UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The dispersing action of polysaccharides in oil sand slurries Arinaitwe, Esau 2013

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

Item Metadata

Download

Media
24-ubc_2013_fall_arinaitwe_esau.pdf [ 5.52MB ]
Metadata
JSON: 24-1.0073935.json
JSON-LD: 24-1.0073935-ld.json
RDF/XML (Pretty): 24-1.0073935-rdf.xml
RDF/JSON: 24-1.0073935-rdf.json
Turtle: 24-1.0073935-turtle.txt
N-Triples: 24-1.0073935-rdf-ntriples.txt
Original Record: 24-1.0073935-source.json
Full Text
24-1.0073935-fulltext.txt
Citation
24-1.0073935.ris

Full Text

THE DISPERSING ACTION OF POLYSACCHARIDES IN OIL SAND SLURRIES  by ESAU ARINAITWE B.Sc., Makerere University, 2000 M.A.Sc., The University of British Columbia, 2008 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY  in The Faculty of Graduate Studies (Mining Engineering) The University of British Columbia (Vancouver)  June 2013  © Esau Arinaitwe, 2013  Abstract  Six carboxymethyl celluloses (CMC) of different degrees of substitution, molecular weights, and molecular weight distributions (MWDs) were researched as dispersants of oil sand slurries. The molecular weights and MWDs were determined by analytical ultracentrifugation. Viscometric studies on dilute solutions indicated semi-flexible, random coiling behavior of the polymers assuming an extended conformation in distilled water and coiled conformation in a dilute sodium chloride solution. It was found that the ionic strength, rather than pH or temperature, had the strongest effect on the intrinsic viscosity and conformation of CMC. Calculations of the persistence length and expansion factors of the polymers showed that the lowest molecular weight CMC was most flexible among the tested samples. As rheological and sedimentation tests showed, addition of CMC stabilized oil sand slurries towards aggregation and settling. All the polymers dispersed oil sand slurries by adsorbing on the solid particles and preventing mineral–mineral interactions. The molecular weight of the polymers was a more important factor than the degree of substitution in dispersing the slurries. The role of CMC was also to enhance the liberation of bitumen from solids. This role was analyzed through contact angle measurements in which it was demonstrated that all the CMC samples accelerated bitumen displacement and detachment from the illite surface. The CMC sample of the lowest molecular weight was found to be most effective in promoting bitumen displacement from the illite surface, and this action was attributed to the small effective size and high flexibility of the chain, which allowed the polymer to very closely approach the three-phase point of contact between the mineral, water, and bitumen. Wettability studies also revealed that CMC interacted very weakly with bitumen and did not permanently change the natural hydrophobicity of bitumen. Since a good rheological dispersant should not render bitumen hydrophilic and prevent bitumen extraction, the weak wetting action of CMC at the bitumen-solution interface is actually highly desirable. Overall, this dissertation demonstrated that CMC could play a significant role in improving the processability of low-grade oil sand ores, particularly since the polymer was also shown to be effective at neutral pH and low temperature.  ii  Preface  This dissertation is based on experimental work that was designed by my research supervisor, Dr. Marek Pawlik with my input. I conducted all the experimental work and analyzed the results under the supervision of Dr. Marek Pawlik. The composition of the oil sand ore studied in this dissertation was conducted by the Alberta Research Council. All copyrighted figures in the dissertation were used with permission of the copyright holders. A version of Chapter 3 has been submitted for publication under the title “Dilute solution properties of carboxymethyl celluloses of various molecular weights and degrees of substitution” with myself as the primary author and Dr. Marek Pawlik as the co-author. Part of the experimental results from Chapter 3 was presented at a conference as: Arinaitwe, E., & Pawlik, M., Solution conformation and flexibility of carboxymethyl cellulose, 2nd International Polysaccharide Conference on Polysaccharides as Source of Advanced and Sustainable Products, Wageningen, Netherlands, 2011. Versions of Chapters 4 and 5 of this dissertation will be submitted for publication as individual manuscripts.  iii  Table of Contents Abstract ................................................................................................................................................ ii  Preface ................................................................................................................................................. iii  Table of Contents ............................................................................................................................... iv  List of Tables ..................................................................................................................................... vii  List of Figures ................................................................................................................................... viii  Acknowledgments ............................................................................................................................ xiii  Dedication ......................................................................................................................................... xiv  1   Introduction ................................................................................................................................. 1  1.1  1.2  1.3   2   Oil sand ores as a source of bitumen .................................................................................................. 1  Objectives ........................................................................................................................................... 4  Scope of the dissertation..................................................................................................................... 4   Literature review......................................................................................................................... 6  2.1  Properties of oil sands and interactions between components ........................................................... 6  2.1.1  Electrical properties of interfaces .................................................................................................. 8  2.1.2  Bitumen–water interface .............................................................................................................. 11  2.1.2.1  Electrochemical properties .................................................................................................. 11  2.1.2.2  Interfacial properties ........................................................................................................... 13  2.1.3  Bitumen–mineral interface ........................................................................................................... 15  2.1.3.1  Bitumen liberation from a solid........................................................................................... 15  2.1.3.2  Surface electrical properties of quartz and clays ................................................................. 21  2.1.3.3  Bitumen–mineral interactions ............................................................................................. 23  2.1.3.4  Mineral–mineral interactions .............................................................................................. 26  2.1.3.5  Implications of bitumen–mineral and mineral–mineral interactions on oil sand slurry aggregation/dispersion .......................................................................................................................... 32  2.1.3.6  Role of chemical additives on bitumen–mineral and mineral–mineral interactions ........... 33  2.1.4  Bitumen wettability ...................................................................................................................... 35  2.2  Organic polymeric dispersants: carboxymethyl cellulose ................................................................ 38  2.2.1  Viscometric properties of carboxymethyl cellulose in solution ................................................... 42  2.2.2  Flexibility of carboxymethyl cellulose in solution ...................................................................... 52  2.2.3  Molecular weight and molecular weight distribution .................................................................. 59  2.3  Experimental assessment of aggregation/dispersion in concentrated suspensions. ......................... 67  2.3.1  Rheology ...................................................................................................................................... 67  2.3.1.1  Overview ............................................................................................................................. 67  2.3.1.2  Theory ................................................................................................................................. 67  2.3.1.3  Description of the flow behavior of suspensions ................................................................ 69  2.3.2  Sedimentation............................................................................................................................... 70  iv  3  Viscometric characterization of carboxymethyl celluloses of different molecular weights and degrees of substitution. .............................................................................................................. 72  3.1  Introduction ...................................................................................................................................... 72  3.2  Materials and methods...................................................................................................................... 73  3.2.1  Polymers and other chemicals...................................................................................................... 73  3.2.2  Viscosity measurements ............................................................................................................... 75  3.2.3  Aging of CMC solutions .............................................................................................................. 76  3.2.4  Sedimentation velocity analytical ultracentrifugation ................................................................. 76  3.3  Results and discussion ...................................................................................................................... 78  3.3.1  Molecular weight and molecular weight distribution .................................................................. 78  3.3.2  Intrinsic viscosities ....................................................................................................................... 88  3.3.2.1  Remarks on the aggregation state of CMC solutions .......................................................... 88  3.3.2.2  Aging of CMC solutions ..................................................................................................... 89  3.3.2.3  Determination of intrinsic viscosities .................................................................................. 91  3.3.3.4  Effect of pH and ionic strength ........................................................................................... 93  3.3.3.5  Expansion factor of CMC ................................................................................................... 98  3.3.3.6  [η] – MW relationship ........................................................................................................... 99  3.3.3.7  Persistence length .............................................................................................................. 102  3.4  Summary ........................................................................................................................................ 104   4   The aggregation/dispersion state of oil sand slurries: Influence of CMC.......................... 106  4.1  Introduction .................................................................................................................................... 106  4.2  Materials and methods.................................................................................................................... 106  4.2.1  Materials .................................................................................................................................... 106  4.2.2  Methods...................................................................................................................................... 108  4.2.2.1  Rheological measurements ................................................................................................ 108  4.2.2.2  Stability/settling tests ........................................................................................................ 113  4.2.2.3  Adsorption tests ................................................................................................................. 114  4.3  Results and discussion .................................................................................................................... 115  4.3.1  Baseline tests for oil sand slurries and influence of addition of CMC....................................... 115  4.3.2  Stability tests .............................................................................................................................. 120  4.3.3  Understanding the role of CMC: Results for bitumen-free solids ............................................. 122  4.3.3.1  Adsorption of CMC on bitumen-free solids ...................................................................... 126  4.3.3.2  Role of bitumen in aggregation of oil sand slurries .......................................................... 130  4.4  Summary ........................................................................................................................................ 132   5   Interaction of carboxymethyl cellulose with solids and bitumen........................................ 133  5.1  Introduction .................................................................................................................................... 133  5.2  Materials and methods.................................................................................................................... 134  5.2.1  Materials .................................................................................................................................... 134  5.2.2  Bitumen displacement from illite ............................................................................................... 135  5.2.3  Bitumen/air contact angle measurements .................................................................................. 137  5.3  Results and discussion .................................................................................................................... 141  5.3.1  Bitumen displacement from illite ............................................................................................... 141   v  5.3.1.1  Effect of the molecular weight of CMC ............................................................................ 141  5.3.1.2  Effect of the degree of substitution of CMC ..................................................................... 146  5.3.1.3  Effect of preadsorbing CMC onto illite ............................................................................. 147  5.3.2  Bitumen/air contact angle measurements .................................................................................. 149  5.3.2.1  Contact angle data analysis ............................................................................................... 149  5.3.2.2  Effect of pH for polymers of different degrees of substitution ......................................... 150  5.3.2.3  Effect of pH for polymers of different molecular weight ................................................. 154  5.3.2.4  Effect of pH on the contact angle evolution rate (from θ0 to θmax) .................................... 155  5.3.3  Summary .................................................................................................................................... 164   6   Conclusions, contributions to knowledge and future directions ......................................... 166  6.1  6.2  6.3   Conclusions .................................................................................................................................... 166  Contributions to knowledge ........................................................................................................... 170  Future directions ............................................................................................................................. 171   Bibliography .................................................................................................................................... 173  Appendices ....................................................................................................................................... 188  Appendix 1 ....................................................................................................................................... 188  Appendix 2 ....................................................................................................................................... 191  Appendix 3 ....................................................................................................................................... 192  Appendix 4 ....................................................................................................................................... 193  Appendix 5 ....................................................................................................................................... 194  Appendix 6 ....................................................................................................................................... 197  Appendix 7 ....................................................................................................................................... 199  Appendix 8 ....................................................................................................................................... 200   vi  List of Tables  Table 2.1. Classification of polymeric modifiers according to their molecular weight .................................... 39  Table 3.1. Molecular weights [g/mol] and polydispersities of the tested polymers.......................................... 86  Table 3.2. Expansion factors of CMC of different DS and MW at different pH ................................................ 99  Table 4.1. Composition of the tested oil sand ore and solids obtained from the ore. ..................................... 107   vii  List of Figures Figure 1.1. Desirable slurry properties in bitumen extraction versus tailings disposal (from Masliyah et al., 2011) .......................................................................................................................................................... 3  Figure 2.1. A schematic of the microscopic structure of the Athabasca oil sand (Takamura, 1982) ................. 7  Figure 2.2. Stern model of the EDL: Variation of potential with distance from the surface of a charged solid particle. .................................................................................................................................................... 10  Figure 2.3. Schematic illustration of the three-phase contact between oil sand components in slurry (solid, bitumen, and aqueous solution). The contact angle is measured through the bitumen phase. A Low contact angle, θ1 as a result of bitumen spreading at the solid–aqueous solution interface. B High contact angle, θ2 beneficial to bitumen liberation from the solid surface. .............................................. 16  Figure 2.4 Schematic illustration of the three-phase contact between bitumen, air, and aqueous phase in the captive bubble technique analogous to bitumen attachment to air during bitumen flotation. The contact angle is measured through the aqueous phase. A Low contact angle, θ1 between the air bubble and bitumen immediately after air bubble–bitumen contact. B High contact angle, θ2 showing spreading of the air bubble on the bitumen surface. C Bitumen – air bubble engulfment. .......................................... 36  Figure 2.5. Flow curves obtained from rheological measurements on concentrated coal suspensions in the presence of CMC (left) and PSS (right). ................................................................................................. 41  Figure 2.6. Yield stresses (obtained from Figure 2.5) as a function of dispersant concentration ..................... 41  Figure 2.7. A typical graph showing the concentration dependence of the reduced viscosity for polyelectrolytes........................................................................................................................................ 47  Figure 2.8. The persistence length Lp and contour length Lc of a linear polymer (from Harding, 1997). Lc represents the length of the fully extended chain. ................................................................................... 53  Figure 2.9. Schematic showing the interference optical system. Collimated light emerges from the source, A, as two narrow stripes. At the cell, B, one stripe passes through the reference solution, R, and the other passes through the sample, S. These two beams are brought together by the condensing lens, C, at its focal point, Iʹ. A cylinder lens, D, magnifies the image at Iʹ, so that the resultant fringes are greatly expanded and in focus on the camera, F. A camera lens, E, brings the the 2/3 plane of the cell, Iʹ, into focus on the camera. Thus, the image seen (inset) is the superposition of the cell and the fringes (from Laue, 1996). ............................................................................................................................................. 61  Figure 2.10. Schematic showing the concentration versus radial displacement at different times after start of an SV experiment. ................................................................................................................................... 62  Figure 2.11. Interference fringes versus radial displacement at different times for one of the CMCs tested in this dissertation. The sample was run at 45,000 rpm. In the XL-I instrument, a Fourier transformation  viii  converts the fringes into a record of concentration as a function of the radial displacement from the axis of rotation. ............................................................................................................................................... 62  Figure 2.12. Illustration of the types of flow behavior of various solid-liquid suspensions. A - Newtonian; B – pseudoplastic or shear thinning; C – shear thickening; D – Bingham plastic; E – yield shear thinning; F – yield shear thickening; and τy is the yield stress. .................................................................................. 69  Figure 3.1. Chemical structure of carboxymethyl cellulose showing a carboxymethyl-substituted unit in the C(6) position and two unsubstituted cellulose units. (Adapted from Coultate (1996)). .......................... 74  Figure 3.2. Sedimentation coefficient distributions ls-g*(s) vs. s for LM-CMC, MM-CMC and HM-CMC, DS 0.7. Sample concentration was 0.1 mg/mL in 0.1 mol/L NaCl, pH 6. The samples were centrifuged at 45,000 rpm (LM-CMC) and 40,000 rpm (MM-CMC and HM-CMC) at 25 °C. The S values indicated at the peaks correspond to the sedimentation coefficient at which most of the material in the sample sediments. ................................................................................................................................................ 80  Figure 3.3 A. The ls-g*(s) vs. s distribution for LM-CMC. The test conditions are the same as indicated in Figure 3.2. ................................................................................................................................................ 83  Figure 3.3 B. The c(s) distribution showing the molecular weight information of the peaks in the lsg*(s).distribution (Figure 3.3 A) for LM-CMC....................................................................................... 84  Figure 3.3 C Transformation of the ls-g*(s) vs. s distribution (Figure 3.3 A) to a molecular weight distribution f(M) versus M for LM-CMC. The f(M) distribution was integrated up to the maximum MW shown in c(s) distribution (Figure 3.3 B). The test conditions are the same as indicated in Figure 3.2.. 85  Figure 3.4. Molecular weight distributions f(M) versus MW for LM-CMC, MM-CMC and HM-CMC (DS 0.7) (Graph A) and for DS0.7, DS0.9 and DS1.2 (Graph B) obtained by transformation of the ls-g*(s) vs. s distribution using the Extended Fujita approach. The scaling law constants b = 0.29 and ks = 0.0101 were used in the transformation. Weight average molecular weights obtained by integration of the areas under the distributions are indicated........................................................................................................ 86  Figure 3.5. Reduced viscosities of DS0.7 and DS1.2 as a function of time in distilled water at 25 °C. ........... 90  Figure 3.6. Reduced viscosity vs. polymer concentration for DS1.2 in distilled water, 25°C at different pH.. 92  Figure 3.7. Fedors representation of the data from Figure 3.6 for DS1.2 in distilled water at 25 °C and different pH. ............................................................................................................................................ 93  Figure 3.8. Effect of pH and temperature on the intrinsic viscosities of the CMCs of different DS. MW = 293,000 – 310,000 g/mol. A: Distilled water; B: 0.01 mol/L NaCl. ....................................................... 94  Figure 3.9. Intrinsic viscosities of DS0.7 and DS1.2 at pH 7 and 25 °C as a function of ionic strength. Constant MW of 295,000 g/mol (DS0.7) and 293,000 g/mol (DS1.2). .................................................... 95  Figure 3.10. Effect of pH and temperature on the intrinsic viscosities of the tested CMCs of different MW (DS = 0.7). A: distilled water; B: 0.01mol/L NaCl. ........................................................................................ 97   ix  Figure 3.11. Logarithmic dependence of the intrinsic viscosity [η] as a function of molecular weight for CMC (MW range 123,000 – 715,000 g/mol) in 0.01 mol/L NaCl at 25 °C and different pH. The exponent α values indicated with an error estimate are obtained by using an average value of duplicate intrinsic viscosities for the lowest molecular weight polymer. The error bars on the data points correspond to the lowest and highest error obtained in log([η]). ....................................................................................... 100  Figure 3.12. Bohdanecky plot for CMC in distilled water and in 0.01 mol/L NaCl at different pH conditions. ............................................................................................................................................................... 102  Figure 4.1. Settling properties of oil sand slurries (55 wt. % solids, pH 8.5) at 0 and 5 minutes in the presence and absence of 250 g/t CMC. ................................................................................................................ 109  Figure 4.2. A schematic illustration of the elongated fixture designed for rheological measurements on settling suspensions. .............................................................................................................................. 111  Figure 4.3. Flow curves of baseline (oil sand ore slurries only) duplicate tests performed at 31.5 vol. % solids content, pH 8.5, 25 °C in distilled water. .............................................................................................. 116  Figure 4.4. Flow curves of oil sand ore slurries (ramping parts only) in the presence of 125 g/t and 250 g/t CMCs of different DS (A) and full flow curves of oil sand slurries in the presence of 125 g/t CMCs of different DS (B). Solids content = 31.5 vol. %, pH = 8.5, CMC MW = 293,000 – 310,000 g/mol........ 117  Figure 4.5. Flow curves for oil sand ore slurries obtained at different dosages of LM-CMC (molecular weight of 123,000 g/mol). The baseline case is also shown. Solids content = 31.5 vol. %, pH = 8.5. ............. 119  Figure 4.6. Flow curves for oil sand ore slurries obtained in the presence of 250 g/t HM-CMC (molecular weight of 715,000 g/mol). The baseline case is also shown.................................................................. 119  Figure 4.7. Reproducibility of flow curves for oil sand ore slurries obtained in the presence of LM-CMC (plot A) and HM-CMC (plot B), both at a dosage of 250 g/t. Solids content = 31.5 vol. %, pH = 8.5. ........ 120  Figure 4.8. Settling of oil sand slurries in the absence and presence of different dosages of CMC (DS0.7, MW = 295,000 g/mol). The slurries were prepared in distilled water. pH = 8.5, T = 25 °C, solids content 31.5 vol. %. Photographs were taken after 1 week of settling. ............................................................. 121  Figure 4.9. Dispersion coefficients and apparent viscosities of oil sand ore slurries plotted versus CMC (DS0.7, MW = 295,000 g/mol) dosage. .................................................................................................. 122  Figure 4.10. Flow curves for oil sand solid (without bitumen) slurries obtained in the presence of 250 g/t CMCs of different DS (DS0.7 and DS1.2) and MW (LM-CMC and HM-CMC). The baseline case is also shown in each case. ............................................................................................................................... 123  Figure 4.11. Reproducibility of flow curves for bitumen-free solid slurries and bitumen-free solids/DS0.7 mixture (left graph) and bitumen-free solids/HM-CMC mixture (right graph). CMC dosage was 250 g/t. ............................................................................................................................................................... 124   x  Figure 4.12. Settling of bitumen-free solids in the absence and presence of different dosages of HM-CMC The slurries were prepared in distilled water at the same conditions as in the rheological tests: pH = 8.5, temperature = 25 °C, solids content 31.5 vol. %. The photographs were taken after 1 week of settling. ............................................................................................................................................................... 125  Figure 4.13. Dispersion coefficients and apparent viscosities of bitumen-free solid slurries plotted versus HM-CMC dosage. ................................................................................................................................. 126  Figure 4.14. Residual CMC concentrations and adsorption densities after adsorption of CMC onto bitumenfree solids. The initial CMC concentration was ~313 mg/L (297 mg/L in the case of HM-CMC), dosage = 250 g/t................................................................................................................................................. 127  Figure 4.15. Flow curves (ramping parts) for oil sand ore and bitumen-free solid slurries as well as oil sand ore/CMC and bitumen-free solids/CMC mixtures. The dosage of CMC is 250 g/t. ............................. 131  Figure 5.1. Schematic showing the setup used for contact angle measurement during bitumen displacement from illite. .............................................................................................................................................. 137  Figure 5.2. Example of an image obtained with the movie-capturing system built into the FTA1000. ......... 139  Figure 5.3. Effect of bubble volume on contact angle evolution. ................................................................... 139  Figure 5.4. Evolution of the dynamic contact angle of bitumen on illite in distilled water (different pH) and CMC of same DS (0.7) but different molecular weights. The polymer concentration was 250 mg/L, temperature = 25 °C. .............................................................................................................................. 142  Figure 5.5. Effect of pH on the evolution of the dynamic contact angle of bitumen on illite in the presence of LM-CMC (Graph A), MM-CMC (Graph B) and HM-CMC (Graph C). The polymers are of the same DS (0.7). The polymer concentration was 250 mg/L, temperature = 25 °C. All tests were performed in distilled water. ....................................................................................................................................... 144  Figure 5.6. Schematic representation of the role of LM-CMC and HM-CMC in bitumen displacement from illite at pH 8. .......................................................................................................................................... 145  Figure 5.7. Evolution of the dynamic contact angle of bitumen on illite in distilled water (different pH) and CMC (pH 6) of similar MW (293,000 – 310,000 g/mol) but different degrees of substitution. The polymer concentration was 250 mg/L, temperature = 25 °C. ................................................................ 146  Figure 5.8. Evolution of the dynamic contact angle of bitumen on illite in distilled water (different pH) and CMC of similar MW (293,000 – 310,000 g/mol) but different degrees of substitution. The polymer concentration was 250 mg/L, temperature = 40 °C. .............................................................................. 147  Figure 5.9. Evolution of the dynamic contact angle of bitumen on illite preconditioned in distilled water (baseline) and in CMC of DS 0.7, MW 295,000 g/mol (in order to preadsorb CMC on illite prior to the bitumen displacement test). The polymer concentration was 250 mg/L, temperature = 25 °C. ........... 148   xi  Figure 5.10. Evolution of the contact angle of an air bubble in contact with a bitumen-coated glass slide immersed in 0.01 mol/L NaCl. Temperature = 25 °C, pH 8.5. The solid line shows the fit to the experimental contact angle data. ........................................................................................................... 150  Figure 5.11. Effect of pH on initial (θ0) and steady-state (θmax) contact angles for the baseline (0.01 mol/L NaCl) case and CMC polymers of different DS. Temperature = 25 °C. ............................................... 151  Figure 5.12. Evolution of the contact angle of an air bubble in contact with a bitumen-coated glass slide immersed in 0.01 mol/L NaCl and CMC of different DS. Data during the first 10 seconds following air bubble-bitumen contact is shown. Temperature = 25 °C, pH = 3, 6, 8.5 and 10.5. .............................. 153  Figure 5.13. Effect of pH on initial (θ0) and steady-state (θmax) contact angles for the baseline (0.01 mol/L NaCl) case and CMC polymers of different MW. Temperature = 25 °C................................................ 155  Figure 5.14. Evolution of the contact angle of an air bubble in contact with a bitumen-coated glass slide immersed in 0.01 mol/L NaCl and CMC of different MW. Data during the first 10 seconds following air bubble–bitumen contact is shown. Temperature = 25 °C, pH = 3, 6, and 8.5. ...................................... 155  Figure 5.15. Contact angle evolution rate, k of an air bubble in contact with bitumen as a function of pH. Data sets used to obtain k are from the time of initial contact of the air bubble with bitumen to the steadystate contact angle. Baseline case (0.01 mol/L NaCl) and CMCs of different DS are shown. Temperature = 25 °C. ............................................................................................................................ 156  Figure 5.16. Contact angle evolution rate, k of an air bubble in contact with bitumen as a function of pH. Data sets used to obtain k are from the time of initial contact of the air bubble with bitumen to the steadystate contact angle. Baseline case (0.01 mol/L NaCl) and CMCs of different MW are shown. Temperature = 25 °C. ............................................................................................................................ 157  Figure 5.17. Illustration of the differences in the evolution of the air bubble contact angle rate k by plotting experimental contact angles at different times. Left graph: pH 3. Right graph: pH 8.5........................ 157  Figure 5.18. Effect of pH on initial and steady-state contact angles for the baseline (0.01 mol/L NaCl) case, polystyrene sulfonate, humic acids, and hydroxypropyl cellulose. Polymer concentration = 50 mg/L. Temperature = 25 °C. ............................................................................................................................ 161  Figure 5.19. Contact angle evolution rate, k of an air bubble in contact with bitumen in the absence and presence of different polymers. A: Baseline, polystyrene sulfonate, humic acids and hydroxypropyl cellulose as a function of pH (polymer concentration = 50 mg/L). B: Hydroxypropyl cellulose as a function of polymer concentration at pH 6. Data sets used to obtain k are from the time of initial contact of the air bubble with bitumen to the steady-state contact angle. Temperature = 25 °C. ...................... 161   xii  Acknowledgments I have been blessed to work with Dr. Marek Pawlik who has guided me through this work and made it a worthwhile journey. Thank you for giving me this opportunity. I greatly appreciate your insightful comments and thoughtful suggestions. Your mentoring has enriched my ability to conduct research and also shaped me as a person, and for that, I am forever grateful. I would like to thank my supervisory committee members, Professors Bern Klein and Ian Frigaard for their valuable comments and expert feedback that have been instrumental in completing this dissertation. This work was accomplished due to the assistance provided by Mrs. Sally Finora. Thank you for making my work in the surface chemistry lab as enjoyable as it could be. You will always have a special place in the memory of my stay at UBC. To my colleagues Dr. Leopoldo Gutierrez, Dr. AJ Gunson, Jophat Engwayu, Claudio Garcia, Avishan Atrafi, Vivian Ferrera, Givemore Sakuhuni and Juliana Parreira: It has been an honor to know you. Many thanks to Mr. Pius Lo for his assistance in the CMP lab and to Ms. Maria Lui and Ms. Leslie Nichols for the administrative support. I would like to express special thanks to Dr. Jennifer Hinton who encouraged me to come and pursue graduate studies at UBC. I also wish to thank the Department of Geological Survey and Mines (Uganda) who gave me permission to come to Canada for my graduate studies. Financial support by UBC (4YF, Kitsault Community Scholarship, Cy and Emerald Keyes Fellowship and Tuition Fee Awards) and NSERC-CNRL (Collaborative Research and Development Grant) is gratefully acknowledged. I cannot find words to express my thanks to Mom and Dad who gave me the foundation to be what I am today. Mom, thank you for all the sacrifices you made (and continue to make) to enable me reach this stage. Dad, thank you for your encouragement and unwavering belief in me. To my family: Henry, Medard, late Doreen, Gloria, Clare, Sam, Mercy, Joshua, Sonia, and Cynthia: Thank you for your love and prayers. To my family-in-law: I greatly appreciate your support. Dr. Jalia Kangave: Thank you for your friendship which places you among the ‘real family’. The years I spent as a graduate student took a lot of time away from my wife Annette and all she gave back was unlimited love, patience, understanding, and support, always encouraging and motivating me to excel. Her inspiration in completing this dissertation is immeasurable. My other inspiration came from my children, Shawn, Asher, and Jesse. They were always there to welcome me back home with warm hugs and happy smiles which gave me energy to keep going. xiii  Dedication  To Mom and Dad: these are the fruits of your sacrifices through the years. Thank you,  To the loving memory of my daughter: Audrey Atuhaire. I miss you every day, and To Annette, Shawn, Asher, Jesse: What you contributed is immense and the least I can do is to dedicate this dissertation to you.  xiv  CHAPTER 1 1 Introduction 1.1 Oil sand ores as a source of bitumen The oil sand deposits in the province of Alberta are estimated at 1.74 trillion barrels out of which about 300 billion barrels are recoverable using current mining and processing technologies. The oil sands are sources of crude bitumen, recovered through a hot water extraction process pioneered by Karl Clark in the 1920’s, and upgraded to produce essential commodities such as gasoline, jet and diesel fuel. Economic and efficient recovery of bitumen from the Alberta oil sands deposits is very critical in meeting Canada’s ever-growing demand for oil estimated at about 2 million barrels per day (International Energy Agency, 2012). The long-term future of oil sands as a sustainable and economic source of oil relies on improvement of current technologies and development of new technologies to process more complex ores as the high-grade (good processing) ores get depleted. Oil sand ores can be viewed as a three-phase mixture consisting of bitumen, solids (silica sand and clays) and water. Ores are defined by the bitumen content and by the fraction of the inorganic mineral phase finer than 44 µm in size. Typically, ores with a bitumen content higher than 10% and fines content of less than 10% are classified as good processing ores whereas those with low bitumen content ranging from ~6 – 10 wt. % and high fines content up to 50 wt. % are classified as poor processing ores. In Alberta’s unconsolidated oil sand ores, bitumen is attached to sand grains and it must be released and recovered through the hot water based extraction process (Clark and Pasternack, 1932). This process is solely based on the differences in the surface properties of the oily organic phase (bitumen) and the mineral inorganic components (sand and clays) of oil sand ores. As currently practiced, the pipeline transport of oil sand slurries from the mine to the extraction plant (hydrotransport) serves as a pre-conditioning stage in which bitumen droplets become liberated from the mineral grains by addition of chemical aids, typically sodium hydroxide. The free hydrophobic bitumen easily attaches to entrained or introduced air bubbles, and the resulting bitumen–air aggregates are recovered in primary separation cells (PSC) as the froth product, leaving the hydrophilic minerals in the pulp as tailings. The tailings from the primary separation stage are often refloated to maximize bitumen recovery. The froth products normally contain on average 60% bitumen, 30% water and 10% solids (Masliyah et al., 2004). Such froth products are de-aerated and 1  mixed with a diluent in a re-cleaning stage to remove water and residual solid (mineral) gangue. The final tailings containing a small amount of unrecovered bitumen are typically discharged into a tailings pond. The hot water extraction process provides high (more than 90%) bitumen recoveries for good processing ores. However, the economic recovery of bitumen from poor processing ores is still a major challenge facing oil sand operators. The low bitumen recovery from poor processing ores can be attributed to the high content of mineral fines, and most critically to the clay fines (< 3 µm) which, through steric hindrance, inhibit bitumen–air attachment resulting in low bitumen recoveries and poor froth quality (Masliyah et al., 2011). Furthermore, the entire bitumen extraction process is controlled by physico-chemical and interfacial properties of the different oil sand components, both at a macro and microscale. The multiphase nature of oil sands suggests that bitumen–mineral, mineral–air and bitumen–air interactions define the processability of the oil sands ores. These interactions, for example, control the aggregation/dispersion state of the oil sand slurries which in turn impacts the performance of the hot water extraction process. The unit operations that are mostly affected by the aggregation/dispersion state of the oil sand slurries are the pipeline hydrotransport and flotation in the PSC. An aggregated slurry exhibits a high viscosity and high yield stress, which are both detrimental to slurry pipelining and bitumen extraction in the PSC. In the pipeline hydrotransport, high viscosities and high yield stresses require high energy requirements to pump the oil sand slurries while in the PSC, aggregation hinders the rise velocity and eventual flotation of the bitumen–air aggregates. On the other hand, a dispersed slurry exhibits a low viscosity and yield stress resulting in low energy requirements during pumping in the hydrotransport stage, and at the same time facilitating bitumen flotation in the PSC by minimizing bitumen–mineral and mineral–air interactions. Therefore, from the viewpoint of slurry pipelining and bitumen extraction, a dispersed slurry with desirable flow properties (low viscosity and yield stress) is favoured as shown in Figure 1.1 (Masliyah et al., 2011). Yet, there is no open literature on the aggregation/dispersion behavior of oil sand slurries perhaps due to the inherent difficulty of testing oil sand slurries and the lack of bench-scale methods and instrumentation, particularly for characterizing rheological properties.  2  Figure 1.1. Desirable slurry properties in bitumen extraction versus tailings disposal (from Masliyah et al., 2011)  Current commercial operations utilize traditional inorganic chemical additives (dispersants) such as sodium hydroxide and sodium silicates to facilitate bitumen liberation and solids dispersion in the hydrotransport pipeline. These traditional dispersants, which act primarily by controlling the pH of the slurry, are ineffective for poor processing ores except at high dosages exceeding 1,000 g/ton (Bichard, 1987). The high dosages cause accumulation of ions in process water that causes an undesirable increase in slurry pH (due to their strong hydrolysis in water) and possible modification of the bitumen surface charge, which can pose difficulties in tailings management and water recycling (Masliyah et al., 2004). Alternatives include organic dispersants such as carboxymethyl cellulose which was shown to be a powerful dispersant for systems with similar mineral components (Pawlik, 2005). The low level of dosages often needed to achieve slurry dispersion make such polymeric dispersants economically attractive. The effectiveness of organic dispersants lies in their dual dispersion mechanism when added to suspensions: (1) they adsorb on the solid surfaces and control the charge density at the solid/liquid interface, and (2) they provide steric hindrance/stabilization. Inorganic dispersants and pH modifiers are only capable of achieving the first purpose. The ability of organic dispersants to disperse and “blind” fine solids and to prevent them from coating bitumen can improve bitumen flotation in poor processing ores containing a high amount of fine solid minerals. 3  In this dissertation, the influence of an organic polymeric dispersant (carboxymethyl cellulose) on the aggregation/dispersion state of a poor oil sand ore was investigated. Carboxymethyl celluloses of different molecular weights and degrees of substitution were characterized, and their ability to disperse oil sand ore components was investigated through rheological measurements and sedimentation tests. The mechanisms through which the polymers dispersed the oil sand components to achieve slurry dispersion were also proposed. 1.2 Objectives The present study aims at understanding the aggregation/dispersion behaviour of oil sand slurries and elucidating the dispersing action of polysaccharides to produce slurries of desirable flow properties. The specific objectives are as follows: 1) To characterize the solution behavior of polysaccharides through intrinsic viscosity measurements, 2) To establish the effect of the polysaccharides on the aggregation/dispersion state of oil sand slurries, 3) To investigate the effect of the tested polysaccharides on the interaction between bitumen and solids under different physicochemical conditions, 4) To investigate the effect of the tested polysaccharides on the wettability of bitumen under different physicochemical conditions. 1.3 Scope of the dissertation In the hydrotransport of oil sand slurries, it is highly desirable to stabilize the slurries against aggregation. It is also desirable to operate at the maximum possible solids loading. Thus the aggregation or dispersion of oil sand slurries is an important area that needs systematic investigation. It is hypothesized that the desired slurry properties can be achieved by use of organic polyelectrolyte additives (dispersants) that are more effective than inorganic additives especially for low-grade oil sand slurries . In the present work, the dispersing action of carboxymethyl cellulose (CMC) on lowgrade aggregated oil sand slurries was investigated. 4  Since solution properties of polymers influence their dispersing efficacy, a systematic characterization of the behaviour of CMC in solutions of different pH, temperature and ionic strength was undertaken. The principle technique used to study the solution properties was dilute solution viscometry. The molecular weights and molecular weight distributions of the polymers were determined by analytical ultracentrifugation. The results from this part of the study, covered in Chapter 3, provided a means of understanding the dispersing action of the polymers. The effect of polysaccharides of different properties (molecular weight and degree of substitution) on the aggregation/dispersion state of oil sand ore slurries and bitumen-free solids extracted from oil sand ores was demonstrated through rheological and sedimentation tests. The results of this part are presented in Chapter 4. To further understand the mechanism of action of the polymers in dispersing oil sand slurries, the interaction of the polymers with a bitumen-coated clay surface was studied. These tests involved measuring the bitumen displacement rate from the clay surface in the presence of the polymers at different solution conditions. In addition, the interaction of the polymers with a bitumen surface to which an air bubble was attached provided information about the effect that the polymers have on the wettability of bitumen. Moreover, the rheological and sedimentation tests on bitumen-free solids extracted by removing bitumen from oil sand ores yielded information about the interaction of the polymers with solids contained in the oil sand ores. Based on these results (presented in Chapter 5), a mechanism of action of the polymers was proposed.  5  CHAPTER 2  2 Literature review 2.1 Properties of oil sands and interactions between components There are three major oil sand deposits in the Western Canada Sedimentary Basin: Athabasca, Peace River, and Cold Lake. The Athabasca oil sands are the largest and most accessible. The reservoir bed is located 0–500m below ground surface making surface mining possible in many locations, although only about 20% of all oil sands resources are recoverable with this method – the remaining 80% are recoverable through in-situ mining technologies. Their origin has been debated by several geoscientists for decades, but the most predominant theory indicates that most of the oil was introduced into the McMurray Formation after it was deposited from highly organic Cretaceous shales, while some of the oil may have been derived from the erosion of seeps in truncated Devonian carbonates (Conybeare, 1966). The history of the geological transport and deposition of the original McMurry sediments had a profound control on the mineralogy and grade of the oil sands (Mossop, 1980). In zones where the mode was sluggish or stagnant, much silt and clay was deposited yielding shale sands of low porosity and permeability. When bitumen was introduced into the McMurray Formation, these shales could not accept high amounts of bitumen and are described today as “poor processing” or “lean” ores. On the other hand, in zones originally occupied by river channels, the depositional flow regime was strong and fine silt and clays were kept in suspension; and only well-sorted sands of high porosity and permeability were deposited. These sands were able to accept high volumes of oil resulting in thick, high-grade oil sands known today as “good processing ores.” Good processing ores have the following features: an average bitumen content of 12–13 wt.% and as high as 18 wt.% in some samples (Camp, 1974), solid fines content less that 10%, low divalent cation content in process water, and minimum weathering. Poor processing ores, on the other hand, are characterized by low bitumen content ranging from ~6–10.5 wt.%, high fines content up to 50 wt.%, and presence of some divalent cations in process water. Therefore, a typical oil sand ore contains up to 10 wt.% bitumen, 85 wt.% inorganic minerals, 5 wt.% water, and small amounts of 6  organic materials, gases and volatiles. The inorganic minerals of the Alberta oil sands are ~90% quartz with minor amounts of K-feldspar, muscovite, clays, and heavy minerals. The clay minerals are predominantly kaolinite and illite with lesser quantities of montmorillonite and smectite while the heavy minerals are predominantly titanium and zirconium bearing minerals such as rutile, zircon and ilmenite (Alberta Chamber of Resources, 1996; Kaminsky et al., 2009). On average, the Athabasca deposits contain 0.35% TiO2 and 0.032% ZrO2. During the hot water extraction process, the heavy minerals are concentrated in the bitumen froth. The study by the Alberta Chamber of Resources showed that further treatment of the bitumen froth to recover bitumen leaves about 11.5% TiO2 and 3.40% ZrO2 in the froth treatment tailings, which makes oil sand ores quite a valuable resource for the heavy minerals. The ores also contain organics which dissolve or become suspended in the water phase during slurrying (Clark, 1944). These organics were found to be surface active by Bowman (1967) who isolated them by foam fractionation techniques. The single most important aspect of the Alberta oil sands lies in the structural arrangement of the oil sand components. The ores consist of mineral grains, mostly quartz covered by bitumen but with a ~10 nm (0.01 µm) water film separating the mineral grains from the bitumen. More water exists at the mineral grain-to-grain boundaries and finally, some water is contained in fine mineral clusters. A microscopic structure proposed by Takamura (1982) is shown in the schematic in Figure 2.1.    Figure 2.1. A schematic of the microscopic structure of the Athabasca oil sand (Takamura, 1982)  The ~10–nm thin water film between the mineral grains and bitumen is stabilized by the repulsive (electric double layer) and attractive (van der Waals) forces at the bitumen–water and mineral–water interfaces. This means that the mineral grains are strongly water wet (hydrophilic) which makes it possible to extract the naturally hydrophobic bitumen using the hot water based extraction process. 