused a shift of the i.e.p. of T1O2 particles towards more acidic or alkaline values depending on the macromolecule nature and the corresponding yield stress/pH curves still exhibited maxima at the "shifted" i.e.p.'s. Interestingly, when the concentration of the polymer/surfactant increased, the maximum value of the yield stress, even at the i.e.p., decreased suggesting that in the presence of macromolecules the steric forces start to play a significant role in the stabilization mechanism. Sterically stabilized dispersions are thermodynamically (permanently) stable while electrostatic repulsive forces alone lead to kinetic (temporary) stability. A combination of these two mechanisms would be most effective. Studies of Taylor et al. (1991) and Trochet-Mignard et al. (1995) in a series of experiments on the rheology of concentrated coal-water slurries suggest that surfactants adsorbed on a coal surface stabilize the suspension through steric forces allowing a higher solids loading. The extent of this effect was shown to be a function of the surfactant concentration. At lower concentrations, nonionic surfactants  caused the hydrophobic  aggregation of the coal particles leading to an increase in the slurry viscosity. At higher concentrations of the surfactants the slurries were clearly redispersed and the viscosity decreased. Ionic surfactants, on the other hand, gave a consistent decrease in the viscosity down to a value that coincided with the maximum adsorption density. According to Taylor et al. (1991) and Trochet-Mignard et al. (1995) this observation suggests stabilization through electrostatic repulsion and, to some extent, steric stabilization. On the other hand, since nonionic surfactants do not have ionizable groups, their stabilizing behavior can be attributed solely to the steric interactions. The mechanism involved in the action of nonionic dispersants can be easily understood based on the work of Clunie and Ingram (1983).  32  On a hydrophobic surface (such as a coal surface), polymers/surfactants having both hydrophobic and hydrophilic groups adsorb through hydrophobic interactions exposing their hydrophilic groups outwards into the water phase and rendering the surface hydrophilic. At this point, further adsorption ceases because, once the surface is saturated with the adsorbate, similar hydrophilic groups do not interact strongly. In contrast, i f adsorption occurs through hydrophilic electrostatically charged groups the hydrophobic parts of the adsorbate molecules are now exposed making the surface hydrophobic. This apparent hydrophobicity may lead to particle aggregation (or flocculation if the surfactant molecule is large enough) and to increased viscosity as observed at lower concentrations of nonionic surfactants. However, i f the concentration of macromolecules is high enough, a second layer of molecules may adsorb through hydrophobic interactions with such a hydrophobized surface leading again to the reversal of surface properties. Multilayer adsorption of such heterogeneous polymers results in strongly hydrophilic, sterically and electrostatically stabilized systems and this phenomenon seems to be involved in the redispersion of coal-water slurries at higher surfactant dosages. It should also be pointed out that when a hydrophobic particle is rendered hydrophilic through surfactant/polymer adsorption, the hydrophobic attractive forces cease to operate thus increasing the stability of such hydrophilic dispersions as predicted by the modified D L V O theory. Polymers may be conveniently classified based on their molecular weight. Three types can be distinguished (Hogg 1999): flocculants with M W ranging from several to «20 million, coagulants with M W between 50,000 and 1,000,000, and dispersants with M W lower than 50,000. Although the molecular weight limits are arbitrary, it should be pointed out that in good solvents only low molecular weight polymers can be used as dispersants. High  33  molecular weight polymers may act as dispersants only at high concentrations (steric stabilization). Since a fully stretched flocculant molecule can be several microns long (Kitchener 1972), the bridging mechanism of flocculation can be easily imagined in better than "theta"solvents, and at low polymer concentrations. Flocculation destabilizes fine particles towards aggregation (sometimes the term "polymer-mediated aggregation" is used instead) and hence facilitates the formation of a mechanical network within the suspension reinforced additionally by the elastic macromolecules. The addition of small quantities of flocculants (high molecular weight polymers) to concentrated suspensions produces highly nonNewtonian systems with non-zero yield stresses and increased viscosities (Killmann and Eisenlauer 1982, Friend and Kitchener 1973, Saeki et al. 1994, 1999). The importance of polymeric additives was also recognized in wet grinding circuits that deal with concentrated coal/mineral pulps. Klimpel (1982, 1983) showed that optimum grinding performance can be achieved by manipulating the rheological properties of the feed slurry. This could be done through controlling solids particle size distribution or by introducing a grinding aid, i.e. a dispersant. In the case of coal grinding, the presence of the yield stress dramatically lowered the grinding rate since some energy was wasted for overcoming the resistance generated by the yield stress and high viscosity of the pulp. After adding a dispersant, the yield stress fell off almost to zero and the rate of grinding was significantly improved. Similar results  on the beneficial effect  of low molecular weight polymers  (polyacrylates) on the efficiency of grinding were reported by Fuerstenau et al. (1985, 1990), and Velamakanni and Fuerstenau (1987, 1993a,b) in their studies on the role of rheology in the wet grinding of mineral (hematite, dolomite and silica) suspensions. 34  The adsorption behavior of long chain polymers (flocculants) is quite complex and not well understood. There are many theories of polymer adsorption, as reviewed by Fleer and Scheutjens (1993) but, generally speaking, it is accepted that a polymer molecule adsorbs simultaneously onto many sites on a solid surface assuming conformations referred to as trains, loops and tails. The actual attachment to the solid surface can take place through many different interactions similarly to surfactant adsorption (Gregory 1987): ionic (electrostatic) interaction, when an adsorbate adsorbs on a surface bearing oppositely charged ionic groups, e.g. amines on silica, hydrophobic bonding, responsible for the adsorption of nonpolar segments of polymers and surfactants on hydrophobic surfaces, e.g. humic acids on hydrophobic coals, hydrogen bonding takes place when the surface and the polymer have suitable H-bonding sites, e.g. hydroxyl groups of polysaccharides and oxidized coal, ion binding, sometimes it is found that a certain amount of divalent metal ions (such as Ca  2+  or M g ) is required to promote the adsorption of anionic collectors onto negatively 2+  charged surfaces. These metal ions are known to bind strongly to carboxylate groups and serve as links between these groups and negative sites on the surface. This mechanism is involved in the adsorption of fatty acids on apatite. dipole-crystal-field effects, this mechanism is common in the adsorption of polymers onto crystal surfaces, e.g. flocculation of aqueous minerals (apatite, fluorite).  35  3.2  FINE COAL CLEANING AND UTILIZATION The purpose of coal cleaning, which removes inorganic impurities from coal, is to  improve coal quality to make it suitable for specific applications. Because of decreasing quality of raw coal, the need for fine coal cleaning has been increasing. Among the factors contributing to the need for more advanced coal processing are (Hower and Parekh 1991): (a) increased demand for quality brought on by market and environmental requirements, (b) increased extraneous dilution caused by mine, health, and safety laws, and (c) depletion of the higher-quality coal seams.  3.2.1  Coal Surface Properties  Coal is the general descriptive term applied to a group of solid fossil fuels, black or brown in color, that consist predominantly of altered plant material and usually occur as seams within other consolidated strata (Osborne 1988). From a geological point of view, coal may be classified as a sedimentary rock (it is not a mineral) consisting of organic components and with only a minor proportion of mineral constituents. Organic Matter of Coal  The metamorphic development of coal from plant material to a rock, also referred to as coalification, is synonymous in chemical terms with progressive enrichment of the coal substance in organically bound carbon. The term "coal rank" is often used to describe the degree of chemical and structural alteration of the organic matrix. As shown by Whitehurst et  36  al. (1980), the progress of coalification is accompanied by a steady increase in the aromatic hydrocarbons content. Simultaneously, the content of oxygen functional groups rapidly declines (Ihnatowicz 1952, Blom et al. 1957). Ihnatowicz and Blom et al. also showed that oxygen can occur in coals in a variety of functional groups. The main ones include O H phenolic, etheric, with a smaller proportion of carboxylic. A l l coals, regardless of their age or origin, can be classified in an ascending order of the elemental carbon content (Barnes et al. 1984). However, in the North American coal classification system ( A S T M D388-77-American Society for Testing Materials) a simplified coal testing is utilized. In this technical analysis, often referred to as proximate analysis ( A S T M D3172), ash, moisture and volatile matter contents ( A S T M D3173, D3174, and D3175) are determined. The fourth parameter, the fixed carbon, is calculated by subtracting the three from 100%, and both the fixed carbon (FC) and the volatile matter (VM) contents are often recalculated to "dry, ash free" (d.f.a.) or "dry, mineral matter free" basis (d.m.m.f). The volatile matter released during the determination consists mainly of a whole range of alkanes (C H2 +2) and alkenes (C H2 ), aromatic hydrocarbons (benzene, toluene, n  n  n  n  naphthalene) and tar (Rees 1966). The fixed carbon (d.m.m.f.) value is not quantitatively equivalent to the elemental carbon content. It merely estimates the amount of solid organic residue after removing moisture and volatile matter from the organic part of coal. Nevertheless, FCdmmf is still indicative of coal rank and, combined with V M d f , is the basis m m  for the A S T M Coal Classification system as shown in Table 1. Sometimes graphite is placed above anthracites in the classification system (Barnes et al. 1984) as it consists (almost 100%) of polyaromatic hydrocarbons. Lower rank coals are ranked according to their calorific values rather than the FC and V M parameters. This results from the fact that for the lower rank coals the FCdmmf and 37  t>fi 00 ^  I-I  </)  CD  in CN  \o  P  in  CD 15 Pi  CO ccj  -  c  CN  <o cn  CN  cn CN  o cn  TT  ON CD  CN  CN CN  o C  V  V  V  CN  cn  -°  c-> „  Id  cn  ON  cn  +•'  00  cn CN o" cn  cn  ON  CN  CN  A  00 |  V  ^ I  CN CN |  V  V  TT  m  •<* ON  <o CN CN  un  o cn ON A  CN  V  A  cn| V  <Nl  ml  V  A  <N|  CN CN |  oo I  A  A  J2  A  A  'o  > oo ON  J O  CN  I  ON  00  oo  V  V  1  I  00 ON  -o  ON VO |  V  c3 PL,  CN  O  ON  T3 CD X  A  I  *0 00  oo  I  A  so | A  ccj  O ccj  O  o CO Pi  o PI o  6  a  u  CM  l-H  P  H  o  l-H  u a  CD c/>  O  o  O O  =3 O  C/2  C/3  O  CD C/3 P!  o  O  a  p  CD  CM CN  cn  H  PI  • l-H  PI  IS CD  ffl CD  O cd  'o  >  I  as  "> 3  CD  CN  O >  i -— tp >  CM  u  O  I  cn  PI  ccj  O  O  CD  CD  C/3  CO  o  O  PI  p  H  pi  PI  1  CO p  u l-H  H  l-H  CM c/3  cd  m O O  CD  O >  o CD  P  I  O  H-1  PI  a  I  >  a  pi  O  CM  CM CN  CM  NH  z o  CD  m  I • l-H  H-l  CD  hH -  CN  CO  u  38  VMdmmf  do not proportionally change with the rank. The calorific value, on the other hand,  gives a good correlation with the lower rank coals and is hence used to classify such coals (Baughman 1975). Chemists view the organic matter of coal as a highly cross-linked polymer, which consists of a number of stable fragments, with predominantly aromatic structures, bonded together by relatively weak cross-links (Davidson 1982). Three types of structures can be distinguished (Larsen and Kovac, 1978). The first-order structure gives the size distribution of the macromolecules in coal and the degree of cross-linking. The second-order structure details the cross-links themselves and the structure of the carbon skeleton. The third-order structure characterizes the nature and distribution of the functional groups. A l l the different organic components of coal microscopically appear as clearly distinct entities referred to as macerals (Stach 1982). The macerals are classified into three groups: vitrinite, exinite (liptinite) and inertinite. Chemical and physical properties of the macerals, such as elemental composition, moisture content, hardness density and petrographic features differ widely and change during the course of coalification (Stach 1982). The recognition of individual macerals is based on microscopic examination. Macerals do not occur in isolation but form associations in various proportions and with different amounts of mineral matter. These associations are termed lithotypes and microlithotypes, and they give rise to the characteristic banded or layered character of most coals, visible even with a naked eye. Lithotypes can be distinguished macroscopically while microlithotypes are identified microscopically.  39 Mineral Matter of Coal  The objective of coal cleaning is to reduce the content of inorganic impurities (the mineral matter). The most common criterion for evaluating coal beneficiation quality is the ash content as determined in proximate analysis. It must be remembered though that during the ashing process at the usual temperatures of about 700-750°C ( A S T M D3174), the original coal minerals are chemically altered through a series of oxidation, decomposition and dehydration reactions. Therefore, the ash content does not reflect the true mineral matter content of the tested coal. Several empirical formulae were developed to relate the ash content to the mineral matter content but probably the two most used are the Parr formula (1932) and the KingMaries-Crossley (KMC) formula (1936). The former is commonly used throughout North America while the latter is frequently applied in Great Britain and Europe. In the Parr formula, the percent content of mineral matter is calculated by:  % MM = 1.08 • Ash% + 0.55 • S%  (35)  where Ash% and S% are the percent ash and sulfur contents in coal, respectively. The K M C formula has a more complex form:  % MM = 1.09 • Ash% + 0.5 • S%  pyr  +U-S0 % 3  atk  +S0 % 3  coal  + 0.8 • C0 % 2  (36)  + 0.5-C/%  40  where S% is the pyritic sulfur content in coal, COi% is the mineral carbon dioxide content, pyr  S0 % h and S0 % i are the sulfur trioxide contents in ash and coal, respectively, and C/% is 3  as  3  coa  the chlorine content in coal. Both formulae indicate that the true mineral matter content is roughly 10% higher than the ash content. The presence of large amounts of carbonates and hydrated clays, however, may cause large errors when using the above equations in estimating the mineral matter content (Rees 1966). Harvey and Ruch (1986) tested a wide range of US coals and identified over 20 different minerals. The most common ones include clay minerals (illite, kaolinite, smectite), sulfides (pyrite), carbonates (dolomite, calcite) and oxides (quartz). As proposed by Cook (1981), the mineral matter in coal occurs at different macroscopic and microscopic levels of heterogeneity: 1) at the seam level - a large proportion of mineral matter arises from its inclusion during mining of roof and floor rock, 2) at the lithotype level - the minerals may occur as deposits in cracks and cleats, or as veins. 3) at the macerals level - the mineral matter may be present in the form of very finely disseminated discrete mineral matter particles. 4) at the submicroscopic level - the mineral matter is present as strongly, chemically bonded elements. The terms "extraneous mineral matter" and "inherent mineral matter" were usually used to describe an ash-forming material, separable and non-separable from coal by means of physical methods. Keller (1984) found that the majority of mineral grains in coal fall within the size range from 1 to 10 um with only 1% of the total being finer than 1 urn. Traditionally, only the mineral fractions from the first two levels of heterogeneity were liberated and 41  removed. However, ultra fine grinding combined with recent advances in fine coal beneficiation technologies may redefine the traditional "separability" limits. Coal Surface Wettability  A l l fine coal beneficiation processes (e.g. froth flotation) are based almost exclusively on the differences between the surface properties of the carbonaceous matrix and mineral components of coal. It is also known that the interaction of a coal surface with water, referred to as coal wettability, is of primary importance in those processes and therefore the wettability of coals by water has been the subject of very extensive research. As soon as the froth flotation process was developed, it was observed that certain minerals were naturally amenable to flotation while others required special pre-treatment. No good explanation of this natural floatability was proposed until the fundamental work of Gaudin et al. appeared in 1957. Gaudin et al. showed that natural floatability results when at least some fracture or cleavage surfaces form without rupture of interatomic bonds other than "residual", weak (dispersion) bonds. On the other hand, the lack of natural floatability appears when all natural fracture or cleavage surfaces offer strong (ionic, covalent, metallic) bonds to the surrounding medium (water) in greater density than some threshold value. In other words, when as a result of grinding strong chemical bonds are broken within a mineral crystal, the exposed surfaces will tend to strongly interact with water. When weak bonds are broken, the freshly obtained surfaces will only weakly interact with water. In the first case of strong mineral-water interaction, the mineral is called hydrophilic (waterattracting). In the latter case, the mineral would be referred to as hydrophobic (water  42  repelling). Hydrophobic minerals, therefore, possess natural floatability while hydrophilic ones are very difficult to float. Zisman (1963) and Fowkes (1964, 1967) later quantified these very empirical observations. The force of interfacial attraction which promotes wetting is the work of adhesion, W , which is the sum of individual terms resulting from the several kinds of A  interfacial attraction (Fowkes 1967):  W = W +W d  A  h  A  +W +W P  A  K  A  +W  e  A  A  (37)  where the superscripts refer to dispersion forces, hydrogen bonds, polar interactions, pi-bonds (common in solids containing aromatic hydrocarbons), and electrostatic interactions. Presently, the electrostatic, polar and hydrogen bond terms are grouped into a single acid-base interaction since these interactions involve a donation/reception of protons (Bronsted acids/bases) or electrons (Lewis acids/bases), as proposed by Good et al. (1990). In 1805, Young using simple vectorial force summation showed that a three phase contact angle between a liquid, its saturated vapor and a solid can be calculated from the following formula, known commonly as the Young equation:  YSV-YSL  =YLV  where y sv, ysL, 7iv  C  °S<9  (38)  are the solid-vapor, solid-liquid and liquid-vapor interfacial tensions,  respectively, and (9 is the three phase contact angle measured through the liquid phase. In their approach, Zisman and Fowkes made use of the Dupre equation. Dupre (1859) showed, using also the Young equation in his analysis, that the work of adhesion per unit area 43  of a liquid with a solid is related to the various interfacial tensions according to the following expression:  A=Y  »+Y  W  S  (39)  LV-YSL  where y „ refers to the solid in vacuum. s  Although the Young and Dupre equations were proposed almost 200 years ago, their validity has never been questioned. In fact, these expressions are two of the most fundamental equations in surface chemistry. Combining the Young and Dupre equations, after a simple rearrangement with respect to W , Zisman (1963) obtained the relationship, also known as the Young-Dupre equation: A  A ={y «  W  s  The [ys"~ysv)  (40)  -rsv)+rLv^+cos0)  difference is sometimes referred to as the equilibrium surface pressure of  water vapor (if the liquid in question is water) on the solid surface and very often it can be assumed to be zero (Fowkes 1964). Therefore, the Young-Dupre equation may simply be written as:  W =r y(l + COS0) A  (41)  L  In the case of water, the water-vapor interfacial tension, y , (or simply the water LV  surface tension) and the three-phase contact angle, 0, can be easily measured. Hence, the work of adhesion of water on any solid can also be estimated. 44  Since the work of adhesion is the sum of several different contributions from different types of interfacial forces, for hydrophilic (non-floatable) solids, strongly interacting with water through polar and electrostatic interactions (e.g. metal oxides), the work of adhesion and cos© will be high (l>cos(x)<-l). This, in turn, means that the three-phase contact angle of water on such a solid would be close to zero. On the same basis, hydrophobic (floatable) solids weakly interacting with water, through dispersion forces for example, would give high contact angles. In fact, all solids would be hydrophobic i f they did not carry polar or ionic groups, as shown by Laskowski and Kitchener (1969). Contact angle measurements offer a powerful tool for probing the interfacial interactions between a solid and a liquid. Contributions of essentially all the components to the work of adhesion can be estimated for different solid-liquid or liquid-liquid systems. Fowkes (1964) measured the interfacial tensions between water and a series of liquid alkanes (from n-hexane to w-tetradecane). Alkanes can be considered as model liquids interacting only through weak dispersion forces (Morrison and Boyd 1966), as their molecules do not possess any polar or ionizable functional groups. Knowing the surface tensions of the alkanes and water, as well as the measured interfacial tensions, Fowkes was able to estimate the dispersion force component to the surface tension of water from the following equation (Fowkes 1964):  (42)  where yu is the water-alkane interfacial tension, y\ and y are the surface tensions of water 2  and the alkane, and y  d  , y  d  are the dispersion force components to the water and alkane  surface tensions, respectively. In the above equation only the dispersion force component for 45  water is unknown since the surface tension of the alkane originates entirely from dispersion forces (^2  =r )2  In this way, Fowkes (1964) found the dispersion force component for water to be equal to 21.8 ± 0.7 mJ/m . Since the surface tension of water at 20°C is 72.8 mJ/m , the 2  2  "remaining" 51 mJ/m can be attributed to hydrogen bonding and dipole-dipole interactions 2  between water molecules. Similar theoretical analysis was also performed for coals. In order to estimate the dispersion force component for coals, Gutierrez-Rodriguez et al. (1984a) used methylene iodide and showed that the values of the contact angle of this liquid did not depend on coal rank or oxidation and were equal to 28 ± 9 deg. This, combined with the methylene iodide surface tension of 50.8 mJ/m , and assuming y Lv = YLV for this compound, gave the 2  d  dispersion force component for coals of 45 ± 4 mJ/m . Using the dispersion components for 2  water (21.8 mJ/m ) and coal («40 mJ/m ), and the water surface tension of 72.8 mJ/m , 2  2  2  Laskowski (1994, 2001) estimated from the Young-Dupre equation that for an ideal (smooth and chemically homogeneous) coal surface the water contact angle should be 100.5 degrees. Such a high value would indicate a strongly hydrophobic solid. Even polished coal surfaces are neither smooth nor chemically homogeneous. Since their composition also varies greatly with coal rank, several systematic studies were undertaken to experimentally find a relationship between coal surface wettability (contact angle) and coal rank. Brady and Gauger (1940) observed that an air bubble placed on the surface of a low rank lignitic coal immersed in water did not adhere at all to the coal surface (0 = 0). They also found that the contact angle values on anthracite (0= 48 deg) were much lower than the angles measured on a bituminous coal surface (0= 60 deg). Brady and Gauger also tested the  46  effect of a strong oxidizing solution, i.e. potassium permanganate, on the wettability of a Pennsylvania bituminous coal and found that the coal became completely hydrophilic ( 0 = 0 ) after treatment with the oxidant. Though based on a limited number of coal samples, this early work suggested that the contact angles measured on coals of different ranks should initially increase from lignites to bituminous coals, and then drop off again for anthracites. Much more detailed works of Russian researchers on coals from Donbass Basin, Ukraine, (Elyashevitch 1941, Elyashevitch et al. 1966, 1967, Klassen 1966) confirm this correlation. When contact angles are plotted as a function of volatile matter content (coal rank), the plots show maxima at a volatile matter content of about 20%, corresponding to medium- to low-volatile bituminous coals. Both subbituminous and anthracitic coals give contact angles lower than those for medium-volatile coals. The same trends were observed by Horsley and Smith (1951), Sun (1954), GutierrezRodriguez and Apian (1984), Gutierrez-Rodriguez et al. (1984), Rosenbaum and Fuerstenau (1984), and Y e and Miller (1988). In these contributions, the contact angle data were plotted as a function of carbon content (more common way in the Western literature) and maxima were found near a carbon content of 80-85% characteristic for medium-volatile bituminous coals. Klassen (1966) explained low contact angles in the low rank range by the high oxygen content in those coals. Hydrogen bonding, dipole-dipole and even electrostatic interactions between water molecules and the oxygen functional groups on the coal surface would facilitate the wetting of such hydrophilic coals. Klassen (1966) also suggested that the "anomalous" behavior of anthracites results from the sharp increase in aromaticity for this class of coals in addition to a marked porosity. One may speculate that the aromatic rings in anthracites can additionally interact with polar 47  water molecules through the ^--electrons thus increasing the total work of adhesion and decreasing the contact angle. Interestingly, according to Whitehurst (1980), the aliphatic hydrocarbons content in coals also varies with rank and reaches a maximum in the range of 80-85% C which falls in the same region as the maximum for the contact angles. It is also worthy of notice that these aliphatic hydrocarbons are mostly higher alkanes, straight-chain or branched, interacting only through weak dispersion forces, that should indeed produce highly hydrophobic surfaces. The experimentally determined contact angle values on real coal samples hardly ever exceed 70 deg for the most hydrophobic ranks, and are always lower than the theoretical value of 100.5 deg obtained under the assumptions discussed above. This fact reflects the highly heterogeneous nature of coal surfaces. The observed maxima on the contact angle-vsrank curves were theoretically predicted by Rosenbaum and Fuerstenau (1984) and Keller (1987) in the patchwork assembly model of the coal surface. In order to calculate the contact angles on heterogeneous coal surfaces, these researchers used the Cassie-Baxter equation (Cassie and Baxter 1944, Cassie 1948) in the following form:  cos(<9) =Y a -cos(0) 5  i  i  (43)  i  i  where (0)s is the contact angle on the coal surface, the a coefficients are the surface coverage t  fractions of each of the components (Ea, = 1), and ((9); are the contact angles on the individual surface components. In their calculations, Rosenbaum and Fuerstenau pictured the coal surface as consisting of discrete, uniformly distributed patches of graphitic, paraffinic and hydrophilic components. Keller (1987) additionally introduced the area occupied by pores. Keller 48  assumed that pores are filled up with water so they would, generally, increase coal surface wettability. However, as shown by He and Laskowski (1992), the effect of porosity depends on coal rank. For high rank coals, water would not be able to displace air from the pores, so the presence of the latter would rather increase coal hydrophobicity. In the case of lower rank, hydrophilic coals the pores fill up quickly so their wettability is increased even further. It is worth noting that Elyashevitch and her coworkers (1941, 1966, 1967) experimentally found that the contact angles measured on coals using the sessile drop method depend on humidity. Electrokinetics and Electrical Charge at Coal/Water Interface  The high heterogeneity of coals also affects their electrochemical behavior. Various heteroatoms, functional groups along with the mineral impurities all contribute to the overall electrical surface charge of coal particles. In order to determine the electrical charge at a solid/liquid interface, one can directly titrate a solid dispersion with a solution(s) of potential determining ions (PDI) and record the changes in the PDI concentrations resulting from their adsorption. The charge can also be determined indirectly with the use of electrokinetic methods. These techniques are much easier to carry out and commercial instruments are widely available. Therefore, electrokinetic measurements are much more common. A l l charged particles in aqueous solutions develop the electrical double layers (EDL) around themselves. In the electrophoretic effect, when a charged particle starts to move in an electric field, the electrical double layer becomes distorted and the diffuse layer of ions is sheared off, leaving ions of the compact (Stern) layer still "attached" to the particle. Direct 49  measurement of the velocity of this motion is the principle of electrophoresis - one of four electrokinetic effects. The velocity of a particle with the Stern potential  moving through a liquid with the  viscosity 77, and the dielectric constant e, in an electric field of the strength E, is given by:  v = le*F E/T] d  (44)  or in terms of the electrophoretic mobility, VE = v/E:  ^  E  = \ ^  d  I V  (45)  This equation is also called the Hiickel equation (Huckel 1924) and allows determination of the Stern potential once the electrophoretic mobility is experimentally measured. Since the surface charge density and the surface potential are related through the Poisson-Boltzmann equation, the surface charge can be finally obtained. As already discussed, the surface potential changes with the distance from the solid surface according to Equation 6. In electrophoretic measurements, the potential measured with the use of the Huckel equation is not really measured at the particle surface but at a certain distance d from the surface. It can be reasonably assumed that d is equal to the thickness of the compact (Stern) layer of the E D L as this layer also moves with the charged particle. The potential at the distance d, at the plane of shear, is called the zeta potential, £ to distinguish this quantity from the Stern potential, *F^. In many practical applications, it is conveniently assumed, though, that the zeta potential is equal to the surface potential (Hunter 1993). 50  In a more detailed analysis, Henry (1931) showed that the Hiickel equation should be written in a more general form, known as the Henry equation, to account for the effect of the thickness of E D L , \IK, and particle size, R:  2£e  v = E  •f(xR)  (46)  The correction function / (KR) varies monotonically from 1 to 1.5 as KR changes from 0 to infinity. At one extreme, when/(KR) = 1, the Henry equation becomes the Hiickel equation. At the other extreme, for/(KR) = 1.5, the Henry formula assumes a simple form:  v =— E  (47)  that is better known as the Smoluchowski equation (1903, 1921). As discussed by Laskowski and Parfitt (1989), for the range of ionic strengths and coal particle sizes encountered in fine coal processing, the Smoluchowski formula is fully applicable. Electrokinetic measurements carried out on model systems offer some insight into the origin of charge at the coal/water interface. Pure paraffin wax (a mixture of higher, solid alkanes, C H2 +2, n>18), Nujol oil (a mixture of liquid alkanes, 18>n>5) as well as pure liquid n  n  alkanes all seem to be negatively charged over the pH range from 3 to 10 with the i.e.p. situated near pH 3 (Douglas and Shaw 1957, Mackenzie 1969, Pereira and Schulman 1961, Arbiter et al. 1975, Wen and Sun 1981, Liu et al. 2001). The origin of the negative charge on such inert surfaces remains disputable but a weaker hydration of OH" ions than H ions may +  51  be responsible for some sort of preferential adsorption of the former at the hydrophobic alkane/water interface (James 1981, Zhou et al. 1998). Graphite composed of polyaromatic hydrocarbons gives a response very similar to the behavior of alkanes. Pure graphite particles are negatively charged in the pH range from 2-10 (Spurny and Dobias 1962, Chander et al. 1975, Solari et al. 1986) although the presence of impurities may shift the p.z.c. to less acidic values (Solari et al. 1986). The origin of the negative charge on the hydrophobic graphite surface may be the same as in the case of alkanes, but graphite oxidation may also be a factor (Groszek 1975). Systematic studies indicate that the electrokinetic behavior of "real" coals is also a function of coal rank. After analyzing the vast literature available on the subject, Laskowski and Parfitt (1989) and Laskowski (1987, 2001) constructed a generalized zeta potential-vs-pH diagram for coals of different ranks (Figure 2B). 9-  Anthracites  +  Bituminous  ° o 133  Subbituminoua Heavily oxidized  " i — 75  80  1  — i — 85  1  — i —  1  — r  90  Carbon Content (d.m.m.f)  95 [%]  100  14  0 pH  Figure 2. Relationship between coal rank and pH of i.e.p. (Figure 2A, Laskowski 2001) and a generalized zeta potential-pH diagram for coals varying in rank (Figure 2B, Laskowski and Parfitt 1989).  According to this diagram, the most hydrophobic bituminous coals have the highest isoelectric points (i.e.p.) falling within the p H range of 5-8. For lower rank coals (sub-  52  bituminous and lignites), the isoelectric points shift towards more acidic values (pH 3-4) as these coals are known to contain weakly acidic groups. Strongly oxidized coals, with their surfaces densely covered with carboxylic (humic) acids, have their i.e.p.'s at even lower pH as such coals approach the electrokinetic behavior of higher alkyl-carboxylic acids. Anthracites, coals of the highest rank, exhibit i.e.p.'s roughly in the same pH range as do the lower rank sub-bituminous coals, similarly to the response of model graphites. It is generally accepted that the negative charge on a coal surface originates from the presence of polar functional groups, predominantly carboxylic and phenolic. In aqueous solution carboxylic acids exist in equilibrium with the carboxylate and hydrogen (hydronium) ions:  RCOOH + H 0±* 2  RCOO~ + H 0  ()  +  48  3  As for any reversible equilibrium, the concentrations of the components are related through the equilibrium constant, K , given by the expression: a  JRCOO-XH^l [RCOOH]  v  '  In the case of acids, the constant is referred to as the acidity constant. For carboxylic acids, K  a  ranges from 10~ to 10" (Morrison and Boyd 1966). 4  5  The above equation can be transformed into a logarithmic form:  [RCOQ-]^  r  \ozK  a  =log  [RCOOH]  [RCOQ-]"  r  + l o g [ / / 0 ] = log +  3  [RCOOH]  pH  (50)  53  Each carboxylic acid may be then characterized by the pH value at which the concentration of the carboxylate [RCOO]  ion is equal to the concentration of the  undissociated acid [RCOOH]. Under such conditions, the above equilibrium can be simplified to: \ogK =- H a  P  or,  (51)  pK = pH a  The pK values for pure carboxylic acids, therefore, range from 4 to 5. For halide- or a  benzyl-substituted carboxylic acids, the pK may be as low as 3 (Morrison and Boyd 1966). a  At any pH higher than the pK„, the concentration of surface -COO" ions will be higher than the concentration of the undissociated form - C O O H . Interestingly, the pK values, and hence a  the chemistry of - C O O H groups, are not significantly affected by the association of the carboxylic groups with the carbonaceous matrix of coal. As discussed by Laskowski and Parfitt (1989), the zeta potential-pH curves for highly oxidized coals show a deflection point in the vicinity of pH = 4-5 indicating a buffering action of the COOH/COO" system. Similar calcuations can be performed for phenolic groups to show that their pK  a  values are around pH = 8-10, depending on the substitutes in the aromatic ring (Morrison and Boyd, 1966). This weak acidity of phenols additionally contributes to the frequent increase of zeta potential of oxidized coal particles towards more negative values in more alkaline solutions (Laskowski and Parfitt 1989). The origin of the positive charge on a coal surface is much more difficult to explain. The presence of nitrogen, as a potential proton acceptor, or polyvalent metal impurities in the form of oxides, were proposed as possible sources of the positively charged sites (Laskowski 1987).  54  In the case of oxidized/lower rank coals, all oxygen groups may serve as proton acceptors to generate positive charges. These oxygen groups are usually polar with a residual negative charge located on the oxygen atom. The mechanism of charging would be identical to the formation of the hydronium ion (i/?(9 ): +  H 0: + H* -^H 0:H  +  2  2  ->H 0  +  3  Such hydrogen complexation reactions would effectively render a surface positively charged. In the case of high rank coals, not containing oxygen, the origin of the positive sites is still not well understood. One may postulate that the ^--electrons of the aromatic rings are capable of attracting hydrogen ions to form positively charged complexes. However, no experimental evidence has been reported to support this hypothesis. Several studies described the effect of mineral impurities. Coal leaching with acids (HF, HC1) to remove mineral constituents results in a dramatic change of the electrokinetic behavior. Fuerstenau et al. (1983), Siffert and Hamieh (1989) and Hamieh and Siffert (1991) showed that demineralization of coals consistently leads to a shift of the i.e.p. towards higher pH values. These researchers postulated that even small quantities of minerals, such as silicates and quartz, may completely mask the electrostatic characteristics of the carbonaceous matrix. Siffert and Hamieh (1989) showed that a coal with an ash content as low as 6%, after demineralization to only 2% can still give a very different zeta potential-pH curve compared with the original sample.  