UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Catalyst recycle in slurry-phase residue upgrading Rezaei, Hooman 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_rezaei_hooman.pdf [ 15.31MB ]
Metadata
JSON: 24-1.0073852.json
JSON-LD: 24-1.0073852-ld.json
RDF/XML (Pretty): 24-1.0073852-rdf.xml
RDF/JSON: 24-1.0073852-rdf.json
Turtle: 24-1.0073852-turtle.txt
N-Triples: 24-1.0073852-rdf-ntriples.txt
Original Record: 24-1.0073852-source.json
Full Text
24-1.0073852-fulltext.txt
Citation
24-1.0073852.ris

Full Text

Catalyst Recycle in Slurry-phase Residue Upgrading by Hooman Rezaei M.Sc, Amirkabir University of Technology, 2006 B.Sc., Sahand University of Technology, 2003  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Chemical and Biological Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) May 2013 ? Hooman Rezaei, 2013 Abstract The application of slurry-phase hydroconversion to upgrade residue oil derived from the Canadian oilsands (CLVR) is hindered by the cost of the catalyst, in part because the catalyst is used once-through before being discarded as part of the solid product (coke). The goal of the present study was to assess the potential of recycling the slurry-phase catalyst under high residue conversion conditions and to identify the cause of catalyst deactivation during catalyst recycle. Catalyst screening in a batch reactor operated at 415 oC and 5.6 MPa initial H2 pressure with CLVR as reactant, showed that MoS2 prepared in reversed micelles was most active for coke suppression and liquid yield among a series of Fe- and Mo-based catalysts. Furthermore, MoS2 derived from Mo-micelle and Mo-octoate precursors had equivalent coke yields, but were more active for coke suppression than a water-soluble ammonium heptamolybdate precursor, as measured in a semi-batch reactor under more severe hydroconversion conditions (T = 445 oC and H2 pressure of 13.8 MPa and 900 mL(STP)/min).  MoS2 prepared using Mo-micelle and Mo-octoate precursors over a range of Mo concentrations (0 ? 1800 ppm) were recycled in the semi-batch reactor to assess the activity of the recycled catalyst in terms of coke suppresion and selectivity toward different products. The MoS2 catalyst remained active for up to 4 reaction cycles, depending on the initial concentration of Mo added to the reactor. Characterization of the coke and catalyst recovered after each recycling step showed that the coke associated with the catalyst undergoes significant chemical and morphological changes during recycle and these changes result in  ii deactivation of the catalyst. A conceptual model of the catalyst deactivation mechanism based on the characterization results was developed and ex-situ simulation of catalyst aging validated the proposed deactivation mechanism in the semi-batch slurry-phase upgrading reactor. Finally, a kinetic model of the CLVR hydroconversion reactions was developed that included the consumption and production of coke as an important step in the overall kinetic scheme.     iii Preface This PhD thesis is divided into nine chapters. Chapters 4, 5, 6 of the thesis have been published in peer reviewed journals, Chapter 7 has been submitted for publication and the materials presented in Chapter 8 are being prepared in the manuscript format for publication. Experimental design, performing experiments, catalyst preparation and characterization, kinetic modeling, literature review, manuscript and thesis preparation was done by Hooman Rezaei under the supervision of Professor Kevin J. Smith in the Chemical and Biological Engineering Department at the University of British Columbia. The list of publications along with a brief description of the contribution of the co-authors is presented below: 1- Rezaei, Hooman; Liu, Xuebin; Jooya Ardakani, Shahrzad; Smith, Kevin J. and Bricker, Maureen; ?A study of Cold Lake Vacuum Residue hydroconversion in batch and semi-batch reactors using unsupported MoS2 catalysts?; Catalysis Today, 150, 3-4, 244-254, 2010. Dr. Xuebin Liu did some preliminary studies to identify the optimum ratio of the solvent to surfactant in preparation of Mo-micelle catalysts. Dr. Shahrzad Jooya Ardakani performed a part of the high resolution transmission electron microscopy (HRTEM) analysis. Dr. Maureen Bricker contributed to the manuscript by taking part in the discussions of the data analysis.  iv 2- Rezaei, Hooman; Jooya Ardakani, Shahrzad and Smith, Kevin J., ?Comparison of MoS2 catalyst prepared in-situ using Mo-micelle and Mo-octoate precursors for hydroconversion of Cold Lake Vacuum Residue: Catalyst activity, produced coke properties and catalyst recyclability?, Energy&Fuels, 26, 2768-2778, 2012. Dr. Shahrzad Jooya Ardakani helped in performing a part of the HRTEM analysis and contributing in the discussions. 3- Rezaei, Hooman; Jooya Ardakani, Shahrzad and Smith, Kevin J., ?A Study of MoS2 Catalyst Recycle in Slurry-phase Residue Hydroconversion?, Energy&Fuels, 26, 6540-6550, 2012. Dr. Shahrzad Jooya Ardakani helped in performing a part of the x-ray diffraction (XRD) characterization and asphaltene analysis. Except for the above mentioned contributions of the co-authors, all the experimental design, catalyst preparation and characterization, performing experiments, literature review, data processing and analysis and preparation of the manuscripts were done by Hooman Rezaei under supervision and final approval of Professor Kevin J. Smith in the Chemical and Biological Engineering Department at the University of British Colombia.      