7  However, it is noteworthy that there is no direct experimental evidence of this structural model and researchers such as Zajic et al. (1981) hold a different view. Zajic et al. investigated some Athabasca oil sand samples using electron microscopy and established that in some ores, water and bitumen exist as a water-in-bitumen emulsion, with bitumen forming the continuous phase and the water forming droplets 0.02 to 0.5 µm in size. However, these authors concurred with Takamura on the thickness of the water layer adsorbed on the sand grains. Czarnecki et al. (2005) analyzed Takamura’s model based on ‘first principles’ and theoretical evidence by analyzing the disjoining pressure isotherms and hence the stability of the water film in the quartz/water/system and concluded that acidic ores i.e., ores that contain connate water with a pH in the acidic range, are an important exception to this model. It is apparent that the mineralogical composition of oil sands plays a vital role in the processability of oil sands. Indeed, researchers involved in the early development of the hot water extraction process qualitatively identified clays as the “most troublesome impurity in bituminous sand from the standpoint of commercial operations” (Clark and Pasternack, 1932). Both the physical, chemical, and electrical properties of the oil sand components influence the bitumen–mineral, bitumen–water, bitumen–air and mineral–air interactions, and collectively contribute to the aggregation/dispersion state of the system. Key to understanding the aggregation/dispersion behaviour of oil sand slurries is an in-depth look at the bitumen–mineral, bitumen–water, and bitumen–air interfacial interactions. 2.1.1  Electrical properties of interfaces  When solid particles are immersed in water, they invariably acquire charges. There are many possible mechanisms that can explain the presence of electrical charges at interfaces, the main ones being (Kitchener, 1969):   Ionization of surface groups such as –COOH (acidic) or –OH (basic). The dissociation of such surface groups is a strong function of the pH in the vicinity of the surface. The magnitude of the surface charge depends on the pH and the acidic or basic strength of the surface groups.    Ion adsorption: preferential adsorption of multivalent ions, surfactants, and polyelectrolytes from a solution on an initially uncharged surface.  8    Ion dissolution: Differential dissolution of ions from the surface of sparingly soluble minerals such as silver iodide (AgI). Preferential dissolution of Ag+ leaves a negatively charged surface while preferential dissolution of I− leaves a positively charged surface.    Isomorphous substitution: Exchange of a structural ion or adsorbed intercalated ion with one of lower valency producing a negatively charged surface. For example, in illite clays, the substitution of Al3+ for Si4+ in the tetrahedral sheet and (Mg2+, Fe2+) for Al3+ in the octahedral sheet is responsible for illite’s negative charge. Kaolinite also becomes charged by isomorphous substitution of silicon atoms by aluminum atoms in the tetrahedral crystal lattice or aluminium atoms by iron or magnesium atoms in the alumina octahedral layer    Charged crystal surfaces: Crystals may be broken by wetting or milling exposing fresh surfaces with different properties. For example in kaolinite, surface breakage at the edges exposes AlOH groups which take up H+ imparting a positive charge on the edges. At the same time, the kaolinite surfaces may be negatively or positively charged (depending on pH) giving rise to surface charge heterogeneities. The formation of an interface between the solid and solution involves rearrangement of species  that result in an ion concentration profile as counter-ions are attracted to the charged solid forming an electrical double layer (EDL). The counter-ions strongly attracted to the solid surface form the Stern layer (Figure 2.2). Further away from the solid surface, weakly attached counter ions form a layer referred to as the diffuse layer. Excellent reviews of the concept of the EDL can be found elsewhere (for example, Hunter, 1981) and will not be repeated here. The electric properties of solid/liquid interfaces can readily be determined from potentiometric titration of the solid with an electrolyte solution and following the change in concentration of the potential determining ions in the solution. Alternatively, the charge at the solid surface can be determined from electrokinetic methods, particularly those based on the electrophoretic effect. In this method, a solid particle suspended in an electrolyte moves in a tangential motion to the aqueous phase and an electric field is induced as a result of a charge (potential) gradient existing within the diffuse layer of the EDL. This potential gradient is known as the surface potential, and its magnitude is related to the surface charge density and the thickness of the EDL. As shown in Figure 2.2, in the Stern model of the EDL, the surface potential decreases with distance from the surface of the charged solid due to the diminishing electrical force through the diffuse layer. Of significance is the 9  electric potential at the boundary (shear plane) between the Stern layer and the diffuse layer. This electric potential, known as the zeta (ζ) potential is related to the mobility υ of the solid particle and determines the aggregation or stability of fine particles suspended in a liquid. The zeta potential can be viewed as an effective electrical potential that is sensed by a charged particle as it approaches another charged particle (Hunter, 1981). Commercial electrophoretic instruments are available to measure the mobility υ of the solid particle at different values of the applied electric field. The ζpotential is obtained from the mobility using the Smoluchowski equation (1903, 1921). [2.1 where  is the dielectric constant, E is the applied electric field and η is the liquid viscosity.  Figure 2.2. Stern model of the EDL: Variation of potential with distance from the surface of a charged solid particle.  The electrophoresis method is based on the assumptions that the solid particles are rigid, have negligible surface conductivity, and the local Debye length (thickness of the EDL) is much smaller than the local radii of curvature. Thus, it should be noted that the method has limitations when applied to solid particles or materials with characteristics that significantly deviate from the assumptions on which the method is based. For example, kaolinite particles, which constitute an important fraction of oil sand ores from a colloid chemistry perspective, have a hexagonal platy-like 10  shape, heterogeneous surface charges, and application of the electrophoresis method to determine their electrokinetic properties has resulted in conflicting results as explicated by Gupta and Miller (2010). In addition, the electrophoretic mobility or ζ-potential measurements of bitumen droplets (Takamura and Chow, 1985; Chow and Takamura, 1988; Dai and Chung, 1995; Lin, 2012) only produce average values due to the bitumen’s surface charge heterogeneities and deformability. Recently, the electrophoresis method has gradually been replaced by Atomic Force Microscopy (AFM) as a method of choice for measuring electrokinetic properties of materials. In the AFM method, forces are measured between a force-sensing probe and a substrate (sample of interest). The substrate is mounted on a piezoelectric scanner and is moved towards or away from the probe mounted on a cantilever. The forces between the substrate and probe cause the cantilever to deflect either upwards or downwards depending on whether the net force is attractive or repulsive. The cantilever deflection is monitored by a laser photodiode detector which records a current shift proportional to the cantilever deflection. The raw data curves (photodiode detector current versus piezoelectric scanner displacement) are converted to force–versus–distance curves using AFM software. Important surface properties such as surface charge density and colloidal interaction forces can be directly obtained from the force curves. 2.1.2  Bitumen–water interface  In the hot water extraction process, the electrochemical properties and surface tension of bitumen determine the type and magnitude of interactions at the bitumen–water interface. The bitumen surface charge, surface tension, and bitumen–water interfacial tension play a role in bitumen detachment from the quartz grains to the aqueous phase, and bitumen spreading at the air bubble surface to facilitate bitumen–air attachment (Drelich and Miller, 1994). 2.1.2.1 Electrochemical properties The surface charge of bitumen has implications in its interaction with the inorganic components of oil sands, mainly quartz. The presence of natural aliphatic carboxylic and sulfonate groups in oil sands is well known. They were reviewed in detail by Shramm et al. (2000). During oil sand processing, increasing the pH through addition of sodium hydroxide triggers the dissociation of carboxylic groups belonging to the natural surfactants, leaving a negative charge on the bitumen surface (Takamura and Chow, 1985). Based on micro-electrophoresis tests on bitumen from 11  Athabasca oil sands, these researchers used the ionizable surface-group model of Healy and White (1978) to show that the surface properties of bitumen developed as a result of the dissociation of potential determining ions from ionizable surface groups (carboxylic groups) into solution (Equation 2.2). .............  [2.2  Dai and Chung (1995) showed that the isoelectric point (pH at which the zeta potential is zero) of Athabasca bitumen occurred at pH = 3, and the surface became more negatively charged at higher pH. The positive bitumen charge at pH < 3 was attributed to cationic, organic amine-type species present in bitumen. The surface also became more negatively charged at high temperature. Drelich et al. (2007) measured colloidal forces between an AFM tip and a bitumen surface conditioned in water or 1mM KCl. They used the bitumen surface charge as one of the fitting parameters in describing the AFM tip–water/1mM KCl–bitumen colloidal forces, and calculated the bitumen surface potential from the fitted surface charge densities using the Graham equation (described in Israelachvili, 1991) At pH 9, the calculated surface potentials varied from -45 to -110 mV across 500 nm from the bitumen–aqueous solution interface. The large variation in the surface potentials was a result of the surface charge heterogeneities at the bitumen–aqueous solution interface. Liu et al. (2004a) conducted electrophoretic mobility measurements on a similar bitumen sample and obtained an average zeta potential value of -80 mV. Other researchers have shown that the zeta potential of bitumen is negative above pH of about 3 with the magnitude of the negative charge changing only due to the different experimental conditions employed (Liu et al., 2002; Takamura and Chow, 1983; Lin, 2012). It is important to note that the presence of additives (such as surfactants, polymers, multivalent cations) can significantly modify the electrical properties of the bitumen–aqueous solution interface depending on the nature and concentration of the additives. For example, Gan et al. (2007) reported that bitumen droplets (isoelectric point at 3.8) were highly negatively charged in alkaline solution with zeta potentials of about -120 mV in the absence of multivalent metal cations. However, the addition of 0.001 mol/L Ca2+ reduced the zeta potential by about 80 mV to about -40 mV in the alkaline pH region. The reduction of the bitumen zeta potential by Ca2+ highlights the role that multivalent cations play in the coagulation of bitumen and mineral particles. A similar result to that of Gan et al. (2007) was obtained by Liu et al. (2004b) who determined that the zeta potential 12  distribution of bitumen droplets dispersed in a 1 mM KCl solution containing 1 mM Ca2+ at pH 8.2 was centred at -38 mV. When Gan et al. added 0.001 mol/L Fe2+ to the bitumen emulsion, the isoelectric point of bitumen droplets shifted from 3.8 to 8.2 but the addition of 0.001 mol/L citric acid, a weak tri-carboxylic organic acid, restored the zeta potentials of bitumen to negative values over the entire pH range tested (3 to 12). This result showed that a reagent such as citric acid could be used to prevent bitumen–mineral coagulation by controlling the charge at the bitumen–water interface. 2.1.2.2 Interfacial properties Several studies (Isaacs and Smolek, 1983; Potoczny et al., 1984; Vargha-butler et al., 1988; Drelich et al., 1994, Wang et al, 2010) were carried out on surface tension measurements of bitumen extracted from oil sands. Isaacs and Smolek, 1983 and Drelich and Miller, 1994 conducted measurements in the presence of chemical additives to investigate the effect of surface–active chemicals on the bitumen–aqueous solution interfacial tension. As discussed by Isaacs and Smolek (1983), the presence of surfactants alters the bitumen– aqueous solution interfacial tension. Isaacs and Smolek determined the surface tension of bitumen extracted from the Athabasca oil sands in the presence of several surfactants and at different temperatures and salinity (NaCl) concentrations. In the absence of surfactants, the bitumen–water interfacial tensions decreased by 0.7 mN/m-1 per °C. However, in the presence of surfactants containing sulfonate functional groups, the interfacial tension decreased to <0.1 mN/m-1, but only in the presence of added electrolyte. The authors attributed this dramatic decrease in interfacial tension to formation of a surfactant-rich third phase at the oleic–aqueous interface. It should be pointed out that the value of 0.1 mN/m-1 corresponded to a sodium chloride concentration of 10 g/L and a sulfonate concentration of 2 g/L. These conditions apply only to the steam injection recovery of bitumen by in-situ technologies, which was the subject of Isaacs and Smolek’s study. However, a similar qualitative effect of lower magnitude can be expected in the hot water extraction process of surface-mined oil sands. A synthetic sulfonate surfactant was also tested in the absence of added electrolyte. Values of interfacial tension between bitumen and D2O at 50 °C decreased from ~22 mN/m-1 at a surfactant concentration of 10 mg/L to ~3 mN/m-1 at ~300 mg/L. The authors noted that the appearance of two discontinuities in the interfacial tension–surfactant concentration curve was a result of a change in the microstructure of the surfactant at a certain critical concentration. The 13  change in the microstructure was attributed to an increase in ionic strength of the aqueous phase with increasing surfactant concentration due to the large amount of ionic impurities in the surfactant sample. These results demonstrate that in addition to reducing the bitumen–water interfacial tension using surfactants, there could be possible changes in the surfactant structure and chemical composition and these could be more significant at high surfactant concentrations. For example, when nonionic surfactants (polyethoxylated nonyl phenols) were used as collectors for oxidized coal flotation (Harris et al., 1995), there existed an optimum collector dosage for maximum recovery above which there was a marked decrease in the recovery. It should be noted that the oxidized coal used in their study was already hydrophilic due to formation of oxygenated functional groups on their surface during the oxidation process and the role of the surfactants was to restore the flotability of the oxidized coal. However, at higher dosages of the surfactants, the authors postulated that the coal surface became more hydrophilic due to the reverse adsorption of a second layer of surfactant molecules in reverse orientation in which the hydrophilic groups of the surfactants oriented towards the aqueous phase. The increase in the hydrophilicity of the coal surface accounted for the decrease in the coal recovery. The same effect can be expected during extraction of bitumen from oil sand slurries. The work of Drelich and Miller (1994) indicated that controlling the interfacial tension of the bitumen–water interface by increasing pH and concentration of sodium tripolyphosphate, an inorganic dispersant, enhanced bitumen recovery from Whiterocks (Utah, USA) tar sands. In the absence of surfactants, the interfacial tension for the bitumen–water interface showed the highest values in the pH range 4.8 to 8.7. The interfacial tension decreased sharply both in acidic and alkaline solutions. The presence of carboxylic acids in bitumen at high pH were thought to be responsible for the interfacial activity in the alkaline range while it was not clear which compounds were responsible for the sharp decrease in the acidic region. There was a significant effect of sodium tripolyphosphate on the interfacial tension of the bituminous phase (diluted with 10 wt. % kerosene)–aqueous phase system at 60 oC. The authors also determined contact angles for bitumen (diluted with kerosene) droplets at mineral particle surfaces, and used them with the interfacial tension data to calculate the work of adhesion, WA, between the bitumen and aqueous phase. This analysis indicated that WA decreased significantly with a decrease in the bitumen/water interfacial tension in the presence of up to 0.05M sodium tripolyphosphate. To correlate the interfacial tension data with bitumen recovery, hot water extraction experiments were performed using an oil sand 14  sample (7.6 wt. % bitumen) from Whiterocks. The interfacial tension was modulated by pH and sodium tripolyphosphate concentration. The results indicated that an increase in bitumen recovery was observed with decreasing interfacial tension. Wang et al. (2010) also determined that the migration of surface-active components from the bulk bitumen to the bitumen–water interface was responsible for the sharp reduction of the surface tension of water at high pH. In the absence of bitumen, the water surface tension was independent of pH while the surface tension of water in contact with bitumen decreased from ~70 mN/m to ~61 mN/m with the sharpest reduction occurring at pH above ~10.5. The preceding studies show that bitumen liberation from solid particles strongly depends on the bitumen–water interfacial tension and that there is an advantageous influence of a decrease in the interfacial tension at the water–bitumen interface on bitumen recovery in the hot water extraction process. 2.1.3  Bitumen–mineral interface  Even though it is understood that a thin film of water separates bitumen from the mineral grains, the existence of direct bitumen–mineral interactions is well known. This section will review the bitumen liberation process from a solid surface, and the role that surface properties of minerals play in oil sand slurries. 2.1.3.1 Bitumen liberation from a solid Bitumen liberation from a solid surface follows three steps as described by Basu et al. (1996) and Wang et al. (2011): (1) the formation of a three-phase contact (TPC) line between bitumen, solid and aqueous phase (Figure 2.3A). The TPC formation largely depends on the generation of natural surfactants from bitumen, (2) bitumen recession or displacement from the solid surface. The rate of displacement from the solid surface is controlled by the viscosity of the bitumen and the quartz– bitumen, bitumen–aqueous solution and quartz–aqueous solution interfacial tensions. The interfacial tensions determine the magnitude of the contact angle at the TPC line, and (3) bitumen detachment from the solid surface. When the bitumen has receded to form a high contact angle (Figure 2.3B), it quickly becomes detached from the solid surface under favorable hydrodynamic conditions.  15  The three-phase equilibrium involving oil sand components (solids [S], bitumen [B] and aqueous phase [Aq]) is illustrated in Figure 2.3 and can be described by Young’s equation. ..[2.3] where θ is the three-phase contact angle measured through the bitumen phase (see Figure 2.3), is the interfacial tension between bitumen and the aqueous phase,  is the interfacial tension  is the interfacial tension between the solid and  between the solid and the aqueous phase, and bitumen.  As shown in Figure 2.3, it is beneficial to have a high contact angle of the bitumen droplet in order to initiate bitumen displacement and detachment from the solid surface. From Young’s equation (Equation 2.3), it can be seen that a high contact angle of bitumen on quartz is obtained when  is reduced and  is increased and/or  is reduced. At a molecular level, a high  contact angle of bitumen on a solid surface results from stronger cohesive forces within the bitumen and weaker adhesion of bitumen with the solid surface. An analysis of such interactions can be made by examining the work of adhesion (WA) and work of cohesion of the bitumen/solid/aqueous phase system shown in Figure 2.3.  Figure 2.3. Schematic illustration of the three-phase contact between oil sand components in slurry (solid, bitumen, and aqueous solution). The contact angle is measured through the bitumen phase. A Low contact angle, θ1 as a result of bitumen spreading at the solid–aqueous solution interface. B High contact angle, θ2 beneficial to bitumen liberation from the solid surface.  The work of adhesion, defined as the work required to separate a column of bitumen from a column of a solid with a unit cross sectional area, is given by the Dupré equation (Equation 2.4)  16  ..[2.4] while the work of cohesion is defined as the work required to break apart a column of bitumen with a unit cross sectional area. Fowkes (1967) proposed Equation 2.5 considering that the intermolecular interactions at interfaces resulted from different phenomena such as dispersion forces (represented by superscript d in Equation 2.5), hydrogen bonding (h), polar (dipole) interactions (p), pi-electron bonding (  and  electrostatic interactions (e). ..[2.5] Combining Equations 2.3 and 2.4, and rearranging leads to: ..[2.6] Equation 2.5 shows the important role that different interactions play on the adhesion of bitumen with a solid surface while Equation 2.6 shows that a high bitumen contact angle is obtained when the work of adhesion of bitumen with the solid surface is weakened and this can be achieved by adjusting any of the components in Equation 2.5 as discussed in subsequent paragraphs. Generally, the interfacial properties of the system are influenced by natural surfactants present in the ore, solution pH, and the composition of the aqueous phase. In commercial operations, the parameter that is often controlled is the  (the solid–aqueous solution interfacial tension). This is  achieved by increasing the pH to alkaline conditions which results in the decrease of  as a result  of quartz hydrolysis and increase in its surface charge. Also, increasing the pH triggers the migration of natural surfactants to the bitumen–aqueous solution interface, resulting in a decrease in the  .  Thus, an increase in pH results in an increase in the bitumen contact angle which facilitates bitumen liberation from quartz (Basu et al. 1996, 1998). In addition, interfacial properties can be controlled by adsorption of an organic surfactant (as discussed in section 2.1.2.2) or a polymer. Adsorption of a surfactant at the bitumen–aqueous solution interface would facilitate bitumen liberation while the effect of an adsorbing polymer on bitumen liberation is not so easy to predict. 17  There are hardly any literature studies on the effect of polymers on bitumen displacement (liberation) from a solid surface. The study by Wang et al. (2011) in which ethyl cellulose, a low molecular weight linear polymer was shown to modify the wettability of silanized hydrophobic silica from oil–wet to water–wet, although studied in the context of heavy oil processing, demonstrated the applicability of polymers in modifying the wettability properties of oil–solid–aqueous solution systems. Majority of studies on removal of oil or bitumen from a solid surface have been carried out in an aqueous surfactant solution environment. However, only a few of them were carried out in the context of oil sands processing (Basu et al., 1998; Wang et al., 2012, Wang et al., 2010) while most of them were conducted in the context of degreasing and solid surface cleaning operations (see for example Rowe et al., 2002, Starkweather et al., 1999, Starkweather et al., 2000, Carroll, 1996). It is generally accepted that at high solution pH and concentration, surfactants facilitate the removal of oil droplets from metal surfaces by reducing the aqueous solution–oil interfacial tension. Basu et al. (1998) used dynamic and static contact angles to study the effect of the methyl– isobutyl carbinol (MIBC)/kerosene surfactant system on bitumen displacement by water from a glass surface conditioned in surfactant solutions of different temperature and pH. At 40 °C and pH 8.2, the dynamic and static contact angles of bitumen (measured through the bitumen droplet) did not significantly change in the presence of 8000 mg/L NaCl. However, at a NaCl concentration of 16000 mg/L, the dynamic and static contact angles decreased. The static contact angle decreased from ~140° to ~105°. Liu et al. (2003) showed that increasing salt concentration reduces the repulsive forces between bitumen and silica leading to stronger adhesion of bitumen to the silica surface, and resulting in a low bitumen contact angle, which correlates with the result by Basu et al. The addition of a mixture of MIBC and kerosene (150 mg/L MIBC/300 mg/L kerosene) increased the dynamic contact angle. The static contact angle also increased from 105°  to 135°, slightly less than the original value without salt. The restoration of the static contact angle to a value close to that of the case with no salt was attributed to the adsorption of the surfactant system at the bitumen–water and glass–water interfaces, thereby preventing the interaction of NaCl with bitumen and glass. It should be noted that the very high NaCl content (equivalent to ~6300 mg/L Na+) under which these tests were conducted is inconsistent with typical NaCl concentrations of ~1500 mg/L (~ 600 mg/L Na+) found in oil sands industrial process water (Zhao et al., 2009). Basu et al. (1996) studied coker feed bitumen displacement from a model glass plate and showed that a higher static contact angle was 18  obtained at pH 11, while a higher bitumen displacement rate but lower static contact angle was obtained at pH 3. A higher bitumen recession rate in the presence of n-butylamine, a short chain amine at pH 8.5 and 70 °C was also obtained by Wang et al. (2010). The amine addition led to an increase in the bitumen displacement rate constant from 0.04 to 0.15 s-1. The beneficial role of the amine was attributed to its adsorption at the bitumen–water interface thereby reducing  . There was only a  marginal change in the steady-state contact angle of bitumen on the glass measured after ~200s. The contact angle (measured through the bitumen phase) increased from 122° to 130°. A more beneficial role can however be seen if the data at shorter times are considered. For example, after 25 seconds, the contact angle increased from 100° to 130°. In a recent study, Wang et al. (2012), reported that kerosene or fatty acid methyl ester enhanced the liberation of a toluene–extracted bitumen film from a glass surface although high concentrations of the diluents (about 20 wt. %) were needed to achieve high bitumen liberation. The decrease in the viscosity of the bitumen by the diluents (through the breakdown of the asphaltene–asphletene and asphaltene–maltene aggregates) was thought to be responsible for improved bitumen liberation. Starkweather et al. (1999) found that triton X–100 enhanced the contact angle of a hydrocarbon on a stainless steel metal surface at high pH and surfactant concentration. At these conditions, the aqueous solution–oil interfacial tension decreased, and the high contact angle of the oil droplet facilitated the droplet detachment from the surface. Rowe et al. (2002) extended the study by Starkweather et al. by studying the effect of surfactants on the solid–aqueous solution interface in addition to the aqueous solution–oil interface. These authors studied the variation of oil droplet shape (through contact angle measurements) and the time required for droplet detachment from a stainless steel surface contained in anionic and cationic organic surfactants. At high pH, faster kinetics of oil droplet detachment were observed when the aqueous solution–oil interface was negatively charged by adsorption of an anionic surfactant. The detachment was driven by repulsive forces between the similarly negatively charged solid surface and oil–aqueous surface. At low pH, faster kinetics were observed when the solid–aqueous solution was positively charged by bilayer adsorption of a positively charged surfactant on the solid surface. The surfactant bilayer was postulated to consist of oppositely oriented two layers of the surfactant. In the first layer, surfactant head groups (hydrophilic) were directed towards the surface and the tail groups (hydrophobic) 19  pointed to the aqueous solution. In the second layer, headgroups presented a hydrophilic surface to the solution and the hydrocarbon tail groups were directed towards the tail groups of the first layer. Rowe’s study demonstrates a beneficial effect of surfactant (or polymer) adsorption on the solid– aqueous solution interface in facilitating oil (or bitumen) displacement from a solid surface. It could be argued that the above studies would apply to the oil sand system (bitumen detachment from quartz or clays) since bitumen and the tested oil droplets are similarly charged as well as the solid surfaces (quartz, glass, or stainless steel surfaces). It needs to be mentioned that the oil sand system is complicated by the presence of natural surfactants and it is not clear how these would interact with the added surfactants. In addition, the majority of the research cited above was performed using model solid surfaces and processed bitumen (coker feed) or solvent extracted bitumen which do not necessarily have the same properties as real oil sand systems, thus the presented results, though useful in understanding or explaining possible physicochemical behaviour in oil sands processing, should be verified with tests using actual oil sands components such as solids extracted from oil sand ores or natural bitumen. This was the motivation for Srivanasa et al. (2012) who recently used an online visualization method to study bitumen liberation from sand grains in a real (good processing) oil sand ore. They essentially reached the same conclusions as those generally drawn by researchers using model solids and processed bitumen. A high solution temperature and high pH led to a faster bitumen liberation rate and a higher degree of bitumen liberation, weathering of a good processing ore lowered the degree of bitumen liberation and bitumen liberation kinetics, but the addition of kerosene (~50 mg/g based on the weight of the ore) to the weathered ore significantly improved the bitumen liberation. Finally, the presence of a higher amount of fines (12.6 wt. % versus 0.1 wt. %) and high salt concentration (16000 mg/L NaCl) were detrimental to bitumen liberation. A similar image analysis technique was employed by Wallwork et al. (2004) and Luthra et al. (2004). It should be noted that applicability of image analysis to oil sand ores still needs to be verified for high solids content (concentrated) slurries such as those encountered in the hydrotransport pipeline or primary separation cells. Such slurries do not provide a clear contrast between bitumen and solid particles which makes it extremely difficult to quantify bitumen liberation. From the foregoing review, it is evident that the whole bitumen liberation process depends on the interactions between bitumen and the mineral solid surface. Strong adhesion between bitumen and 20  quartz prevents bitumen recession from quartz. In addition, interactions between the liberated bitumen and free quartz or clay particles is detrimental to the bitumen froth quality. Therefore, the surface electrical properties of the individual oil sand components become important and are worth reviewing. 2.1.3.2 Surface electrical properties of quartz and clays The ionizable surface-group model was used to describe the surface properties of quartz obtained from oil sand ores (Takamura and Wallace, 1988). Quartz is the predominant inorganic mineral contained in the Alberta oil sands. It carries a negative charge above pH 2 resulting from the pH dependent ionization of surface silanol groups as shown by the following reactions. ..[2.7] ..[2.8 The clay mineralogy in oil sand ores varies but several researchers identified kaolinite and illite as the primary clay minerals (Omotoso and Mikula, 2004; Bayliss and Levinson, 1976; Ignasiak et al., 1985). Kaolinite is the predominant clay mineral in oil sand ores. The anisotropy of the kaolinite crystal structure is well known due to the existence of the edge and face surfaces. It is traditionally believed that the edges and basal planes (faces) of kaolinite particles show different electrical properties. Due to the protonation/deprotonation of surface hydroxyl ions, the edges show a net positive charge at low pH and negative charge at high pH with a point of zero charge (pzc, defined as the pH at which net surface charge is zero) in the range 4 to 8 (Olphen, 1963; Rand and Melton, 1977; Taubaso et al, 2004; Tombacz and Szekeres, 2006, Williams and Williams, 1978). It has long been generally accepted that the silica tetrahedral and alumina octahedral faces carry a permanent negative charge throughout the pH ranges due to the isomorphous substitution of silicon atoms by aluminum atoms in the tetrahedral crystal lattice or aluminium atoms by iron or magnesium atoms in the alumina octahedral layer (Williams and Williams, 1978). However, recently, Gupta and Miller (2010) showed that the two kaolinite faces carry different charges depending on the solution pH. Gupta and Miller designed an experiment in which interaction forces between a silicon nitride tip and the exposed silica and alumina faces were measured using AFM. They found that the silica tetrahedral face was negatively charged at pH > 4 while the alumina octahedral face was positively 21  charged at pH < 6 and negatively charged at pH > 8. The complex electric properties of the basal planes and edges of kaolinite contribute to the formation of the card–house structure and the gelation of clays, which are important practical issues in many industrial applications (van Olphen, 1963). Illite is a 2:1 layer, non-expanding hydrous aluminium silicate mineral. The 2:1 layer contains an octahedral sheet enclosed by two silica tetrahedral sheets, with potassium (K) as the major interlayer cation. Potassium balances the negative layer charge generated by the substitution of Al3+ for Si4+ in the tetrahedral sheet and (Mg2+, Fe2+) for Al in the octahedral sheet (Srodon, 2003). Illite differs from other mica minerals in having less substitution for Al3+ or Si4+, in containing more water and in having less potassium adsorbed in the interface position. The cation deficiency in illite results in weaker interlayer forces, as the intensity of the negative interlayer forces is proportional to the amount of cation deficiency resulting from isomorphous substitution. The weak interlayer negative charge present on the illite surface can be increased by anion adsorption as a result of electrostatic attraction of negatively charged anions to positively charged cations exposed at the broken edges (Miller, 1960). There is a wide range of zeta potentials reported for illite in literature. For example, Ding et al. (2006) determined the zeta potential distributions of illite dispersed in solutions of varying pH by electrophoresis. At pH 4.9 and 8.5, measurements in 1 mM KCl solutions containing 24 mg/L Mg2+ ions at 25 °C indicated that illite was negatively charged. Stephan and Chase (2001) used electrophoretic measurements to determine the zeta potentials of pure illite dispersed either in distilled water or in 0.14 mol/L NaCl. The values ranged from about -15 mV to -40 mV in the pH range 3 to 11 with no apparent pzc over the pH range tested. Hussain et al. (1996) determined the pzc of illite to be 2.5 using electrophoresis while Taubaso et al. (2004) found it to be 6.4 by potentiometric titration. It should be mentioned that the illite samples tested by the authors came from different sources and thus contained different levels of contamination which could explain the disparity in the results. It is also apparent from the result by Taubaso et al. that the zeta potentials measured depend on the technique used. Studies reviewed in the above sections generally show that bitumen and mineral solids are negatively charged over most of the industrially relevant pH range. However, it is important to note that the electrokinetic properties of kaolinite presented by Gupta and Miller (2010) can lead to bitumen–mineral and/or mineral–mineral interactions in aqueous suspensions. It also needs to be mentioned that the magnitude of the surface charge on mineral solids is an indicator of interactions 22  between the particles, and zeta potential measurements are routinely used to assess the stability of suspensions. For example, Long et al. (2005) reported the surface potentials of mineral fines obtained from oil sands tailings. The authors used AFM to determine force–distance curves for fine tailing particles in industrial process water at pH 8.2. After fitting the curves with the classical DLVO theory, they obtained small surface potentials ranging from -5 to -7 mV. Such small surface potentials (at a high pH) can enhance particle aggregation as the repulsive forces diminish. The process water in which the measurements were made contained 47.0 mg/L Ca2+ and 15.0 mg/L Mg2+, and these inorganic divalent cations were most likely responsible for the small surface potentials by specifically adsorbing on the clay particles and making them less negatively charged In comparison, Liu et al. (2004a) measured zeta potentials of fine solids obtained from tailings of good and bad processing ores and reported a high value of -50 mV in a simple electrolyte without multivalent ions (1 mM KCl) at the same pH as in Long et al.’s study. The above two studies highlight the role that divalent cations play in the electrical properties of oil sand components and hence the aggregation/dispersion of oil sand slurries. It should also be noted that experimental evidence exists to suggest that non-DLVO forces can enhance aggregation in systems whose components all have high negative zeta potentials (Churaev and Derjaguin, 1985). In such systems, the total force balance according to the DLVO theory is expected to be repulsive. However, the non-DLVO forces dominate over the repulsive forces and cause aggregation of the particles. The attractive hydrophobic forces are an example of such forces. Xu and Yoon (1989) showed that attractive hydrophobic forces were responsible for the coagulation of coal particles in a wide range of pH even when the coal zeta potentials were highly negative. Since the magnitude of these non-DLVO hydrophobic forces is a function of the wettability of the surface, the balance of forces in oil sand systems containing bitumen is expected to be dominated by the attractive hydrophobic forces, more so for poor processing ores also containing hydrophobic solids. The following sections will review bitumen–mineral and mineral–mineral interactions in light of the preceding discussion. 2.1.3.3 Bitumen–mineral interactions Bitumen extraction from oil sand ores is typically carried out at pH 8.5. At such high pH, as reviewed in section 2.1.3.2, the bitumen and quartz surfaces should be highly negatively charged and are therefore separated by charge repulsion. This should prevent the wetting of the quartz surface by 23  bitumen and facilitate bitumen liberation. The bitumen–silica repulsive forces decrease with decreasing pH as the adhesion forces start becoming significant (Liu et al., 2003). Studies on model bitumen–silica systems (Zhou et al., 1999) showed that bitumen–silica coagulation occurred at pH values lower than 7 where both bitumen and silica are expected to carry a negative charge. Masliyah et al. (2004) measured interaction and adhesion forces between bitumen and silica using AFM and suggested that at pH values below 7, a combination of strong adhesion forces and weak repulsive forces between bitumen and silica are responsible for the coagulation observed by Zhou et al. (1999). The presence of multivalent metal ions is known to alter the surface properties of quartz. For example, the presence of calcium ions was determined to be responsible for coagulation of bitumen and silica at pH >10.5 as a result of chemisorption of the positively charged calcium monohydrate ions on silica (Zhou et al., 1999). Calcium present in process water acts as a binder between the negatively charged bitumen and quartz surfaces (Zhao et al., 2006). At higher calcium ion concentrations, charge reversal of quartz from negative to positive could further enhance bitumen– silica interactions. Takamura and Chow (1983) observed the same effect when they studied the displacement of bitumen from sand in the presence of calcium ions. They measured the adjoining pressure between bitumen and sand and found that in the presence of 0.01 mol/L CaCl2, an attractive disjoining pressure existed which reduced the receding velocity of bitumen from the sand grains. The fundamental studies of Liu et al. (2003) and Kasongo et al. (2000) using zeta potential measurements and force measurements by AFM generally showed that bitumen–quartz and bitumen–clay interactions in the presence of multivalent metal ions play a critical role in the colloidal behavior of oil sand slurries, but the studies did not indicate or predict any detrimental effects in the pH range ~8–9 in which bitumen extraction is performed. However, in the presence of montmorillonite clays, adhesion force distribution measurements on model clay systems by Liu et al. (2004b) showed that at pH 8.2, the presence of calcium lowered the repulsive forces between bitumen and montmorillonite clay. These results show that calcium can act as a bridge between the negatively charged bitumen and montmorillonite clay thereby facilitating undesired slime coating of bitumen by montmorillonite leading to poor bitumen recoveries. The same authors (Liu et al., 2004a) extended the bitumen–model clay interaction studies to systems involving oil sand-derived fines and demonstrated that hetero-coagulation was observed between bitumen and fines extracted from poor 24  processing ores while only weak coagulation was observed between bitumen and fines derived from good processing ores. This result partly explains the high recoveries obtained when processing good ores. The strong adhesion between bitumen and fines from poor processing ores was correlated to the hydrophobic nature of the fines and the high concentration of divalent cations in the process water, and these effects seem to be responsible for the observed low recoveries when processing poor processing ores. The hydrophobicity of fines from a poor processing ores was demonstrated by Dang-vu (2009a) through a series of independent tests based on contact angle, critical surface tension, hydrophilic/hydrophobic partitioning, and measurements of water droplet penetration time. Similar results were observed by Ren et al. (2009) who measured direct colloidal interaction forces between bitumen and fine solids extracted from a laboratory-weathered good processing ore. The dominant long-range repulsive forces measured between bitumen and fines from a good processing ore became attractive when the measurements were conducted with fines extracted from the weathered ore. Higher adhesion forces were also measured between bitumen and fines from the weathered ore. It was postulated that the reversal of the repulsive forces to attractive forces and increase in the adhesion forces was due to the change in the wettability of the solids from hydrophilic to hydrophobic. The hydrophobicity of the fines due to weathering is thought to result from the loss of formation water and adsorption of organic matter transferred from the bulk bitumen onto the solids (Liu et al., 2005a; Dang-vu et al, 2009b). The change in solids wettability creates an additional long-range attractive hydrophobic force between the hydrophobic bitumen and fines, and consequently explains the attractive force profiles and increased bitumen–fines adhesion at high pH (8.0–8.6) where only repulsive DLVO forces could be expected to exist. When the Stern potential values of silica particles were used to fit colloidal force profiles between bitumen and weathered and nonweathered model silica particles, the classic DLVO overestimated the interaction forces indicating the presence of an unaccounted attractive hydrophobic force. The force profiles were well fitted with the extended DLVO theory that includes extra hydrophobic forces in the total force balance. Also, significantly, a higher magnitude of the hydrophobic force (as seen from the hydrophobic force constant) was obtained for the bitumen–silica (weathered) force profiles than the bitumen–silica (nonweathered) force profiles This explains the higher adhesion forces measured between bitumen and fines extracted from a weathered good processing ore.  25  The hydrophobicity of bitumen is a function of pH. The contact angle of water on bitumen decreases as pH increases (Masliyah et al., 2011, Liu et al., 2005b), which diminishes the hydrophobic bitumen–bitumen forces at high pH. However, a sharp decrease is seen above pH 9, and in fact a close look at the data by Masliyah et al. and Liu et al. reveals that the contact angle of water on bitumen only changes by about 5° up to pH 8.5, the operating pH for the bitumen extraction process. The data by Liu et al. also show that the hydrophobic force constant is virtually the same up to pH 8.5. This observation indicates that bitumen is still very hydrophobic within the pH range for oil sand operations and considerable aggregation arising from hydrophobic attractive bitumen– bitumen, bitumen–solids can be expected. 2.1.3.4 Mineral–mineral interactions In addition to bitumen–mineral and bitumen–bitumen interactions in an oil sands slurry, mineral– mineral interactions should be considered. In order to control the aggregation/dispersion state of oil sand slurries, an understanding of the behaviour of mineral–mineral interactions is necessary. Quartz is the dominant mineral found in oil sands ores. The clay minerals kaolinite, K-feldspar, muscovite are the other notable minerals present in the ores. The clays tend to accumulate in the fines fraction; however, quartz still remains the dominant mineral in the fines fraction. In concentrated slurries such as those encountered in the pipeline hydrotransport of oil sand slurries (about 60% by weight), the interparticle spacing is sufficiently small that a high frequency of particle collisions can lead to particle-particle aggregation since in such high volume suspensions, the effect of nonhydrodynamic (interparticle) forces is quite significant (Russel, 1978). The balance between hydrodynamic and interparticle forces governs the aggregation/dispersion state of such concentrated suspensions. The solids in oil sand ores can range in size from 2.5 mm to submicron-size, thus the contents of coarse and fine particles determines the balance between hydrodynamic and interparticle forces and subsequently the aggregation/dispersion state of the ore slurries. Coarse particles interact only through hydrodynamic forces while fine particles can interact through intermolecular and surface forces. It is clear that in poor oil sand ores containing a high fraction of fine particles, the interparticle forces dominate over the hydrodynamic forces, and thus the slurry properties are governed by the balance between attractive (van der Waals and hydrophobic) and repulsive (electrostatic, hydration and steric) forces. The solids from poor oil sands are known to be hydrophobic as shown by Dang-vu (2009a). The initial water contact angles measured on a 26  compressed disc of fines from a poor processing ore were as high as ~120° compared to ~35° for fines from a good processing ore. For such highly hydrophobic solids, the attractive hydrophobic force dominates over the electrostatic repulsive forces (Yoon et al, 1997) as demonstrated experimentally by Xu and Yoon (1989) who showed that fresh hydrophobic coal and methylated silica (hydrophobic silica) coagulated at high pH where electrostatic repulsive forces were expected to dominate due to the high zeta potential. Skvarla and Kmet (1991) demonstrated similar nonDLVO induced aggregation for fine magnetite and dolomite particles hydrophobicized by sodium oleate. In addition, the properties of oil sand slurries are governed by the chemistry of the process water. The most important ions are divalent cations, Ca2+ and Mg2+ which are known to reduce the magnitude of the zeta potential of quartz and could thus promote particle aggregation through weak van der Waals forces. For example, the adsorption of Ca2+ from a 10mM CaCl2 solution onto negatively charged silica through electrostatic attraction makes the silica surface less negatively charged at pH 8 than at pH 6 (Masliyah et al., 2011). At pH above 8, calcium ions are converted to calcium monohydroxy ions (CaOH+). The specific adsorption of CaOH+ onto the already hydrolyzed silica is more effective in reducing the zeta potential of silica by forming covalent SiOCa links and making the silica surface positively charged, and data by Masliyah et al. (2011) indicate that a very rapid reduction in the zeta potential of silica occurs at pH above 8.5 at calcium ion concentrations as low as 1.0 mM. Under such conditions, the reduction of the zeta potential of quartz contained in oil sand slurries can be expected to cause quartz–quartz interactions through van der Waals forces. Surface hydration in the presence of different cations also plays a role in the interaction of quartz particles (Farrow et al., 1989). The adsorption of the poorly hydrated K+ and Cs+, both water structure breaking cations onto quartz reduces the surface hydration of quartz, an effect that reduces the repulsive hydration forces between quartz particles and allows the quartz particles to approach one another to within the range of attractive van der Waals forces (Farrow et al., 1989). In concentrated slurries where particles are already close enough, the reduction in the repulsive hydration forces results in weak particle–particle interactions and formation of a structure within the suspension. In the study by Farrow et al., quartz suspensions prepared with 0.1 mol/dm3 electrolyte solutions of the strongly hydrated K+ and Cs+ cations displayed a small yield stress and higher viscosity than those prepared with 0.1 mol/dm3 electrolyte solutions of the strongly hydrated Li+ and 27  Na+ cations. The major cation in oil sand process water is Na+ while K+ is present in low concentrations (Zhao et al., 2009), thus the role of hydration forces on quartz-quartz interactions would be negligible owing to the low affinity of the large, strongly hydrated sodium toward quartz. It should however be noted that the process water analyzed by Zhao et al. (2009) was obtained from the Aurora plant, Syncrude Canada Ltd. and so is presumably from a good processing ore. It is known that the high content of fines from poor processing ores contributes to a high ion concentration in the process water arising from the cation exchange properties of clays, and as shown by connate water analysis of two ores of different grades, the poor processing oil sand connate water contained more K+ than the good processing oil sand connate water (Zhao et al., 2009). Also, soluble ion concentration data for the 2003-2004 Imperial Oil Ltd. drill core hole program showed a high K+ concentration (almost similar to the Na+ concentration) for samples with an average bitumen grade of 12 wt.% (Imperial Oil Ltd., 2005). Thus, the hydration effects of K+ on quartz particles in oil sand slurries cannot be entirely dismissed. It was suggested that a small fraction of fine and ultrafine (commonly classified as -0.3 µm particles) clay minerals has a major influence on the colloidal state of oil sand slurries, and that the unique crystal structures and distinctive surface properties are responsible for the gelation encountered in the primary separation cell during extraction of bitumen from oil sands (Tu et al., 2005; Mercier et al., 2012). The gelation of clays was linked to the “house-of-cardsˮ arrangement of the particles, or a threedimensional, aggregated structure arising from edge–edge or edge–face associations (Olphen, 1977) but the exact mechanism of gelation is still unclear. The edge–edge interactions are expected at the iso-electric point of the clay edges while edge–face interactions can occur at pH values less than the iso-electric point of the edges. In addition, face–face interactions become significant in highly concentrated slurries. The present state of understanding of clay particle interactions is complicated by recent data by Gupta and Miller (2010) that show that kaolinite silica and alumina faces, just like the edges, carry different charges depending on pH. The silica face was shown to have an isoelectric point at pH 4 while the alumina face had an isoelectric point between pH 6 and 8. In concentrated suspensions, depending on the edge isoelectric point, face–face interactions that often lead to stacking of kaolin particles could significantly enhance edge–silica face and edge–alumina face interactions. 