55  3.2.2  Traditional Fine Coal Processing  From the utilization point o f view, all coals can be broadly classified into four groups (Osborne 1988): steam (thermal) coals, coking (metallurgical) coals, conversion coals and special coals. The conversion coals are used to produce gaseous or liquid fuels derived from coal. The special coals serve as the matrix for making active carbons and special types o f electrodes. The majority o f coals, however, fall within the first two groups as power generation and coke-making i n blast-furnace operations are the main uses o f coal today. The terms "coking" and "coke-making" refer to a process in which a bituminous coal is heated in the absence o f oxygen to drive off the volatile matter to leave a porous solid residue. This residue serves as a source o f carbon in subsequent metal ore blasting to produce pure metals. In the case o f thermal coals, the primary objective o f cleaning is to increase the calorific value and to reduce shipping cost by rejecting inorganic rock and separating troublesome impurities, e.g. sulfur. The main objective o f metallurgical coal cleaning is a reduction o f its ash content to maintain the optimum ratio o f reactives to inerts in the feed to a coke-making plant. In practice then, for both thermal and metallurgical coals, the primary objective o f their cleaning is rejection o f inorganic rock and sulfur. Most fine coal beneficiation processes, e.g. flotation or o i l agglomeration are based on the  differences  between the  surface properties o f carbonaceous matrix  and  mineral  components o f coal. The basic relationship between coal rank, wettability and the flotation response o f coal is well established; lower rank/oxidized coals are very difficult to float while higher rank bituminous coals float easily. The same qualitative trends apply generally to the process o f oil  56  agglomeration; hydrophobic coals are usually easier to process than hydrophilic, lower rank ones (Qiu 1992, Labuschagne 1986, Sadowski et al. 1988). In oil agglomeration, oil droplets attach preferentially to hydrophobic coal particles, merge and bridge the particles into larger agglomerates. Three main factors affect oil agglomeration (Capes 1989, 1991): solid wettability, the amount of bridging oil, the type and intensity of conditioning. Only very hydrophobic solids can be beneficiated. Oil dosages of at least 10% (by feed weight) are needed to produce large and strong agglomerates. Oil emulsification, prior to agglomeration, was shown to reduce the oil dosage and energy input (Bensley et al. 1977, Laskowski and Y u 1998, Y u and Laskowski 1999). Cationic (amines) and nonionic (alcohols) emulsifiers were reported to improve the oil agglomeration of oxidized coal (Good et al. 1994, Laskowski and Y u 1998, Y u and Laskowski 1999). Generally, high shear mixers are used to create intense mixing conditions. Several industrial processes have been developed, as reviewed by Mehrotra et al. (1983), in which various types of oils and additives (surfactants) were utilized. Since oil agglomeration is carried out in water, dewatering by centrifuging followed by pelletization is applied in the final stages. The main function of polymeric flocculants in mineral processing is to produce large and strong aggregates, or floes. The principal application of the flocculation process in coal preparation lies in the areas of solid/liquid separation (thickening and filtration) and waste water treatment. These operations, first of all, deal with tailings and hence it is the behavior of the tailings minerals (clays, quartz, carbonates etc.), and not coal, that controls the process. Nevertheless, Attia et al. (1987), Attia and Y u (1987), Palmes and Laskowski (1993) and Laskowski and Y u (1998) reported that certain flocculants may be used for the selective flocculation of coal. Hydrophilic flocculants, such as polyacrylamide, were shown to flocculate any type of coal regardless of rank or oxidation. On the other hand, hydrophobic  57  latices were selectively flocculating only hydrophobic coals. The same latices were completely ineffective against oxidized coals. Froth flotation of coal is, by far, the most widely used technology for fine coal beneficiation. As already pointed out, however, flotation is practically limited only to coals possessing natural floatability/hydrophobicity, i.e. to bituminous, metallurgical coals. Froth flotation is a physicochemical process which exploits the differences in the wettability of hydrophobic, organic parts of coal and the hydrophilic, mineral refuse. Separation is achieved by means of air bubbles which selectively attach to the hydrophobic particles and carry them to the top, where the concentrate is collected in the form of froth. The hydrophilic gangue is fully wetted and as such remains in the pulp as tailings. In coal practice, a frother and a collector are usually added to the pulp. The frother allows the production of stable, more uniform bubbles, while the collector enhances the hydrophobicity of coal particles. Flotation frothers are usually aliphatic, cyclic or aromatic alcohols (Laskowski 2001). Aliphatic hydrocarbons and various fractions from crude oil distillation are commonly used as coal collectors. When necessary, promoters and modifiers may be added as well. The former facilitate collector emulsification and its spreading over the coal surface while the latter include pyrite depressants and pH regulators (Laskowski 2001). Typically, only particles below 500 microns are processed by means of froth flotation (Apian and Arnold 1991,Nicol 1997). Since it is a worldwide trend to process more and more difficult-to-clean coals it is also expected that the amount of fine coal processed will increase as well. It was reported that the amount of fines (size fraction -500 um) in such a run-off-mine coal may be as high as 40% (Apian 1987), or even 60% for some western Canadian coals (Holuszko 1994). Furthermore, when a difficult-to-clean coal is processed, the amount of fines in the cleaning 58  circuits may be even higher since the extraction of middlings followed by recrushing is often required in such cases (Laskowski and Walters 1987, Osborne 1988). Therefore, any coal flotation plant actually treats a slurry consisting of different fine coal fractions generated from the various stages of coal preparation. So far, it has been common practice to dewater the cleaned coal and then to thermally dry it because the amount of water in the product (e.g. filter cake) is still too high, usually around 25-30% (Apian 1987). Even after chemical pretreatment (flocculation) the water content in the filtration cake can be as high as 22-23% (Groppo and Parekh 1996). Needless to say, thermal drying is a very expensive operation and very environmentally unfriendly.  3.2.3  Seam-to-Steam Strategy and Coal-Water Fuels (CWF)  One of many possible solutions to the dewatering problem is the "seam-to-steam" strategy (Marnell et al. 1983) in which fine coal is utilized in the form of Coal-Water Fuels (CWF), or simply Coal-Water Slurries. Coal-Water Fuel is a concentrated suspension of 6570% by weight of fine coal in water. In Marnell's idea, the product of fine coal cleaning is converted into a Coal-Water Fuel (Coal-Water Slurry) at the coal preparation plant and then the CWF is pipelined (or transported in any other way) to customers where it can be directly burned as a fuel oil. This kind of approach has already been foreseen as an integral part of modern, advanced coal cleaning technologies (Osborne et al. 1996). It is known that a Coal-Water Slurry (CWS) must meet certain rheological requirements in order to be efficiently pipelined and stored. Firstly, the viscosity and the yield stress must be low, and secondly, the settling of solids has to be minimal (Pommier et al. 1984, Papachristodoulou and Trass 1987). The first objective can be achieved by using a  59  suitable dispersant, normally a low molecular weight polymer, that adsorbs on solid surfaces and disperses the particles electrostatically and/or sterically. The settling stability of C W F can be improved by introducing another chemical additive, usually a high molecular weight polymer. Such a polymer forms some kind of weak networking that prevents particles from settling thus slightly increasing the viscosity and yield stress of the slurry and keeping the particles in suspension. Apparently the action of the dispersant opposes the action of the settling stability modifier, so the final product is usually a trade-off between low viscosity and good stability towards settling. On an industrial scale, various polyacrylates, polyolefm- and polystyrene sulfonates and their derivatives (MW<50,000) have been widely utilized as dispersants (Ohki et al. 1996, Tadros et al. 1995, Hayashi et al. 1993, Ukigai et al. 1994, Yoshihara 1999, Saeki et al. 1999). Natural high molecular weight polysaccharides, e.g. gums, (MW>1,000,000) have been commonly added as settling stability modifiers (Saeki et al. 1994, 1999, Usui et al. 1997a). Many nonionic, cationic and anionic surfactants were also tested as potential CWS additives, as reviewed by Botsaris and Glazman (1989). The rheological properties of concentrated coal-water slurries also depend strongly on coal surface properties. It was found by several authors that the amount of water adsorbed on coal, or absorbed in pores, is directly proportional to the surface oxygen group content and porosity, and hence to coal rank (Kaji et al. 1986, Igarashi 1984). Consequently, Kaji et al. (1983) and Seki et al. (1985) showed that the viscosity of CWS was strongly affected by the amount of water adsorbed on coal or trapped in pores. These researchers found that lower rank coals characterized by high equilibrium moisture contents gave low solids loadings in CWS. In  60  contrast, bituminous coals that had inherently low moisture contents produced highly concentrated suspensions compared to the lower rank coals. As predicted from the modified D L V O theory, hydrophobic bituminous coals tend to agglomerate in water suspensions apparently due to the hydrophobic attractive forces while lower rank/oxidized (hydrophilic) coals are well dispersed mainly through electrostatic repulsion. Thus, the hydrophilic coals should theoretically produce low viscosity- and highly concentrated suspensions. This conclusion is in sharp contrast to the correlations found experimentally by the Japanese researchers and this apparent contradiction will be discussed further in the experimental part of this thesis. It must be pointed out that low rank coals, when mixed with water to prepare a CWS, in a way dilute the suspension with the moisture adsorbed on their surfaces. It should also be remembered that this surface water (moisture) content may be as high as 40% for some sub-bituminous coals. Therefore, coal-water slurries prepared from  low rank coals exhibit much lower coal contents (per dry solids  weight/volume) at the same viscosity as those obtained from high rank coals (Seki et al. 1985, Schwartz et al. 1985). This qualitative trend agrees well with the lubrication theory proposed by Potanin and Uriev (1991) who postulated that the viscosity of concentrated solid-liquid suspensions is controlled by the availability of the suspending medium (i.e. water) that acts as a lubricating agent. Seki et al. (1985) found the following expression to calculate the amount of the free (lubricating) water:  M  F  = 100 — C 1 + —±v 100 J C  (52)  where M is the free water content in suspension (%), Cc is the dry coal content (%), and MQ F  is the coal equilibrium moisture content (% at 80% humidity and 20°C). 61  A s Seki's relationship shows, the amount o f free water i n the system decreases as the coal moisture content increases (coal rank decreases). For highly concentrated coal-water suspensions, this lubricating water is needed to reduce viscosity. Hence, i n the case o f low rank coals characterized by high values o f Mc, the free water content can be maintained at the required level only by lowering the coal content i n the slurry. Therefore, it is not surprising that so far only bituminous coals have been utilized in the form o f Coal-Water Fuels on an industrial scale (Usui et al. 1997b, Hashimoto 1999). Another important reason behind this fact is that bituminous fines are easy to clean by means o f traditional methods, primarily by froth flotation. However, since Coal-Water Fuels are supposed to replace fuel oils for power generation, it is understandable that C W F technology should primarily utilize thermal, difficult-to-fioat coals. A t the present time, though, thermal coal fines are often rejected without any beneficiation. Several upgrading processes o f lower rank coals were designed i n an attempt to facilitate the production o f C W F from those coals (Usui et al., 1997b, 1999, W i l l s o n et al. 1997, Ohki et al., 1999, Guo et a l , 1999, Ono et al., 1999, Suwono and Hamdani, 1999). These methods experimental  are technologically very complex and have not gone far beyond the  level. Nevertheless,  they  clearly reflect  the  increasing interest  in  the  beneficiation and utilization o f lower rank coals. It should also be noted that these low rank coal upgrading technologies are designed to deal specifically with the water immobilization problem. The so-called hot water drying (Wilson et al. 1997) is based on the exposure o f coal particles to very high temperatures at elevated pressures for several minutes. Under such conditions some o f the surface oxygen groups are removed as a result o f decarboxylation. More importantly, during this process, devolatilized oils and hydrocarbons condense on the surfaces o f coal particles immersed in water, covering the coal with a tar-like coating which  62  seals pores and prevents water from re-absorbing. A similar low rank coal upgrading technology was developed by Usui et al. (1997b). Their process involves a combination of high temperature (200°C) vacuum drying of coal followed by tar coating at 270-350°C. At such high temperatures tar is present in the form of vapor so when the cooling stage starts the coal is very uniformly coated with liquefied tar that fills up all pores. Such a tar coating would also render the coal surface hydrophobic. The maximum coal loading of CWS increased from 45%, for raw coal, to about 65% for the upgraded product  3.2.4  Effect of Coal Cleaning on Properties of CWF  If the seam-to-steam strategy is to be applied successfully, the transition from the wet fine product to C W F must be as technologically simple as possible. A typical process of froth flotation requires - depending on coal floatability - a variety of chemical additives (Wheeler 1994). As mentioned earlier, in oil agglomeration some surface active agents (e.g. dispersants or emulsifiers) may be needed to improve the separation of coal from gangue minerals. In other words, the product of fine coal cleaning that can be used for the preparation of CWF inevitably contains a certain set of chemical additives. Surprisingly, the literature on the effect of coal cleaning on the rheological properties of Coal-Water Fuels is very limited, while both the rheology of C W F and coal cleaning technologies are discussed separately in great detail and the literature on the two subjects is vast. The few existing reports indicate, however, that the chemicals used in conventional fine coal cleaning, and sometimes the cleaning technique itself, are not compatible with C W F technology. The main disadvantages include: -  Flotation collectors increase coal's hydrophobicity which leads to an enhanced aggregation of fine coal.  63  After flotation, flocculation or oil agglomeration, recrushing followed by redispersion of the clean fine coal would be needed to convert the coal to C W F (Trass and Papachristodoulou 1986). This problem even led to the formulation of a separate class of Coal-Water-Oil Fuels. -  Nonionic surfactants (aliphatic alcohols and ethers) are usually needed to deal with coal cleaned by flotation (or oil agglomeration) and these are known to form gels at the temperatures used to preheat C W F (Dooher et al. 1985) so the final combustion also suffers (Winters et al. 1985). From the chemical point of view, any industrial fine coal flotation produces highly  hydrophobic coal that would undergo aggregation and would need redispersion to reduce viscosity. In the case of the hydrophilic lower rank coals, making them hydrophobic for flotation and then redispersing, or rendering them hydrophilic again, for C W F preparation does not seem to be the right practice. In a critical review of the coal cleaning methods, and coal flotation in particular, Laskowski (1999) concluded that modern flotation processes have never been optimized with regard to the needs of C W F technology. Additionally, flotation reagents and viscosity reducers are not compatible. According to Laskowski (1999), some further modifications of flotation circuits would also be needed to obtain a product with a bimodal particle size distribution. Despite the fact that Coal-Water Fuels have been proven experimentally to be very attractive, the concept has not yet found worldwide recognition. To the author's knowledge only Japan commercially uses Coal-Water Fuels on a large scale and C W F technology is advanced the most in that country. A new plant has also recently been built in Sardinia, Italy (Bozano et al. 2000) employing conventional flotation for coal cleaning. A simplified plant flowsheet is presented in Figure 3. 64  Cjeed)— - 6mm Raw Coal  -ri P / G  Primary Grinding T, /Sieve Bend (0.5 mm) S e p  Rougher Flotation  Sieve Bend (0.2 m m Sieve Bend  Water Recycling Tank Drum \ ~  Horizontal Filter di>r  Filtrate  Coal-Water Fuel -'. C W F .  Vertical Mixers  Figure 3. A flowsheet of a CWF preparation plant at the Oristano harbor on Sardinia, Italy (Bozano et al. 2000).  As the flowsheet illustrates, the raw coal is ground below 0.5 mm and sent to rougher flotation. The concentrate is then deep-cleaned with the use of flotation columns. The clean coal is re-ground below 0.2 mm and partly dewatered by settling in a thickener. No flocculants are used in this stage to avoid any contamination of the coal with the polymers. About one-third of the coal-containing underflow is further ground to produce a suitable proportion of ultrafines for optimization of the coal particle size distribution. The fines are then combined with the main stream underflow and the slurry is filtered in a drum filter. The  65  moisture content in the cake is about 25%. Rheology modifiers are subsequently added to the wet filter cake and the final coal-water fuel is produced in high intensity mixers. The top size of 0.5 mm is optimal for high yield flotation. The secondary grinding below 0.2 mm is dictated by combustion characteristics of the power generation plant that utilizes the coal-water fuel. Particles coarser than 0.2 mm would not be completely burned producing significant quantities of unwanted soot in the combustion boiler. The flowsheet is an excellent example of an actual application of the seam-to-steam strategy.  3.2.5  Coal Reverse Flotation  In this thesis, an alternative fine coal cleaning technique for the preparation of CWS is studied, in which the gangue particles are floated instead of the organic carbonaceous ones. The technique is referred to as coal reverse flotation and requires a simultaneous depression of coal and the use of collectors to float the gangue particles (Eveson 1961, Stonestreet and Franzidis 1988, 1989, 1992). Different quaternary amines were apparently shown to act as both coal depressants and mineral activators in such a method. Since amines can be used as fairly good cationic dispersants of coal (Tadros 1985), the presence of this group of chemicals in the cleaned coal may help in the direct preparation of CWF from the coal reverse flotation reject (clean coal). Other coal depressants can also be potentially used. The idea of coal reverse flotation is not entirely new. So far, the process has found only limited application to the selective removal of pyrite from coal. K.J.Miller (1975) patented a two stage froth flotation procedure which involves the conventional flotation of coal followed by the second stage re-flotation of pyrite from the first stage concentrate. In the  66  second stage, separation is achieved using dextrin (Aero Depressant 633) as a coal depressant, potassium amyl xanthate as a pyrite collector and MJJ3C (methyl isobutyl carbinol) as a frofher. Miller reported that the pyritic sulfur content decreased by as much as 80% after the two-stage flotation process, with simultaneous ash rejection from 31.8 to 7.8%. It is noteworthy, that Miller's study was one of the first to deal with the desulfurization of coal by flotation. K . J . Miller and Deurbrouck (1982) later discussed the advantages of two-stage reverse flotation, as a pyrite removal method, over the conventional flotation techniques. This process was also studied by J.D. Miller et al. (1984). This time a well-defined sample of dextrin was used as a coal depressant together with potassium amyl xanthate and MIBC. Although dextrin was shown to adsorb on both coal and pyrite, the flotation response of pyrite to xanthate as a collector was not affected at all, whereas coal was strongly depressed. In a review of pyrite recovery mechanisms in coal flotation, Kawatra and Eisele (1997) list more than 40 techniques of pyrite removal, including a series of tests conducted by Chander and Apian (1989) on the use of different starches (cationic, xanthated, high-amylose) as organic coal depressants. Experiments of Good et al. (1994) on pyrite rejection by means of oil agglomeration showed that in the presence of starch, a coal surface becomes hydrophilic, as indicated by decreasing values of the contact angles. This suggests a possible use of starch as a coal depressant. It is interesting to note that both starch and dextrin belong to the same class of polysaccharides which differ by molecular weight only. The above cited studies on pyrite reverse flotation show that starch and dextrin can be successfully used as selective coal depressants. Moreover, their adsorption was shown to be  67  very weak on such a common mineral component of coal as quartz (J.D. Miller et al. 1984, Liu and Laskowski 1989). A work of Solari et al. (1986) indicates that carboxymethyl cellulose (CMC) - a modified polysaccharide - is able to depress the flotation of hydrophobic minerals such as graphite. The depressant behavior of C M C and other organic colloids was first studied by Klassen (1966), who also summarized the experimental results of Russian researchers on coal flotation. Based on those studies, Klassen showed that organic colloids at very small dosages (5-20 g/ton of coal) actually improved coal flotation while at higher concentrations the depression of coal took place and flotation recovery dropped. Recent flocculation studies strongly suggest that certain flocculants should be effective  as coal depressants. Pradip and Fuerstenau  (1987) tested the effect  of  polyacrylamide (PAM) adsorption on coal wettability through contact angle measurements. They concluded that the coal surface became more hydrophilic in the presence of P A M . Water contact angles measured on a flat polished surface decreased from 60 deg to 30 deg at a P A M concentration of 1.6 g/dm . Palmes and Laskowski (1993) investigated the effect of coal surface and flocculant properties on the selective flocculation of coal. Palmes and Laskowski showed that polyacrylamide can be used for the flocculation of any type of coal regardless of the rank. Very similar flocculation behavior of P A M was reported very recently by Raichur et al. (1997), who also found that at higher dosages of P A M the flocculation efficiency dropped substantially due to coal redispersion by the excess of the polymer. Some results on the effect of P A M on coal flotation were reported by Moudgil (1983). Strong depression of coal was observed in the presence of P A M at neutral pH and P A M dosage of 10 mg per kg of coal. The depressing action of P A M in coal flotation was also  68  described by Pikkat-Ordynsky and Ostry (1972) who found that higher dosages of P A M (150 g/m of the pulp) reduced both the concentrate yield and ash content in tailings. Humic acids are another group of very promising coal depressants. Laskowski et al. (1986) and L i u and Laskowski (1988) discussed in detail the effect of humic acids on coal flotation. In the presence of humic acids the zeta potential values became more negative and almost complete coal depression was observed although staged reagent addition and manipulation of pH improved the flotation recovery. Similar results were obtained by Lai et al. (1989) who found that humic acids, even at low concentrations («10 ppm), decreased the flotation recovery from 92% to less than 20% and that the adjustment of pH to more alkaline values improved the flotation response of the tested coals. Consistent with these observations, Liu and Laskowski (1988) and Laskowski and Y u (1994) showed that the depressing effect of humic acids was most pronounced at low pH values, most likely due to the precipitation and increased adsorption of the free humic acids. Other researchers also reported the depressant action of humic acids in the flotation of other hydrophobic minerals such as graphite (Wong and Laskowski 1984) and molybdenite (MoS ) (Lai et al. 1984). 2  The cationic flotation of various aluminosilicates and silica (the main constituents of coal mineral matter) with the use of primary amines as collectors is very well documented (Gaudin and Fuerstenau 1955, Manser 1975, Smith and Lai 1966, Smith 1973). As electrokinetic measurements indicate, the surfaces of the minerals associated with coal are negatively charged over a wide pH range (above pH = 2-3) while amine molecules are present in solution as ionized, positively charged species below pH « 10. This creates ideal conditions for the adsorption of amines on the negatively charged mineral surfaces over a wide pH range. Since amine molecules orient themselves on such surfaces with their aliphatic chains pointed towards the water phase, the surfaces "look" more hydrophobic and flotation easily occurs. At 69  the same time the orientation of amines is reversed on the hydrophobic coal surface. It appears that the ionized amine groups point outwards rendering the coal surface hydrophilic. The problem with the use of amines, however, is that they were reported to be good collectors or promoters in the flotation of oxidized coals (Sun 1954b, Wen and Sun 1977) whose surfaces are usually negatively charged over almost the same pH range as the surfaces of mineral constituents (Laskowski and Parfitt 1989). Therefore the affinity of amines towards both types of surfaces and the adsorption mechanism would be similar, and the selectivity in the reverse flotation of oxidized/lower rank coals could be lost completely, although adjustment of pH and amine concentration may give improved results (Wen and Sun 1977). Coal reverse flotation has not been adequately studied. In 1961, Eveson patented an obscure flotation process in which shale from a bituminous coal was floated while the depressed clean coal was left as tailings. Eveson tested a wide range of quaternary amines (some 8 different samples plus their mixtures) as shale collectors, together with several strong inorganic oxidants (permanganates, hypochlorites, persulfates and dichromates) as coal depressants. It remains unclear, though, why only certain combinations of collectors and oxidants were effective as Eveson did not provide any discussion reporting only "optimum results". In the work of Stonestreet and Franzidis (1988, 1989, 1992), coal depression and mineral matter flotation were supposedly achieved using only quaternary amines. Again, however, the mode of action of such amines was not elucidated. Although they reported that a hydrophobic bituminous coal was used throughout the work, the yields of their conventional "forward" flotation runs with M I B C did not exceed 10% which indicates that the coal was actually of a low rank.  70  Apparently coal reverse flotation should be able to solve the general problem with the processing of thermal coal fines since it is the mineral matter, and not the coal, that is floated in this process. Also, the reverse flotation reject containing the clean coal and some coal flotation depressants, could be ready-to-use and not require any extra treatment for C W F production. This work is based on the assumption that coal flotation depressants and C W F dispersants belong to the same group of agents, i.e. they have the same effect on coal surface properties. Such a flotation process would constitute the first stage of C W F preparation and would thus simplify the whole seam-to-steam strategy.  71  3.3  CHEMISTRY AND MODE OF ACTION OF AMINES  Amines have found several applications in different industries. Initially, they were used as sanitizing and antiseptic agents, as components in cosmetic formulations and as germicides and fungicides. More recent applications include the use of amines as antistatic agents, textile softeners, corrosion inhibitors, foam depressants, asphalt and petroleum additives, and flotation reagents.  3.3.1  Classification of Amines  Amines are organic compounds which contain a nitrogen atom and several (up to four) straight chain hydrocarbon or cyclic (aromatic) radicals attached to it. A l l amines can be looked at as organic derivatives of ammonia NH3. In the ammonia molecule, the nitrogen atom has four sp  hybrid orbitals directed to the corners of a  tetrahedron. Three of the four orbitals contain unpaired electrons that form bonds with hydrogen to produce ammonia. The fourth orbital is filled by a pair of unshared electrons as shown below (Figure 4) (Morrison and Boyd 1966). For the ideal tetrahedron the angles between H - N - H bonds should be 109.5 deg. However, in the case of ammonia the electron pair compresses the bond angles slightly to 107 deg. The unshared electron pair is a region of high electron density and is a source of electrons for electron seeking atoms and molecules, and thus gives ammonia its basic properties.  72  le  Figure 4. Geometric representation of ammonia molecule. Since amines are simply ammonia in which one or more of the hydrogen atoms are replaced by alkyl/aryl groups, the geometry around the nitrogen atom is also preserved with C-N-C bond angles of 108 deg (Morrison and Boyd 1966). Amines are classified as primary, secondary and tertiary according to the number of groups, R, attached to the nitrogen atom. H  Primary 1°  H  Secondary 2°  R,  Tertiary 3°  In all three cases, the unshared electron pair would be pointing outwards, perpendicular to the page plane. As mentioned earlier, this electron pair may serve as an attachment point for electron acceptors, or it may be used to form a fourth covalent bond. Since the nitrogen atom in amines is trivalent, the fourth bond would impart a positive charge to the nitrogen atom. This class of amines is sometimes referred to as quaternary amines although the term ammonium salts would be a more appropriate name, by analogy to N H / ammonium ion.  73  t  3  R:—N-—R, R  4  In mineral processing the primary and "quaternary" amines are mainly utilized. In the case o f industrial primary amines, the hydrocarbon chain length ranges from Cio to C i s . Three of the four radicals i n the ammonium salts are usually simple methyl (CH3-) groups while the fourth one can be as long as C20. This type o f ammonium salts is also known as alkyltrimethyl ammonium halides since frequently a halide ion is needed to neutralize the positive charge on the nitrogen atom:  H Cs^..|....:^ CH3 3  >  /  H,C  \  /  H C 2  Br" CH  \  CH  2  2  H C 2  \  CH,  /  H C 2  \ CH,  / H C 2  \ CH,  / H,C \  CH  3  Figure 5. Molecular structure of Dodecyl-Trimethyl Ammonium Bromide, [Cj2H sN(CH3)3j Br (DTAB) with the characteristic tetrahedral arrangement of the three methyl groups around the nitrogen atom. +  2  74  3.3.2  Chemistry of Amine Solutions  The unshared electron pair on the nitrogen atom determines the behavior o f primary, secondary and tertiary amines i n aqueous solutions. Similarly to ammonia, amines are converted into their salts by mineral acids. The mechanism involves a simple coordination reaction o f the hydronium ion to the unshared electrons on the nitrogen atom:  NH  3  R-NH  + H 0  <-> N H  +  3  2  + H Q  + 4  +  <-> R - N H  +  3  H 0 2  + 3  + H 0  (53)  2  The above reversible reactions strongly depend on p H and can be characterized by their corresponding pK  a  values. In the case o f aliphatic amines, the values range from 10 to 11  (Morrison and B o y d 1966). For longer chain amines ( C i to Cn), the pK 2  a  does not vary  significantly with the chain length as shown i n Table 2 (Hoerr et al. 1943).  Table 2. Ionization constants for selected primary amines (C H2 +iNH2) Primary Amines Kb values for reactions Calculated pKa values n  given by Eq. 53  n  (pKa+pKb=U)  4.4 • 10"  4  10.64  «-dodecyl amine ( C i )  4.3 • 10"  4  10.63  «-tetradecyl amine ( C M )  4.2 • 10"  4  10.62  n-hexadecyl amine (Ci6)  4.1 • 10"  4  10.61  n-octadecyl amine (Cis)  4.0 • 10"  4  10.60  w-decyl amine (Cio) 2  However, in the case o f primary amines, the solubility o f the neutral species R N H in 2  water is very low and quickly decreases with the hydrocarbon chain length. A s soon as the solubility limit is exceeded, a colloidal amine precipitate appears i n solution.  75  Table 3. Solubilities of selected primary amines (Apian and Fuerstenau 1962). Amine Solubility, S, of the neutral species [mol/dm ] 3  n-decyl amine (Cio)  5 • 10-  w-dodecyl amine (C12)  2 • 10"  w-tetradecyl amine (C14)  1 • lO"  4  5  6  Leja (1983) pointed out that other, significantly different solubility values were also reported and that the differences could be attributed to reagent purity, uniformity with respect to the chain length, variable pH (dissociation state) and/or the extent of equilibrium reached. The above values are most frequently quoted and are generally accepted. The pH of solid amine precipitation, pH ,,  can be calculated for any amine from the  precip  respective pK values and solubilities S. For equation 53, K can be expressed by: a  a  _ [ R N H  K  2  ] . [ H  +  ]  (54)  [RNH ] +  3  Assuming that Crotai,  and  [RNH ] 2  [RNH ] +  3  are the only species in solution, i.e.  the above constant can be rearranged with respect to  [RNH ]  =  2  Ka  Since the precipitate starts to appear only when  g _ CTotal ' K  a  [H ] + K +  2  [RNH ] +  2  3  to yield:  (55)  -^f_ ' l  [RNH ]  [RNH ]+  [RNH ] 2  > S, then at the onset of precipitation:  (56)  a  and thus:  76  [H ] = K +  a  ^ L - l j  (57)  which for a dodecylamine concentration of 10" mol/dm gives p H 4  3  pre  cip.  = 10.03.  Above the solubility limit, the neutral species concentration remains constant while the ionized form, RNF£ , achieves equilibrium with the colloidal precipitate through the +  3  following dissolution reaction (Smith and Akhtar 1976):  RNH (p ecip.) 2  r  + H 0 <-> R N H 2  + 3  + OH"  (5 8)  and the system can be characterized by the solubility product, K = [RNH ]-[OH"] = 8.6-10" +  so  9  3  for ^-dodecylamine (Smith and Akhtar 1976). For such weak electrolytes the equilibria given by the above reactions can be represented by a species distribution diagram (Smith 1963, Smith and Akhtar 1976, M . C . Fuerstenau  1982). For a total w-dodecylamine concentration of 10" mol/dm , the 4  3  concentrations of the individual species are given by the following equations: Crotai = [RNH ] . 2  Below  + [RNH ]  ag  2  pB. r ip.([RNH2]preci =  0):  r DATU +1 — CTotal '[H  1  P  ec  P  .  precip  + [RNH ] +  3  =  10"  4  Above the pH,precip.'  77  (61)  [RNH ] =S 2  aq  [RNH ]= ' +  (62)  Kso [H+]  3  10 -14  The amount of precipitate must also be taken into account: K -[H ]  (63)  +  [RNH ] 2  - C  precip  -S-  Total  S0  10 -14  The species distribution diagram for n-dodecylamine can be constructed as shown in Figure 6:  i i i i | i i i i | i i r i | i i i i | i i I I | i i i i J i i r i | i i i i | i i i i | i i II  4  5  6  7  8  9 pH  10  11  12  13  14  Figure 6. Species distribution diagram for a total dodecylamine concentration of 10~ mol/dm at 25°C.  4  3  The diagram indicates that the amine is 50% ionized ([RNH ] =[RNH ]) +  2  aq  3  at pH =  pK . Below the pK , the ionic form dominates, while the neutral amine is the dominant a  a  species above the pK„.  78  A somewhat more practical way of representing the equilibria in aqueous solutions of amines was proposed by Laskowski and his coworkers. Equation 56 can be transformed with respect to  C taiTo  [H ] + K +  (64)  a  ^Total -  $  which, when plotted as log(Cr to/) 0  =  f(pH), gives the domain diagram (Castro et al. 1986,  Laskowski 1994a, Laskowski 1999a). For dodecylamine the domain diagram assumes the form shown in Figure 7: •  •  I •  I . I . I .  Micelles  T3  |  0.01-=1  Colloidal Precipitate  a o '-*-> a § c o  0.001 -d  u  True Solution [RNH ]+[RNH ] +  (D  3  a  2  a  % o.oooi H o H  Solubility Limit for [RNH ] 2  1E-005-  ' J ' I ' I ' I ' I ' I ' I ' I ' 6  7  8  9  10  11  12  13  14  pH Figure 7. Domain diagram for dodecylamine.  The domain diagram allows the calculation of the pH of precipitation for any total amine concentration.  79  Surfactant molecules in solution exhibit some peculiar behavior due to the presence of the hydrophobic hydrocarbon chain and the hydrophilic head group. The hydrocarbon chain is non-ionic and a mutual incompatibility exists between polar water molecules and non-polar hydrocarbon chains. In aqueous solutions, such surfactant molecules are "expelled" from the aqueous phase and preferentially adsorb at the air-water interface with their hydrophilic groups pointing towards the aqueous phase and the hydrocarbon chains oriented towards the liquid surface. Such oriented adsorption dramatically decreases the liquid surface tension and the extent of this effect is given by the Gibbs equation:  r  =- ^ - ^ mRT d\nc  s  (65)  where r is the surface excess concentration of the surfactant, R is the gas constant, T is the s  temperature, y is the liquid (water) surface tension and c is the surfactant concentration. The parameter m is called the "salt" parameter, to account for the presence of an electrolyte, which is given by:  m = l+ s  where C  s  and  (66)  C s +  '-'salt  C it sa  are the concentrations of the surfactant  and the salt/electrolyte,  respectively. A typical y= f(log c) relationship is shown in Figure 8. As Figure 8 and the Gibbs isotherm indicate, the surface tension will continuously decrease as long as the surfactant adsorbs at the water/air interface: r > 0 when dyldln c < 0. s  However, as the graph illustrates, there is a concentration above which the surface tension does not change any further. 80  log(Surfactant Concentration) Figure 8. Typical experimental relationship between the solution surface tension and the concentration of a surfactant.  This phenomenon results from the simple physical fact that there must be a saturation concentration above which the surfactant adsorption density does not increase any further at the water/air interface. This limiting concentration is referred to as the critical micelle concentration (cmc) and is a characteristic property of a given surfactant. At surfactant concentrations slightly higher than the cmc, the surfactant molecules form the so-called micelles, multimolecular aggregates, in which the hydrocarbon chains point towards the inside of the spherical micelles. At much higher concentrations, the micelles change their shape to cylindrical or even lamellar and the transition concentrations between the different forms are often termed 2nd and 3rd cmc's. For a given surfactant, the cmc can be regarded as a measure of the stability of its micellar form relative to its monomeric form (Zielifiski et al. 1989). Surfactants with longer hydrocarbon chains should therefore form micelles more readily than the ones with shorter  81  chains since the more hydrophobic molecules would be more strongly "expelled" from the water phase. Generally, the size and shape of the micelles are determined by the equilibrium between the attractive hydrophobic forces among the hydrocarbon chains and the repulsive forces between the often charged head groups of the surfactant ions (Debye 1949, Stigter 1964, 1967). Using this DLVO-type of reasoning, for a homologous series of «-alkyl surfactants, Shinoda et al. (1963) derived the following relationship:  c c)  H  m  =-^^--K lnC kT s ' g  i  +  K  <) 67  l  '  where n is the number of carbon atoms in the hydrocarbon chain, co is the change in cohesive energy per one methylene (-CH -) group on transfer from a hydrocarbon environment of a 2  micelle to an aqueous medium (usually of the order of 1.08AT at 25°C), K is a constant g  related to the work of introducing a charge onto the micelle surface (K = 0.4-0.6), C, is the g  total ionic strength, and Kt is an experimental constant. As discussed by Lin and Somasundaran (1971) the value of l.OSkT for co is common for all surfactants since it expresses the free energy change due to the ever-present van der Waals attraction. It does not include, however, the electrostatic component of the transfer energy. Hence, for ionic surfactants the total energy is actually of the order of 0.61-0.69AT as the electrostatic repulsive forces between the charged groups oppose micelle formation. Shinoda et al. (1963) also proposed a simplified empirical formula for a given homologous series:  log(cmc) = A-Bn  (68) 82  where A and B are constants for a given series. Indirectly, Shinoda's expressions imply that the longer chain surfactants are more effective in decreasing the surface tension than the shorter chain homologues at the same concentration. Table 4 shows the critical micelle concentrations for quaternary (alkyltrimefhyl ammonium bromides) and primary amines. Table 4. Critical micelle concentrations for selected primary and quaternary amines. Quaternary Amines, C H2n+iN(CH3) Br Primary Amines, C H2„+iNH2 Temperature Temperature Chain Length cmc [mol/dm ] cmc [mol/dm ] °C °C 25 60 Cio 4 • 10" 6.63 • 10" 25 60 C 1.3 • 10" 1.46 • 10" 25 60 3.72 • 10" 4 • 10" Cl4 25 60 Ci6 8 • 10" 0.96 • 10" References: primary amines - Ralston (1948), quaternary amines - Zielinski et al. (1989) n  n  3  3  2  2  2  2  l2  3  4  3  3  It can be seen from Table 4 that the cmc's for primary amines were determined at rather high temperatures. Ionic surfactants often show a strong dependence of their solubilities on temperature. It was shown by Eggenberger and Harwood (1951) that the solubility of dodecylamine remains low until a temperature of about 23°C is reached above which the amount of dissolved amine increased very rapidly. This critical temperature is called the Krafft point. At least two different approaches to the phenomenon of the increased solubility were proposed, as discussed by Anacker (1970). In the first approach (the phase separation model), the Krafft point is viewed as the temperature at which solid hydrated surfactant, micelles, and a solution saturated with unaggregated surfactant are in equilibrium. In the second view (the mass action model), micelles and individual surfactant molecules are considered to be in association-dissociation equilibrium so the mass action law can be applied. Since the solubility of micelles is greater than the solubility of the monomeric form, the solubility of the surfactant as a whole will not increase markedly with temperature until it 83  attains the cmc region. Thus, the Kraft point is the temperature at which the surfactant solubility is equal to the cmc. According to Moroi et al. (1984), the Krafft temperature can be defined as the temperature at which the solubility of surfactant molecules as monomers becomes high enough for the individual molecules to start aggregation/micellization. The mass action model, according to Moroi et al. (1984), is much more accurate than the phase separation model in characterizing a surfactant behavior above the Krafft point. Following the mass action model, it can be concluded that below the Krafft point, the amount of surfactant molecules is not sufficient to form micelles and hence it has been widely accepted that the cmc's of ionic surfactants can not be measured at temperatures below the Krafft point (Leja 1982, Smith 1988, Smith and Scott 1990, Dai and Laskowski 1991). For primary amines, the Krafft point is a function of the hydrocarbon chain length. It changes from 23°C (Eggenberger and Harwood 1951), or 26°C (Gerrens and Hirsch 1975) for dodecylamine, to 56°C for octadecylamine (Gerrens and Hirsch 1975). Because of the relatively high Krafft temperatures, the cmc values reported by Ralston (1948) were all measured at the temperatures higher than the Krafft point of the "longest" amine tested. There  is some  significant difference  between  the  Krafft  temperatures for  dodecylamine cited above and the data of Dai and Laskowski (1991) who also studied the effect of pH. Dai and Laskowski determined the Krafft point for dodecylamine to be 17.5°C at pH = 5. They also showed that the Krafft point changes with pH from 17.5°C at pH = 5 to only 7°C at pH = 9. This effect was also mirrored by a decrease of the cmc with pH. At pH 10, the solubility of the cationic species was so low that micellization could not be detected even at temperatures as high as 55°C. Above the Kraft point and for a given pH, the cmc for dodecylamine did not significantly change with temperature although a shallow minimum  84  could be observed. Similar results on the effect of pH on the cmc were reported by Watson and Manser (1968). Typical solubility-cmc-Krafft point diagram is shown in Figure 9. When the cmc data of Dai and Laskowski (1991) and of Watson and Manser (1968) are plotted as a function of pH on the domain diagram, as shown in Figure 10, another important observation can be made.  c.m.c. Krafft y Temperature  Temperature Figure 9. A general phase diagram for a surfactant near its Krafft point.  As pointed out by Laskowski (1994a, 1999a), micellization and precipitation may be mistaken for each other, especially at alkaline pH values and higher amine concentrations. The domain diagram (Figure 10) clearly illustrates the problem as some of the data of Watson and Manser fall well within the precipitate domain. One may argue, however, that when large colloidal micelles are formed (these are known to exist at concentrations greatly higher than the cmc), the process of micellization seems to be synonymous with precipitation. In the case of alkyl-trimethyl ammonium halides, their solution chemistry is dramatically different. Since the nitrogen atom in this case does not possess the unshared 85  electron pair (used for the fourth covalent bond between the nitrogen atom and the alkyl group), the quaternary amines are strong electrolytes, fully dissociated in water regardless of pH and concentration.  5  6  7  8  9  10  11  12  13  14  pH Figure 10. Experimental cmc vs pH data superimposed on the domain diagram.  For these compounds, the domain or species distribution diagrams do not make physical sense as only one ionic species, C H2 +iN(CH3)3 , is present in true solution over the +  n  n  entire p H range. The solubilities of selected quaternary amines at various temperatures are shown in Table 5 (Shapiro 1968). Table 5. Solubilities of Alkyl-Trimethyl-Ammonium Chlorides in water (g of amine per 100 g of water) , Ci H N(CH )3Cl C , 4 H N ( C H ) C 1 C H33N(CH )3C1 C H 7N(CH )3C1 2  o°c  25  3  40 °C 60 °C o ° c  29  3  3  16  40 °C 60 °C o ° c  76.1g 68.1g 68.1g 56.7g 60.8g 63.3g  -  3  18  40 °C 60 °C o ° c 53.1g 55.1g  -  3  3  40 °C 60 °C 42.1g 44.5g  86  The lack of data at 0°C for C\e and Cig homologues results from the fact that their Krafft points were estimated at 25°C (Adam and Pankhurst 1946) and «37°C (Reck and Harwood 1953), respectively. Interestingly, the solubilities of shorter chain quaternary amines are very high, even at 0°C (the Krafft points can not be defined), and do not change markedly with temperature. The cmc values for quaternary amines of varying hydrocarbon chain length (Cs-Ci6) at a given temperature also follow the linear relationship proposed by Shinoda (see Equation 67) as shown by Zielinski et al. (1989), Evans et al. (1984) and Bashford and Woolley (1985). The temperature dependence of the cmc is similar to that of primary amines: there is a clear minimum value of the cmc for all quaternary amines. The temperature at the minimum cmc systematically decreases as the alkyl chain length increases (Zielinski et al. 1989). Based on the analysis of the solution chemistries of the two classes of amines, it can be concluded that quaternary amines are more convenient reagents for studying flotation phenomena. The effect of pH can be entirely neglected whereas their high solubilities make the preparation of solutions quite straightforward.  3.3.3  Adsorption of Amines at Mineral/Water Interface.  A l l amines belong to the class of cationic surfactants. In mineral processing, they are almost exclusively used as flotation collectors for oxides, oxidized metal sulfide ores, silicates and sylvite. Probably the most frequently studied model system is that of silica and aliphatic, primary amines. Silica (SiCh) is a typical hydrophilic solid whose surface charge strongly depends on the pH of solution. Fuerstenau (1970), based on a review of the available 87  literature, reported that the point of zero charge for silica is located in the acidic pH range, anywhere between 1.3 and 3.7. Such a low p.z.c. value indicates that the silica surface is negatively charged over the entire practical pH range. This creates ideal conditions for the adsorption of cationic agents (amines) on the silica surface. The early systematic work of Fuerstenau and his coworkers on the flotation of silica with primary amines indicates that the cationic species, RNF£3 , adsorbs on the silica surface +  through the electrostatic attraction between the positively charged amine group and the negatively charged sites on the silica surface (Gaudin and Fuerstenau 1955, Fuerstenau 1956, 1957). The adsorption mechanism involves a simple ion exchange in which the hydrogen ions from the surface (silanol) - O H groups are exchanged for RNH3 cations from solution. When +  measured as a function of pH, the maximum flotation recovery coincided with the maximum amine adsorption density. The maximum surface coverage by the amine also produced highly hydrophobic surfaces, as determined from the contact angle measurements, suggesting that the alkyl chains of the adsorbed amine molecules point away from the surface. These basic observations led to the formulation of the hemi-micelle hypothesis (Gaudin and Fuerstenau 1955, Fuerstenau et al. 1964, Somasundaran et al. 1964). According to this approach, in dilute amine solutions, the surfactant ions weakly adsorb as individual molecules. At higher adsorption densities, though, the amine concentration at the silica-solution interface becomes comparable with the bulk critical micelle concentration and the surface amine aggregates may eventually form due to the lateral hydrophobic interactions between the hydrocarbon chains. Gaudin and Fuerstenau (1955) termed these surface aggregates hemi-micelles. As the formation of the hemi-micelles progresses at the interface, the zeta potential starts to reverse charge and both the amine adsorption density and flotation recovery sharply increase. A l l these phenomena are shown below in Figure 11.  88  CONCENTRATION OF DA A, MOLE/LITER  Figure 11. Correlation diagram of contact angle, adsorption density, flotation response, and zeta potential for quartz as a function of dodecylammonium acetate concentration at pH 6-7, 20-25°C (Fuerstenau et al. 1964) Similar experimental findings for the silica/primary amine system were reported by de Bruyn (1955) and L i and de Bruyn (1966). Somasundaran et al. (1964) and Fuerstenau et al. (1964) also showed that longer chain amines required much lower dosages than the shorter homologues for complete silica recovery by flotation. Modi and Fuerstenau (1960) studied the flotation of alumina with the use of both anionic and cationic surfactants. They showed that primary amines were effective as collectors only when the alumina surface was negatively charged, i.e. in the pH range higher than 9.4 (the p.z.c. for alumina). In highly alkaline solutions the alumina flotation recovery dropped to zero which was attributed to a rapid decrease in the concentration of the cationic species. Interestingly, Modi and Fuerstenau also noticed that quaternary amines did not act as effectively as primary amines, indicating that the presence of neutral amine molecules in solution plays a role in the collection mechanism. The studies of Smith (1963, 1973) and Smith and Lai (1966) showed convincingly that the coadsorption of the neutral species on the silica surface is an important factor. As  89  pointed out by Smith (1963), dense adsorption of the cationic species would be unfavorable because of the electrostatic repulsion between two adjacently adsorbed amine groups, unless some sort of charge screening takes place. Also, the concentration of the cationic species at pH = 9-10 - the optimum pH for the highest flotation recovery - is simply too low to bring about the observed large contact angles. Using dodecanol (C12H25OH) as a "screening" additive in combination with dodecylamine, Smith measured contact angles of about 75 deg at pH values as low as 5. Under the same pH, dodecylamine alone gave contact angles of only 10-30 deg. Smith and his coworkers explained that the nonionic molecules are "captured" by the hemi-micelles formed from the ionic (RNH3 ) species. They also suggested that as pH +  increases to highly alkaline regions (>11) the concentration of the charged species is too low to form hemi-micelles and thus to hold as many amine molecules. Watson and Manser (1968) and Manser (1975) carried out systematic vacuum flotation tests with the use of primary amines on a wide range of silicates including orthosilicates, pyroxenes, amphiboles as well as sheet and framework silicates. Although the optimum flotation conditions (amine dosage and pH) were case-specific, the two researchers observed that all the silicates showed a common "alkaline flotation limit". Detailed research of Laskowski and his group provided experimental evidence that the sudden drop in the flotation recovery of silica and silicates in the alkaline range can be attributed entirely to the formation of a colloidal amine precipitate whenever the solubility of the amine is exceeded (Castro et al. 1986, Castro and Laskowski 1988, Laskowski 1994a, 1999a). Using the amine domain diagram (Figure 7), Smith and Scott (1990) estimated that the concentration of the neutral amine at the silica/solution interface may actually exceed the solubility limit at pH values much lower than those of bulk precipitation - analogously to the formation of hemimicelles. For example, at a bulk dodecylamine concentration of 4 • 10" mol/dm , precipitation 90  occurs at pH = 10.85, while surface precipitation may already take place at pH — 8.2. This observation suggests that the "capture" mechanism for the coadsorption of the neutral amine proposed earlier by Smith (1963) is actually the surface precipitation of the colloidal amine. According to Smith and Scott (1990), the adsorption mechanism of amines on silica and silicates also depends on temperature. Below the Krafft point, there is a great increase in both contact angle and flotation when pH is increased to the surface precipitation limits. Silica hydrophobicity and flotation will increase even further until the conditions for bulk precipitation are reached. Below the Krafft point, the amine concentration is too low for either surface or bulk micellization. Above the Krafft point, the hydrophobicity and flotation of silica will rise until a pH is reached at which the surface cmc is exceeded locally at the quartz (silica) surface due to the adsorption of amine ions. This moment corresponds to the formation of hemi-micelles. A further increase of pH would bring about bulk amine micellization. Under these conditions, a second layer of amine ions would adsorb onto the hydrophobic hemi-micelles through hydrophobic interactions between the hydrocarbon chains. The second layer of the outwards-oriented amine molecules would render the surface hydrophilic and dramatically depress silica flotation. It is widely accepted that the adsorption isotherms of ionic surfactants  on  homogenous, charged surfaces can be divided into three, sometimes four, distinct regions as shown schematically in Figure 12 below (Fuerstenau 1970, Fan et al. 1997, Somasundaran and Krishnakumar 1994, Wangnerud et al. 1995). In Region I amine ions adsorb as individual molecules, most likely through electrostatic attraction. This process is often referred to as "ion-exchange" (Rosen 1989) since the surface silanol groups (Si-OH) exchange their hydrogen cations for the amine cationic  91  species. The weak hydrogen bonding between the surface silanol groups and the ammonium head group was also postulated (Rao et al. 2001).  Region I  II  III  Region IV  cmc  log(Equilibrium Concentration [mol/dm ]) 3  >  Figure 12. Schematic adsorption isotherm of an ionic surfactant on a charged homogeneous surface. In Region II, the already adsorbed amine molecules reach some critical concentration and the formation of aggregates at the surface is initiated. These aggregates may be termed hemi-micelles (Gaudin and Fuerstenau 1955). Surfactant molecules adsorbed at this stage produce most hydrophobic surfaces, easily amenable to flotation. Adsorption in Region III occurs through the growth of the aggregates already formed in Region II. Some bilayer adsorption may already take place (Koopal et al. 1999) and these tail-to-tail associations are sometimes called admicelles (Scamehorn et al. 1982, Harwell et al. 1985a,b). Finally, in Region IV, a plateau is reached near the bulk cmc (Tamamushi and Tamaki 1957, Connor and Ottewill 1971). The adsorbed surfactant molecules form a uniform bilayer that renders the  92  surface completely hydrophilic (Elton 1957, Ralston and Kitchener 1975, Dobias 1986, Parachuri et al. 2001, Fuerstenau 2001). Since the cmc and the solubility of amines are a function of pH, the amine concentration ranges, as well as the adsorption modes within the four regions, also strongly depend on pH. Claesson et al. (1992) showed through surface force measurements that the formation of a hydrophilic bilayer may be induced either by increasing the amine concentration or by increasing the pH. At pH values higher than 10.3, Claesson et al. also observed that hydrated amine droplets formed as a separate phase which resulted in a highly chaotic, poorly defined structure at the silica/solution interface. This observation is in excellent agreement with the findings of Laskowski and his group on colloidal amine precipitation. At p H =10, the bilayer forms at the dodecylamine concentration of only 10"  4  mol/dm while some 60 times higher amine concentrations are needed to produce the bilayer at pH 5. In the former case, the outer layer consists mainly of neutral amine molecules. In more acidic environments, the cationic form is the main constituent. Quite often region III is not clearly defined, while region II may appear as an almost vertical jump on a log-log graph. Such two-step isotherms were also reported for the adsorption of alkyltrimethyl ammonium bromides on alumina at p H =10 (Fan et al. 1997). In the case of heterogeneous surfaces, the adsorption isotherm can be split into a series of steps corresponding to successive adsorption on the individual surface components. Such stepwise isotherms of primary amines on natural and synthetic calcite and on phosphate oolites were reported by Cases et al. (1975) who later developed the condensation theory to describe surfactant adsorption on such complex substrates (Cases and V i l l i eras 1992). Several studies dealt with the adsorption of cationic surfactants on hydrophobic surfaces. Elton (1957) studied the adsorption of quaternary amines on octadecane (one of the 93  model solid hydrocarbons) through contact angle measurements. He showed that the contact angle systematically decreased with increasing concentration of the ammonium salts reaching zero at the corresponding bulk cmc's. This trend suggests that cationic surfactants adsorb primarily through hydrophobic interactions of the hydrocarbon chains with the hydrophobic hydrocarbon surface. Similar conclusions were drawn by Parachuri et al. (2001), who found that the contact angle on a graphite surface continuously decreased to zero in the presence of H T A B (hexadecyl-trimethyl ammonium bromide) at concentrations near its cmc. The results of Elton (1957) and Parachuri et al. (2001) are shown in Figure 13 below.  Figure 13. Effect of quaternary amines (HTAB and DTAB) on the wettability of hydrophobic (graphite and octadecane) and hydrophilic, charged solids (mica, silica). Plot A is taken from Parachuri et al. (2001) while plot B is based on tabulated results of Elton (1957). Vertical, dashed lines indicate the critical micelle concentrations for the two surfactants.  Laskowski  and  Konieczny (1970)  found  that  the  adsorption  density  of  hexadecyltrimethyl ammonium bromide (HTAB) on a hydrophobic coal decreased with coal oxidation time. This finding implies that H T A B adsorbs through hydrophobic interactions on the coal surface. As oxidation progresses, the number of hydrophobic sites declines rapidly leading to the observed drop of amine adsorption. Similar findings were reported by Bennett  94  and Abram (1967) who used the magnitude of H T A B adsorption as a measure of bone char quality (used as a decolorization adsorbent in sugar refining). Bennett and Abram postulated that H T A B adsorbs only on the hydrophobic, active components of bone char. Their results correlated very well with the carbon content, and the relative surface areas calculated from the adsorption isotherms agreed with the specific surface areas obtained from the BET method. According to Latiff Ayub et al. (1985a,b), quaternary amines can interact with heterogeneous coal surfaces through both electrostatic and hydrophobic interactions. Latiff Ayub et al. performed systematic adsorption and electrokinetic measurements on a highvolatile bituminous coal using several quaternary amines. They determined that even below the i.e.p. of the coal, the cationic surfactants were still able to change the zeta potential towards more positive values. Such a result is consistent with adsorption taking place through hydrophobic attraction. The outwards-oriented charged groups would strongly increase the positive charge of the coal surface. On the other hand, as p H is gradually increased above the i.e.p. towards more alkaline values, the amine adsorption density also increases indicating that when the coal surface is negatively charged, adsorption occurs due to electrostatic attraction. Interestingly, Latiff Ayub et al. also observed that the adsorption isotherms did not exhibit any characteristic regions. This demonstrates that the adsorption of amines at the coal/water interface does not fit the traditional models. Other effects of amine adsorption on the solid surface properties were already discussed in Sections and 3.2.5.  95  4 EXPERIMENTAL PROGRAM 4.1 REAGENTS 4.1.1  Dodecyl-Trimethyl Ammonium Bromide (DTAB)  D T A B was obtained from Acros Organics (distributed by Fisher Scientific). The sample was 99% pure as reported by the manufacturer. The sample was used without any purification. Tap water was used for the preparation of stock solutions and fresh solutions of D T A B were prepared every 3 days. Stock solutions of practically any D T A B concentration could be easily prepared thanks to the exceptional solubility of this amine. D T A B was investigated as a potential coal depressant and mineral matter collector in the reverse flotation process.  4.1.2  Humic Acids  The sodium salt of humic acids of technical grade was provided by Aldrich Chemical Company. In a preliminary test, 5 grams of the salt were dissolved in 1 dm of distilled water 3  and the solution was filtered on a 0.45-pm Millipore filter. The residue on the filter (mainly coarse, sand-like, hard particles) was dried and weighed. The amount of water-insoluble material was found to be 1.0%. Since the content of water-insolubles was rather low, the humic acids were subsequently used throughout the tests without taking any corrections. Typically, 1 g/dm stock solutions in tap water were prepared daily (pH = 9-9.1). Humic acids 3  were used as coal depressants and coal dispersants.  96  4.1.3  Polymers and Other Reagents  Several polymers were utilized in rheological and flotation experiments as CWS dispersants and coal depressants. Dextran, carboxymethyl cellulose and hydroxyethyl cellulose samples were obtained from Polysciences and were used without any purification. A dextrin sample was obtained from A.E. Staley Manufacturing Company. The trade name of the sample was Tapioca Dextrin 12. Using gel permeation chromatography,  Nyamekye  (1992) found the molecular weight of this dextrin to be 56,000. A l l the polysaccharides were water-soluble and suitable stock solutions were prepared daily to avoid their biodegradation. A sample of polystyrene sulfonate of very high purity was kindly provided by Dr. Junichi Yamada of L I O N Corporation, Japan. The polymer was supplied as a 37% (wt) aqueous solution. The sample came from a batch that was commercially used in Japan as a coal dispersant in the preparation of Coal-Water Fuels. Some properties of the tested polymers are listed below in Table 6. Their structural formulae are shown in Appendices 11 and 12. Table 6. Molecular weight and electric charge of the tested polymers. Polymer Molecular Weight Electric Charge Humic Acids  Low  Strongly Anionic  Dextrin  56,000  Nonionic  Dextran  15,000-20,000  Nonionic  Hydroxyethyl Cellulose  24,000-27,000  Nonionic  Carboxymethyl Cellulose  80,000  Strongly Anionic  Polystyrene Sulfonate  14,000  Strongly Anionic  Hydrochloric acid and potassium hydroxide were used to adjust pH when necessary. Potassium chloride was used to control the ionic strength only in electrokinetic tests. A l l other  97  experiments were carried out in tap water (pH = 5.5-6.2) to reflect natural processing conditions. Methyl-isobutyl carbinol (MIBC) was also used in some flotation tests as a frother. This reagent was added as a 1 g/dm stock solution in tap water.  4.2 COAL AND SILICA SAMPLES 4.2.1  Silica  Fine silica, trade name SilcoSil 395 obtained from Ottawa Silica Company (Illinois, USA), was particularly useful for adsorption and electrokinetic tests due to its relatively high specific surface area (see section 4.3.3 for details). The silica was used as received. Since the particle size distribution was approximately described by the supplier as "95% passing 325mesh (44 pm)", the actual size analysis was performed using an E L Z O N E 280 PC (Micromeritics) sizing and counting system. The analysis was carried out using 1% and 4% NaCl solutions as the conducting media. The results are presented in Figure 14 (section 4.2.2), together with the particle size distributions of the coals, and tabulated in Appendix 8.  4.2.2  Coals  Three coal samples of different ranks were selected for this study. A high rank, metallurgical coal was obtained from the Fording Mine (seam 4) in South-eastern British Columbia. A lower rank coal sample was provided by Luscar Stereo Ltd. from the Coal Valley Mine (pit 43) in Alberta (foothills of the Rocky Mountains). Some rheological experiments were carried out with the use of a Fording oxidized coal (seam 13). This coal, 98  however, had a high bulk ash content of 30.5% and as such would be of little use for any type of fundamental study. Therefore, larger hand-picked pieces of F13 coal were utilized, but the amount collected was about 5 kg which was enough only for a limited number of experiments. The ash content in this "clean" F13 coal batch was 14.8%. Proximate analyses of the coals are given in Table 7. The analyses were carried out following A S T M Standards 3172 through 3175, as briefly described in section 4.3.1. According to the A S T M Coal Classification System, F4 coal can be classified as a medium-volatile bituminous coal. LS43 is a sub-bituminous coal, while F13 is an oxidized metallurgical coal. The calorific values for F4 and LS43 coals were found to be 33,100 kJ/kg and 26,000 kJ/kg, respectively. The high moisture content for LS43 coal compared to that of Fording 4 coal also indicates that LS43 is indeed of low rank. Table 7. Proximate analyses of the tested coals Coal Moisture Ash Volatile  Fording 4 (F4) Luscar Stereo 43 (LS43) Fording 13 (F13)  Volatile Matter,  Fixed Carbon,  [%]  [%]  Matter [%]  d.m.m.f [%]  d.m.m.f. [%]  0.6  7.7  23.4  25.5  74.5  5.6  14.4  31.9  39.9  60.1  2.0  14.8  29.2  35.1  64.9  F4 and F13 coals have been used in many studies in this Department. The total acidity of these samples after demineralization was 0.04 meq/g for F4 coal and 1.22 meq/g for F13 coal (Laskowski et al. 1992). The equilibrium moisture content (determined at room temperature after equilibration over a saturated K2SO4 solution for 72 hours) was found to be 0.94%) and 4.59% for F4 and F13 coals, respectively. The higher acidity and moisture content  99  for F13 coal indicate that this sample was oxidized and should be fairly hydrophilic compared to F4 coal. Large pieces, 10-cm or larger, of the coals were crushed and dry-pulverized below 212 microns (65 mesh). After homogenization through five "cone-and-quartering" steps, 50-kg batches of the coals were representatively split into five portions that were then sealed in separate polyethylene bags and stored in a refrigerator. For a short term, the ground coals were kept in sealed buckets under nitrogen.  Particle Size [microns] Figure 14. Particle size distributions of the tested coals (wet screening) and silica (Elzone apparatus)  Three representative 200-gram samples of the coals were wet-screened on a series of sieves with apertures ranging from 212 (65 mesh) to 38 microns (400 mesh) to obtain the particle size distribution. The averaged numerical results are shown in Figure 14, and tabulated in Appendix 6. The experimental error (standard deviation) ranged from 0.3% for coarser fractions to 1.2% for finer ones.  100  Since only a limited quantity of F13 coal was available, once the particle size distribution of F4 coal was determined, the size distribution for F13 coal was adjusted to the same proportions as those of F4 by regrinding and blending to avoid any material losses. This was critical for rheological studies as the particle size distribution has a profound effect on the rheology of concentrated solid-liquid suspensions. As Figure 14 indicates, the coal particle size distributions did not follow the GGS distribution function. These particle sizes could, however, be linearized using the R R B function (see Appendix 7). For F4 (and F13) coal, the size modulus, ^3.2, and the distribution modulus, m, were found to be 59.8 pm and 1.41, respectively. These two moduli for LS43 coal were d^.i = 71.5 urn and m = 1.38. F4 and LS43 coals were used throughout all flotation, adsorption and electrokinetic experiments without any additional grinding, screening or cleaning . B E T specific surface areas were also determined for the two coals as described in 1  section 4.3.3.  4.2.3  Coal Surface Characterization - FT-IR Study  In infrared spectroscopy, the presence of various functional groups on a coal surface is detected by recognizing the characteristic bands and peaks on the IR spectrum. These IR bands are assigned to deformations of chemical bonds in terms of bond stretches, bond angle bends, oscillations, etc. The qualitative analysis is based on correct band assignments to  Coal is a highly heterogeneous solid and grinding leads to the liberation of its mineral matter. Chemical/mineralogical analyses of the individual size fractions always show significant differences in composition. In most cases, clays accumulate in very fine fractions. Therefore, sub-sieve fractions may be mineralogically very different and do not have to represent the coal particle size distribution. Since the R R B distribution moduli (m and d63.2) were determined for the coal samples, the complete distribution can also be obtained from the R R B function, as it is commonly done in coal preparation.  101  functional groups. Table 8 shows the band assignments accepted in this work (from Painter et al. 1985). Table 8. Band assignments for the infrared spectra of coals. Oxygen-containing Functional Groups Aliphatic and Aromatic Groups Assignment Wavenumber Assignment Wavenumber [cm" ] [cm" ] Various - O H stretching Aromatic C-H stretch 3100-3000 3800-3200 modes 3000 - 2700 Aliphatic CH, -CH - and C=0, anhydride CH - stretches 1835 1  1  2  3  1600  Aromatic ring stretch  1490  Aromatic ring stretch  1450  -CH -, CH - bends  1375  CH - bend  2  1775-1765  C=0, ester with electronwithdrawing group attached to single-bonded oxygen, e.g. Aryl-O-CO-R  1735  C=0, ester  1690-1720  C=0 from ketones aldehydes and carboxylic acids  1700-1710  Carboxylic group in acidic form, i.e. COOH  1650-1630  C=0, highly conjugated, e.g. Aryl-CO-Aryl  1600  Highly conjugated, hydrogen bonded C=0  1560-1590  Carboxyl group in salt form, i.e. -COO"  1330-1100  C=0 stretch and OH bend in phenoxy structures and ethers  3  3  900 - 700  Aromatic C-H out-of-plane bending modes  The 3100-3000 cm" band for the aromatic C-H stretch often appears as a single peak 1  at «3030 cm" . The 3000-2700 band for the various aliphatic stretches can be resolved into at 1  least three separate peaks at 2864 cm" , 2923 cm and 2956 cm . The 1490 cm' and 1450 1  -1  -1  1  cm" peaks often appear as a single broad band, particularly for high rank coals and vitrinites 1  (Painter et al. 1985, Lowenhaupt and Gray 1989). Similarly, the 900-700 cm band for the C-1  102  H out-of-plane bending modes can be expanded into three distinct peaks at «750 cm" , «810 1  cm" and «870 cm" . 1  1  There is an ongoing discussion surrounding the assignment of the 1600 cm" peak. As 1  Table 8 shows, this peak can be assigned to two, very different groups - from the coal chemistry point of view. According to Painter and his coworkers, the peak originates from aromatic ring stretching, while other researchers assign it to the presence of the carbonyl, C=0, groups (Dryden 1963, Lowenhaupt and Gray 1989). Strong evidence to support both approaches has been presented but a more detailed discussion on this controversy is beyond the scope of this work. As discussed by Painter et al. (1985), the actual assignment should be made based on "experience, common sense and informed intuition". Based on the results of this thesis it is assumed that, in the case of F4 coal, the 1600 cm" peak can indeed be attributed to the aromatic ring stretches. 1  As far as the oxygen groups are concerned, it must be pointed out that the majority of deformations occurs in a relatively narrow wavenumber range, i.e. within the 1650-1800 cm"  1  region. Therefore, when all these oxygen groups are present, the individual peaks blend into a broad band stretching from 1650 cm" to almost 1900 cm" . Carboxylic groups, whose main 1  1  peak is located at «1700 cm" , also give a weak broad band at 3300-2500 cm" . 1  1  Infrared spectra were collected for F4, F4-oxidized and LS43 coals, as well as for solid humic acids (sodium salt). The results are shown in Figures 15 and 16. Figure 15 presents the spectra for F4 and F4-oxidized coals to illustrate the changes taking place as a result of oxidation. F4 coal was oxidized following a procedure described in section 4.3.2. The LS43 spectrum is shown together with the humic acids spectrum in Figure 16. The main bands and peaks are also marked in the figures.  103  b  o  b  b  o  o  '  b  ©  104  105  The infrared tests were carried out to qualitatively investigate the surface composition of the tested coals. FT-IR (Fourier Transform Infrared) techniques are also widely used to assess the degree of coal oxidation as well as the resulting changes in the coal surface composition. The measurements were carried out using a Perkin-Elmer System 2000 FT-IR spectrophotometer. The samples were prepared by mixing 0.05 g of coal (or HA) with 0.5 g of KC1 crystals. 50 scans were run to collect the IR-spectra in the wavenumber range from 400 to 4000 cm" . The spectral resolution was 0.2 cm" . 1  1  The IR spectrum of F4 coal is a textbook example of a high rank coal. The dominant features are the multiple peaks in the 700-900 cm" and 2700-3100 cm" frequency ranges. 1  1  The presence of the strong 3040 cm" peak, and the weak overtone at 2730 cm" in particular, 1  1  is an indication of an unusually high content of polyaromatic structures (Bellamy 1975). Based on this, the peak at 1600 cm" can be assigned to aromatic ring stretching. The high 1  aromaticity of F4 coal is also confirmed by the presence of the peaks corresponding to the aromatic -C-H- out of plane bending modes. Additionally, the coal appears to contain substantial amounts of aliphatic hydrocarbons as evidenced by the strong multiple peaks in the 2700-3000 cm" band, and the peak at 1450 cm" . 1  1  Some trace amounts of oxygen groups (-C=0- and -COOH) also seem to be present as indicated by the weak shoulders and knees on the 1600 cm" peak (1600-1800 cm" range). 1  1  Upon oxidation, the most obvious difference is the appearance of the strong and broad band stretching from 1650 to almost 1900 cm" . The maximum on the peak is at 1700 cm" 1  1  which is the fingerprint of carboxylic groups (R-COOH). It is important to observe that all the aromatic or aliphatic - C H = , - C H 2 - and -CH3 bands partially or completely vanished as a result of oxidation. A new broad band also 106  appeared in the 1100-1300 cm" region characteristic of various -C-O- (ethers) and -OH 1  (phenoxy) groups. This indicates that the methyl- and methylene groups are the primary sites/targets for oxygen. However, it seems that the aromatic skeleton of F4 coal was preserved since the intensity of the 1600 cm" peak did not change after oxidation. This, in 1  turn, implies that oxidation took place only superficially while the internal structure of F4 coal remained intact. As opposed to the spectrum of the high rank coal, the peaks for the aromatic and aliphatic -CH=, - C H 2 - and -CH3 groups of LS43 coal are very weak although still present. Since the frequencies for the aromatic groups are not strong, the 1600 cm" peak in this case 1  can be assigned to - C = 0 group enhanced by some weak aromatic ring stretching. Surprisingly, the carboxylic group bands are not clearly outlined although a small peak shows at 1700 cm" . 1  In the case of humic acids (sodium salt), the most unexpected feature is the lack of any bands of aromatic origin. The double aliphatic peak in the 2800-3000 cm" region is the only 1  clear evidence of the presence of long-chain hydrocarbons. However, the spectrum obviously exhibits a band at 1567 cm" which is typical for carboxylates. The presence of these 1  dissociated carboxylic groups would impart a strongly anionic character to the humic acids. Interestingly, this peak does not show on the spectrum of F4-oxidized coal indicating that the carboxylic groups on the surface of the oxidized coal are not neutralized and occur mainly in the acidic form.  107  4.3 EQUIPMENT AND METHODS 4.3.1  Proximate Analysis - Ash Content Determination  Proximate analysis is a standard method for testing coal ( A S T M D-3172). In the course of this analysis three quantities are directly determined: moisture, ash and volatile matter. Moisture ( A S T M D-3173) is determined by heating a 1-g fine coal sample (below 6065 mesh) at 105-110°C for one hour in an oven providing natural air circulation. After heating, the sample should be immediately placed in a desiccator and cooled down to room temperature. The dried coal should be weighed quickly to minimize the re-absorption of moisture from air. The moisture content is calculated from the difference in coal weight before and after heating. Volatile matter ( A S T M D-3175) is determined by establishing the loss of weight resulting from heating coal at 950°C for exactly 7 minutes in a platinum crucible without access to air. Some preheating at lower temperatures may be required i f highly volatile samples are analyzed. This was found to be the case with LS43 coal which produced the so called "sparking", or explosive devolatalization, when the standard procedure was followed. For this coal, 3 minutes of preheating at 600°C were required to drive off the most volatile fractions and prevent sparking. The ash content is obtained by weighing the residue remaining after burning coal under rigidly controlled conditions of sample weight, temperature, time and air circulation in a furnace as prescribed by A S T M ( A S T M D-3174). It is recommended that the sample be slowly preheated from room temperature to 500°C during the first hour. The heating rate  108  should be such that a temperature of 750°C is reached at the end of the second hour and coal ashing should continue for two more hours at 750 ± 10°C. Ash content determinations were routinely carried out for all concentrates and tailings obtained in flotation tests to assess the effectiveness of cleaning. In practice, it was found that a total heating time of two hours was adequate for samples not exceeding a weight of 1 g. However, the initial preheating to 500°C turned out to be critical for obtaining reproducible ash data for highly volatile samples, such as those of LS43 coal. The fixed carbon value is obtained by subtracting the moisture, volatile matter and ash contents from 100%.  4.3.2  Alkali Extraction Tests and Coal Oxidation  The alkali extraction test, sometimes referred to as the US Steel method, was developed by Lowenhaupt and Gray (1980, 1989) to reliably detect oxidized metallurgical coal. The method is based on the fact that oxidized coals yield a yellowish-brown solution of humic acids when treated with strong alkali solutions. The intensity of the color is proportional to the amount of humic acids leached from coal and hence to the degree of coal oxidation. Lowenhaupt and Gray found an excellent agreement between the petrographically determined degree of oxidation and the optical transmittance of the alkaline extract at 520 nm. As discussed by Lowenhaupt and Gray, the alkali extraction test can detect only strongly oxidized coal, since humic acids form only under the most severe oxidation conditions. The procedure was as follows: 1 gram of the ground coal was mixed with 100 cm of I N NaOH solution and the mixture was brought to boil. Boiling continued for exactly 3 109  minutes after which the hot slurry was filtered on two (one #40 on one #42) Whatman filter papers. The filtrate was allowed to cool down to room temperature and the transmittance at 520 nm was measured with the use of a Cary 50 (Varian) UV-VIS spectrophotometer. Since lower rank coals, even when not oxidized, are also known to release humic acids, the alkali extraction test can be used to directly estimate ranks of different coals. In this work, some tests were also carried out with the use of oxidized F4 coal. The idea was, simply, to oxidize F4 coal to the same degree of hydrophilicity as that of LS43 coal. Since LS43 coal yielded a significant amount of humic acids, the degree of F4 coal oxidation was controlled by measuring the amount of humic acids at different stages of the oxidation process. Oxidation was stopped when the amount of humic acids leached from F4 coal was equal to the humic acids concentration obtained from LS43 coal. In the oxidation experiments, about 3 kg of ground F4 coal were spread uniformly on several metal pans and the oxidation of the coal was carried out in a convection oven at 150°C. It was found that an oxidation time of 645 hrs («4 weeks) was needed to produce the desired degree of oxidation. Obviously, higher temperatures would generate humic acids much faster but it was observed that oxidation under such conditions was very difficult to control. Once exothermic oxidation was initiated, the temperature in the oven would locally increase to the coal selfcombustion levels. As a result, a couple of initial runs at 180°C led to almost complete ashing of the coal within the first 24 hrs, which also raised serious safety concerns. Therefore, it was decided to maintain the temperature at the lower level. Samples for the alkali extraction tests were taken every 24-48 hours. A sample of fresh F4 coal did not produce any measurable quantity of humic acids. LS43 and oxidized-F4 coals yielded humic acids solutions with transmittances of 64.0% at  110  520 nm. This transmittance value was equivalent to the optical density of an 88 mg/dm  3  Aldrich humic acids solution. In this work, the alkali extraction test was further modified for the determination of humic acids concentration in adsorption experiments as discussed in section 4.3.4.  