28  A number of studies show that kaolinite particle–particle interactions are possible close to the pH at which oil sand operations are carried out. Rand and Melton (1977) investigated the effect of electrolyte concentration and pH on flow properties (Bingham yield stress and plastic viscosity) of homoionic sodium kaolinite suspensions and determined conditions under which edge–edge, edge– face and face–face coagulated structures existed. They determined the edge surface isoelectric point to be at a pH value of 7.3 since at this value, the yield stress was invariant with ionic strength. Their rheological data showed that edge–edge interactions resulting from van der Waals forces were preferred at the edge isoelectric point. At pH values above the edge isoelectric point, addition of a small amount of NaCl promoted edge–edge structures and the yield stress increased until it attained the value at the edge isoelectric point. For example, at pH 8.5 and in distilled water, a small yield stress was measured owing to electrostatic repulsive forces between the edges and faces of the kaolinite particles. However, addition of about 0.04 mol/L NaCl increased the yield stress to that measured at the edge isoelectric point. The rheological data by Williams and Williams (1982) and James and Williams (1982) were in contrast to those of Rand and Melton (1977). They performed their rheological experiments in the pH range between 6 and 8 to embrace the edge isoelectric point determined by Rand and Melton. Their data demonstrated a decline of the yield stress with salt concentration at pH values near or at the zero point of charge of the edge, suggesting edge–face interactions in the pH range 6–8, in addition to the edge–edge interactions at the edge isoelectric point. Chow (1991) investigated the stability of kaolinite dispersions using a light scattering instrument that monitored fluctuations in the transmitted light intensity of the kaolinite dispersion. The measured stability was compared with the stability calculated by the DLVO theory based on zeta potentials of the kaolinite particle edges predicted by the Ionizable Surface Group model (Healy and White, 1978). At a NaCl concentration of 0.001 mol/L, there was an agreement between the measured and calculated stabilities but only at pH >8. At pH <8, light scattering showed that the dispersion coagulated while the DLVO calculation predicted a stable system. At a NaCl concentration of 0.01 mol/L, the disagreement between the two techniques was noted at pH >8 with the light scattering technique showing a weakly coagulated dispersion between pH 8.6 and 9.1 while the DLVO calculation predicted a stable dispersion at pH 8.2. The disagreement between the two techniques was attributed to the well-known inability of the electrophoresis method to differentiate between the origin of charge on the faces and edges of the kaolinite particles. Thus, the calculations using the DLVO model did not predict the different modes 29  of interaction occurring between the edges and faces of the kaolinite particles. The most important observation from this work was the fact that weak coagulation was observed at pH 8.6 – 9.1 where the kaolinite particles should repel each other due to the highly negatively charged surfaces. Zbik and Frost (2009) also observed non-DLVO edge–edge interactions in aqueous suspensions of Birdwood kaolinite from South Australia. Electron microscopy micrographs showed edge–edge orientations that resulted in a honeycomb like structure containing discrete particles linked together to form a coagulated network. High-resolution micrographs of the same structure showed a stair step connection between the aggregated particles forming a long string like aggregate. It was proposed that the non-DLVO edge-edge interactions were likely due to hydrophobic interactions between aerated edges which rolled on top of each other to build the long strings observed in the highresolution micrographs. The gaseous micro bubbles attached to the edges were thought to originate from hydrocarbon contamination at the kaolinite interface. Such contamination results in wettability alteration of kaolinite particles from hydrophilic to hydrophobic as recently demonstrated by Lebedeva and Fogden (2011). In another study, Zbik et al. (2009) showed that air micro bubbles attached to kaolinite particles led to particle aggregation. The role of air bubbles in particle aggregation has been shown elsewhere. Experimental evidence from Zhou et al. (1996) showing that air bubbles attached to hydrophobic coal particles increase the viscosity and yield stress of concentrated coal–water slurries (i.e., enhance aggregation of coal particles) supports the postulation by Zbik et al. (2009). Recent studies by Yin and Miller (2012) show that the silica tetrahedral face of kaolinite displays a modest level of hydrophobicity as determined from direct colloidal force measurements owing to the lack of hydrogen bonding sites while the alumina octahedral face shows no hydrophobicity due to the surface hydroxyl groups at the alumina face that strongly interact with water molecules. Earlier studies by Saada et al. (1995) and Kaminsky et al. (2009) provided indirect evidence of the hydrophobicity of kaolinite particles. In the study by Saada et al., the hydrophobicity of the kaolinite was shown by its high affinity to asphaletenes and a lower affinity to water while Kaminsky et al. (2009) showed through clay mineralogical analysis of bitumen froth and tailings that kaolinite tended to report in larger amounts to the bitumen froth compared to other clay minerals, an indirect evidence of possible interactions between hydrophobic sites on kaolinite and bitumen. If indeed the  30  silica tetrahedral faces of kaolinite display some level of hydrophobicity, silica face–silica face associations can be expected to occur through hydrophobic interactions. The foregoing studies show that in addition to edge–edge, edge–face and face–face interactions arising from DLVO interactions, additional non-DLVO interactions such as hydrophobic forces can be expected in oil sand slurries. The aggregation of clay particles resulting from such interactions affect oil sand operations in two ways: (a) the aggregated structure in the primary separation cells hinders the bitumen rise velocity hence reducing bitumen recovery (Schramm, 1989; Tu et al., 2005; Mercier et al. 2012), and (b) the structure inhibits the settling of coarser solids in the middlings zone of the PSC and keeps them in suspension. Also, organic matter adsorbs on clay fines to produce biwetted fines (Kotlyar et al., 1995; Majid and Sparks, 1996) that are strongly attracted to the bitumen–water interface resulting in undesired slime coating of bitumen usually associated with poor froth quality. Tu et al. (2005) and Kotlyar et al. (1996) demonstrated that ultrafines (<0.3 µm) are especially detrimental to the slurry properties in the oil sands extraction. In the presence of a critical cation concentration, a small amount of ultrafines (between 1.5 and 3.0 w/w %) was required to form a thixotropic structure. At concentrations up to the critical ultrafine concentration (CUC) for the onset of the structure, Kotlyar et al. (1996) reported that the slurry viscosity increased progressively, thus even concentrations less than the CUC could be detrimental to the flotation of bitumen. For coarser clays (~2–3 µm), Tu et al. (2005) showed that concentrations higher than 10 wt. % are needed to facilitate the structure formation. This indicates that most poor processing ores are highly susceptible to gelation as they usually contain high concentrations of ultrafine clays. For example, Gutierrez and Pawlik (2012a) recently tested four poor oil sands ores with contents of the size fractions finer than 3 µm all greater than 10 wt. % and found that the slurry viscosity increased with the fines content of the ores primarily due to the increased interparticle aggregation. The trend was the same irrespective of the pH (8.5 or 10) and temperature (25 °C or 50 °C). This result clearly points  towards  the  significance  of  the  finest  aggregation/dispersion state of oil sand slurries.  31  mineral  fractions  in  controlling  the  2.1.3.5 Implications of bitumen–mineral and mineral–mineral interactions on oil sand slurry aggregation/dispersion In good processing oil sand ores characterized by high bitumen and low fines content, interactions between the ore components at the operating pH do not lead to appreciable aggregation of the slurry, at least in the PSC where typical solids contents of ~35% by weight are maintained. However, some particle aggregation could result in the pipeline hydrotransport at high solids loading. For poor processing ores, it becomes clear that long-range hydrophobic forces, much stronger than the van der Waals forces are responsible for the increased adhesion forces measured between bitumen and solids even when the surfaces of bitumen and the solids are both negatively charged. Thus, a certain level of aggregation resulting from bitumen–bitumen, bitumen–solids, and solids– solids hydrophobic interactions can be expected for poor oil sand ore slurries at the operating pH of bitumen extraction. In the presence of multivalent cations such as Ca2+ and Mg2+, the reduction of the zeta potentials of bitumen and quartz could promote aggregation in the slurry. As reviewed by Mercier et al. (2012), critical concentrations of fines and cations present in process water are capable of forming aggregated structures that are detrimental to bitumen recovery as a result of increased slurry viscosity in the pipeline hydrotransport and in the PSC. Thus, the effective volume fraction of fine solids and the interaction energy between particles are the critical factors that control the aggregation/dispersion state of oil sand ores (Masliyah et al., 2011). As shown in Figure 1.1 (Chapter 1), the slurry properties desired in the extraction stage are well-dispersed fines and low viscosity in order to achieve separation of bitumen and solids and efficient aeration of the slurry. To achieve this, current strategies used to control the aggregation state of oil sand slurries are to operate at low solids loading (lowering the fines content) and/or increasing the dosage of sodium hydroxide to promote stronger repulsions between the solids (Masliyah et al., 2011). Other inorganic dispersants such as sodium hexa-meta phosphate or sodium silicate can be used to obtain similar results. As discussed in section 1.1 (Chapter 1), these traditional dispersants are ineffective for poor processing ores except at high dosages. A number of other inorganic process aids (and some polymers) have been researched as reviewed in the following section.  32  2.1.3.6 Role of chemical additives on bitumen–mineral and mineral–mineral interactions As discussed above, interactions between oil sand components affect bitumen recoveries and in most ores, chemical additives are needed to enhance bitumen extractability. Long et al. (2007) reviewed the role of chemical additives on bitumen recovery. A summary of that review and recently researched additives will be presented in the following paragraphs. The earliest known additive used in bitumen extraction was caustic soda (sodium hydroxide) and its role was to adjust the slurry pH and promote the generation of natural surfactants. High recoveries were easily obtained provided a high temperature was maintained. To operate at a low temperature, the increase in bitumen viscosity was responsible for low recoveries even with the use of sodium hydroxide. It was shown that the use of sodium hydroxide in combination with bitumen viscosity diluents, primarily organic solvents such as kerosene, increased bitumen recoveries to acceptable levels, particularly when the bitumen viscosity was reduced to less than ~3 Pa·s (Hupka et al., 1983, 1987). For poor or average oil sand ores in which bitumen–mineral and mineral–mineral interactions affect ore processability, the addition of a diluent increases bitumen recovery but not to acceptable levels as shown by Schramm et al., 2003. In the presence of 0.06% NaOH only, a bitumen recovery of 8% was obtained in a batch extraction experiment at 25 °C. With the addition of a diluent (20000 mg/L kerosene), the recovery improved to 79% and with further addition of 1000mg/L MIBC, the recovery reached 98%. Long et al. (2005) measured colloidal forces between bitumen and fines and attributed the beneficial role of methyl-isobutyl carbinol (MIBC) to its ability to prevent bitumen– fines heterocoagulation by changing the long-range interactions between the two from attractive to repulsive, thereby preventing coating of fines onto bitumen that would otherwise retard bitumen–air attachment. However, no exact mechanism of the action of MIBC was suggested. It is also important to note that in the study by Schramm et al. (2003), higher dosages of MIBC were detrimental to bitumen recovery. Long et al.’s study also showed that at a concentration of 5000 mg/L, MIBC promoted adhesion forces between bitumen and fines which would lead to low bitumen recoveries. These results highlight the disadvantage of using inorganic additives and surfactants in processing of poor processing ores as the high dosages required would ultimately have negative consequences on the whole process. Thus, the usual strategy is to operate at low solids loading.  33  Zhao et al. (2009) showed that the detrimental effect of Ca2+ on bitumen recovery could be minimized by using bicarbonate (HCO3-) ions to precipitate Ca2+ from process water. At the same time, HCO3- ions apparently dispersed the fine solids and decreased solid–bitumen adhesion. Flotation tests conducted on a poor processing ore in the presence of HCO3- ions showed a marginal improvement in the bitumen flotation rate and bitumen froth quality but there was no improvement in the overall flotation recovery. Li et al. (2005) showed a similar beneficial effect of an inorganic process aid (acidified sodium silicate). The authors demonstrated through flotation and zeta potential measurements that the use of sodium silicate as a process aid instead of the traditional sodium hydroxide resulted in higher bitumen recovery and froth quality. They attributed the advantageous effect of acidified sodium silicate to its low impact on solution pH, effective precipitation of Ca2+ and Mg2+ from process water and dispersion of clays thereby minimizing bitumen–clay coagulation. It should be noted that high chemical dosages were needed to achieve the results obtained by Zhao et al. (2009) and Li et al. (2005). Zhao et al. needed 12 mM of NaHCO3 (about 730 mg/L HCO3-) while Li et al. determined an optimum dosage of 10 mM acidified sodium silicate (about 3600 g/t) from chemical dosage experiments. Such high dosages lead to accumulation of ions in process water and to a gradual increase in the ionic strength of process water. Since higher ionic strengths promote extensive aggregation between all the ore components, it is not clear whether the beneficial effects reported would be sustained under such conditions. In the study by Li et al. (2008) the use of a hydrolyzed polyacrylamide (HPAM) as a process aid on a poor processing ore led to a deteriorated bitumen froth quality, while an attempt to use a hybrid Al(OH)3-polyacrylamide (Al-PAM) improved the froth quality, but at the cost of bitumen recovery. The authors then used a combination of Al-PAM and HPAM and achieved a simultaneous improvement in both bitumen recovery and froth quality. The role of the dual polymer system was to reduce the bitumen–clay and bitumen–bitumen adhesion and increase solid–solid adhesion forces. Reducing the bitumen–clay and bitumen–bitumen adhesion through addition of Al-PAM resulted in higher bitumen recovery and low bitumen–bitumen coalescence, respectively, while higher solid– solid adhesion forces enhanced tailings settling due to the flocculating abilities of HPAM. At the recommended dual polymer dosage (5 mg/L Al-PAM + 5 mg/L HPAM), it needs to be mentioned that although the bitumen recovery increased from 50% to 78%, there was only a marginal 34  improvement in the bitumen/solid ratio indicating that the froth quality was not affected much by the dual polymer system. In fact, the data show that keeping the Al-PAM dosage constant and increasing the HPAM dosage deteriorated the froth quality. In addition, increasing the solid–solid adhesion forces does not favor the bitumen extraction stage. In terms of the aggregation/dispersion state of the oil sand slurries, increasing solid–solid adhesion forces increases the viscosity and yield stress of the slurries, which is undesirable for the slurry transport and processing. In summary, it seems that the principal role of process aids in oil sands processing is to reduce the bitumen viscosity and/or promote bitumen liberation from solids by adsorbing on solid (and possibly bitumen) surfaces and keeping them apart so as to avoid the detrimental effects of slime coatings, bitumen–solid heterocoagulation, and solid–solid aggregation. The process aid utilized should not depress the floatability of bitumen, thus it should not alter the bitumen–air interface (discussed in section 2.1.4 below). A convenient way to assess the floatability of bitumen is to determine its surface hydrophobicity through contact angle measurements of an air bubble attached to a bitumen surface immersed in an aqueous solution containing the process aid. Such a determination closely mimics the bitumen flotation process in a primary separation cell. A review of bitumen wettability and determination of the bitumen contact angle is given in section 2.1.4. As was mentioned in section 1.1, organic polymeric dispersants can potentially be utilized as process aids for tuning the aggregation/dispersion characteristics of oil sand slurries. The key potential problem with the use of polymeric dispersants for oil sand slurries is that such an additive may also depress the floatability of bitumen. Thus the effect of the dispersants on the surface properties and hence wettability of bitumen needs to be known. Contact angle measurements provide information about the surface hydrophobicity of bitumen, which is an indication of its floatability. 2.1.4  Bitumen wettability  The hot water extraction process is primarily based on the difference between the wettability of bitumen and the wettability of the inorganic minerals. The term wettability refers to the interaction of solids with water. In cases where the solid has a strong affinity for water, water tends to spread on or wet the solid surface. If the solid has a weak affinity for water, the water will tend to retract from such a surface. The inorganic mineral fraction found in oil sands is an example of the first case and is therefore hydrophilic or water-attracting, while bitumen is an example of the second case and is 35  therefore hydrophobic or water-repelling. It should be noted that some clays in poor processing oil sands have been found to be hydrophobic (Liu et al., 2004a) which would also explain poor recoveries of bitumen that are often obtained when processing these types of ores. The contact angle is used as a convenient measure of wettability. Theoretically, a contact angle of water of 0° indicates perfect wetting, a condition of complete hydrophilicity. On the other hand, an angle of 180° indicates perfect dewetting, a condition of complete hydrophobicity. Most systems exhibit contact angles that lie between the two extremes. Once bitumen is liberated from a solid surface, its attachment to air bubbles requires the bitumen and air surfaces to be hydrophobic. If a fine layer of hydrophilic clays coats the bitumen, it becomes difficult for bitumen to attach to air bubbles. The surface properties of bitumen and air, which are dependent on the pH, ionic strength, temperature and presence of polymers or surfactants, then become important. Even if clay-coated bitumen becomes attached to air bubbles and floats to the surface, the result is a bitumen froth contaminated with solids which have to be removed before upgrading the bitumen. The situation opposite to that depicted in Figure. 2.3 (section 2.1.3.1) is desirable for strong bitumen–air attachment (Figure 2.4 A and B at lower temperatures) and bitumen–air bubble engulfment (Figure 2.4 C at higher temperatures).  Figure 2.4 Schematic illustration of the three-phase contact between bitumen, air, and aqueous phase in the captive bubble technique analogous to bitumen attachment to air during bitumen flotation. The contact angle is measured through the aqueous phase. A Low contact angle, θ1 between the air bubble and bitumen immediately after air bubble–bitumen contact. B High contact angle, θ2 showing spreading of the air bubble on the bitumen surface. C Bitumen – air bubble engulfment.  36  The foregoing information shows that contact angle measurements provide a means of determining the wettability of bitumen which should be a good indication of the hydrophobicity and hence the floatability/extractability of bitumen without interference from solids. Such measurements can be performed in the presence of polymeric dispersants to understand their effect on bitumen hydrophobicity. Many techniques have been developed to measure the contact angles of liquids on solids. These include the captive bubble and sessile drop, axisymmetric drop shape analysis profile, Wilhelmy plate, Washburn and microscopy methods. Chau (2009) provided a comprehensive review of these methods. Drelich et al. (1994) measured contact angles of bitumen from five Utah oil sands using the sessile drop technique. The contact angles were determined by placing water droplets on slides covered with bitumen films, and were found to vary from 94o to 98o with an exception of 85o for one ore. These values are consistent with measurements by other researchers: Dang-Vu et al. (2009b) reported values ranging from 95o to 99o while Vargha-Butler et al. (1988) reported contact angles ranging from 93o to 96o. Drelich et al. (1994) noted that when contact angles were used to calculate surface tension, the measured surface tension values could not be compared with those determined directly. The inaccuracies in the contact angle measurements were attributed to possible interactions between molecules of water and bitumen during the measurements. As well, factors such as solubility of some bitumen compounds in water and/or water in bitumen, deformation of the bitumen film at the three-phase contact line and spreading of the bitumen at the water surface could affect the contact angle measurements. However, in the study by Vargha-Butler et al. (1988), no such problems were encountered. These researchers also used the sessile drop method to determine the contact angles of water and glycerol on semi-solid bitumen films extracted from Canadian oil sands. They were able to provide a good comparison of surface tension values determined directly and those calculated from contact angle measurements, indicating that the contact angles were accurately measured. However, the authors reported some key observations: the contact angles measured depended on the solvent used for the bitumen extraction, and the presence of clays in bitumen can affect the composition of the tested bitumen surface and hence the contact angle. This was especially the case for low grade ores known to contain a high amount of fine solids and clays. The effects described by  37  Drelich et al. (1994) present difficulties in defining the thermodynamic equilibrium for contact angle measurements of water droplets on a bitumen film. It should be noted that different procedures and techniques produce different values of measured contact angles. For example, Kasongo et al. (2000) reported a much smaller bitumen contact angle of 60° using the captive bubble method. A note on their measurements is that they were made in an aqueous suspension containing 0.5 wt. % clay (compared to distilled water) and for a much longer time than the measurements discussed above. The low bitumen contact angle measured by Kasongo et al. was attributed to the use of the captive bubble method which is more or less a pseudo-water receding process. Gutierrez-Rodriguez et al. (1984) evaluated the hydrophobicity of coal by sessile drop and captive bubble methods and proposed that the difference between the larger water sessile drop contact angle and the smaller captive bubble contact angle was due to the differences in the wettability of different components of the coal surface. They postulated that the coal surface has three types of sites: strongly hydrophobic, weakly hydrophobic, and hydrophilic. The sessile drop method measures the total hydrophobic sites whereas the captive bubble method measures only the strongly hydrophobic sites leaving out the weakly hydrophobic sites due to their interaction with water. This claim agrees well with the observation made by Drelich et al. (1994) for contact angle measurements of water on bitumen films. The captive bubble method was found to be less sensitive to imperfections (heterogeneity, presence of impurities) of coal surfaces (Drelich et al., 2000). In addition, the effect of relative humidity on the measured contact angle is essentially eliminated in the captive-bubble technique. As such, the captive bubble method is very convenient and modern instrumentation enables high accuracy to be achieved . The captive bubble method can also be considered as more representative of the state of the bitumen surface under extraction conditions since initially all slurry components are immersed in water and then attachment of air bubbles takes place. The same procedure is basically employed in the captive bubble technique: the tested specimen is first immersed in the tested solution, and an air bubble is attached from below to the surface, as shown in Figure 2.4. 2.2 Organic polymeric dispersants: carboxymethyl cellulose Organic additives such as starch, modified celluloses, dextrins have been extensively used in mineral processing as dispersants and depressants primarily due to their better performance at a 38  lower dosage (Pugh, 1989), low cost per tonne of ore treated and their biodegradability (Nagaraj, 2000) in comparison to the traditionally used inorganic reagents such as sodium sulphide, sodium silicate and cyanide. Their application in oil sands processing, to the author’s knowledge, is not documented. In this section, a review of organic polymeric dispersants will be given. Specific emphasis will be placed on carboxymethyl cellulose, a biodegradable polymer that was tested in this dissertation. A comprehensive review of slurry modifiers and their use in the mineral industry was provided by Klein and Pawlik (2005). In the review, a classification of polymeric modifiers (Table 2.1) based on their molecular weight was given. A distinction should be made between dispersants and flocculants/coagulants. Dispersants decrease the viscosity and yield stress of suspensions and produce Newtonian flow behavior whereas flocculants and coagulants produce the opposite effect. The terms viscosity and yield stress are parameters of a suspension that describe its flow properties. An aggregated suspension exhibits a high viscosity and yield stress due to interparticle attractive forces while a dispersed one shows a low viscosity and yield stress. In oil sands hydrotransport and bitumen extraction processes, the desired effect is to decrease the viscosity and yield stress of slurries in order to improve bitumen liberation and produce good flow characteristics. In this case, the preferred additives would be dispersants. In other applications such as formulation of commercial coal-water slurries, the desired outcomes are good flow characteristics and stability towards settling. In this case, both dispersants and flocculants are used in the preparation process (Klein and Pawlik, 2005). Table 2.1. Classification of polymeric modifiers according to their molecular weight  Low molecular weight MW < 105 (approx.) Dispersants Organic Inorganic  High molecular weight MW > 106 (approx.) Flocculants Natural Synthetic  Dextrins,  Polyphosphates,  Starches,  Polyacrylamides  Polyacrylates  Polysilicates  Gums  Polyalkylene  Medium molecular weight, MW 105 - 106 Coagulants Synthetic Polyamines and cationic derivatives of various polymers  Oxides  Generally, when only low viscosity and yield stress are required, the choice of which dispersants to use is important. As shown in Table 2.1, good polymeric dispersants should be of low molecular weight and they should not have any significant flocculating properties that would facilitate 39  aggregation of particles and subsequent increase in viscosity and yield stress. They should also be capable of adsorbing on the surface of interest. Therefore, the most effective dispersants are anionic in nature, although non-ionic and cationic dispersants are not uncommon. The use of inorganic dispersants such as sodium hydroxide and sodium silicate has led to improved bitumen recoveries in oil sands processing, but the resulting ionic build-up in process water is expected to negatively affect the hydrotransport and bitumen flotation stages by gradually increasing ionic strength which promotes aggregation. It can thus be said that the most effective dispersants are low molecular weight anionic or non-ionic organic dispersants such as polysaccharides and polyacrylates. However, polymeric modifiers of high molecular weight can act as dispersants at high dosages. Polymeric dispersants adsorb on solid surfaces and create a physical barrier that keeps two particles apart at such a distance that van der Waals attractive forces become negligible (steric stabilization). The effectiveness of charged organic dispersants is derived from their dual action. In addition to steric stabilization, their adsorption on surfaces facilitates electrostatic repulsion between the surfaces. It is known that electrostatic repulsion creates suspensions that are kinetically stable while steric stabilization provides thermodynamic stability to the suspensions (Laskowski, 1988). For example, Figures 2.5 and 2.6 (Pawlik, 2005) show the dispersing effect of CMC (MW of 80,000 g/mol) and polystyrene sulfonate – PSS (MW of 14,000 g/mol) on aqueous suspensions of fine coal. The hydrophobic coal particles in the suspensions (59 wt. % solids) initially interacted strongly to form a network structure within the suspensions giving rise to high yield stresses. With the addition of dispersants, the yield stress gradually reduced and at high dispersant dosages, the flow curves became Newtonian with no yield stress. It should be noted that although these dispersants are effective at controlling the aggregation/dispersion properties of suspensions, there are issues related to effective mixing of the dispersants in concentrated suspensions and optimum choice of dosages. Such issues can be resolved by systematic characterization of the dispersants to understand their conformation in solution, chemistry and mode of action, and their effect on wettability of the surface of interest.  40  80  80  A  70  B C  50  60  Shear Stress [Pa]  Shear Stress [Pa]  60  40  D  30  50 40 30  20  20  10  10  0  C  0 0  A: B: C: D:  A B  70  30  60  90  -1  Shear Rate [sec ]  120  150  0  No CMC,  33.07Pa, Bingham -3 5.41*10 % of CMC,  = 25.03Pa, Bingham -3 9.36*10 % of CMC,  = 9.82Pa, Casson -2 4.51*10 % of CMC,  = 0.00Pa, Newton  30  60  90  -1  Shear Rate [sec ]  120  150  A: No PSS,  32.85Pa, Bingham -3 B: 4.64*10 % of PSS,  = 16.75Pa, Casson -1 C: 1.06*10 % of PSS,  = 0.00Pa, Newton  Figure 2.5. Flow curves obtained from rheological measurements on concentrated coal suspensions in the presence of CMC (left) and PSS (right).  Extrapolated Yield Stress [Pa]  40  30  20  10  0 10  100  10 00  10 00 0  Po lymer Do sage [g/t] C a rb ox ym e thyl C e llulos e H um ic Ac id s P o lystyr ene S ulf ona te D ex trin H ydr ox ye th y l C e llulos e H ydr ox ypr opy l Ce llulose  Figure 2.6. Yield stresses (obtained from Figure 2.5) as a function of dispersant concentration  41  2.2.1  Viscometric properties of carboxymethyl cellulose in solution  A systematic characterization of CMC not only enables one to understand its ionic character and molecular weight, but also to assess its effective macromolecular size and conformation in solution (which has implications on adsorption) and its effect on the properties of the adsorbent. Carboxymethyl cellulose is a synthetic anionic polysaccharide obtained from cellulose. A typical preparation reaction involves two steps: the first step involving treatment of cellulose with sodium hydroxide disrupts the crystalline structure of cellulose to produce an alkali-cellulose complex that is easily accessible for subsequent treatment. The second step involves the treatment of the alkalicellulose with sodium monochloroacetate (ClCH2COONa) to produce CMC. During the reaction, hydrogen atoms in one or more of cellulose’s hydroxyl groups are substituted by CH2COONa to produce CMCs of different degrees of substitution (DS). This is achieved by addition of different amounts of sodium monochloroacetate to the cellulose (Hoogendam et al., 1998b). Addition of different amounts of hydrogen peroxide to CMC of a given DS cleaves the cellulose chain at random positions and produces a series of CMCs varying in molecular weight (Hoogendam et al., 1998b). The substituted carboxy-methyl groups not only improve the solubility of CMC in water but also impart a strong polyelectrolyte character to CMC. The electrostatic repulsion between the carboxylic groups along the CMC chains is responsible for the polyectrolyte effect. This effect, first described by Fuoss and Strauss (1948a) makes the characterization of CMC an experimental challenge. The properties of CMC in solution depend largely on its macromolecular conformation which is influenced by the DS, MW, and solution chemistry (pH, temperature, ionic strength, types of ions). One of the most convenient techniques to assess polymer conformation in solution is through viscosity measurements. It is well established that dilute solution viscometry is one of the simplest and quickest methods for characterizing polymers in solution (Kulicke and Clasen, 2004; Lovell, 1989). The viscosity of a solution is a measure of its resistance to flow when a shearing force is applied. It reflects the frictional forces of all the molecules in the solution, including those of the dissolved polymer. The measurement of solution viscosity can be used to determine the intrinsic ability of the polymer to increase the viscosity of a particular solvent at a given temperature. Typically, a coiled conformation of the polymer results in a decrease in solution viscosity while a stretched, rod-like conformation results in an increase in viscosity.  42  The most commonly used devices for viscosity measurements on dilute polymer solutions are capillary viscometers (Kulicke and Clasen, 2004). There are numerous types of capillary viscometers on the market, the most common being the U-tube and suspended level viscometers. The choice of the viscometer depends on the cost, construction and usability, and volume of solution they can handle. Most suspended-level viscometers are based on the design by Übbelohde (1937). Modifications have been made to allow for control of the atmosphere above the liquid, for measurements at high temperatures, and automation of dilution. Examples of U-tube viscometers are Ostwald and Cannon-Fenske types. The Cannon-Fenske viscometers have been used in this dissertation to measure the kinematic viscosities of polymer solutions at ambient as well as elevated temperatures. These types of viscometers were previously used with satisfactory results by Ma and Pawlik (2007) to measure the kinematic viscosities of dilute guar gum solutions at room temperature as well as high temperatures of 45 °C, 70 °C, and 90 °C, and by Arinaitwe (2008) to investigate the viscometric behavior of high molecular weight, dilute polyacrylamide-based flocculants. Viscosity measurements, although seemingly easy to execute, need a high level of precise and tidy sample preparation procedures to produce reliable data. In laboratory viscometric studies, the choice of the concentration range of the polymer solutions should be such that the solutions should give a clear Newtonian response. For this reason, measurements in a capillary viscometer are taken under the assumption that the viscosity of the tested solution remains constant over the entire range of shear rates encountered as they pass down the capillary tube. A typical viscosity measurement of a polymer involves measuring the time it takes for a volume of the polymer solution to flow between two fixed marks on the capillary tube. This time is compared to the time it takes for the same volume of solvent to flow between the two marks on the capillary tube. The flow time, t, for the solution and the solvent is proportional to the viscosity, η, and inversely proportional to the density, ρ as shown in Equation 2.9 (Kulicke and Clasen, 2004). , [2.9]  43  The relative viscosity is defined as the ratio solution solvent . For dilute solutions, it is true that  solution solvent ≈ 1. Thus to a good approximation, the relative viscosity is a simple time ratio: [2.10] The specific viscosity is defined as the fractional change in viscosity upon addition of the polymer:   sp   rel  1  [2.11]  The reduced viscosity, red   sp c , is the ratio of the specific viscosity to the polymer concentration, c, and is a measure of the specific capacity of the polymer to increase the relative viscosity. The intrinsic viscosity, [η] is the limit of the reduced viscosity as the polymer concentration approaches zero. It is also the limit of the inherent viscosity (ln ηrel /c) as the solution polymer concentration tends to zero: [η] = lim c 0   sp c   lim c 1 ln rel c 0  [2.12]  For purposes of polymer characterization, the intrinsic viscosity, [η] (also called the limiting viscosity number according to the International Union of Pure and Applied Chemistry nomenclature) is of great importance. It is a useful measure of the volume demand of the polymer in a dilute solution and is the most relevant parameter used to describe the viscosity behaviour of a dilute polymer solution. It can be evaluated by measuring the reduced viscosity at different polymer concentrations and extrapolating the experimental data to zero concentration (to eliminate polymer intermolecular interactions). As will be discussed in the next paragraphs, various extrapolation methods exist and each method has its limitations. Extrapolation of experimental data for charged polymers (polyelectrolytes) such as CMC is not straightforward. The units of [η] are inverse concentration, the most commonly used being dL/g. There are a number of factors that influence the flow pattern of a polymer solution. These factors are more pronounced when there are intermolecular interactions between the polymer molecules. The aim of the intrinsic viscosity measurement is to probe the tested solution in a state where interactions between neighboring polymer molecules are minimized and only the polymer–solvent interactions are significant. This is essentially the definition of a dilute solution as given by Huggins 44  (1942) and Kulicke and Clasen (2004). In such dilute solutions, the thermal motion of the molecules predominates over the hydrodynamic and inter-molecular forces, and the intrinsic viscosity of a polymer characterizes the behavior of a single polymer chain. As the polymer concentration is increased, individual molecules are brought into contact with one another to form intermolecular entanglements, and this is accompanied by a dramatic change in flow behaviour and, in particular, by a sudden increase in the concentration dependence of viscosity. Robinson, Ross-Murphy, and Morris (1982) showed that the concentration (c*) at which the formation of entanglements occurs is inversely proportional to the volume occupied by the isolated polymer particles, and for a range of polysaccharides, they showed that c*  4/[η]. Kulicke and Clasen (2004) also recommended that in order to evaluate viscosity data with confidence, the polymer stock solution should be diluted to concentrations that give a range of relative viscosities varying from 1.2 – 2.5. The concept of the critical transition concentration between dilute (Newtonian) and moderately concentrated (non-Newtonian) polymer solutions was also discussed by Bohdanecky and Kovar (1982) who explained that the product [η]*c can be employed as a simple, approximate overlap criterion. For coiling polymers such as CMC, they showed that the borderlines of the different concentration regimes (dilute, moderately concentrated, and concentrated systems) begin at [η]*c  1 (dilute – moderately concentrated), and ends at [η]*c   10 (moderately concentrated – concentrated).  When determining the intrinsic viscosity of a polymer from dilute solution viscometry, it is therefore critical to ensure that the highest concentration of the polymer in solution, cmax, is lower than the critical transition concentration as approximately given by 1/[η]. Unfortunately, several published studies on polymers suffer from this fundamental omission, and the reported intrinsic viscosity values are thus quite questionable. The viscosity properties of a polymer solution can generally be represented by a power series in concentration, c, as follows (Lovell, 1989):  sp c     k1  2 c  k 2  3 c 2  .......  [2.13]  where [η] is the intrinsic viscosity, and k1, k2 …. are dimensionless constants. For sufficiently dilute solutions of non-ionic polymers, the higher order terms in Equation [2.13] can be neglected. There are some interactions between polymer molecules even in dilute solutions  45  (Kulicke and Clasen, 2004), but these are captured by the second term. Recasting Equation [2.13] and neglecting second order and higher terms gives a linear dependence of  red on the concentration.  sp c     k1  2 c  [2.14]  Equation [2.14] is in accordance with the well known Huggins (1942) equation in which the constant k1 is equivalent to the Huggins coefficient kH.  sp c     k H  2 c  [2.15]  The Huggins coefficient is a constant for a given polymer-solvent system and it represents the hydrodynamic interactions between the polymer molecules and the solvent. The Huggins equation is strictly applied when [η]*c <<1 because at higher concentrations, experimental data show upward curvature (Lovell, 1989). Simple plots of (ηsp/c) versus c according to the Huggins equation should yield a straight line whose intercept and slope are [η] and kH[η]2, respectively. It should be noted that higher-order terms in Equation [2.13] cannot be neglected for high molecular weight non-ionic polymers characterized by high intrinsic viscosities. Also, polyelectrolytes in dilute solution behave differently from non-ionic polymers due to the dissociation of the ionic groups that causes coil expansion, the extent of which depends on the degree of dissociation of the groups, and on the ionic strength of the solution. Thus the extrapolation of the (ηsp/c) versus c plots cannot be applied to polymers such as CMC. Charged polymers such as CMC exhibit a typical behaviour showing an increase in reduced viscosity with decreasing concentration in the low concentration region. At higher concentrations, the reduced viscosity decreases with decreasing concentration as expected for a polymer solution. Thus two extremes, a minimum and a maximum are observed at high and low concentrations, respectively as depicted in Figure 2.7. There appears to be no universal concentration range at which the (ηsp/c) versus c plots for polyelectrolytes change shape. The range is specific for each polymer–solvent system, but it is quite clear that extrapolation of the (ηsp/c) versus c plots described in the previous paragraph for non-ionic polymers would not fit such a set of experimental results.  46  100  Reduced Viscosity [dL/g]  80  60  40  20  0 0  0.002  0.004 0.006 Concentration [g/dL]  0.008  0.01  Figure 2.7. A typical graph showing the concentration dependence of the reduced viscosity for polyelectrolytes.  The universal method used to go around this problem is the iso-ionic dilution technique developed by Pals and Hermans (1948). This method was used by the same authors (Pals and Hermans, 1952) to determine intrinsic viscosities of CMC by keeping the effective ionic strength constant or nearly constant by adding salt to the polyelectrolyte solution. When simple inert electrolytes such as NaCl or KCl are added to polyelectrolyte solutions, the dissociated ionic groups are shielded by the electrolyte and at high enough concentration of the salt, the (ηsp/c) vs. c plots become linear allowing for the determination of [η] by extrapolation of ηsp/c to c→0. However, Kulicke and Clasen (2004) pointed out that the intrinsic viscosity determined in this way does not reflect the coil expansion in more dilute or salt-free solutions. It is clear that the iso-ionic dilution technique does not allow determination of intrinsic viscosities in low ionic strength solutions such as distilled water. A number of empirical relationships have been developed to evaluate viscosity data for polyelectrolyte solutions and to determine their intrinsic viscosities. Fuoss and co-workers (1948b, 1949) proposed an equation [2.16] to describe the extrapolation of the reduced viscosity of polyelectrolytes from a series of experiments.  47    sp   c      1  1         c 2 1  1  [2.16]  Where  is a polymer–solvent dependent constant. According to this equation, the intrinsic viscosity can be obtained by taking the reciprocal of the y-axis intercept of the  vs.  plot. The Fuoss equation was used by Rattanakawin (2002) to determine intrinsic viscosities of anionic polyacrylamide with degrees of anionicity up to 30%. Rey and Machado (2000) also used the Fuoss equation to determine intrinsic viscosities of fulvic acid samples in the pH range 3.0 – 8.0 where the samples exhibited strong polyelectrolyte behaviour. However, Bohdanecky and Kovar (1982) showed that the  vs.  plot is in fact not linear due to the maximum and  minimum in the (ηsp/c) vs. c plots, and that the Fuoss equation fails in the neighborhood of the maximum at low polymer concentrations. The failure of the Fuoss equation was evident in the data by Vink (1970) where a decrease in the reduced viscosity of a CMC sample tested in pure water manifested itself by an upturn of the Fuoss plot in the low concentration region. At higher concentrations, Vink’s data conformed well with the Fuoss equation. Consequently, the value of [η] evaluated according to Equation [2.16] is not quite reliable, especially in the lower concentration region that is of interest in dilute solution viscometry. Still, the values of [η] obtained using the Fuoss equation can be taken as a relative measure of a coil expansion for comparison with other methods (Kulicke and Clasen, 2004). Fedors (1979) working with latex solutions developed an equation [Equation 2.17] to describe the viscosity behaviour of uncharged polymers over a wide range of polymer concentrations. [2(ηrel1/2 -1)]-1 = ([η]c) -1 - ([η]cm) -1  [2.17]  where cm is a polymer concentration parameter. Rao (1993) and Ghimici and Popescu (1998) applied the Fedors equation to polyelectrolytes by plotting [2(ηrel1/2 -1)]-1 vs. 1/c and concluded that it can be used to describe the viscosity of some polyelectrolyte solutions over a wide concentration range including the dilute solution range. Ghimici and Popescu (1998) also fitted viscometric data with the Fuoss equation and observed that the Fedors equation gave a better fit than the Fuoss equation over a greater concentration range. Arinaitwe and Pawlik (2013) and Arinaitwe (2008), using linear polyacrylamide flocculants, tested several methods with the aim of identifying one that would best 48  describe raw viscosity data in order to determine the intrinsic viscosity without having to apply the iso-ionic dilution technique. An analysis and comparison of intrinsic viscosities obtained using different equations showed that the Fedors equation could quite universally be used to determine intrinsic viscosities under all the experimental conditions (pH, temperature, ionic strength). As stated above, the intrinsic viscosity of a polymer can be used to analyze the conformation of the polymer chains in solution. In general, five ‘conformation zones’ can be described (Pavlov et al., 1997): extra-rigid rod, rigid rod, semi-flexible coil, random coil, and compact sphere. Additional information about the conformation of the polymer can be obtained by determining the dimensions of the macromolecule, and the parameter determined depends on the conformation type. For a rodlike conformation, the length is relevant; if it is a coil, its flexibility; if it is sphere-like, its radius; and if its conformation is between a sphere and a rod or ellipsoid, its axial ratio is relevant. It should be noted that coiling (in high ionic strength solutions) results in a decrease of the effective size of the polymer molecules and it is therefore easy to imagine that coiling should result in high adsorption densities of the polymer on adsorbent surfaces. Such a relationship was obtained by Pawlik et al. (2003) who observed that adsorption densities of coiling anionic CMC on illite increased dramatically with increasing ionic strength while those of the non-coiling nonionic guar gum were practically independent of ionic strength. Similar results were obtained by Parolis et al. (2008) who found that increased coiling of CMC in calcium electrolytes resulted in an increase in the adsorption density on talc. Microflotation experiments were used to confirm that an increase in ionic strength (resulting in CMC coiling) resulted in enhanced talc depression. The authors attributed the enhanced talc depression to the ability of high amounts of the coiled CMC molecules to closely pack on the talc surfaces. The intrinsic viscosity also allows the determination of the polymer molar mass (M) to be made through the empirical relation (Equation 2.18) first proposed by Staudinger and Heuer (1930). [η] = constant × M  [2.18]  The constant depends on the polymer-solvent system. However, later experimental work showed that Staudinger’s equation was limited and a more general relationship, the Mark-Houwink-Sakurada (MHS) equation (Equation 2.19) was proposed (Mark, 1938; Houwink, 1940).  49  [η] = KMα  [2.19]  where K and α are constants for a given polymer/solvent/temperature system. The molecular weight determined through this equation is a viscosity-average molecular weight (Mv) defined by Equation 2.20. It lies between the weight-average molecular weight, MW and the number-average molecular weight, Mn. The number-average molecular weight is the mean value of the molar mass and can be calculated from the ratio of the sum of all molecules and the number, n of all molecules (Equation 2.21), while the weight-average molecular weight can be determined if the fractions of mass mi = ni × Mi of the single molecules are averaged over the total mass as shown in Equation 2.22 (Kulicke and Clasen, 2004). 1    mi M i   Mv      mi   .....[2.20]    ni M i Mn     ni      .....[2.21]    mi M i Mw     mi      .....[2.22]  where mi is the total weight of molecules of molecular weight Mi. The exponent α is a measure of the solvent quality and is therefore indicative of the shape of the polymer coil in solution – whether stiff or flexible chains. It is also a measure of the interactions between the polymer and the solvent. Equation 2.19 can be written as log [η] = log K + α·log M  [2.23]  to facilitate determination of α if [η] and M of the polymer system can be determined. Different methods for determining the molecular weight are available including light scattering, size exclusion chromatography or analytical ultracentrifugation. The different methods have limitations which have to be taken into account. For example, light scattering is sensitive to the presence of polymer  50  aggregates and tends to yield higher than expected molecular weights (Sitaramaiah and Goring, 1962). A plot of log [η] versus log M yields a straight line with slope α. Kulicke and Clasen (2004) provided a review of the influence of the solvent quality and solution structure on the [η]–M relationship. When the polymer chain–solvent interactions are so small that the coil is not contracted or expanded, the so-called theta conditions are reached and the coil has unperturbed dimensions in solution. In such a solution, the theoretical value of the exponent α should be 0.5. The exponent α increases as the coil expands in good solvents. Generally, 0.5≤α ≤0.8 for flexible chains in a good solvent, 0.8≤α ≤1.0 for inherently stiff molecules and 1.0≤α ≤1.7 for highly extended chains (Lovell, 1989). According to Flory (1953), the exponent α = 1 indicates the limit of free-draining and is the highest obtainable for a random coil. A polymer molecule is said to be freedraining when solvent molecules flow freely past each segment of the chain, and non-draining when solvent molecules within the coil move with the polymer molecule (Lovell, 1989). Generally, the characterization studies done on CMC indicate an extended configuration at low ionic strength due to expansion of the polyelectrolyte molecule (Sitaramaiah and Goring, 1962; Schneider and Doty, 1954; Vink, 1970; Brown and Henley, 1964; Kästner et al., 1997). Sitaramaiah and Goring (1962) determined the intrinsic viscosities of CMC samples with molecular weights (sedimentation-diffusion averages) ranging between ~45,000 and 450,000 g/mol and similar DS (0.66 – 0.73). For ionic strength solutions below 0.01 mol/L NaCl, the iso-ionic dilution was employed to determine [η]. From the logarithmic dependence of [η] on sedimentationdiffusion molecular weights, the authors found the MHS exponents (α) to be 0.91 and 1.40 in 0.1 mol/L and 0.001 mol/L NaCl, respectively, indicating a free-draining configuration at 0.1 mol/L NaCl and an extended configuration at low ionic strength. Kulicke et al. (1996) determined the exponent α to be 0.9 for CMC samples ranging in molecular weight from 200,000 to 2,000,000 g/mol at a constant DS of 1. Brown and Henley (1964) using CMCs of DS 1.06 and MW ranging from ≈150,000 – 1,000,000 g/mol also found that α increased from 0.74 (0.2 mol/L NaCl) to 0.95 (0.005 mol/L NaCl) and attributed the increase to the random coil – extended rodlike transition. The exponent α at infinitely high ionic strength was 0.6. From the intrinsic viscosities and ionic strength, the authors determined [η] at infinite ionic strength ([η]I=∞) by graphical extrapolation of [η] versus 1/x, where x is the 51  thickness of the Debye-Huckel ionic atmosphere. As well, they determined the [η]θ, the intrinsic viscosities of CMC in the unperturbed state (at the theta point) by utilizing the Stockmayer-Fixman theory (1963) where the MHS exponent (α) is fixed at 0.5. A comparison of [η]I=∞ with [η]θ showed that aqueous solutions of CMC are generally very far from the theta point even when the polymer chain carries no charge as expected at infinite ionic strength. Their conclusion was substantiated by the exponent α at infinite ionic strength (0.6) which is higher than the value of 0.5 that would be expected in the absence of any charge effects. The high polarities of the ionizable groups on the polymer chain were thought to contribute to the high degree of solvent–polymer association, which contributes to the molecular extension in solution. Kästner et al. (1997) showed that CMC macromolecules (molecular weights between 9,000 and 360,000 gmol-1, DS between 0.75 and 1.47) coiled upon increase in ionic strength. Generally, the macromolecules coiled with addition of salt (NaCl), anionic and cationic surfactant molecules or acid (HCl). The level of coiling was lowest for the anionic surfactants and highest for HCl. The pronounced effect of pH on the conformation of CMC was related to the protonation (through H+) and screening of the carboxylic groups by sodium ions. In this work, the coiling of the CMC macromolecules manifested itself in decreasing zero-shear viscosity at higher ionic strengths. However, as discussed above, a more useful parameter to infer coiling in polymer molecules is by determining intrinsic viscosities at different ionic strengths. The above-mentioned studies did not systematically investigate the whole range of physicochemical parameters needed to elucidate the behavior of CMC in solution. For example, there was no systematic study done on the influence of pH on the conformation of CMC in distilled water and salt solutions. Moreover, there is no literature data describing the accurate determination of intrinsic viscosity of dilute solutions of CMC in low ionic strength solutions without using the isoionic dilution method. 2.2.2  Flexibility of carboxymethyl cellulose in solution  The viscosity properties of linear polymer solutions are influenced by the flexibility of the polymer backbone. A very useful representation of the linear flexibility of polymers is in terms of the persistence length of the equivalent worm-like chain (Kratky and Porod, 1949). Generally, polymers with a higher persistence length will be more extended while those with a lower persistence length 52  tend to be more coiled. The extended polymer chain moves much slower than the solvent leading to a higher viscosity. Since intrinsic viscosity measurements can adequately describe changes in the dimensions of the polymer chain resulting from changing conformations, and since the polymer dimensions are related to the persistence length (Lp), viscometry can be used to accurately determine L p. In the worm-like chain model the polymer chain is treated as a continuous cylinder with the length of each segment l → 0 and the number of segments n → ∞. The persistence length Lp is defined as the average projection length along the initial direction of a chain of contour length Lc and in the limit of Lc → ∞ (Figure 2.8) (Yamakawa, 1971). Thus, Lp → 0 and Lp → ∞ correspond to a perfect coil and perfect rod, respectively (Harding, 1997). The stiffness of a polymer chain can also be assessed by an alternative parameter, the Kuhn length Lk (Kratky and Porod, 1949), defined as [2.24]  where Nk is the number of Kuhn segments in the polymer molecule. For a semi-flexible chain, the Kuhn length is about twice the persistence length.  Figure 2.8. The persistence length Lp and contour length Lc of a linear polymer (from Harding, 1997). Lc represents the length of the fully extended chain.  It should be noted that for charged polymers (polyelectrolytes) like CMC, there are two contributions to the chain persistence length: one resulting from the repulsive interactions between charged segments of the chain, the so-called electrostatic persistence length Lpe, and the other representing the chain stiffness intrinsic to the polymer backbone Lp0 and denoted as the intrinsic persistence length (Odijk, 1977; Skolnick and Fixman, 1977). Theoretical considerations of the 53  persistence length of worm-like polyelectrolytes, independently described by Odijk (1977) and Skolnick and Fixman (1977), showed that the two contributions are additive, i.e.,  Lp  Lp 0  Lpe  [2.25]  Lk  Lk 0  Lke  [2.26]  Similarly for the Kuhn length,  where Lk0 is the bare Kuhn length that represents the chain stiffness of the uncharged polymer backbone and Lke, the electrostatic Kuhn length arises from the repulsive coulombic interactions between fixed charges on the chain (Davis, 1991). Both the electrostatic Kuhn length and persistence length tend to zero at infinite ionic strength, thus one way of determining Lp0 is to extrapolate experimental Lp at different ionic strengths to infinite ionic strength. Most available models were developed to estimate the structural dimensions either of ideal flexible coils or rigid chains and thus they usually need to be tested for their applicability to semiflexible or stiff polymers. The wormlike chain model was initially described by Kratky and Porod for cellulose, and has since been applied to a number of other macromolecules such as the semi-flexible poly(tert-butyl crotonate) with a persistence length of 5 – 6 nm (Noda et al., 1981), the ‘locally stiff’ carboxymethyl cellulose as assessed by Skolnick and Fixman, 1977, the semi-flexible carboxymethyl cellulose with a persistence length of 16 nm (Hoogendam et al., 1998b) and the more rigid DNA (Marko and Siggia, 1995; Hagerman, 1988). In the study by Skolnick and Fixman (1977), the electrostatic persistence length of a theoretical polyelectrolyte represented by a continuous uniform charge distribution was determined using the wormlike model. The authors neglected excluded volume effect (volume around a polymer chain segment from which other chains are excluded) in their analysis which limits the applicability of their theory to data in low ionic strength and theta solvents where the polyelectrolyte chain is sufficiently stiff that excluded volume effects are negligible. The theoretical values were compared with experimental data on CMC (DS = 1.15, MW = 440,000 g/mol) dimensions determined by Schneider and Doty (1954) in aqueous NaCl, and a reasonably good agreement was obtained. The theoretical values of Lpe ranged from 0.15 nm in a solution of high ionic strength (I = 0.5) to 14.6 nm in I = 0.005, while the experimental values ranged from 0.14 nm in I = 0.05 to 16.8 nm in I = 0.005. 