4.3.3  BET Specific Surface Area Determination  A Quantasorb sorption system was used to measure the specific surface area of all the samples. In this apparatus, nitrogen is adsorbed from a nitrogen-helium mixture flowing through a powdered (ground) sample. The powder surface area is determined from the adsorption/desorption process of nitrogen molecules which is monitored by measuring the change in thermal conductivity of the nitrogen/helium mixture. In the experiment, a stable recorder baseline is established for a given adsorbate (TSy-to-carrier gas (He) ratio. Adsorption is then initiated by immersing the sample, placed in a U-tube glass sample holder, in liquid nitrogen. The adsorption peak appears as a result of the change in the thermal conductivity of the mixture. The conductivity change results from a decrease in N content in the mixture due to its adsorption on the tested 2  material. The area of the adsorption peak is directly proportional to the amount of nitrogen adsorbed. Typically, however, the desorption peak is actually used for calculating the surface area of the sample. The desorption peak is generated by removing the sample from liquid nitrogen. In this desorption process, the mixture leaving the sample holder is richer in the adsorbate and the signal is opposite in polarity to that of the adsorption. At the end of several adsorption/desorption cycles, a known volume (amount) of nitrogen is injected into the flow 111  stream in order to calibrate the desorption signal. To avoid any non-linearity effects in the detector signal, the calibration volume should be such that the area of the calibration peak is approximately equal to the area of the desorption peak. The determination of the surface area is a straightforward application of the B E T (Brunauer, Ernrnett, Teller) equation in the following form (Brunauer et al. 1938):  C-l  J  P_  x,„c P  X  0  1 XC m  where; X - weight of nitrogen adsorbed at a pressure P (obtained from the desorption peak) P - partial pressure of adsorbate (nitrogen) PQ - saturated vapor pressure of adsorbate X  M  - weight of adsorbate adsorbed at a monolayer coverage  C - a constant which is a function of the heat of the adsorbate condensation and heat of adsorption. It can be seen from the BET equation that a plot of  \IX(PQIP)  versus  P/PQ  should be a  straight line with the slope S and intercept / given by: C-l S = -  xc m  I-  1  xc m  Solving the set of the above equations forX yields: m  X  = m M  S  +I  112  Since both S and / can be easily found graphically (see Appendix 1 for the BET plots for the tested samples), X Smai,  M  can also be calculated, and the total surface area of the sample,  can be obtained from:  _ X S  M  A  <  M  ~  - N - A M  where iV~ is the Avogadro constant (6.023-10 molecules/mo 1), A is the cross-sectional area of 23  the adsorbate molecule (16.2-10" m /molecule for N molecule), and M is the molecular 20  2  2  weight of the adsorbate (28.01 g/mol for N ). 2  Finally, the specific surface area of the sample can be obtained by dividing the total surface area,  Sjotai,  by the weight of the sample.  In order to reliably estimate the slope and intercept from the B E T equation, several P/Po ratios should be tested. In this multipoint BET method three nitrogen/helium mixtures are usually used with PIPQ ratios between 0.05 and 0.35. In the single point method, only one mixture is used to obtain the surface area and it is also assumed that the intercept of the B E T equation is zero (X = 1/5). This assumption is M  valid when the constant C sufficiently large. As discussed by Lowell and Shields (1998), for C = 100 and a mixture of P/PQ = 0.3, the single point method would produce an error of only 2% compared to the multipoint approach. In this work, the coal and silica samples were tested after air-drying at room temperature. Preliminary tests showed that no moisture was present on the solid surfaces as indicated by the lack of "valleys" at the base of the desorption peaks. Drying at higher temperatures, 90-110°C, was avoided as some coals are known to change their structure/porosity under such conditions. Three nitrogen/helium mixtures were used with P/PQ ratios of 0.1, 0.2 and 0.3 for all samples.  113  In the case of F4-oxidized coal, one sample was air-dried at room temperature while another one was dried under partial vacuum (0.07 atm) at 30°C for 12 hours. It was expected that the amount of moisture on the oxidized coal surface would significantly increase compared to the fresh coal, and that air-drying alone would not remove the moisture trapped deep within the pores. The N2-BET specific surface areas of the coals and silica are summarized in Table 9. Table 9. BET specific surface areas of the tested coal and silica samples. Material BET Specific Surface Area [m /g] 2  SilcoSil 395 (Silica)  1.444  F4 Fresh  1.850  F4-Oxidized, Air Dried  0.877  F4-Oxidized, Vacuum Dried  0.880  LS43  2.070  Assuming that silica porosity is negligible, the BET surface area practically gives the external surface area of the silica. In the case of coals, however, their porosity may be quite significant and thus the BET surface area will include both the internal and external surfaces. Since D T A B and H A molecules are much larger than N2 molecules, the surface accessible to nitrogen does not have to be accessible to such large organic molecules, as further discussed in section 5.3.2. It is interesting to note at this point that the oxidation of F4 coal reduced its specific surface area more than twice (this effect will be discussed in later sections). Also, the vacuum-dried sample gave the same result as the air-dried one indicating that essentially no moisture was permanently trapped in pores.  114  4.3.4  Humic Acids (HA) Adsorption  In these experiments, the interaction of humic acids as coal depressants with the coals and silica was investigated. In order to accurately measure the adsorption of humic acids on the studied materials, an analytical method was developed for determination of humic acids concentration in solution. The starting point was the alkali extraction test of Lowenhaupt and Gray (1980, 1989) discussed in section 4.3.2. Firstly, Lowenhaupt and Gray did not provide any explanation why the wavelength of 520 nm was used in their transmittance measurements. Secondly, humic acids are a mixture of various organic polyelectrolytes that can be expected to exhibit very different light absorption characteristics, so the choice of a single wavelength to characterize such multicomponent systems seemed to be inadequate. As a first step, a full absorption spectrum of a 150 mg/dm Aldrich humic acids 3  solution in tap water was collected in the visible portion of the spectrum from 350 to 700 nm. The absorbances of the solution were automatically measured every 0.5 nm so that each spectrum consisted of 700 data points. A Cary 50 (Varian) UV-VIS spectrophotometer with a 1-cm quartz cell was used in all spectrophotometric measurements. Next, the same 150 mg/dm H A solution was conditioned with 5 grams of F4 coal for one hour and the spectrum of the residual humic acids after adsorption was also collected. It was again suspected that different components of the humic acids might selectively, or preferentially, adsorb on the coal so the residual solution would have a different composition and hence a different optical signature. The same was done for LS43 coal. The three absorption spectra are presented in Figure 17. Only every 30th point is shown for clarity. The spectrum of the pure humic acids solution is surprisingly featureless for such a complex 115  substrate. More importantly, however, the spectra for the residual humic acids after adsorption are also "smooth" and the different "trajectories" result from partial adsorption of the humic acids on these two coals. Clearly, adsorption on F4 coal is lower than adsorption on LS43 coal. The three spectra were then normalized by dividing the absorbances at different wavelengths by the absorbance at 350 nm (the highest value). The resulting curves are shown in Figure 18. It can be seen that the normalized spectra fall practically on a single curve. This unexpected observation indicates that the composition of the residual solutions after adsorption is the same as that of the original H A solution, and that essentially any wavelength can be used to measure the residual humic acids concentration.  300  400  500  600  700  800  F4 and LS43 coals.  116  The small differences at higher wavelengths result from the fact that the absorbances in this range are comparable to the absorbance of the reference tap water sample. Such monotonically changing spectra for humic substances have been previously reported in the literature (Bailey 1964, Kumada 1987). Some weak bands or shoulders were observed in the ultraviolet range of the spectra (at 250 nm) due to the presence of polycyclic aromatic structures (Friedel and Queiser 1959, Bailey 1964). As shown by Bailey (1964), such components, though, would make only a minor contribution to the structures of humic acids. Kumada (1987) concluded that the featureless character of the spectra is actually an intrinsic attribute of humic acids.  h0.8  h0.6  h-0.4  h0.2  300  400  i — • — T 500 600 Wavelength [nm]  700  800  Figure 18. Normalized absorption spectra of humic acids solutions before and after adsorption on F4 and LS43 coals.  117  Consequently, several calibration curves were obtained at various wavelengths as shown i n Appendix 2. Since the curve at 350 nm had the largest slope (sensitivity), the absorbances at 350 n m were used to measure the equilibrium H A concentrations. In the actual adsorption tests, 5 grams o f the tested materials were pre-conditioned for 20 minutes with 25 c m o f tap water in 100-cm glass bottles, placed i n an Environ shaker at 300 rpm, to ensure complete wetting o f the solids (coals in particular). Then 25 cm o f humic acids solutions o f different concentrations were added to each sample (two-fold dilution o f the original solution) and the mixtures were conditioned for one hour. In the kinetic tests, the reacting mixtures were taken out from the shaker after 5, 10, 15, 30 and 60 minutes to determine the time needed to reach the adsorption equilibrium. These tests were carried out at a constant initial H A concentration o f 150 mg/dm . 3  Afterwards, the solids were separated from solution by vacuum filtration on a 0.22-pm Millipore polycarbonate filter (Millipore Corporation, Bedford, M A , U S A ) . The equilibrium humic acids concentration i n the filtrate was determined at 350 n m from the calibration curve. The blanks were prepared by mixing 5 g o f the tested solids with 50 c m o f tap water only, 3  and subjecting the mixture to the same conditioning/filtration procedure. The amount o f humic acids adsorbed was then calculated from the following formula:  ^Ads  _ (Clnitial ~ Equil.) " ~ 5§' ABET  CAds  is the amount adsorbed (mg/m ),  C  r  where;  (mg/dm ),  CEquii.  0.05dm  2  Ci uiai n  is the initial humic acids concentration  is the equilibrium humic acids concentration (mg/dm ),  A T BE  is the specific  surface area (m /g).  118  The standard deviation of three absorbance readings for a given humic acids solution, or filtrate, was below 0.8%.  4.3.5  DTAB Adsorption  The conditioning/filtration procedure for amine adsorption tests was the same as in the case of humic acids adsorption. The amount of solids and the total volume of amine solution were 5 grams and 50 cm , respectively. A series of kinetic tests was also carried out at a 3  constant dodecyl-trimethyl ammonium bromide (DTAB) concentration of 154 mg/dm . The equilibrium D T A B concentration was determined with the use of a general method for amines developed by Mukerjee (1956) and Mukerjee and Mukerjee (1962). In this complex procedure, 20 cm of 0.01N hydrochloric acid are mixed with 5 cm of 3  3  a 1.5-g/dm Bromophenol Blue (BPB) solution (in distilled water) and 1 cm of 0.05N HC1, 3  3  all in a 100-cm volumetric flask. Then, 4 cm of the amine containing filtrate are added to the 3  3  mixture followed by 20 cm of chloroform (HPLC grade, Fisher Scientific). The basis for the method is a reaction between the anionic dye (BPB) and the cationic surfactant (DTAB) to form a water-insoluble dye-amine complex. In the acidic environment, the complex has a 1:1 (BPB:DTAB) molar ratio (Mukerjee and Mysels, 1955). The concentration of the B P B stock solution (1.5 g/dm = 2.17-10" mol/dm ) was chosen such that 3  3  3  the molar concentration of B P B in the acidified mixture was at least 4 times higher than the highest molar amine concentration tested (180 mg/dm = 5.84-10" mol/dm ). This excess 3  4  3  BPB would guarantee the complete capture of all D T A B molecules by the dye, as discussed by Mukerjee and Mukerjee (1962).  119  The water insoluble B P B - D T A B complex can be extracted into the chloroform phase by vigorously shaking the flask for about 45 seconds. This time was found to be sufficient to completely transfer the complex into the organic phase. The chloroform-water emulsion was then allowed to separate - the flasks were left to stand for 20 minutes. The chloroform layer turned yellow from the presence of the B P B - D T A B complex. The acidified water phase (30 cm in total) was easily poured off (the density of chloroform is 1.473 g/cm at 25°C) and the 3  3  organic phase was withdrawn with a pipette and transferred to a 1-cm quartz cell. The concentration of the complex in the organic phase was spectrophotometrically determined at 416 nm. The D T A B equilibrium concentration was read from a calibration curve (see Appendix 3). The amount of D T A B adsorbed was calculated in the same way as in the case of humic acids adsorption:  _  R  ^Ads  where;  CM*  —  (.C  I  N  I  T  I  A  L  -C  5gA  E q u i l  )-0mdm^  BET  is the amount of D T A B adsorbed (M/m ), 2  concentration (mol/dm ),  CEquii.  CMHOI  is the initial D T A B  is the equilibrium D T A B concentration (mol/dm ), ABET is the  specific surface area (m /g) of the solid. 2  Since the adsorption tests were carried out at D T A B concentrations as high as its cmc value (*4500 mg/dm ), and the calibration curve was linear only up to 180 mg/dm , such highly concentrated D T A B filtrates had to be diluted, usually 5, 10 or 25 times, in 50 cm 3  flasks. The diluted solutions were then used for determining the original D T A B concentration in the filtrate, taking the dilution factor into account.  120  Bromophenol Blue is usually supplied in two forms: as a sodium salt (dark, almost black powder), or as an acid (pink-yellow powder). In the course of the work, it was found that only the sodium salt (electrophoresis grade, Sigma-Aldrich Chemical Company) gave a linear calibration curve over a satisfactory D T A B concentration range under the above experimental conditions. This was quite surprising as these two molecular forms of BPB should be easily converted into each other just by changing the p H (BPB is a common acidbase indicator). It was also found that the BPB stock solution was stable for the first 12 hours after which it slowly lost its "activity". Using a fresh 1.5 g/dm B P B and a 40 mg/dm D T A B 3  3  solutions, the absorbance of chloroform at 416 nm was 0.6908. After 24 hours, the assay was repeated and the absorbance value decreased to 0.6629 (-4%), and to 0.6223 after 72 hrs. (10%). For this reason, fresh BPB solutions in distilled water were prepared a few hours before testing to minimize the experimental error. The standard deviation of three absorbance readings was between 0.5 and 0.7%, and seemed to increase slightly for low amine concentrations: A B S = 3.0089 ± 0.015 at 180 mg/dm of D T A B (0.5%), and A B S = 0.1440 ± 3  0.001 at 10 mg/dm of D T A B (0.7%). 3  4.3.6  Electrokinetic Studies  In order to investigate the interaction of D T A B with the tested solids, the effect of DTAB  on the  surface  charge was studied through electrokinetic (zeta potential)  measurements.  121  The experiments were carried out using a Zeta-Meter System 3.0 (Zeta-Meter Inc., New York) and a quartz electrophoretic cell. The zeta potential was measured as a function of pH and D T A B concentration. The use of a supporting electrolyte is necessary in electrokinetic measurements to maintain the thickness of the electrical double layer around coal particles constant. This also ensures that the zeta potential measured at the plane of shear is affected only by adsorbing ions ( H , OH" or amine cations) and not by changes in the ionic strength. It should also be +  borne in mind that the ionic strength also varies with pH and one of the purposes of using a "background" electrolyte is to mask this effect. A small amount of the ground coal (or silica), usually 0.02 g, was placed in a 200-cm beaker and 100 cm of a 0.005N KC1 solution were added as a supporting electrolyte. A few 3  drops of 0.1N K O H or 0.1N HC1 were introduced to adjust the pH. The dispersion was then conditioned until a constant pH value was achieved, normally within 15-25 minutes. These equilibrium values were subsequently reported. The dispersion was left to stand for 5 minutes so that the coarsest particles could settle, and then the upper portion of the dispersion, roughly 25-30 cm , was carefully poured into the electrophoretic cell. External voltage was applied to 3  the cell and 25 particles would be tracked and timed along a calibrated scale under the microscope to find their electrophoretic mobilities. The measuring system would then average the values and automatically calculate the zeta potential from the Smoluchowski equation (Equation 47). The standard deviation for 25 particles was always between 3-5 mV for the coals, and only 1-2 mV for the fine silica. When the effect of D T A B was investigated, the procedure was essentially the same. The solids were conditioned with 100 cm of a D T A B solution (in 0.005N KC1) of known 3  concentration for 30 minutes. The pH was adjusted only when necessary, to bring its values to 122  those used in the adsorption and flotation tests. It was assumed that D T A B adsorption on the tested material did not markedly change the D T A B concentration in solution. This assumption was justified by the fact that a large volume of D T A B solution and a very small quantity of solids were used in the tests. Based on this, the initial D T A B concentrations were reported as equal to the equilibrium values. The measurements were made as duplicates for a given D T A B concentration. Similarly, 25 fine particles were tracked to obtain the zeta potentials. The experimental error (standard deviation) was between 2-5 mV regardless of the material or D T A B concentration.  4.3.7  Contact Angle Measurements  In order to elucidate the effect of D T A B on the wettability of the coals and quartz (silica), a series of contact angle measurements on larger polished pieces was performed. In the case of the two coals, the specimens were hand-picked to make sure that clean coal pieces were used. In the case of silica, a quartz glass slide was used for the measurement. The slide was treated with a chromic acid mixture (1:1 sulfuric acid saturated with potassium dichromate) to remove any organic and mineral impurities from its surface. Then, the quartz slide was thoroughly washed with and stored in distilled water. The contact angles were measured with the use of a Rame-Hart goniometer. The captive bubble method was employed throughout the tests. In this method, a solid piece is immersed in solution and an air bubble from a capillary is produced. The bubble is allowed to attach to the surface while still in contact with the tip of the capillary. As the bubble size is increased (using a micrometric syringe), the solution/bubble interface recedes along the  123  surface until an equilibrium position is reached. At this moment a contact angle measurement is made through a magnifying ocular with a special cross-hair, and the measured angle is referred to as the water-receding contact angle (as the air bubble expands, water recedes). When the bubble size is decreased, the water advancing contact angle is measured. The measurements were performed in tap water and D T A B solutions with the concentrations ranging from 1.5-10" mol/dm to 1.5-10" mol/dm (approx. the cmc for 5  3  2  3  DTAB). The contact angles were measured on several places on the solid surface, and about 5-10 measurements were made in a given spot for different air bubble sizes so that about 2030 contact angle values were collected at each D T A B concentration. The experimental error (standard deviation) was as high as 5 degrees.  Coal Surface Preparation  Coal specimens for contact angle measurements were prepared following a special procedure developed by Dr. J. Drelich in cooperation with our group (Drelich et al. 1997). In this method, a larger coal piece was wet polished on a series of abrasive papers with the grit numbers varying from 60 to 1200, followed by polishing with a range of suspensions of fine alumina (particle size 5, 1, and 0.05 pm). After each polishing step with the use of alumina, the piece was washed with distilled water and cleaned ultrasonically for 10 minutes to remove any alumina particles attached to the coal surface. In the final stage, the piece was wet-polished with the use of a C H E M O M E T cloth (Buehler, Lake Bluff, IL, USA), cleaned in an ultrasonic bath, and washed with a stream of distilled water. As shown by Drelich et al. (1997), the polishing procedure produced a surface with a micro-roughness below 0.5 nm as determined by the Atomic Force Microscopy (AFM). More 124  importantly, the contact angle values measured on such "perfectly smooth" specimens were more reproducible and less scattered than the values measured on surfaces polished only with abrasive papers.  Between the measurements, the polished specimens were kept wet in distilled water.  4.3.8 Batch Flotation Tests In preliminary experiments, a number of flotation tests was carried out with the use of M I B C (Methyl Iso-Butyl Carbinol) for both F4 and LS43 coals to assess their relative floatabilities. Then, several polymers were studied as potential coal depressants (see section 4.1.3). These tests were done with the use of F4 coal only. In the second batch, a series of flotation runs was carried out with the use of D T A B at varying dosages. These experiments were performed on the clean coals and SilcoSil silica so that their natural responses to D T A B could be determined. In the main series, artificial coal-silica mixtures were floated in the reverse mode. The coals and the silica were mixed in such proportions that the ash content in the feed was always 40%. A l l coal-silica combinations were studied. Usually, only D T A B was used as a silica collector and its excess also provided frothing. Humic acids were used as a coal depressant. The depressant dosage was set at 400 g/ton of coal in the feed, as determined in the earlier tests. The pH of the flotation pulp was also measured but was not adjusted to reflect natural conditions. A l l flotation experiments were carried out in a standard Denver laboratory flotation machine. A modified 2-dm flotation cell, designed by Roberts et al. (1982), was used throughout. As described by Roberts et al., the standard cell's geometry was significantly 125  altered by means of a perspex insert embracing the impeller shaft. The insert prevents the formation of froth at the back of the cell, and near the shaft, where the froth is most difficult to collect. The bottom of the perspex block is sloped at 45° to guide the air/coal aggregates to the front of the cell. The froth could thus be removed by means of a scraper designed to cover the full width of the cell at a fixed depth. The size (or the reach) of the scraper was such that its bottom edge would just touch the pulp surface once the cell was filled. Another modification was the addition of a side water reservoir for constant volume control. This was done by connecting the side vessel, using P V C tubing, through a valve at the bottom of the cell. The vessel could be freely moved up or down along a laboratory burette stand to maintain a constant pulp volume in the cell. During a test, the reservoir would be constantly refilled with water directly from a tap. A schematic illustration of the modified cell is shown in Appendix 4. In all tests, the feed weight was 200 g, the pulp volume was approximately 2 dm , and the impeller's speed was 1100 rpm which produced an aeration rate of 2 dm /mm.  Flotation with D T A B and Polymeric Depressants  Since F4 coal was expected to float well due to its high rank, several polymers were tested as possible depressants for this coal. LS43 coal was tested only with D T A B and M I B C as its natural floatability was very poor, and no depressants were needed in this case. A 200-g batch of coal (-212 pm) was mixed with 1.4 dm of tap water in the flotation 3  cell. The slurry was initially conditioned for 15 minutes at 1500 rpm to completely wet the coal particles. After that, the impeller speed was lowered to 1100 rpm, and 200 cm of a polymer (or D T A B ) solution of known concentration were added to the pulp. Mixing 126  continued for another 15 minutes to allow the polymer to adsorb. At the end of this second 15-minute conditioning period, the pH was also measured. If M I B C needed to be used, 200 cm of M I B C solution would be added at this moment. If only natural floatability was to be 3  investigated, 200 cm of tap water were added instead and flotation commenced 30 seconds 3  later by opening the air valve on the impeller shaft and, immediately afterwards, the side reservoir valve at the bottom of the cell. The total conditioning time was 30 minutes (plus 30 seconds) and the solids content in the pulp was 10% by weight (200g of feed + a total of 1800 cm of tap water). These 3  parameters were rigidly observed throughout the tests. It was quickly realized that the collection of concentrates had to be precisely timed to obtain reproducible results. As soon as flotation started, the froth was scraped manually every 5 seconds. The cumulative concentrates were collected after total flotation times of 30 seconds, 1 minute, 2 minutes and 5 minutes (i.e. after exactly 60 "scrapings") to investigate the kinetics of flotation. The solids remaining in the pulp were treated as tailings. The wet concentrates, and the tailings slurry, were directly collected into metal pans, dried overnight in a convection oven at 100-110°C, and then weighed. Since both coals had rather low ash contents (see section 4.2.2), no ash assays were performed when coals alone were tested, and only the concentrate yields were recorded. The absolute yield difference between duplicate runs, after employing the precisely timed scraping sequence, never exceeded 5%.  127  Reverse Flotation  Reverse flotation tests were performed on artificial mixtures of coal and silica with an ash content of 40%, taking into account the ash contents of the "pure" coals. Similarly to the previous procedure, 200 grams of coal/silica feed were mixed with 1.4 dm of tap water in the 3  flotation cell at 1500 rpm. After 15 minutes of mixing, 200 cm of a depressant solution were 3  added to the pulp and conditioning (at 1100 rpm) continued for another 15 minutes. In the next stage, 200 cm of D T A B solution were introduced and the pulp was mixed for the final 3  15 minutes (total conditioning time was 45 minutes). Normally, only D T A B was used in these tests and its excess provided frothing as well. Flotation commenced at this point. In some instances, however, M I B C was also utilized along with D T A B and a depressant. In such cases, the initial conditioning of the pulp was done in 1.2 dm of tap 3  water, and 200 dm of M I B C solution were added at the end of the 45-minute mixing period. 3  Flotation started 30 seconds later. The pulp density was again 10% by weight. The concentrates were collected after flotation times of 30 seconds, 1 minute, 2 minutes and 5 minutes, following the previous 5-second scraping intervals. After drying and weighing (the weights of empty pans had been recorded earlier so the weights of the concentrates could be easily determined from the weight difference), the individual concentrates and tailings were homogenized on the pans by at least five "cone-andquartering" steps (using a knife and a 2-inch wide paint brush). Afterwards, representative 10-g samples of each of the concentrates and tailings were taken and sent for an ash assay. In some instances, however, the yield of a given concentrate was well below 10 grams, and could be as low as 1.5 grams. For such samples, the entire  128  amount of concentrate was sent for an ash content determination. Ash assays were done as duplicates for every sample. Using the ash contents and flotation yields, the ash content in the feed was backcalculated from a simple mass balance equation to estimate the accuracy of the entire flotation-homogenization-sampling procedure. The calculated average ash content of 34 flotation tests (almost all were done as duplicates) was 40.1 + 0.3% (standard deviation).  4.3.9  Rheological Tests  In this thesis, it is assumed that coal flotation depressants and coal dispersants belong to the same class of additives, i.e. they have the same effect on coal surface properties. Therefore, rheological tests were carried out to investigate the effect of coal depressants on the properties of coal-water slurries. The experiments were conducted either as a function of coal content in the slurry, or as a function of an additive concentration at a constant coal loading. F4 coal suspensions were known to exhibit high yield stresses due to the hydrophobic aggregation of the fine coal particles (Pawlik et al. 1997). Therefore, this coal seemed to be a natural choice for studying further the effect of different additives on the hydrophobic/hydrophilic transition on the coal surface, and the resulting changes in yield stress values. The measurements were performed with the use of a Haake Rotovisco R V 20 rotational viscometer. The elongated fixture (Klein 1992, Klein et al. 1995) was chosen as the measuring geometry. The fixture was designed specifically to test settling suspensions. The original design (Klein 1992, Klein et al. 1995) is a concentric cylinder, bob-in-cup, doublegap arrangement with the gap sizes of 1.0 and 1.1 mm for the inner and outer gaps, 129  respectively. Since the coarsest coal particles in the slurries had a diameter of 212 microns, the gap-to-particle size ratio would be less than 5. It is required that this ratio be at least 10 (Van Wazer 1963), and thus the fixture was redesigned following the guidelines of Klein (1992), and the gap sizes were increased to 2.50 and 3.03 mm for the inner and outer gap, respectively. Nevertheless, some tests were done with the use of both fixtures. The elongated, fixture is schematically shown in Figure 19. The inner and outer gap sizes are determined by the dimensions of the four radii; r\ = 1.650 cm, r = 1.900 cm, r = 2.000 cm and r = 2.303 cm. Thus, the inner gap width is equal 2  3  4  to r - r\ (0.250 cm), while the outer gap width equals r - r (0.303 cm). The dimensions of 2  4  3  the four radii satisfy the following criteria for achieving similar shearing conditions in the two annular gaps (Moore and Davis 1956):  and (Whorlow 1980):  The Haake Rotovisco RV20 applies a rotational speed, Q%, to the bob which, for a Newtonian fluid, is directly proportional to the shear rate, D:  £>= M-Q„  /o  where M i s the fixture constant given by (Klein 1992, Klein et al. 1995): 130  Plan View  Side View  Figure 19. The elongated fixture for testing settling suspensions (original design by Klein 1992).  131  Putting the actual dimensions into the above equation, M w a s determined to be 3.736 for the redesigned fixture. The shear stress, r, is directly proportional to the torque, 7%, exerted on the bob by the tested suspension:  r=A-T  %  where A is a constant. Using three standard oils having viscosities of 0.05, 0.5 and 1 Pa-sec, the value of A for the new geometry was found to be 1.471. This was obtained by dividing the actual viscosity of a standard by the measured viscosity under the assumption that A was equal to 1. Once the constants M and A are determined and put into the Haake software controlling the viscometer, the shear rate and shear stress are automatically calculated from the rotational speed and torque, respectively. The double-gap, elongated fixture was designed with consideration of the following potential errors (Klein 1992): Errors arising due to settling of particles; the dimensions of the fixture are such that the bob is positioned within the constant density zone for the duration of the measurement, even when some settling takes place. The total length of the outer cylinder is 23 cm. A l l shearing surfaces are grooved to reduce wall-slip errors. Temperature is controlled by a water jacket surrounding the outer cylinder. 132  -  End effects are minimal due to the bob design; the bob is essentially a hollow cylinder supported by three thin spokes attached to a torque-transducer shaft. In the case of the original fixture, non-Newtonian shear rate effects were shown to be negligible due to the narrow gaps (1.0 and 1.1 mm) and low shear stresses (0-10 Pa) encountered in the tested magnetite suspensions (Klein 1992). However, for the redesigned geometry used in this work, it was still assumed that the  tested coal suspensions were fully sheared along the whole width of the gaps (2.5 and 3.03 mm) at all rotational speeds of the bob, and that the shear rate decay along the gaps was linear. Because of the last assumption, the simplified terms shear rate and shear stress used throughout this work actually denote the apparent shear rate and apparent shear stress, respectively. Since the rheological tests were used only to assess qualitative changes in the rheological parameters (apparent viscosity and yield stress), it was also assumed that any errors resulting from the use of uncorrected data would not affect the final conclusions. The coal-water slurries were prepared under strictly controlled conditions of impeller speed and mixing time, as such suspensions exhibit strong thixotropic (time-dependent) behavior. The suspensions were prepared by mixing 400-g batches of the ground coal with a predetermined amount of tap water in a 2-dm glass jar. The slurry was then stirred for 15 3  minutes at an impeller speed of 200 rpm (Stedi-Speed adjustable stirrer, Fisher) to completely wet the coal. In the tests with additives, a polymer stock solution was slowly added at this moment (over a period of 2 min) while mixing, and the slurry was stirred for 10 more minutes at 200 rpm. When no polymers were used, the suspension was continuously conditioned for 27 minutes at 200 rpm. Finally, the stirring speed was lowered to 100 rpm and conditioning continued for 15 minutes. The total conditioning time was 42 minutes in all cases. The mixing 133  speeds and times were chosen arbitrarily but were strictly followed in all the tests to keep the pre-shearing history constant.  This procedure  produced  slurries, especially highly  concentrated ones, that could be almost freely poured from the jar into the measuring fixture. Conditioning times shorter than 20 minutes gave suspensions with extremely high yield stresses that made their handling difficult. Standard flow curves were obtained as quickly as possible after transferring the slurry into the measuring system. The sample volume was about 400 cm . The shear rate was increased over 3.5 minutes from 0 to 250 sec" for the original fixture, or to 120 sec" for the 1  1  modified system, and then decreased back to zero over another 3.5 minutes. The shearing history was programmed using a Haake Rheocontroller R C 20. A ramping time of at least 2.5 minutes is recommended by Haake for achieving steady-state conditions under continuous shearing. The upper values of the shear rates differed because the laminar flow limits were found to be around 300 sec" (Klein 1992), and near 150 sec" for the original and redesigned 1  1  fixture, respectively. The upper shear rate limit for the redesigned geometry was determined by collecting flow curves for a standard oil with a viscosity of 0.5 Pa-sec and a 55% F4 coalwater slurry in the presence of 0.8% humic acids (dispersant), over the shear rate range from 0 to 250 sec" . Both flow curves turned upwards in a dilatant (shear-thickening) fashion around 1  150 sec" indicating an onset of a turbulent flow. 1  The ramping parts of the flow curves were then fitted with either the Casson or Bingham models to estimate the yield stresses. In some cases, the apparent viscosities at 100 sec" were also calculated. 1  Since all rheological experiments were carried out to qualitatively describe certain relative trends, the reproducibility of the rheograms was not as thoroughly examined as in the case of other experimental techniques. Only one flow curve at one coal concentration was 134  determined three times (curves A in Fig. 28, 29, and 30 - section 5.1.2) and this comparison shows good reproducibility of the tests.  4.3.10 Surface Tension Measurements  The surface tensions of D T A B and humic acids solutions were measured with the use of a C E M C O ring (du Noiiy) tensiometer. After taking raw readings, corrections were applied following the procedure of Zuidema and Waters (1941). These measurements were carried out to assess relative surface-activities of the two reagents. Three measurements were taken for every D T A B concentration and the standard deviation was calculated. The results are presented in Appendix 5.  135  5 RESULTS 5.1 RHEOLOGY OF COAL-WATER SLURRIES (CWS) 5.1.1  Effect of Humic Acids and Coal Surface Properties  Figures 20 and 21 show sets of flow curves for suspensions of F4 and F13 coals, respectively, in tap water at varying coal content. These results, as well as those in Figures 22, 23, 24 and 25 were obtained with the use of the original double-gap fixture (gap sizes 1.0 and 1.1 mm).  Shear Rate [sec ] 1  Figure 20. Flow curves (ramping parts) for suspensions of F4 coal in water at various coal contents (weight % solids in suspension).  136  160 140 H  F13 Coal in Water Original Fixture 1.0 and 1.1 mm gaps  120 H  PL,  100 H  60.1 %  59.4 % 58.7 % 58.0 %  0 0  50  100  150  200  250  300  Shear Rate [sec ] -1  Figure 21. Flow curves (ramping parts) for suspensions of F13 coal in water at various coal contents (wt.%).  The results of similar tests but in 0.8% humic acids solutions are presented in Figures 22 and 23. Figure 24 shows a comparison of the full flow curves obtained in tap water and in the presence of humic acids for F4 coal suspensions at a constant coal loading of 55.2%. In this particular case, an extra shearing "segment" was added. When the shear rate reached its highest value of 250 sec" after the first 3.5 min, the shearing continued at 250 sec" for an 1  1  additional 3 minutes before the final descent to zero, to enhance the thixotropic response of 2  It is to be noted that the slurry preparation procedure generally gave reproducible flow curves for freshly prepared suspensions in water without any additives. However, when the experiment was repeated using the same suspension immediately after the first flow curve was collected, the second curve was always significantly different indicating that the observed "thixotropic" behavior was not reversible. Moreover, reconditioning and retesting of the coal suspensions, after a day or two, resulted in even poorer reproducibility of the subsequent flow curves relative to the baseline curve. Such time-dependent effects can probably be attributed to a slow  137  the coal-water suspensions. The arrows on the graph indicate the shearing history. Analogous data for F13 coal in water and in a H A solution are shown in Figure 25.  140 H  ^  F4 Coal in 0.8% H A Original Fixture 1.0 and 1.1 mm gaps  100-  I  n  • i •i  250 300 150 200 Shear Rate [sec ] Figure 22. Flow curves (ramping parts) for suspensions of F4 coal in the presence of 0.8% of HA (per coal weight) at various coal contents. 50  100  1  It can be seen that the flow curves in water are quite "rough" and irregular particularly at lower shear rates. This effect is to some extent related to the bridging of the gaps by the coarsest/aggregated particles. In these difficult cases, in order to estimate the yield points, a Haake computer program was used to select the model giving the best fit.  penetration of micro-bubbles may affect the responses. In significant.  pores by water and hence to an increase in wetting of the coal surfaces. It is also known that of air nucleate on hydrophobic particles in water in an open system. Therefore, any shearing that distribution of such micro-bubbles on coal surfaces may also lead to non-reproducible rheological the presence of dispersants that make coal surfaces hydrophilic, this effect becomes less  138  As Figure 24 shows, the "return" part of the full flow curve is relatively smooth and it is often recommended that the descending shear rate segment be used to calculate the yield stress. The rationale usually given for such an approach is that the initial two segments should be considered as pre-shearing (conditioning), and only the third descending part is the actual flow curve. It must be pointed out, however, that the preparation procedure of the slurries involved intense conditioning just before the measurements, and it was, therefore, believed that  the  ramping  parts  of  the  curves  better  reflected  the  original  state  of  aggregation/dispersion in the tested suspensions. This was also one of the reasons why the yield stresses were calculated from the ascending portions of the flow curves.  0  50  100  150 200 Shear Rate [sec ]  250  300  1  Figure 23. Flow curves (ramping parts) for suspensions of F13 coal in the presence of 0.8% of HA (per coal weight) at various coal contents.  139  Interestingly, the ramping portions of the flow curves for F4 coal-water suspensions in the presence of 0.8% of humic acids (Figure 22) are smoother than those obtained in water (Figure 20). This suggests that the bridging of the gaps does not take place when the coal is dispersed by the addition of H A . This observation also validates the assumption that the ramping parts are much more sensitive to the changes in the aggregation/dispersion of the slurries.  -  F4 Coal Content 55.2 % Original Fixture  3 min.  0  50  100  150 200 Shear Rate [sec ]  250  300  -1  Figure 24. Full flow curves for suspensions of F4 coal in water and in the presence of 0.8% of HA (per coal weight) at the coal content of 55.2% (wt). Arrows indicate shearing history.  As Figure 24 shows, in the case of F4 coal-water suspensions, the shear stress hysteresis, characteristic of strongly thixotropic systems, mostly disappears in the presence of humic acids. This effect is only weakly pronounced for F13 coal suspensions (Figure 25).  140  Even concentrated suspensions of F13 coal in water («60%) are not as thixotropic as the suspensions of F4 coal at much lower solids content («55%).  1  160-  140 H  F13 Coal Content 60.1 % Original Fixture 1.0 and 1.1 mm gaps  1  120 H 4? cn CD  a  Water only  100-  80 H  GO CS CD  CM  60-  0.8% H A solution  40 H 20 H  n  • i •i  250 300 150 200 Shear Rate [sec ] Figure 25. Full flow curves for suspensions of F13 coal in water and in the presence of 0.8% of HA (per coal weight) at the coal content of 60.1% (wt). 100  1  From the flow curves presented in Figures 20-23, the yield stresses and apparent viscosities (at 100 sec" ) were calculated and the results are summarized in Figure 26 for two 1  different measuring geometries. Since F4 coal was known from our earlier work (Pawlik et al. 1997) to aggregate the most, and thus to bridge the narrow gaps of the Haake viscometer, the tests were repeated with the use of the wider-gap system for this coal only. It can be clearly seen that the data from the redesigned fixture differ from those obtained using the original geometry, particularly for "water only" suspensions. This effect appears to be due to the partial plugging of the gaps by the coarsest/aggregated particles. 141  50  52  54  56 58 Coal Content [wt %]  60  62  64  Figure 26. Effect of humic acids (HA) on the yield stress and apparent viscosity of F4 and F13 coal-water suspensions at varying solids content. Results for two different, double-gap, measuring geometries are shown.  142  As Figure 26 demonstrates, both the apparent viscosity and yield stress increase exponentially with coal concentration as expected from the Krieger-Dougherty, or Mooney equations. Interestingly, the oxidized coal (F13) produces suspensions in water with coal contents 5-6% higher than those of the bituminous coal (F4) at the same viscosity. Moreover, the yield stress values for F13 coal suspensions in water are clearly lower than the yield stresses of F4 coal suspensions. Since the humic acids content is a function of coal rank (Lawson and Stewart 1989, Lowenhaupt and Gray 1980, 1989) - it increases as the rank decreases - humic acids were used as a model substance to modify the coal surface properties. It was assumed that humic acids adsorption onto the bituminous F4 coal surface would render it similar to the surface of a lower rank/oxidized coal. The effect of such a surface modification on the rheology of coal-water slurries is also illustrated in Figure 26. The action of humic acids is demonstrated in Figure 27 which shows the changes in zeta potential and contact angles in the presence of humic acids (HA). It can be seen that the effect of humic acids is much more pronounced in the case of F4 bituminous coal. The zeta potential of F4 coal particles decreases sharply towards more negative values as the H A concentration increases. At the same time, the oxidized coal is less affected and only very large amounts of humic acids bring the zeta potential down to, roughly, the same value as in the case of F4 coal. The effect of humic acids on the zeta potential of LS43 coal particles is also illustrated - the trend is quite similar to the results for F13 coal. The wettability of the bituminous coal is also strongly affected. The advancing contact angle measured on a polished piece of F4 coal declines more markedly than the advancing contact angle on F13 coal. Clearly, humic acids render F4 coal more hydrophilic.  143  Figure 27. Effect of humic acids on surface properties (contact angle and zeta potential) of F4 and F13 coals. Effect of HA on the zeta potential ofLS43 coal particles is also shown.  144  Undoubtedly, F13 coal is naturally more hydrophilic than F4, as indicated by the water advancing contact angles of 45 and 80 degrees measured in water on F13 and F4 coals, respectively. This observation is in good agreement with the fact that F13 coal came from a strongly weathered seam. Contact angle measurements at higher H A concentrations (>300 mg/dm ) were very difficult to make since more concentrated H A solutions were too opaque (tea-like appearance) for a reliable visual reading using the goniometer.  5.1.2  Effect of Low Molecular Weight Polymers  Figures 28, 29 and 30 illustrate the ramping parts of the flow curves for 55% (wt) F4 coal-water suspensions obtained in the presence of some of the tested low molecular weight polymers, i.e. carboxymethyl cellulose, polystyrene sulfonate and humic acids. The flow curves in Figures 28 through 30 were obtained with the use of the modified fixture with 2.50 and 3.03-mm gaps. The models used to calculate the yield stresses are also indicated in the legends. As curve C in Figure 28 shows, the yield stress for such a downward running curve is very often obtained by extrapolating the linear portion of the flow curve to zero shear rate - as i f the Bingham model was used only over the linear shear rate range. This would clearly lead to a significant overestimation of the yield point. As shown in Figure 28, the Casson model fits the data much better, and thus provides more reliable yield stress values.  145  J  0  I  I  I  I  I  30  I  I  I  I  60 90 Shear Rate [sec ]  I  I  I  120  L  150  -1  A : Water only, T = 33.07 Pa (Bingham) 0  B: C M C dosage: 54.1 g/t, T = 25.03 Pa (Bingham) 0  C: C M C dosage: 93.6 g/t, x = 9.82 Pa (Casson) 0  D: C M C dosage: 451 g/t, T = 0.00 Pa (Newton) 0  Figure 28. Examples of flow curves obtained in the presence of carboxymethyl cellulose (MW = 80,000). F4 coal content: 55% (wt).  It is worth noting that the flow curves obtained with the wider-gap geometry are much smoother than the curves shown in section 5.1.1 (Figure 20) obtained with the use of the original fixture. This again reflects the bridging effect by the coarsest/aggregated particles. Such flow curves also allowed an easier determination of the flow parameters. It can be seen from Figures 28-30 that F4 coal in water produces non-Newtonian suspensions whose flow curves follow the Bingham plastic model. As the dosage of the 146  polymers increases the flow behavior becomes more Casson-like, and this change is also accompanied by a decrease in the yield point. At sufficiently high polymer concentrations the rheological response is essentially Newtonian, in terms of viscosity, with a zero yield stress.  J  I  I  I  I  I  I  I  I  I  I  I  L  Shear Rate [sec ] -1  A : Water only, T = 32.85 Pa (Bingham) 0  B: PSS dosage: 46.4 g/t, x = 16.75 Pa (Casson) 0  C: PSS dosage: 1061 g/t, x = 0.00 Pa (Newton) 0  Figure 29. Examples of flow curves obtained in the presence of polystyrene sulfonate (MW = 14,000). F4 coal content: 55% (wt).  It is noteworthy that the flow curves measured in water were very reproducible with almost identical yield stress values (32.75 to 33.03 Pa). This result can be attributed to the rigidly observed CWS preparation procedure.  147  The other two polymers (dextran and hydroxyethyl cellulose) gave qualitatively similar results.  J  0  I  I  I  I  30  I  L  60 90 Shear Rate [sec ]  120  150  -1  A : Water only, x = 32.75 Pa (Bingham) 0  B: H A dosage: 110 g/t, T = 16.01 Pa (Casson) 0  C: H A dosage: 208 g/t, T = 6.88 Pa (Casson) 0  D: H A dosage: 930 g/t, x = 0.00 Pa (Newton) 0  Figure 30. Examples of flow curves obtained in the presence of humic acids (low MW). F4 coal content: 55% (wt).  148  5.1.3  Effect of Polymers on Natural Floatability of F4 Coal.  The same polymers that were used in the rheological measurements were then investigated in a series of flotation tests involving F4 coal. Figure 31 presents a combined plot of the yield stress values calculated from the flow curves and the F4 coal flotation yield (after 2 min of flotation). The flotation tests were carried out without any frother or collector to study the natural floatability of F4 coal. The yield stresses were calculated from the flow curves shown in section 5.1.2. The coal content was maintained constant at 55.0% (wt). A l l the additives affected flotation in a similar manner. F4 coal flotation was gradually depressed at polymer dosages as low as 10-20 g/t and practically complete depression was observed at dosages of 200 to 600 g/t. In tap water, the yield of F4 coal flotation was 33% as indicated by the dashed lines. Simultaneously, the yield stress of F4 coal suspensions was effectively lowered to zero in approximately the same range of polymer concentrations where the complete depression  of flotation took place. Apparently, the anionic polymers (humic acids,  polystyrene sulfonate and carboxymethyl cellulose) were stronger coal dispersants than the nonionic ones (dextran, hydroxyethyl-cellulose) as the yield stress decreased sharply at relatively lower doses of these anionic additives. The trend for the flotation depressants, however, was not so obvious suggesting that the electrostatic interactions are not so important in the depression mechanism of the bituminous coal.  149  J  I  I  I  I  _i  I I  I  I  I  i  •35  i I  Dextrin ( M W = 56,000) was used instead of Dextran  -30  _  :  £  - 2 5  4  J-20 L  15  ~ L  40-  _i  i  •• • • I  _l  O O • V A  PH  I  I  _1  I—L  I  I  1  1  1  1  0  5 •0  Polystyrene Sulfonate, M W = 14,000 Carboxymethyl Cellulose, M W = 80,000 Humic Acids, Low M W Hydroxyethyl Cellulose, M W = 24,000 Dextran, M W = 15,000  C/J  03  bO g  m o  a o  <z> CO  03  U  i  Water Only  100 1000 Polymer Dosage [g/t of coal]  1 — i — i — i — i  i i  10000  Figure 31. Natural floatability of F4 coal in the presence of various polymers. Corresponding changes in the yield stress values of 55% (wt) F4 coal-water suspensions are also shown.  150  5.2 BATCH FLOTATION 5.2.1  Flotation of LS43 and F4 coals with MIBC.  In the main part of this work, two coals were investigated; F4 (medium volatile bituminous) and LS43 (sub-bituminous). Their flotation in the presence of M I B C is illustrated in Figure 32. It should be noted that the natural floatability of F4 coal (no MIBC) is high as indicated by a relatively high flotation yield of 33% achieved without the aid of a frother or a collector. In fact, the yield for F4 coal was over 60% after 10 minutes of flotation.  0  50  100 M I B C Dosage [g/t]  150  200  Figure 32. Flotation of F4 and LS43 coals with the use of MIBC as a frother.  LS43 coal floats poorly and its response does not improve even at high frother dosages. While the yields of 90% or more could be easily obtained for F4 coal, the  151  concentrate yields for LS43 coal did not exceed 8%. These results are typical for high rank and low rank coals, respectively.  5.2.2 Flotation of Coals and Silica with DTAB.  Figure 33 shows the flotation response of "pure" coals (F4, F4-oxidized, LS43), F4 coal depressed with 400 g/t of H A , and silica to varying dosages of D T A B . cmc  0 —I—i W  a  t  0 n l  e  y  i 111 r  100  1  1—i—i—i  i 111  1  1—i—i—i  i 111  1  1000 10000 D T A B Dosage [g/t] • • • • O  1—I—i—i  i 11  100000  F4, p H = 7.0-7.3 F4-Oxidized, p H = 3.7-4.0 F4 +400 g/t H u m . A c . , p H = 7.1-7.4 L S 4 3 , p H = 8.4-8.7 Silica, p H = 7.0-7.3  Figure 33. Flotation of F4 coal, LS43 coal and silica with the use of DTAB. Flotation of F4oxidized coal, and F4 coal in the presence of400 g/t of humic acids is also shown.  152  It can be seen from Figure 33 that silica and F4 coal float well over a wide range of D T A B dosages. LS43 coal exhibits only a very narrow flotation window at rather high dosages of D T A B . Interestingly, F4 coal depressed with 400 g/t of humic acids responds similarly to the oxidized F4 coal. In these two cases, the onset of flotation was shifted towards higher D T A B concentrations compared to the untreated F4 coal. It must be noted, however, that the natural pH of the flotation pulp was quite acidic in the case of the oxidized F4 coal (3.7-4.0), as opposed to the neutral pH for the HA-depressed coal. Also, LS43 coal generated slightly alkaline pulps (pH = 8.4-8.7) which was attributed to a high carbonate content in this coal. A chemical analysis of the mineral matter in LS43 coal is shown in Table 10. Table 10. Chemical composition of mineral matter from LS43 coal (percent of coal weight). Aluminum Ferric Oxide Calcium Magnesium Silica Carbonate Carbonate Oxide Fe 0 Si0 A1 0 CaC0 MgC0 2.1% 0.5% 8.1% 1.9% 1.0% 3  3  2  2  3  2  3  As Figure 33 indicates, the flotation of all samples drops dramatically as the D T A B dosage approaches the critical micelle concentration (marked with the dashed vertical line).  5.2.3  Effect of pH on Flotation of F4 and F4-Oxidized Coals  The pH of the flotation pulp for LS43 coal was very difficult to control. Large volumes of concentrated HCI were needed to bring the pH to only mildly acidic values (~4.5). Flotation concentrates collected under these conditions contained  large amounts of  hygroscopic calcium and magnesium salts that made accurate weighing very difficult. Therefore, F4-oxidized coal was used instead and its behavior was assumed to be indicative of that of LS43 coal. The results are shown in Figure 34.  153  o  —11  3  i  i  i  i 4  i  i  i  i  i  i  i  i  i  5  i 6  i  i  ii  i i  7  i  i  i  i  i  i i  8  i  ii i  i i  9  i 10  i  i  i  i  ii 11  i  i i  i 12  i  i i  i  — I 13  pH of Pulp Figure 34. Effect of pH on flotation of F4 and F4-oxidized coal in the presence of 1500 g/t of DTAB.  F4 coal floats very well over a wide pH range. Only at pH = 12 does the flotation yield drop slightly. In contrast, the flotation of the oxidized coal strongly depends on pH with almost no flotation taking place at pH = 7.2.  5.2.4  Flotation of Coal/Silica Mixtures with DTAB  Figure 35 shows the cumulative ash content in the concentrate and cumulative flotation yield at varying dosages of D T A B , for the F4/Silica mixture. No coal depressant (humic acids) was added to the pulp. 154  The results obtained in the presence of 400 g/t of humic acids are presented in Figure 36 for F4/Silica mixtures. In order to obtain an ash content of 40% in the feed, 130.1 g of F4 coal (ash content 1.1%) were mixed with 69.9 g of silica. The dosage of H A was per weight of coal (130.1 g) in the feed. The flotation kinetics of LS43/Silica mixtures is illustrated in Figure 37. The mixtures were prepared by blending 140.2 g of the coal (ash content 14.4%) with 59.8 g of silica to obtain an ash content of 40% in the feed. Humic acids were not added to the pulp. It is evident from Figures 35-37 that only the first concentrates exhibit high ash contents. As flotation continues the cumulative ash content systematically decreases reaching a level of 40% at longer flotation times. At the same time, the cumulative yield continuously increases up to an almost complete recovery of solids at higher D T A B dosages. There also appears to be an optimum D T A B dosage for every mixture. Above the optimum D T A B concentration the cumulative ash content decreases. These trends are indicated in Figure 38 which shows the cumulative yield and the cumulative ash content in the concentrates after 2 minutes of flotation as a function of D T A B concentration. A l l ash curves exhibit maxima while the yield curves sharply increase. The dashed line for the F4/Silica mixture runs down to an ash content of 15.5% obtained in a test carried out in tap water only (no DTAB). A series of additional flotation tests was carried out on LS43/silica mixtures to investigate the effect of conditioning time on the separation efficiency. The results are shown in Appendix 10. These experiments were also to offer some further insight into the mechanism of amine adsorption under the flotation conditions.  155  DTAB Dosage O Water Only 125 g/t • 250 g/t A 500 g/t V 800 g/t  80 H  O  2  3 4 Flotation Time [min]  Figure 35. Flotation kinetics of F4/Silica mixture (40% of ash in feed) at various dosages of DTAB, pH = 7.0-7.2.  156  i—•—i—•—r 2  3 4 Flotation Time [min]  Figure 36. Flotation kinetics of F4(400 g/t of humic acids)/Silica mixture (40% of ash in feed) at varying dosages of DTAB, pH = 7.1-7.3.  157  rj 10  2  3 4 Flotation Time [min]  Figure 37. Flotation kinetics ofLS43'/Silica mixture (40% of ash in feed) at different dosages of DTAB, pH = 8.3-8.6.  158  _l  I  I  I  _J_  l _ L .  _l  I  I  I  100  L.  h-80 £  O • •  >  'M  I  u  10  F4 + Silica, p H = 7.0-7.3 F4 + 400 g/t of Hum. Ac. + Silica, p H = 7.0-7.4 LS43 + Silica, p H = 8.3-8.6  H  0  I  Water Only 100  I™' I I '""I" I I I  ~i  1—i—i—i  i11  1000 10000 D T A B Dosage [g/t]  ~i  1—i—i—i  i11  100000  Figure 38. Yields and ash contents of concentrates after 2 min of flotation for the tested coal/silica mixtures at varying DTAB dosages.  159  5.3 ADSORPTION STUDIES 5.3.1  Adsorption of Humic Acids on Coals and Silica  Figure 39 shows the kinetics of humic acids adsorption on LS43 and F4 coals. Adsorption isotherms of H A onto LS43 and F4 coals, as well as on silica, are plotted in Figure 40.  0  10  20  30  40 50 60 70 80 Conditioning Time [min]  90  100  110 120  Figure 39. Adsorption kinetics of humic acids onto LS43 and F4 coals. Temperature: 23° C  As Figure 39 demonstrates, in the case of F4 coal, adsorption is very fast and equilibrium is reached within the first 5-10 minutes. On the other hand, LS43 coal still adsorbs humic acids even after 2 hrs of conditioning. It seems, however, that almost all humic  160  acids adsorb within the first hour. For this reason, the residual H A concentration in the actual adsorption tests was measured after 1 hr of conditioning in the shaker.  Equilibrium Humic Acids Concentration [mg/dm ] 3  Figure 40. Adsorption isotherms of humic acids onto silica, LS43 and F4 coals, T = 23° C  As Figure 40 demonstrates, the adsorption of H A on silica is low compared to the two coals but it continuously increases with the H A concentration. Adsorption on LS43 coal also steadily increases, while adsorption on F4 coal appears to reach a plateau. Clearly, adsorption on LS43 coal is the highest among the tested samples.  161  5.3.2  Adsorption of DTAB on Coals and Silica  The kinetics of D T A B adsorption on the coals and silica is shown in Figure 41. The adsorption density is expressed in micromoles per m . The initial D T A B concentration was 154 mg/dm (0.5 mmol/dm ). 3  3  As Figure 41 indicates, D T A B adsorbs very quickly on all the tested solids. The adsorption equilibrium is achieved within the first 5 minutes for all the samples. It is also important to note that LS43 coal adsorbed practically the entire amount of D T A B introduced into the adsorption mixture.  0  10  20  30 Time [min]  40  50  60  Figure 41. Adsorption kinetics of DTAB onto coals and silica. Initial DTAB concentration was 154 mg/dm (0.5 mmol/dm ), room temperature (23°C). 3  3  162  Figure 42 presents D T A B adsorption isotherms on the coals and silica. The D T A B adsorption isotherm on F4-oxidized coal at pH = 7.1-7.4 is also included. Additionally, the adsorption density of D T A B at the air/water interface, calculated from the Gibbs equation (see Appendix 9), is also shown. Equilirbium Concentration [mg/dm ] 10 100 1000 J i i i I • I 3  1x10"'  ^  I  10000  1  Trendline for adsorption of Dodecylamine on silica from de Bruyn (1955)  hio  1x10"'H o  •3  Q  — fK  o o  S 1x10"'  CO  a  o  ho.1 3*  & O co T3  < 1x10  H • • O O V  1x10"  n  1x10-  i—I I i i i i  T  I  F 4 , p H = 7.1-7.4 F4 Oxidized, p H = 7.1-7.4 LS43, p H = 8.3-8.6 Fine Silica, p H = 7.0-7.3 Air/Water Interface  I I I II I  T  i  i  lxlO" 1x10"' Equihbrium Concentration [mol/dm ] 4  i  r  b-0.01  r  lxl0"  :  3  Figure 42. Adsorption isotherms of DTAB onto coals, silica and at the air/water interface, T = 23°C.  If negligible porosity of the silica used in the experiments is assumed, the adsorption isotherm for silica will reflect the true adsorption density. On the other hand, the adsorption isotherms for the tested coal samples should be regarded as "apparent" since D T A B most  163  likely adsorbs only on the external surfaces of the coals. Thus, the true adsorption densities on the tested coals will actually be higher than those shown in Figure 42. In the case of LS43 coal, D T A B adsorption was complete below an initial D T A B concentration of 150 mg/dm . Only above this concentration, some residual D T A B could be 3  detected in equilibrium with the coal, but the equilibrium concentrations were below 10"  5  mol/dm (w3 mg/dm ). The first point on the LS43 isotherm was obtained at an initial D T A B 3  3  concentration of 270 mg/dm . Interestingly, F4-oxidized coal also gave a range of complete 3  adsorption. A n 80-mg/dm D T A B solution did not yield any D T A B remaining in the solution. 3  At 160 mg/dm , the equilibrium D T A B concentration was only 6.7 mg/dm («2.2-10" 3  3  5  mol/dm ) and this is the first point on the isotherm. 3  Neither F4 coal nor silica exhibited this kind of total adsorption, even at D T A B concentrations as low as 5 mg/dm . 3  Obviously, the adsorption of D T A B on LS43 is very high, some ten times higher than adsorption on F4 coal. However, after oxidation, D T A B adsorption on F4 coal increased to the level of LS43 coal and the two isotherms almost perfectly overlap. The adsorption density on silica is very low and it becomes equal to that of F4 coal at concentrations near the critical micelle concentration of D T A B . The adsorption density of D T A B at the air/water interface is clearly higher than the adsorption density at the silica/water interface. Interestingly, adsorption on F4 coal is comparable to adsorption at the air/water interface. Since some flotation tests were carried out at varying pH values (section 5.2.3), additional adsorption experiments were performed as a function of pH for F4 and F4-oxidized coals. A n initial D T A B concentration of 166 mg/dm was chosen to make sure that some 3  residual DTAB could be detected for the F4-oxidized coal samples. Also, the initial concentration 164  was not very different from the D T A B dosage of 1500 g/t in the flotation tests. Under the experimental conditions, the initial concentration of 166 mg/dm of D T A B corresponds to a 3  D T A B dosage of 1660 g/t in the flotation tests. The results are shown in Figure 43. As in the flotation tests, the carbonates present in LS43 coal effectively buffered pH of the reacting mixtures and prevented pH adjustments towards acidic values, unless large quantities of concentrated hydrochloric acid were used (e.g. 5 cm to bring pH to 4). This also 3  significantly changed the volume of the adsorption mixture and thus the concentration of D T A B . To complicate the matter even further, after adding so much HCI, high amounts of Ca  2+  and M g  cations would be released to solution and could strongly interfere with the  2 +  adsorption process. Therefore, only F4 and F4-oxidized coals were tested.  .  7_|  I  •  I  .  I  .  I  .  I  .  I  .  I  .  |_  Complete Adsorption Line  2  3  4  5  6  7 pH  8  9  10  11  Figure 43. Effect of pH on DTAB adsorption for F4 and F4-oxidized coals, T = 23° C.  165  The adsorption of D T A B onto F4 coal does not change significantly with pH. Only at strongly alkaline pH values, does the adsorption density slightly increase. In contrast, the adsorption on the oxidized coal depends strongly on pH. As pH decreases the adsorption density drops quite sharply and becomes comparable to the adsorption density on F4 coal in acidic solutions. In neutral solutions (pH « 7), the adsorption density on the oxidized coal is over 5 times higher than the adsorption density on F4 coal.  5.4 ZETA POTENTIAL MEASUREMENTS  The electrokinetic measurements were carried out to estimate the sign and magnitude of the surface charges on the coals and silica. Figure 44 shows the zeta potentials as a function of pH in 0.005N KC1 as the background electrolyte. It can be seen that F4 coal has an isoelectric point (i.e.p.) at pH = 5.4. After oxidation the i.e.p. of the oxidized coal shifts towards much more acidic values (pH « 2.2), which also appears to be equal to the i.e.p. for LS43 coal. Based on the experimental data, the i.e.p. for .silica can only be approximately estimated (by extrapolation) at around 1.8. The effect of D T A B on the zeta potential of the coals and silica is shown in Figure 45. In these tests, the pH was adjusted in all cases to the values equal to those from the flotation tests. The ionic strength was kept constant using a 0.005N KC1 solution.  166  '  1 1 1  I •'  1 1  I'  1 1 1  I '•  1 1  I'  1 1  'I  1 1 1 1  I  1 1 1 1  I  1 1  •• I •  1 1 1  •  1  2  3  4  5  6  7  8  I  1 I  1  1  F4  9  10  11  pH Figure 44. Zeta potential as a function of pH for the tested coals and silica. Numbers above the dashed line indicate estimated values of the i.e.p's.  As Figure 45 indicates, charge reversal quickly takes place on the F4 coal particles at D T A B concentrations as low as 2-10" mol/dm . In the same concentration range, the zeta potential of silica remains almost unaffected and the surface charge on silica reverses its sign at a D T A B concentration 100 times higher than that for F4 coal. Both LS43 and F4-oxidized coals showed almost the same response. A l l curves tend to converge on approximately the same zeta potential value as the D T A B concentration approaches the cmc.  167  Figure 45. Effect of DTAB on the zeta potentials of coals and silica.  168  5.5 WETTABILITY (CONTACT ANGLE) STUDIES  The effect of D T A B on contact angles on coals and quartz is shown in Figure 46. Both the advancing and receding contact angles are reported. D T A B Concentration [mol/dm ] 3  lxlO" on  ,  i  Water Only  i  i  I  lxlO"  5  i  i  i  I  lxlO"  4  i  lxlO"  3  i i i i i 111  i  i  i i i i 111—.—  0.001 0.01 0.1 D T A B Concentration [fraction of cmc] Quartz Plate • Advancing O Receding  LS43 Coal • Advancing O Receding  2  1  F4 Coal • Advancing • Receding  Figure 46. Contact angles on coals and quartz plate in the presence of DTAB.  Low concentrations of D T A B («0.001cmc) have a negligible effect on the wettability of F4 coal. However, the contact angles steadily decrease with increasing D T A B concentration, and eventually fall to zero at the cmc.  169  On the other hand, the contact angles on quartz quickly increase rendering the quartz surface as hydrophobic as the surface of F4 coal. Again, the contact angles on the quartz surface decline to zero near the cmc. The LS43 coal surface is naturally hydrophilic as indicated by the low contact angles measured in water. As the D T A B concentration is gradually raised, the surface seems to become even more hydrophilic but exhibits some increased degree of hydrophobicity at a D T A B concentration of two-tenths of the cmc. A l l surfaces become fully hydrophilic when the D T A B concentration approaches the cmc. Both the advancing and receding contact angles show qualitatively similar trends.  5.6 COAL OXIDATION 5.6.1  Alkali Extraction Tests  Figure 47 shows the results of controlled F4 coal oxidation as determined by the alkali extraction method of Lowenhaupt and Gray (1980, 1989). The transmittance (at 520 nm) of the alkaline extracts obtained after different oxidation times is plotted together with the equivalent Aldrich H A concentration. It can be found from Figure 47 that oxidation at 150°C is rather slow during the first 450 hrs and then increases exponentially over the last 200 hrs. The concentration of humic acids in the extract after 645 hrs was equal to the amount of humic acids leached from LS43 coal. Thus, this was the end point of the oxidation process.  170  0  100  0  100  200  300  400  500  200 300 400 500 F4 Coal Oxidation Time at 150 °C [hours]  600  700  600  700  Figure 47. Humic acids concentration and transmittance of alkaline extracts at different stages of F4 coal oxidation.  In order to better characterize the changes taking place on the F4 coal surface as a result of oxidation, several humic acids leaching tests were carried out at room temperature and at different pH values. For comparison, a series of such low temperature leaching tests was also performed on LS43 coal. The results are summarized in Figure 48. Obviously, F4-oxidized coal releases humic acids at much lower p H values than LS43 coal does. Even in weakly acidic and neutral solutions (pH = 5-7), there is always some amount of H A extracted from the F4-oxidized coal surface. LS43 coal is completely inert in the same pH range until strongly alkaline solutions, with pH =10 or higher, are used.  171  172  6 DISCUSSION 6.1 DISPERSING/DEPRESSING ACTION OF POLYMERS  In the preparation of coal-water slurries, the main focus is on obtaining a highly loaded product with a coal content of at least 65% (wt). At the same time, the viscosity and yield stress of such a concentrated suspension should be as low as possible to facilitate pipelining. As Figures 20-26 reveal, it is more difficult to obtain a low-viscosity coal-water slurry from a bituminous coal (e.g. F4 coal). Aggregation of coal particles resulting from hydrophobic forces increases the yield stress and apparent viscosity. C W S obtained from F13 coal exhibit much better rheological properties. Grimanis and Breault (1992) reported that lower rank CWS were inherently more stable towards settling and some types of subbituminous coals required no modifying agents at all. As Figure 27 shows, coal surfaces become strongly hydrophilic as indicated by the decreasing contact angles after adding humic acids. Humic acids also make the zeta potential of coal much more negative. In other words, in the presence of H A both hydrophilic and hydrophobic coals become similar in terms of their surface properties. It must pointed out that the equation of Seki et al. (1985) (Equation 52, section 3.2.3) for the free water content in suspensions was derived based on results obtained in the presence of 1% (per coal wt) of various dispersants. Additionally, in the work of Seki et al. the coal content was expressed per dry coal weight. After taking corrections for the moisture content, Seki et al. showed that the actual loading of CWS in the presence of 1% of additives is higher for higher rank coals. This situation can be better understood after examining the  173  raw data in Figure 26. In the presence of humic acids, the surface properties of F4 and F13 coals are similar. Also, the particle size distribution is the same for these two coals and thus it is not surprising that the yield stress values and apparent viscosities tend to fall on a single curve. However, i f a correction for the moisture content is taken, the actual coal loading will be lower for F13 coal (moisture content « 2.0%) compared with the dry coal content of F4 coal (moisture « 0.6%). The resulting changes in coal surface wettability and surface charge should have a pronounced effect on the aggregation/dispersion equilibria in concentrated coal suspensions. As follows from the modified D L V O theory (section 3.1.2), an overall interparticle repulsion can be induced in several ways: by increasing the surface charge (electrostatic repulsion), -  by introducing an adsorbing polymer (steric repulsion), and  -  by increasing the wettability of solid particles to eliminate hydrophobic attraction. For the hydrophobic F4 coal, characterized by a water advancing contact angle of 72-  75 degrees (Figures 27 and 46), the viscosity and the yield stress increase with coal content as a result of coal aggregation due to the hydrophobic attractive forces. When the yield stress of F4 coal suspensions was measured as a function of pH (Pawlik and Laskowski 1998), and hence electrostatic repulsion, its values remained constant even when p H was as low as 2.5 and the zeta potential was «42 mV. Clearly, electrostatic repulsion alone was unable to stabilize the system towards aggregation and the hydrophobic attractive forces dominated the force balance. F13 coal suspensions, on the other hand, seem to be dispersed not only through electrostatic repulsion - the zeta potential in water was -32 mV for this coal (Figure 27) - but  174  also due to the hydrophilic character of its surface characterized by an advancing contact angle of about 40 deg. It is known that lower rank/oxidized coals contain significant amounts of humic acids on their surfaces, while mature, high rank bituminous coals do not possess these "native" oxygen-containing compounds. As these results indicate, however, once humic acids adsorb onto the bituminous coal surface, both the surface charge (zeta potential) and surface wettability quickly increase (contact angle decreases) and, in the presence of humic acids, the bituminous coal surface resembles that of the oxidized/lower rank coal. This combined effect of the increased surface charge and wettability leads to a net repulsion between the coal particles which, in turn, brings about the observed decrease of the yield stress and viscosity of the coal-water suspensions at a given coal content. As humic acids may have molecular weights as high as 50,000 (Lawson and Stewart 1989), their steric contributions may also play a role in the dispersion mechanism. Since F13 coal particles are already hydrophilic and electrostatically charged, the addition of humic acids does not markedly alter the coal's surface properties, and thus, does not significantly affect the rheology of F13 coal-water suspensions. The induced hydrophilicity of F4 coal surface should be directly linked to the coal's flotation response. It is evident from Figure 31 that humic acids - as expected - depress the natural floatability of F4 coal. Since several other polymers also depressed the flotation of the coal, it may be concluded (using humic acids behavior as a reference) that these polymers also rendered the coal's surface hydrophilic. Moreover, their effect on the yield stress of F4 coalwater suspensions is qualitatively similar to that of humic acids within the same dosage range. It should also be pointed out that even non-ionic polymers, such as dextrin or hydroxyethyl cellulose, can both depress coal flotation and decrease the yield stress. Since these 175  macromolecules can not stabilize the coal particles through electrostatic repulsion, their dispersing action must result from the steric and wetting effects. The anionic polymers (polystyrene sulfonate and carboxymethyl cellulose) are more efficient coal dispersants due to the contribution from electrostatic repulsion, but their flotation response is not so clearly related to their electrostatic charge (Figure 31). In light of these results, it is obvious that coal particles in highly concentrated coalwater slurries must be hydrophilic and electrostatically/sterically dispersed to keep the yield stress and apparent viscosity at low levels. As mentioned in the earlier sections, fine coal beneficiation by flotation does not produce clean coal that would meet the above specifications. However, coal depression is the basis for coal reverse flotation and it seems that any potential coal depressants should also work as coal dispersants in the preparation of CWS from clean coal.  6.2 INTERACTION OF DTAB WITH COALS AND SILICA  In the work of Stonestreet and Franzidis (1988, 1989, 1992) on coal reverse flotation, it was postulated that quaternary amines can simultaneously act as coal depressants and mineral matter activators/collectors. This assumption was based on a simple model according to which D T A B adsorbs at the bituminous coal/water interface in a very oriented fashion with the hydrophobic chain attached to the hydrophobic coal surface and the hydrophilic group pointing towards the bulk solution. Such an orientation would effectively render the coal surface hydrophilic and depress its flotation. On the other hand, the adsorption of the quaternary amine on silica, according to Stonestreet and Franzidis, should take place through electrostatic interactions between the 176  negatively charged silica surface and the positively charged group of the amine cation. As a result, the amine molecules would be oriented with their hydrophilic groups attached to the silica surface, while their hydrophobic chains would point away from the silica/water interface rendering the silica surface hydrophobic. This simple picture was reinforced by the flotation results, such as those obtained in this thesis and shown in Figure 33. Stonestreet and Franzidis would argue that, for example, since LS43 coal floats poorly with D T A B , while silica floats well, a dosage of 500 g/t of D T A B should give a very selective flotation of silica from a LS43/silica mixture. The poor floatability of LS43 coal would thus be attributed to the depression of the coal by the above mechanism. The main problem with such an interpretation is that a typical bituminous coal, such as F4 in this work, floats very well with D T A B and practically no depression was observed in the concentration range considered by Stonestreet and Franzidis. In order to obtain the kind of response discussed by these two authors, one has to use a lower rank, naturally hydrophilic coal. It can be seen from Figure 33 that at 500 g/t of D T A B , F4 coal floats with over 90% yield while LS43 does not float at all. It seems, therefore, that the model for the depression of a hydrophobic coal was applied to explain the results obtained with a hydrophilic, lower rank coal. In fact, Stonestreet and Franzidis also observed some improved flotation of their coal at higher D T A B dosages (as in the case of LS43 coal in Figure 33 at dosages above 1500 g/t). They did not extend their work, though, to dosages as high as the cmc and were unable to provide any explanation for this trend.  177  Because of these literature discrepancies and the lack of fundamental information on the reverse flotation process, one of the main objectives of this thesis was to elucidate the role and mechanism of action of the quaternary amine, D T A B .  6.2.1  Adsorption of DTAB and Its Effect on Silica Surface Properties  As the adsorption isotherms show (Figure 42), the adsorption density of D T A B on silica is surprisingly low compared to the coals. It seems that the isotherm is S-shaped suggesting that adsorption takes place in four distinct stages, as discussed in section 3.3.3 (Figure 12). However, a comparison of the results for the silica/DTAB system with the data of de Bruyn (1955) for primary dodecyl-amine (DDA) indicates that D D A adsorption is much higher and the hemimicelle region raises more steeply for D D A (at C mme > 2 • 10" mol/dm ). 