54  At the ionic strength used in most tests in this dissertation (I = 0.01 mol/L NaCl), the theoretical value was 7.3 nm while the experimental value was 7.4 nm. Skolnick and Fixman extrapolated their data to infinite electrolyte concentration and obtained an Lp0 value of 16.6 nm, thus the total persistence length of CMC at 0.01 mol/L NaCl (Lpe + Lp0 = 23.9 nm) indicates a semi-flexible chain. Hoogendam et al. (1998b) utilized the electrostatic wormlike model as described by Davis (1991) to determine the intrinsic persistence length of CMC from size exclusion chromatography data. The electrostatic wormlike chain model was developed to extend the wormlike chain model (originally developed for non-ionic polymers) to dilute solutions of polyelectrolytes where electrostatic interactions change the size and solution conformation of polymer molecules leading to increases in such parameters as the radius of gyration (Rg) and intrinsic viscosity as ionic strength decreases (Davis, 1991). Hoogendam et al. obtained Rg and MW of different CMC fractions by multi-angle laser light scattering via a Zimm plot. CMCs of different DS (0.75 – 1.25) and MW (120,000 – 1,200,000 g/mol) were utilized in their study. The intrinsic persistence length (without electrostatic contributions) was assessed to be 17 nm (in 0.02 mol/L NaNO3) and 15 nm (in 0.1 mol/L NaNO3), irrespective of the DS. Using Odijk’s model (1979) for the swelling of polyelectrolytes, lower values of Lp0 (13 nm in 0.02 mol/L NaNO3 and 11 nm in 0.1 mol/L NaNO3) were obtained. The Lp0 values indicate that CMC can be considered as a semi-flexible polymer. For highly charged polyelectrolytes, the Davis theory was preferred as it gives a complete description of electrostatics and takes into account the molecular properties (such as chain length and cross section of the molecule) into consideration. In the worm-like chain and Odijk’s models, excluded volume effects are taken into consideration by introducing a line expansion factor which is a function of the excluded volume parameter. To calculate this parameter, Odijk’s model assumes the probability of contact between segments in the chain to be at the limiting value ( ). Data by Hoogendam et al. (1998b) illustrated that the assumption was not valid for the whole range of CMCs that they tested which partly explained the lower Lp0 values obtained with Odijk’s model. It should be noted that Hoogendam et al. evaluated their size exclusion chromatography data by using Lp0 as an adjustable parameter and the values reported above are those that gave the best fits to the experimental data (Rg vs. MW). Thus, Lp0 was not determined in an absolute manner. The Lpe values calculated using the two models were comparable at a given DS and salt (NaCl) concentration, and ranged from ≈ 0.3 – 2.9 nm.  55  Majority of the literature data described above for CMC dimensions were obtained by light scattering with viscometry most times applied as a complementary technique. One way of analyzing light scattering data involves extrapolation of the intensity data to zero angle to determine the radius of gyration and the weight-average molecular weight. However, as described by Mijnlieff and Coumou (1968) and Davis (1991), such an extrapolation is only valid with data measured at sufficiently low angles, and most reported data seems to suffer from this limitation. For example, by calculating the light scattering form factor (P[θ]-1) range, Davis (1991) analyzed the data reported by Brown and Henley (1964) and Brown et al. (1963) in their characterization studies on four fractions of CMC and concluded that most of the data derived from light scattering were invalid as they fell out of the acceptable range (1<P[θ]- 1  1.3). The invalidity of Brown and Henley’s light scattering  data was probably responsible for the divergence observed in the CMC chain segment lengths computed from light scattering and sedimentation experiments. However, the lowest molecular weight fraction (MW = 147,000 g/mol, DS = 1.06) tested by Brown and Henley in aqueous solutions of cadoxen (triethylenediamine cadmium hydroxide) lay approximately within the acceptable range. Davis (1991) used the Rg (extrapolated to infinite ionic strength) and MW data for this sample and applied the electrostatic wormlike chain model to calculate the Lk0 value for that fraction and found it to be 10.8 nm, consistent with a semi-flexible chain, and similar to the Lp0 value determined by Hoogendam et al.(1998b) in 0.1 mol/L NaNO3. Brown and Henley (1964) determined the unperturbed dimensions of CMC using the StockmayerFixman theory by plotting [η]/MW1/2 versus MW1/2. The magnitudes of the Kuhn equivalent chain segment length and the steric factor (measure of hindrance to internal rotation about the intermonomer bonds of the chains) indicated that CMC is a typical flexible polymer. Tricot (1984) fitted Brown and Henley’s [η]-MW data to the hydrodynamic theory of Yamakawa and Fujii (1974) and obtained Lk in the range 16.0 – 46.2 nm for solutions with ionic strength ranging from 0.2 – 0.005 mol/L, respectively. In the calculations by Tricot, the excluded volume effects were not considered and the polymer backbone radius was allowed to vary with ionic strength. Davis (1991) recalculated the Lk values by applying the electrostatic wormlike model which takes into account the excluded volume effects and fixes the backbone radius at a value determined by atomic bond lengths (0.5 nm for CMC). For this reason, Davis’s method is considered more rigorous. At low ionic strength, Tricot’s values of Lk were much larger than those calculated by Davis but the difference between the two sets of values narrowed as the ionic strength increased. It should be noted that 56  excluded volume effects for polyelectrolytes are negligible near the θ-state or at high salt concentrations (Davis, 1991), thus the close Lk values obtained by Tricot and Davis at high ionic strength (0.5 mol/L NaCl) were explained by the reduced electrostatic excluded volume effects. It was also reported by Sitaramaiah and Goring (1962) that the presence of aggregates in CMC solutions caused anomalies in the light scattering data. For example, the authors noted a marked discrepancy in the light-scattering MW data and those computed from hydrodynamic parameters from sedimentation, diffusion and viscometry. While the sedimentation-viscosity molecular weights agreed well with the sedimentation-diffusion data, extremely high values obtained from lightscattering data did not agree with either of the other techniques used. Sitaramaiah and Goring attributed the presence of aggregates to unreacted cellulose left during the preparation of alkali cellulose and subsequent conversion to CMC. With regard to the effect of this small proportion of aggregates on the intrinsic viscosity, the authors observed that aggregates would behave as compact spheres, and thus their contribution to the intrinsic viscosity would be negligible. It should be noted that attempts at removing the aggregates by rigorous purification and centrifugation procedures did not eliminate the irregularities noted in the light scattering molecular weight data. Thus, from the study by Sitaramaiah and Goring, it would seem that the presence of aggregates in CMC solutions obscures light scattering data more than hydrodynamic techniques such as viscometry. A number of theories relating the dependence of [η] on MW allow the evaluation of polymer molecular dimensions. Some of them involve extrapolation of MW/[η] versus MW1/2 or MW1/2/[η] versus MW-1/2 and extrapolating MW1/2 or MW-1/2 to zero respectively, while others involve finding theoretical best-fit curves to the [η]-MW data, and as explained by Bohdanecky (1983), these procedures suffer from very restrictive assumptions and are not free from ambiguity. For this reason, Bohdanecky (1983) described a method to determine persistence length from [η] data for stiff-chain polymers based on the theory of Yamakawa and Fujii (1974). The method involves expressing the intrinsic viscosity data in a simple analytical form, which on rearrangement provides a basis for evaluation of Lp with good accuracy. Bohdanecky’s method is valid over a broad range of chain lengths and has been widely applied to a range of polymers including flexible and semi-flexible polymers such as the charged xanthan (Sato et al., 1984), sodium hyaluronate (Mendichi et al., 2003), methylcellulose (Patel et al., 2008) and neutral galactomannans (Patel et al., 2006).  57  Using the Yamakawa-Fujii theory, Bohdanecky (1983) showed that the dependence of [η] on the molecular mass of stiff-chain polymers can be expressed in a simple relation that provides a basis for the estimation of the persistence length. In its simplest form, Bohdanecky’s equation is: [2.27] Where Aη and Bη are parameters related to the molar mass per unit length (ML) and the persistence length (Lp) through the following equations: [2.28] and [2.29] where ɸ0,∞ = 2.86 × 1023 mol-1 is the limit of ɸ0 for Lc,r→∞. ɸ0 is the Flory-Fox viscosity constant (Yamakawa and Fujii, 1974) and Lc,r is the reduced contour length (Lc,r= Lc/2Lp). Both A0 and B0 are tabulated functions of the reduced hydrodynamic diameter, dr = d/2Lp, and can be found in Bohdanecky’s article (1983). According to the above equations, Lp can easily be evaluated from the slope  of the plot of  vs.  . The parameter B0 is a slowly decreasing function of dr  while the dependence of A0 on dr is more pronounced. A mean value, B0 = 1.05 can be used in the evaluation of Lp (Bohdanecky, 1983). The molar mass per unit length, ML can be obtained from other measurements such as light scattering or electron microscopy. It should be noted that Bohdanecky’s method does not account for the excluded volume effect and the expansion of the polymer molecular dimensions. These two effects are shown in the following equation: [2.30] where  is the viscosity expansion factor,  the perturbed state,  is the intrinsic viscosity for expanded molecules in  is the intrinsic viscosity of unperturbed chains, and z is the excluded volume  parameter. The dependence of the dimensionless coefficient, 58  on the chain contour length is not  exactly known. However, the comparable first order coefficient factor  expression (  in the end-distance expansion  is the ratio of the end-to-end distance in the perturbed state to that in the  unperturbed state) is known to be close to unity at low chain contour lengths and close to the limiting value, 1.276, at high chain contour lengths (Bohdanecky, 1983). On the expectation that  follows  the viscosity expansion parameter is expected to  the same dependence on chain contour length as  be very close to unity at low chain contour lengths, in which case the initial slope of the vs.  will not be significantly affected by the expansion of the molecular  dimensions. 2.2.3  Molecular weight and molecular weight distribution  As shown in the preceding paragraphs, knowledge of the molecular weights of the polymers is needed in order to determine Lp from intrinsic viscosity. The polymer molecular weights need to be determined separately. In addition, it should be noted that viscometry becomes a suitable method to determine Lp if applied to monodisperse polymers, and in fact all the preceding discussions about the evaluation of molecular dimensions from [η]-MW data assume monodisperse chains. For polydisperse samples containing a mixture of chains with different molecular weights, information about the molecular weight distribution of the polymers is needed (Hoogendam et al., 1998b). Both the molecular weight and molecular weight distribution of polymers can be obtained from analytical ultracentrifugation, a technique that has many advantages over the conventional methods such as size exclusion chromatography and light scattering since it is based on fundamental hydrodynamic and thermodynamic principles, does need any calibration, and does not require use of any columns or membranes. Size exclusion chromatography data can be affected by specific interactions between polymer chains and the filling gel. Dynamic light scattering was applied widely to characterize polymers, but the technique is sensitive to the presence of molecular aggregates which, as has been described earlier, could obscure the data. Analytical ultracentrifugation, hereafter abbreviated as AU, was a method of choice for polymer characterization up to the 1970’s but greatly suffered due to lack of new instrumentation capable of digital data acquisition (Lebowitz et al., 2002). In the 1990’s, new instrumentation and software for recording and analyzing AU data greatly improved the capabilities of the technique. The technique is based on the optical detection of concentration profiles of a sample contained in specially designed 59  cells and subjected to very high centrifugal forces of up to 1,000,000g. Excellent reviews on AU can be found in Lebowitz et al. (2002) and Harding (2005a). Harding’s review focused on characterization of polysaccharides and so is relevant to the polymeric additives used in this dissertation. AU data can be obtained through two types of complementary experiments: sedimentation velocity (SV) and sedimentation equilibrium (SE). From an SV experiment, the rate of movement of macromolecules in a solution can be monitored from which sedimentation coefficients can be obtained. Sedimentation coefficients contain information regarding the size and shape of the macromolecules. SE data allows for the determination of molecular weight and molecular weight distribution as well as self-association and heterogeneous macromolecular interactions. It should be noted that recent computations allow for manipulation of SV data to obtain fairly accurate molecular weights and molecular weight distributions when the solution conformation of the solute is known or can be assumed via the sedimentation coefficient–molecular weight scaling law (Harding et al., 2011). Analytical ultracentrifuges are equipped with both UV/visible absorption optics and interference optics. The availability of different optics offers the option of investigating a large range of concentrations. One of the features of the interference optics is that macromolecules can be detected through refractive index changes enabling nonabsorbing polymers such as polysaccharides to be analyzed (Lebowitz et al., 2002). By adjusting the rotor speed, macromolecules ranging in size from proteins to carbohydrates or synthetic polymers can be investigated as long the absorbance or refractive index of the solute differs from that of the solvent. Thus, AU can be used to determine a wide range of molecular weights from several hundreds to many millions, unlike other methods such as SEC-MALLS (size exclusion chromatography coupled to multi-angle laser light scattering) that have limitations in the range of molecular weights that they can analyze. A schematic of the interference optical system is shown in Figure 2.9. Sedimentation velocity experiment In an SV experiment, a polymer solution is placed in an ultracentrifuge cell and a high rotor speed is applied to the cell causing rapid sedimentation of the solute to the cell bottom and depletion of the solute at and near the meniscus. A concentration boundary forms between the depleted region and 60  the plateau where the concentration of the solute is uniform. The boundary moves towards the bottom of the cell with time as shown in the schematic in Figure 2.10 and in Figure 2.11 showing SV data (fringes vs. radial distributions) obtained for one of the CMC samples tested in this dissertation.     Figure 2.9. Schematic showing the interference optical system. Collimated light emerges from the source, A, as two narrow stripes. At the cell, B, one stripe passes through the reference solution, R, and the other passes through the sample, S. These two beams are brought together by the condensing lens, C, at its focal point, Iʹ. A cylinder lens, D, magnifies the image at Iʹ, so that the resultant fringes are greatly expanded and in focus on the camera, F. A camera lens, E, brings the the 2/3 plane of the cell, Iʹ, into focus on the camera. Thus, the image seen (inset) is the superposition of the cell and the fringes (from Laue, 1996).  61  Fringes  Figure 2.10. Schematic showing the concentration versus radial displacement at different times after start of an SV experiment.  0.4  0.4  0.3  0.3  0.2  0.2  0.1  0.1  0  0  -0. 1  -0. 6  6. 5  7  Radius  Figure 2.11. Interference fringes versus radial displacement at different times for one of the CMCs tested in this dissertation. The sample was run at 45,000 rpm. In the XL-I instrument, a Fourier transformation converts the fringes into a record of concentration as a function of the radial displacement from the axis of rotation.  62  In the figures, it can be seen that the plateau region decreases in concentration as the boundary moves towards the bottom of the cell. Given sufficient time, the solute begins "piling up" at the bottom of the cell and as the solute concentration increases, diffusion of the solute slowly begins to oppose the sedimentation process. Diffusive broadening of the boundary as it moves towards the bottom of the cell is apparent in Figure 2.10. The rate of movement of the boundary can be measured from which the sedimentation coefficient, s, can be evaluated. A balance of the forces (gravitational, buoyancy and frictional) acting on a macromolecule suspended in a solvent gives the Svedberg equation (Equation 2.31) (Svedberg and Pedersen, 1940): [2.31] where u is the radial velocity of the solute,  is the angular velocity of the rotor,  is the radial  position (distance of the macromolecule from the axis of rotation), MW is the molecular weight of the macromolecule in g/mol,  is the partial specific volume (the volume in mL that one gram of the  macromolecule occupies in solution),  is the density of the solvent in g/ml,  number, and f is the frictional coefficient. The term  is Avogadro’s  is the sedimentation coefficient, s, and is  defined as the velocity of the macromolecule per unit gravitational acceleration. Generally, macromolecules with different sizes, shapes, and molecular weights will move with different velocities in a given centrifugal field, and thus will have different sedimentation coefficients. The sedimentation coefficient is commonly expressed in Svedberg (S) units, which corresponds to 10-13 seconds. The SV data obtained using one of the AU optical systems consists of scans of relative concentration (in units of fringes displacement perpendicular to the direction of sedimentation in the interference optical system) versus the radial displacement, r, from the axis of rotation at different times. The concentration, c, profiles versus radial direction, r, and versus time, t, distribution, c(r,t), can be modelled directly with the underlying transport function known as the Lamm equation (Equation 2.32) (Lamm, 1929) [2.32] where D is the translational diffusion coefficient and 63  is the angular velocity.  The solution to Equation 2.32 can be achieved using nonlinear least-squares regression using several software packages that have been well reviewed by Lebowitz et al. (2002). The program SEDFIT developed by Schuck and co-workers (Schuck, 1998; Dam and Shuck, 2004) has several advantages over the other programs and was thus used in this dissertation to analyze the SV data for carboxymethyl cellulose. The main advantage of SEDFIT is its generality – the software allows sedimentation coefficients to be determined even when no clearly visible boundary is formed. Only a molecular redistribution of the macromolecules as a result of the applied centrifugal force is needed (Lebowitz et al., 2002). In addition, the software can be used to model a wide spectrum of sedimentation processes and is capable of analyzing very small molecules (< 0.1 S) and large particles (>1000 S). The SEDFIT solution to Equation 2.32 produces two distributions. The first distribution is obtained through least-squares fitting of the sedimentation boundaries and is designated ls-g*(s). The procedure involves direct boundary modeling with Equation 2.32 taking D = 0, thus no correction for diffusion is made in the ls-g*(s) distribution. The second distribution c(s) is obtained by assigning one frictional ratio to the whole macromolecular distribution in order to deconvolute diffusion effects. The mathematical description of the SEDFIT solution to Equation 2.32 is described in more detail in Appendix 1. The assignment of one frictional ratio to the whole macromolecular distribution is limited only to monodisperse single solutes whose translational frictional ratio can be accurately determined and used in the data analysis to produce reliable diffusion-corrected sedimentation distribution coefficients (Harding et al., 2011). For polydisperse systems the macromolecules possess different frictional ratios such that describing them using only one parameter is not applicable, thus the c(s) versus s profiles obtained as described above are not so reliable. However, as noted by Harding et al. (2011), the sedimentation coefficients of large polymeric systems such as those investigated in this dissertation are not largely affected by diffusion effects such that the ls-g*(s) profiles obtained from their analysis should give a good representation of the distribution. Also, the width of the sedimentation coefficient distribution rather than the peak is most affected by sedimentation as demonstrated by Patel et al. (2006) for guar gum, thus in case diffusive effects are significant, the correct value of s is still reported. In addition to polydispersity, repulsive macromolecular interactions in charged macromolecular systems, such as for CMC, decrease the measured sedimentation coefficients. In order to suppress 64  the non-ideality arising from such charge effects, a high ionic strength is recommended by Harding (2005a). On the other hand, sample concentration depends on the optical path length of the cell and the optical properties of the sample. A concentration of 200 mg/L is recommended for sedimentation velocity experiments on the analytical ultracentrifuge used in this dissertation. However, this concentration is well above the critical overlap concentration (c* ≈ 1/[η]) above which CMC molecules start interacting, thus reliable data can only be obtained at a concentration lower than 200 mg/L. Therefore, to minimize hydrodynamic and thermodynamic non-ideality resulting from molecular co-exclusion and CMC polyelectrolyte behaviour, the ionic strength and sample concentration should be carefully chosen. Sedimentation equilibrium experiment In an SE experiment, the macromolecular solution is centrifuged at much lower angular velocity than is required for an SV experiment. As already stated, the sedimentation of the macromolecules towards the bottom of the cell gradually results in a polymer concentration gradient that causes the diffusion of the polymer in the opposite direction. Equilibrium is reached when there is no net mass transport in the cell, i.e., the sedimentation process is balanced by diffusional transport. The steadystate pattern achieved at equilibrium is a function only of MW and not the molecular shape since there is no net transport or frictional effects (Harding, 2005a). The mathematical approach to analyzing SE data is described in Appendix 1. For polydisperse polysaccharides, at equilibrium, there is a redistribution of species of different molecular weight in the cell. The higher MW species appear at the bottom of the cell while the lower MW fractions concentrate near the meniscus. Thus in order to obtain the molecular weight the complete concentration distribution profile across the cell height has to be considered (Harding, 2005a). In addition, non-ideality arising from co-exclusion and polyelectrolyte effects in polysaccharides has to be corrected. The constraints stated above result in an apparent weight average molecular weight, MW,app that has to be corrected to MW. If the SE experiment is performed at a sufficiently low concentration, the effects of thermodynamic non-ideality are reduced and MW,app ~ MW. However, the optical path length (12mm) of the cells used in most analytical ultracentrifuges limits the minimum concentration to ~0.5 mg/ml (~500 mg/l) which is too high for polymeric  65  systems. Performing the SE experiment at this concentration would lead to underestimates of MW due to the earlier mentioned non-ideality effects. A convenient alternative to sedimentation equilibrium is to transform the sedimentation coefficient distribution from a sedimentation velocity experiment into a molecular weight distribution from which MW can be evaluated. Reliable measurements at much lower concentrations of polymeric systems (0.1 mg/ml and below) can be tested in sedimentation velocity experiments with the available 12 mm path length cells. The method employed was described by Harding et al. (2011) and is based on the Fujita method (Fujita, 1962) for transformation of sedimentation coefficient distribution into a molecular weight distribution for linear polymers with a random coil conformation. Molecular weight, MW and molecular weight distribution (f(M)) from SV data The sedimentation coefficient distribution, ls-g*(s) can be transformed into a molecular weight distribution, f(M) using the Extended Fujita method (Harding et al., 2011). This method is an extension of the Fujita (1962) transformation of a differential distribution g(s) of sedimentation coefficients into a differential distribution f(M) of molecular weight for linear polymers with a random coil conformation. The g(s) is the population of species with a sedimentation coefficient between s and s+ds, and f(M) is the population of species with a molecular weight between M and M+dM. The transformation requires prior knowledge of a pair of s and M values from other sources. The method is built into the sedimentation velocity analysis program SEDFIT (Schuck, 2000, Brown and Schuck, 2006) and was recently successfully applied to neutral, polyanionic and polycationic polysaccharides (Harding et al., 2011; Gillis et al., 2012). The approach to using sedimentation velocity data also provides an alternative to rapid molecular weight determination for materials that cannot be analyzed with SEC-MALLS such as in cases where column interactions are suspected or in cases where the molecular weight of very large macromolecular systems exceeds the separation range in SEC-MALLS (Harding et al., 2011). A detailed description of the transformation of a g(s) distribution into a molecular weight distribution is given in Appendix 1.  66  2.3 Experimental assessment of aggregation/dispersion in concentrated suspensions. Because oil sand slurries are concentrated multiphase systems, also containing natural surfactants and other added chemicals, it becomes complicated to study the dispersion/aggregation behaviour using traditional optical based systems such as turbidity or image analysis. Techniques such as those utilized by Wallwork et al. (2004) and Luthra et al. (2004) to study bitumen liberation are suited for dilute slurries that provide a clear contrast between bitumen and solid particles and as such, their applicability to high solids content (concentrated) slurries is rather limited. In the present work, the dispersion/aggregation state of concentrated oil sand slurries was determined using rheological methods. The rheological measurements were complemented by sedimentation tests to confirm the stability of the oil sand slurries towards aggregation and settling. 2.3.1  Rheology  2.3.1.1 Overview It is well established that there is a relationship between the rheological behavior of a suspension and the inter-particle forces acting within the suspension (Rand and Melton, 1977; Nguyen and Boger,  1983).  Therefore,  bitumen–mineral–water  interactions  which  determine  the  aggregation/dispersion state of oil sand slurries can be studied through rheological measurements. Gutierrez (2009) and Gutierrez and Pawlik (2012a) in their studies on quartz-bitumen model mixtures as well actual oil sand ores showed that bitumen liberation from the solids results in a measurable rheological response. Thus, it is possible, by following the rheological response of an ore slurry, to probe the aggregation and dispersion state of the slurry. Since addition of dispersants is expected to alter the inter-particle interactions in aggregated suspensions, rheology can be applied to confirm the change in the state of the slurry from aggregation to dispersed due to the action of the dispersant. 2.3.1.2 Theory The properties of suspensions can be thought of as being in between those of purely elastic solids and purely viscous liquids. When a force is applied to an ideal elastic solid, it deforms reversibly. The relation between the applied force and the resulting deformation can be described by Hooke’s law: 67  [2.33] where  is the force per unit area (stress),  is the relative length change (strain) and G is the constant of proportionality (elastic modulus). In case of a purely viscous liquid, an applied stress deforms the liquid irreversibly and results in flow that can be described by Newton’s law of viscosity: [2.34] where  is the constant of proportionality (Newtonian viscosity) and   d  / dt is the rate of straining, often referred to as the shear (or strain) rate. The Newtonian viscosity is often simply called viscosity. When a small amount of particles is added to a fluid, the pure viscous flow of the liquid is altered due to the resulting hydrodynamic disturbances. There is a corresponding increase in the viscosity that is dependent on the size of the solid particles among other parameters. It can then be said that the resulting suspension has deviated from the Newtonian behavior described by Equation 2.34. At very low solids addition, the inter-particle interactions and hydrodynamic disturbances are low due to the large separation between individual particles. This is a case of a dilute suspension. However, as the solids content increases, interactions between the particles become more important and a further increase in solids content will result in a concentrated suspension in which many particles can interact simultaneously through the ever present hydrodynamic effects as well as: attractive dispersion (London-van der Waals) and hydrophobic forces; and repulsive electrostatic, hydration and steric (in the presence of a polymer) forces. According to the modified DLVO theory (Yotsumoto and Yoon, 1993, Yoon et al., 1997), the net forces present in the suspension are determined by the balance of the attractive and repulsive forces. Depending on the magnitude of the interaction forces, concentrated suspensions can develop a network structure that is highly ordered and exhibits a mechanical strength. Such a structure is responsible for a yield stress, which can be viewed as the minimum stress that must be applied to the suspension to initiate flow (van Wazer, 1963).  68  By applying a constant shear to a suspension with a structure, the suspension can show two types of time-dependent flow behavior: thixotropy in which the shear stress decreases over time and rheopexy in which the stress increases. 2.3.1.3 Description of the flow behavior of suspensions In a typical rheological measurement, a shear rate is applied to a suspension and the resulting stress is measured using a controlled rate rheometer. Alternatively, in a controlled stress test, a stress of known magnitude is applied and the resulting shear rate is measured. The raw data (torque vs. rate of shear) are usually recalculated to the well known ‘shear stress-shear rate’ flow curves (also called rheograms) from which important rheological parameters such as apparent viscosity and yield stress can be extracted. The simplest flow curve (line A, Figure 2.12) is that of a Newtonian liquid or suspension for which the viscosity is constant as described by Equation 2.34.  Figure 2.12. Illustration of the types of flow behavior of various solid-liquid suspensions. A - Newtonian; B – pseudoplastic or shear thinning; C – shear thickening; D – Bingham plastic; E – yield shear thinning; F – yield shear thickening; and τy is the yield stress.  However, for many industrial suspensions, the viscosity of a suspension varies with changing rate of shear resulting in non-Newtonian behavior characterized by curves B-F. The mathematical treatment 69  for non-Newtonian fluids for rotational and tube viscometers was described by van Wazer (1963). The curves (B and C) which start at the origin represent viscous fluids which flow no matter how small the shear. In the case of B (pseudoplastic or shear-thinning), an increase in the rate of shear results in a more ordered structure which reduces the apparent viscosity. Typical fluids or suspensions characterized by curve B are solutions of long chain high molecular weight polymers and coarse particle suspensions. Suspensions depicting the flow behavior such as that in curve C are said to be shear thickening. Such suspensions become more viscous with increase in shear rate. This behavior is often referred to as dilatancy. The next group of flow curves includes those that exhibit a yield stress (D-F). Of these, the simplest is the Bingham plastic flow curve (D). It can be seen that once the yield stress has been exceeded, indicating breakdown of the network structure, the material flows like a Newtonian fluid. Fluids or suspensions with flow behavior such as that described by curve E are yield shear thinning. At shear stresses higher than the yield stress, the fluid behaves as non-yielding shear thinning (curve B). Several models, most of which are empirical, have been developed to describe the different flow behaviors and have been well reviewed by Klein (1992). In this dissertation, the shear stress-shear rate data obtained experimentally were presented as depicted in Figure 2.12, and a comparison between flow curves obtained at different conditions allowed conclusions to be drawn about the aggregation/dispersion state of oil sand slurries.  2.3.2  Sedimentation  The stability of suspensions is affected by particle–particle interactions in the same way as the rheology, thus sedimentation tests provide a good way to confirm the extent of aggregation or dispersion when carried out in conjunction with rheological tests. In a typical sedimentation test, the suspension is allowed to settle under gravity in a measuring cylinder. Generally, aggregated (unstable) suspensions settle more rapidly than dispersed (stable) ones, except at very high solids contents where hindered settling takes place. Several parameters can be obtained that describe, qualitatively or quantitatively the extent of aggregation/dispersion. The 70  settling rate is the commonly measured parameter obtained by monitoring the rate of descent of the supernatant-suspension interface with time. The settled or sediment volume, measured after the suspension has settled for a long time provides information about the nature of the flocs. Aggregated suspensions generally produce soft and loosely parked sediments, usually large in volume and easy to redisperse. On the other hand, dispersed suspensions usually form closely packed, compact sediments that are hard to redisperse (Tadros et al., 1995). Quantitatively, it can be reported as a relative value i.e., Vrelative = Vsediment/ Vtotal. The percent weight settled also provides information about the stability of the suspension. It is obtained by withdrawing a certain amount of the suspension after a fixed amount of time and calculating the amount of solids settled within that time (Somasundaran et al., 2009). The turbidity of the supernatant, measured by light transmittance techniques is another parameter that is particularly useful in applications where the clarity of the supernatant is important. However, this parameter is only useful for relatively clear supernatants. Dispersed suspensions usually produce supernatants with turbidity beyond the range of the measuring instruments. Pawlik et al. (2003) reported a method to describe the dispersion state of such suspensions. The method involves performing a settling test for a fixed time and measuring the Dispersion Coefficient (DC) defined as the ratio of the density difference between the supernatant and the pure medium, to the density difference between the fully dispersed slurry and the pure medium. For a fully dispersed suspension, DC = 100% while a suspension in which all the solids settle leaving a clear supernatant has a DC of 0%. In this dissertation, the aggregation/dispersion state of oil sand slurries was determined through rheological and sedimentation tests. The combined use of the two methods is recommended for a thorough evaluation of particle-particle interactions in colloidal suspensions (Somasundaran et al., 2009).  71  CHAPTER 3  3 Viscometric characterization of carboxymethyl celluloses of different molecular weights and degrees of substitution 3.1 Introduction Carboxymethyl cellulose (CMC) is an anionic polysaccharide obtained from cellulose. The nontoxicity, biodegradability and biocompatibility of CMC make it one of the most important cellulose derivatives. CMC readily dissolves in water to form viscous solutions with a range of thickening, stabilizing and film-forming properties. These properties make CMCs attractive polymers for industrial and consumer applications, including textile (Stelzer and Klug, 1980), food (Gilbert, 1994), pharmaceutical (Wade and Weller, 1994), carbon nanotubes (Minami et al., 2006), mineral processing (Pugh, 1989) and the pulp and paper industry (Watanabe et al., 2004). In mineral processing, CMC is commonly used as a depressant of carbonate and talcaceous gangue minerals in the froth flotation of sulfide and Platinum Group Metal ores by adsorbing on the gangue minerals, rendering them hydrophilic and preventing their flotation (Leppinen et al. 2005). Another important application of CMC is in potash flotation as a “blinderˮ of waste fine water-insoluble minerals such as clays and dolomite. The role of CMC is to adsorb on the waste minerals and render them inaccessible to the approaching collector thereby allowing the collector to only adsorb on the valuable mineral particles. The effect of the solution physicochemical properties such as pH and ionic strength on the adsorption of polymers on gangue minerals and their depression is well known (Morris et al., 2002; Khraisheh et al., 2005; Beaussart et al., 2010; Hoogendam et al., 1998a). The effect of the DS and molecular weight of CMC on its depressant action has also been studied (Khraisheh et al., 2005; Steenberg and Harris, 1984; Shortridge et al., 2000; Mierczynska-Vasilev and Beattie, 2010). However, knowledge of the physical and chemical characteristics of CMC itself is still lacking despite its widespread applications. The functional properties of polymers are generally related to structural and physicochemical properties such as conformation, chain flexibility and molar mass distribution (Fleer et al., 1993), thus knowledge of such properties is required to adequately describe the mechanism of action of the polymers. A few studies published on the molecular properties of 72  CMC have dealt with persistence length (Hoogendam et al., 1998b) and conformation in aqueous solution (Brown and Henley, 1964; Kästner et al., 1997; Sitaramaiah and Goring, 1962). However, these studies did not systematically investigate the whole range of physicochemical parameters needed to elucidate the behavior of CMC in solution. For example, to the author’s knowledge, there was no systematic study done on the influence of pH and ionic strength on the conformation of CMC in distilled water and salt solutions. In this dissertation, the molecular conformation, chain flexibility, and molar mass distribution of a series of CMCs of different DS and molecular weights were determined by dilute solution viscometry and analytical ultracentrifugation to better understand the solution chemistry of CMC. 3.2 Materials and methods 3.2.1  Polymers and other chemicals  Three carboxymethyl cellulose samples with different degrees of substitution (DS = 0.7, 0.9 and 1.2, manufacturer’s data; designated hereon as DS0.7, DS0.9, DS1.2, respectively) but similar molecular weight (250,000 g/mol), and three samples of similar DS (0.7) but different molecular weights (80,000, 250,000 and 700,000 g/mol, manufacturer’s data, designated hereon as LM-CMC, MM-CMC, and HM-CMC, respectively) were evaluated. Since the glucose unit of cellulose has three hydroxyl groups available for substitution, the degree of substitution of CMC is typically defined as the average number of hydroxyl groups per glucose unit substituted by carboxylic groups, and the maximum degree of substitution of CMC is therefore 3 (Salmi, 1994). The actual molecular weights used in the data analysis were independently determined by sedimentation velocity using the analytical ultracentrifuge and are presented in section 3.3.1. The samples of different DS were obtained from Acros Organics while the samples of different molecular weights were supplied by Polysciences. The samples were received as dry solids and were used without any further treatment or purification. A schematic structure of carboxymethyl cellulose is shown in Figure 3.1.  73  Figure 3.1. Chemical structure of carboxymethyl cellulose showing a carboxymethyl-substituted unit in the C(6) position and two unsubstituted cellulose units. (Adapted from Coultate (1996)).  The samples for viscosity measurements were prepared by mixing an appropriate amount of CMC in de-ionized water or in 0.01 mol/L sodium chloride solution at room temperature to give concentrations of 1 g/L. In order to prepare a 1 g/L solution, 0.5 g of CMC was weighed in a small sample bottle. De-ionized water (~350 mL) was then added into a 600-mL beaker, and a magnetic stirring bar was used to create a deep vortex reaching nearly the bottom of the beaker. The CMC granules were slowly sprinkled into the sides of the vortex making sure that the granules did not form a gel but were individually dispersed in water. The exact amount of polymer added was determined by weighing the amount of CMC left in the sample bottle and subtracting it from the initial amount weighed. This small correction was used to calculate the exact concentration of the stock solution. The top of the beaker was then sealed with a parafilm to prevent contamination with dust. The CMC solution was then gently mixed for a further 10 hours using a magnetic stirrer at a constant low mixing setting to achieve complete dissolution of the polymer. The solution from the beaker was transferred to a 500-mL volumetric flask, and the beaker was washed with several small volumes (~ 50 mL) of either de-ionized water or 0.01 mol/L sodium chloride solution. Each wash solution was poured into the flask thus ensuring that the entire amount of CMC was transferred into the flask. In this way, the final volume was kept at 500 ml. Finally, the stock solution was allowed to stand overnight (~ 10 hours) to ensure complete hydration of CMC after which working solutions were prepared by dilution with the appropriate background electrolyte and tested within 2-4 hours after dilution. Reagent grade sodium chloride was used to adjust the ionic strength of the CMC solutions while small amounts of analytical grade sodium hydroxide and hydrochloric acid were used to adjust the pH of the working solutions. The pH range investigated (3–7) covered the range over which the 74  CMC carboxylic units are either fully protonated/associated (pH 3) or fully dissociated (pH 7), as shown by Hoogendam et al. (1998b) who tested CMC samples of different charge densities and molar masses and found that the degree of dissociation approached 1.0 (full dissociation) at pH 6 and remained constant up to pH 10. Recent viscosity measurements on a similarly charged polyelectrolyte (hydrolyzed polyacrylamide) showed that there is no significant polymer chain expansion above ~ pH 7 (Arinaitwe, 2008) with the viscosity behaviour in high alkaline solutions being influenced by sodium from NaOH used to adjust pH. All the reagents were obtained from Fisher Scientific. 3.2.2  Viscosity measurements  Dilute CMC solutions were prepared by mixing a desired volume of the stock solution in deionized water or in 0.01 mol/L NaCl. The kinematic viscosities of the CMC solutions in de-ionized water and 0.01 mol/L NaCl were measured using two Cannon-Fenske capillary viscometers (ShottGeräte, GmbH, Germany) of different calibration constants. The viscometers were calibrated by carrying out carefully controlled viscosity measurements with de-ionized water at 25 °C. When the determined kinematic viscosities deviated from accepted literature values (Weast, 1970), the viscometer constants were re-calculated accordingly. The corrections for the viscometer constants were on the order of 3% but such rather small adjustments were considered very important since small differences in kinematic viscosities can have a dramatic impact on the calculated specific viscosities in very dilute polymer solutions, especially since two different capillaries were used, which would then affect the accuracy of the extrapolated intrinsic viscosities. After dilution of the stock solution to the desired polymer concentration, 7-mL aliquots were transferred to the capillary viscometer, the viscometer was placed in a water bath whose temperature was kept constant at 25 ± 0.1 oC for an equilibration time of 30 minutes. The kinematic viscosity was then determined by allowing the CMC solution to flow down the capillary under gravity. A PVS1 Lauda photo-timing and processing system interfaced with a computer was used to automatically measure three flow times, an average of which was used to calculate the kinematic viscosity. The standard deviation of the flow times was always less than 1 second. In addition, three solutions were usually tested at a given concentration.  75  For high temperature measurements, with the capillary and sample still in the bath, the temperature of the bath was increased to 50 °C. The CMC solution was equilibrated at this temperature for 30 minutes before testing. The CMC solutions were tested in the concentration range from 10 mg/L to 100 mg/L. This range was chosen to satisfy the general critical concentration (c*) requirement for dilute solution viscometry (c* ≈ 1/[η]) for the onset of entanglement formation. The lower limit of the concentration range was arbitrarily chosen to avoid possible adsorption of the CMC macromolecules on the capillary walls during viscosity measurements. From the kinematic viscosities, the relative, specific and reduced viscosities of the CMC solutions were calculated at each concentration, pH value (approximately 3, 4.5, 7), and temperature (25 °C, 50 °C). Finally, the intrinsic viscosities were determined by extrapolation with the use of Fedors equation (see the results and discussion section). 3.2.3  Aging of CMC solutions  Aging studies were performed to investigate the solution stability of the CMC solutions in distilled water at 25 °C. Solutions for the aging tests were prepared and diluted in exactly the same manner as for the viscosity measurements except that the stock solutions were not allowed to equilibrate overnight after preparation. Only the CMC solution with the lowest degree of substitution (DS0.7, 150 mg/L) and highest degree of substitution (DS1.2, 100 mg/L) were investigated. The diluted solutions were transferred to a capillary viscometer and their viscosities were determined at 1 hour intervals over a period of 24 hours without removing the capillary from the bath. The measurements were started immediately following preparation of the stock solutions. 3.2.4  Sedimentation velocity analytical ultracentrifugation  Sedimentation velocity experiments were conducted using a Beckman Coulter ProteomeLab XL-I analytical ultracentrifuge equipped with absorbance and interference optics. Stock solutions of 1 g/L CMC were prepared by agitation of the solid CMC in distilled water for 12 h at room temperature. Aliquots of the stock solutions were diluted to the final concentration prior to analysis. In order to suppress the non-ideality arising from charge effects, a high ionic strength is recommended for analytical ultracentrifugation analysis of charged macromolecular systems such as CMC (Harding, 2005a). On the other hand, the sample concentration depends on the optical path length of the cell and the optical properties of the sample. A minimum concentration of 200 mg/L is recommended 76  (Harding, 2005b) for sedimentation velocity experiments on the XL-I analytical ultracentrifuge. However, this concentration is well above the critical overlap concentration (c* ≈ 1/[η]) above which CMC molecules start interacting. To accommodate the above requirements, an ionic strength of 0.1 mol/L NaCl and sample concentration of 100 mg/L were chosen in order to minimize hydrodynamic and thermodynamic non-ideality resulting from molecular co-exclusion and CMC polyelectrolyte behaviour. All experiments were run at 25 °C and pH 6. It should be noted that at a high electrolyte concentration, pH effects on polymer conformation are suppressed (see results and discussion) hence the choice of only one pH value. The diluted CMC solutions (460 µL) and solvent (0.1 mol/L NaCl, 450 µL) were injected into the sample and reference channels of the 12-mm double sector centerpieces (assembled with sapphire windows) and loaded into a 4-hole titanium (AnTi60) rotor. The slightly different solution and solvent volumes facilitate the process of locating the position of the solution meniscus at the top of the cell during data analysis. The samples were equilibrated at 25 °C for 1 hour prior to the sedimentation velocity run to minimize convection effects in the cells at the start of the sedimentation run. The rotor speed was 40,000 rpm for the higher molecular weight CMC polymers (MM-CMC and HM-CMC), and 45,000 rpm for the lower molecular weight polymer (LM-CMC). Rayleigh interference scans were acquired in the continuous mode and the resulting data (fringe displacement versus radial position) were analyzed by the least-squares regression method incorporated in the program SEDFIT (Schuck and Rossmanith, 2000). The method involves direct fitting of the data to the solution of the Lamm equation (Equation 2.32, section 2.2.3) to yield a sedimentation coefficient distribution, ls-g*(s). The ls-g*(s) model was run with the TikhonovPhillips 2nd derivative regularization at a confidence level of p = 0.95. The resolution of the ls-g*(s) distribution (number of points on the s-grid) was 200. The sedimentation coefficient distribution, ls-g*(s) was transformed into a molecular weight distribution, f(M) using the Extended Fujita method (Harding et al., 2011) for general conformation types of macromolecules. This method is an extension of the Fujita (1962) transformation of a differential distribution g(s) of sedimentation coefficients into a differential distribution f(M) of molecular weight for linear polymers with a random coil conformation i.e. with b = 0.5 in the sedimentation coefficient – molecular weight relation, s = ks MWb, where ks is a constant for a given polymer/solvent/temperature system, and b (conformation of the polymer) ranges from 0.4 to 5 for a coil, ~0.15 to 0.2 for a rod and ~0.67 for a sphere (Kulicke and Clasen, 2004). A more detailed 77  review of the method is given in section 2.2.3, Chapter 2, and in Appendix 1. The method is built into the sedimentation velocity analysis program SEDFIT (Schuck, 2000, Brown and Schuck, 2006) and was recently applied to polysaccharides and other polymeric systems (Harding et al., 2011; Gillis et al., 2012). Values of b and ks used in the transformation were obtained by linear regression of a plot of log s versus log MW (Appendix 2) using published s and MW data (Brown and Henley, 1964). These authors characterized CMC samples of DS 1.06, with molecular weights ranging from 147,000 g mol-1 to 1,060,000 g mol-1 under different physicochemical conditions. The s-MW data obtained under conditions closest to those of the present dissertation were chosen for use in the transformation. The CMC partial specific volume (0.565 mL/g) used during the fitting process was that determined by Sitaramaiah and Goring (1962) using CMC samples with intrinsic viscosities and sedimentation coefficients similar to those used in this research. 3.3 Results and discussion 3.3.1  Molecular weight and molecular weight distribution  Sedimentation coefficient distributions were obtained from SV data following the method described by Schuck and Rossmanith (2000). The method involves direct linear boundary modelling of the data to obtain an apparent sedimentation coefficient distribution, ls-g*(s). The method was shown to produce precise sedimentation coefficients from SV experiments, exhibits a high resolution and is particularly useful for analysis of SV data of large polymers and samples with a high degree of heterogeneity (Schuck and Rossmanith , 2000; Harding et al., 2011). The SV data were analyzed with the program SEDFIT using the ls-g*(s) model. The analyzed scans spanned the entire sedimentation process from the initial depletion of the meniscus at the start of the analysis to the completion of the edge of the sedimentation boundary at the bottom of the cell in order to encompass the sedimentation process of all sedimenting macromolecules. The fitting limits were chosen to exclude any optical artifacts near the meniscus and bottom regions. Where necessary, the meniscus position was floated in the software to improve the quality of the data fit as determined from the root mean square deviation, rsmd. Generally, rsmd values closer to zero indicate fits of excellent quality. The obtained rsmd values were of optimal quality especially for LM-CMC (rsmd = 0.0054) while those of the MM-CMC and HM-CMC were slightly higher at  78  0.0074 and 0.0089, but are considered relatively good considering the heterogeneous nature (composition and molecular weight) of the samples. The resulting sedimentation distribution functions for LM-CMC, MM-CMC and HM-CMC are shown in Figure 3.2. Examples of screen shots of the actual data analysis windows are shown later in Figure 3.3 (for LM-CMC) and in Appendix 3 (MM-CMC) and Appendix 4 (HM-CMC). A few general observations can be noted in Figure 3.2. As expected, and in accordance with the Svedberg equation (Equation 2.31), the high molecular weight sample sediments faster based on the sedimentation coefficient value, S at the peak (shown on the graphs). This is attributed to the large size of HM-CMC compared to the lower molecular weight polymers which sediment at much smaller S values. Similar sedimentation coefficients were obtained by Brown and Henley (1964) for CMC of comparable molecular weights. All the samples show a single major peak but at different sedimentation coefficients. The ls-g*(s) distributions, especially for the higher molecular weight polymer, show several minor but broad peaks consistent with the polydisperse nature of the samples. All the samples show a small fraction sedimenting at higher S values outside of the main peak, but the higher MW samples show more small peaks over a broader span of S values. The range of S is 0 – 30 S for LM-CMC, 0 – 100 S for MM-CMC and 0 – 150 S for HM-CMC. The limit of S was determined as the highest value at which the distribution became zero indicating that no more material was sedimenting. By integrating the areas under the peaks, an estimate of the weight average sedimentation coefficient, Sav and the loading concentration could be obtained, thus revealing information about the amount of material sedimenting at a particular sedimentation coefficient. For example, for LM-CMC, 90% of the polymer sediments at 3.6 S but 10% sediments at 18.3 S for a total Sav of 5.1. At higher molecular weights, the same percentages sediment at much higher S values. Thus for MM-CMC, 90% sediments at 5.7 S and 10% sediments at 67 S (total Sav ≈ 11.7), while for HM-CMC, 90% sediments at 10.7 S and 10% sediments at 112 S (total Sav ≈ 21).  