4  3  a  This observation suggests that the adsorption of these two amines on silica takes place through different mechanisms. It is recognized that in the case of primary amines, the presence of the molecular amine species in equilibrium with the cationic form enhances amine adsorption by screening the positive charges on adjacently adsorbing amine cations, and hence facilitates the formation of hemimicelles (Smith 1963, 1973, Smith and Lai 1966, Smith and Scott 1990). Since the quaternary amines are strong electrolytes, only the cationic form is present in solution and the formation of hemimicelles on the silica surface does not seem to be favorable. Using ACD-3D Viewer visualization software (Advanced Chemistry Development Inc., Toronto, Canada), the geometrical dimensions of D T A B molecules were calculated. The head group cross-sectional area was found to be 0.2865 nm . Assuming a cylindrical shape for 2  178  2  the amine chain, the area of the longitudinal section was estimated at 1.1144 nm . The software gave the C-C, C-H and C - N bond lengths of 0.154, 0.11 and 0.147 nm, respectively, values which are in excellent agreement with literature data (Morrison and Boyd, 1966). These two areas were then used to calculate the magnitude of adsorption corresponding to monolayer coverage for both the vertical and flat D T A B orientations on the surfaces. It is evident that the D T A B adsorption density on silica does not exceed monolayer coverage for the flat orientation which is in striking contrast to the traditional picture that D T A B molecules orient themselves vertically with their head groups at the surface and the hydrocarbon tails pointing towards the solution. One may argue, though, that the vertically oriented D T A B molecules are only scarcely distributed on the silica surface. Koopal and his coworkers (Goloub et al. 1996, Goloub and Koopal 1997, Koopal et al. 1999) showed, however, that the quaternary amine molecules (pyridinium derivatives) may indeed, under certain conditions, assume a flat orientation on silica surfaces. First of all, as pointed out by Goloub and Koopal (1997), silica surfaces possess some weakly hydrophobic siloxane sites that can attract the hydrocarbon chains of quaternary amines. Laskowski  and Kitchener (1969)  elegantly  showed,  through  contact  angle  measurements, that hydrophobic-hydrophilic transitions on silica were associated with the degree of hydration of silica surfaces. Even methylated silica could be rendered hydrophilic by prolonged conditioning in water, but its hydrophobicity could be easily restored by heating at 110°C to remove the weakly adsorbed water molecules. Griot and Kitchener (1965a,b) also observed that as the hydration of a silica surface progressed, the flocculation of silica fines with polyacrylamide (PAM) worsened. This was attributed to a competitive adsorption of P A M and water on the silanol sites. Strongly hydrated, or "aged", silica did not absorb P A M at all, and thus could not be flocculated.  179  Secondly, as discussed by Goloub and Koopal (1997), the lack of the hemimicelle region was observed only at low ionic strengths (0.001N KC1). Higher p H values increased the magnitude of adsorption but the characteristic S-shaped isotherms were still not observed. At such low salt concentrations, the adsorption density reached a plateau at »1 umol/m for dodecyl-pyridinium chloride which agrees very well with the "pseudoplateau" value for D T A B (Figure 42). When the electrolyte concentration was raised to 0.1N KC1, the adsorption isotherms appeared to follow the four-region model and the hemimicelles could form as indicated by the sharp increase in the slope of the adsorption isotherm. Goloub and Koopal (1997) proposed the following mechanism. At low ionic strength, electrostatic repulsion between the charged amine groups prevents the molecules from dense adsorption. Also, the weak localized hydrophobicity of the silica surface promotes a flat orientation of the adsorbed amine molecules. Only at concentrations near the cmc (C > «0.5cmc), does the adsorption density increase steeply as, in this region, complete micelles seem to adsorb on the silica surface. It must be remembered that the amine concentration at the surface, resulting from amine adsorption, is actually higher than the bulk concentration and "surface micellization" may occur at concentrations lower than the bulk cmc. At high salt concentrations, on the other hand, the electrostatic charge on the amine head group is effectively screened, the tail-to-tail lateral hydrophobic interactions start to play a role, and the surfactant molecules may now adsorb more closely. Although the initial adsorption density is lower at high ionic strengths than it is at low ionic strengths - the potassium cations also screen the negatively charged surface sites - it increases very rapidly once the formation of hemimicelles is initiated. Since the adsorption tests in this work were done in tap water, the low adsorption of D T A B on the SilcoSil silica seems to closely follow the adsorption mechanism described by 180  Goloub and Koopal (1997) for alkyl-pyridinium chlorides at low ionic strengths and neutral pH. Also, the zeta potential on silica particles reverses sign at a D T A B concentration of about 0.1-0.2cmc, which according to Goloub and Koopal (1997) indicates an onset of the tail-totail adsorption of D T A B on silica surfaces. A similar D T A B concentration range for the charge reversal on silica was obtained by Ter-Minassian-Saraga (1975). Interestingly, in this concentration range, the contact angles on quartz slides also start to sharply decrease suggesting that the silica surface becomes gradually hydrophilic as a result of the bilayer adsorption of D T A B . From simple geometrical considerations, it should also be pointed out that the nitrogen atom in dodecyl-trimethyl ammonium halides is in the geometric center of a tetrahedron with three methyl- and one dodecyl group in the four corners. The positive charge on the alkyltrimethyl ammonium ion is therefore partially screened by the four hydrocarbon radicals. Also, the tetrahedral head group may actually adsorb through any of the four planes leading, statistically speaking, to a flat orientation on the silica surface. Nevertheless, the presence of local hydrophobic sites would be an additional driving force for a flat orientation as typical hydrophilic solids are usually covered by layers of strongly adsorbed water molecules that would "repel" the hydrocarbon chains.  Flotation of Silica with D T A B  Despite a very low amine adsorption density at the water/silica interface, the silica surface becomes hydrophobic in the presence of D T A B at concentrations as low as 10" -10" 5  4  mol/dm . Under these conditions, the silica particles float very well.  181  Several researchers (Apian and de Bruyn 1963, Ter-Minassian-Saraga 1964, 1975, Somasundaran 1968, Somasundaran and Fuerstenau 1968, Digre and Sandvik 1968) pointed out that flotation results are often related only to collector adsorption at the water/solid interface, while adsorption at other interfaces, i.e. solid/air and water/air, is largely neglected. As observed by Somasundaran and Fuerstenau (1968), high flotation yields for silica with the use of primary amines may be obtained at concentrations much lower than those corresponding to the formation of hemimicelles. More recently, Koopal et al. (1999) also concluded that the observed increase in the hydrophobicity of silica in the presence of quaternary amines, as indicated by high contact angles and flotation recoveries, is not accompanied by the formation of hemimicelles. Consequently, Somasundaran (1968) showed, using the Gibbs and Young equations and experimentally determined interfacial tensions (solid/liquid and liquid/air), that a higher collector adsorption density at the air/solid interface compared with that at the solid/liquid interface is the necessary condition for an efficient bubble-particle attachment. Digre and Sandvik (1968) and Somasundaran (1968) also calculated that, in the case of the primary amine/silica system, amine adsorption at the silica/air interface is significantly higher, up to ten times, than adsorption at silica/water interface. Ter-Minassian-Saraga (1964, 1975) directly measured the adsorption of quaternary amines at the quartz/air interface. In her experiments, quartz slides were slowly withdrawn from solutions of C - and Br -labeled quaternary ammonium bromides, and the amount of 1 4  82  amine adsorbed on the dry slide surface was measured via the radioactivity of the slides. In this way, Ter-Minassian-Saraga showed that quaternary amines also preferentially adsorb at the silica/air interface and the transfer of the amine molecules from solution to the silica/air  182  interface takes place at the line of contact between silica, air and amine solution, as the slides are removed from solution. Eskilsson and Yaminsky (1998) carried out a study on the adsorption of hexadecyltrimethyl ammonium bromide (HTAB) onto silica. They essentially followed the slide retraction procedure of Ter-Minassian-Saraga and, through ellipsometric measurements (light polarization), also observed the preferential adsorption of H T A B at the air/silica interface. Eskilsson and Yaminski pointed out that the adsorption of the quaternary amine at the air/silica interface is already large starting from very small H T A B concentrations at which adsorption at the silica/water interface is negligible. As the results in Figure 42 reveal, the D T A B adsorption density at the air/water interface is indeed higher than the adsorption density at the silica/water interface. These two isotherms become comparable only when the D T A B concentration is higher than 0.1 cmc. In a way, the slide withdrawal procedure corresponds to a situation where an aminecovered air bubble collides with a silica particle and makes the water film recede from the particle surface. The main difference is the amine transfer rate between the interfaces under the quiescent conditions of a lab test, and the turbulent conditions of a flotation process (Digre and Sandvik 1968). This last conclusion is important from the flotation point of view. It implies that the hydrophobicity of silica in the presence of amines is actually induced by the introduction of air bubbles and the resulting deposition of amine molecules at the air/silica interface. It also explains why low concentrations of amines produce high contact angles, and hence result in good flotation. Since the contact angles can be measured only by attaching an air bubble to the solid surface, the presence of the air phase immediately initiates the transfer of the amine molecules between the interfaces and thus produces a high contact angle. This also means that 183  silica particles in dilute amine solutions are not hydrophobic per se, and that their apparent hydrophobicity manifests itself only i n the presence o f air bubbles. Such an explanation is consistent with the "bubble transfer" hypothesis put forward by Digre and Sandvik (1968) who postulated that, during a "real" flotation process, amine molecules are introduced onto silica surfaces by repeated collisions o f the amine-coated air bubbles with silica particles until the silica particles become sufficiently hydrophobic. Thermodynamic analysis by L i n and Somasundaran (1971) revealed that the free energy change o f amine molecules due to hemimicelle formation at the water/silica interface was o f the order o f -OJlkT. The respective change for the air/silica interface was equal to 1.01AT. This suggests that adsorption, and hemimicelle formation, at the air/silica interface are thermodynamically more favorable. Interestingly, the highest  advancing contact  angle was  obtained at a  DTAB  concentration (O.Olcmc) at which the zeta potential changed from -53 m V (in 0.005 N KC1) to -45 m V indicating that charge neutralization as a result o f D T A B adsorption was quite ineffective, and that D T A B adsorption itself was quite low. Nevertheless, such charged silica surfaces were still very hydrophobic. A s the contact angle and flotation results indicate, when the D T A B dosages approach the critical micelle concentration, the quartz/silica surfaces become hydrophilic and flotation is inhibited. A t such high amine concentrations, entire D T A B micelles adsorb on the silica surface to form a uniform, highly hydrophilic bilayer (Dobias 1986).  184  6.2.2  Adsorption of DTAB on F4 Bituminous Coal  Mutual hydrophobic attraction between the coal surface and the aliphatic hydrocarbon chain of the amine molecule seems to be the major component of the driving force for the adsorption of D T A B on a bituminous coal surface. The adsorbed D T A B molecules appear to assume a flat orientation on the coal surface for maximum attraction. Only near the cmc, does some rearrangement take place and the quaternary amine molecules adsorb as semicylindrical aggregates, as shown by Parachuri et al. (2001) through atomic force microscopy for hexadecyl-trimethyl ammonium bromide (HTAB) on a graphite surface. The flat amine orientation at lower concentrations does not markedly affect the hydrophobicity of the coal as indicated by the rather slowly changing values of the advancing contact angle. This is also confirmed by the fact that F4 coal flotation shows only a shallow minimum at very low D T A B dosages but as the amount of D T A B increases to about 0.1 cmc, complete coal recovery can still be achieved. This observation indicates that an advancing contact angle of approximately 40 deg (Figure 46) is sufficient for efficient flotation. The zeta potential on F4 coal undergoes the most dramatic changes in the presence of D T A B (Figure 45). According to Arbiter et al. (1975), the electrokinetic behavior of hydrophobic solids is primarily affected by the aqueous phase since the electrical double layer is entirely on the water side of the solid/water interface. The sign of the charge and its magnitude are determined by the composition of the aqueous phase with little or no influence from the "hydrophobic wall" - as i f the species in solution were "unaware" of the presence of a solid behind the water curtain. In other words, the hydrophobic solid/water interface resembles closely the air/water interface. This analogy is also supported by the results in Figure 42. The adsorption isotherms for F4 coal/water and air/water interfaces appear to 185  coincide, particularly at lower D T A B concentrations, indicating that the adsorption densities at these two interfaces are indeed similar. The electrokinetic results for DTAB/F4 coal system can be explained in a similar way. It is known that surfactants (DTAB) adsorb at the water/air interface with their hydrophobic chains at the surface and the charged, hydrophilic groups oriented towards the aqueous phase. When a quaternary amine solution is brought into contact with a hydrophobic solid, such as F4 coal, the hydrophobic surfactant chains act as anchors for the adhesion of the solution to the solid surface. The hydrophilic, positively charged head groups embedded in the aqueous phase would then determine the sign and magnitude of the electrokinetic potential on the "water side" of the solid/solution interface. Since the amine molecules would be the only adsorbing ions at the interface, very small quantities of D T A B would be sufficient to impart a positive charge to the coal surface. Despite the relatively low D T A B adsorption density on the F4 coal surface, the zeta potential increases very rapidly towards highly positive values. As the surface  density of the anchoring chains increases  with the  DTAB  concentration, the enhanced adhesion of the amine solution to the coal surface also results in systematic lowering of the contact angles. The effect of D T A B on the contact angles on F4 coal is in excellent agreement with the wettability data of Elton (1957) obtained for D T A B on octadecane (a model hydrophobic solid) shown in Figure 13B. As in the case of silica, F4 coal flotation is strongly depressed only at D T A B dosages approaching the critical micelle concentration when the hydrophilic micelles render the coal surface completely hydrophilic (contact angle drops to zero). Similarly, Parachuri et al. (2001) observed that a graphite surface turned hydrophilic near the cmc for H T A B (see Figure 13 A).  186  It should be pointed out that in the case of naturally hydrophobic solids, the condition of higher surfactant adsorption at the solid/air interface compared to the solid/water interface does not have to be met. Even when no adsorption occurs at the solid/air interface, the native hydrophobicity of the solid surface will still ensure its floatability. As discussed by Elton (1957), it is very likely that a layer of adsorbed surfactant molecules, oriented tail-to-surface, will be stripped off together with the receding liquid from the octadecane surface. Based on his adhesion tension calculations, Elton also implied that there is actually little adsorption, i f any, of quaternary amines at the hydrophobic solid/air interface (Elton assumed that the air/octadecane interfacial tension does not significantly change with the addition of DTAB). Therefore, in relation to the Young equation, Elton attributed the observed decrease of the contact angles on octadecane to the decrease of the solid/liquid and liquid/air interfacial tensions. Likewise, in the case of F4 coal/DTAB system, D T A B molecules adsorb primarily at the water/air and coal/water interfaces and thus decrease the respective interfacial tensions. Assuming Elton's reasoning' (Elton 1957), it may be postulated that the relatively high hydrophobicity of the F4 coal surface, despite the presence of moderate amounts of D T A B , is caused by a low/weak adsorption of the surfactant at the coal/air interface. This explains the good flotation of F4 coal over a wide range of D T A B dosages. This behavior of D T A B on the coal surface is in sharp contrast to its action on the silica surface. In the latter case, flotation is favorable when substantial amounts of D T A B molecules still remain adsorbed at the solid/gas interface even after rupture and recession of the surface water film. On a hydrophobic coal, however, the natural hydrophobicity of the surface overcomes the effects of low/weak surfactant adsorption and determines the flotation response of the coal. 187  In conclusion, the main role of D T A B in the F4/DTAB flotation system appears to be simply that of a frother. In the silica/DTAB system the quaternary amine acts as a typical collector.  6.2.3  Action of DTAB in Flotation of Low Rank/Oxidized Coals.  Surface Composition of Coals and Coal Oxidation  The alkali extraction test of Lowenhaupt and Gray (1980,1989), described in section 4.3.2, yields the entire amount of humic substances that can be leached from a given coal. Under the test conditions, LS43 coal and F4-oxidized coal released the same amounts of humic acids. In contrast, at room temperature and at gradually increasing pH, these two coals responded very differently (Figure 48). First of all, when the oxidized coal was placed in water, the pH changed from 5.8 to 3.8 indicating that the oxidized coal behaves as a weak acid. This is fully in line with the IR results showing the presence of carboxylic groups. Also, the isoelectric point for F4-oxidized coal is at pH = 2.2-2.4 characteristic of highly oxidized coals. As the pH is steadily increased by adding a I N NaOH solution (starting at pH = 3.8 for F4-oxidized coal), F4-oxidized coal releases humic salts at pH values as low as 5-6, while only highly alkaline solutions (pH > 10) start to react with the humic substances from LS43 coal. This indicates that the humic acids on the oxidized coal are much more leachable than the acids from LS43 coal. Based on the IR results (section 4.2.3), it may be concluded that the humic acids from F4-oxidized coal are present mainly on the coal surface, as coal oxidation seemed not to 188  proceed deep into the coal structure. As such, these surface acids are easily accessible and become quickly neutralized by even weakly alkaline solutions. LS43 coal is naturally rich in humic acids. For such low rank coals, exposed for thousands of years to high pressures and temperatures, humic substances ("oxygen containing groups" would be a more appropriate term) are part of the coal structure. Therefore, only highly alkaline solutions can destroy the structural bonds and free the humic salts. It should also be noted that the content of the acidic (carboxylic) groups on LS43 coal surface is rather low which would make its "humic acids" even less reactive. The dry oxidation of F4 coal also results in a significant decrease of its BET specific surface area from 1.85 to 0.88 m /g. The most reasonable explanation is that, during 2  oxidation, oxygen primarily attacks pores as their edges in particular are highly reactive. As a result, the pores are simply etched off to form larger cavities whose effective surface area is much smaller. Some evidence of this effect was observed during the BET measurements. The desorption peaks for the fresh F4 coal sample showed some elongated tails and steps that are characteristic of porous, energetically heterogeneous materials. After oxidation, the oxidized coal samples gave perfectly Gaussian desorption peaks (as did the silica and LS43 coal samples) suggesting that the porosity of F4 coal was significantly reduced.  Adsorption of Humic Acids  Further insight into the surface properties of F4 and LS43 coals can be obtained from the adsorption isotherms of humic acids (Figure 40). Unexpectedly, the highest adsorption density was observed for LS43 coal. The kinetic tests on this coal showed that even after 2 hrs  189  of conditioning, the adsorption of humic acids did not reach equilibrium although the initial adsorption rate was very high (Figure 39). F4 coal, on the other hand, reached adsorption equilibrium very quickly and the isotherm appears to exhibit a plateau. However, the adsorption density of humic acids on F4 coal is much lower than on LS43 coal. Generally, the shape of the isotherms for these two coals indicates a rather high affinity of humic acids towards both coals, although the hydrophilic, lower rank coal surface appears to be a much more "friendly" environment for adsorption. The adsorption process on LS43 coal is in a sense similar to a crystal growth. Since humic acids essentially form this coal, their adsorption on it may be considered as a gradual built-up of the coal structure. The "growth" would slowly continue for as long as residual humic acids (structural units) are present in solution. Such an interpretation agrees well with the experimental findings of Lai et al. (1989): coal-derived humic acids give higher adsorption densities than technical humic acids, as the former would naturally show a higher affinity towards the source coal surface. The F4 coal surface is a mixture of highly hydrophobic, polyaromatic and aliphatic components so, as opposed to LS43 coal, humic acids should exhibit a poorer affinity towards it. As opposed to D T A B , humic acids are very weak surfactants. They reduce the surface tension of water from 71.8 mJ/m to 68.7 mJ/m in a 400 mg/dm H A solution (see Appendix 2  2  3  5). In light of the "hydrophobic wall" concept of Arbiter et al. (1975) and the similarity to the air/water interface, as discussed for D T A B adsorption, it is somewhat surprising that humic acids can still adsorb at the water/F4 coal interface. This supports the idea of some chemical interaction involved in the adsorption mechanism.  190  Moreover, the strongly anionic humic acids despite their relatively low adsorption, sharply change the zeta potential of F4 coal particles towards very negative values as shown in Figure 27. Assuming that F13-oxidized and LS43 coals are equivalent in terms of their surface properties, it is quite obvious that the higher adsorption of humic acids on the more hydrophilic coals does not result in an equally dramatic drop of the zeta potential. This suggests that a hydrophilic coal/water interface is not as sharply defined as the hydrophobic wall of F4 coal. As discussed by Tadros and Lyklema (1968) and Lyklema (1968), the "surface" charges at such poorly defined, often porous interfaces are located not only in a two-dimensional array in the solid surface and adjacent liquid layer, but also in a volume phase of finite thickness.  Adsorption of D T A B on LS43 and F4-Oxidized Coals  As the adsorption isotherms in Figure 42 demonstrate, the D T A B adsorption density is about 10 times higher on LS43 and F4-oxidized coals compared to the adsorption density on the hydrophobic F4 coal. As simple geometrical considerations indicate, the adsorbed D T A B molecules appear to be perpendicular to the coal surface. There was a range of initial D T A B concentrations (up to 150 mg/dm ) in which complete adsorption took place on LS43 and F43  oxidized coals (no D T A B in equilibrium). As Figure 42 suggests, this initial complete adsorption seems to produce approximately a monolayer of amine molecules. Because the adsorption density slowly increases even further, it appears that additional molecular layers can still develop on the surfaces of these two coals. It is evident from Figure 43 that D T A B adsorption on F4-oxidized coal strongly depends on pH, while adsorption on fresh F4 coal is practically constant over the entire pH 191  range. This observation immediately indicates that the adsorption mechanisms are very different for the two coals. As the electrokinetic measurements show (Figure 44), the particles of F4-oxidized coal are negatively charged above p H « 2.2 and their surface charge quickly increases towards more negative values as the p H is gradually raised. Carboxylic and other oxygen functional groups become gradually ionized with increasing p H to produce negative sites on the oxidized coal surface. Since F4 coal becomes negatively charged only above pH = 5.4, it would seem that the cationic surfactant should easily adsorb in the pH ranges where the surfaces of the coals are negatively charged. While this is true for F4-oxidized coal, it is not so for the fresh coal. Moreover, F4 coal adsorbs the amine even when its surface is positively charged and when electrostatic repulsion between the amine head group and the surface should prevent any adsorption. In other words, the amine adsorption density on the F4 coal surface does not depend on the sign and magnitude of the surface charge and therefore the adsorption mechanism does not involve electrostatic interactions. Such adsorption results are the evidence of hydrophobic forces acting between the hydrocarbon chains of D T A B and the hydrophobic coal surface. In contrast, the changes in D T A B adsorption density on F4-oxidized coal closely follow the electrokinetic potential-pH curve. The adsorption density quickly decreases as the zeta potential values approach zero at the i.e.p.. Obviously, electrostatic attraction between the cationic head group and the negatively charged surface governs the adsorption mechanism. The D T A B adsorption density on F4-oxidized coal at pH = 3.8 can be calculated based on the observation that the addition of the coal to water results in a decrease of the pH of the flotation pulp from 5.8 to 3.8. Assuming that every hydrogen ion released into solution leaves a single negatively charged site on the coal surface, the adsorption density of D T A B on 192  the surface should be equal to the surface concentration of the ionized sites. Simple calculations show that, under the conditions of the adsorption tests, the adsorption density on the oxidized coal should amount to 1.80 umol/m (at pH 3.8) which is in remarkable 2  agreement with the isotherm in Figure 43. The high adsorption density of D T A B on the oxidized coal and the presence of the complete adsorption range indicate a strong interaction between the surface sites and the cationic molecules. The above calculation suggests that some relatively strong acidic groups, most likely the carboxylic ones, are responsible for the negative charge and the magnitude of adsorption on the oxidized coal surface. It is also noteworthy that the analytical method for determination of amines (section 4.3.5) is based on a reaction between an anionic dye and a cationic amine and subsequent formation of a 1:1 (amine:dye) complex. The method utilizes the high natural affinity of cationic amines towards ionizable acidic groups. Interestingly, the adsorption results for F4 and F4-oxidized coals are mirrored by the flotation responses of the coals in the presence of 1500 g/t of D T A B at varying pH (Figure 34). Flotation yields of F4 coal are very high over the entire pH range. Simultaneously, D T A B adsorption is constant and relatively low. Only under highly alkaline conditions, does the flotation yield slightly decrease. The flotation of F4-oxidized coal is good only in the acidic environment where D T A B adsorption is low. As the pH increases, the D T A B adsorption density increases as well and this is accompanied by a dramatic fall of the flotation yields. In neutral flotation pulps, F4-oxidized coal does not float at all. It should also be pointed out that in the case of very hydrophobic surfaces, the hydrophobic attractive forces are known to be much larger than the van der Waals dispersion forces (Xu and Yoon 1989, 1990). The weak dependence on pH of F4 coal flotation is very characteristic for highly hydrophobic materials. In the case of hydrophilic and charged 193  surfaces, the effect of electrostatic forces predominates and maximum hydrophobicity, and hence floatability, is observed around the isoelectric point. As Figure 34 demonstrates, while the flotation of the very hydrophobic F4 sample does not depend on pH, the floatability of the F4-oxidized sample dramatically improves around the i.e.p. (pH = 2.2). Several important observations were made during the flotation experiments. In the case of F4 coal flotation, frothing was excellent throughout the tests regardless of pH. For F4oxidized coal, the froth thickness and its strength systematically declined with increasing pH. There was almost no froth present at pH = 7. At a D T A B dosage of 1500 g/t, the flotation of LS43 coal was also very poor although the natural pH in this case was slightly higher («8.5). At this amine dosage, D T A B adsorption on LS43 coal is still complete and there is no amine in equilibrium with the coal. Frothing is also very poor as in the case of F4-oxidized coal at pH = 7. Only at D T A B dosages higher than 1500 g/t, above the limit for complete amine adsorption, does LS43 coal flotation markedly increase. It appears that the flotation of low rank/oxidized coals is influenced by the amount of free amine in the flotation system, which in turn depends on the magnitude of amine adsorption, and hence on pH. It should be pointed out that in the case of F4 coal and silica there was always some residual amine present in the system since the adsorption of D T A B on these two substrates was much lower, and no region of complete adsorption was actually observed. However, the lack of amine to provide good frothing does not satisfactorily explain the results. As the contact angle data show, LS43 coal remains hydrophilic over a wide range of D T A B concentrations. Some weak hydrophobicity on the surface of LS43 coal is induced at about 0.2cmc (Figure 46), as indicated by the increasing contact angles, and this region appears to correspond with the narrow floatability window for LS43 coal in Figure 33. It 194  seems, therefore, that in order to initiate flotation the coal surface must be to some extent rendered hydrophobic in addition to the presence of free amine molecules in the system. According to Sun (1954b) and Wen and Sun (1977), primary amines act as coal collectors in the flotation of oxidized/low rank coals. Primary amines were assumed to adsorb through electrostatic attraction between the positively charged head groups and the negatively charged coal surface. As these researchers suggested, such an amine orientation should render the coal surface hydrophobic, analogously to the amine/silica system. The results of some additional flotation tests shown in Table 11 indicate that the above explanation is generally true. Table 11. Flotation ofLS43 coal in the presence of DTAB and MIBC. LS43 + 1500 g/t of LS43 only DTAB 100 200 M I B C Dosage [g/t] 100 200 Flotation Yield (2 min) [%] 33.5 3.7 7.7 23.9  It can be seen from Table 11 (see also Figure 32) that LS43 coal floats poorly with M I B C as a frother. However, after adding 1500 g/t of D T A B followed by M I B C , the yields increased significantly suggesting that the coal surface became mildly hydrophobic as a result of complete amine adsorption. At 1500 g/t, D T A B alone would be unable to improve the flotation of LS43 coal. Only when D T A B (a collector) and M I B C (a frother) are added together, does the flotation yield start increasing. Despite very high amine adsorption and the supposed head-to-surface orientation of D T A B molecules on the surface of LS43 coal, the flotation yields in Table 11 still seem to be much lower than expected. Such dense amine adsorption should theoretically produce a strongly hydrophobic surface, and yet, as the results of contact angle measurements show, and using the flotation of F4 coal as a reference, the highest hydrophobicity of LS43 coal in  195  D T A B solutions is nowhere near that of a typical metallurgical coal. It can be seen from Figure 32 that, without any additives, F4 coal gives a flotation yield of over 30% in tap water. Neither M I B C nor D T A B (up to the limit of complete adsorption) can separately produce such results for LS43 coal. This leads to the conclusion that although the mechanism of adsorption involves electrostatic interactions, the final orientation of amine molecules on the hydrophilic coal surface cannot be so simply inferred. Leja (1957, 1982) in his review on the flotation of sulfide minerals, pointed out that although oxidation of sulfides causes a deterioration of their floatability with the use of xanthates, some unusual results were sometimes observed. These can be briefly summarized as follows: a) heavily oxidized sulfides show an ever-increasing abstraction of xanthates from solution, b) only unusually large additions of xanthates (3-5 lb/t) may greatly improve recoveries of the oxidized minerals. Although Leja's observations concerned systems very different from coal, the similarities to the flotation of heavily oxidized/low rank coals are quite striking. Firstly, as the results of this thesis show, coal oxidation also leads to a dramatic increase of D T A B adsorption. Secondly, LS43 coal floats only when "unusually large additions" of D T A B (>1.5 kg/t) are made. Leja (1982) used the following mechanism to explain the flotation of oxidized sulfides. Strong chemisorption of xanthate-type collectors on surfaces of oxidized sulfides takes places on the most active sites forming isolated patches. Because of the strong chemical interactions between the adsorbing species and the surface atoms in the active sites, the chemical bonds between the adsorption sites and the rest of the solid are broken leading to a weakening of adhesion between the surface patches and the internal solid structure. Under the dynamic conditions of flotation, these heavily disturbed areas are simply "peeled o f f from 196  the surface and only the collector adsorption "products" may be floated. However, when the collector addition is increased to such an extent that the multilayer patches of adsorbed collector are no longer isolated but a continuous layer envelopes each particle, the floatability of the particles improves. The reason for this is that the initially weak adhesion to the solid is now strengthened by lateral bonds between the collector molecules existing within the continuous layer surrounding the particle. Despite the fact that these lateral bonds are only van der Waals bonds, they are sufficiently numerous to be capable of withstanding the disruptive hydrodynamic forces acting on the particle itself or on the particle-bubble aggregate. There are also other strong indications that a similar mechanism applies to the DTAB/low  rank (or oxidized) coal system. Undoubtedly, any coal surface is highly  heterogeneous so the formation of isolated patches is certainly plausible. The mechanism of D T A B adsorption on low rank/oxidized coals also involves strong chemical interactions between the ionized acidic groups and the positively charged ammonium cations. The adsorption density at higher amine dosages corresponds to multilayer coverage. Petrographic studies on coal show (Gray and Lowenhaupt 1989, Stach 1982) that even dry low rank/oxidized coals inherently exhibit microscopic features consistent with physical and chemical degradation of their surfaces (cracks, fractures, discolorations, presence of fines and secondary minerals). Such surfaces may become chemically unstable in solution, as evidenced by the gradual leaching of humic acids from F4-oxidized coal. Moreover, the earlier mentioned model of the "gel-like interface" of Tadros and Lyklema (Tadros and Lyklema 1968, Lyklema 1968) would supplement well such a mechanism of amine adsorption on a highly heterogeneous coal surface. The zeta potential values on F4-oxidized and LS43 coals increase rather slowly with the D T A B concentration  197  (Figure 4 5 ) despite the very high D T A B adsorption density. According to Tadros and Lyklema, charge neutralization within such an extended interface takes place randomly at different "depths" of the gel layer, so D T A B molecules would have to penetrate (or diffuse into) the entire volume of the interface to completely neutralize and then reverse the charge. In the case of low rank/oxidized coals, this poorly defined interface may be pictured as consisting of long, strongly hydrated, anionic, polymeric species (humic acids) branching out from the porous, cracked coal surface. D T A B adsorption on such "dangling" polyelectrolyte chains and fractured surfaces would obviously result in a highly chaotic orientation of the amine molecules generating only a limited hydrophobicity of the surface. Once the charged, hydrophilic groups are neutralized, and some water molecules are expelled from the surface by D T A B adsorption, the lateral van der Waals bonds between the amine hydrocarbon chains would be "sufficiently numerous" to strengthen the surface layer. The lower hydration of the interface resulting from the presence of the weak van der Waals forces would render the surface hydrophobic, at least to some extent. As the zeta potential and contact angle measurements indicate, the maximum hydrophobicity of the L S 4 3 coal surface occurs in the D T A B concentration range where the coal is slightly positively charged. This indicates that charge neutralization already took place in the whole volume of the interface, and additional D T A B molecules adsorbed through tailto-tail hydrophobic attraction. This moment marks also the beginning of lateral interactions between the adsorbed D T A B molecules and the formation of a reinforced surface layer. It is understandable that due to the chaotic orientation of the initially adsorbed D T A B molecules, some of the additional molecules would have their head groups oriented away from the surface rendering it positively charged.  198  It is interesting to note that although there are significant differences in the surface composition of LS43 and F4-oxidized coals, the flotation, adsorption and electrokinetic responses of the two coals are essentially identical.  6.2.4  DTAB versus Humic Acids - Their Behavior at Different Interfaces  The experimental results indicate that both humic acids and D T A B exhibit certain common features in terms of their action at different interfaces. These similarities can be itemized as follows: a) Even though the adsorption of both D T A B and H A at the hydrophobic coal/water interface is low, their effect on the zeta potential is very dramatic. Low concentrations of both additives quickly change the zeta potential values towards more positive or negative values for D T A B and H A , respectively. b) In contrast, the much higher adsorption of D T A B and H A on low rank/oxidized coals does not affect the zeta potential so strongly. Only very high concentrations of these reagents produce results comparable with the effect on the hydrophobic coal. c) Both D T A B and H A affect the wettability of the F4 coal surface while their effect on the contact angles on the hydrophilic coal is moderate at best. It appears that all these trends are related to the structure of the coal/water interface. Generally, the poorly developed hydrophilic coal/water interface is not "amenable" to any modification by either D T A B or humic acids. The sharply defined hydrophobic coal/water interface (F4), on the other hand, responds "instantaneously" to the presence of very small concentrations of these two additives.  199  One aspect of the behavior of humic acids and D T A B deserves an additional comment. Both the anionic H A and the cationic D T A B affect the hydrophobic coal/water interface in a qualitatively similar manner and yet only humic acids can act as coal depressants in coal flotation. Humic acids are strongly hydrated molecules, as are other coal depressants such as dextrin. Even low concentrations of these hydrophilic macromolecules on a hydrophobic coal surface can stabilize the surface water film and render the surface hydrophilic. In contrast, D T A B is quite hydrophobic due to the presence of the long hydrocarbon chain. Its adsorption at a hydrophobic coal/water interface does not, therefore, affect coal floatability so dramatically. As the surface tension data show (Appendix 5), D T A B sharply reduces the water surface tension which, according to the Gibbs equation, is equivalent to a high adsorption density at the air/water interface (Figure 42, Appendix 9). The adsorption of D T A B at the air/water interface appears to be only slightly higher than the adsorption density at the F4 coal/water interface. The tendency to adsorb on the surfaces of air bubbles gives D T A B its frothing capabilities. The effect of humic acids on the surface tension is very weak indicating that humic acids do not adsorb at the air/water interface. The coal/water interface is the primary adsorption "area" for humic acids. Even upon introduction of air bubbles to the pulp, these strongly hydrated macromolecules remain on the coal surface. Therefore, typical depressants should selectively adsorb only at the coal/water interface. The magnitude of this adsorption seems to be of secondary importance.  