79  Figure 3.2. Sedimentation coefficient distributions ls-g*(s) vs. s for LM-CMC, MM-CMC and HM-CMC, DS 0.7. Sample concentration was 0.1 mg/mL in 0.1 mol/L NaCl, pH 6. The samples were centrifuged at 45,000 rpm (LM-CMC) and 40,000 rpm (MM-CMC and HM-CMC) at 25 °C. The S values indicated at the peaks correspond to the sedimentation coefficient at which most of the material in the sample sediments.  The ls-g*(s) distributions in Figure 3.2 were transformed into a molecular weight distribution by following the Extended Fujita method (Harding et al., 2011). The transformation was performed by switching to the f(M) vs. M model in SEDFIT and utilizing prior data in the form of the s-MW scaling law as described before. In all cases, the rsmd values of the fits (indicating the quality of the fits) were very similar to those of the ls-g*(s) fits indicating that the resulting f(M) vs. M distributions 80  quite accurately described the raw sedimentation data. The weight-average and z-average molecular weights could be extracted from the data by integrating the area under the peaks, however, the f(M) vs. M distributions showed a broad range of small peaks at very high molecular weights which made it difficult to determine the cut-off point at which to integrate the distribution. In order to determine the cut-off for molecular weight in the f(M) distribution, an estimate of the molecular weight corresponding to the highest peak obtained in the ls-g*(s) distribution was obtained from the corresponding c(s) distribution, since in the c(s) distribution, the program SEDFIT can provide information about the molecular weights of the observed peaks by utilising the S values and a weight-average frictional ratio of the sedimenting macromolecules. As described before, the c(s) distribution assumes one frictional ratio for the entire distribution, which is not entirely accurate for polydisperse polysaccharides, but nonetheless can provide a reliable distribution of sedimentation coefficients for slow-diffusing large macromolecules, which enables an estimate of the maximum molecular weight to be made. To obtain the c(s) distributions, the SV data were analyzed with the solutions of the Lamm equation (Equation 2.32, section 2.2.3) with maximum entropy regularization at a confidence level P = 0.95 and with an estimate for the frictional ratio of 2.0. The frictional ratio can be taken as a measure of the extent to which the shape of a macromolecule deviates from that of a compact sphere of the same mass and density. The estimate for CMC was deemed appropriate considering the coiled conformation of the CMC macromolecules in 0.1 mol/L NaCl. By comparison, the frictional ratio of a globular protein is 1.2 – 1.4 while that of an elongated protein ranges from 2 to 3. The rsmd values for the fits obtained with the frictional ratio estimate were similar or better than those obtained from the ls-g*(s) fits confirming that the estimate fairly described the frictional properties of the macromolecules. The resulting c(s) distributions exhibit a single major peak at the conditions and fitting parameters chosen. The peaks in the c(s) distribution are slightly narrower than those obtained in the ls-g*(s) distributions but the S value at the peak does not change much. A similar trend was observed in the comparison of ls-g*(s) and c(s) distributions for guar gum (Patel et al., 2006). Thus, although the assumption of an average frictional ratio may not be optimal for CMC, the fits to the SV data of the tested CMC samples and the similarity of the ls-g*(s) and c(s) distributions confirm the use of the c(s) distribution to be an appropriate strategy for estimating the average molecular weights of the peaks obtained in the ls-g*(s) distributions.  81  Figures 3.3 A, B and C show the ls-g*(s), c(s) and f(M) distributions obtained for LM-CMC. Similar analysis was done for MM-CMC and HM-CMC (see ls-g*(s), c(s) and f(M) screenshots in Appendices 3 and 4). Generally, it should be recalled that c(s) distributions do not account for the heterogeneity in the samples and the obtained distributions are dependent on the data selected for analysis as well as the fitting limits chosen. Consequently, this is reflected in the value, but not in the distribution of the molecular weight obtained from the f(M) distribution. At the selected fitting conditions for the LM-CMC, the c(s) distribution shows a small secondary peak within the S range of the major peak in the ls-g*(s) distribution. This is not seen in the c(s) distributions of the other CMC polymers. The quality of the fit is excellent (rsmd = 0.0040) thus the peak most likely represents true sedimenting macromolecules at the particular fitting conditions. However, it should be noted that the second peak disappears when the regularization level is increased by increasing the confidence level P from 0.95 to 1.1. Thus it appears that in the c(s) distribution for LM-CMC, a lower level of regularization (P = 0.95) produces more details by splitting the single major peak observed in the ls-g*(s) distribution. Brown and Schuck (2006) analyzed the sedimentation velocity data of a glycosylated natural killer receptor filament of MW ~ 50,000 g/mol and observed a similar effect of regularization on its size distributions. The difference in quality of fits of the c(s) distributions at the two levels of regularization is insignificant and the detailed information at lower regularization does not significantly alter the main information required from the c(s) distribution, thus the gain in detail at higher regularization is not warranted by the data (Brown and Schuck, 2006).  82  A  Figure 3.3 A. The ls-g*(s) vs. s distribution for LM-CMC. The test conditions are the same as indicated in Figure 3.2.  83  B  Figure 3.3 B. The c(s) distribution showing the molecular weight information of the peaks in the lsg*(s).distribution (Figure 3.3 A) for LM-CMC.  84  C  Figure 3.3 C Transformation of the ls-g*(s) vs. s distribution (Figure 3.3 A) to a molecular weight distribution f(M) versus M for LM-CMC. The f(M) distribution was integrated up to the maximum MW shown in c(s) distribution (Figure 3.3 B). The test conditions are the same as indicated in Figure 3.2.  The resulting molecular weight distributions are plotted in Figure 3.4A. The same analysis as described above was done for DS0.7, DS0.9 and DS1.2 and the molecular weight distributions are plotted in Figure 3.4B. The data is conveniently plotted to show the logarithmic spread of the MW. In order to do that, the x-axis scale is plotted as a log scale, however the y-axis cannot simply be plotted on a log scale since it is a distribution i.e., f(M). dM represents the moles of the polymer with MW between M and M+dM and thus the y-axis should represent polymer chains with logMW, between logM and logM+dlogM. To achieve this, the y-axis should be plotted as f(M)*MW as in Figure 3.4. A comparison of the manufacturer’s molecular weights and those obtained using the 85  Extended Fujita approach is provided in Table 3.1. Also shown in Table 3.1 are the polydispersity indices (PDI) calculated as the ratio of MZ to MW. The MZ values were also obtained by integration of the f(M) distributions.  Figure 3.4. Molecular weight distributions f(M) versus MW for LM-CMC, MM-CMC and HM-CMC (DS 0.7) (Graph A) and for DS0.7, DS0.9 and DS1.2 (Graph B) obtained by transformation of the ls-g*(s) vs. s distribution using the Extended Fujita approach. The scaling law constants b = 0.29 and ks = 0.0101 were used in the transformation. Weight average molecular weights obtained by integration of the areas under the distributions are indicated.  Table 3.1. Molecular weights [g/mol] and polydispersities of the tested polymers.  MW (Manufacturer’s data)  Weight-average MW (Extended Fujita method)  PDI  LM-CMC  80,000  123,000  2.27  MM-CMC  250,000  285,000  2.74  HM-CMC  700,000  715,000  4.30  DS0.7  250,000  295,000  4.42  DS0.9  250,000  310,000  3.99  DS1.2  250,000  293,000  4.52  Polymer  86  It is evident in Figure 3.4A that the width of molecular weight distributions increases in the order LM-CMC<MM-CMC<HM-CMC. Quantitatively, this is shown by the increasing polydispersity index with molecular weight in Table 3.1. The central portion of the distribution is sensitive to the overall weight-average molecular weight which is shown by the shift of the major peak in the distributions to higher molecular weights as the molecular weight increases. The shift in the major peak from the LM-CMC (peak at ~ 87,000 g/mol, total MW = 123,000 g/mol) to the MM-CMC (peak at ~ 184,000 g/mol, total MW = 285,000 g/mol) is more pronounced than that observed for a further increase in molecular weight to HM-CMC (peak at 274,000 g/mol, total MW = 715,000 g/mol). Clearly, the small high molecular weight peaks seen in the distribution tail for HM-CMC account more for the polymer’s high molecular weight than the chains around the major peak. In all the distributions, the obtained weight-average MW lies well beyond the peak. There is no significant difference in the molecular weight distributions of DS0.7, DS0.9, DS1.2 which results in quite similar polydispersity index values as can be seen in Table 3.1. Generally, the Extended Fujita approach produces fair estimates of the molecular weights provided by the manufacturer. Surprisingly, the estimate for the higher molecular weight, more polydisperse sample (HM-CMC) is much closer to the manufacturer’s value than the estimate for LM-CMC and MM-CMC (less polydisperse samples). However, it should be mentioned that the method assumes that translational diffusion effects are negligible, which may be true for the larger HM-CMC coils but not entirely accurate for the smaller LM-CMC. The disparity in the size of the polymer coil between LM-CMC and HM-CMC can be deduced from viscosity measurements performed in 0.01 mol/L NaCl at pH 7 (presented in subsequent sections) which indicate that the intrinsic viscosity of HM-CMC is on the order of four times that of LM-CMC, while that of MMCMC is about twice that of LM-CMC. In addition, the method is primarily intended for large polydisperse macromolecules (Harding et al., 2011) and it seems from the present results that HMCMC, and to some extent, MM-CMC are more suited to the method than the lower molecular weight sample. It can also be seen that the obtained molecular weight values for DS0.7, DS0.9 and DS1.2 are quite similar which confirms that they are indeed of the same molecular weight. From the polymer adsorption point of view, it is pertinent to note that differences in the molecular weight and molecular weight distributions of polymers have implications to the adsorption behaviour of the polymers. It is generally agreed that higher molecular weight polymers show higher 87  adsorption than lower molecular weight polymers (see Khraisheh et al., 2005; Chibowski, 1990, for example). However, it was also demonstrated that when a polydisperse sample adsorbs on a surface, the higher molecular weight fractions will be preferentially adsorbed at the expense of the lower molecular weight fractions (Hlady et al., 1982). On this account, it would be expected that the differences in the molecular weights and polydispersities of the tested polymers can be utilized in understanding the dispersing action of the polymers.  3.3.2  Intrinsic viscosities  3.3.2.1 Remarks on the aggregation state of CMC solutions One of the main aims of the derivatization of native cellulose is to improve its solubility. The poor solubility of native cellulose arises from the strong hydrogen bonding within the cellulose crystalline structure. It can be expected that full derivatization results in complete solubility of cellulose to produce molecularly dispersed solutions. Partial derivatization, as is often the case for most cellulose derivatives including CMC, leaves unsubstituted OH groups that can interact by intra- and inter-molecular hydrogen bonding thereby limiting the solubility. According to Schulz et al. (1998) and Schulz and Burchard (1993), the specific interaction between OH groups in partially substituted celluloses results in formation of aggregates that can be significantly affected by the solvent. The literature on the aggregation of CMC is contradictory. Schulz and Burchard (1993) determined the exponent v in the scaling law relating the radius of gyration Rg and MW (Rg = KRGMWv). From the low value obtained in 0.1 mol/L NaCl (v = 0.28) and measurements of the hydrodynamic radius, they concluded that the CMC solutions contained aggregates. The exponent v is sensitive to the presence of aggregates and can be taken as a good measure of the solution structure. However, Hoogendam et al. (1998b) obtained a value of 0.59 in 0.1 mol/L KOH and NaNO3 indicating that CMC contained very little or no aggregates under those conditions. Clasen and Kulicke (2001) determined the value of v for CMC from Rg-MW data in 0.1 mol/L NaCl and found it to be 0.53. The authors also reported a literature value of 0.70 in 0.1 mol/L NaCl. From their data, CMC may be considered as a linear polymer in a good solvent, in contrast to Schulz and Burchard’s result.  88  The exponent v can be correlated to the exponent α in the [η]-MW relationship (Equation 2.19) through the following equation (Kulicke and Clasen, 2004): ..[3.1] Using the values of exponent α determined in this work for CMC in 0.01M NaCl (Figure 3.11, section 3.3.3.6), the exponent v is found to range from 0.57 to 0.62 depending on the pH of the solution, indicating good solvent conditions. The data also agree well with values from Clasen and Kulicke (2001) and Hoogendam et al. (1998b). It is also worth noting that all intrinsic viscosity data presented in this work were obtained using quite dilute stock and working solutions prepared carefully by gentle mixing in water over a long period of time. These conditions favor the dissolution of the polymer, and though the presence of aggregates in the tested solution cannot be entirely ruled out, it can be concluded that no or very little aggregates were present in the tested solutions, and that the ensuing discussion of the data will be based on the assumption that the presence of any aggregates did not significantly affect the experimental data. This is in line with the assumption that since aggregates behave as compact spheres, any contribution of aggregates to the intrinsic viscosity would be negligible (Kratochvíl, 1972).  3.3.2.2 Aging of CMC solutions Figure 3.5 shows the effect of time on the reduced viscosities of 150 mg/L DS0.7 and 100 mg/L DS1.2 solutions in distilled water. The highest concentrations tested in all the viscosity measurements were chosen for the aging tests. In general, the two samples are stable over the time scale of the test (hourly for 24 hours). However, a closer look at the data indicates that the rate of decrease of the reduced viscosity with time increases slightly with DS. The decrease of the reduced viscosity for sample DS0.7 amounted to only 0.15% over the 24-hour period while that of DS1.2 amounted to 2.5%. The slightly higher viscosity loss in the highest DS sample can be attributed to a small polymer conformational change from a more extended conformation to a coiled one. Similar trends were reported by Kulicke and Kniewske (1981), Arinaitwe and Pawlik (2013) and Arinaitwe (2008) who observed a viscosity loss in aqueous solutions of anionic hydrolyzed polyacrylamide.  89  150 DS = 0.7, c = 150 mg/L, T = 25oC  Reduced Viscosity [dL/g]  125 100  DS = 1.2, c = 100 mg/L, T = 25oC  75 50 25 0 0  5  10 15 Time [hrs]  20  25  Figure 3.5. Reduced viscosities of DS0.7 and DS1.2 as a function of time in distilled water at 25 °C.  It should be noted that the viscosity loss observed by these authors was significantly higher than that observed in the present study. The viscosity loss was attributed to a conformational change from more extended configuration to a coiled one over the time scale of the test. Kulicke and Kniewske indicated that the conformational change involved intra-molecular hydrogen bonding while Arinaitwe did not observe a significant viscosity loss in 0.01 mol/L NaCl due to the screening of charged carboxylic groups by sodium ions that resulted in a coiled conformation that was already stable against further coiling. Thus, the viscosity loss in distilled water was postulated to involve the hydrolysis of the weakly acidic carboxylic groups releasing hydroxyl ions into solution that would have the same effect on conformation as sodium ions. Vink (1970) also noticed a slightly higher viscosity loss for a more substituted CMC, but the addition of 10-5 mol/L NaOH inhibited a further decrease. Overall, the effect of aging on the tested CMCs can be considered negligible and most importantly the small aging effect in DS0.7 and DS1.2 does not seem to involve polymer degradation but rather a small change in the configuration of the polymers. Nonetheless, strict  90  experimental protocols involving similar sample preparation procedures and times, dilution steps and testing times were followed. It should also be stressed that all viscometric measurements were performed in very dilute solutions with relative viscosities not exceeding 2.0 in accordance with Kulicke and Clasen’s (2004) recommendation for relative viscosity limits needed to obtain accurate data. In such dilute solutions, polymer-solvent interactions, and not polymer-polymer chain interactions, determine the flow properties of the polymer solutions, thus the observed trends in the intrinsic viscosities result from changes in the conformation of single macromolecular chains. 3.3.2.3 Determination of intrinsic viscosities The viscosity properties of the CMC solutions can generally be represented by a power series in concentration, c, (similar to Equation 2.10) as follows:  red     k1 2 c  k2  3 c 2  .......  ..[3.2]  where [η] is the intrinsic viscosity, c is the polymer concentration, ki are constants, ηred is the reduced viscosity. The intrinsic viscosity of dilute nonionic polymers can routinely be determined graphically by truncating Equation 3.2 to a linear approximation (the first two terms on the right side of Equation 3.2), and plotting ηred versus the polymer concentration, c which allows evaluation of [η] from the intercept. However, the tested CMC solutions showed typical polyelectrolyte behaviour and thus could not be described by the procedure for nonionic polymers. Previous results by Arinaitwe and Pawlik (2013) and Arinaitwe (2008) on the intrinsic viscosity of anionic polyacrylamide showed that the Fedors equation (Equation 2.17: [2(ηrel1/2 -1)]-1 = ([η]c)  -1  - ([η]cm) -1) reliably described the  raw data by plotting [2(ηrel1/2 -1)]-1 against 1/c, and therefore the equation was applied to anionic CMC. The intrinsic viscosity could then be determined from the slope, 1/[η], of the plot. It should be noted that very good fits to the experimental data were obtained at all experimental conditions (ionic strength, temperature and pH), and most importantly, the equation allowed the intrinsic viscosities to be determined without having to apply the iso-ionic dilution technique (Pals and Hermans, 1948) almost universally used for polyelectrolyte solutions in order to minimize the polyelectrolyte effect especially in low ionic strength solutions.  91  Figure 3.6 shows an example of the concentration dependence of the reduced viscosity for DS1.2 in distilled water at different pH. Similar plots were obtained for the other polymers (Appendix 5). The errors bars shown at the lowest concentration tested (10 mg/L) generally reflect the absolute experimental errors in determining the reduced viscosities. The reduced viscosity at the lowest concentration in distilled water at natural pH was always the most difficult to reproduce. For DS1.2, the highest absolute error in obtaining the reduced viscosity was ~ 6.7 dL/g corresponding to an error in the intrinsic viscosity of ~ 9.1 dL/g.  Figure 3.6. Reduced viscosity vs. polymer concentration for DS1.2 in distilled water, 25°C at different pH.  The reduced viscosity increases with decreasing polymer concentration as typically seen for polyelectrolytes. Since the degree of dissociation of the weakly acidic carboxylic groups increases at lower polymer concentrations, the increase in ηred as concentration decreases is attributed to the gradual expansion of the polymer coil brought about by increasing repulsion between the dissociated carboxylate groups. Figure 3.7 illustrates Fedors representation of the data from Figure 3.6 in which [2(ηrel1/2 -1)]-1 is plotted against 1/c to produce linear fits whose reciprocal slopes give the intrinsic viscosities. It is interesting to note that the intrinsic viscosities determined by fitting a second-order polynomial equation (essentially the first three terms in Equation 3.2) to the experimental data, as 92  shown by the solid curves in Figure 3.6, gave very similar intrinsic viscosities as those determined using the Fedors equation. This provided a good cross-check of the intrinsic viscosities obtained using the Fedors equation. A comparison of the intrinsic viscosities determined from fits using the second-order polynomial and Fedors equations can be found in Appendix 6.  Figure 3.7. Fedors representation of the data from Figure 3.6 for DS1.2 in distilled water at 25 °C and different pH.  3.3.3.4 Effect of pH and ionic strength Figure 3.8 shows the intrinsic viscosities for all the CMCs in distilled water and in 0.01 mol/L NaCl at different DS. In distilled water, as expected, an increase in the DS from 0.7 to 1.2 brings about a steady increase in the intrinsic viscosity of the CMCs resulting from the increasing content of carboxylic groups that cause stronger repulsion between the polymer chains. The increased repulsion further stretches the chains resulting in an increase in intrinsic viscosity. Compared to the effect of pH, temperature does not seem to have any significant effect on the intrinsic viscosities in distilled water. Under all experimental conditions, the differences between the results at 25 °C and  93  50 °C differ utmost by 3dL/g which is well within the experimental error. As plotted, the differences in the curves at different pH values are quite evident compared to those at different temperatures.  Figure 3.8. Effect of pH and temperature on the intrinsic viscosities of the CMCs of different DS. MW = 293,000 – 310,000 g/mol. A: Distilled water; B: 0.01 mol/L NaCl.  At all degrees of substitution, the effect of pH in distilled water is apparently quite strong. Increasing the pH from 3 to 4.5 results in a large increase in [η] and a further increase is noted as the pH is increased to 7. This trend is consistent with the degree of dissociation of CMC which increases from pH 3 to ~pH 5 and steadily increases to a maximum around pH 6-7 depending on the ionic strength (Hoogendam et al., 1998b). The natural pH of the tested polymer solutions was about 6 indicating that they were already highly dissociated. The increasing dissociation of the carboxylic groups induces strong electrostatic repulsions between the groups and therefore produces an increasingly stretched conformation of higher viscosity at higher pH. At pH values higher than 7, the carboxylic groups are completely dissociated and further increase in pH should not induce additional repulsion between the carboxylic groups, thus the effect of pH on the intrinsic viscosity above pH 7 should be correspondingly weak. For this reason, higher pH values were not tested.  94  In Figure 3.8B, the effect of ionic strength is shown by plotting intrinsic viscosities in 0.01 mol/L alongside those in distilled water, keeping the same scale to highlight the impact of NaCl. The increase in ionic strength through the introduction of sodium counterions in solution shields the intramolecular electrostatic repulsive forces, which leads to coiling of the polymer chains, and a dramatic decrease in the intrinsic viscosities.  Figure 3.9. Intrinsic viscosities of DS0.7 and DS1.2 at pH 7 and 25 °C as a function of ionic strength. Constant MW of 295,000 g/mol (DS0.7) and 293,000 g/mol (DS1.2).  Figure 3.9 shows the effect of ionic strength on the intrinsic viscosities of DS0.7 and DS1.2. The strong impact of sodium ions in suppressing the electrostatic interactions between the polymer coils can clearly be seen. The data in Figures 3.8 and 3.9 allow the effect of pH on the intrinsic viscosity to be de-coupled from the effect of ionic strength. It should be noted that pH 3 in distilled water is equivalent to an ionic strength of 10-3 mol/L, and as Figure 3.8A shows the intrinsic viscosity of DS1.2 decreases to about 30 dL/g at pH 3 from nearly 120 dL/g at pH 7. If, however, 10-3 mol/L sodium chloride is used as the background electrolyte at pH 7, the intrinsic viscosity of DS1.2 also decreases to about 44 dL/g as seen in Figure 3.9. The difference between the value at pH 3 in 95  distilled water and the value at pH 7 in 10-3 mol/L sodium chloride can then be attributed to pH effects. In other words, 10-3 mol/L hydrochloride acid (HCl) has almost the same effect on the intrinsic viscosity as 10-3 mol/L sodium chloride (NaCl), which strongly suggests that the apparent effect of pH in Figure 3.8A is primarily caused by changes in the ionic strength of CMC solutions, and to a smaller extent by association/dissociation of carboxylic groups. The same trends can be observed for the DS0.7 sample. The intrinsic viscosities of DS0.7 and DS1.2 decrease by almost 6and 4-fold respectively with the addition of 0.01 mol/L NaCl. Very interestingly, the strong effect of pH observed in distilled water completely disappears as evidenced by comparing Figures 3.8A and B. It is clear that once the CMC polymer chains are coiled by addition of 0.01 mol/L NaCl, increasing the pH from 3 to 7 does not fully uncoil the chains. Thus, it is the presence of background counterions, not the pH that seems to determine the conformation of CMC over the studied pH range. It should also be noted that the effect of temperature remains very weak under all conditions implying that solvency effects are not significant in the tested CMC solutions. Along the same lines, Figure 3.10A shows that in distilled water, at any pH, an increase in the molecular weight of CMC results in an increase in the intrinsic viscosity. When the molecular weight of a polymer is increased, the effective size of the polymer coil increases which is reflected in the increasing intrinsic viscosity at a given pH value. However, pH seems to have an effect on this expansion. At a low pH value of 3, an increase in the molecular weight results in a moderate increase in intrinsic viscosity, but quite a sharp increase is seen at higher pH values (4.5 and 7). The curves obtained at higher pH values are steeper than at pH 3. This effect can be explained by ionic strength effects and association of the polymer at low pH. Adding HCl to adjust the natural pH of the CMC solutions (~pH 6) to pH 3 increases the ionic strength of the solutions which limits stretching of the chain with increasing molecular weight, thus the intrinsic viscosities obtained for LM-CMC, MMCMC, and HM-CMC at pH 3 do not differ much. Moreover, the anionic CMC macromolecules are associated at low pH, and an increase in the molecular weight does not seem to lead to a corresponding change in intrinsic viscosity as seen in the small difference in the intrinsic viscosity between LM-CMC and HM-CMC. On the other hand, at pH 7 close to the natural pH, CMC is fully dissociated and there are few counterions in solution to shield negatively charged carboxylic group. As a result, expansion of CMC due to increasing molecular weight is much more pronounced as can be seen from the intrinsic 96  viscosity difference between LM-CMC and HM-CMC. Overall, especially at higher pH values, increasing the polymer coil size (increasing MW) affects the conformation of the polymers more than increasing the number of ionic groups on the polymer chain (DS) at a given molecular weight.  Figure 3.10. Effect of pH and temperature on the intrinsic viscosities of the tested CMCs of different MW (DS = 0.7). A: distilled water; B: 0.01mol/L NaCl.  Figure 3.10B shows that adding 0.01 mol/L NaCl has a similar effect to that seen in Figure 3.8 – the sharp increase in [η] with MW observed in distilled water again disappears with addition of salt. The absolute values of [η] decrease due to the screening of the carboxylic groups and therefore coiling of the CMC macromolecules. However, even in 0.01 mol/L NaCl, there is still a significant increase of [η] with MW, compared to the smaller increase of [η] with DS (Figure 3.8B) in salt solution again indicating the more significant role of MW over DS in determining the conformation of CMC.  97  3.3.3.5 Expansion factor of CMC The polyelectrolyte expansion can conveniently be expressed in terms of the expansion factor αE, given for cellulose derivatives by (Pals and Hermans, 1952): ..[3.3] where  is the intrinsic viscosity in distilled water and  ionic strength. The values of  is the intrinsic viscosity at infinite  can be determined graphically from the relation of  with  ionic strength (Pals and Hermans, 1952) given by: ..[3.4] where I is the ionic strength. It can be seen that as I → ∞, axis of a plot of  ~ A, thus A, the intercept with the y-  versus I-1/2 can be taken as the intrinsic viscosity at infinite ionic strength.  Replotting the data from Figure 3.9 according to Equation 3.4 (see Appendix 7) for DS0.7 and DS1.2 gives  as 15.6 dL/g and 26.1 dL/g respectively (only DS0.7 and DS1.2 were tested at  different NaCl concentrations). By comparison, the intrinsic viscosities  , for the DS0.7 and  DS1.2 in 0.01 mol/L NaCl are 16.8 dL/g and 29.2 dL/g respectively which indicates that 0.01 mol/L NaCl is sufficient to bring coiling of CMC to intrinsic viscosity values close to those at infinite ionic strength. This comparison also indicates that the expansion factors estimated using Equation 3.3 instead of  in  should quite closely describe the expansion of the CMC  macromolecules. Thus, the expansion factors presented in Table 3.2 were obtained as square roots of the ratios of the intrinsic viscosity in distilled water ( NaCl  to the intrinsic viscosity in 0.01 mol/L  (Equation 3.5). ..[3.5]  Table 3.2 shows values of αE, evaluated from Equation 3.5. A few trends can be noted in the table. Generally, the expansion factors increase with pH at any DS or MW. However, the apparent effect of pH on chain expansion needs to be viewed with caution since it is more likely related to the different ionic strengths of the solutions as pH is increased from 3 to 7. As discussed earlier, the ionic strength at pH 3 in distilled water produces nearly the same effect on the intrinsic viscosity as a 98  10-3 mol/L NaCl solution. Since the determination of the expansion factor involves taking a ratio of the intrinsic viscosity in distilled water to the intrinsic viscosity in 0.01 mol/L NaCl, the αE values at pH 3 would be lower than the values at pH 7 primarily due to the stronger effect of ionic strength on the intrinsic viscosities at pH 3 in distilled water compared to the weaker effect of ionic strength on the intrinsic viscosity at pH 7 in distilled water. However, the effect of DS and MW on the chain expansion can be seen by comparing the expansion factors at constant pH thus ensuring a constant ionic strength. Table 3.2. Expansion factors of CMC of different DS and MW at different pH Polymer  αE pH 3  pH 4.5  pH 7  DS0.7  1.15  2.23  2.39  DS0.9  1.24  1.99  DS1.2  1.27  1.93  Polymer  αE pH 3  pH 4.5  pH 7  LM-CMC  1.87  2.53  2.86  2.28  MM-CMC  1.28  2.06  1.90  2.19  HM-CMC  1.17  1.72  1.92  The effect of DS on the chain expansion is not so pronounced as that of the molecular weight. The CMC with the lowest molecular weight has the highest ability to expand while the highest molecular weight polymer generally exhibits lower expansion factors. It should be noted that the expansion factor can also be looked at as the ability of the polymer chains to coil or stretch in solution upon transition from low to high ionic strength solutions. Thus, the low expansion factors of HM-CMC are probably due to the increased stiffness within the polymer chain resulting from intrachain hydrogen bonding. The increased rigidity limits the ability of the higher molecular weight polymer (and to some extent the higher DS polymer) to coil or stretch compared to the more flexible lower molecular weight and lower DS polymers.  3.3.3.6 [η] – MW relationship Figure 3.11 shows the relation between the intrinsic viscosity and weight-average molecular weight for CMC samples LM-CMC, MM-CMC, and HM-CMC with DS 0.7. A linear relation is observed in the molecular weight range tested in accordance with the scaling law [η] = KMWα (Equation 2.19). As already indicated, the exponent α in the scaling law contains information about 99  the solvent quality and the structural properties of the polymer coil in solution (Kulicke and Clasen, 2004). The high values of α indicate that 0.01 mol/L NaCl is a good solvent for CMC and that CMC assumes a randomly coiling, semi-flexible conformation under the tested conditions.  Figure 3.11. Logarithmic dependence of the intrinsic viscosity [η] as a function of molecular weight for CMC (MW range 123,000 – 715,000 g/mol) in 0.01 mol/L NaCl at 25 °C and different pH. The exponent α values indicated with an error estimate are obtained by using an average value of duplicate intrinsic viscosities for the lowest molecular weight polymer. The error bars on the data points correspond to the lowest and highest error obtained in log([η]).  The magnitude of α is generally independent of pH indicating the relatively stable conformation of CMC in salt solution. In the study by Eremeeva and Bykova (1998), the exponent α for low molecular weight CMCs (MW = 8,000 – 92,000 g/mol) was found to decrease from 0.83 in a 0.4 mol/L acetate buffer solution at pH 5, to 0.73 in a strongly alkaline solution of 0.5 mol/L NaOH (pH ~ 13.7). In the present study, there is no clear correlation of the exponent α with pH, with an average value of 0.83. It is difficult to directly compare published α values to the ones determined in the present study due to discrepancies arising from differences in sample sources (commercial, as in this study vs. well 100  fractionated homogenous samples) and different experimental conditions such as pH and ionic strength. Also important is the range of the molecular weights used in the studies since α values are only valid for a limited range of molecular weights (Kulicke and Clasen, 2004). All these factors account for the differences noted whenever scaling law constants are compared. For example, Brown and Henley (1964) obtained a value of 0.92 at the same ionic strength as that used in the present dissertation. However, their samples had a DS of 1.06 and measurements were performed at 21 °C. Also, there was no mention of the pH at which the measurements were made. The samples utilized in their study had a much wider range of MW (~ 150,000 – 1,100,000 g/mol) than those studied in this dissertation. The PDI (Mz/MW) of their samples was 1.53 – 1.77 compared to 2.27 – 4.30 (LM-CMC – HM-CMC) in the present study. Kulicke et al. (1996) found α to be 0.9 for CMC samples ranging in MW from ~200,000 – 2,000,000 g/mol with a DS of 1.0 in 0.01 mol/L NaCl and at 25 °C. It should be pointed out that the differences in the published and present results are not overly significant considering the varied conditions under which the studies were performed. However, an explanation for the slight differences is warranted. From the presented intrinsic viscosity values, it can be seen that temperature does not have a pronounced effect on the conformation of CMC. It has also been established that pH does not have a strong influence on the conformation of CMC as long as a constant ionic strength is maintained. Thus it seems that pH and temperature differences do not account for the small differences in the values of α. Regarding DS, it was shown in the intrinsic viscosity results in section 3.3.3.4 that at a given pH, an increase in DS of CMC has a measurable effect on the conformation of CMC, thus the difference in DS of 0.7 in the present study and DS~1.0 in the studies cited above offers one possible explanation for the discrepancy. The other source of the discrepancy between the α values obtained in the present study and those published can be related to the molecular weight range studied and the polydispersity indices of these samples. The effect of molecular weight range on the exponent α was demonstrated by Bothner (1988) and Mendichi et al. (2003) in their studies on hyaluronan, a semistiff, negatively charged linear biopolymer. Both authors determined different values of α at the low and high ends of the molecular weight ranges tested than in the intermediate MW range, and a similar trend could be expected in the case of CMC. From this point of view, it should be noted that the molecular weight range tested in this work, albeit using fewer samples, is much narrower than in the 101  studies reported above, which is reflected in the linear relationship of the log [η] – log MW plot (Figure 3.11). 3.3.3.7 Persistence length Figures 3.12 shows the molar mass dependence of the intrinsic viscosity of CMC according to Bohdanecky’s method for determination of chain persistence length. The data are plotted according to Equation 2.27 and the persistence length values shown in the figures were calculated from the slopes, Bη [Equation 2.29: Bη = B0Φ0,∞-1/3(2Lp/ML)-1/2] of the linear fits to the data. The Flory’s constant,  used was  (Yamakawa and Fujii, 1974) while the parameter B0 =  1.05 was obtained from Bohdanecky’s article (Bohdanecky, 1983). The molar mass per unit length, ML = m/l where m is the glucose monomer mass (~162 g/mol) and l is the monomer length or the diameter of a monosaccharide (~0.54 nm). For a DS of 0.7, the average monomer mass is approximately 172 g/mol, thus ML ~ 319  Figure 3.12. Bohdanecky plot for CMC in distilled water and in 0.01 mol/L NaCl at different pH conditions.  102  It should be recalled that excluded volume effects are not accounted for in Bohdanecky’s method for determining persistence length, thus it is worth noting that the data presented above were obtained using very dilute solutions in which excluded volume effects were negligible. Moreover, the relative stiffness of the CMC chain (as indicated by the persistence length results) should diminish excluded volume effects. This can also be illustrated by the study of Davis (1991) who analyzed CMC intrinsic viscosity data produced by Brown and Henley (1964). Davis used the electrostatic wormlike chain theory which accounts for excluded volume effects and determined the CMC Kuhn length, Lk (see equation 2.24) to be 19.6 nm in 0.01 mol/L NaCl. The equivalent persistence length, Lp = Lk/2, was therefore 9.8 nm. This value was much lower than that determined by Tricot (1984) who employed the Yamakawa-Fujii (simply referred to as Yamakawa’s model) wormlike chain model to determine Lk for the same CMC data and found it to be 36.8 nm (Lp = 18.4). Davis attributed Tricot’s much higher value to excluded volume effects which are ignored in Yamakawa’s model. Bohdanecky’s method for stiff chain polymers involves a simple analytical form of Yamakawa’s theory obtained by redefining Flory’s constant in Yamakawa’s equation to include parameters dependent on the reduced hydrodynamic diameter (Equations 2.27 – 2.29), and integrating the approach into Yamakawa’s model to obtain Equation 2.27. Bohdanecky’s method, when applied to determine the persistence length for the same CMC data analyzed by Davis gives a persistence length of 11.2 nm, not significantly different from Davis’s value of 9.8 nm obtained with excluded volume effects accounted for. This validates the applicability of Bohdanecky’s method to the present data. It can be seen that the slopes, and therefore the persistence length Lp, of the linear fits in distilled water increase with pH while they are statistically constant in 0.01 mol/L NaCl. This trend can be expected in accordance with data in Figure 3.10 which shows a strong effect of pH in distilled water and a weak effect of pH in salt solutions. The independence of Lp from pH in 0.01 mol/L NaCl solutions indicates that the electrostatic persistence length Lpe is much lower than the intrinsic persistence length Lp0. To a good approximation, it can be assumed that the Lp value obtained at pH 3 is equal to Lp0 since at pH 3, the polymer is in an associated state and the presence of 0.01 mol/L NaCl is sufficient to screen the few dissociated carboxylic groups. With this assumption, Lpe values at pH 4.5 and pH 7 calculated using Equation 2.25 ( Lp  Lp 0  Lpe ) are 3.0 nm and 3.5 nm respectively. These values are slightly higher than the Lpe value (1.37 nm in 0.02 mol/L NaCl) 103  determined by Hoogendam et al. (1998b) but the small difference can be explained by the difference in ionic strength between the two studies. Hoogendam et al.’s value was obtained by setting Lp0 at 13nm which gives an Lp value of 14.4 nm, similar to the values obtained in 0.01 mol/L at pH 4.5 and pH 7. Thus, the obtained persistence lengths are in reasonable agreement with those obtained using carefully prepared CMC samples and employing more involved techniques and theories. In distilled water, Lp increases dramatically from 8.8 nm to 19.2 nm as pH is increased from 3 to 4.5. A further increase in pH to a value of 7 increases the Lp to 24.5 nm. The observed increase in Lp with pH can be explained by the increasing coulomb repulsive forces with increasing pH which leads to a higher Lp at higher pH values of 4.5 and 7, though the change in ionic strength of the solutions as pH changes complicates such a comparison. However, the Lp values in distilled water are considerably higher than those in 0.01mol/L NaCl at the same pH and illustrate the importance of ionic strength in affecting Lp. The dependence of Lp on ionic strength indicates that Lpe > Lp0 in distilled water. This can be demonstrated using the assumption made before for the data in 0.01mol/L NaCl that the Lp value at pH 3 fairly well reflects the chain length in the absence of electrostatic interactions and can be approximated to be close to Lp0. Thus, assuming Lp0 to be 8.8 nm, the calculated Lpe values at pH 4.5 and 7 are 10.4 nm and 15.7 nm respectively which are all higher than Lp0, and are significantly higher than the Lpe values estimated in 0.01 mol/L NaCl. The results presented above indicate that CMC is a semi-flexible, weakly stiff chain rather than a flexible chain extended by intramolecular excluded volume effects. Admittedly, the number of samples used for the analysis is low, and the method could be sensitive to the molecular weight distribution of the samples. However, the agreement with some of the data obtained using more involved methods, and the presentation of data as a function of solution pH, which is often lacking in many published works, should enhance the understanding of the dilute solution behavior of CMC. 3.4 Summary Six CMCs of various degrees of substitution (DS) and molecular weights (MW) were characterized by dilute solution viscometry. The intrinsic viscosities [η] of the polymers were measured as a function of ionic strength (distilled water and 0.01 mol/L NaCl), pH (3, 4.5, 7) and temperature (25 °C, 50 °C). The molecular weights and molecular weight distributions (MWD) of the polymers were determined from analytical ultracentrifugation (AU). 104  The molecular weights determined from sedimentation velocity (AU) were in fair agreement with those supplied by the manufacturer, and were thus used in subsequent data analysis. The MWDs of the polymers of different MW (LM-CMC, MM-CMC and HM-CMC) were broad with polydispersities increasing with MW from 2.27 for the lowest MW to 4.30 for the highest MW. The MWDs of the polymers of different DS (DS0.7, DS0.9 and DS1.2) but similar MW were also broad and the obtained polydispersities (3.99 – 4.52) did not significantly change with DS. In distilled water, the intrinsic viscosities increased with DS and pH indicating an extended conformation. In 0.01 mol/L NaCl, CMC assumed a coiled conformation resulting in a decrease in the intrinsic viscosities. The effect of pH on the conformation of CMC was negligible provided that constant ionic strength was maintained during the tests. Temperature also had a very weak effect on the intrinsic viscosity of CMC. Of the tested CMCs, the lowest molecular weight polymer (LMCMC) was the most flexible as inferred from its high ability to coil or stretch in solution. The Mark-Houwink-Sakurada (MHS) exponent α in the [η] ~ MW scaling law was found to be on the order of 0.83 without a clear variation with pH in 0.01 mol/L NaCl. Estimates of the chain persistence length (Lp) showed that chain flexibility decreased with increasing pH in distilled water (Lp ≈ 8.8 nm – 24.5 nm) while there was no significant change with pH in 0.01 mol/L NaCl (Lp ≈ 11.3 nm – 14.8 nm). These MHS exponent and Lp results indicate that CMC is semiflexible randomly coiling chain. Overall, the results suggest that ionic strength rather than pH is the main parameter affecting the conformation and flexibility of the tested CMC samples.  105  CHAPTER 4  4 The aggregation/dispersion state of oil sand slurries: Influence of CMC 4.1 Introduction It is well established that there is a relationship between the rheological behavior of a suspension and the inter-particle forces acting within the suspension. Thus, rheology can be used as a tool to examine the aggregation-dispersion properties of oil sand slurries. In this dissertation, rheological tests were performed on slurries prepared from an average-processing oil sand ore. The main aim was to probe interactions between oil sand components, in the presence and absence of carboxymethyl cellulose. The type of oil sand ore chosen was characterized by a moderate amount of bitumen and a high fines content (- 44 µm size fraction). At the same time, this type of ore represents problematic ores encountered in the oil sands industry by account of its high amount of fines. In order to determine the contribution of bitumen to the aggregation-dispersion state of oil sand slurries, additional rheological measurements were also performed on bitumen-free solids extracted from the oil sand ore. The rheological tests were complimented with cylinder settling tests in order to confirm the stability of the slurries towards aggregation and settling. Adsorption tests of CMC onto bitumen-free solids were also performed under the same experimental conditions as in the rheological and settling tests in order to evaluate the dispersing capabilities of different CMC polymers. 4.2 Materials and methods 4.2.1  Materials  The oil sand ore was provided by Canadian Natural Resources Ltd. (Calgary, AB). The ore was kept in a freezer maintained at about -15 °C to minimize oxidation. A representative sample was obtained and characterized in terms of the bitumen, water, and solids contents using the Dean-Stark analysis (Bulmer and Starr, 1979). The Dean-Stark method is a standard industrial technique for analyzing oil sands.  106  Solids from the oil sand ore were obtained by repeated washing with toluene, filtering, and evaporation of toluene under a fumehood. The clean solids were free of bitumen and toluene. A representative sample was obtained for mineralogical analysis. The sample was screened at 44 microns and the - 44 µm and + 44 µm size fractions, as well as the total sample were analyzed by xray diffraction using the Rietveld refinement method. Table 4.1 shows selected properties of the ore and solids. Table 4.1. Composition of the tested oil sand ore and solids obtained from the ore.  Ore composition (Dean-Stark analysis)  Solids mineral composition, wt. % (XRD) Total sample  – 44 µm size fraction  Bitumen (wt. %)  9.4  Quartz  87.4  65.1  Water (wt. %)  3.9  Kaolinite  4.2  15.9  Solids (wt. %)  86.1  Muscovite  3.0  10.8  K-feldspar  5.1  6.5  Anatase  0.1  0.9  The data show that the oil sand ore can be described as an average-processing ore, as indicated by the moderate bitumen content and high amounts of solids and fines. The mineralogical analysis indicates that quartz is the dominant mineral in the ore, but it is important to observe that clay minerals, particularly kaolinite and muscovite, tend to accumulate in the fines fraction. The amount of kaolinite and muscovite increased by about 11% and 7% respectively in the fines fraction, compared to the total sample. The volume mean diameter of the particles of the clean solids from the ore was measured using a Malvern Mastersizer 2000 and was found to be 70.5 μm with a top size of ~275 μm. The volume fractions below 44 μm and 3 μm were 35.1% and 6.3% respectively. The relatively high volume fraction below 3 μm is particularly significant as this size fraction is most sensitive to any surface chemistry changes in the slurries and thus significantly contributes to the rheological behavior of oil sand slurries (Gutierrez and Pawlik, 2012a). The specific surface area of the solids was measured from nitrogen adsorption using a Quantachrome 1MP BET (Brunauer Emmett Teller) analyzer and was found to be 2.24 m2/g. 107  Six CMC samples described in section 3.2.1 were tested to determine their influence on the aggregation-dispersion state of oil sand and bitumen-free solid slurries. Stock solutions of the polymers were prepared as detailed in section 3.2.1. 4.2.2  Methods  4.2.2.1 Rheological measurements The oil sand slurries were prepared by mixing a mass of the ore with a given mass of water to reach a solids content of 55 wt. %. The solids content was calculated as the mass of solids only (not including bitumen) divided by the total mass of the slurry. The corresponding volume concentration was calculated by dividing the volume of solids by the total volume of the slurry (including the volume of bitumen) and was ~31.5 vol. %. The mixing process was done at ambient temperature, typically 23 °C, for 30 minutes at an impeller speed of 450 rpm using a Lightnin Labmaster mixer. Several mixing speeds were tried and 450 rpm was found to effectively shear the bitumen–mineral lumps and keep the oil sand components well suspended in distilled water. The mixing time of 30 minutes was found to produce slurries that were well wetted, easy to handle, and with a stable pH. The slurry pH was adjusted to 8.5 using small amounts of a concentrated sodium hydroxide solution to avoid dilution of the slurry. The selected pH is essentially that of the industrial bitumen extraction process. The slurry volume, mixing speed, and impeller rotational speed were closely controlled to minimize any time-dependent behaviour and potential systematic errors. After 30 minutes of mixing, the slurry was quickly transferred to the rheometer for testing. In tests involving CMC, a predetermined amount of the polymer stock solution was added after 20 minutes of mixing. The impeller speed was then lowered to 400 rpm and mixing continued for a further 10 minutes after which the slurry was transferred to the viscometer. The time interval between the end of slurry mixing and the start of the measurements in the rheometer was kept as short as possible, usually between 20 and 30 seconds, in order to minimize slurry settling. Prior to the rheological testing, settling tests were performed in order to qualitatively determine the stability of 55 wt. % suspensions of the ores and solids towards settling, and hence determine the best measuring time needed to avoid substantial settling in the viscometer. Figure 4.1 shows a comparison of the ore and solid slurries just after mixing and after 5 minutes. Close visual examination indicated that very coarse particles in the coarsest fraction of the slurry (approximately 108  two-thirds of the entire volume of the slurry) started settling immediately after transferring the slurry to the cylinder but there was no indication of substantial settling until after about 5 minutes when a visible supernatant-sediment interface in the baseline ore slurries formed. No supernatant-sediment interface was visible in the slurries to which CMC was added. Based on the settling tests, rheological measurement time of 3 minutes was used.  Ore+CMC Ore+CMC at 0 min at 5 min  Ore at 0 min Ore at 5 min  Figure 4.1. Settling properties of oil sand slurries (55 wt. % solids, pH 8.5) at 0 and 5 minutes in the presence and absence of 250 g/t CMC.  Baseline steady-state rheological measurements were carried out on oil sand and bitumen-free solid slurries in the absence of CMC at 55 wt. % solids content. Measurements were also carried out with CMC addition at dosages of 125 g/t and 250 g/t (based on the weight of ores or solids). When there was a difference in the rheograms at these two dosages, higher concentrations were tested to further probe the impact of CMC on rheology. The measurements were performed using a standard 109  Haake Rotovisco VT550 rotational viscometer, connected to an elongated fixture (Klein, 1992; Klein et al., 1995) consisting of a concentric, double-gap, bob-in-cup, Couette type arrangement with inner and outer gap sizes of 2.50 mm and 3.03 mm, respectively. Figure 4.2 illustrates a schematic of the device. The dimensions of the fixture allow the bob to be positioned within a constant density zone for the duration of the experiment to minimize errors arising due to settling of particles. The outer gap dimensions satisfy the general requirement that the gap size should be at least ten times the largest particle diameter in the tested slurry in order to prevent the particles from plugging the rheometer (Van Wazer et al., 1963). In the tested slurries, the top particle size was ~275 μm, which satisfies the requirement for the outer gap. The effect of particle size on the inner gap was neglected since the potentially problematic coarsest particles settled within seconds inside the rheometer. It should be noted that the original fixture designed by Klein (1992) had much smaller inner and outer gap sizes of 1.0 and 1.1 mm as the suspensions tested by Klein had a top size of about 100 μm. Pawlik (2002) redesigned the fixture to test coarser suspensions by increasing the inner and outer gap sizes to 2.50 mm and 3.03 mm, respectively, thereby making it suitable to test suspensions with top sizes up to about 300 μm. However, increasing the gap sizes potentially increases nonNewtonian shear rate effects due to non-linear and incomplete shearing across the width of the gaps. In this work, the tested suspensions produced low shear stresses in the range 0 - 20 Pa (0 – 8 Pa for most of the suspensions to which CMC was added), and thus non-Newtonian shear rate effects were considered to be negligible.  110     Figure 4.2. A schematic illustration of the elongated fixture designed for rheological measurements on settling suspensions.  Another salient design feature of the fixture is that the shearing surfaces were modified to include vertical grooves, 300 μm deep, and 500 μm apart in order to minimize wall slip effects. In addition, the hollow bob eliminates end effects. However, the ore and solids slurries in which no CMC was added were highly non-Newtonian and there was indication of possible wall slip as was seen from the low shear stress values at very low shear rates, thus the first two points of the rheograms were omitted. In steady state measurements, the Haake Rotovisco viscometer applies a rotational speed (Ω) to the bob and the resulting stress (  ) is measured as the torque exerted on the suspension by the rotation of the bob. The rotational speed is proportional to the shear rate,  (   M ) while the shear stress is proportional to the torque, T (   AT ) where M and A are constants. The value of M was determined from the elongated fixture dimensions and was calculated as 3.736 (Appendix 8). The value of A was calculated to be 3532.99 Pa/Nm by determining the viscosities of two standards, 111  CANNON N75 (viscosity = 125 mPas) and CANNON N415 (viscosity = 828.3 mPas) at 25 °C and dividing the actual viscosities of the standards by the obtained viscosities, assuming that A was equal to 1 Pa/Nm. The values of M and A were entered into the Haake software to facilitate recalculation of the measured torque and applied rotational speed data to shear stress and shear rate, respectively. The maximum applied shear rate (120 s–1) was less than the critical shear rate at which Taylor vortices appear and the flow between the shearing surfaces becomes turbulent. The critical shear rate was calculated using the simplified Taylor equation (Taylor, 1923; Akroyd, 2004). .        ..[4.1]  where Vθ is the velocity of the rotating bob, η is the slurry viscosity, ρ is the slurry density, Rc is the radius of the cup, and k is the ratio of the outer gap dimensions. According to Equation 4.1, the lowest possible critical shear rate occurs at the outer gap with slurries of low viscosity. The lowest slurry viscosity was anticipated to be about 20 mPas for the tested slurries (density of about 1500 kg/m3). It was assumed that the outer gap was narrow enough that non-Newtonian shear rate effects were negligible and that the velocity gradient was constant across the gap. The angular velocity of the rotating bob calculated from Equation 4.1 is 0.467 m/s (23.3 rad/s). The critical shear rate, Dc can be taken as the average shear rate across the outer gap given by Equation 4.2 (Macosko, 1994), and is calculated to be 165.8 s–1. ..[4.2] Where  (Rc is the radius of the cup, Rb is the radius of the bob),  is the angular velocity  of the bob. It turned out that the lowest slurry viscosity tested was about 13 mPas for the Newtonian slurries dispersed with CMC. Crosschecking the above calculation with η = 13 mPas and using the expression for shear rate for Newtonian fluids  , the critical shear rate is  calculated to be 123.5 s–1, thus it can be concluded that the flow curves for all tested slurries were obtained below the limiting maximum shear rate for the onset of turbulent flow. The following procedure was followed to obtain flow curves: With the slurry in the viscometer, a rest time of 1 minute was allowed during which no shear was applied. Since mixing breaks down 112  any structure present in the slurries, the 1 minute rest time allowed for structure reformation. The slurry was then pre-sheared at 100 s–1 for a further 1 minute followed by a rest time of 10 seconds after which the shear rate was increased from 0 to 120 s–1 over a period of 50 seconds, and then decreased back to 0 s-1 over a further 50 seconds. It should also be noted that the rest times incorporated in the procedure served to remove micro-bubbles of air entrapped within ore lumps, but as explained later, it was not easy to completely remove the air bubbles which led to poor reproducibility of the baseline flow curves. The rheological data collected (torque exerted on the bob vs. rotational speed of the bob) were recalculated to ‘shear stress vs. shear rate’ flow curves. This procedure was found to produce a measurable rheological response for the baseline slurries and this facilitated comparison of the flow curves obtained in the presence and absence of CMC. It should be emphasized that the rheological analysis was performed to ascertain the aggregation/dispersion state of oil sand slurries in the absence and presence of CMC rather than to quantitatively determine the rheological properties (yield stress and apparent viscosity) of the slurries. However, in some cases, the apparent viscosities at a specified shear rate were calculated for comparison of the two states of the slurries. The reproducibility of the flow curves was checked by obtaining duplicate flow curves for each slurry tested. Data on the reproducibility of the rheological measurements are presented in the results and discussion sections. 4.2.2.2 Stability/settling tests The rheological measurements were supplemented by sedimentation tests in order to confirm the aggregation/dispersion state observed for the tested slurries. Settling tests were performed both on oil sand ores and on bitumen-free solids under the same conditions as in the rheological measurements. Slurries for settling tests were prepared following the procedures described in section 4.2.2.1. After mixing, the ore and solid slurries were transferred to 300-mL graduated cylinders. The cylinders were flipped ten times to re-suspend any settled particles during the transfer stage. The slurries were then allowed to stand until there was no further settling (one week was sufficient) after which three 30-mL aliquots of the supernatants were carefully withdrawn and weighed using calibrated 10-mL syringes to determine their densities. In addition, digital pictures of the settled slurries were taken. Due to the high turbidities of the supernatants (especially those in which CMC was added), the commonly used methods for assessing the stability of the slurries (for example turbidity or transmittance measurements) could not be used. Instead, the measured densities were 113  used to determine dispersion coefficients of the slurries following the method described by Pawlik et al. (2003). The dispersion coefficient (DC) is a relative measure of the dispersion state of the slurries and is calculated from Equation 4.3 (Pawlik et al., 2003). ..[4.3] where δsample is the density of the supernatant of the tested slurries, δmedium is the density of the pure medium, in this case distilled water, and δSlurry is the density of the fully dispersed slurries (δSlurry) determined from the densities of the ore or solids and distilled water. It can be seen from Equation 4.3 that DC = 100% for fully dispersed slurries (δsample = δSlurry) and DC = 0 for fully aggregated or coagulated slurries in which all solids settle leaving supernatants of similar density as that of the pure medium (δsample = δmedium). 4.2.2.3 Adsorption tests In the adsorption tests, 40-g of bitumen-free solids were mixed with 22 mL (32 mL in the case of background tests) of distilled water for 10 minutes after which 10 mL of 1 g/L CMC stock solutions of different MW and DS were added to achieve the same solids content (~ 55 wt. % or 31.6 vol. %) as in the rheological and stability tests. The CMC dosage was 250 g/t corresponding to an initial CMC concentration of ~313 mg/L in the aqueous phase. The solution pH was adjusted to 8.5 before adding the CMC solution. The mixtures were then stirred for an additional 20 minutes after which the solids were centrifuged at 10,000g for 30 minutes. The supernatants after this centrifuging time still contained suspended solids. A visual assessment showed that there were more suspended solids in the supernatants containing CMC than in the baseline supernatants. The supernatants were therefore vacuum-filtered on a 0.45-micron filter and the clear filtrates were assayed for total organic carbon (TOC) using a Shimadzu TOCCPH analyzer. The filter paper used for filtration contained a small amount of surfactants that contributed to the measured TOC. To correct for this, distilled water (same volume as the filtered supernatants) was vacuum-filtered using the same type of filter paper and the TOC of the filtrate was subtracted from the measured TOCs. Also, it was determined that vacuum-filtration decreased the TOC of the CMC solutions. This was determined by diluting a CMC stock solution to 100 mg/L, and assaying the filtered and unfiltered solutions for TOC. The difference in the TOC was added to the TOC values obtained for the vacuum-filtered supernatants. Finally, the corrected TOC values were recalculated to the residual CMC concentrations using the 114  TOC assays for the 100 mg/L unfiltered CMC solutions, and the nitrogen BET surface area of the bitumen-free solids was used to calculate the adsorption densities. 4.3 Results and discussion 4.3.1  Baseline tests for oil sand slurries and influence of addition of CMC  The baseline aggregation/dispersion state for oil sand slurries was determined through rheological measurements of an oil sand ore/water slurry at a solids content of 31.5 vol. %, maintained at pH 8.5 and 25 °C, and the results are plotted in Figure 4.3. Also shown in Figure 4.3 is a duplicate test of the shear stress-shear rate flow curve. As can be seen from the ramping and descending portions of the flow curves, the slurries show strong shear stress hysteresis indicative of thixotropic behaviour. The flow curve pattern depicts non-Newtonian shear thinning behaviour with a yield stress which indicates the presence of a structure due to aggregation in the slurry. The very poor reproducibility of the flow curves is typical for aggregated slurries. In such concentrated oil sand slurries at high pH, the poor reproducibility results from the presence of micro-bubbles of air that can nucleate on the liberated hydrophobic bitumen and affect the shearing pattern in the slurry. As explained in section 4.2.2.1, the air bubbles entrapped in the slurries were minimized by pre-shearing the slurries and allowing them to rest before obtaining the flow curves, but as shown in Figure 4.3, it was not possible to completely remove the air bubbles which led to the poor reproducibility. De-aerated water can be used to prepare the slurries but the presence of ore lumps in the system makes it almost impossible to remove the micro air bubbles. As shown later in this section, it was much easier to reproduce flow curves for dispersed slurries (those in which CMC was added) because there was no space for air bubbles to be trapped. Good reproducibility was also obtained for the baseline slurries prepared with bitumen-free solids, that is, solids obtained by removing bitumen from the ore, as well as bitumen-free solids dispersed with CMC (section 4.3.3) which confirms that the poor reproducibility of the baseline ore slurries was caused by the nucleation of micro air bubbles on bitumen within the ore lumps. It should be noted that the baseline flow curves shown in later figures are averages of the two curves from Figure 4.3. In some cases, only the ramping parts are shown for clarity.  115  Figure 4.3. Flow curves of baseline (oil sand ore slurries only) duplicate tests performed at 31.5 vol. % solids content, pH 8.5, 25 °C in distilled water.  Figure 4.4A shows the effect of addition of CMC of different DS on the flow curves of oil sand slurries. For comparison, the ramping part of the baseline flow curve is also shown in the same figure. Also, for clarity, only the ramping parts of the flow curves are presented. As can be seen in Figure 4.4B (for clarity, only every second point is shown), the shear stress hysteresis observed in the baseline slurries completely disappears in presence of 125 g/t CMC. It is evident that with addition of CMC at a dosage of 125 g/t (based on the weight of the ore), the baseline flow curve changes from non-Newtonian yield shear thinning to Newtonian with no indication of a yield stress. An increase in the CMC dosage to 250 g/t does not result in a further decrease in the shear stress and apparent viscosity indicating that a low CMC dosage of 125 g/t is sufficient to change the rheology of the baseline slurries. It is also noted that there is no significant effect of DS on the rheological response. The standard deviation of the shear stress measurements on slurries in which CMC was added was always lower than 1.0 Pa, thus the small differences in DS seen in Figure 4.4 are within the experimental error and should be considered insignificant.  116  The flow curves in Figures 4.3 and 4.4 portray two different systems in accordance with the DLVO theory of colloidal stability. Firstly, the baseline oil sand slurries are aggregated due to the attractive forces between the oil sand components.  Figure 4.4. Flow curves of oil sand ore slurries (ramping parts only) in the presence of 125 g/t and 250 g/t CMCs of different DS (A) and full flow curves of oil sand slurries in the presence of 125 g/t CMCs of different DS (B). Solids content = 31.5 vol. %, pH = 8.5, CMC MW = 293,000 – 310,000 g/mol.  It was expected at the high pH (8.5) at which the rheological measurements were performed that repulsive forces would dominate over any inter-particle attraction. The mineral–mineral and bitumen–mineral interactions are supposedly repulsive assuming that no other attractive forces are present. The study by Dang-vu et al. (2009a) on wettability of poor oil sands indicates that solids from ores such as the one tested in this dissertation are hydrophobic. Thus, attractive mineral– mineral and bitumen–mineral hydrophobic interactions can be expected. More important are bitumen–bitumen forces. Although it was determined that the hydrophobicity of bitumen decreases with increasing pH (Masliyah et al., 2011, Liu et al., 2005b), a dramatic drop was only seen above pH 9. In the study by Liu et al., the hydrophobic force constant was not affected up to pH 8.5 after which a significant decrease was noticed. The results obtained in this dissertation (presented in Chapter 5) also indicate that bitumen retains significant hydrophobicity at pH 8.5 with a sharp drop measured at pH 10.5. Thus, the observed results for the baseline slurries can be attributed to mineral–mineral, bitumen–mineral and bitumen–bitumen attractive hydrophobic forces. It should 117  also be noted that in such concentrated systems, the interparticle spacing is so small that only small magnitudes of van der Waals and hydrophobic attractive forces are required to induce extensive aggregation. Secondly, the slurries in which CMC was added are dispersed as can be seen from their flow curves which generally show no yield stress and much lower apparent viscosities than those measured in the baseline case. In these slurries, the repulsive forces dominate over the attractive forces mentioned above, thus the observed effect of CMC is due to the dispersion of the aggregated system. A mechanism through which CMC disperses the slurries will be elucidated in later sections. In order to study the effect of CMC molecular weight on the aggregation/dispersion state of oil sand slurries, rheological tests were also performed with addition of LM-CMC (MW = 123,000 g/mol) and HM-CMC (MW = 715,000 g/mol). Figures 4.5 and 4.6 show the obtained flow curves. For comparison, the baseline case is plotted in the same figures. As seen in the figures, the flow curves obtained in the presence of LM-CMC and HM-CMC generally display lower shear stresses and apparent viscosities compared to the baseline case. In contrast to the other polymers, the flow curves obtained at lower dosages of the LM-CMC (125 and 500 g/t) show marked hysteresis similar but opposite in direction to that obtained in the baseline case. The downward part of the flow curve falls above the upward part indicating a shear-thickening response and formation of some form of a structure with time. However, it should be noted that addition of LM-CMC at all dosages breaks up the structure originally present in the baseline case as seen in the Newtonian upward parts of the flow curves in Figure 4.5. Only when the shearing rate is decreased back to 0 s-1 does the structure start forming again but at a much lower magnitude than is seen in the baseline case. At a higher dosage of 750 g/t, the build up of the structure observed at lower dosages completely disappears and the upward and downward parts of the flow curves overlap. As is seen in Figure 4.6, adding 250 g/t of the HM-CMC produces a slurry that is already as dispersed as that of 750 g/t for the LM-CMC. This clearly illustrates that the dispersing capabilities of the polymers primarily depend on the polymer chain length (molecular weight) rather than on the number of anionic functional groups (DS) as illustrated in Figure 4.4 where a change in the DS from 0.7 to 1.2 (equivalent to a change in the degree of anionicity from about 23% to 40%) does not result in a noticeable change in the rheological properties of the slurries.  118  Figure 4.5. Flow curves for oil sand ore slurries obtained at different dosages of LM-CMC (molecular weight of 123,000 g/mol). The baseline case is also shown. Solids content = 31.5 vol. %, pH = 8.5.  Figure 4.6. Flow curves for oil sand ore slurries obtained in the presence of 250 g/t HM-CMC (molecular weight of 715,000 g/mol). The baseline case is also shown.  119  Figure 4.7 shows duplicates of flow curves obtained from rheological measurements on slurries in which CMC was added. In comparison to the baseline case (Figure 4.3), it can be seen that the reproducibility in slurries without a structure is much better. Specifically, the best reproducibility was always obtained with slurries that produced Newtonian flow behaviour compared to those that deviated from Newtonian behaviour. This can also be seen in Figure 4.7A and B where the reproducibility obtained in Figure A, though much better than the baseline case, is not as good as that in Figure B. 12  A  LM-CMC: dosage 250 g/t -Test 1 LM-CMC: dosage 250 g/t -Test 2  Shear Stress [Pa]  9  6  3  0 0  20  40 60 80 Shear Rate [s -1 ]  100  120  Figure 4.7. Reproducibility of flow curves for oil sand ore slurries obtained in the presence of LM-CMC (plot A) and HM-CMC (plot B), both at a dosage of 250 g/t. Solids content = 31.5 vol. %, pH = 8.5.  4.3.2  Stability tests  In order to confirm the aggregated state of oil sand ore slurries and the dispersing action of CMC, stability tests were performed in settling cylinders as shown in Figure 4.8. In each case immediately after starting the settling tests, the coarse sands settled quickly to the bottom of the cylinders leaving a suspension of fine particles at the top. Without CMC addition (baseline case), the fine suspension further settled leaving a clear supernatant at the top indicative of an unstable aggregated slurry (Somasundaran et al., 2009). With 125 g/t CMC addition, the fine fraction largely remained suspended. There was indication of limited settling leaving a very turbid supernatant at the top. As the dosage of the polymer was further increased, the fine fraction remained suspended with very 120  minimal settling as shown by the very small volume of supernatant at the top. Clearly, the slurries in which CMC was added were very stable against aggregation and settling, particularly the fine fraction, confirming the dispersing action of the polymer. In order to further confirm the results in Figure 4.8, the Dispersion Coefficients (DC) for the slurries were determined and are plotted in Figure 4.9. Also plotted are the apparent viscosities of the slurries calculated at a shear rate of 100 s-1 from the flow curve for DS0.7 in Figure 4.4. The values of the DC are low due to the fact that the calculation of the DC values takes into account the coarsest portion of the slurries that settled instantaneously. Even with the low DC, it can still be seen that the DC values increase at higher CMC dosages and this increase is accompanied by a decrease in the apparent viscosities of the slurries, clearly demonstrating the strong aggregation of the baseline oil sand ores and the efficient dispersion of the slurries at CMC dosages higher than 125 g/t.  No Polymer  125 g/t  250 g/t  500 g/t  Figure 4.8. Settling of oil sand slurries in the absence and presence of different dosages of CMC (DS0.7, MW = 295,000 g/mol). The slurries were prepared in distilled water. pH = 8.5, T = 25 °C, solids content 31.5 vol. %. Photographs were taken after 1 week of settling.  121  Figure 4.9. Dispersion coefficients and apparent viscosities of oil sand ore slurries plotted versus CMC (DS0.7, MW = 295,000 g/mol) dosage.  An important qualitative observation made during the sedimentation tests was that there appeared to be more bitumen in the suspended fine fractions of the slurries in which CMC was added, and after a week of settling, a thin layer containing bitumen droplets could be seen at the surface of the slurries. However, it was not possible to quantify the amount of bitumen in this layer as it would collapse into the slurry with any small disturbance. This observation indicates that under mixing conditions, more bitumen was liberated in slurries in which CMC was added. During the course of the sedimentation period (one week), this “free” bitumen rose through the slurry to the surface. 4.3.3  Understanding the role of CMC: Results for bitumen-free solids  In order to more clearly understand the dispersing action of CMC, rheological measurements were performed on slurries prepared using solids extracted from the oil sand ore, that is, with bitumen removed. The same solids volume concentration as in the ore slurries was tested in order to determine the contribution of bitumen to the aggregation of oil sand ore slurries. Figure 4.10 shows the flow curves obtained for the polymers of different DS and MW. The slurries were also tested at pH 8.5 and 25 °C in distilled water. The baseline case, that is, the slurry containing only bitumen-free solids shows the same time-dependency seen in the baseline oil sand 122  Figure 4.10. Flow curves for oil sand solid (without bitumen) slurries obtained in the presence of 250 g/t CMCs of different DS (DS0.7 and DS1.2) and MW (LM-CMC and HM-CMC). The baseline case is also shown in each case.  ore slurries indicating the presence of a structure that is broken down with shear. Generally, the same dispersing effect of CMC as seen in the ore slurries is obtained in all cases. However, the weak structure build-up seen in Figure 4.5 for LM-CMC in ore slurries can also be seen in the return parts of the flow curves for DS0.7 and DS1.2, though it is much less pronounced. Thus it seems that on removal of bitumen, the amount of polymer needed to ensure a stable dispersed slurry increases even for the DS0.7 and DS1.2, thus dosages higher than 250 g/t are required to prevent weak structure build-up. For example, at a dosage of 500 g/t DS0.7, the structure disappears and the downward part 123  of the flow curve overlaps with the upward portion. The higher MW (HM-CMC) CMC still performs very well at a dosage of 250 g/t as seen from the overlapping Newtonian flow curves. The reproducibility of the results above was checked by performing duplicate tests on the baseline case and the slurries containing the DS0.7 and HM-CMC polymers. As shown in Figure 4.11, the reproducibility in each case is very good. The good reproducibility obtained with the aggregated bitumen-free solids shows that the presence of bitumen substantially contributes to the very poor reproducibility obtained for the flow curves of the baseline oil sand ore slurries (Figure 4.3).  Figure 4.11. Reproducibility of flow curves for bitumen-free solid slurries and bitumen-free solids/DS0.7 mixture (left graph) and bitumen-free solids/HM-CMC mixture (right graph). CMC dosage was 250 g/t.  124  No Polymer  125 g/t  250 g/t  500 g/t  Figure 4.12. Settling of bitumen-free solids in the absence and presence of different dosages of HM-CMC The slurries were prepared in distilled water at the same conditions as in the rheological tests: pH = 8.5, temperature = 25 °C, solids content 31.5 vol. %. The photographs were taken after 1 week of settling.  Further evidence of the dispersive effect of CMC was obtained through settling tests as shown in Figure 4.12. Just like in the oil sand ore slurries, the coarse portion of the slurry settled immediately after starting the settling tests leaving a fine suspension that slowly settled with time. The clarity of the supernatant deteriorates with increasing CMC dosage and this is also shown by the Dispersion Coefficients for the slurries which steadily increase with increasing CMC dosage as shown in Figure 4.13. Also, as can be seen in the figure, the apparent viscosity of the baseline slurry calculated at a shear rate of 100 s-1 decreases in the presence of 250 g/t HM-CMC which demonstrates the efficient dispersion of the slurries by CMC. The supernatant of the baseline case (Figure 4.12) is not as clear as that obtained with the oil sand ore slurries (Figure 4.8) indicating the less aggregated nature of the solid particles in the absence of bitumen.  125  Figure 4.13. Dispersion coefficients and apparent viscosities of bitumen-free solid slurries plotted versus HM-CMC dosage.  4.3.3.1 Adsorption of CMC on bitumen-free solids In order to explain the different dispersing capabilities of the LM-CMC and HM-CMC as well as the results in Figure 4.10 for the DS0.7 and DS1.2, residual CMC concentrations and adsorption densities after adsorption of CMC onto bitumen-free solids were determined and are shown in Figure 4.14. As is seen, the residual CMC concentrations after adsorption of the DS0.7, DS0.9, DS1.2, LMCMC, and MM-CMC polymers are much higher than that of the HM-CMC indicating that the adsorption of the polymer is the highest in the case of HM-CMC. The corresponding adsorption densities of the DS0.7, DS0.9, DS1.2, LM-CMC, and MM-CMC polymers on the solids are lower than that of the HM-CMC. It is also important to note that the residual CMC concentrations and adsorption densities of the DS0.7, DS0.7 and DS1.2 polymers are quite similar whereas a systematic decrease in the CMC residual concentration and corresponding increase in the adsorption density is seen as molecular weight increases from LM-CMC to HM-CMC. The residual CMC concentration after adsorption of the HM-CMC is about 2.5 times lower than that of the LM-CMC and the adsorption density of LM-CMC is about 1.5 times lower than that of HM-CMC. 126  Figure 4.14. Residual CMC concentrations and adsorption densities after adsorption of CMC onto bitumen-free solids. The initial CMC concentration was ~313 mg/L (297 mg/L in the case of HM-CMC), dosage = 250 g/t.  Under the experimental conditions, the complete adsorption density for all the CMCs was about 0.11 mg/m2, thus the solids almost adsorbed the entire amount of the HM-CMC introduced into the solids/distilled water mixtures, but only about one-half of the DS0.7, DS0.9, DS1.2 and about twothirds of the LM-CMC and MM-CMC was adsorbed. The adsorption results can be correlated with the rheological flow curves presented in Figure 4.10. As long as there is a high amount of residual CMC left in solution, and the amount of CMC adsorbed is low as in the case of the DS0.7, DS1.2 and LM-CMC, the flow curves show the shear induced structure. Upon shearing, and with incomplete surface coverage by the polymer, the residual CMC in solution is able to absorb on two or more particles leading to weak shear-induced flocculation in the slurries. With an increase in the dosage of the polymers as seen in the case of the DS0.7 (500 g/t, Figure 4.10), sufficiently high coverage of the particles by the polymer, probably 127  near the plateau adsorption level ensures stable dispersion of the slurry and CMC adsorption on several particles at the same time becomes very difficult if not sterically impossible. In the case of the HM-CMC, there is very little CMC left in solution and the adsorption density is already high. In this case, particles are almost fully covered with the polymer which results in adequate dispersion of the slurry, and the shear-induced structure does not form. It should also be noted that in the case of the oil sand ores (containing bitumen), the shear induced structure is seen only in the case of the LM-CMC polymer but not for the DS0.7 and DS1.2 polymers despite very similar adsorption densities of the polymers on the bitumen-free solids. Thus the formation of the observed shear-induced structure by LM-CMC is also most likely related to the higher flexibility of the LM-CMC, or in other terms, its higher ability to coil or stretch in aqueous solution (Table 3.2, section 3.3.3.5) compared to the other polymers. It should be observed that although the rheological measurements were performed in distilled water, the resulting process water in equilibrium with the ore had sufficient ionic strength to bring about conformational changes in the polymer chains. The ionic strength of distilled water increases with mixing due to the gradual release of ions from the ore into the aqueous phase. The conductivity of the aqueous phase after mixing with the oil sand ore can be used as a convenient measure of the ionic strength of the slurry. The conductivity of the aqueous phase was found to be 3463.0 μS/cm which is equivalent to ~ 0.03 mol/L NaCl. In comparison, the conductivity of distilled water used to prepare the slurries was only about 5 μS/cm. The conductivity obtained when distilled water was mixed with the bitumen-free solids was 2246.5 μS/cm which is equivalent to ~ 0.02 mol/L NaCl. As already mentioned, the structure forms most likely due to weak flocculation resulting from adsorption of the residual polymer on two or more particles. The stability of the structure depends not only on the adsorption density but also on the ability of the adsorbed polymer chains to conform to the size, shape, and morphology of the individual particles, and this capability is enhanced in the case of the LM-CMC due to the higher flexibility of the polymer and its ability to stretch or coil in solution. This polymer behaviour can be related to the well-known role that adsorbed polymer chain conformations (loops, trains, and tails) play in flocculation (Scheutjens and Fleer, 1979, 1980), as only flexible polymers can be expected to assume such shapes. The adsorption data also support the postulation that one of the mechanisms through which the polymers disperse the oil sand slurries is by adsorbing on the solid particles and dispersing them as a 128  result of repulsive electric double layer forces due to the negatively charged CMC, as well as from steric forces arising from the physical barrier formed by the adsorbed CMC layers. The adsorption of polyelectrolytes on similarly charged surfaces, like in this case, is not uncommon as evidenced through many studies. Pawlik et al. (2003) for example determined that a CMC of molecular weight 250,000 g/mol and DS 0.7 (similar to the MM-CMC tested in this dissertation) adsorbed on negatively charged illite and efficient dispersion of illite was achieved at high CMC concentrations exceeding the plateau adsorption densities. At low CMC concentrations (below the adsorption plateau), CMC destabilized the illite suspensions by flocculation as also seen in this work from the results obtained with LM-CMC. Although negatively charged polymers can be repelled from negatively charged surfaces (in which case little adsorption will take place), appreciable adsorption can be facilitated by either low pH conditions under which the polymer’s negative charge becomes neutralized by protonation, or by increasing the ionic strength under which the negative charge is sufficiently screened by the electrolyte (Theng, 1982). In the study by Pawlik et al. (2003), an increase in the CMC uptake was achieved by increasing the ionic strength. Similarly, in the study by Hoogendam et al. (1998a), the plateau values in the adsorption of CMC on hematite at pH > 8.3, when the mineral’s net charge was negative, was increased by increasing the salt concentration from 0.01 mol/L to 1 mol/L. In the present study, the ionic strength (~0.03 mol/L NaCl) was sufficient to screen the negative charges on CMC but the adsorption data show that a higher adsorption density was achieved with the polymer of the highest molecular weight (HM-CMC). Theng (1982) also noted that the adsorption of negatively charged polymers can be promoted by the presence of polyvalent cations which act as bridges between the anionic groups on the polymer and the negative sites on the particles. However, also important are positively charged surface sites available on the particles as illustrated in the study by Areas and Galembeck (1991) in which two CMCs of low DS and MW were assumed to adsorb on hydroxylapatite by forming small loops on the mineral surface and binding their carboxylate groups to the surface Ca2+ sites. The widely accepted view is that natural polysaccharides interact with minerals via surface metal sites, the nature of the interaction likely being an acid/base interaction (Liu et al., 2000). As postulated by Liu et al. (2000), the extent of the interaction determines whether adsorption proceeds through hydrogen bonding, as illustrated for example by the work of Somasundaran et al. (2005) on 129  adsorption of guar gum on talc, or chemical complexation, as illustrated by Rath et al. (1997). For the systems studied in this dissertation, it can be postulated that the adsorption of CMC on the solid particles probably occurs through a combination of hydrogen bonding (rather than polar or electrostatic forces) particularly since the main constituent of the solids in the ore is negatively charge quartz, and possible surface metal sites present on the solid particles or even polyvalent cations such as calcium and magnesium introduced into solution from the oil sand components. 4.3.3.2 Role of bitumen in aggregation of oil sand slurries It should be noted that whereas the rheological tests on bitumen-free solids show that dosages higher than 250 g/t of DS0.7, DS1.2 and LM-CMC would be required to fully disperse the solids without any weak flocculation, the rheological flow curves obtained for the oil sand ore slurries show that a polymer dosage as low as 125 g/t was needed to substantially disperse the slurries (containing the same volume of solids) and to completely change the rheology of the slurries. Only the LM-CMC required a higher dosage (750 g/t) to fully disperse the oil sand slurries. Since a portion of the solids in the oil sand slurries was covered by bitumen, it could be argued that the surface area available for adsorption was much lower than on bitumen-free solids thus requiring lower polymer dosages for efficient dispersion of the slurries. However, bitumen–mineral interactions would still be present and the adsorption of the polymer only on portions not covered by bitumen would not necessarily prevent aggregation in the slurry. As will be discussed in later sections, adsorption of CMC on bitumen indeed appears to be very low, if any. The contribution of bitumen to the aggregation of oil sand slurries can be seen in Figure 4.15 A and B by plotting the ramping portions of the flow curves for slurries of first, oil sand ore and bitumen-free solid slurries and second, the same slurries but in the presence of CMC (Figure 4.15A: DS0.7 and DS1.2, Figure 4.15B: LM-CMC and HM-CMC). The removal of bitumen from the oil sand ore results in a decrease in the yield stress and apparent viscosity. However, the flow curve for the bitumen-free solids shows that a structure is still present even in the absence of bitumen suggesting the importance of small van der Waals attractive forces at small interparticle spacing in such concentrated slurries and attractive forces between the hydrophobic surfaces of the solids. The flow curves in which CMC was added to the oil sand ore and bitumen-free solids both display Newtonian behaviour with no yield stress. However, 130  Figure 4.15. Flow curves (ramping parts) for oil sand ore and bitumen-free solid slurries as well as oil sand ore/CMC and bitumen-free solids/CMC mixtures. The dosage of CMC is 250 g/t.  the apparent viscosity is always higher in the case of the flow curves for the oil sand ore/CMC mixture, obviously due to the presence of bitumen. An important observation from Figure 4.15 is that the rheological measurements were performed at 25 °C. At this temperature, the liberation of bitumen from solids can be expected to be very low even at pH 8.5, and as demonstrated by Gutierrez and Pawlik (2012a), who tested bitumen–quartz mixtures and obtained flow curves with marked thixotropy and yield stress at pH 8.5 and 25 °C, a combination of high pH and temperature was required for sufficient bitumen liberation from solids in the absence of a dispersing polymer. At pH 10 and a temperature of 50 C, when the liberation of bitumen from solids was high, Gutierrez and Pawlik obtained Newtonian flow curves very similar to those in Figure 4.15 in the presence of CMC. With addition of CMC to the oil sand ores, the flow curves exhibit a Newtonian flow behavior and thus do not indicate the presence of un-liberated bitumen in the slurries. The action of CMC is not only limited to the dispersion of solids but also possibly to enhancing the liberation of bitumen from the solids. This additional role of CMC is supported by visual observations during sedimentation tests that a layer of free bitumen formed as a result of CMC addition, and will be analyzed in Chapter 5.  131  4.4 Summary This chapter described the aggregation/dispersion state of oil sand ore and bitumen-free solid slurries. As well, the influence of carboxymethyl cellulose on the aggregation/dispersion of the slurries was investigated. The rheological behavior of the oil sand ore and bitumen-free solid slurries in the absence of CMC indicated that the oil sand components interacted to produce aggregated slurries. The aggregated nature of the slurries was confirmed through settling tests. Rheological measurements and settling tests on slurries in which CMC was added indicated that the polymers stabilized the slurries towards aggregation presumably by a combination of electrostatic and steric effects, despite the low adsorption densities of the polymers on the solids. The dispersing ability of the polymers was independent of the degree of substitution. At lower dosages (125 g/t and 500 g/t), the lower molecular weight polymer produced dispersed slurries that were susceptible to shear-induced weak flocculation. Higher dosages (750 g/t) were required to efficiently disperse the slurries. The addition of 250 g/t of the higher molecular weight polymer was effective in dispersing the particles and preventing aggregation. The different dispersing abilities of the low and higher molecular weight polymers were primarily related to the different adsorption densities of the polymers and also to some extent to the differences in their flexibility. The results suggest that the mechanism through which the polymers disperse the oil sand slurries is by adsorbing on the solid particles and preventing interparticle (mineral–mineral) interactions which prevents aggregation in the slurries. The adsorption of the polymers on the solids produces slurries without any yield stress indicating that mineral-bitumen interactions are also eliminated, thus the role of CMC is also related to liberation of bitumen from the solids. In Chapter 5, the interaction between CMC, solids, and bitumen is probed further to confirm the role of CMC in bitumen liberation.  132  CHAPTER 5  5 Interaction of carboxymethyl cellulose with solids and bitumen 5.1 Introduction In order to determine the role of CMC in bitumen–solid interactions, bitumen displacement from illite (representing solids) in the presence and absence of CMC was studied by measuring the static and dynamic contact angle of bitumen on an illite surface immersed in either distilled water or an aqueous CMC solution. The measurements were performed as a function of pH, time, and temperature, and in some cases, CMC concentration. These tests can be viewed as a simulation of the bitumen liberation process from solids in oil sands ores. Illite was chosen since it is mineralogically similar to muscovite, one of the major clay components of the oil sand ore tested in Chapter 4, and was reported to be detrimental to bitumen recovery at low pH and temperature due to its acidic nature (Ding et al., 2006). In addition, illite is a layered silicate mineral, is anisotropic and is therefore a good representation of the clay minerals found in oil sand ores. Since bitumen hydrophobicity is critical to its extraction from oil sand ores, CMC, though beneficial in preventing aggregation should not alter the wetting characteristics of the bitumen–water interface so as to depress the floatability of bitumen. Therefore, it was also necessary to determine the effect of CMC on the wettability of the bitumen surface. Such an interaction can also conveniently be studied through contact angle measurements of an air bubble attached to a bitumencoated glass surface immersed in an aqueous solution containing either a background electrolyte or CMC. The bitumen–air interactions were studied under dynamic conditions in which both static and dynamic contact angles, measured through the solution phase, were obtained by attaching an air bubble to a glass slide coated with bitumen, and monitoring the evolution of the air/solution/bitumen contact line with time under different physicochemical conditions.  133  5.2 Materials and methods 5.2.1  Materials  The polymers tested in this work were the same as those described in section 3.2.1. Stock solutions of 1 g/L were prepared as described in the same section. An appropriate amount of the stock solution was diluted to working concentrations of 100 mg/L. Dilutions were made using background electrolyte or distilled water. The glassware used in solution preparation and contact angle measurements were carefully cleaned with a piranha solution (1:1 mixture of 35% H2O2 and 36% HCl) to remove any organic contaminants that could introduce errors in contact angle measurements. Bitumen was extracted from a good processing oil sand ore obtained from Alberta Research Council. To obtain the bitumen, the oil sand ore and warm water (50 °C) were mixed in a beaker for 30 minutes in the absence of any additional reagents. The resulting bitumen froth was carefully scrapped from the top of the beaker to minimize solids carry-over to the froth. The froth was cleaned five times by such reagentless flotation until the fines content in the bitumen was less than 0.5% by mass. The obtained bitumen was considered clean enough for contact angle measurements. It should be noted that no solvents were used to clean the bitumen thus eliminating possible artifacts in contact angle measurements resulting from residual solvent left in the bitumen or from losses of the light components of bitumen. Illite rocks of different sizes were obtained from Ward’s Natural Science Establishment. It was easier to prepare specimens of illite for this work compared to kaolinite which is practically unavailable as large crystals that can be cut into pieces suitable for such tests. The illite pieces used for bitumen displacement tests were handpicked and cut into ca. 4 cm × 4 cm pieces. The pieces were rough and irregular and were thus dry-polished with 180-grit and 1500-grit silicon carbide papers until the surfaces became flat and smooth. The polished pieces were then cleaned using TEXMET polishing cloths (#40-7602, Buehler, USA) to remove abraded particles of illite generated by the polishing process. The illite pieces were used immediately after the cleaning process to avoid possible surface contamination.  134  Bitumen coated-glass substrates for bitumen/air bubble contact angle measurements were prepared using the following procedure. Rectangular glass slides (~4 cm × 4 cm) were cleaned with a 1:1 mixture of HCl and H2O2 for 30 minutes after which they were rinsed with distilled water and dried at 100 °C for 5 minutes on a hot plate to remove excess water from the surface of the glass substrates. The contact angle of a sessile water droplet on such cleaned glass slides was 18°.  A droplet of hot bitumen (60 °C) was then deposited onto the centre of the glass slide and the slide was quickly transferred to a 6800-series spin-coater (Specialty Coating Systems, Indianapolis, IN, USA) programmed to spin-coat the bitumen onto the slide at 4000 rpm for 3.5 minutes. The resultant bitumen-coated slides were stored in a dust-free environment at ambient temperature for 12 hours. After this storage time, the bitumen surface was visually checked and found to be free of fine solids. The surface was also observed under a microscope and determined to be smooth enough for contact angle measurements. 5.2.2  Bitumen displacement from illite  To determine the bitumen contact angle on illite, a droplet of hot bitumen at about 60 °C was placed on the illite surface using a syringe tip. The syringe tip was dipped in the hot bitumen and the attached bitumen droplet was quickly placed on the illite surface. Thus, the sessile drop method was used and the contact angle was measured through the bitumen sessile droplet as explained later. The amount of bitumen deposited on the illite surface was standardized by heating the bitumen to the same temperature in all tests and using the same syringe. When it was visually determined that the amount of bitumen deposited was lower or higher than usual, the process was repeated. Triplicate tests done using three different illite pieces showed that small deviations in the amount of deposited bitumen did not have a significant effect on the measured contact angles. The illite surface around the bitumen droplet was then wetted with a small amount of distilled water previously adjusted to the desired pH (6 and 8.5) and maintained at room temperature, usually about 25 °C. The illite surface with bitumen was then allowed to stand for 5 minutes during which the bitumen droplet spread to a low steady-state contact angle of ~30°. Finally, the illite specimen with the bitumen droplet was mounted on a glass stand and transferred to a quartz cell containing the test solution. Measurements with the use of precision fine wire thermocouples showed that the temperature inside the bitumen droplet decreased rapidly to that of the test solution within a few 135  seconds of introducing the illite sample into the solution. The position of the glass stand was quickly adjusted until a clear image of the illite and bitumen droplet was seen on the movie-capturing window of an FTA1000 Drop Shape Analyzer (First Ten Angstroms, Inc., Portsmouth, VA, USA). The schematic setup of the experiment is shown in Figure 5.1. The FTA1000 video system was then started and the change in the bitumen droplet shape over time was captured as a movie clip until steady-state conditions were attained (the bitumen droplet shape did not change with time). At the start of the test, images were obtained in increments of about 1 second in order to probe any physicochemical phenomena taking place during the early bitumen displacement stages. The time interval between the images was increased accordingly since the changes in droplet shape were not so pronounced as steady state was approached. The FTA32 software was used for image acquisition and processing. The software utilizes the edge finder algorithm to accurately locate the edge of the captured images and develop a profile of the images. The contact angle is determined by drawing a line tangent at the point of intersection of the baseline with the bitumen droplet profiles. In most cases, the captured images of the bitumen droplet were asymmetric, that is the contact angles at the left and right ends of the droplet were different, which necessitated individual image analysis of each movie frame by manually fitting the Laplace equation to obtain the static and dynamic contact angles. It should be noted that the contact angles were measured through the bitumen droplet, thus a higher angle indicates greater bitumen displacement from illite as shown in Figure 2.3 (section 2.1.3.1) and also Figure 5.1 by comparing the contact angle frames at 0 and t times. The effect of temperature on bitumen displacement from the illite surface was studied by performing similar tests at 40 °C. In the tests performed at higher temperature, the same procedures as described above were followed, the only modification being that the test solution was kept at the target temperature by placing the quartz cell containing the illite and bitumen droplet in a small bath through which hot water was circulated in order to maintain the temperature of the test solution in the quartz cell at 40° 1 °C.  136  Bitumen  Illite  Time = 0  FTA1000  Quartz cell containing test solution  Time = t Sample stage  Microscope with High speed video camera  Figure 5.1. Schematic showing the setup used for contact angle measurement during bitumen displacement from illite.  In another test, the CMC was preadsorbed on the polished illite surface prior to depositing bitumen on the surface. The polished illite was conditioned in a quartz cell containing a 250 mg/L CMC solution at room temperature and natural pH (~6) for 15 minutes. After conditioning, the illite was removed from the cell and allowed to dry for 5 min before depositing a hot bitumen droplet. The illite with the bitumen droplet was left to equilibrate for 5 minutes followed by testing on the FTA1000 Drop Shape Analyzer, as described above. A control test was also performed by preconditioning the illite in distilled water maintained at the same pH and temperature as in the tests with CMC. 5.2.3  Bitumen/air contact angle measurements  The contact angles of an air bubble attached to bitumen were measured by the captive bubble method using the same FTA1000 Drop Shape Analyzer as used in the bitumen displacement tests described above. The contact angles were measured by introducing a bitumen-coated glass slide into 137  a quartz cell containing the background electrolyte solution (0.01 mol/L NaCl) or 100 mg/L CMC at different pH conditions (pH 3 – 10.5). Contact angle measurements at different CMC concentrations (25 mg/L – 150 mg/L) indicated that the steady-state contact angle and the contact angle evolution rate were independent of polymer concentration whereas the initial contact angle slightly increased with increasing polymer concentration. A CMC concentration of 100 mg/L was considered sufficient to study differences in contact angles under different solution conditions. In one set of tests, the effect of calcium ions on bitumen–air interactions in the presence of CMC was investigated at calcium concentrations of 40 – 80 mg/L. In each test, the bitumen was allowed to equilibrate in the test solution for 5 minutes at ambient temperature (~25 °C)  after which an air bubble, generated from a highly accurate and automated glass microsyringe via a stainless steel dispensing needle, was introduced into the solution and left to equilibrate for 1 minute while still attached to the needle. The syringe pump and needle arrangement was then gently raised allowing the bubble to contact the bitumen surface. Simultaneously, the movie-capturing system of the FTA1000 was started enabling the acquisition of bubble images at different times until steady-state was reached. Figure 5.2 shows an example of an image captured and analyzed with the FTA32 software. The contact angles were measured at three (sometime four) different spots on the bitumen-coated slide keeping the same time interval at different spots so that three contact angle values could be obtained at each time on three different spots. The three measurements were averaged to obtained contact angles as a function of time for one bitumen coated slide. Triplicate measurements on three different bitumen coated slides were performed. In most cases, the measurement error in the contact angle was between 1° and 2° with the highest observed error being 3°. In order to determine the influence of the air bubble volume on the measured contact angles and the kinetics of the wetting process, different bubble volumes ranging from 2.5 µL to 7.5 µL were tested on a bitumen-coated glass slide and the kinetics of development of the steady-state contact angle were obtained as shown in Figure 5.3. It can be seen that the bubble volume had a marginal effect on the kinetics of development of the steady-state contact angle with the bubble of volume 7.5 µL exhibiting slower kinetics than the smaller bubbles. There was virtually no effect of bubble volume on the steady-state contact angle. To minimize any changes in the contact angle kinetics with bubble volume, all analyses were performed with a bubble volume of 5 µL. It should be noted that 138  Figure 5.2. Example of an image obtained with the movie-capturing system built into the FTA1000.  µ  Figure 5.3. Effect of bubble volume on contact angle evolution.  all contact angles reported were measured through the liquid phase. Contact angles measured in this way provide a better representation of the bitumen–air bubble attachment process in the oil sands flotation process.  139  A first order rate equation (Equation 5.1) was fit to the experimental data to describe the kinetics  θt = θmax – a·(exp(−kt))  ..[5.1]  of contact angle evolution and to quantify the spreading behaviour of the air bubble on the bitumen surface. At time t = 0, Equation 5.1 becomes 5.2 from which θ0 can be calculated once θmax and a are determined. θ0 = θmax – a  ..[5.2]  The contact angle data (θ versus t) were fitted with Equation 5.1 using the non-linear curve-fitting tool in Origin 8.5 software (Northampton, MA, USA). Excellent fits were obtained with the coefficient of correlation greater than 0.99. Interestingly, the effect of pH and CMC on bitumen wettability could be determined from the parameter k [s-1]. The highest experimental error in the values of k was 0.005 s-1. This high magnitude of error was only obtained in two cases with CMC of the highest DS (1.2) and MW (715,000 g/mol) at pH 8.5 and 10.5, respectively. The typical error was 0.003 s-1. The highest error in the initial and steady-state contact angles (3°) was also obtained for the CMC with the highest DS. This observation indicates that high experimental errors were only obtained in cases where the bitumen-coated slides were tested in solutions of high viscosity (high polymer DS or MW and high pH). It should be noted that k in Equation 5.1 has units of reciprocal time and can be viewed as a rate constant. The observed changes in the bitumen-bubble contact angles were mainly attributed to the changes at the bitumen–solution interface due to changes in pH and presence of CMC, and not to the leaching of organic materials from bitumen into the test solution. To confirm this, a bitumen coated glass slide was conditioned in the background electrolyte at natural pH (5.3) and pH 10.5 for 30 minutes with occasional stirring. An aliquot of the solution at each pH was then withdrawn from just below the bitumen surface and analyzed for total organic carbon content (TOC, Shimadzu) and absorbance using a UV-visible spectrophotometer (Cary50 Varian). At pH 5.3, the TOC was 0.59 mg/L and it slightly increased to 2.13 mg/L at pH 10.5 while there was no substantial difference in the absorbance of the two solutions. These results indicate that very little of organic materials were introduced into the solution during the time scale of the contact angle experiments. 140  In order to better understand the interaction between CMC and bitumen, other polymers with selected chemical properties were tested following the same procedures as explained above. Humic acids (low MW), hydroxypropyl cellulose (HPC, MW = 60,000 g/mol), and polystyrene sulfonate (PSS, MW = 70,000 g/mol) were used in these tests. A humic acids sample was obtained from Aldrich. Nonionic hydroxypropyl cellulose and strongly anionic polystyrene sulfonate were supplied by Polysciences. 5.3 Results and discussion 5.3.1  Bitumen displacement from illite  At the beginning, the displacement of bitumen from illite was monitored over a long period of time (~ 2 hours), but it became quite clear that differences in contact angles between the tests in the presence and absence of CMC occurred over much shorter timescales. Thus, contact angle analysis for all subsequent tests was limited to ~33 minutes which conveniently corresponds to the timescale of the rheological tests described in the previous chapter. The use of distilled water in the bitumen displacement tests is justified because most of the testwork was performed at a temperature of 25 °C at which it was shown that the role of water chemistry in bitumen recession from a solid substrate becomes less significant (Masliyah et al., 2011). Masliyah et al. showed that as temperature decreased, the time required for recession of bitumen from a microscope slide was independent of water chemistry at low temperatures. The water samples tested by the authors were deionized water and industrial process water at pH 8.2. 