200  6.3 COAL REVERSE FLOTATION  As Figures 35,36 and 37 show, the separation of silica from coal in reverse flotation is a kinetic process. Only the first concentrates collected after 1-2 minutes contain significant amounts of silica as indicated by their high ash contents. The initial flotation of silica is followed by the flotation of coal, which leads to the "dilution" of the concentrates with clean coal after longer flotation times. It can be seen from Figure 35 that for F4/silica mixtures, when F4 coal is not depressed with humic acids, the initial ash contents do not exceed 60%. When humic acids are added (400 g/t) and coal is rendered hydrophilic, the ash contents in the concentrates increase to about 75% (Figure 36). This indicates that this process requires coal depression for efficient separation. When the naturally hydrophilic LS43 coal is used no depressant is needed and the ash contents in the initial concentrates reach 80% (Figure 37). However, prolonged flotation in this case also leads to a dramatic loss of selectivity. In all cases, the highest ash content is attained when the flotation yield is about 30-50%, when essentially the whole amount of silica has already been floated. Higher yields are inevitably accompanied by the decrease of selectivity. Two clear trends are visible in Figure 38. As the coal surface becomes more hydrophilic in the order F4<(F4 + 400 g/t HA)<LS43, higher and higher dosages of D T A B are needed to initiate flotation. At the same time, the highest ash content attainable for the tested coal/silica mixtures also increases as the coal in the mixture becomes more hydrophilic. True reverse flotation, with high ash contents in the concentrates, can only be obtained in very narrow ranges of D T A B concentration. 201  Interestingly, the cumulative yield-DTAB dosage curves for the coal/silica mixtures in Figure 38 follow closely the trajectories of the yield curves for respective coals from Figure 33. This trend is most pronounced for LS43 coal (Figure 33) and the LS43/silica mixture (Figure 38). Significant amounts of solids in both cases can be floated only when the D T A B dosage is higher than about 1.5 kg/t. As in the case of the LS43/DTAB system, the poor flotation of LS43/silica mixtures was accompanied by the lack of frothing suggesting absence of free amine in the pulp. Since D T A B adsorption on silica is very low compared to adsorption on LS43 coal, it may be concluded that the amount of residual amine in solution is determined by the magnitude of adsorption on the coal. Because the adsorption of D T A B on LS43 coal is so extensive that the coal is capable of completely adsorbing D T A B from solution, it can be deduced that as increasing quantities of D T A B are introduced into the pulp, the amine preferentially adsorbs on LS43 coal and not on silica. Only when the coal surface is saturated with D T A B , can the free amine molecules adsorb on silica. Since even low adsorption on silica renders its surface highly hydrophobic, it is silica that selectively floats at the onset of flotation. In the presence of D T A B , silica is always more hydrophobic than LS43 coal as can be seen from the contact angle results. These conclusions are further supported by the results from Appendix 10. It can be seen that when D T A B is given time (15 min) to adsorb, the surfactant will first adsorb on LS43 coal and then on silica as discussed above. However, when aeration is started first followed by the addition of D T A B , even low dosages of the amine seem to more selectively pick out silica as indicated by the higher ash contents in the concentrates. Also, at a dosage of 800 g/t, the zero conditioning time gives a higher yield while after 15 minutes of conditioning there is no flotation at all. These observations suggest that immediately after D T A B addition,  202  its adsorption can actually proceed onto silica. At longer conditioning times, however, the high affinity of D T A B towards LS43 coal takes over and the initial selectivity is lost. The addition of 400 g/t of H A to depress F4 coal in a mixture with silica shifts the yield curve towards higher D T A B dosages. In the presence of humic acids the surface of F4 coal "looks" like the surface of a low rank/oxidized coal. The coal particles become hydrophilic and highly negatively charged. Hence, such a modified coal would adsorb the cationic amine more strongly than the original coal. At low D T A B dosages no measurable flotation takes place and it seems that the onset of flotation is again determined by the coal's interaction with D T A B . Without depressants, F4 coal floats well with D T A B , as does silica. It is not surprising then that separation from a mixture is poor. D T A B adsorption on both components of the mixture is relatively low, there is always excess amine present to generate froth, and neither component seems to preferentially adsorb the amine. Poor selectivity results from good flotation of both constituents. Generally, the best separation conditions occur at such D T A B dosages that the hydrophobicity of silica is higher than that of coal. This requirement is met when a small amount of free amine is present in the system. However, the preferential chemisorption of D T A B on the surfaces of hydrophilic coals leaves residual amounts of amine in solution only when the coal surface is already weakly hydrophobic (Table 11). A small excess of D T A B to provide moderate frothing creates the narrow window for kinetic separation - the more hydrophobic silica floats much better than the weakly hydrophobic coal. Higher amine doses, of about 0.1-0.2cmc, increase the coal's hydrophobicity even further and the more intense frothing completely masks any differences in the wettability of the two components.  203  7 CONCLUSIONS  The rheology of concentrated coal-water suspensions is influenced by coal surface properties. Suspensions of high rank, bituminous coals exhibit high yield stresses and apparent viscosities due to the hydrophobic aggregation of coal particles. The obtained results suggest a correlation between the yield stress of CWS and water contact angle on coal. In contrast, lower rank/oxidized coals produce slurries with much lower yield stresses and apparent viscosities. The rheological properties of suspensions of lower rank/oxidized coals can be attributed to the better wettability by water of the surfaces of such coals. Coal flotation which makes the floating coal particles even more hydrophobic is not an appropriate cleaning method for CWS preparation. Humic acids, strongly anionic polyelectrolytes, can be used to modify coal surface properties. The adsorption of humic acids on a hydrophobic coal renders its surface strongly hydrophilic and electrostatically charged. As a result, in the presence of humic acids both the yield stress and apparent viscosity of bituminous coal suspensions decline. In the presence of humic acids, the surfaces of high rank coals closely resemble the surfaces of low rank coals. The effect of humic acids on the wettability and surface charge of coal surfaces is, in terms of the D L V O theory, characteristic of dispersants. Since even nonionic polymers can reduce the yield stress of coal-water suspensions, it may be concluded that coal dispersants should primarily affect the wettability of a coal surface by rendering it strongly hydrophilic. This eliminates hydrophobic attraction as the dominant interparticle force controlling the rheology of coal-water suspensions prepared from bituminous coals. The increase of the surface charge density is of secondary importance although is highly advantageous.  204  Coal dispersants can also be used as coal depressants in the coal reverse flotation process. Quaternary amines, such as dodecyltrimethyl ammonium bromide (DTAB), act as collectors for silica. D T A B adsorbs at all interfaces of the flotation system, i.e. air/silica, air/water and water/silica interfaces. However, it is the adsorption of D T A B at the air/silica interface that renders the silica surface hydrophobic and thus facilitates its flotation. The adsorption mechanism involves a fast deposition of D T A B molecules on air bubbles and subsequently at the air/silica interface as the water film ruptures and recedes upon collisions with air bubbles. Adsorption at the water/silica interface is too low to induce the observed high hydrophobicity of silica at low amine concentrations. In contrast to primary amines, the adsorption of D T A B at the silica/water interface does not involve the formation of hemimicelles under neutral pH values and low ionic strength in the flotation pulp. Electrostatic repulsion between the positively charged head groups prevents the molecules from adsorbing at high-density. Under certain conditions, a weak hydrophobicity of the silica surface may enhance the probability of a flat orientation of amine molecules at the water/silica interface. Contrary to some literature  reports, it has been shown in this work that  dodecyltrimethyl ammonium bromide cannot be simultaneously used as a coal depressant and silica activator (collector). D T A B is unable to depress the flotation of hydrophobic coals, such as F4. The contact angle measurements show that the bituminous coal surface becomes gradually hydrophilic in the presence of D T A B . However, the resulting hydrophilicity is not sufficient to depress coal flotation over a wide range of D T A B dosages. Only near the critical micelle concentration of D T A B , can the adsorbing micelles finally render the coal surface hydrophilic. 205  The separation of a hydrophobic coal from a mixture with silica is impossible when only D T A B is used. Both coal and silica are hydrophobic in the presence of D T A B and thus their flotation with D T A B is not selective. Naturally hydrophilic coals strongly adsorb D T A B through chemical interactions between the oxygen functional groups on the coal surface and the positively charged ammonium head group of the amine. D T A B adsorption, even when complete, cannot render the coal strongly hydrophobic due to a highly chaotic orientation of the amine molecules on such a surface. This poorly ordered adsorption results from a gel-like, hydrated structure of the coal/water interface in which the adsorption sites are distributed in the entire volume of the often porous and extended interface. The separation of silica from low rank coals, such as LS43 coal (or coals depressed with humic acids), is a kinetic process occurring in a very narrow range of D T A B dosages. In a coal/silica mixture, D T A B preferentially and completely adsorbs on the coal leaving no free amine to activate silica. Due to the mechanism of amine adsorption on such coals, the appearance of residual amine for activation of silica takes place when the coal is already to some extent hydrophobic. Since both the coal and silica can separately float under these conditions, it is the more hydrophobic component that selectively floats from a mixture. As contact angle measurements show, silica is more hydrophobic than LS43 coal at any D T A B concentration and therefore silica floats first. However, at too high amine concentrations frothing is too intense, coal becomes more floatable and the selectivity of reverse flotation drops dramatically. Additives, such as D T A B , adsorbing at the air/water interface as well, are unable to depress the flotation of a hydrophobic coal. These will merely act as frothers at dosages below their critical micelle concentrations. 206  8 RECOMMENDATIONS FOR FUTURE W O R K  Further studies should be extended on a wider range of quaternary amines with various hydrocarbon chain lengths. Coals of intermediate ranks should also be tested as the depressing action of these amines appears to be very rank-specific. In order to substantiate the conclusions regarding the mechanism of action of amines, more research is needed on carefully selected model systems. The role of quaternary amines in the flotation of low rank coals should be investigated in more detail due to the apparent complexity of the phenomena involved. "Real" coals should eventually be tested and the effect of liberation of mineral matter should be closely looked at. The composition of mineral matter is also a factor to be dealt with since the various coal minerals are known to exhibit different floatabilities with amines as collectors. The effect of hydrodynamic conditions certainly deserves special attention. Aeration rate, solids content in the pulp and the mixing rate are very important factors. Staged reagent addition should also be employed. As the results from Appendix 10 show, the conditioning time with quaternary amines plays a role in the kinetics of the reverse flotation process. Dynamic surface tension measurements and adsorption studies in the presence of gaseous phase (bubbles) should yield more information on such time-dependent effects. The use of column flotation would be highly advantageous, especially because the separation is strongly kinetic. Column flotation also effectively eliminates the entrainment of gangue fines by the addition of wash water to the top of the column. Column flotation would thus be ideal whenever fine grinding was required to fully liberate the coal minerals. Because  207  about 70% of feed is recovered in "forward" coal flotation, the column carrying capacity is commonly a limiting factor, and for this reason flotation columns have not found much use in conventional coal flotation. In contrast, this concentrate-to-tailings ratio is reversed in the reverse mode so the columns could be easily operated at higher throughputs. In light of the experimental results, it appears that depressants other than humic acids should also be investigated. The amine dosages required for flotation are extremely high due to the strong interactions of the cationic surfactants with the anionic surface groups. Suitable blinders, such as cationic dextrins, could block these negatively charged sites on coal surfaces and hence would direct amine molecules to adsorb only onto coal mineral matter. Dextrins, however, are also strong depressants for carbonates (calcite, dolomite), and since these minerals are commonly associated with coal this aspect must also be considered.  208  9 REFERENCES Adam N . K . and Pankhurst G.A., 1946; "The Solubility of Some Paraffin Chain Salts", Discussions of the Faraday Society, vol.42, p.523. Anacker E.W., 1970; "Micelle Formation of Cationic Surfactants in Aqueous Media", in "Cationic Surfactants", E.Jungermann (Ed.), Chapter 7, Marcel Dekker, New York, p.203. Apian F.F. and de Bruyn P.L., 1963; "Adsorption of Hexyl Mercaptan on Gold", Transactions SME, vol.226, p.235. Apian F.F., 1987: ""Fine Coal Preparation - Its Present Status and Future", in "Fine Coal Processing", S.K.Mishra, R.R.Klimpel (Eds.); Noyes Publications, Park Ridge. Chapter 1, p.1-18. Apian F.F. and Arnold B.J., 1991; "Flotation", in "Coal Preparation", J.W.Leonard and B.C. Hardringe (Eds.), 5-th Edition, SME, Littleton, p.450. Arbiter N . , Fujii U . , Hansen B . and Raja A., 1975; "Surface Properties of Hydrophobic Solids", in "Advances in Interfacial Phenomena of Particulate/Solution/Gas Systems; Applications to Flotation Research", P.Somasundaran and R.B.Grieves (Eds.), American Institution of Chemical Engineers, vol.71, p. 176. A S T M Standard D-388-77, "Classification of Coals by Rank". A S T M Standard D-3172, "Standard Practice for Proximate Analysis of Coal and Coke". A S T M Standard D-3173, "Standard Test Method for Moisture in the Analysis Sample of Coal and Coke". A S T M Standard D-3174, "Standard Test Method for Ash in the Analysis Sample of Coal and Coke from Coal". A S T M Standard D-3175, "Standard Test Method for Volatile Matter in the Analysis Sample of Coal and Coke". Attia Y . A . , Y u S. and Vecci S., 1987; "Selective Flocculation Cleaning of Upper Freeport Coal With a Totally Hydrophobic Flocculant", in "Flocculation in Biochemistry and Separation Systems", Y.A.Attia (Ed.), Elsevier, Amsterdam, p.547. Attia Y . A . and Y u S., 1987; "Production of Super-Clean Coal Slurries From the HighSulfur Coals by Selective Flocculation", in "Processing and Utilization of HighSulfur Coals II", Y.P.Chugh, R.D. Caudle (Eds.), Elsevier, Amsterdam, p.402.  209  Bailey N.T., 1964; "Studies in the Chemistry of Humic Acids", Ph.D. Thesis, University of Birmingham. Barnes M . A . , Barnes W.C. and Bustin R . M . , 1984; "Chemistry and Evolution of Organic Material", Geoscience Canada, vol.11, no.3, p.103. Bashford M . T . and Woolley E . M . , 1985; "Enthalpies of Dilution of Aqueous Decyl-, Dodecyl-, Tetradecyl-, and Hexadecyltrimethyl Ammonium Bromides at 10, 25, 40 and 55 °C", Journal of Physical Chemistry, vol.89, p.3173. Baughman G.L., 1975; "Synthetic Fuels Data Handbook", Cameron Engineers, Denver CO. Bellamy L.J., 1975; "The Infrared Spectra of Complex Molecules", 3-rd Edition, Vol.1, Chapman & Hall, London. Bennett M . C . and Abram J.C., 1967; "Adsorption From Solution on the Carbon and Hydroxyapatite Components of Bone Char", Journal of Colloid and Interface Science, vol.23, p.513. Bensley C.N., Swanson A.R. and Nicol S.K., 1977; "The Effect of Emulsification on the Selective Agglomeration of Fine Coal", International Journal of Mineral Processing, vol.4, p.173. Blom L . , Edelhausen L . and van Krevelen D.W., 1957; "Chemical Structure and Properties of Coal Oxygen Groups in Coal and Related Products", Fuel, vol.36, p.135. Bozano L., Bozano S., Holmberg K . , Bozano P. and Ferraras G., 2000; "Production of C W F with Low Ash and High Percentage Solids", 25-th International Conference on Coal Utilization and Fuel Systems, Clearwater, FL, U S A , March 2000. Brady G.A. and Gauger A . W . , 1940; "Properties of Coal Surfaces", Industrial and Engineering Chemistry, vol.32, p. 1599. Brunauer S., Emmett P. and Teller E., 1938; "Adsorption of Gases in Multimolecular Layers", Journal of the American Chemical Society, vol.60, p.309. de Bruyn P.L., 1955; "Flotation of Quartz by Cationic Collectors", Transactions AIME, vol.202, p.291. Capes C.E., 1989; "Liquid Phase Agglomeration: Process Opportunities for Economic ad Environmental Challenges", in "Challenges in Mineral Processing" K.V.S.Sastry and M.C.Fuerstenau (Eds.) SME, Littleton, p.237.  210  Capes C.E., 1991; "Oil Agglomeration Process, Principles and Commercial Application for Fine Coal Cleaning", in "Coal Preparation", J.W.Leonard and B.C. Hardringe (Eds.), 5-th Edition, SME, Littleton, p. 1020. Cases J.M., Goujon G. and Smani S., 1975; "Adsorption of n-Alkylamine Chlorides on Heterogeneous Surfaces", in "Advances in Interfacial Phenomena of Particulate/Solution/Gas Systems; Applications to Flotation Research", P.Somasundaran and R.B.Grieves (Eds.), American Institution of Chemical Engineers, vol.71, p. 100. Cases J . M . and Villieras F., 1992; "Thermodynamic Model of Ionic and Nonionic Surfactant Adsorption-Abstraction on Heterogeneous Surfaces", Langmuir, vol.8, p.1251. Cassie A.B.D. and Baxter S., 1944; "Wettability of Porous Surfaces", Transactions of the Faraday Society, vol.40, p.546. Cassie A.B.D., 1948; "Contact Angles", Discussions of the Faraday Society, vol.3, p . l 1. Casson N . , 1959; " A flow equation for pigment-oil suspensions of the printing ink type", in "Rheology of Dispersed Systems", C.C.Mill (Ed.), Pergamon Press, New York, p.84. Castro S.H., Vurdela R . M . and Laskowski J.S., 1986; "The Surface Association and Precipitation of Surfactant Species in Alkaline Dodecylamine Hydrochloride Solutions", Colloids and Surfaces, vol.21, p.87. Castro S.H. and Laskowski J.S., 1988; "Application of the Dodecylaminium Ion Surfactant Selective Electrode in Oxide Flotation Research", in "Developments in Mineral Processing 9 - Froth Flotation", Proceedings of the Second Latin American Congress on Froth Flotation, Concepcion, Chile, Aug. 1985, S.H.Castro-Flores, J.A.Alvarez (Editors), Elsevier, p. 141. Chander S., Wie J . M . and Fuerstenau D.W., 1975; "On Native Floatability and the Surface Properties of Naturally Hydrophobic Solids", in "Advances in Interfacial Phenomena of Particulate/Solution/Gas Systems; Applications to Flotation Research", P.Somasundaran and R.B.Grieves (Eds.), American Institution of Chemical Engineers, vol.71, p.183. Chander S. and Apian F.F., 1989; "Surface and Electrochemical Studies in Coal Cleaning", Final Report to the U.S. Department of Energy, D O E PC/80523-T11 (DE90007603). Chong J.S., Christiansen E.B., Baer A.D., 1971; "Rheology of concentrated suspensions", Journal of Applied Polymer Science, vol.15, p.2007.  211  Churaev N . V . , Derjaguin B.V., 1985; "Inclusion of structural forces in the theory of stability of colloids and films", Journal of Colloid and Interface Science, vol.103, no.2, p.542. Churaev N.V., 1995; "The Relation Between Colloid Stability and Wetting", Journal of Colloid and Interface Science, vol.172, p.479. Claesson P . M . , Blom C.E., Herder P.C., Ninham B.W., 1986; "Interaction between water-stable hydrophobic Langmuir-Blodgett monolayers on mica", Journal of Colloid and Interface Science, vol.114, no.l, p.234. Claesson P.M., Herder P.C., Rutland M.W., Waltermo A . and Anhede B., 1992; "Amine Functionalized Surfactants - pH effects on Adsorption and Interaction", Progress in Colloid & Polymer Science, vol.88, p.64. Clunie J.S., Ingram B.Y., 1983; "Adsorption of Nonionic Surfactants", in "Adsorption from Solution at the Solid/Liquid Interface", G.D.Parfitt, C.H.Rochester (Eds.), Academic Press, p. 105-152. Connor P. and Ottewill R.H., 1971; "The Adsorption of Cationic Surface Active Agents on Polystyrene Surfaces", Journal of Colloid and Interface Science, vol.37, no.3, p.642. Cook A . C . , 1981; "What Are We Trying to Separate ?", Separation Science and Technology, vol.16, p.1545. Dai Q. and Laskowski J.S., 1991; "The Krafft Point of Dodecylammonium Chloride; pH Effect", Langmuir, vol.7, p. 1361. Davidson R . M . , 1982; "Molecular Structure of Coal", in "Coal Science". M.L.Gorbaty, J.W.Larsen, I.Wender (Eds.), Academic Press, New York, Vol.1. Debye P., 1949; "Light Scattering in Soap Solutions", Ann. N.Y. Acad. Sci., vol.51, p.575. Derjaguin B . V . , Landau L . D . , 1941; "Theory of the stability of strongly charged lyophobic sols and the adhesion of strongly charged particles in solutions of electrolytes", Acta Physicochimica URSS, vol.14, p.633. Derjaguin B.V., 1989; "Theory of Stability of Colloids and Thin Films", translated from Russian by R.K. Johnston, Plenum Publishing. Derjaguin B.V., Churaev N.V., 1989; "The current state of the theory of long-range surface forces", Colloids and Surfaces, vol.41, p.223. Digre M . and Sandvik K . L . , 1968; "Adsorption of Amine on Quartz Through Bubble Interaction", Transactions IMM, sec.C, vol.77, p.C61.  212  Dobias B., 1986; "Adsorption, Electrokinetic and Flotation Properties of Minerals Above the Critical Micelle Concentration", in "Phenomena in Mixed Surfactant Systems", Chapter 16, p.216, J.F.Scamehorn (Ed.), A C S Symposium Series vol.311, American Chemical Society, Washington DC. Dooher J.P. et al, 1985; "Rheological Evaluations of Proprietary and Generic Coal Water Fuels", Seventh International Symposium on Coal Slurry Fuel Preparation and Utilization. New Orleans, May 1985. Douglas H.W. and Shaw D.J., 1957; "Electrokinetic Studies on Model Particles", Transactions of the Faraday Society, vol.53, p.512. Drelich J., Laskowski J.S., Pawlik M . and Veeramasuneni S., 1997; "Preparation of a coal surface for contact angle measurements", Journal of Adhesion Science and Technology, vol. 11, no. 11, p. 1399. Dryden I.G.C., 1963; "The Chemistry of Coal Utilization", H.H.Lowry (Ed.), Suppl. Volume, p.232, Wiley, New York. Dubois M , Gilles A . , Hamilton J.K., Rebers P.A. and Smith F., 1956; "Colorimetric Method for Determination of Sugars and Related Substances", Analytical Chemistry, vol.28, no.3, p.350. Dupre A , 1869; "Theorie Mechanique de la Chaleur", Gauthier-Villars, Paris, p.369. Eggenberger D.N. and Harwood H.J., 1951; "Conductometric Studies of Solubility and Micelle Formation", Journal of the American Chemical Society, vol.73, p.3353. Elton G.A.H., 1957; "The Adsorption of Cationic Surface Active Agents by Silica and by Octadecane", in "Electrical Phenomena and Solid/Liquid Interface", J.H.Schulman (Ed.), Proceedings of the Second International Congress of Surface Activity, vol.Ill, Butterworths, London, p. 161. Elyashevitch M . G . , 1941; "Contact Angles as a Criterion of Coal Floatability", Transactions of Donetsk Industrial Institute, Gosgortiekhizdat, no.32, p.225. (in Russian) Elyashevitch M . G . , Ofengenden M . E . and Zubkova J.N., 1966; "Some physicochemical Surface Properties of Donetsk Coals Affecting Their Floatability", Rozrabotka Mestorozdeni Polyeznykh Iskopaemykh, Tekhnika, Kiev, no.6, p.5. (in Russian). Elyashevitch M.G., Zubkova J.N. and Kucher R.V. 1967; "On Hydration and Natural Floatability of Donbass Coals", Khimia Tverdogo Topliva, vol.3, p. 126. (in Russian).  213  Eskillson K . and Yaminsky V . V . , 1998; "Deposition of Monolayers by Retraction from Solution: Ellipsometric Study of Cetyltrimethylammonium Bromide Adsorption at Silica-Air and Silica-Water Interfaces", Langmuir, vol.14, p.2444. Evanko C.R., Delisio R.F., Dzombak D.A., Novak J.WJr., 1997; "Influence of Aqueous Solution Chemistry on the Surface Charge, Viscosity and Stability of Concentrated Alumina Dispersions", Colloids and Surfaces A, vol.125, p.95. Evans D.F., Allen M . , Ninham B.W. and Fouda A , 1984; "Critical Micelle Concentrations for Alkyltrimethyl Ammonium Bromides in Water From 25 to 160 °C", Journal of Solution Chemistry, vol.13, no.2, p.87. Everett D.H., 1988; "Basic Principles of Colloid Science", The Royal Society of Chemistry, London. Eveson G.P., 1961; "Removing Shale Particles from Coal or From Coal-Washing Effluent by Froth Flotation", British Patent 863,805. Fan A . , Somasundaran P. and Turro N.J., 1997; "Adsorption of Alkyltrimethylammonium Bromides on Negatively Charged Alumina", Langmuir, vol.13, p.506. Farris R.J., 1968; "Prediction of the Viscosity of Multimodal Suspensions from Unimodal Viscosity Data", Transactions of the Society of Rheology, vol.12, no.2, p.281. Ferrini F., Battarra V . , Donati E. and Piccinini C , 1984; "Optimization of Particle Grading For High Concentration Coal Slurry", paper B2, 9-th International Conference on Hydraulic Transport of Solids in Pipes, Rome, Italy, October 1984. Fleer G.J. and Scheutjens J.M.H.M., 1993; "Modeling Polymer Adsorption, Steric Stabilization and Flocculation", in "Coagulation and Flocculation", Surfactant Science Series 47, p.209-263, B.Dobias (Ed.), Marcel Dekker. Flory P.J., 1953; "Principles of Polymer Chemistry", Cornell University Press, Ithaca, NY. Firth B.A., 1976; "Flow properties of coagulated colloidal suspensions. II. Experimental properties of the flow curve parameters", Journal of Colloid and Interface Science, vol.57, no.2, p.257. Firth B.A., Hunter R.J., 1976a; "Flow properties of coagulated colloidal suspensions. I. Energy dissipation in the flow units", Journal of Colloid and Interface Science, vol.57, no.2, p.248. Firth B.A., Hunter R.J., 1976b; "Flow properties of coagulated colloidal suspensions. III. The elastic floe model", Journal of Colloid and Interface Science, vol.57, no.2, p.267.  214  Fowkes F . M . , 1964; "Attractive Forces at Interfaces", Industrial and Chemistry, vol.56, no. 12, p.40.  Engineering  Fowkes F . M . , 1967; "Attractive Forces at Solid Liquid Interfaces", in "Wetting", Society of Chemical Industry Monograph no.25, Gordon and Breach, New York, p.3. Friedel R.A. and Queiser J.A., 1959; "Ultraviolet-Visible Spectrum and the Aromaticity of Coal", Fuel, vol.38, p.369. Friend J.P. and Kitchener J.A., 1973; "Some Physico-Chemical Aspects of the Separation of Finely-Divided Minerals by Selective Flocculation", Chemical Engineering Science, vol.28, p. 1071. Fuerstenau D.W., 1956; "Streaming Potential Studies on Quartz in Solutions of Aminium Acetates in Relation to the Formation of Hemi-Micelles at the Quartz-Solution Interface", Journal of Physical Chemistry, vol.60, p.981. Fuerstenau D.W., 1957; "Correlation of Contact Angles, Adsorption Density, Zeta Potentials, and Flotation Rate", Transactions AIME, vol.208, p.1365. Fuerstenau D.W., Healy T.W. and Somasundaran P, 1964; "The Role of the Hydrocarbon Chain of Alkyl Collectors in Flotation", Transactions AIME, vol.229, p.321. Fuerstenau D.W., 1970; "Interfacial Processes in Mineral/Water Systems", Pure Applied Chemistry, vol.24, no.l, p.135. Fuerstenau D.W., Rosenbaum J.M. and Laskowski J.S., 1983; "Effect of Surface Functional Groups on the Flotation of Coal", Colloids and Surfaces, vol.8, p.153. Fuerstenau D.W., Venkataraman K.S. and Velamakanni B.V., 1985; "Effect of Chemical Addtives on the Dynamics of Grinding Media in Wet Ball M i l l Grinding", International Journal of Mineral Processing, vol.15, p.251. Fuerstenau D.W., Kapur P.C. and Velamakanni B., 1990; " A Multi-Torque Model for the Effects of Dispersants and Slurry Viscosity on Ball Milling", International Journal of Mineral Processing, vol.28, p.81. Fuerstenau D.W., 2001; "Excess Nonequilibrium Collector Adsorption and Flotation Rates", Minerals & Metallurgical Processing, vol.18, no.2, p.83. Fuerstenau M . C . , 1982; "Chemistry of Collectors in Solution", in "Principles of Flotation", R.P.King (Ed.), Chapter 1, p . l , South African Institute of Mining and Metallurgy, Johannesburg. Gaudin A . M . and Fuerstenau D.W., 1955; "Quartz Flotation with Cationic Collectors", Transactions AIME, vol.202, p.958.  215  Gaudin A . M . , Miaw H . L . and Spedden H.R., 1957; "Native Floatability and Crystal Structure", in "Electrical Phenomena and Solid/Liquid Interface", J.H.Schulman (Ed.), Proceedings of the Second International Congress of Surface Activity, vol.Ill, Butterworths, London, p.202. Gerrens H . and Hirsch G., 1975; "Critical Micelle Concentration", in "Polymer Handbook", J.Brandrup and E.H.Immergut (Eds.), Wiley, vol.11, p.482. Goloub T., Koopal L.K., Bijsterbosch B.H. and Sidorova M.P., 1996; "Adsorption of Cationic Surfactants o Silica. Surface Charge Effects", Langmuir, vol.12, p.3188. Goloub T. and Koopal L . K . , 1997; "Adsorption of Cationic Surfactants on Silica. Comparison of Experiment with Theory.", Langmuir, vol.13, p.673. Good R.J., Srivasta N.R., Islam M . , Huang H.T.L. and van Oss C J . , 1990; "Theory of the Acid-Base Hydrogen Bonding Interactions, Contact Angles, and the Hysteresis of Wetting: Application to Coal and Graphite Surfaces", Journal of Adhesion Science and Technology, vol.4, p.607. Good R.J., Badgujar M . K . , Huang T.L.H. and Kulkarni S.N.H., 1994; "Hydrophilic Colloids and Elimination of Inorganic Sulfur from Coal: A Study Employing Contact Angle Measurements", Colloids and Surfaces A, vol.93, p.39. Gregory J., 1987; "Flocculation by Polymers and Polyelectrolytes", in "Solid Liquid Dispersions", Th.F.Tadros (Ed.), p. 163, Academic Press. Grimanis M.P. and Breault R.W., 1992; "Effect of Coal Type and Beneficiation Process on Storage and Handling", Proceedings of 17-th International Conference on Coal Utilization and Slurry Technology, Clearwater, p. 19. Griot O. and Kitchener J.A., 1965a; "Role of Surface Silanol Groups in the Flocculation of Silica by Polyacrylamide. Part I - Chemistry of the Adsorption Process", Transaction of the Faraday Society, vol.61, p.1026. Griot O. and Kitchener J. A., 1965b; "Role of Surface Silanol Groups in the Flocculation of Silica by Polyacrylamide. Part II - Surface Changes of Silica Suspensions on Ageing", Transaction of the Faraday Society, vol.61, p. 1032. Groppo J.G. and Parekh B.B., 1996; "Surface Chemical Control of Ultra Fine Coal to Improve Dewatering", Coal Preparation, vol.17, p. 103. Groszek A.J., 19''5; Far.Disc.Chem.Soc. vol.59, p.109. Guo J., Tiu C , Hodges S. and Uhlhlerr P.H.T., 1999; "Hydrothermal-Mechanical Upgrading of Brown Coal", Coal Preparation, vol.21, no.l, p.35.  216  Gutierrez-Rodriguez J.A., Purcell R.J.Jr. and Apian F.F., 1984; "Estimating the Hydrophobicity of Coal", Colloids and Surfaces, vol.12, p . l . Gutierrez-Rodriguez J.A. and Apian F.F., 1984; "The Effect of Oxygen on the Hydrophobicity and Floatability of Coal", Colloids and Surfaces, vol.12, p.27. Hamaker H.C., 1937; "The London-van der Waals Attraction Between Spherical Particles", Physica, vol.IV, no.10., p.1058. Hamieh T and Siffert B., 1991; "Determination of Point of Zero Charge and Acid-Base Superficial Coal Groups in Water", Colloids and Surfaces, vol. 61, p.83. Harvey R.D. and Ruch R.R., 1986; "Mineral Matter in Illinois and Other US Coals", in "Mineral Matter and Ash in Coal", K.S.Vorres (Ed.), A C S Symposium Series, vol.301, Washington DC. Harwell J.H., Schechter R. and Wade W.H., 1985a; "Recent Developments in the Theory of Surfactant Adsorption on Oxides", in "Solid-Liquid Interactions in Porous Media". J.M.Cases (Ed.), Technip, Paris, p.371. Harwell J.H., Hoskins J.C., Schechter R. and Wade W.H., 1985b; Langmuir, vol.1, p.251. Hashimoto N . , 1999; " C W M : Its Past, Present and Future", Coal Preparation, vol.21, p.3. Hayashi K . , Yamada S., Toriya S. and Tahara H . , 1993; "Effect of Molecular Distributions of Dispersants for Slurryability", Japan-China-Australia Joint Symposium on Preparation and Transportation of Coal Slurries, Japan, p. 101. He Y . B . and Laskowski J.S., 1992; "Contact Angle Measurements on Discs Compressed From Fine Coal", Coal Preparation, vol.10, p.19. Henry D.C., 1931; "The Cataphoresis of Suspended Particles", Proceedings of the Royal Chemical Society of London A, vol.133, p. 106. Hoerr C.W., McCorkle M.R. and Ralston A . W . , 1943; "Studies on High Molecular Weight Aliphatic Amines and Their Salts. X . Ionization Constants of Primary and Symmetrical Secondary Amines in Aqueous Solution", Journal of the American Chemical Society, vol.65, p.328. Hoffman R.L., 1992; "Factors affecting the viscosity of unimodal and multimodal colloidal dispersions", Journal of Rheology, vol.36, no.5, p.947. Hogg R., 1999; "Polymer Adsorption and Flocculation", in "Polymers in Mineral Processing", Proceedings of the Third U B C - M c G i l l Bi-Annual International  217  Symposium on Fundamentals of Mineral Processing, J.S. Laskowski (Ed.), Metallurgical Society of CTM, p.3. Holuszko M.E., 1994; "Washability Characteristics of British Columbia Coals", Province of British Columbia - Ministry of Energy, Mines and Petroleum Resources, Mineral Resources Division, Paper 1994-2. Horsley R . M . and Smith H.G., 1951; "Principles of Coal Flotation", Fuel, vol.30, p.54. Hower J.C. and Parekh B.K., 1991; "Chemical/Physical Properties and Marketing", in "Coal Preparation". J.W.Leonard and B.C. Hardringe (Eds.), 5-th Edition, S M E , Littleton, p . l . Huckel E., 1924; "Die Kataphorese der Kugel", Physikalische Zeitschrifte, vol.25, p.204. Hunter R.J., 1981; "The Zeta Potential in Colloid Science", Academic Press, London. Hunter R.J., 1982; "The Flow Behavior of Coagulated Colloidal Dispersions", Advances in Colloid and Interface Science, vol.17, p. 197. Hunter R.J., 1987; "Foundations of colloid science". Volumes I and II, Clarendon Press, Oxford. Hunter R.J. 1993; "Introduction to Modern Colloid Science", Oxford University Press Inc., New York. Hunter R.J., Frayne J., 1979; "Couette Flow Behavior of Coagulated Colloidal Suspensions. TV. Effect of Viscosity of the Suspension Medium", Journal of Colloid and Interface Science, vol.71, no.l, p.30. Hyunh L . , Jenkins P., Ralston J., 1999; "Rheological Properties of Metal Sulfide Slurries", in "Polymers in Mineral Processing", Proceedings of the Third U B C M c G i l l Bi-Annual International Symposium on Fundamentals of Mineral Processing, J.S. Laskowski (Ed.), Metallurgical Society of CEVI, p.525. Igarashi T., Kiso N . , Hayasaki Y . , Ogata T., Fukuhara K . and Yamazaki S., 1984; "Effects of Weathering of Coals on Slurryabilities", Proceedings of the 6-th International Symposium on Coal Slurry Combustion and Technology, p.283, U.S.Dept. of Energy, Orlando. Ihnatowicz A., 1952; "Studies of Oxygen Groups in Bituminous Coals", Bulletin of the Central Research and Mining Institute, no. 125, Katowice (in Polish). Israelachvili J.N., Pashley R . M . , 1984; "Measurement of the Hydrophobic Interaction Between Two Hydrophobic Surfaces in Aqueous Electrolyte Solution", Journal of Colloid and Interface Science, vol.98, no.2, p.500.  218  James A . M . , 1981; "Electrophoresis of Particles in Suspension", in "Surface and Colloid Science", E.Matijevic (Ed.), Plenum Press, New York, vol.11, p.121. Johnson S.B., Franks G.V., Scales P.J., Boger D . V . and Healy T.W., 2000; "Surface Chemistry-Rheology Relationships in Concentrated Mineral Suspensions", InternationalJournal of Mineral Processing, vol.58, p.267. Kaji R., Muranaka Y . , Otsuka K . , Hishinuma Y . , Kawamura T., Murata M . , Takahashi Y . , Arikawa Y . , Kikkawa H., Igarashi T. and Higuchi H., 1983; "Effects of Coal Type, Surfactant and Coal Cleaning on the Rheological Properties of Coal Water Mixture", Proceedings of the 5-th International Symposium on Coal Slurry Combustion and Technology, vol.1, p251, U.S.Dept. of Energy, Tampa. Kaji R., Muranaka Y . , Otsuka K . and Hishinuma Y . , 1986; "Water Absorption by Coals: Effects of Pore Structure and Surface Oxygen", Fuel, vol.65, p.288, 1986. Kapur P.C., Scales P.J., Boger D . V . and Healy T.W., 1997; "The Yield Stress of Suspensions Loaded with Size Distributed Particles", AIChE Journal, vol.43, p.1171. Kawatra S.K. and Eisele T.C., 1997; "Pyrite Recovery Mechanisms in Coal Flotation", International Journal of Mineral Processing, vol.50, p. 187. Keller D.V., 1984; "The Separation of Mineral Matter From Pittsburgh Coal by Wet Milling", in "Chemistry of Mineral Matter and Ash in Coal", A C S Division of Fuel Chemistry, vol.29, no.4, p.326. Keller D.V.Jr., 1987; "The Contact Angle of Water on Coal", Colloids and Surfaces, vol.22, p.21. Killmann E. and Eisenlauer J., 1982; "The Stability of Silica-(Aerosil)-Hydrosols Under The Influence of Polymer Adsorption", in "The Effect of Polymers on Dispersion Properties", Th.F.Tadros (Ed.), Academic Press, p.221. King J.G., Maries M . B . and Crossley H.E., 1936; "Formulae for the Calculation of Coal Analyses to a Basis of Coal Substance Free From Mineral Matter", Journal of the Society of Chemical Industry, vol.55, p.277T Kitchener J.A., 1969; "Colloidal Minerals: Chemical Aspects of Their Dispersion, Flocculation and Filtration", Filtration and Separation, vol.6, no.5, p.553 Kitchener J.A., 1972; "Principles of Action of Polymeric Flocculants", British Polymer Journal, vol.4, p.217. Klassen W.I., 1966; "Coal Flotation", Wydawnictwo Slask, Katowice, 1966, Translated from Russian by J.S. Laskowski (Polish text).  219  Klein B., 1992; "Rheology and Stability of Magnetite Dense Media", Ph.D. Thesis, The University of British Columbia, Vancouver, Canada. Klein B., Laskowski J.S. and Partridge S.J., 1995; " A New Viscometer for Rheological Measurement on Settling Suspensions", Journal of Rheology, vol.39, no.5, p.827. Klein B., 2001; "Observations on the Rheopectic Properties of Nickel Laterite Suspensions", in "Polymers in Mineral Processing", Proceedings of the Third U B C - M c G i l l Bi-Annual International Symposium on Fundamentals of Mineral Processing, J.S. Laskowski (Ed.), Metallurgical Society of C I M , p.279. Klimpel R.R., 1982; "Slurry Rheology Influence on the Performance of Mineral/Coal Grinding Circuits", Mining Engineering, p. 1665, December 1982. Klimpel R.R., 1983; "Slurry Rheology Influence on the Performance of Mineral/Coal Grinding Circuits - Part II", Mining Engineering, p. 21, January 1983. Koopal L.K., Goloub T., de Keizer A . and Sidorova M.P., 1999; "The Effect of Cationic Surfactants on Wetting, Colloid Stability and Flotation of Silica", Colloids and Surfaces A, vol.151, p.15. Krieger I.M. and Dougherty T.J., 1959; " A Mechanism for Non-Newtonian Flow in Suspensions of Rigid Spheres", Transactions of the Society of Rheology, vol. Ill, p.137. Kumada K., 1987; "Chemistry of Soil Organic Matter" Japan Scientific Societies Press, Tokyo, and Elsevier, Amsterdam. Labuschagne B.C.J., 1986; "Relationship Between Oil Agglomeration and Surface Properties of Coal: Effect of pH and Oil Composition", Coal Preparation, vol.3, p.l. Lai R.W., Stone L . C . and Rimmasch B.E., 1984; "Effect of Humus Organics on the Flotation Recovery of Molybdenite", International Journal of Mineral Processing, vol.12, p. 163. Lai R.W., Wen W.W. and Okoh J.M., 1989; "Effect of Humic Substances on the Flotation Response of Coal", Coal Preparation, vol.7, p.69. Larsen J.W. and Kovac J., 1978; "Polymer Structure of Bituminous Coal", in "Organic Chemistry of Coal", J.W.Larsen (Ed.), A C S Symposium Series, vol.71, Washington DC. Laskowski J.S. and J.A. Kitchener, 1969; "The Hydrophilic-Hydrophobic Transition on Silica", Journal of Colloid and Interface Science, vol.29, no.4, p.670.  220  Laskowski J.S. and Konieczny E., 1970; " A n Attempt at the Determination of AirOxidation of Coal by Adsorption of Surfactants", Archiwum Gornictwa, vol.15, p.29 (English text). Laskowski J.S. et al., 1985; "Desulfurizing Flotation of Eastern Canadian High Sulfur Coal", in "Processing and Utilization of High Sulfur Coals", p.247, Y.A.Attia (Editor), Elsevier, 1985. Laskowski J.S., Sirois L . L . and Moon K.S., 1986; "Effect of Humic Acids on Coal Flotation. Part I. Coal Flotation Selectivity in the Presence of Humic Acids", Coal Preparation, vol.3, p. 133. Laskowski J.S., 1987; "Coal Electrokinetics: The Origin of Charge at Coal/Water Interface", Preprints of A C S Storch Award Symposium, Denver CO, April 5-10, 1987, vol.32, no.l,p.367. Laskowski J.S. and Walters A.D., 1987; "Coal Preparation", in Encyclopedia of Physical Science and Technology, vol.3, p.38, Academic Press. Laskowski J.S., 1988; "Dispersing Agents in Mineral Processing", in "Developments in Mineral Processing 9 - Froth Flotation", Proceedings of the Second Latin American Congress on Froth Flotation, Concepcion, Chile, Aug. 1985, S.H.Castro-Flores, J.A.Alvarez (Eds.), Elsevier. Laskowski J.S. and Parfitt G.D., 1989; "Electrokinetics of Coal-Water Suspensions", in "Interfacial Phenomena in Coal Technology", G.D.Botsaris and Y.M.Glazman (Eds.), Surfactant Science Series, vol.32, Marcel Dekker, p.279. Laskowski J.S., He Y . B . and Zhan Y , 1992; "Coal Wettability and Its Correlation with Floatability", S M E Annual Meeting, February 24-27, Phoenix A Z , Preprint 92125. Laskowski J.S., 1994; "Coal Surface Chemistry and Its Role in Fine Coal Beneficiation and Utilization", Coal Preparation, vol.14, p . l 15. Laskowski J.S., 1994a; "Flotation of Potash Ores", in "Reagents for Better Metallurgy", Chapter 5, P.S.Mulukutla (Ed.), Society for Mining, Metallurgy and Exploration Inc., Littleton, p.225. Laskowski J.S. and Y u Z., 1994; "The Effect of Humic Acids on the Emulsion Flotation of Inherently Hydrophobic Minerals", in "Flotation", A Volume in Memory of Alexander Sutulov, Proceedings of the IV-th Meeting of the Southern Hemisphere on Mineral Technology; and III Latin-American Congress on Froth Flotation, Vol.11, S.Castro and Alvarez (Eds.), p.397, University of Concepcion, Chile.  221  Laskowski J.S. and Y u Z., 1998; "Fine Coal Particle Agglomeration in Coal Preparation Plant Circuits", in "Proceedings of the 13-th International Coal Preparation Congress", Australian Coal Preparation Society, Brisbane, vol.2, p.591. Laskowski J.S., 1999; "Does It Matter How Coals Are Cleaned for CWS ?", Coal Preparation, vol.21, no.l, p.105. Laskowski J.S., 1999a; "Weak Electrolyte Collectors", in "Advances in Flotation Technology", B.K.Parekh and J.D.Miller (Eds.), SME, Littleton, p.59. Laskowski J.S., 2001; "Coal Flotation and Fine Coal Utilization", Developments in Mineral Processing, vol.14, Elsevier. Latiff Ayub A , A l Taweel A . M . and Kwak J.C.T., 1985a; "Surface Properties of Coal Fines in Water. I. Electrokinetics and Surfactant Adsorption", Coal Preparation, vol.1, no.2, p.117. Latiff Ayub A , Hayakawa K . , A l Taweel A . M . and Kwak J.C.T., 1985b; "Surface Properties of Coal Fines in Water. II. Isotherms, Electrokinetics and Chainlength Dependence", Coal Preparation, vol.2, no.l, p. 1. Lawson G.J. and Stewart D., 1989; "Coal Humic Acids", in "Humic Substances II", Chapter 23, p.642, M.H.B.Hayes, P.McCarthy, R.L.Malcolm, R.S.Swift (Editors), Wiley. Lee D.I., 1970; "Packing of Spheres and Its Effect on the Viscosity of Suspensions", Journal of Paint Technology, vol.42, no.550, p.579. Leja J. 1957; "Interactions at Interfaces in Relation to Froth Flotation", in "Electrical Phenomena and Solid/Liquid Interface", J.H.Schulman (Ed.), Proceedings of the Second International Congress of Surface Activity, vol.III, Butterworths, London, p.273. Leja J., 1982; "Surface Chemistry of Froth Flotation", Plenum Press, New York and London. Leja J., 1983; "On the Action of Long Chain Amines in Potash Flotation", in "Potash '83 - Proceedings of the First International Potash Technology Conference", Oct. 3-5, 1983, Saskatoon, Sask., Canada., McKercher R . M . (Ed.), Pergamon Press, p.623. Leong Y . K . and Boger D.V., 1990; "Surface Chemistry Effects on Concentrated Suspension Rheology", Journal of Colloid and Interface Science, vol.136, p.249. Leong Y . K . , Boger D . V . and Parris D., 1991a; "Surface Chemistry and Rheological Properties of Zirconia Suspensions", Journal of Rheology, vol.35, no.l, p.149.  222  Leong Y . K . , Boger D.V. and Parris D., 1991b; "Surface and Rheological Properties of Zirconia Suspensions", Transaction of Institution of Chemical Engineers", vol.69, part A , p.381. Leong Y . K . , Scales P.J., Healy T.W. and Boger D.V., 1993; "Rheological Evidence of Adsorbate-mediated Short-range Steric Forces in Concentrated Dispersions", Journal of the Chemical Society. Faraday Transactions, vol.89, no.14, p.2473. Leong Y . K . , Scales P.J., Healy T.W. and Boger D.V., 1995; "Interparticle Forces Arising From Adsorbed Polyelectrolytes in Colloidal Suspensions", Colloids and Surfaces A, vol.95, p.43. L i H.C. and de Bruyn P.L., 1966; "Electrokinetic and Adsorption Studies on Quartz", Surface Science, vol.5, p.203. Lin I.J. and Somasundaran P., 1971; "Free-Energy Changes on Transfer of SurfaceActive Agents Between Various Colloidal and Interfacial States", Journal of Colloid and Interface Science, vol.37, no.4, p.731. Liu J., Zhou Z. and X u Z., 2001; "Electrokinetic Study of Hexane Droplets in Surfactant Aqueous Solutions", in "Interactions in Mineral Processing", J.A.Finch, S.R.Rao, L.Huang (Eds.), Proceedings of the Fourth U B C - M c G i l l International Symposium on Fundamentals of Mineral Processing, Metallurgical Society of C I M , Montreal, p.375. Liu Q. and Laskowski J.S., 1988; "Effect of Humic Acids on Coal Flotation. Part II. The Role of pH", Society of Mining Engineers, Littleton, Colorado, Preprint Number 88-31. Liu Q. and Laskowski J.S., 1989; "The Role of Metal Hydroxides at Mineral Surfaces in Dextrin Adsorption I. Studies on Modified Quartz Samples", International Journal of Mineral Processing, vol.26, p.297. Lowell S. and Shields J.E., 1998; "Powder Surface Area and Porosity", B.Scarlett (Ed.), Chapman and Hall, Third edition. Lowenhaupt D.E. and Gray R.J., 1980; "The Alkali-Extraction Test as a Reliable Method of Detecting Oxidized Metallurgical Coal", International Journal of Coal Geology, vol.1, p.63. Lowenhaupt D.E. and Gray R.J., 1989; "Aging and Weathering", in "Sample Selection, Aging, and Reactivity of Coal", p.255, R.Klein and R.Wellek (Eds.), John Wiley & Sons. Lyklema J., 1968; "The Structure of the Electrical Double Layer on Porous Surfaces", Journal of Electroanalytical Chemistry, vol.18, p.341.  223  Mackenzie J.M.W., 1969; "Electrokinetic Properties of Nujol-Flotation Collector Emulsion Drops", Transactions of AIME, vol.244, p.393. Manser R . M . , 1975; "Handbook of Silicate Flotation", Warren Spring Laboratory, Stevenage, Herts, England. Marnell P., Poundstone W.N. and Halverson W., 1983; "Coal Water Slurries: A Seam to Steam Strategy", 5-th Annual Symposium on Industrial Coal Utilization, Pittsburgh, Pennsylvania, June 1983, p.. Mehrotra V.P., Sastry K . V . S . and Morey B.W., 1983; "Review of Oil Agglomeration Techniques for Processing of Fine Coals", International Journal of Mineral Processing, vol.11, p.175. Miller J.D., Laskowski J.S. and Chang S.S., 1983; "Dextrin Adsorption by Oxidized Coal", Colloids and Surfaces A, vol.8, p.137. Miller J.D., L i u C L . and Chang S.S., 1984; "Coadsorption Phenomena in the Separation of Pyrite From Coal by Reverse Flotation", Coal Preparation, vol.1, p.21. Miller K.J., 1975; "Coal-Pyrite Flotation", Transactions AIME, vol.258, p.30. Miller K . J . and Deurbrouck A . W . , 1982; "Froth Flotation to Desulfurize Coal", in "Physical Cleaning of Coal". Y . A . L i u (Ed.), Marcel Dekker, New York, p.255. Mishra S.K. and Klimpel R.R. (Editors), 1987; "Fine Coal Processing", Noyes Publications, Park Ridge. Modi H.J. and Fuerstenau D.W., 1960; "Flotation of Corundum, an Electrochemical Interpretation", Transactions AIME, vol.217, p.381. Mooney M . , 1951; "The viscosity of a concentrated suspension of spherical particles", Journal of Colloid Science, vol.6, p. 162. Moore F. and Davies L.J., 1956; " A New Rotational Viscometer and Some Preliminary Results", Transactions of the British Ceramics Society, vol.55, p.313. Moroi Y . , Sugii R. and Matuura R., 1984; "Examination of Micelle Formation by Phase Rule", Journal of Colloid and Interface Science, vol.98, no.l, p. 184. Morrison R.T. and Boyd R.N., 1966; "Organic Chemistry", 2-nd Edition, Allyn and Bacon, Boston. Moudgil B . M . , 1983; "Effect of Polyacrylamide and Polyethylene Oxide Polymers on Coal Flotation", Colloids and Surfaces A, vol.8, p.225.  224  Mukerjee P. and Mysels K . , 1955; " A Re-evaluation of the Spectral Change Method of Determining Critical Micelle Concentration", Journal of the American Chemical Society, vol.77, p.2937. Mukerjee P., 1956; "Use of Ionic Dyes in the Analysis of Ionic Surfactants and Other Ionic Organic Compounds", Analytical Chemistry, vol.28, no.5, p.870. Mukerjee A . and Mukerjee P., 1962; "Spectrophotometric Analysis of Long-Chain Amines by Dye-Extraction Method", Journal of Applied Chemistry, vol. 12, p. 127. Muster T.H. and Prestidge C.A., 1995; "Rheological Investigations of Sulphide Mineral Slurries", Minerals Engineering, vol.8, no.12, p.1541. Napper D.H., 1977; "Steric Stabilization", Journal of Colloid and Interface Science, vol.58, no.2, p.390. Napper D.H., 1981; "Polymeric Stabilization", in "Colloidal dispersions", Goodwin J.W. (Ed.) Royal Chemical Society, London, p.99. Nguyen Q.D., 1983; "Rheology of Concentrated Bauxite Residue Suspensions", Ph.D. Thesis, Monash University, Australia. Nguyen Q.D. and Boger D.V., 1983; "Yield Stress Measurement for Concentrated Suspensions", Journal of Rheology, vol.27, no.4, p.321. Nguyen Q.D., Logos C. and Semmler T., 1997; "Rheological Properties of South Australian Coal Water Slurries", Coal Preparation, vol.18, p. 185. Nguyen Q.D. and Boger D.V., 1998: "Application of Rheology to Solving Tailings Disposal Problems", International Journal of Mineral Processing, vol.54, p.217. Nicol S.K., 1997; "The Principles of Coal Preparation", A.R.Swanson (Ed.), The Australian Coal Preparation Society, Brisbane. Nyamekye G.A., 1992; "Adsorption of Dextrin onto Sulphide Minerals and Its Effect on the Differential Flotation of the INCO Matte", Ph.D. Thesis, The University of British Columbia, Vancouver, Canada. Ohki A., Fukuda S., Naka K. and Maeda S., 1996; "Studies on Coal Slurry Fuel (Part 4). Effect of Additive and Particle Size Distribution on Characteristics of Coal-Water Mixture (CWM)", Sekiyu Gakkaishi, vol.39, p. 129. Ohki A . , Xie X.F., Nakajima T., Itahara T. and Maeda S., 1999; "Change in Properties and Combustion Characteristics of an Indonesian Low-rank Coal due to Hydrothermal Treatment", Coal Preparation, vol.21, no.l, p.23.  225  Ono T., Makino H . and Takinami T., 1999; "Development of C W M Producing Method by Dry Type Pulverizer and Application for Low Rank Coal", Coal Preparation, vol.21, no.l, p.53. Osborne D.G., 1988; "Coal Preparation Technology", vol.I/II, Graham and Trotman Ltd. Osborne D.G., Graham J.M. and Elliot L.K., 1996; "New Coal Utilization Technologies", Minerals Engineering, vol.9, no.2, p.215. Painter P., Starsinic M . and Coleman M . , 1985; "Determination of Functional Groups in Coal by Fourier Transform Interferometry", Chapter 5 in "Fourier Transform Infrared Spectroscopy", J.R. Ferraro, L.J. Basile (Eds.), Vol. 4, p. 169, Academic Press. Palmes J.R. and Laskowski J.S., 1993; "Effect of the Properties of Coal Surface and Flocculant Type on the Flocculation", Minerals and Metallurgical Processing, vol.10, no.4,p.218. Papachristodoulou G. and Trass O., 1987; "Coal Slurry Fuel Technology", The Canadian Journal of Chemical Engineering, vol.65, p. 177. Parachuri V.K.-, Nalaskowski J. and Miller J.D., 2001; "The Structure and Characteristics of Adsorbed Surfactants as Revealed by A F M Imagery", S M E Annual Meeting, Feb. 26-28, Denver CO, Preprint 01-48. Parkinson C , Matsumoto S. and Sherman P., 1970; "The influence of particle size distribution on the apparent viscosity of non-Newtonian dispersed systems", Journal of Colloid and Interface Science, vol.33, no.l, p. 150. Parr S.W., 1932; "The Analysis of Fuel, Gas, Water and Lubricants", McGraw-Hill Book Co., Inc., 4-th Edition, New York. Pashley R . M . and Israelachvili J.N. 1981; " A Comparison of Surface Forces and Interfacial Properties of Mica in Purified Surfactant Solutions", Colloids and Surfaces, vol.2, pl69. Pashley R . M . , 1981; " D L V O and Hydration Forces Between Mica Surfaces in L i , N a , K and C s Electrolyte Solutions: A Correlation of Double-Layer and Hydration Forces with Cation Exchange Properties", Journal of Colloid and Interface Science, vol.83, no.2, p.531. +  +  +  +  Pawlik M . , Laskowski J.S. and Liu H., 1997; "Effect of Humic Acids and Coal Surface Properties on Rheology of Coal Water Slurries", Coal Preparation, vol.18, p. 129. Pawlik M . and Laskowski J.S., 1998; "Direct yield stress measurements as a method of determining contact angles on fine coal particles", Presentation at 216-th National  226  A C S Meeting, Apparent and Microscopic Contact Angles, Boston, M A , August 24-27. Pawlik M . and Laskowski J.S., 1999; "Evaluation of Flocculants and Dispersants Through Rheological Tests", in "Polymers in Mineral Processing", Proceedings of the Third U B C - M c G i l l Bi-Annual International Symposium on Fundamentals of Mineral Processing, J.S. Laskowski (Ed.), Metallurgical Society of CEM, Montreal, p.541. Pereira H.C. and Schulman J.H., 1961; "Streaming Potential Measurements on Paraffin Wax", in "Solid Surfaces". R.F.Gould (Ed.), A C S Advances in Chemistry Series vol.33, Washington DC, p. 160. Pikkat-Ordynsky G.A. and Ostry V . A . , 1972; "Technology of Coal Flotation". Izd. Nedra, Moscow (in Russian). Pommier L., Frankiewicz T. and Weissberger W., 1984; "Coal Water Slurry Fuels - A n Overview", Minerals and Metallurgical Processing, vol.1, no.l, p.62. Poslinski A.J., Ryan M . E . , Gupta R . K . , Seshadari S.G. and Frechette F.J., 1988; "Rheological Behavior of Filled Polymeric Systems II. The Effect of a Bimodal Size Distribution of Particulates", Journal of Rheology, vol.32, no.8, p.751. Potanin A . A . and Uriev N . B . , 1991; "Microrheological Models of Aggregated Suspensions in Shear Flow", Journal of Colloid and Interface Science, vol.142, no.2, p.385. Pradip and Fuerstenau D.W., 1987; "The Effect of Polymer Adsorption on the Wettability of Coal", in "Flocculation in Biotechnology and Separation Systems", p.95, Process Technology Proceedings 4, Y.A.Attia (Editor), Elsevier. Prestidge C.A., 1997; "Rheological Investigations of Galena Particle Interactions" Colloids and Surfaces A, vol.126, p.75. Probstein R.F. and Sengun M.Z., 1987; "Dense Slurry Rheology With Application to Coal Slurries", Physicochemical Hydrodynamics, vol.9, nol/2, p.299. Probstein R.F., Sengun M.Z. and Tseng T.C., 1994; "Bimodal Model of Concentrated Suspension Viscosity for Distributed Particle Sizes", Journal of Rheology, vol.38, no.4,p.811. Qiu X . , 1992; "Surface Properties of Coal and Their Effects on the Selective Oil Agglomeration", Ph.D. Thesis, Iowa State University.  227  Raichur A . M . , Misra M . , Davis S.A. and Smith R.W., 1997; "Flocculation of Fine Coal Using Synthetic and Biologically Derived Flocculants", Minerals and Metallurgical Processing, p.22, February 1997. Ralston A.W., 1948; "Fatty Acids and Their Derivatives", Wiley, New York. Ralston J. and Kitchener J.A., 1975; "The Surface Chemistry of Amosite Asbestos, an Amphibole Silicate", Journal of Colloid and Interface Science, vol.50, no.2, p.242. Rao K . H . , Vidyadhar A., Chernyshova I.V. and Forssberg K.S.E., 2001; "Interactions of Long-Chained Primary Alkylamine on Silicate Minerals", in "Interactions in Mineral Processing", J.A.Finch, S.R.Rao, L.Huang (Eds.), Proceedings of the Fourth U B C - M c G i l l International Symposium on Fundamentals of Mineral Processing, Metallurgical Society of CIM, Montreal, p.343. Reck R.A. and Harwood H.J., 1953; "Antimicrobial Activity of Quaternary Ammonium Chlorides Derived From Commercial Fatty Acids", Industrial and Engineering Chemistry, vol.45, no.5, p. 1022. Rees O.W., 1966; "Chemistry, Uses, and Limitations of Coal Analyses", Report of Investigations 220, Illinois State Geological Survey, Urbana IL. Rehbinder P. A., 1964; "Coagulation and Thixotropic Structures", Discussions of Faraday Society, no. 18, p. 151. Rehbinder P. A., 1965; "Formation of Structures in Disperse Systems", Pure and Applied Chemistry, vol.10, no.4, p.337. Roberts T., Firth B . A . and Nicol S.K., 1982; " A Modified Laboratory Cell for the Flotation of Coal", International Journal of Mineral Processing, vol.9, p.191. Rodriguez B.E., Kaler E.W. and Wolfe M.S., 1992; "Binary Mixtures of Monodisperse Latex Dispersions", Langmuir, vol.8, p.2382. Rosen M.J., 1989; "Surfactants and Interfacial Phenomena", 2-nd Edition, Wiley. Rosenbaum J.M. and Fuerstenau D.W., 1984; "On the Variation of Contact Angles With Coal Rank", International Journal of Mineral Processing, vol.12, p.313. Sadowski Z., Venkatadri R., Druding J.M., Markuszewski R. and Wheelock T.D., 1988; "Behavior of Oxidized Coal During Oil Agglomeration", Coal Preparation, vol.6, p.17.  228  Saeki T., Usui H. and Ogawa M , 1994; "Effect of Molecular Structure of Polysaccharide on the Stability of Coal-Water Mixtures", Journal of Chemical Engineering of Japan, vol.27, no.6, p.773. Saeki T., Tatsukawa T. and Usui H . , 1999; "Preparation Techniques of Coal Water Mixtures with Upgraded Low Rank Coals", Coal Preparation, vol.21, no.l, p.161. Sato T. and Ruch R., 1980; "Stabilization of Colloidal Dispersions by Polymer Adsorption", Surfactant Science Series, vol.9, Marcel Dekker. Scales P.J., Kapur P.C., Johnson S.B. and Healy T.W., 1998; "The Shear Yield Stress of Partially Flocculated Suspensions with Size Distributed Particles", AIChE Journal, vol.44, p.538. Scales P.J., S.B. Johnson and Kapur P.C., 1999; "The Influence of Surface Chemistry on the Rheology an Flow of Flocculated Particulate Suspensions", Mineral Processing and Extractive Metallurgy Review, vol.20, p.27. Scamehorn J.F., Schechter R.S. and Wade W.H., 1982; "Adsorption of Surfactants on Mineral Oxide Surfaces from Aqueous Solutions. I. Isomerically Pure Anionic Surfactants", Journal of Colloid and Colloid Science, vol.85, p.463. Schwartz E., Stork K . and Crema S., 1985; "Physical Properties of Coal Surfaces and Their Interactions with Stabilizers", Proceedings of the 7-th International Symposium on Coal Slurry Fuels Preparation and Utilization, p41, U.S.Dept. of Energy, New Orleans. Scott K J . , 1982; "The Effect of Surface Charge on the Rheology of Concentrated Aqueous Quartz Suspensions", CERG-CSIR Report C E N G 423, Pretoria, South Africa. Seki M . , Kiyama K . and Nishino J., 1985; "Effects of Coal Property and Additive on the Rheological Characteristics of Coal-Water Mixtures", 7-th International Symposium Coal Slurry Fuels Preparation and Utilization, New Orleans, p. 136. Sengun M . Z . and Probstein R.F., 1989a; "Bimodal Model of Slurry Viscosity with Application to Coal Slurries. Part I. Theory and Experiment". Rheologica Acta, vol.28, p.382. Sengun M . Z . and Probstein R.F., 1989b; "Bimodal Model of Slurry Viscosity with Application to Coal Slurries. Part II. High Shear Limit Behavior". Rheologica Acta, vol.28, p.394.  229  Shapiro S.H., 1968; "Commercial Nitrogen Derivatives of Fatty Acids", in "Fatty Acids and Their Industrial Applications", E.S.Pattison (Ed.), Marcel Dekker, New York, p.77. Shinoda K., Nakagawa T., Tamamushi B. and Isemura T., 1963; "Colloidal Surfactants; Some Physicochemical Properties", Academic Press, New York/London. Siffert B . and Hamieh T., 1989; "Effect of Mineral Impurities on the Charge and Surface Potential of Coal", Colloids and Surfaces, vol.35, p.27. Smith R.W., 1963; "Coadsorption of Dodecylamine Ion and Molecule on Quartz", Transactions AIME, vol.226, p.427. Smith R.W. and Lai R.W., 1966; "On the Relationship between Contact Angle and Flotation Behavior", Transactions AIME, vol.238, p.413. Smith R.W., 1973; "Effect of Amine Structure in Cationic Flotation of Quartz", Transactions AIME, vol.254, p.353. Smith R.W. and Akhtar S., 1976; "Cationic Flotation of Oxides and Silicates", in "Flotation - A . M . Gaudin Memorial Volume", Vol.1, Chapter 5, M.C.Fuerstenau (Ed.), A I M E , New York, p.87 Smith R.W., 1988; "Cationic and Amphoteric Collectors", in "Reagents in Mineral Technology", P.Somasundaran and B.M.Moudgil (Eds.), Marcel Dekker, New York, Surfactant Science Series vol.27, Chapter 8, p.219. Smith R.W. and Scott J.L., 1990; "Mechanisms of Dodecylamine Flotation of Quartz", Mineral Processing and Extractive Metallurgy Review, vol.7, p.81. Smitham J.B. and Napper D.H., 1979; "Steric stabilization in worse than theta-solvents", Colloid and Polymer Science, vol.257, p.748. Smoluchowski M . , 1903; "Contribution a la Theorie de l'Endosmose Electrique et de Quelques Phenomenes Correlatifs", Bulletin International de VAcademie des Sciences de Cracovie, vol.8, p. 182. Smoluchowski M . , 1921; in "Handbuch der Electrizitat und des Magnetismus", (Graetz) vol.11, p.366, Barth, Leipzig. Solari J.A., deAraujo A . C . and Laskowski J.S., 1986; "The Effect of Carboxymethyl Cellulose on Flotation and Surface Properties of Graphite", Coal Preparation, vol.3, p.15.  230  Somasundaran P., Healy T.W. and Fuerstenau D.W., 1964; "Surfactant Adsorption at the Solid-Liquid Interface - Dependence of Mechanism on Chain Length", The Journal of Physical Chemistry, vol.68, no. 12, p.3562. Somasundaran P. and Fuerstenau D.W., 1968; "On Incipient Flotation Conditions", Transactions SME, vol.241, p.102. Somasundaran P., 1968; "The Relationship Between Adsorption at Different Interfaces and Flotation Behavior", Transactions SME, vol.241, p. 105. Somasundaran P. and Krishnakumar S., 1994; "In Situ Spectroscopic Investigations of Adsorbed Surfactant and Polymer Layers in Aqueous and Non-Aqueous Systems", Colloids and Surfaces A, vol.93, p.79. Spurny J. andDobias B., 1962; Coll.Czechoslov.Chem.Comm. vol.27, p.931 Stach E., 1982; "The Macerals of Coal", in "Stach's Textbook of Coal Petrology". Gebriider Borntraeger, Berlin-Stuttgart. Steelink C , 1985; "Implications of Elemental Characteristics of Humic Substances", in "Humic Substances in Soil, Sediment, and Water", Chapter 18, p.457, G.R.Aiken, D.M.McKnight, R.L.Wershaw and P.McCarthy (Eds.), Wiley. Stigter D., 1964; "On the Adsorption of Counterions at the Surface of Detergent Micelles", Journal of Physical Chemistry, vol.68, p.3603. Stigter D., 1967; "Density, Hydration, Shape, and Charge of Micelles of Sodium Dodecyl Sulfate and Dodecylammonium Chloride", Journal of Colloid and Interface Science, vol.23, p.379. Stonestreet P. and Franzidis J.P., 1988; "Reverse Flotation of Coal - A Novel Way for the Beneficiation of Coal Fines", Minerals Engineering, vol.1, p.343. Stonestreet P. and Franzidis J.P, 1989; "Development of the Reverse Coal Flotation Process: Depression of Coal in the Concentrates", Minerals Engineering, vol.2, p.393. Stonestreet P. and Franzidis J.P., 1992; "Development of the Reverse Coal Flotation Process: Application to Column Flotation", Minerals Engineering, vol.5, p.1041. Sudduth R., 1993a; " A Generalized Model to Predict the Viscosity of Solutions With Suspended Particles I", Journal of Applied Polymer Science, vol.48, p.25. Sudduth R., 1993b; " A New Method to Predict the Maximum Packing Fraction and the Viscosity of Solutions With a Size Distribution of Suspended Particles II", Journal of Applied Polymer Science, vol.48, p.37.  231  Sudduth R., 1993c; " A Generalized Model to Predict the Viscosity of Solutions With Suspended Particles III. Effects of Particle Interaction and Particle Size Distribution", Journal of Applied Polymer Science, vol.50, p.123. Sun S.C., 1954a; "Hypothesis for Different Floatabilities of Coals, Carbons, and Hydrocarbon Minerals", Transactions AIME, vol.199, p.67. Sun S.C., 1954b; "Effects of Oxidation of Coals on Their Flotation Properties", Transactions AIME, vol.199, p.396. Suwono A . and Hamdani, 1999; "Upgrading the Indonesian's Low Rank Coal by Superheated Steam Drying with Tar Coating Process and Its Application for Preparation of C W M " , Coal Preparation, vol.21, no.l, p.149. Tadros Th.F. and Lyklema J., 1968; "Adsorption of Potential-Determining Ions at the Silica-Aqueous Interface and the Role of Some Cations", Journal of Electroanalytical Chemistry, vol.17, p.267. Tadros Th.F., 1982; "Polymer Adsorption and Dispersion Stability", in "The effect of polymers on dispersion properties", Tadros Th.F. (Ed.), Academic Press, p . l . Tadros Th.F., 1985; "Use of Surfactants and Polymers for Preparation and Stabilization of Coal Suspensions", Second European Conference on Coal Liquid Mixtures, September 1985, London, Proceedings, IChemE Symposium Series No.95, p . l , Pergamon Press, Oxford, England 1985. Tadros Th.F., 1986; "Control of the Properties of Suspensions", Colloids and Surfaces, vol.18, p. 137. Tadros Th.F., Taylor P. and Bognolo G., 1995; "Influence of Addition of a Polyelectrolyte, Nonionic Polymers, and Their Mixtures on the Rheology of Coal/Water Suspensions", Langmuir, vol.11, p.4678. Tadros Th.F., 1996; "Correlation of Viscoelastic Properties of Stable and Flocculated Suspensions With Their Interparticle Interactions", Advances in Colloid and Interface Science, vol.68, p.97. Tamamushi B . and Tamaki K . , 1957; "Adsorption of Long-Chain Electrolytes at the Solid/Liquid Interface", in "Electrical Phenomena and Solid/Liquid Interface", J.H.Schulman (Ed.), Proceedings of the Second International Congress of Surface Activity, vol.III, Butterworths, London, p.449. Taylor P., Liang W., Bagnolo G. and Tadros Th.F., 1991; "Concentrated Coal Water Suspensions Containing Non-Ionic Surfactants and Polyelectrolytes. Part I. A Preliminary Study Using Rheology and Adsorption Isotherms", Colloids and Surfaces A, vol.61, p.147.  232  Ter-Minassian-Saraga L . , 1967; "Chemisorption and Wetting and Non-Wetting", in "Wetting", Society of Chemical Industry Monograph no.25, Gordon and Breach, New York, p.272. Ter-Minassian-Saraga L . , 1975; "Dewetting Reaction and Flotation", in "Advances in Interfacial Phenomena of Particulate/Solution/Gas Systems; Applications to Flotation Research", P.Somasundaran and R.B.Grieves (Eds.), American Institution of Chemical Engineers, AIChE Symposium Series No. 150 vol.71, p.68. Trass O. and Papachristodoulou G., 1986; "Economical Coal Slurry Fuels by Oil Agglomeration", Eighth International Symposium on Coal Slurry Fuel Preparation and Utilization. Proceedings, May 1986, Orlando, Florida, p.232. Trochet-Mignard L., Taylor P., Bagnolo G. and Tadros Th.F., 1995; "Concentrated Coal Water Suspensions Containing Non-Ionic Surfactants and Polyelectrolytes. Part II. Adsorption of Nonyl Phenyl Propylene-Ethylene Oxide on Coal and the Rheology of the Resulting Suspension", Colloids and Surfaces A, vol.95, p.37. Ukigai  T., Sugawara H . and Tobori N . , 1994; "Effect of Additives on Dispersed/Coagulated State in C W M " , "Proceedings of I E A - C L M Workshop'94 on Coal Water Mixture (CWM)", p.84.  Usui PL, Saeki T., Hayashi K . and Tamura T, 1997a; "Sedimentation Stability and Rheology of Coal Water Slurries", Coal Preparation, vol.18, p.201. Usui FL, Tatsukawa T., Saeki T. and Katagiri K . , 1997b; "Rheology of Low Rank Coal Slurries Prepared by and Upgrading Process", Coal Preparation, vol.18, p. 119. Usui FL, Tatsukawa T., Saeki T. and Katagiri K., 1999; "Upgrading of Low rank Coal by a Combined Process of Vacuum Drying and Tar Coating", Coal Preparation, vol.21, p.71. Van Wazer J.R., 1963; "Viscosity and Flow Measurement: a Laboratory Handbook of Rheology", Interscience Publishers, New York. Velamakanni B.V. and Fuerstenau D.W., 1987; "The Influence of Polymeric Additives on the Rheology of Dense Slurries", "Process Technol. Proc." (Flocculation Biotechnol. Sep. Syst.), p.211. Velamakanni B . V . and Fuerstenau D.W., 1993a; "The Effect of the Adsorption of Polymeric Additives on the Wet Grinding of Minerals. 1. Mechanism of Suspension Stabilization", Powder Technology, vol.75, p . l .  233  Velamakanni B . V . and Fuerstenau D.W., 1993b; "The Effect of the Adsorption of Polymeric Additives on the Wet Grinding of Minerals. 2. Dispersion and Fine Grinding of Concentrated Suspensions", Powder Technology, vol.75, p . l 1. Walters A . B . , 1986; "Chemical Effects of Beneficiation on Coal-Water Fuels Properties", International Symposium on Coal Slurry Fuel Preparation and Utilization. Proceedings, p.222, May 1986, Orlando, Florida. Wang Z., Zhang R.Z., Jiang Z. and Jiang S., 1993; "Preparation of Coal-Water Fuels From Coal Preparation Plant Fines", in "Proceedings of the 6-th Australian Coal Preparation Conference', Davis J.J. (Ed.), Australian Coal Preparation Society, Mackay, p.418. Wangnerud P., Berling D and Olofsson G., 1995; "Adsorption of Alkyl-trimethylammonium Bromides on Silica: Calorimetric Study of Effect of Coions", Journal of Colloid and Interface Science, vol.169, p.365. Watson D. and Manser R . M . , 1968; "Some Factors Affecting the Limiting Conditions in Cationic Flotation of Silicates", Transactions of the Institution of Mining and Metallurgy, vol.77, p.C57. Wen W.W. and Sun S.C., 1977; " A n Electrokinetic Study on the Amine Flotation of Oxidized Coal", Transactions AIME, vol.262, p. 174. Wen W.W. and Sun S.C., 1981; " A n Electrokinetic Study on the O i l Flotation of Oxidized Coal", Separation Science, vol.16, p.1491. Wheeler T.A., 1994; "Coal Floats by Itself - Doesn't It ?", in "Reagents for Better Metallurgy". Chapter 14, P.S.Mulukutla (Ed.), SME, Littleton, p.131. Whitehurst D.D., Mitchell T.O. and Farcasiu M . , 1980; "Coal Liquefacation", Academic Press, New York. Whorlow R.W., 1980; "Rheological Techniques", Halsted, New York. Willson W.G., Walsh D. and Irwin W., 1997; "Overview of Low Rank Coal (LRC) Drying", Coal Preparation, vol.18, no. 1/2, p . l . Winters P.J., Bailey R.T. and Olen K.R., 1985; "The Effect of Fuel Formulation on the Atomization Characteristics of Coal Water Mixtures", Seventh International Symposium on Coal Slurry Fuel Preparation and Utilization. Proceedings, p.430, New Orleans, May 1985. Wong K . and Laskowski J.S., 1984; "Effect of Humic Acids on the Properties of Graphite Aqueous Suspensions", Colloids and Surfaces A, vol.12, p.319.  234  Woskoboenko F., Siemon S.R. and Creasy D.E., 1987; "Rheology of Victorian Brown Coal Slurries 1. Raw Coal-Water", Fuel, vol.66, p. 1299. Woskoboenko F., Siemon S.R. and Creasy D.E., 1989; "The Rheology of Victorian Brown Coal Slurries 2. Effect of pH", Fuel, vol.68, p. 121. Verwey E.J.W. and Overbeek J.Th.G., 1948; "Theory of the Stability of Lyophobic Colloids", Elsevier, Amsterdam. X u Z. and Yoon R . H . 1989; "The Role of Hydrophobic Interactions in Coagulation", Journal of Colloid and Interface Science, vol.132, no.2, p.532. X u Z. and Yoon R.H., 1990; " A Study of Hydrophobic Coagulation", Journal of Colloid and Interface Science, vol.134, no.2, p.427. Ye Y . and Miller J.D., 1988; "Bubble/Particle Contact Time in the Analysis of Coal Floatation", Coal Preparation, vol.5, p. 147. Yoon R.H., Flinn D.H. and Rabinovich Y.J., 1996; "Hydrophobic Interactions Between Dissimilar Surfaces", Journal of Colloid and Interface Science, vol.185, p.363. Yoshihara H . , 1999; "Graft Copolymers as Dispersants for C W M " , Coal Preparation, vol.21, p.93. Yotsumoto H . and Yoon R . H . , 1993a; "Application of Extended D L V O Theory I. Stability of Rutile Suspensions", Journal of Colloid and Interface Science, vol.157, p.426. Yotsumoto H . and Yoon R.H., 1993b; "Application of Extended D L V O Theory II. Stability of Silica Suspensions", Journal of Colloid and Interface Science, vol.157, p.434. Young T., 1805, Philosophical Transactions of the Royal Society of London, vol.95, p.65. Y u Z. and Laskowski J.S., 1999; "The Effects of Surfactants on Oil Agglomeration and Filtration of Fine Coal", in "Proceedings of the International Symposium on Beneficiation, Agglomeration and Environment", S.R.S.Sastry, S.Mohanty, and B.K.Mohapatra (Eds.), Regional Research Lab, Bhubaneswar, p. 115. Zhang R.Z., Zeng F., Hu K . M . and Gao M.Q., 1984; "Research on C W M Preparation Techniques With Chinese Coals", in "Proceedings of the 6-th International Symposium on Coal Slurry Combustion and Technology", Orlando, p.234. Zhou Z.A., Hussein H . , X u Z., Czarnecki J. and Masliyah J.H., 1998; "Interaction of Ionic Species and Fine Solids With Low Energy Hydrophobic Surface From  235  Contact Angle Measurement", Journal of Colloid and Interface Science, vol.204, p.342. Zhou Z., Scales P.J. and Boger D.V., 2001; "Chemical and Physical Control of the Rheology of Concentrated Metal Oxide Suspensions", Chemical Engineering Science, vol.56, p.2901. Zielinski R., Ikeda S., Nomura H. and Kato S., 1989; "Effect of Temperature on Micelle Formation in Aqueous Solutions of Alkyltrimethyl Ammonium Bromides", Journal of Colloid and Interface Science, vol.129, no.l, p. 175. Zisman W.A., 1963; Industrial and Engineering Chemistry, vol.55, no.10, p.19. Zuidema H . H . and Waters G.W., 1941; "Ring Method for the Determination of Interfacial Tension", Industrial and Engineering Chemistry, vol.13, no.5, p.312.  236  APPENDIX 1 BET plots for the tested solids. An example of the specific surface area calculation for LS43 coal is also shown.  1  1000-  800 H  ^  Fine Silica; Y = 1025.20 * X F4 O x i d : Y = 2884.56 * X F4 O x i d V a c : Y = 2488.48 * X F 4 : Y = 1437.57 * X +43.57 L S 4 3 : Y = 1080.11 * X +29.86  1  1  600 H o  PH  H  400  LS43 Fine Silica  200 H  00  0.2  0.1  P/P  0.4  0.3 n  Calculations for LS43 Coal: Calibration volume: 0.7 cm Gas Weight: 0.000875 g Solids Weight: 1.5170 g Sample: LS43 Coal, -212 microns  Slope: 1080.11  Intercept: 29.86  P/Po  Po/P  Po/P-1  Signal Area  Calib. Area  X  X(Po/P-l)  l/[X(Po/P-l)]  X =l/(S+I)  0.1  10  9  1358.6667  1483.5220  0.0008014  0.0072122  138.653395  0.000900927  0.2  5  4  1843.3257  1576.2580  0.0010233  0.0040930  244.3189702  0.3  3.33  2.33  2093.2657  1515.7913  0.0012084  0.0028195  354.6747171  m  Total Surface Area: Srotai = (6.023 10 molec./mol 16.2 10" nrVmolec.- 0.000900927 g)/28.01 g/mol = 3.139 m 23  20  2  The Specific Surface Area: ABET  =  3.139 m /1.5170 g = 2.070 m7g 2  237  APPENDIX 2 Calibration curves for humic acids at different wavelengths. Only the curve for 350 nm was used for determining the equilibrium humic acids concentration.  • Equation of the calibration curve at 350 nm:  0  20  40 60 80 100 120 140 Humic Acids Concentration [mg/dm ]  160  180  3  238  APPENDIX 3 Calibration curve for Dodecyl-Trimethyl Ammonium Bromide (DTAB). Experimental conditions are also given.  Concentration of D T A B in Aqueous Phase [ixmol/dm ] 3  0  8.65  17.3 25.95 34.6 43.25 51.9 60.55 69.2 77.85 86.5  0 20 40 60 80 100 120 140 160 180 200 D T A B Concentration in 4-cm Aliquots Taken for Assay [mg/dm ] 3  3  Conditions: 20 cm of 0.01 mol/dm HCI + 1 cm of 0.05mol/dm HCI + 5 cm 1.5 g/dm BPB(Na salt) in distilled water + 4 cm D T A B in tap water + 20 cm Chloroform or 30 cm of the aqueous phase + 20 cm of the organic phase 3  3  3  3  3  3  3  3  3  3  239  APPENDIX 4 The modified laboratory flotation cell designed by Roberts et al. (1982).  Locating Bolts P e r s p e x Insert  Top Lip ofCell Base of Froth Top of Pulp Internal Width of Cell  Scraper, cut out f r o m ~ 6 m m t h i c k transparent P l e x i g l a s sheet  Constant Level Control Perspex Insert  • Overflow ^ Tap Side View  Supply Water  Front View Complete Assembly  240  APPENDIX 5 Surface tension of DTAB and humic acids solutions in tap water, T = 23 deg. DTAB/Humic Acids Concentration [mg/dm ] 3  lxlO  1x10'  1  i  j  i  'I  i  • 111111  lxlO  1x10*  1x10' i  i  i 1111il  i  i  5  111111  i  •75 T70  Humic Acids  r65 ^  r60  60-  r55  'to  I  r50  50-  O  T a p Water,  U 45-  r  45  r  40  p H = 6.0-7.5, T = 23°C  40-  -B  B  •  DTAB  -35  3530-  -i—i—i  1x10"  i 11111  -i—i—i  1x10"  -i—i—111111  i1111  lxl0"  1x10  1x10"  :  1  r  I  T-30  1x10"  D T A B Concentration [mol/dm ] 3  DTAB Concentration rmol/dm l water 0.0000146 0.000146 0.00146 0.00292 0.0073 0.0146 0.0292 0.073 0.146 3  Numerical data for DTAB Surface Tension [mJ/m l 71.70 69.10 64.20 56.13 51.37 44.03 37.30 37.20 37.15 37.10  Standard deviation  2  0.15 0.12 0.15 0.20 0.10 0.13 0.12 0.10 0.13 0.15  241  APPENDIX 6 Particle size distributions of the tested coals. Particle size distributions of LS43 and F4 coals. PSD for F13 coal was exactly the same as that of F4. Size [microns] F4 Cumulative Undersize [%] LS43 Cumulative Undersize [%] 212  100.0  100.0  180  99.6  98.4  150  96.6  92.4  106  87.0  78.6  53  55.1  47.6  38  43.6  36.1  242  APPENDIX 7  Graphical determination of the RRB distribution parameters for F4 and LS43 coals. The RRB function is usually given in the following form:  F(d) = 100 1 - exp  d '63.2 J  where d is the particle size, is the size modulus, m is the distribution modulus and F(d) is the cumulative fraction passing size d. In order to find the size modulus and the distribution modulus, the above equation can be rearranged in several simple steps to obtain the following form:  100  In In  100-F(J)  = m \n(d) - m ln( J  6 3  2  )  When the expression on the left-hand side is plotted versus ln(rf), using the data from Appendix 6, the plot should be a straight line with the slope equal to the distribution modulus, m, and the intercept equal to -  m\n(d(ft2)-  Once m is determined from the slope,fi?63.2 plots for LS43 and F4 coals are shown below. c  1  a  n  he easily found from the intercept. The resulting  1  _L  Y = 1.415 * X - 5.788 for F4 Coal  Y = 1.378 * X - 5.883 for LS43 Coal  I  o o o o  -l  3.6  1 4  •  i  1  r  4.4 4.8 ln(particle size) [microns]  5.2  The slopes give the distribution moduli, i.e. 1.415 and 1.378 for F4 and LS43 coals, respectively. From the intercepts and the slopes, the size moduli can be calculated to be 59.8 pm and 71.5 pm for F4 and LS43 coals, respectively. The so-called RRB-graph paper is often used to only approximately estimate the moduli. The above method is based on actually solving the R R B equation.  243  APPENDIX 8 Particle size distribution of SilcoSil 395 silica as determined with the use of an Elzone counting and sizing system.  Particle size distribution (volume basis) of SilcoSil 395 silica Size [microns]  Cumulative Undersize [%]  67.3  100  60  99.34  54  98.09  50  97.02  44  94.14  40  91.52  38  89.89  35  86.88  30  81.02  25  73.80  20  65.06  15  54.98  12.5  49.73  10  43.72  7.5  35.25  5  25.34  2.5  12.07  2  8.75  1  2.63  244  APPENDIX 9 Graphical determination of surfactant adsorption at air/water interface from the surface tension data using the Gibbs equation. The relationship between the surface tension of a surfactant solution, y, surfactant concentration, c, and the surfactant adsorption density at the air/water interface, f , s  1  r. = — mRT  dy  dy  dine  mRT dc  is given by the Gibbs equation:  dine = — c  where m is the salt parameter (m - 2 when only the surfactant is present in solution, see Equation 66), R is the gas constant and T is the temperature. The equation states that in order to find the adsorption density at an air/solution interface, one has to determine the relationship between the surface tension and surfactant concentration (dy/dc). The surface tension data from Appendix 5 are replotted below on a linear scale in the concentration range below the cmc. 75-  70  1  1 1  S  Mid-point Concentration  a a  45-d oo  40-^ 35 300.0x10°  I  '  I  '  I  4.0xl0" 8.0xl0" 1.2xl0~ D T A B Concentration [mol/dm ] 3  3  :  1.6x10"  3  The calculation of the adsorption density was done as follows. As shown on the graph, a straight line was drawn between two data points and the slope of the line, equal to dyidc, was calculated. It was assumed that this slope was equal to the "real" slope of a line tangent to the curve at the mid-point concentration. The midpoint concentration was then taken as c in the Gibbs equation. If the actual functional relationship between y and c was known, the exact value of dy/dc at any concentration c could be easily obtained by differentiation. Several different functions were tried to best fit the data but none of them gave a satisfactory fit over the entire concentration range. The values of c and dy/dc obtained graphically were then put into the Gibbs equation and the adsorption density was calculated at T = 296K (23°C) taking c as the equilibrium D T A B concentration. It is important to use the proper units to correctly calculate the adsorption. If the surface tension is expressed in J/m , the gas constant in J/(K-mol) (R = 8.31434 J/molK) and the D T A B concentration in mol/m , then dy/dc will be given in Jm/mol, and the adsorption density will be obtained in mol/m . 2  3  2  245  APPENDIX 10 Effect of conditioning time on flotation of LS43/silica mixture, j  I  i  I  i  I  i  I  i  I  i  I  i  I  90  ^  D T A B Dosage [g/t] The results for 15 min conditioning time are those from Figure 38. 246  APPENDIX 11 Structures of polymers used in this work CH OH 2  branch point, (1-6) linkage (1-4) linkage  Branched structure of dextrin CH OCH COONa 2  2  CH OCH COONa 2  2  J  Carboxymethyl cellulose, sodium salt  Polystyrene sulfonate, sodium salt  n  APPENDIX 12 Proposed structure of humic acids according to Steelink (1985)  Tetramer structure for humic acids  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items