5.3.1.1 Effect of the molecular weight of CMC Figure 5.4 shows the contact angle evolution for the baseline cases, that is, bitumen contact angles measured on illite immersed in distilled water at natural pH (~6) and pH 8.5, and for the tests in which illite was immersed in CMC solutions of different MW at natural pH. The baseline case at pH 8.5 was included since pH 8.5 reflects conditions under which the rheological tests were performed. It should be noted that for each curve, about 170 images (and hence 170 contact angles) were analyzed during the period shown in the figure, but only every second data point is shown for clarity. The baseline data represent averages of four measurements while the CMC data are averages of two measurements, that is, each data point is an average of four contact angle measurements (baseline) 141  and two contact measurements (CMC data). The measurement errors (standard deviation) between the contact angle measurements at each data point were between 2° and 10° which is acceptable considering the surface roughness and possible chemical heterogeneity of the illite surface, as well as the charge heterogeneity of the bitumen surface.  120 110  Contact Angle [deg]  100 90 80 70 60 50  Distilled water: pH 6 Distilled Water: pH 8.5 LM-CMC: pH 6 MM-CMC: pH 6 HM-CMC: pH 6  40 30 20 0  400  800 1200 Time [s]  1600  2000  Figure 5.4. Evolution of the dynamic contact angle of bitumen on illite in distilled water (different pH) and CMC of same DS (0.7) but different molecular weights. The polymer concentration was 250 mg/L, temperature = 25 °C.  By comparing the two baseline cases (distilled water) , it can be seen that the evolution rate of the bitumen contact angle at pH 6 is higher than at pH 8.5 which was unexpected since increase in pH should promote repulsion between bitumen and illite. However, the experimental errors between four contact angle measurements at pH 6 and pH 8.5 after 2000 seconds were 6.8° and 7.1° respectively which makes the differences in contact angles at these two pH values not very significant. Generally, it is evident that the evolution of the bitumen contact angle for the cases where illite was immersed in CMC solutions is much faster than the baseline cases indicating that the presence of CMC increases the displacement rate of bitumen from the illite surface. In addition, the maximum contact angle after 2000 seconds is higher for the tests with CMC than in the baseline cases, which 142  indicates that bitumen detachment from the illite surface is favored in the presence of CMC. Particularly, the larger differences between the tests with CMC and distilled water at pH 8.5 (which reflects conditions under which rheological tests were performed) should be noted. In addition, noticeable differences between the three CMCs of different molecular weights can be seen after about 200 seconds. The LM-CMC displaces bitumen much faster than the higher molecular weight samples, but the differences disappear with time and after 2000 seconds, the final contact angle is virtually the same. Thus, bitumen displacement from illite occurs at a much faster rate, and to a greater extent in the presence of CMC which is beneficial for bitumen liberation from solid particles. Within the timescale of the measurements shown in Figure 5.4, the lower molecular weight CMC provides faster bitumen displacement rates at shorter times but the final maximum contact angle does not depend on the molecular weight. A simple explanation for the differences in the bitumen recession rates could be related to the viscosity of the polymer solutions. By comparison, the viscosity of a 100 mg/L solution of HM-CMC at pH 7 is 10% higher than the viscosity of a solution of LM-CMC of similar concentration. Thus, the recession of bitumen from the illite surface and the simultaneous spreading of the polymer solution along the illite surface could be expected to be slowed down more in the case of the higher viscosity solution of the HM-CMC than in the case of the MM-CMC and LM-CMC. Another possible explanation for the results in Figure 5.4 is that the LM-CMC adsorbs faster on illite than the higher molecular weight polymers owing to its smaller chain dimensions and faster diffusion towards the surface, which explains the higher contact angles in the initial stages of bitumen displacement. It should be noted that tests described above were obtained under quiescent conditions. Under shearing conditions typically encountered during bitumen extraction, any slight increase in the bitumen droplet contact angle as seen in the presence of CMC should result in higher bitumen liberation. Figure 5.5 shows that increasing the pH of CMC to 8.5 affects the higher molecular weight polymers (MM-CMC and HM-CMC) more than the LM-CMC, as seen from the lower bitumen contact angles at pH 8.5. In the presence of HM-CMC (Graph C), the bitumen contact angle evolution curve is essentially the same as the curve for distilled water at pH 8.5 indicating that the presence of HM-CMC did not have any influence on bitumen recession from illite at pH 8.5 despite the higher adsorption density of the HM-CMC which was about 1.5 times higher than that of the LM-CMC (Table 4.1, Section 4.3.3.1, Chapter 4). The slightly extended configuration of the higher molecular weight polymers at high pH (in comparison to the LM-CMC) and the presence of very 143  high molecular weight fractions in the HM-CMC, as seen in its molecular weight distribution (Figure 3.4A, section 3.31), mean that the HM-CMC chains are most likely anchored onto the illite surface through very few trains with long tails extending into solution. This type of configuration is ineffective in displacing bitumen from the illite surface at high pH.  Figure 5.5. Effect of pH on the evolution of the dynamic contact angle of bitumen on illite in the presence of LM-CMC (Graph A), MM-CMC (Graph B) and HM-CMC (Graph C). The polymers are of the same DS (0.7). The polymer concentration was 250 mg/L, temperature = 25 °C. All tests were performed in distilled water.  144  The preceding discussion leads to the mechanistic model presented in Figure 5.6. The presence of the LM-CMC promotes bitumen displacement from the illite surface due to the small effective size and higher flexibility of the chains, which allow the polymer chains to very closely approach the three-phase point of contact between the illite, water, and bitumen. Once the polymer chains have penetrated into the illite/bitumen interface, the recession of the bitumen is driven by repulsive forces between the negatively charged polymer, illite and bitumen surfaces. On the other hand, the presence of the more rigid HM-CMC is not as effective as the LM-CMC in promoting bitumen displacement from the illite surface except at longer displacement times at pH 6. The larger size and higher rigidity of the chains, and the higher viscosity of the solution slow down the diffusion of the polymer to the bitumen–illite interface, and the displacement of bitumen from the illite surface is driven only by the higher affinity of hydrophilic illite towards the aqueous phase which results in a final contact angle similar to the baseline case.  Figure 5.6. Schematic representation of the role of LM-CMC and HM-CMC in bitumen displacement from illite at pH 8.  The present results and the results presented in the previous chapter suggest that the lower molecular weight polymer (LM-CMC) disperses oil sand slurries by accelerating the liberation and detachment of bitumen from solids rather than just by dispersing the solids. On the other hand, at pH 8.5, the higher molecular weight polymer (HM-CMC) disperses the oil sand slurries more by dispersing the solids through a steric mechanism than by accelerating the liberation of bitumen from the solids. At a lower pH of 6, the higher molecular weight polymers would also accelerate bitumen liberation and detachment from the solids.  145  5.3.1.2 Effect of the degree of substitution of CMC Figure 5.7 shows the effect of the degree of substitution on the bitumen contact angle evolution. For comparison, the baseline cases are also plotted. Clearly, the presence of CMC leads to higher bitumen contact angles hence higher bitumen displacement from illite compared to the baseline cases. The effect of DS is not as pronounced in the initial stages of bitumen displacement as that seen with the polymers of different MW. Slight differences between contact angles in the presence of the lower DS samples (0.7 and 0.9) and higher DS sample (1.2) are seen at higher bitumen displacement times (> 1000 seconds), with the lower DS polymers showing slightly higher bitumen contact angles. However, the differences are within the experimental errors and are therefore not statistically significant. Overall, all the polymers accelerate the recession of bitumen from the illite surface. In the previous chapter, it was demonstrated that the polymers prevent aggregation in oil sand slurries by dispersing solids in the slurry. The present results strongly suggest that the additional role of all the polymers in dispersing oil sand slurries is to enhance the liberation of bitumen and detachment from the solids. 120 110  Contact Angle [deg]  100 90 80 70 60 50  Distilled Water: pH 6 Distilled Water: pH 8.5 DS0.7 DS0.9 DS1.2  40 30 20 0  400  800 1200 Time [s]  1600  2000  Figure 5.7. Evolution of the dynamic contact angle of bitumen on illite in distilled water (different pH) and CMC (pH 6) of similar MW (293,000 – 310,000 g/mol) but different degrees of substitution. The polymer concentration was 250 mg/L, temperature = 25 °C.  146  Figure 5.8 shows the effect of temperature on bitumen displacement from illite in the presence and absence of CMCs of different DS. A dramatic difference in the kinetics of bitumen displacement from illite is noted at higher temperature. For example, it takes less than 150 seconds at 40 C to achieve the same bitumen contact angles as those obtained after 2000 seconds at 25 °C (Figure 5.7). However, at the higher temperature of 40 °C, the beneficial effect of CMC addition is still visible but not as much as at 25 °C. In this case, temperature seems to be the main driving force for bitumen displacement and detachment from illite. However, it should be emphasized that the tests were carried out under quiescent conditions and any improvement in bitumen displacement such as that seen in Figure 5.8 could have significant kinetic implications to the bitumen liberation process. Such a small improvement in bitumen displacement can result in appreciably higher bitumen liberation under industrial mixing conditions. 160  Contact Angle [deg]  140 120 100 80 60  Distilled Water: pH 6 Distilled Water: pH 8.5 DS0.7: pH 6 DS0.9: pH 6 DS1.2: pH 6  40 20 0  300  600 900 Time [s]  1200  1500  Figure 5.8. Evolution of the dynamic contact angle of bitumen on illite in distilled water (different pH) and CMC of similar MW (293,000 – 310,000 g/mol) but different degrees of substitution. The polymer concentration was 250 mg/L, temperature = 40 °C.  5.3.1.3 Effect of preadsorbing CMC onto illite In order to show that it was indeed CMC adsorption on illite that accelerated bitumen displacement from illite, a piece of illite was preconditioned in distilled water at pH 6 and in a 147  solution of CMC (DS0.7) at pH 6, followed by bitumen displacement tests, as described earlier. Figure 5.9 shows the bitumen contact angle evolution for an illite sample on which CMC was preadsorbed, and for an illite specimen conditioned in distilled water. In comparison to previous baseline cases shown in Figures 5.4 and 5.7, bitumen displacement from the illite preconditioned in distilled water proceeded much faster. This can be explained by the weak adhesion of bitumen onto the illite due to adsorption of water molecules on the external surfaces of illite. Hatch et al. (2012) recently demonstrated that adsorption of water can occur on the external surfaces of illite and that desorption of the adsorbed water molecules by drying takes a long time. Thus, it can be concluded that water adsorbed on the illite surface and thus the hydration of the illite surface was responsible for the faster bitumen displacement rate relative to the earlier baseline cases in which illite was not preconditioned in the test solution.  Figure 5.9. Evolution of the dynamic contact angle of bitumen on illite preconditioned in distilled water (baseline) and in CMC of DS 0.7, MW 295,000 g/mol (in order to preadsorb CMC on illite prior to the bitumen displacement test). The polymer concentration was 250 mg/L, temperature = 25 °C.  It is quite evident that preadsorbing CMC on illite still results in much faster bitumen displacement. The error bars on the initial contact angle data points indicate that the difference between the initial bitumen contact angles for the two cases is not significant. However, very small 148  errors were obtained at higher bitumen displacement times as shown by the small error bar at the final data point for the illite on which CMC was preadsorbed, indicating that the difference in the final bitumen contact angles is statistically significant. With the CMC preadsorbed on illite, in addition to the faster bitumen displacement kinetics, the high bitumen contact angle (~150°) indicates that it would be easier to detach bitumen from illite compared to the lower contact angle in the baseline case. It is remarkable to note that the bitumen kinetics and final bitumen contact angle obtained by preadsorbing CMC on illite at 25 °C are comparable to the data obtained at a much higher temperature (40 °C) (Figure 5.8) which clearly demonstrates the beneficial role of CMC in bitumen displacement (and detachment) from illite at low temperature. The above results suggest that the role of CMC is to modify the interaction between illite and bitumen. Once CMC is preadsorbed on illite, the deposited bitumen droplet does not directly interact with the illite surface which results in very low illite-bitumen adhesion forces which favor faster bitumen displacement and detachment. The results presented in the previous sections indicate that CMC disrupts the three-phase contact line at the illite/bitumen/aqueous interface by adsorbing on illite and facilitating faster displacement of bitumen from illite. Under these conditions, illite preferentially interacts with CMC rather than with bitumen. 5.3.2  Bitumen/air contact angle measurements  In the previous sections, it was demonstrated that bitumen displacement and detachment can be enhanced by CMC suggesting that bitumen liberation from oil sand solids would increase in the presence of CMC. This was related to the action of the polymers in preventing aggregation in oil sand slurries. In industrial operations, bitumen liberation is followed by aeration in which bitumen is attached to air bubbles, transported to the top of the slurry, and recovered as a froth product. At all stages of bitumen aeration and recovery, the high bitumen surface hydrophobicity, critical for effective bitumen–air attachment, should be maintained and should not be diminished by any added reagents. The present section presents the effect of the tested polymers on the hydrophobicity of bitumen, as determined from measurements of contact angles of an air bubble attached to bitumen. 5.3.2.1 Contact angle data analysis Figure 5.10 shows an example of a typical contact angle versus time plot for an air bubble in contact with the bitumen surface. Every second data point is shown for clarity. The kinetics of the 149  experimental contact angle data were described by fitting a three-parameter equation to the data (see experimental section). The steady-state contact angle (θmax) and the contact angle evolution rate constant k were directly read from the fitted equation while the initial contact angle, θ0 was determined by solving the fitted equation with t = 0. Because of the excellent fits to the data, it should be noted that θ0 and θmax determined from the fitted equation were very similar to the initial and final experimental data points, respectively. 5.3.2.2 Effect of pH for polymers of different degrees of substitution Figure 5.11 shows the initial and steady-state contact angles for the baseline case and CMCs of different DS. It should be noted at this point that the determination of true contact angles on surfaces such as the tested bitumen-coated glass slide can be complicated by surface roughness, chemical and charge heterogeneity (Xu et al., 1996; Lin, 2012; Drelich et al., 2007). Therefore, further analysis of the data will focus on changes in contact angles as a result of changes in solution conditions, primarily pH, and polymer properties (DS, MW).  Figure 5.10. Evolution of the contact angle of an air bubble in contact with a bitumen-coated glass slide immersed in 0.01 mol/L NaCl. Temperature = 25 °C, pH 8.5. The solid line shows the fit to the experimental contact angle data.  150  Figure 5.11. Effect of pH on initial (θ0) and steady-state (θmax) contact angles for the baseline (0.01 mol/L NaCl) case and CMC polymers of different DS. Temperature = 25 °C.  A few trends can be noted in Figure 5.11. Firstly, there is a slight but measurable effect of pH on the initial contact angle in the baseline case. The contact angle decreases from ~50° at pH 3 to ~42° at pH 10.5. The contact angle measurement errors were surprisingly very small. The highest observed measurement error in the contact angle was 3°, therefore, the difference in the initial contact angle is statistically significant. This suggests that it becomes slightly harder to attach an air bubble to a bitumen surface with increasing pH. Once the air bubble is attached and allowed to interact with the bitumen surface, very high steady-state contact angles are measured in the pH range 3 – 8.5 indicating high hydrophobicity of the bitumen surface, but a large decrease is observed at pH 10.5. Liu et al. (2005b) observed a similar effect of pH on the surface hydrophobicity of bitumen. The authors attributed the sharp drop of the contact angle at pH 10.5 to the higher ionization of functional groups of macromolecules on the bitumen surface resulting in stronger electrostatic repulsions between the ionized groups. Since an air bubble has a rather low iso-electric point (1.5 – 2.5) (Nguyen and Schulze, 2004), its surface is highly negatively charged at high pH. Thus, the repulsive electrostatic forces between the similarly charged air bubble and bitumen surfaces are responsible for the weak adhesion between the two surfaces. In addition, Liu et al. (2005b) measured colloidal forces between two bitumen surfaces and fitted the resulting force profiles with the 151  extended DLVO theory by including the hydrophobic force. The fitted hydrophobic force constant (which indicates the magnitude of the hydrophobic force between bitumen surfaces) followed the same trend as the contact angle, decreasing dramatically at high pH and confirming the decreasing hydrophobicity of bitumen at high pH. The results in Figure 5.11 indicate the detrimental effect of high pH on bitumen aeration. In addition to the contributing factors discussed above, the bitumen–water interfacial tension decreases at pH values higher than ~10.5 as a result of natural surfactant accumulation at the bitumen–solution interface (Wang et al. 2010). If the natural surfactant molecules migrate into the aqueous phase, they can also decrease the water–air surface tension as demonstrated by Wang et al. (2010). Both the decrease of bitumen–water interfacial tension and water-air surface tension are unfavourable for bitumen aeration. Conversely, it is well established that an increase in pH promotes bitumen liberation from solids (Basu et al., 1996) though this assertion seems to hold more for high temperatures (>40 °C) and not so much for low temperatures, e.g. 25 °C, as shown in Figure 5.4, section 5.3.1.1 (low temperature) and Figure 5.8, section 5.3.1.2 (high temperature). This conflict in pH requirements (bitumen liberation versus bitumen aeration) is a reason why current hot waterbased oil sands operations operate at pH values between 8 – 8.5 (Masliyah et al., 2011). Within this pH range, bitumen still retains sufficient hydrophobicity to achieve high bitumen recovery as seen from the contact angle measurements in this dissertation and from the study by Liu et al. (2005b). The second trend to note in Figure 5.11 is the effect of CMC. At pH 3, there is a small but statistically significant difference in the initial contact angles between the baseline case and the cases in the presence of CMC. The contact angle decreases from ~50° (baseline case) to ~38° in the presence of CMC. At higher pH values, the effect of CMC on the initial contact angles disappears indicating that the polymers do not affect the wetting of bitumen by water at higher pH. There is no significant effect of the degree of substitution on either the initial contact angles or steady-state contact angles. However, as can be seen in Figure 5.12, plotting the contact angle evolution from the time of bubble attachment to about 10 seconds after the attachment reveals that the higher DS sample (DS1.2) slows down the contact angle evolution rate in comparison to the baseline case, DS0.7, and DS0.9, and that this effect is more pronounced at pH 3 than at higher pH values. It is unlikely that the slowing down of the contact angle evolution rate to a greater extent by the DS1.2 polymer is due to the higher viscosity of the polymer solutions. As the viscosity data show, a 100 152  Figure 5.12. Evolution of the contact angle of an air bubble in contact with a bitumen-coated glass slide immersed in 0.01 mol/L NaCl and CMC of different DS. Data during the first 10 seconds following air bubble-bitumen contact is shown. Temperature = 25 °C, pH = 3, 6, 8.5 and 10.5.  mg/L solution of the DS1.2 has a viscosity only about 2% higher than the viscosity of a DS0.7 solution of similar concentration. It can also be seen that the contact angle evolution rates for the baseline case (with only 0.01 mol/L NaCl) and for the polymers are practically the same at high pH values of 8.5 and 10.5, where the polymer solutions are more viscous than at low pH, which shows that changes in the contact angle evolution rate are not related to the viscosity of the polymer solutions. The plots in Figure 5.12 thus show that there is a somewhat stronger interaction between 153  bitumen and the DS1.2 polymer relative to the lower DS polymers, but it should be noted that this interaction is rather weak as the steady-state contact angle for the DS1.2 is not much different from that of the lower DS polymers. 5.3.2.3 Effect of pH for polymers of different molecular weight Very similar trends with pH, as observed with polymers of different DS, are observed for polymers of different molecular weights as can be seen in Figure 5.13 and Figure 5.14. The measurement error for the initial contact angle data points in Figure 5.13 ranges between 0.2° and 2° which makes the differences observed, particularly at pH 3, statistically significant. At pH 3, the initial contact angle decreases from ~50° to ~35° in the presence of the LM-CMC and decreases further to 27° in the presence of the HM-CMC. At higher pH values, the initial contact angles slightly decrease in the presence of CMC but the effect of molecular weight is rather weak. Figure 5.14 further shows that the polymers slow down the evolution of the contact angle in the early stages of bubble-bitumen interaction. A visible effect of molecular weight can be seen at pH 3, a weaker effect can be observed at pH 6, and it disappears at pH 8.5. As noted with the CMCs of different DS, the steady-state contact angles are not affected by changes in pH or polymer MW indicating that the effects on the initial contact angle and on the movement of the three-phase contact line in the early stages of bubble-bitumen interaction are weak.  154  Figure 5.13. Effect of pH on initial (θ0) and steady-state (θmax) contact angles for the baseline (0.01 mol/L NaCl) case and CMC polymers of different MW. Temperature = 25 °C.  Figure 5.14. Evolution of the contact angle of an air bubble in contact with a bitumen-coated glass slide immersed in 0.01 mol/L NaCl and CMC of different MW. Data during the first 10 seconds following air bubble–bitumen contact is shown. Temperature = 25 °C, pH = 3, 6, and 8.5.  5.3.2.4 Effect of pH on the contact angle evolution rate (from θ0 to θmax) Finally, the effect of pH on contact angle evolution from the moment of bubble-bitumen contact to the steady-state contact angle as a function of pH is shown in Figure 5.15 for polymers of 155  different DS and Figure 5.16 for polymers of different MW. Also plotted are data points for two tests in which calcium was added to a CMC solution to investigate whether it would enhance CMCbitumen interactions or not. The most evident trend from Figures 5.15 and 5.16 is that the presence of CMC (regardless of DS or MW) slows down the evolution of the contact angle at all pH values. To more clearly illustrate the magnitude of differences in k, two examples of the raw data for LM-CMC at pH 3 and pH 8.5 are plotted in Figure 5.17. Each curve contains data points averaged from duplicate tests. Each test is an average of measurements taken at four different spots on the bitumencoated slides. At pH 3, the difference in k is 0.027 s-1 while the difference at pH 8.5 is 0.020 s-1. From these two examples, the differences in k in Figures 5.15 and 5.16 can easily be seen. Overall, it can be seen that the addition of CMC slows down the development of the steady-state contact angles which shifts the k curves to lower values. However, the steady-state contact angles are not affected by CMC (as explained before) suggesting that the interactions between CMC and the bitumen surface, which account for the observed decrease in k values, are very weak.  Figure 5.15. Contact angle evolution rate, k of an air bubble in contact with bitumen as a function of pH. Data sets used to obtain k are from the time of initial contact of the air bubble with bitumen to the steadystate contact angle. Baseline case (0.01 mol/L NaCl) and CMCs of different DS are shown. Temperature = 25 °C.  156  Figure 5.16. Contact angle evolution rate, k of an air bubble in contact with bitumen as a function of pH. Data sets used to obtain k are from the time of initial contact of the air bubble with bitumen to the steadystate contact angle. Baseline case (0.01 mol/L NaCl) and CMCs of different MW are shown. Temperature = 25 °C.  Figure 5.17. Illustration of the differences in the evolution of the air bubble contact angle rate k by plotting experimental contact angles at different times. Left graph: pH 3. Right graph: pH 8.5.  Figure 5.15 shows that the k values are more sensitive to changes in DS than the initial or steadystate contact as pH is increased from 3 to 10.5. It can be seen that the DS1.2 is most affected by the 157  change in pH and the DS0.7 is least affected. The DS1.2 interacts more with the bitumen surface at pH 3 but as pH increases, the bitumen surface becomes more and more electrostatically charged, as does the DS1.2 and electrostatic repulsion reduces the interaction between the polymer and the bitumen surface and at pH 10.5, the measured k value is the same as the baseline case. It can also be noticed in Figures 5.15 and 5.16 that a maximum in k appears at pH 6 for the baseline case and for most of the polymers. This maximum indicates that bitumen is most hydrophobic at pH 6 which explains the highest measured spreading rate of the air bubbles over the bitumen surface. It is important to note that the k values are more indicative of bitumen surface hydrophobicity rather than the initial or steady-state contact angles which do not change much in the pH range from 3 to 8.5. Interestingly, the maximum in k disappears in the presence of two polymers (DS0.7 and HMCMC) that decrease the initial contact angle the most at pH 6. The DS0.7 polymer has the biggest effect on the early stages of the contact angle evolution (first 10 seconds) as can be seen in Figure 5.12 (pH 6 graph), while the HM-CMC polymer decreases the initial contact angle as well as the contact angle evolution rate (first 10 seconds) more than the LM-CMC (Figure 5.13 and Figure 5.14, pH 6). It is likely that the least anionic/charged DS0.7 and the highest molecular weight HM-CMC interact more strongly with bitumen at pH 6 than the other polymers. In the case of DS0.7, the lower charge on the molecule facilitates polymer adsorption on bitumen. In the case of HM-CMC, the polymer is also weakly anionic with a DS of 0.7 and it possess the highest molecular weight among the tested samples. Both the low charge of the polymer and the high molecular weight should promote adsorption of CMC on the bitumen surface. Compared to the LM-CMC, the higher ability of the HM-CMC to further reduce the spreading rate of air bubbles on the bitumen surface at pH 6 when bitumen is most hydrophobic can be attributed to its higher molecular weight which allows it to adsorb at the bitumen-aqueous solution interface with a higher segment density and reduce the spreading rate of the air bubble on the bitumen surface more than the LM-CMC. With regard to the effect of calcium and whether it enhances the interaction between CMC and bitumen, it can be seen in Figure 5.15 that the opposite effect is obtained for the case where 80 mg/L of calcium (in the form of calcium chloride) was added to the DS1.2 CMC at pH 8.5. The k value 158  increased to almost the baseline value indicating that the interaction observed with only DS1.2 in solution disappeared with the addition of calcium. Most likely, the added calcium was bound by the carboxylate groups on the highly dissociated CMC instead of forming an active adsorption site on the bitumen surface. However in the case of the LM-CMC (Figure 5.16), a significant decrease in k was observed at pH 6 in the presence of 40 mg/L calcium suggesting that the polymer-bitumen interactions were enhanced. This can be explained by the fact that LM-CMC has a lower degree of substitution hence possesses fewer carboxylate groups with which calcium can react. The remaining small amount of unbound calcium was then attached to the bitumen surface facilitating polymerbitumen interactions and reducing the contact angle evolution rate but not sufficient to significantly reduce the initial or steady-state contact angles. The initial and steady-state contact angle values in the presence of only LM-CMC were 96.1° and 42.6° respectively and the values decreased to 92.0° and 41.7° respectively after the addition of 40 mg/L calcium. From the results above, it is clear that the CMC-bitumen interactions that bring about significant changes in bitumen hydrophobicity occur at pH 3. It is also clear that these interactions affect the initial attachment of the air bubble onto the bitumen surface but do not affect the steady-state contact angle. Once attachment is established, the first 10 seconds of the three-phase contact line movement indicate that the higher DS and higher molecular weight samples slow down the contact angle evolution in comparison to the lower DS and LM-CMC samples. Again, this effect is more pronounced at pH 3 and is not observed at the highest pH tested. When the entire timescale of the contact angle evolution is analyzed, it becomes clear that the presence of CMC generally reduces the contact angle evolution rate at all pH values, but no clear trends in the initial and steady-state contact angle values can be attributed to the different polymer properties. Overall, the contact angle results for CMC indicate that none of the tested samples permanently changes the wettability of the bitumen surface. The final contact angle on the bitumen surface is not affected by the molecular weight or the degree of anionicity of the polysaccharide. However, all the CMC samples affect the kinetics of the evolution of the final contact angle value. In the presence of CMC the spreading rate of air bubbles is significantly lower than in a background electrolyte solution alone. Perhaps the simplest mechanism that could be proposed to explain that kinetic effect is due to the higher viscosity of polymer solutions. As the viscosity data show, a 100 mg/L solution of the highest MW sample has a viscosity about 25% higher than the viscosity of the background 159  electrolyte. It could be expected that the spreading of an air bubble and the simultaneous receding flow of solution along the bitumen surface should be slowed down by the higher viscosity of the solution. However, there is no clear trend in the wettability with the molecular weight of the samples, or with the viscosity of solutions of the tested samples. In order to further probe the kinetic effect of CMC on the contact angles, a set of tests was performed using three polymers with selected chemical properties. Humic acids, hydroxypropyl cellulose (HPC), and polystyrene sulfonate (PSS) were used in these tests. Humic acids are known to be powerful depressants of bituminous coal flotation (Laskowski, 2001; Pawlik, 2002; Pawlik, 2005), they were also shown to affect the extractability of bitumen from oil sands (Gutierrez and Pawlik, 2012b). Humic acids are strongly anionic and they contain both aliphatic and aromatic hydrocarbons, and such structures are also very common in bitumen. Hydroxypropyl cellulose (HPC) is very similar to CMC. However, the carboxymethyl (anionic) group of CMC is replaced by a hydroxypropyl (nonionic) hydrocarbon group in HPC. Also, the hydroxypropyl group is more hydrophobic than the carboxymethyl group. Finally, PSS is strongly anionic and is entirely composed of aromatic styrene rings basically identical with the aromatic structures of bitumen. In contrast to CMC and HA, the presence of sulfonate groups in PSS makes PSS a strong electrolyte whose dissociation is not affected by pH. It should also be mentioned that all three polymers are of molecular weights lower than the molecular weight of the CMC samples, and the concentration of these polymers, 50 mg/L, was also much lower than the concentration of CMC. The wettability results for these three polymers are shown in Figures 5.18 and 5.19 A and B. In contrast to CMC, all three polymers much more strongly affect the initial and final contact angle values on the bitumen surface. Therefore, a permanent change in the wettability of the bitumen surface occurred in the presence of these polymers particularly at lower pH. In these three cases, it is very difficult to argue that solution viscosity had any role in this effect. Therefore, the kinetic effect of CMC is caused by some form of weak interaction between the polymer and the bitumen surface. The effect of HPC is particularly noteworthy, since in this case the replacement of a short, anionic carboxymethyl group with a longer, uncharged, and hydrophobic propyl group makes HPC a powerful modifier of bitumen wettability. It seems that the HPC chains more strongly attach themselves to the bitumen surface, and the attachment most likely proceeds through hydrophobic  160  Figure 5.18. Effect of pH on initial and steady-state contact angles for the baseline (0.01 mol/L NaCl) case, polystyrene sulfonate, humic acids, and hydroxypropyl cellulose. Polymer concentration = 50 mg/L. Temperature = 25 °C.  Figure 5.19. Contact angle evolution rate, k of an air bubble in contact with bitumen in the absence and presence of different polymers. A: Baseline, polystyrene sulfonate, humic acids and hydroxypropyl cellulose as a function of pH (polymer concentration = 50 mg/L). B: Hydroxypropyl cellulose as a function of polymer concentration at pH 6. Data sets used to obtain k are from the time of initial contact of the air bubble with bitumen to the steady-state contact angle. Temperature = 25 °C.  161  forces between the side propyl group of HPC and the hydrophobic bitumen surface, while the main cellulose chain of HPC renders the bitumen surface hydrophilic. It is also important to observe that higher concentrations of HPC almost completely prevent any spreading of air bubbles over the bitumen surface (Figure 5.19B) from an initial contact angle of 32.4°. The anionic polymers, HA and PSS, permanently change the wettability of bitumen at lower pH. The effect of humic acids can be explained by two factors. At lower pH, HA become less anionic as a result of the association of acidic groups, while the bitumen surface become less negatively charged due to the same association of surface acidic groups (note that the i.e.p. of bitumen is at pH 2-3). These simultaneous processes basically eliminate electrostatic repulsion between HA and the bitumen surface and the polymer can more easily adsorb on the bitumen surface. Even though the same association phenomena can be expected to take place in the CMC-bitumen system, it is the presence of aromatic and aliphatic groups in HA that seem to allow the polymer to more strongly interact with bitumen compared to the very weak interaction of CMC. The PSS results also highlight the importance of aromatic structures in interactions between the polymer and the bitumen surface. The dissociation and anionicity of PSS is not affected by pH, so at low pH values PSS remains highly charged, while the bitumen surface is nearly uncharged. Under these conditions, electrostatic repulsion is very weak and PSS adsorption seems to be facilitated by the aromatic rings of PSS and the aromatic rings of bitumen. At higher pH values, electrostatic repulsion prevents interactions between PSS and bitumen, and the wettability of bitumen is not affected by the polymer. The above comparative analysis strongly suggests that the kinetic effect of CMC on the wettability of the bitumen surface is caused by very weak interaction/adsorption of CMC with/on bitumen. This interaction is not strong enough to permanently change bitumen wettability. However, other types of polymers (including modified polysaccharides), containing functional groups more chemically compatible with the bitumen surface, are capable of rendering bitumen hydrophilic. From this point of view, it can be concluded that CMC is not capable of strongly interacting with the bitumen surface through hydrophobic forces in a way similar to interactions between HPC and bitumen.  162  There is very limited information on the adsorption of polymers on bitumen. Recently, Wang (2012) measured adsorption of two cationic polymers and one anionic polyacrylamide on bitumen using quartz crystal microbalance with dissipation and found that the mass uptake of the cationic polymers was higher than that of the anionic polymer. The cationic polymers were Fe and Al hybrids of polyacrylamide (Fe-PAM and Al-PAM). However, this conclusion is slightly misleading since different polymer concentrations were used in the adsorption tests. The initial polymer concentration for the cationic polymers was 100 mg/L while it was only 25 mg/L for the anionic polymer and so the observed differences in the adsorbed amounts could be due to differences in the initial polymer concentration. However, there were differences in the adsorbed amounts for Fe-PAM and Al-PAM with the former showing higher affinity for bitumen. This was explained by the higher adsorptive capacity of Fe3+ due to its lower hydrated ionic radii relative to Al3+. The study by Wang (2012) suggests that electrostatic charge neutralization was the mechanism through which the cationic polymers adsorbed on bitumen. It is still noteworthy that some adsorption was measured for the anionic polymer on the negatively charged bitumen surface at pH 8.5. The adsorption was attributed to hydrogen bonding. Such a type of adsorption can be facilitated by the presence of surface oxygen sites on bitumen which would form hydrogen bonds with the hydrogen atoms on amide units of polyacrylamide molecules (Long et al., 2006). Another explanation for the adsorption of negatively charged polymers on bitumen is the presence of positively charged metal sites on bitumen. It is known from several adsorption studies on solid surfaces such as coal, graphite or quartz that the presence of active metal sites enhances polysaccharide adsorption through chemical complexation mechanisms (Liu and Laskowski, 1989a; Liu and Laskowski, 1989b, Solari et al., 1986; Cuba-Chiem et al., 2008). The presence of such sites on bitumen would facilitate chemical complexation reactions with the negatively charged polymer resulting in strong binding of the polymer to the bitumen surface. As discussed above, the present results indicate a rather weak interaction of CMC with bitumen thus chemical interaction with metal sites on the bitumen surface can be ruled out. It is also reasonable to expect that contamination of the tested bitumen with any metallic impurities was insignificant, and an effort was made during bitumen preparation to remove fine mineral particles. Rather, the interaction between CMC and bitumen is more likely through a weak form of hydrogen bonding between hydroxyl groups on the polymer chain and similar polar groups on the bitumen 163  surface. It should be noted that the amount of oxygen in bitumen is only about 1% by weight (Masliyah et al., 2011). This small amount would limit the extent of the hydrogen bonding. 5.3.3  Summary  Contact angles of a bitumen droplet in contact with illite indicated that the presence of CMC accelerated the displacement or recession of bitumen from the illite surface under conditions of low pH (6) and low temperature (25 °C). The polymers with different DS (DS0.7, DS0.9 and DS1.2) all accelerated bitumen displacement from illite but the effect of DS was insignificant. The lower molecular weight polymer (LM-CMC) caused a faster bitumen displacement rate than the HM-CMC but the final measured contact angle did not depend on the MW. This result meant that both polymers would eventually promote liberation/detachment of bitumen from the illite surface. An increase in pH from 6 to 8.5 did not affect the kinetics of bitumen displacement from illite or the final bitumen contact angle for the LM-CMC. However, an increase in pH in the presence of the HM-CMC slowed down the bitumen displacement rate and reduced the final bitumen contact angle to that of the baseline case. The differences in results obtained with the LM-CMC and the HM-CMC were related to the effective size and flexibility of the polymer chains. The LM-CMC promoted bitumen displacement more than the HM-CMC due to its small effective size and higher flexibility which allowed it to very closely approach the three-phase point of contact between the illite, water, and bitumen. At higher temperature, the beneficial effect of the polymers was not as evident as at 25 °C suggesting that at 40 °C, temperature is the main driving force for bitumen liberation from illite. However, the slightly higher bitumen contact angles observed in the presence of the polymers at high temperature could prove beneficial for bitumen liberation under application of shear compared to the quiescent conditions under which the present results were obtained. Preadsorbing CMC on illite provided even faster bitumen displacement from illite and the very high bitumen contact angles indicated that it would be easier to detach bitumen in the presence of CMC. From these results, it was suggested that the additional role of CMC in dispersing oil sand slurries as experimentally determined in Chapter 4 was to accelerate the liberation and detachment of bitumen from solids. Contact angles of an air bubble in contact with a bitumen surface indicated that the presence of CMC led to a decrease in the initial contact angles but this decrease was more significant at pH 3 164  than at higher pH values. The HM-CMC polymer noticeably decreased the initial contact angle more than the LM-CMC did at pH 3. There was no effect of the polymers on the steady-state contact angles. Within the first 10 seconds of contact of the air bubble with bitumen, an effect of DS and MW was observed in the evolution of the contact angle, but this effect disappeared over the entire contact angle evolution period where no clear effect of DS or MW could be pinpointed. Over the entire timescale of the contact angle evolution, all the polymers generally slowed down the contact angle evolution rate at all pH values. A comparative analysis of the effect of polymers of selected chemical properties indicated that the kinetic effect of CMC on the wettability of the bitumen surface was caused by very weak interactions /adsorption of CMC with/on bitumen. It is important to note that the effects of CMC on the initial contact angle and the contact angle evolution rate occur at a much shorter timescale than the bitumen displacement tests (section 5.3.1) and the rheological measurements (Chapter 4). Overall, this chapter demonstrated the beneficial role of CMC in accelerating bitumen displacement from illite which indicates that CMC facilitates bitumen liberation from solids in oil sand ores. Also, it was demonstrated that CMC does not affect bitumen hydrophobicity at the longer timescale under which its beneficial role in accelerating bitumen displacement from illite and dispersing oil sand slurries was established. Since a good rheological dispersant should not render bitumen hydrophilic and prevent bitumen extraction, the weak wetting action of CMC at the bitumen-solution interface is actually highly desirable.  165  CHAPTER 6  6 Conclusions, contributions to knowledge and future directions 6.1 Conclusions The production of bitumen from oil sands involves several stages including hydrotransport of oil sand slurries, recovery of bitumen in flotation cells and disposal of tailings. At each stage, the properties of the oil sand slurries are very important. The interparticle aggregation/dispersion and bitumen liberation phenomena during the hydrotransport stage determine the efficiency with which bitumen is recovered in the flotation stage. Generally, well dispersed slurries are desirable in the hydrotransport and flotation stages. An understanding of the slurry properties and finding means of achieving the desired slurry properties is critical to improved bitumen recoveries and should be beneficial especially to processing of challenging low grade (poor processing) oil sand ores. This dissertation investigated the aggregation/dispersion properties of poor processing oil sand slurries and explored the role of polysaccharides in preventing aggregation in concentrated slurries such as those encountered in industrial operations. Six carboxymethyl cellulose (CMC) polymers of various degrees of substitution (DS) and molecular weights (MW) were first characterized by dilute solution viscometry. The intrinsic viscosities [η] of the polymers were measured as a function of ionic strength (distilled water and 0.01 mol/L NaCl), pH (3, 4.5, 7) and temperature (25 °C, 50 °C). The molecular weights and molecular weight distributions (MWD) of the polymers were determined from analytical ultracentrifugation. The MWDs of the polymers were broad with polydispersities ranging from 2.27 to 4.52. In distilled water, the intrinsic viscosities increased with DS and pH indicating an extended conformation. In 0.01 mol/L NaCl, CMC assumed a coiled conformation resulting in a decrease in the intrinsic viscosities. The effect of pH on the conformation of CMC was negligible provided that constant ionic strength was maintained during the tests. Temperature also had a very weak effect on the intrinsic viscosity of CMC indicating that solvency effects are not important in controlling the conformation of CMC under the experimental conditions. Of the tested CMCs, the lowest molecular 166  weight polymer (LM-CMC) was the most flexible as inferred from its high ability to coil or stretch in solution. The Mark-Houwink-Sakurada (MHS) exponent α in the [η] ~ MW scaling law was found to be on the order of 0.83 in 0.01 mol/L NaCl without a clear correlation with pH. Estimates of the chain persistence length (Lp) showed that the chain flexibility decreased with increasing pH in distilled water (Lp ≈ 8.8 nm – 24.5 nm) while there was no significant change with pH in 0.01 mol/L NaCl (Lp ≈ 11.3 nm – 14.8 nm). The MHS exponent and Lp results indicate that the tested CMCs are semiflexible polymers with randomly coiling chains. Overall, the results suggest that ionic strength is the main parameter affecting the conformation and flexibility (through the electrostatic persistence length) of the tested CMC samples. The rheological behavior of oil sand ore slurries in the absence of CMC indicated that the oil sand components interacted to produce aggregated slurries with non-Newtonian flow behaviour. The slurries exhibited pseudoplastic behaviour with a yield stress and high apparent viscosities. The slurries also exhibited pronounced thixotropic properties. The aggregated nature of the slurries was confirmed through settling tests. Rheological measurements and settling tests on slurries in which CMC was added indicated that the polymers stabilized the slurries towards aggregation, thereby producing highly stable dispersions with no yield stress and significantly lower apparent viscosities. The dispersing ability of the polymers was independent of the degree of substitution. However, a dependence on the MW was observed. At lower dosages (125 g/t and 500 g/t), the lowest molecular weight polymer was only effective at reducing the yield stress and apparent viscosities, but not the thixotropy of the slurries. The flow curves showed pronounced hysteresis with indication of shearinduced weak flocculation being the contributing factor to the observed hysteresis. Higher dosages (750 g/t) were required to efficiently disperse the slurries and eliminate the shear-induced weak flocculation. The addition of 250 g/t of the higher molecular weight polymer was very effective at suppressing the thixotropy and reducing the yield stress and apparent viscosities resulting in highly dispersed slurries. The different dispersing abilities of the low and higher molecular weight polymers were primarily due differences in their molecular weight (hence different adsorption densities as indicated by the adsorption results of CMC on bitumen-free solids) and to some extent due to differences in their flexibility.  167  In order to more clearly understand the role of CMC in dispersing the oil sand slurries, two different tests were performed. Firstly, similar rheological tests as those performed on oil sand slurries were performed on bitumen-free solid slurries in the presence and in absence of CMC. Secondly, bitumen displacement tests from a model clay solid as well as contact angles of an air bubble in contact with a bitumen surface were performed both in the presence and in absence of CMC. The aim of these tests was to study the interactions between CMC, bitumen and solids. From these tests, the action of CMC in controlling the flow properties of oil sand slurries was attributed to two main mechanisms: Mechanism 1 The anionic CMC adsorbs on the solid surfaces, presumably through a combination of hydrogen bonding and possible metal sites present on the solid particle surfaces, rendering them electrostatically charged and creating a steric barrier that disperses the solids and stabilizes them against van der Waals and possibly hydrophobic attractive forces. Rheological measurements show the effective dispersing effect of CMC polymers when added to the aggregated bitumen-free solid slurries. The flow curves obtained in the presence of CMC are essentially Newtonian with no yield stress and low apparent viscosities. Sedimentation tests on oil sand and bitumen-free solid slurries showed an increasing amount of solid particles in the supernatant with increasing CMC concentration which confirmed the dispersing action of CMC and correlated well with the rheological changes observed in the presence of the polymers. While all CMC polymers tested were both effective at stabilising the bitumen-free solid slurries, the lower molecular weight polymer (LM-CMC) was not as effective as the higher molecular weight (HM-CMC) polymer in suppressing the time dependence of the solid slurries which explains the earlier mentioned ineffectiveness of the LM-CMC in dispersing oil sand slurries to the same extent as the HM-CMC. Mechanism 2 CMC facilitates the liberation of bitumen from solid particles. This diminishes the particlebitumen and bitumen–bitumen interactions, which reduces aggregation and formation of structures in oil sand slurries. Contact angles of a bitumen droplet in contact with illite indicated that the 168  presence of CMC accelerated the displacement or recession of bitumen from the illite surface under conditions of relatively low pH (6) and low temperature (25 °C). Adsorption of CMC at the illite/water interface prevented bitumen adhesion to the illite surface resulting in high steady-state contact angles required for bitumen detachment from solid surfaces. The polymers with different DS (DS0.7, DS0.9 and DS1.2) accelerated bitumen displacement from illite and the effect of DS was insignificant. The lower molecular weight polymer (LM-CMC) provided a faster bitumen displacement rate than the HM-CMC but the final measured contact angle did not depend on the MW. An increase in pH from 6 to 8.5 did not affect the kinetics of bitumen displacement from illite or final bitumen contact angle for the LM-CMC. However, increase in pH in the presence of the HM-CMC slowed down the bitumen displacement rate and reduced the final bitumen contact angle to that of the baseline case. At higher temperature, the beneficial effect of the polymers was not as evident as at 25 °C suggesting that at 40 °C, temperature was the main driving force for bitumen liberation from illite. Preadsorbing CMC on illite facilitated even faster bitumen displacement from illite and the very high bitumen contact angles indicated that it would be easier to detach bitumen in the presence of CMC. In this respect, CMC seems to adsorb on illite rather than on bitumen and effectively forms a steric barrier on the illite surface preventing illite-bitumen attachment. Contact angles of an air bubble in contact with a bitumen surface indicated that the presence of CMC led to a decrease in the initial contact angles but this decrease was more significant at pH 3 than at higher pH values. The HM-CMC polymer noticeably decreased the initial contact angle compared to the LM-CMC at pH 3. There was no effect of the polymers on the steady-state contact angles. Over a very short timescale, an effect of DS and MW was observed in the evolution of the contact angle, but this effect disappeared over the entire contact angle evolution period where no clear effect of DS or MW could be identified. Over the entire timescale of the contact angle evolution, all the polymers generally slowed down the contact angle evolution rate at all pH values. A comparative analysis of the effect of polymers of selected chemical properties indicated that the kinetic effect of CMC on the wettability of the bitumen surface was caused by very weak interactions /adsorption of CMC with/on bitumen rather than the viscosity of the polymer solutions. Overall, it is important to note that the effects of CMC on the initial contact angle and the contact angle evolution rate occur at a much shorter timescale (< 5 minutes) than the bitumen displacement tests and the rheological measurements (~30 minutes). Industrially, the average residence time of an 169  oil sand slurry in a hydrotransport pipeline (where air and chemicals are added and during which bitumen liberation and aeration occur) is about 17 minutes (average slurry speed of ~4 m/s in a 4-km long pipeline). Thus, it can be concluded that CMC does not affect bitumen hydrophobicity at the longer timescale over which its beneficial role in accelerating bitumen displacement from illite and dispersing oil sand slurries could be very beneficial to bitumen liberation and extraction. Finally, it can be concluded that although all the six carboxymethyl celluloses show very good dispersing capabilities, it should be noted that choosing the best dispersant for oil sand slurries is a trade-off between the dispersing and bitumen liberation capabilities of the polymer and the extent to which the polymer depresses the hydrophobicity of bitumen. Since excellent dispersion of the oil sand slurries is undesirable from the tailings disposal point of view, the best polymer should be able to give good slurry dispersion, promote bitumen liberation from solids, and not depress bitumen hydrophobicity. A look at the polymer capabilities in dispersing the oil sand slurries, displacing bitumen from illite and changing the wettability of bitumen shows that the LM-CMC is the best overall dispersant, since it was shown to give good (not excellent) slurry dispersing abilities at low dosages and also gave the best kinetics during the initial stages of bitumen displacement from the illite surface. At the same, it interacted very weakly with the bitumen surface and thus did not depress bitumen hydrophobicity. 6.2 Contributions to knowledge This dissertation investigated several technically important areas including aggregation/dispersion phenomena in slurry rheology, the role of polysaccharide in preventing interparticle aggregation in oil sand slurries, bitumen wettability, solution chemistry of polysaccharides, and stability of solid/liquid dispersions. Specifically, this work has contributed to existing knowledge in the following ways: 1. The solution chemistry of carboxymethyl cellulose polymers of different DS and MW was systematically studied as a function of pH, temperature, and ionic strength. To the author’s knowledge, this was the first study that used the analytical ultracentrifuge as a novel technique to systematically characterize the molecular weights and molecular weight distributions of industrial carboxymethyl cellulose samples. Important polymer parameters such as MHS constants, persistence lengths, polydispersity indices, and expansion coefficients were 170  determined. Such data is invaluable in understanding the application of the polymers in a wide range of processes such as mineral processing, food and cosmetics, pulp and paper processing, nanotechnology etc. 2. The aggregation/dispersion state of slurries prepared using an ore that is considered difficult to process (average bitumen content, high fines content) was studied and it was confirmed that inter-particle, particle-bitumen, and bitumen–bitumen interactions were prevalent even at high pH where colloid stability theories predict repulsion between the oil sand components. The attractive interactions were responsible for aggregation in the slurries. This aggregation between all slurry components accounts for the poor processability encountered in the commercial recovery of bitumen from similar ores. 3. A detailed and systematic study was undertaken on the role of carboxymethyl cellulose in preventing aggregation in poor processing oil sand slurries. Carboxymethyl cellulose is well known as a dispersant in coal and mineral processing operations but its dispersing action in oil sand slurries has not been researched. CMC can play a critical role in improving the processability of poor processing oil sands since its dispersing action not only involves preventing particle-particle aggregation but also facilitating bitumen liberation from solids. Its role could even be more significant since it was shown to be more effective at a lower pH and temperature than what current commercial operations use. This could have far-reaching cost and energy implications. 6.3 Future directions This work generated key contributions but not every aspect could be investigated. The following aspects arising out of this dissertation deserve to be followed-up in future research work: 1. Bitumen extraction (flotation) tests in the presence of CMC can be performed to show whether its beneficial role in preventing interparticle aggregation and facilitating bitumen liberation actually translates into gains during the bitumen recovery stage. Such tests can be carried out in a bench-scale Denver flotation machine which seems to generate more reliable bitumen ‘extractability’ data particularly for poor processing ores at lower temperatures (Zhou et al., 2004) compared to the Laboratory Hydrotransport Extraction System (LHES) which is 171  considered a standard unit for bench scale bitumen extraction studies (Wallwork et al., 2004). Qualitative observations during the sedimentation tests (Chapter 4) suggested that a higher amount of ‘free’ bitumen was present in the supernatant of the slurries containing CMC relative to those without CMC. However, it was not possible to quantify this amount of free-bitumen. A good starting point would be to devise a simple but quick test to quantify the amount of bitumen liberated as a result of adding CMC to an oil sand slurry. 2. Systematic adsorption studies of polysaccharides on oil sand components can be carried out to provide more details about the trends observed in the testwork (i.e., rheological, bitumen displacement and bitumen wettability tests) conducted in this dissertation. A technique that could be useful for such adsorption tests is the quartz crystal microbalance with dissipation (QCM-D). The technique measures the change in frequency of a quartz crystal from which the amount of adsorbed polymer can be determined. More importantly, the dissipation parameters obtained from the test can provide information about the structural (conformational) properties of the adsorbed polymer layers. 3.  The role of other types of polymers in the aggregation/dispersion of oil sand slurries should be explored so as to extend the use of such polymers in the oil sands industry where the use of chemicals used in the hydrotransport and bitumen flotation has largely been limited to inorganic additives. Potential polymers to consider would be of the anionic or nonionic low molecular weight type such as polystyrene sulfonate, lignosulfonates, hydroxypropyl cellulose, and hydroxyethyl cellulose.  4. The dispersion of oil sand components by CMC (and other polymers as suggested above) could make it difficult to separate the solids from water by flocculation during the tailings disposal stage. Thus, the effect of CMC concentration, DS, and molecular weight on flocculation of oil sand tailings should be investigated with particular emphasis on the settling rates of the solids and the quality/turbidity of the recycled water.  172  Bibliography Akroyd, J. T. (2004). Continuous flow rheometry for settling slurries (PhD Thesis, The University of Adelaide, Australia). Retrieved from http://hdl.handle.net/2440/37944. Alberta Chamber of Resources. (1996). Mineral Developments Agreement Co-Products Study. Final Report. Arêas, E. P. G., & Galembeck, F. (1991). Adsorption of carboxymethylcellulose onto hydroxylapatite. Journal of Colloid and Interface Science, 147(2), 370-377. Arinaitwe, E. (2008). Characterization of industrial flocculants through intrinsic viscosity measurements (Master’s thesis, University of British Columbia, Canada). Retrieved from https://circle.ubc.ca/handle/2429/1399. Arinaitwe, E., & Pawlik, M. (2013). A role of flocculant chain flexibility in flocculation of fine quartz. Part I. Intrinsic viscosities of polyacrylamide-based flocculants, International Journal of Mineral Processing, doi:10.1016/j.minpro.2013.01.006. Basu, S., Nandakumar, K., & Masliyah, J. H. (1996). A study of oil displacement on model surfaces. Journal of Colloid and Interface Science, 182(1), 82-94. Basu, S., Nandakumar, K., & Masliyah, J. H. (1998). Effect of NaCl and MIBC/kerosene on bitumen displacement by water on a glass surface. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 136(1), 71-80. Bayliss, P., & Levinson, A. A. (1976). Mineralogical review of the Athabasca oil sands deposits (Lower Cretaceous, Mannville group). Bulletin of Canadian Petroleum Geology, 24, 211-214. Beaussart, A., Mierczynska-Vasilev, A., & Beattie, D. A. (2010). Evolution of carboxymethyl cellulose layer morphology on hydrophobic mineral surfaces: Variation of polymer concentration and ionic strength. Journal of Colloid and Interface Science, 346(2), 303-310. Bichard, J. A. (1987). Oil sands composition and behavior research. The research papers of John A. Bichard, 1957–1965, AOSTRA technical publication series 4, Alberta Oil Sands Technology and Research Authority, Edmonton, AB. Bohdanecky, M. (1983). New method for estimating the parameters of the wormlike chain model from the intrinsic viscosity of stiff-chain polymers. Macromolecules, 16(9), 1483-1492. Bohdanecky´, M., & Kovar´ R, J. (1982). Viscosity of polymer solutions. Amsterdam: Elsevier. Bothner, H., Waaler, T., & Wik, O. (1988). Limiting viscosity number and weight average molecular weight of hyaluronate samples produced by heat degradation. International Journal of Biological Macromolecules, 10(5), 287-291. Bowman, C. W. (1967). Proceedings, 7th World Petroleum. Congress. 3, 583. Brown, P. H., & Schuck, P. (2006). Macromolecular size-and-shape distributions by sedimentation velocity analytical ultracentrifugation. Biophysical Journal, 90(12), 4651-4661. Brown, W., & Henley, D. (1964). Studies on cellulose derivatives. Part IV. The configuration of the polyelectrolyte sodium carboxymethyl cellulose in aqueous sodium chloride solutions. Die Makromolekulare Chemie, 79(1), 68–88. 173  Brown, W., Henley, D., & Öhman, J. (1963). Studies on cellulose derivatives. Part I. The dimensions and configuration of sodium carboxymethyl cellulose in cadoxen and the influence of the degree of substitution. Die Makromolekulare Chemie, 62(1), 164-182. Bulmer J.T, & Starr, J. (1979). Syncrude Analytical Methods for Oil Sand and Bitumen Processing. Edmonton: AOSTRA. Camp, F. W. (1974). The Tar sands of Alberta, Canada. Denver: Cameron Engineers. Carroll, B. (1996). The direct study of oily soil removal from solid substrates in detergency. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 114, 161-164. Chau, T. T. (2009). A review of techniques for measurement of contact angles and their applicability on mineral surfaces. Minerals Engineering, 22, 213-219. Chibowski, S. (1990). Dependence of the adsorption behavior of polyvinyl alcohol at the polystyrene latex—solution interface on the molecular weight. Journal of Colloid and Interface Science, 134(1), 174-180. Chow, R. S. (1991). The stability of kaolinite-in-water dispersions. Colloids and Surfaces, 61(4), 241-253. Chow, R. S., & Takamura, K. (1988). Electrophoretic mobilities of bitumen and conventional crudein-water emulsions using the laser doppler apparatus in the presence of multivalent cations. Journal of Colloid and Interface Science, 125(1), 212-225. 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, 103(2), 542-553. Clark, K. A. & Pasternack, D.S. (1932). Hot water separation of bitumen from Alberta bituminous sand. Industrial and Engineering Chemistry, 24, 1410-1416. Clark, K. A. (1944). Hot water separation of Alberta bituminous sand. CIM Transactions, XLVII, 257-274. Clasen, C., & Kulicke, W. -M. (2001). Determination of viscoelastic and rheo-optical material functions of water-soluble cellulose derivatives. Progress in Polymer Science, 26(9), 1839-1919. Conybeare, C. E. B. (1966). Origin of Athabasca oil sands: A review. Bulletin of Canadian Petroleum Geology, 14, 145-163. Coultate, T. P., & Royal Society of Chemistry (Great Britain). (2009). Food: The chemistry of its components. Cambridge, UK: Royal Society of Chemistry. Cuba-Chiem, L. T., Huynh, L., Ralston, J., & Beattie, D. A. (2008). In situ particle film ATR FTIR spectroscopy of carboxymethyl cellulose adsorption on talc: Binding mechanism, pH effects, and adsorption kinetics. Langmuir, 24(15), 8036-8044. Czarnecki, J., Radoev, B., Schramm, L. L., & Slavchev, R. (2005). On the nature of Athabasca oil sands. Advances in Colloid and Interface Science, 114-115, 53-60. Dai, Q., & Chung, K. H. (1995). Bitumen—sand interaction in oil sand processing. Fuel, 74(12), 1858-1864.  174  Dam, J., & Schuck, P. (2004). Calculating sedimentation coefficient distributions by direct modeling of sedimentation velocity concentration profiles. In: M.L. Johnson, & L. Brand, (Eds.), Numerical computer methods, Part E. New York: Academic Press. Dang-Vu, T., Jha, R., Wu, S., Tannant, D. D., Masliyah, J., & Xu, Z. (2009a). Wettability determination of solids isolated from oil sands. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 337(1), 80-90. Dang-Vu, T., Jha, R., Wu, S., Tannant, D., Masliyah, J., & Xu, Z. (2009b). Effect of solid wettability on processability of oil sands ores. Energy & Fuels, 23, 2628-2636. Davis, R.M. (1991). Analysis of dilute solutions of (carboxy-methyl)cellulose with the electrostatic wormlike chain theory. Macromolecules, 24(5), 1149-1155. Ding, X., Repka, C., Xu, Z., & Masliyah, J. (2006). Effect of illite clay and divalent cations on bitumen recovery. The Canadian Journal of Chemical Engineering, 84(6), 643-650. Drelich, J. & Miller, J. D. (1994). Surface and interfacial tension of the Whiterocks bitumen and its relationship to bitumen release from tar sands during hot water processing”, Fuel, 73(9), 15041510. Drelich, J., Bukka, K., Miller, J. D., & Hanson, F.V. (1994). Surface tension of toluene-extracted bitumens from Utah oil sands as determined by Wilhelmy plate and contact angle techniques. Energy & Fuels, 8(3), 700-704. Drelich, J., Laskowski, J.S., & Pawlik, M. (2000). Improved sample preparation and surface analysis methodology for contact angle measurements on coal (heterogeneous) surfaces. International Journal of Coal Preparation and Utilization, 21(3), 247-275. Drelich, J., Long, J., & Yeung, A. (2007). Determining surface potential of the Bitumen Water interface at nanoscale resolution using atomic force microscopy. The Canadian Journal of Chemical Engineering, 85(5), 625-634. Eremeeva, T. E., & Bykova, T. O. (1998). SEC of mono-carboxymethyl cellulose (CMC) in a wide range of pH; Mark–Houwink constants. Carbohydrate Polymers, 36(4), 319-326. Farrow, J., Horsley, R., Meagher, L., & Warren, L. (1989). The effect of alkali and alkaline-earth cations on the rheology of concentrated quartz suspensions. Journal of Rheology, 33(8), 12131230. Fedors, R.F. (1979). An equation for describing the viscosity of dilute to moderately concentrated polymer solutions. Polymer, 20(2), 225-228. Fleer, G. J., Cohen-Stuart, M. A., Scheutjens, J. M., Cosgrove, T., & Vincent, B. (1993). Polymers at interfaces. London: Chapman and Hall. Flory, P. J. (1953). Principles of polymer chemistry. Ithaca: Cornell University press. Fowkes, F. M. (1967). Attractive forces at solid-liquid interfaces. In: Wetting (pp. 3-30). London: Society of Chemical Industry Monograph No. 25. Fujita, H. (1962). Mathematical theory of sedimentation analysis. New York: Academic Press.  175  Fuoss, R. M. & Strauss, U. P. (1948a). Electrostatic interaction of polyelectrolytes and simple electrolytes. Journal of Polymer Science, 3, 602-603. Fuoss, R. M. & Strauss, U. P. (1948b).Polyelectrolytes II: Poly-4-vinyl pyridonium chloride and poly-4-vinyl-N-n butyl-pyridonium bromide. Journal of Polymer Science, 3, 246-263. Fuoss, R. M., & Cathers, G. I. (1949). Polyelectrolytes III: Viscosities of butyl bromide addition compounds of 4-vinyl pyridine –styrene copolymers in nitromethane-dioxane mixtures. Journal of Polymer Science, 3, 97-120. Gan, W., Cao, M., Crozier, B., & Liu, Q. (2007). Inhibiting quartz-bitumen coagulation by complexing agents. Canadian Metallurgical Quarterly, 46(3), 207-214. Ghimici, L., & Popescu, F. (1998). Determination of intrinsic viscosity for some cationic polyelectrolytes using Fedors equation. European Polymer Journal, 34(1), 13-16. Gilbert, R. D. (1994). Cellulosic Polymers, Blends, and Composites, Cincinnati: Hanser/Gardner Publications. Gillis, R. B., Rowe, A.J., Harding, S. E., Adams, G. G., Wolf, B., Berry, M., Besong, T. M. D., Corfield, A., Kök, S. M., Sidebottom, R. & Lafond, D. (2012). Molecular weight distribution analysis by ultracentrifugation: Adaptation of a new approach for mucins. Carbohydrate Polymers, 93(1), 178-183. Gupta, V., & Miller, J. D. (2010). Surface force measurements at the basal planes of ordered kaolinite particles. Journal of Colloid and Interface Science, 344(2), 362-371. Gutierrez, L., & Pawlik, M. (2012a). Influence of pH and temperature on the rheology of aqueous quartz-bitumen suspensions. Journal of Rheology, 56(4), 687-706. Gutierrez, L., & Pawlik, M. (2012b). Alkali extraction test to detect oxidation of bitumen in oil sand ores. Paper presented at the XXVI International Mineral Processing Congress, September 24-28, New Delhi, India. Gutierrez, L. (2009). Probing mineral-bitumen liberation using rheological measurements (Master’s thesis, University of British Columbia, Canada). Retrieved from https://circle.ubc.ca/handle/2429/7705. Gutierrez-Rodriguez, J. A., Purcell Jr, R. J., & Aplan, F. F. (1984). Estimating the hydrophobicity of coal. Colloids and Surfaces, 12(1-2), 1-25. Hagerman P. J. (1988). Flexibility of DNA. Annual Review of Biophysics and Biophysical Chemistry, 17, 265-286. Harding, S. E. (1997). The intrinsic viscosity of biological macromolecules. Progress in measurement, interpretation and application to structure in dilute solution. Progress in Biophysics and Molecular Biology, 68(2–3), 207-262. Harding, S. E. (2005a). Analysis of polysaccharide size, shape and interactions. In D.J. Scott, S.E. Harding, & A.J. Rowe (Eds.), Analytical ultracentrifugation. Techniques and methods (pp. 231252). Cambridge: Royal Society of Chemistry.  176  Harding, S. E. (2005b). Challenges for the modern analytical ultracentrifuge analysis of polysaccharides. Carbohydrate Research, 340(5), 811-826. Harding, S. E., Schuck, P., Abdelhameed, A. S., Adams, G., Kök, M. S., & Morris, G. A. (2011). Extended fujita approach to the molecular weight distribution of polysaccharides and other polymeric systems. Methods, 54(1), 136-144. Harris, G. H., Diao, J., & Fuerstenau, D.W. (1995). Coal flotation with non-ionic surfactants. Coal Preparation, 16, 135-147. Hatch, C. D., Wiese, J. S., Crane, C. C., Harris, K. J., Kloss, H. G., & Baltrusaitis, J. (2012). Water adsorption on clay minerals as a function of relative humidity: Application of BET and freundlich adsorption models. Langmuir, 28(3), 1790-1803. Healy, T. W., & White, L. R. (1978). Ionizable surface group models of aqueous interfaces. Advances in Colloid and Interface Science, 9(4), 303-345. Hlady, V., Lyklema, J., & Fleer, G. J. (1982). Effect of polydispersity on the adsorption of dextran on silver iodide. Journal of Colloid and Interface Science, 87(2), 395-406. Hoogendam, C. W., de Keizer, A., Cohen Stuart, M. A., Bijsterbosch, B. H., Batelaan, J. G., & van der Horst, P. M. (1998a). Adsorption mechanisms of carboxymethyl cellulose on mineral surfaces. Langmuir, 14(14), 3825-3839. Hoogendam, C. W., de Keizer, A., Cohen Stuart, M. A., Bijsterbosch, B. H., Smit, J. A. M., van Dijk, J. A. P. P., van der Horst, P. M., & Batelaan, J. G. (1998b). Persistence length of carboxymethyl cellulose as evaluated from size exclusion chromatography and potentiometric titrations. Macromolecules, 31(18), 6297-6309. Houwink, R. (1940). Journal für Praktische Chemie, 157, 15-18. Huggins, M. L. (1942). The viscosity of dilute solutions of long-chain molecules. IV. Dependence on concentration. Journal of the American Chemical Society, 64, 2716-2718. Hunter, R. J. (1981). Zeta potential in colloid science: Principles and applications. London: Academic Press. Hupka, J., Miller, J. D. & Cortez, A. (1983). Importance of bitumen viscosity in the hot water processing of domestic tar sands. Mining Engineering, 35, 1635-1641. Hupka, J., Oblad, A. G. & Miller, J. D. (1987). Diluent-assisted hot-water processing of tar sands. AOSTRA Journal of Research, 3, 95-102. Hussain, S. A., Demirci, S.,Oezbayoglu, G. (1996). Zeta potential measurements on three clays from turkey and effects of clays on coal flotation. Journal of Colloid and Interface Science, 184(2), 535-541. Ignasiak, T.M., Zhang, Q., Kratochvil, C., Montgomery, D.S. & Strausz, O.P. (1985). Chemical and mineral characterization of bitumen free Athabasca oil sands related to the bitumen concentration in the sand tailings from the Syncrude batch extraction test. AOSTRA Journal of Research, 2, 2134.  177  Imperial Oil Ltd. (2005). AEUB Application for Approval for Kearl Oil Sands Project. Vol. 1. Calgary: Imperial Oil Ltd. International Energy Agency. (2012). Oil market report. Paris. Isaacs, E. E. & Smolek, K. F. (1983). Interfacial tension behavior of Athabasca bitumen/aqueous surfactant systems. The Canadian Journal of Chemical Engineering, 61, 233-240. Israelachvili, J. N. (1991). Intermolecular and surface forces. London: Academic Press. James, A. E., & Williams, D. J. A. (1982). Particle interactions and rheological effects in kaolinite suspensions. Advances in Colloid and Interface Science, 17(1), 219-232. Kaminsky, H. A. W., Etsell, T. H., Ivey, D. G., & Omotoso, O. (2009). Distribution of clay minerals in the process streams produced by the extraction of bitumen from Athabasca oil sands. The Canadian Journal of Chemical Engineering, 87(1), 85-93. Kasongo, T., Zhou, Z., Xu, Z., & Masliyah, J. (2000). Effect of clays and calcium ions on bitumen extraction from Athabasca oil sands using flotation. The Canadian Journal of Chemical Engineering, 78(4), 674-681. Kästner, U., Hoffmann, H., Dönges, R. & Hilbig, J. (1997). Structure and solution properties of sodium carboxymethyl cellulose. Colloids and Surfaces A, 123-124, 307-328. Khraisheh, M., Holland, C., Creany, C., Harris, P., & Parolis, L. (2005). Effect of molecular weight and concentration on the adsorption of CMC onto talc at different ionic strengths. International Journal of Mineral Processing, 75(3), 197-206. Kitchener, J.A. (1969). Colloidal minerals: Chemical aspects of their dispersion, flocculation and filtration. Filtration and Separation, 6, 553-558. Klein, B. 1992. Rheology and stability of magnetite dense media (PhD thesis, University of British Columbia, Canada). Retrieved from https://circle.ubc.ca/handle/2429/2991. Klein, B., & Pawlik, M. (2005). Rheology modifiers for mineral suspensions. Minerals & Metallurgical Processing, 22(2), 83-88. Klein, B., Laskowski, J. S. & Partridge, S. J. (1995). A new viscometer for rheological measurements on settling suspensions. Journal of Rheology, 39, 827-840. Kotlyar, L. S, Sparks, B. D., & Schutte, R. (1996). Effect of salt on the flocculation behavior of nanoparticles in oil sands fine tailings. Clays and Clay Minerals, 44(1), 121-131. Kotlyar, L. S., Sparks, B. D., Woods, J., Capes, C. E., & Schutte, R. (1995). Biwetted ultrafine solids and structure formation in oil sands fine tailings. Fuel, 74(8), 1146-1149. Kratky, O., & Porod, G. (1949). Röntgenuntersuchung gelöster Fadenmoleküle, Recueil des Travaux Chimiques des Pays Bas, 68, 1106-1123. Kratochvil, P. (1972). Particle scattering functions. In M.B. Huglin (Ed.), Light scattering from polymer solutions (pp. 333-384). London: Academic Press. Kulicke W.-M., & Clasen C. (2004). Viscosimetry of Polymers and Polyelectrolytes. Heidelberg: Springer Laboratory. 178  Kulicke, W., Kull, W., Kull, A. H., Thielking, H., Engelhardt, J., & Pannek, J. (1996). Characterization of aqueous carboxymethylcellulose solutions in terms of their molecular structure and its influence on rheological behaviour. Polymer, 37(13), 2723-2731. Kulicke, W.-M., & Kniewske, R. (1981). Long-term conformation of macromolecules in solution, 2. Poly(acrylamide-co-sodium acrylate)s. Die Makromolekulare Chemie, 182, 2277-2287. Lamm, O. (1929). Die Differentialgleichung der Ultrazentrifugierung. Ark. Mat. Astr. Fys., 21B(2), 1-4. Laskowski, J. S. (1988). Dispersing agents in mineral processing. In: Developments in Mineral Processing 9 – Froth Flotation, Proceedings of the 2nd Latin American Congress on Froth Flotation, Concepcion, Chile, Aug. 1985, Castro-Flores, S.H. & Alvares, J.A. (Eds.), Elsevier. Laskowski, J. S. (2001). Coal flotation and fine coal utilization. Amsterdam: Elsevier. Laue, T. M. (1996). Choosing which optical system of the Optima™ XL-I analytical ultracentrifuge to use. Retrieved from Beckman Coulter website: https://www.beckmancoulter.com/wsrportal/wsr/research-and-discovery/ products-and-services/centrifugation-analytical/index.htm Lebedeva, E. V., & Fogden, A. (2011). Wettability alteration of kaolinite exposed to crude oil in salt solutions. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 377(1), 115-122. Lebowitz, J., Lewis, M. S., & Schuck, P. (2002). Modern analytical ultracentrifugation in protein science: A tutorial review. Protein Science, 11(9), 2067-2079. Li, H., Long, J., Xu, Z., & Masliyah, J. H. (2008). Novel polymer aids for low grade oil sand ore processing. The Canadian Journal of Chemical Engineering, 86(2), 168-176. Li, H., Zhou, Z., Xu, Z., & Masliyah, J. (2005). Role of acidified sodium silicate in low temperature bitumen extraction from poor-processing oil sand ores. Industrial & Engineering Chemistry Research, 44(13), 4753-4761. Lin, F. (2012). Surface charge heterogeneities and shear-induced coalescence of bitumen droplets (PhD thesis, University of Alberta, Canada). Retrieved from http://hdl.handle.net/10402/era.28116. Liu, J., Xu, Z., & Masliyah, J. (2003). Studies on bitumen-silica interaction in aqueous solutions by atomic force microscopy. Langmuir, 19(9), 3911-3920. Liu, J., Xu, Z., & Masliyah, J. (2004a). Interaction between bitumen and fines in oil sands extraction system: Implication to bitumen recovery. The Canadian Journal of Chemical Engineering, 82(4), 655-666. Liu, J., Xu, Z., & Masliyah, J. (2004b). Role of fine clays in bitumen extraction from oil sands. AICHE Journal, 50(8), 1917-1927. Liu, J., Xu, Z., & Masliyah, J. (2005a). Processability of oil sand ores in Alberta. Energy & Fuels, 19(5), 2056-2063. Liu, J., Xu, Z., & Masliyah, J. (2005b). Colloidal forces between bitumen surfaces in aqueous solutions measured with atomic force microscope. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 260(1), 217-228. 179  Liu, J., Zhou, Z., Xu, Z., & Masliyah, J. (2002). Bitumen-clay interactions in aqueous media studied by zeta potential distribution measurement. Journal of Colloid and Interface Science, 252(2), 409-409. Liu, Q., & Laskowski, J. S. (1989a). The interactions between dextrin and metal hydroxides in aqueous solutions. Journal of Colloid and Interface Science, 130(1), 101-111. Liu, Q., & Laskowski, J. S. (1989b). The role of metal hydroxides at mineral surfaces in dextrin adsorption, I. studies on modified quartz samples. International Journal of Mineral Processing, 26(3), 297-316. Liu, Q., Zhang, Y., & Laskowski, J. S. (2000). The adsorption of polysaccharides onto mineral surfaces: An acid/base interaction. International Journal of Mineral Processing, 60(3), 229-245. Long, J., Drelich, J., Xu, Z., & Masliyah, J. H. (2007). Effect of operating temperature on Water Based oil sands processing. The Canadian Journal of Chemical Engineering, 85(5), 726738. Long, J., Xu, Z., & Masliyah, J. (2005). On the role of temperature in oil sands processing. Energy & Fuels, 19(4), 1440-1446. Long, J., Xu, Z., & Masliyah, J. H. (2006). Adhesion of single polyelectrolyte molecules on silica, mica, and bitumen surfaces. Langmuir, 22(4), 1652-1659. Lovell, P. A. (1989). Dilute solution viscometry. In C. Price, & C. Booth (Eds.), Comprehensive polymer Science: Polymer characterization, Vol. 1 (pp. 173-197). Oxford: Pergamon Press. Luthra, M., Lopetinsky, R. J. G., Sanders, R. S., Nandakumar, K., & Masliyah, J. H. (2004). A new device to determine bitumen extraction from oil sands. The Canadian Journal of Chemical Engineering, 82(4), 752-762. Macosko, C. W. (1994). Rheology: Principles, measurements, and applications. New York: WileyVch. Mark, H. (1938). In: Der Feste Körper, Leipzig: S. Hirzel. Ma, X., & Pawlik, M. (2007). Intrinsic viscosities and Huggins constants of guar gum in alkali metal chloride solutions. Carbohydrate Polymers, 70(1), 15-24. Majid, A., & Sparks, B. D. (1996). Role of hydrophobic solids in the stability of oil sands fine tailings. Fuel, 75(7), 879-884. Marko, J. F., & Siggia, E. D. (1995). Stretching DNA. Macromolecules, 28, 8759-8770. Masliyah, J., Czarnecki, J., & Xu, Z. (2011). Handbook on Theory and Practice of Bitumen Recovery from Athabasca Oil Sands Volume 1: Theoretical Basis. Cochrane: Kingsley publishing services. Masliyah, J., Zhou, Z. J., Xu, Z., Czarnecki, J., & Hamza, H. (2004). Understanding Water Based bitumen extraction from Athabasca oil sands. The Canadian Journal of Chemical Engineering, 82(4), 628-654.  180  Mendichi, R., Soltés, L., & Giacometti Schieroni, A. (2003). Evaluation of radius of gyration and intrinsic viscosity molar mass dependence and stiffness of hyaluronan. Biomacromolecules, 4(6), 1805-1810. Mercier, P., McCracken, T., Ng, S., Moran, K., Sparks, B., Kingston, D. Kotlyar, L.S., Kung, J., Woods, J., Patarachao, B., & McCracken, T. (2012). Colloidal clay gelation: Relevance to current oil sands operations. Petroleum Science and Technology, 30(9), 915-923. Mierczynska-Vasilev, A., & Beattie, D. A. (2010). In situ atomic force microscopy of modified dextrin adsorption on hydrophobic and hydrophilic layered silicate minerals. Journal of Colloid and Interface Science, 344(2), 429-437. Mijnlieff, P. F., & Coumou, D. J. (1968). Extrapolation of light-scattering data to zero scattering angle. Journal of Colloid and Interface Science, 27(3), 553-572. Miller, R.F. (1960). Theory of clay mineral suspensions in dilute suspensions. In: B.N. Rolfe, R.F. Miller, & I.S. McQueen (Eds.), Dispersion characteristics of montmorillonite, kaolinite, and illite clays in waters of varying quality, and their control with phosphate dispersants, Geological Survey Professional Paper 334-G, Washington: United States Government printing Office. Minami, N., Kim, Y., Miyashita, K., Kazaoui, S., & Nalini., B. (2006). Cellulose derivatives as excellent dispersants for single-wall carbon nanotubes as demonstrated by absorption and photoluminescence spectroscopy. Applied Physics Letters, 88(9), 93123. Morris, G. E., Fornasiero, D., & Ralston, J. (2002). Polymer depressants at the talc–water interface: Adsorption isotherm, microflotation and electrokinetic studies. International Journal of Mineral Processing, 67(1), 211-227. Mossop, G. D. (1980). Geology of the Athabasca oil sands. Science, 207(4427), 145-152. Nagaraj, D. R. (2000). New synthetic polymeric depressants for sulfide and non-sulfide minerals. In: Proceedings of the XXI International Mineral Processing Congress, Rome, Italy, July 23-27, Massacci, P. (Ed.), Elsevier. Nguyen, A. V., & Schulze, H. J. (2004). Colloidal science of flotation. New York: Marcel Dekker. Nguyen, Q. D., & Boger, D. V. (1983). Yield stress measurement for concentrated suspensions. Journal of Rheology, 27, 321-349. Noda, I., Imai, T., Kitano, T., & Nagasawa, M. (1981). Solution properties of poly(tert-butyl crotonate) with a semiflexible backbone and sharp molecular weight distributions. 1. light scattering. Macromolecules, 14(5), 1303-1306. Odijk, T. (1977). Polyelectrolytes near the rod limit. Journal of Polymer Science Part b-Polymer Physics, 15(3), 477-483. Olphen, H. V. (1963). An introduction to clay colloid chemistry, for clay technologists, geologists, and soil scientists. New York: Interscience Publishers. Olphen, H. V. (1977). An introduction to clay colloid chemistry, for clay technologists, geologists, and soil scientists. New York: Wiley Omotoso, O. E., & Mikula, R. J. (2004). High surface areas caused by smectitic interstratification of kaolinite and illite in Athabasca oil sands. Applied Clay Science, 25(1), 37-47. 181  Leppinen, J., Palosaari, V., Kreutzman, N., & Moilanen, A., Kloow, G., Kalapudas, R., & Luukkanen, S. (2005). Depression of gangue minerals using tailor-made CMCs in sulfide ore flotation. In G.J. Jameson (Ed.), Centenary of flotation symposium, 6-9 June 2005 (pp. 977-982). Brisbane, Australia: The Australasian Institute of Mining and Metallurgy. Pals, D. T. F., & Hermans, J. J. (1948). Concentration dependence of the viscosity of sodium pectinate in solutions of sodium chloride. Journal of Polymer Science, 3, 897-898. Pals, D. T. F., & Hermans, J. J. (1952). Sodium salts of pectin and of carboxymethyl cellulose in aqueous sodium chloride, I. Viscosities. Recueil, 71, 433-457. Parolis, L. A. S., van der Merwe, R., Groenmeyer, G. V., & Harris, P. J. (2008). The influence of metal cations on the behaviour of carboxymethyl celluloses as talc depressants. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 317(1), 109-115. Patel, T. R., Morris, G. A., de la Torre, Jose Garcia, Ortega, A., Mischnick, P., & Harding, S. E. (2008). Molecular flexibility of methylcelluloses of differing degree of substitution by combined sedimentation and viscosity analysis. Macromolecular Bioscience, 8(12), 1108-1115. Patel, T. R., Picout, D. R., Ross-Murphy, S. B., & Harding, S. E. (2006). Pressure cell assisted solution characterization of galactomannans. 3. Application of analytical ultracentrifugation techniques. Biomacromolecules, 7(12), 3513-3520. Pavlov, G. M., Rowe, A. J., & Harding, S. E. (1997). Conformation zoning of large molecules using the analytical ultracentrifuge. Trends in Analytical Chemistry, 16(7), 401-405. Pawlik, M. (2002). Reverse flotation as a method of coal cleaning for preparation of coal-water slurries (PhD thesis, University of British Columbia, Canada). Retrieved from https://circle.ubc.ca/handle/2429/13213. Pawlik, M. (2005). Polymeric dispersants for coal–water slurries. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 266(1), 82-90. Pawlik, M., Laskowski, J. S., & Ansari, A. (2003). Effect of carboxymethyl cellulose and ionic strength on stability of mineral suspensions in potash ore flotation systems. Journal of Colloid and Interface Science, 260(2), 251-258. Potoczny, Z. M., Vargha-Butler, E. I., Zubovits, T. K., & Neumann, A. W. (1984). Surface tension of bitumens, I. Temperature dependence. AOSTRA Journal of Research, 1, 107-115. Pugh, R. J. (1989). Macromolecular organic depressants in sulphide flotation—A review, 1. principles, types and applications. International Journal of Mineral Processing, 25(1), 101-130. Rand, B., & Melton, I. E. (1977). Particle interactions in aqueous kaolinite suspensions, I. Effect of pH and electrolyte upon the mode of particle interaction in homoionic sodium kaolinite suspensions. Journal of Colloid and Interface Science, 60(2), 308-320. Rao, M.V.S. (1993). Viscosity of dilute to moderately concentrated polymer solutions. Polymer, 34(3), 592-596. Rath, R., Subramanian, S., & Laskowski, J. (1997). Adsorption of dextrin and guar gum onto talc. A comparative study. Langmuir, 13(23), 6260-6266.  182  Rattanakawin, C. (2002). Polymeric Flocculation: Effects of Chemical and Physical Variables on Floc Formation and Growth, (PhD thesis, The Pennsylvania State University, USA). Ren, S., Zhao, H., Long, J., Xu, Z., & Masliyah, J. (2009). Understanding weathering of oil sands ores by atomic force microscopy. AIChE Journal, 55(12), 3277-3285. Rey, F., & Machado, A. (2000). Influence of the temperature on the viscosity measurements of solutions of a soil fulvic acid. Canadian Journal of Chemistry, 78(7), 1013-1018. Robinson, G., Ross-Murphy, S. B., & Morris, E. R. (1982). Viscosity-molecular weight relationships, intrinsic chain flexibility, and dynamic solution properties of guar galactomannan, Carbohydrate Research, 107, 17-32. Rowe, A., Counce, R., Morton, S., Hu, M., & DePaoli, D. (2002). Oil detachment from solid surfaces in aqueous surfactant solutions as a function of pH. Industrial & Engineering Chemistry Research, 41(7), 1787-1795. Russel, W. B. (1978). The rheology of suspensions of charged rigid spheres. Journal of Fluid Mechanics, 85(2), 209-232. Saada, A., Siffert, B., & Papirer, E. (1995). Comparison of the hydrophilicity/hydrophobicity of illites and kaolinites. Journal of Colloid and Interface Science, 174(1), 185-190. Salmi, T., Valtakari, D., Paatero, E., Holmbom, B., & Sjoholm, R. (1994). Kinetic study of the carboxymethylation of cellulose. Industrial & Engineering Chemistry Research, 33(6), 14541459. Sato, T., Norisuye, T., & Fujita, H. (1984). Double-stranded helix of xanthan: Dimensional and hydrodynamic properties in 0.1 M aqueous sodium chloride. Macromolecules, 17(12), 26962700. Scheutjens, J. M. H. M., & Fleer, G. J., (1979). Statistical theory of the adsorption of interacting chain molecules. 1. Partition function, segment density distribution, and adsorption isotherms. Journal of Physical Chemistry, 83, 1619-1635. Scheutjens, J. M. H. M., & Fleer, G. J., (1980). Statistical theory of the adsorption of interacting chain molecules. 2. Train, loop, and tail size distributions. Journal of Physical Chemistry, 84, 178-190. Schneider, N. S, & Doty, P. (1954). Macro-ions. IV. The ionic strength dependence of the molecular properties of sodium carboxymethylcellulose. Journal of Physical. Chemistry, 58(9), 762-769. Schramm, L. L, Stasiuk, E. N, Yarranton, H., Maini, B., & Shelfantook, B. (2003). Temperature effects from the conditioning and flotation of bitumen from oil sands in terms of oil recovery and physical properties. Journal of Canadian Petroleum Technology, 42(8), 55-61. Schramm, L. L. (1989). The influence of suspension viscosity on bitumen rise velocity and potential recovery in the hot water flotation process for oil sands. Journal of Canadian Petroleum Technology, 28(3), 73-80. Schramm, L. L., Stasiuk, E. N. & MacKinnon, M. (2000). Surfactants in Athabasca oil sands extraction and tailings processes. In L.L. Schramm (Ed.), Surfactants: Fundamentals and applications in the petroleum industry, Cambridge: Cambridge University Press. 183  Schuck, P. (1998). Sedimentation analysis of noninteracting and self-associating solutes using numerical solutions to the Lamm equation. Biophysical Journal, 75(3), 1503-1512. Schuck, P. (2000). Size-distribution analysis of macromolecules by sedimentation velocity ultracentrifugation and Lamm equation modeling. Biophysical Journal, 78(3), 1606-1619. Schuck, P., & Rossmanith, P. (2000). Determination of the sedimentation coefficient distribution by least squares boundary modeling. Biopolymers, 54(5), 328-341. Schulz, L., & Burchard, W. (1993) Solution structure of various cellulose derivatives. Das Papier, 47, 1-10. Schulz, L., Burchard, W., & Dönges, R. (1998). Evidence of supramolecular structures of cellulose derivatives in solution; cellulose derivatives. In T. J. Heinze, & W. G. Glasser (Eds.), Cellulose derivatives (pp. 218-238). Washington: American Chemical Society. Shortridge, P. G., Harris, P. J., Bradshaw, D. J., & Koopal, L. K. (2000). The effect of chemical composition and molecular weight of polysaccharide depressants on the flotation of talc. International Journal of Mineral Processing, 59(3), 215-224. Sitaramaiah, G. & Goring, D. A. I. (1962). Hydrodynamic studies on sodium carboxymethyl cellulose in aqueous solutions. Journal of Polymer Science, 58 (166), 1107–1131. Skolnick, J., & Fixman, M. (1977). Electrostatic persistence length of a wormlike polyelectrolyte. Macromolecules, 10(5), 944-948. Škvarla, J., & Kmet', S. (1991). Influence of wettability on the aggregation of fine minerals. International Journal of Mineral Processing, 32(1), 111-131. Smoluchowski, M. (1903). Contribution à la théorie de l’endosmose électrique et de quelques phenomènes corrélatifs, Bulletin International del’Academie des Sciences de Cracovie, 8, 182200. Smoluchowski, M. 1921. Electro-osmosis and current flow. In L. Graetz (Ed.), Handbuch der Elektrizitat und des M agnetismus, Leipzig: Barth-Verlag. Solari, J. A., Correa de Araujo, A. & Laskowski, J. S. (1986). The effect of carboxymethyl cellulose on the flotation and surface properties of graphite. Coal Preparation, 3, 15-31. Somasundaran, P., Mehta, S. C., Yu, X. & Krishnakumar, S. (2009). Colloid systems and interfaces stability of dispersions through polymer and surfactant adsorption. In K.S. Birdi (Ed), Handbook of surface and colloid chemistry, Boca Raton: CRC Press. Somasundaran, P., Wang, J., & Nagaraj, D. R. (2005). Adsorption and conformation of polysaccharide depressants on mineral. In: Innovations in natural resource processing: Proceedings of the Jan. D. Miller symposium, Courtney A., Young, J., Kellar, J., & Michael L. (Eds), SME Annual Meeting, Salt Lake City, Utah, United States, 28 February - 2 March 2005. Srinivasa, S., Flury, C., Afacan, A., Masliyah, J., & Xu, Z. (2012). Study of bitumen liberation from oil sands ores by online visualization. Energy & Fuels, 26(5), 2883-2890. Środoń J. (2003). Illite group clays. In G.V. Middleton, & M.J. Church (Eds.), Encyclopedia of sediments and sedimentary rocks, Dordrecht: Kluwer Academic Publishers. 184  Starkweather, B., Counce, R., & Zhang, X. (1999). Displacement of a hydrocarbon oil from a metal surface using a surfactant solution. (1999). Separation Science and Technology, 34(6-7), 14471462. Starkweather, B., Zhang, X., & Counce, R. (2000). An experimental study of the change in the contact angle of an oil on a solid surface. Industrial & Engineering Chemistry Research, 39(2), 362-366. Staudinger, H., & Heuer., W. (1930). Relationship between the viscosity and the molecular weight of polystyrene (in German). Ber. Dtsch. Chem. Ges., 63(11), 222-234. Steenberg, E., Harris, P. J. (1984). Adsorption of carboxymethylcellulose, guar gum, and starch onto talc, sulfides, oxides, and salt-type minerals. South African Journal of Chemistry, 37, 85-90. Stelzer, G.I. & Klug, E.D. (1980). Carboxymethylcellulose. In R.L. Davidson (Ed.), Handbook of water-soluble gums and resins, New York: McGraw-Hill. Stephan, E.A. & Chase, G.G. (2001). A preliminary examination of zeta potential and deep bed filtration activity. Separation and Purification Technology, 21, 219–226. Stockmayer, W. H., & Fixman, M. (1963). On the estimation of unperturbed dimensions from intrinsic viscosities. Journal of Polymer Science, Part C, 1(1), 137-141. Svedberg, T., & Pedersen, K. O. (1940). The ultracentrifuge. Oxford: The Clarendon Press. Tadros, Th. F., Taylor, P., & Bognolot, G. (1995). Influence of addition of a polyelectrolyte nonionic polymers, and their mixtures on the rheology of coal/water suspensions. Langmuir, 11(12), 46784684. Takamura, K. (1982). Microscopic structure of Athabasca oil sand. The Canadian Journal of Chemical Engineering, 60, 538-545. Takamura, K., & Chow, R. S. (1983). A mechanism for initiation of bitumen displacement from oil sand. Journal of Canadian Petroleum Technology, 22(6), 22-30. Takamura, K., & Chow, R. S. (1985). The electric properties of the bitumen/water interface part II. Application of the ionizable surface-group model. Colloids and Surfaces, 15, 35-48. Takamura, K., & Wallace, D. (1988). The physical chemistry of the hot water process. Journal of Canadian Petroleum Technology, 27(6), 98-106. Taubaso, C., Dos Santos Afonso, M., & Torres Sánchez, R. M. (2004). Modelling soil surface charge density using mineral composition. Geoderma, 121(1), 123-133 Taylor, G. I. (1923). Stability of a viscous liquid contained between two rotating cylinders. Philos. Trans. R. Soc. London, A223, 289-343. Theng, B. K. G. (1982). Clay-polymer interactions; summary and perspectives. Clays and Clay Minerals, 30(1), 1-10. Tombácz, E., & Szekeres, M. (2006). Surface charge heterogeneity of kaolinite in aqueous suspension in comparison with montmorillonite. Applied Clay Science, 34(1), 105-124. Tricot, M. (1984). Comparison of experimental and theoretical persistence length of some polyelectrolytes at various ionic strengths. Macromolecules, 17(9), 1698-1704. 185  Tu, Y., O'Carroll, J. B., Kotlyar, L. S., Sparks, B. D., Ng, S., Chung, K. H., & Cuddy, G. (2005). Recovery of bitumen from oil sands: Gelation of ultra-fine clay in the primary separation vessel. Fuel, 84(6), 653-660. Übbelohde, L. (1937). The principle of the suspended level: Applications to the measurement of viscosity and other properties of liquids. Industrial & Engineering Chemistry Analytical Edition, 9(2), 85-90. Van Wazer, J. R. (1963). Viscosity and flow measurement: A laboratory handbook of rheology. New York: Interscience Publishers. Vargha-Butler, E. I., Potoczny, Z. M., Zubovits, T. K., Budziak, C. J., & Neumann, A. W. (1988). Surface tension of bitumen from contact angle measurements on films of bitumen. Energy & Fuels, 2(5), 653-656. Vink, H. (1970). Viscosity of polyelectrolyte solutions. Die Makromolekulare Chemie 131, 133-145. Wade, A., & Weller, P. J. (1994). Handbook of pharmaceutical excipients. London: Pharmaceutical Society of Great Britain. Wallwork, V., Xu, Z., & Masliyah, J. (2004). Processibility of Athabasca oil sand using a laboratory hydrotransport extraction system (LHES). Canadian Journal of Chemical Engineering, 82(4), 687-695. Wang, L., Dang-Vu, T., Xu, Z., & Masliyah, J. (2010). Use of short-chain amine in processing of Weathered/Oxidized oil sands ores. Energy & Fuels, 24, 3581-3588. Wang, L., Englert, A. H., Masliyah, J.H., & Xu, Z. (2011). Oil sand processing: Role of colloidal chemistry. In: Encyclopedia of Surface and Colloid Science, Second Edition: Taylor & Francis. Wang, S. (2012). Study on adsorption of inorganic-organic hybrid polymers and flocculation of oil sands tailings (Master’s thesis, University of Alberta, Canada). Retrieved from http://hdl.handle.net/10402/era.29364. Wang, S., Segin, N., Wang, K., Masliyah, J., & Xu, Z. (2011). Wettability control mechanism of highly contaminated hydrophilic Silica/Alumina surfaces by ethyl cellulose. Journal of Physical Chemistry C, 115(21), 10576-10587. Wang, Y., Jia, W., Ding, M., Yang, H., Hu, B., & Ren, S. (2012). Effect of diluent addition on bitumen liberation from a glass surface. Energy & Fuels, 26(2), 1019-1027. Watanabe, M., Gondo, T., & Kitao, O. (2004). Advanced wet-end system with carboxymethylcellulose. TAPPI journal, 3(5), 15-19. Weast, R., (1970). CRC Handbook of Chemistry and Physics. 50th Ed., Cleveland: CRC Press. Williams, D. J. A. & Williams, K. P. (1978). Electrophoresis and zeta potential of kaolinite. Journal of Colloid and Interface Science, 65, 79–87. Williams, D. J. A., & Williams, K. P. (1982). Colloid stability and rheology of kaolinite suspensions. Transactions and Journal of the British Ceramic Society, 81(3), 78-83. Xu, Z., & Yoon, R. (1989). The role of hydrophobic interactions in coagulation. Journal of Colloid and Interface Science, 132(2), 532-541. 186  Xu, Z., Liu, Q., Ling, J., & Summers, A. (1996). Wetting of mercury surfaces by halide electrolyte solutions. Langmuir, 12(2), 547-554. Yamakawa, H. (1971). Modern theory of polymer solutions. New York: Harper & Row. Yamakawa, H., & Fujii, M. (1974). Intrinsic viscosity of wormlike chains. determination of the shift factor. Macromolecules, 7(1), 128-135. Yin, X., & Miller, J. (2012). Wettability of kaolinite basal planes based on surface force measurements using atomic force microscopy. Minerals & Metallurgical Processing, 29(1), 1319. Yoon, R., Flinn, D. H., & Rabinovich, Y. I. (1997). Hydrophobic interactions between dissimilar surfaces. Journal of Colloid and Interface Science, 185(2), 363-370. Yotsumoto, H. & Yoon, R. H. (1993). Application of extended DLVO theory I. Stability of rutile suspensions. pp.426-433, Part II. Stability of silica suspensions. pp.434-441, Journal of Colloid and Interface Science, 157. Zajic, J. E., Cooper, D. G., Marshall J. A., & Gerson, D. F. (1981). Microstructure of Athabasca bituminous sand by freeze-fracture preparation and transmission electron microscopy. Fuel, 60, 619-623. Zbik, M. S., & Frost, R. L. (2009). Micro-structure differences in kaolinite suspensions. Journal of Colloid and Interface Science, 339(1), 110-116. Żbik, M. S., Du, J., Pushkarova, R. A., & Smart, R. S. C. (2009). Observation of gaseous films at solid–liquid interfaces: Removal by ultrasonic action. Journal of Colloid and Interface Science, 336(2), 616-623. Zhao, H., Dang-Vu, T., Long, J., Xu, Z., & Masliyah, J. H. (2009). Role of bicarbonate ions in oil sands extraction systems with a poor processing ore. Journal of Dispersion Science and Technology, 30(6), 809-822. Zhao, H., Long, J., Masliyah, J., & Xu, Z. (2006). Effect of divalent cations and surfactants on silicabitumen interactions. Industrial & Engineering Chemistry Research, 45(22), 7482-7490. Zhou, Z. A., Xu, Z., & Finch, J. A. (1996). Effect of gas nuclei on hydrophobic coagulation. Journal of Colloid and Interface Science, 179(1), 311-314. Zhou, Z. A., Xu, Z., Masliyah, J. H., & Czarnecki, J. (1999). Coagulation of bitumen with fine silica in model systems. Colloids and Surfaces A: Physicochemical and Engineering Aspects, 148(3), 199-211. Zhou, Z., Kasongo, T., Xu, Z., & Masliyah, J. (2004). Assessment of bitumen recovery from the Athabasca oil sands using a laboratory Denver flotation cell. Canadian Journal of Chemical Engineering, 82(4), 696-703.  187  Appendices Appendix 1 Mathematical description of SV and SE data The c(s) distribution from an SV experiment can be defined as: [A1.1] where a(r,t) denotes the experimental sedimentation data as a function of radial position r and time t, c(s) is the concentration of macromolecules with sedimentation coefficient between s and s+ds and is the Lamm equation solution at unit loading concentration of a macromolecule with sedimentation coefficient s and diffusion coefficient D. Mathematical methods using maximum entropy regularization are implemented in SEDFIT and can be used to obtain a solution to equation A1.1 (Schuck, 2000). In both distributions, integration of a peak of interest or the whole distribution gives the weight average sedimentation coefficient for the sedimenting macromolecules. In an SE experiment, the signal measured for a mixture of non-interacting ideally sedimenting species is a function of radial position, a(r), and follows the equation below: [A1.2] where the summation is over all n species; cn,0 denotes the molar concentration of species n at a reference position r0; Mn,  ,  denote the molar mass, partial specific volume, and the molar  extinction coefficient, respectively ; d is the optical path length; and δ is a baseline offset. In equation A1.2, the exponential distribution of SE is the sum of the exponentials of the macromolecular species present in solution. The term  , is the buoyant or reduced molar  mass in accordance with Archimedes principle, i.e., the mass of a macromolecule acted on by the applied centrifugal force in the cell is reduced by the mass of solvent displaced. For materials such as single proteins, only one exponential distribution will be present, and fitting an exponential model to the data can readily produce a weight average molar mass. Global nonlinear regression fitting of multiple data sets extends the application of SE to more complex systems such as self-associating proteins. 188  Transformation of a g(s) distribution into a molecular weight distribution The transformation from g(s) versus s for materials with randomly coiled polymers is as follows: [A1.3] from which, [A1.4] For a randomly coiled polymer, [A1.5] where ks is a constant for a given polymer/solvent/temperature system and the exponent 0.5 corresponds to a randomly coiled polymer under theta conditions. The differential required in Equation A1.4 can be obtained from Equation A1.5 as follows: [A1.6] In the Extended Fujita method, Equation A1.5 is rewritten to describe polymers with a b value different from 0.5. [A1.7] where b = 0.4-0.5 for a coil, ~0.15-0.2 for a rod and ~0.67 for a sphere (Kulicke and Clasen, 2004). The differential is: [A1.8] In order to do the transformation, b (conformation of the polymer) and at least one pair of s-M values need to be known in order to obtain ks in Equation A1.7. The s values used to define ks should be obtained at solution and temperature conditions similar to those under which the data to be transformed were obtained. The s values should also be obtained at low enough concentration to minimize non-ideality effects. The s-M pair of values (tab or space-delimited) can be read as an ASCII file in SEDFIT and the software provides b by linear regression from the slope of a log s versus log M plot. Once the ls-g*(s) versus s distribution is obtained from the sedimentation data (as 189  explained before), the contribution of diffusion needs to be corrected for by specifying a frictional ratio in the data analysis to obtain the c(s) versus s distribution. As described before, large polymeric, non-globular systems do not show significant diffusion that would affect the width of the peak of the sedimentation coefficient distribution and so the ls-g*(s) profiles will give a good representation of the sedimentation coefficient distribution. The transformation to f(M) versus M can then be implemented by numerically solving the following integral equation: [A1.9] where a(r,t) denotes the experimental data as a function of radial position r and time t, f(M) is the is the Lamm equation solution at unit  unknown molecular weight distribution,  loading concentration of a macromolecule with sedimentation coefficient s and diffusion coefficient D, and b(r) and (t) are systematic baseline noise contributions.  190  Appendix 2 Linear regression of a plot of log s versus log MW using the data from Brown and Henley (1964). The b and ks values obtained from the plot were used in the transformation of the sedimentation coefficient distribution, ls-g*(s) into a molecular weight distribution, f(M).  191  Appendix 3 Transformation of a ls-g*(s) vs. s distribution (A) to a MW distribution f(M) versus M (C) for MM-CMC. Also shown is the c(s) distribution (B) showing the MW information of the peaks in the ls-g*(s) distribution. The f(M) distribution was integrated up to the maximum MW shown in the c(s) distribution.  A  B  C  192  Appendix 4 Transformation of a ls-g*(s) vs. s distribution (A) to a MW distribution f(M) versus M (C) for HM-CMC. Also shown is the c(s) distribution (B) showing the MW information of the peaks in the ls-g*(s) distribution. The f(M) distribution was integrated up to the maximum MW shown in the c(s) distribution.  A  B  C  193  Appendix 5 Figure A5.1. Reduced viscosity vs. polymer concentration for DS0.7 in distilled water, 25 °C at different pH.  Figure A5.2. Reduced viscosity vs. polymer concentration for DS0.9 in distilled water, 25 °C at different pH.  194  Figure A5.3. Reduced viscosity vs. polymer concentration for LM-CMC in distilled water, 25 °C at different pH.  Figure A5.4. Reduced viscosity vs. polymer concentration for MM-CMC in distilled water, 25 °C at different pH.  195  Figure A5.5. Reduced viscosity vs. polymer concentration for HM-CMC in distilled water, 25 °C at different pH.  196  Appendix 6 A comparison of the intrinsic viscosities obtained at different pH, temperature and ionic strength using the Fedors equation and by fitting the experimental data with a 2nd order polynomial. (a) Polymers of different DS (DS0.7, DS0.9, and DS1.2) Intrinsic Viscosity [dL/g] in Distilled Water DS 0.7 0.9 1.2  °  T ( C) 25 50 25 50 25 50  pH 3 Fedors Polynomial 15.5 15.4 15.2 15.0 22.9 22.5 20.7 20.6 28.6 28.4 27.4 26.3  pH 4.5 Fedors Polynomial 69.4 68.4 68.5 66.9 79.4 81.5 79.4 80.7 84.7 84.6 82.0 81.7  Fedors 96.2 95.2 106.4 103.1 114.9 113.6  pH 7 Polynomial 93.9 92.4 105.3 102.1 121.5 117.3  Intrinsic Viscosity [dL/g] in 0.01 mol/L NaCl DS 0.7 0.9 1.2  °  T ( C) 25 50 25 50 25 50  pH 3 Fedors Polynomial 11.8 11.8 11.2 11.1 14.9 14.7 14.0 14.2 17.6 18.0 16.5 16.4  pH 4.5 Fedors Polynomial 14.0 13.9 13.4 13.3 20.0 21.3 18.9 20.2 22.8 23.9 21.1 22.0  197  pH 7 Fedors Polynomial 16.8 17.6 16.0 16.9 20.5 20.6 19.6 19.5 23.9 23.8 23.3 22.8  (b) Polymers of different MW (LM-CMC, MM-CMC, and HM-CMC)  T (°C) 25 50 MM-CMC 25 50 HM-CMC 25 50 MW LM-CMC  Intrinsic Viscosity [dL/g] in Distilled Water pH 3 pH 4.5 pH 7 Fedors Polynomial Fedors Polynomial Fedors Polynomial 20.5 19.2 45.5 46.2 78.1 77.7 20.6 18.8 44.6 45.1 76.9 76.4 23.9 21.2 74.6 73.1 86.2 87.4 22.3 19.4 74.1 72.8 85.5 86.6 34.1 33.0 97.1 98.3 144.9 145.9 32.1 30.6 94.3 96.4 144.9 145.7  Intrinsic Viscosity [dL/g] in 0.01 mol/L NaCl pH 3 pH 4.5 pH 7 ° MW T ( C) Fedors Polynomial Fedors Polynomial Fedors Polynomial LM-CMC 25 5.9 7.3 7.1 6.3 9.5 9.6 6.6 7.1 6.8 6.1 9.3 10.8 50 MM-CMC 25 14.6 13.9 17.5 16.2 23.9 24.0 13.9 12.7 16.6 15.5 21.3 22.0 50 HM-CMC 25 25.1 25.3 32.8 34.7 42.7 44.1 24.2 24.6 31.3 32.7 40.2 42.0 50  198  Appendix 7 Intrinsic viscosity at infinite ionic strength for DS0.7 and DS1.2 (intercept with the y-axis) according to Equation 3.4 (section 3.3.3.5)  199  Appendix 8 Calculation of the constant M from the elongated fixture (double gap geometry) dimensions  The figure below shows the top view of the elongated fixture (Figure 4.2, section 4.2.2.1) with dimensions r1 (radius to the outside of the inner cylinder) = 16.5 mm, r2 (radius to the inside of the bob) = 19.0 mm, r3 (radius to the outside of the bob) = 20.0 mm, and r4 (radius to the inside of the cup) = 23.03 mm.  The constant M is given by (Klein et al., 1995): [A8.1] Putting the fixture dimensions into the above equation, the value of M was calculated as 3.736.  200  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0073935/manifest

Comment

Related Items