v Table of Contents Abstract ..................................................................................................................................... ii Preface ...................................................................................................................................... iv Table of Contents ..................................................................................................................... vi List of Tables .......................................................................................................................... xv List of Figures ......................................................................................................................... xx Nomenclature ....................................................................................................................... xxxi Acknowledgements ............................................................................................................. xxxv Dedication ......................................................................................................................... xxxvii Chapter 1 ................................................................................................................................... 1 Introduction ............................................................................................................................... 1 1.1 Background ........................................................................................................ 1 1.2 Objectives of the Thesis ..................................................................................... 6 1.3 Approach of the Thesis ....................................................................................... 6 Chapter 2 ................................................................................................................................... 9 Literature Review...................................................................................................................... 9 2.1 Petroleum, Heavy Oil, Residue and Synthetic Crude Oil .................................. 9 2.2 Definition of Primary and Secondary Upgrading ............................................. 11 2.3 Heavy-oil and Residue Primary Upgrading Processes ..................................... 14 2.3.1 Thermal Cracking (Carbon Rejection) Processes ........................................ 14  vi 2.3.2 Catalytic Hydroconversion (Hydrogen Addition) Processes ....................... 14 2.3.2.1 Fixed-bed Processes ........................................................................... 16 2.3.2.2 Ebullated-bed Processes ..................................................................... 18 2.3.2.3 Slurry-phase Processes ....................................................................... 20 2.3.2.3.1 Exxon M-coke Process ................................................................ 23 2.3.2.3.2 Eni Slurry Technology (EST) ...................................................... 24 2.4 Catalysts Used for Hydroconversion ................................................................ 26 2.4.1 Conventional Catalysts Used in Hydroprocessing ....................................... 26 2.4.2 Non-conventional Catalysts for Hydroprocessing ....................................... 28 2.4.2.1 Soluble Catalysts ................................................................................ 29 2.4.2.1.1 Oil-soluble Precursors ................................................................. 29 2.4.2.1.2 Water-soluble Precursors ............................................................. 32 2.4.2.1.3 Finely-powdered Precursors ........................................................ 33 2.5 Recovery of Dispersed/ Dissolved Catalyst ..................................................... 34 2.6 Catalyst Recycle in Heavy-oil and Residue Upgrading ................................... 36 2.7 Chemistry of Upgrading Reactions .................................................................. 38 2.7.1 Thermal Reactions ....................................................................................... 39  vii 2.7.1.1 Carbon-Carbon Bond Breakage .......................................................... 40 2.7.1.2 Reaction of Mixtures of Hydrocarbons .............................................. 41 2.7.1.3 Reactions of Sulfur, Nitrogen and Oxygen Compounds .................... 42 2.7.2 Catalytic Reactions ...................................................................................... 43 2.7.2.1 Reactions of Aromatics ...................................................................... 44 2.7.2.2 Reactions of Sulfur and Nitrogen Compounds ................................... 46 2.7.3 The Mechanism of Coke Formation during Upgrading ............................... 47 2.7.4 The Mechanism of Catalytic Coke Suppression .......................................... 51 2.8 Lumped Kinetic Modeling of Residue Hydroconversion ................................ 54 2.9 Summary .......................................................................................................... 56 Chapter 3 ................................................................................................................................. 58 Experimental Materials, Apparatus and Procedures ............................................................... 58 3.1 Catalyst Precursor Preparation ......................................................................... 58 3.1.1 Micelle Precursor ......................................................................................... 58 3.1.2 Commercial Oil-soluble (Mo-octoate) Precursor ........................................ 63 3.1.3 Water-soluble Precursor ............................................................................... 63 3.1.4 Finely-powdered precursor .......................................................................... 63  viii 3.2 Batch Reactor Experimental Procedure ........................................................... 64 3.3 Semi-batch Reactor Experimental Procedure ................................................... 66 3.4 Experimental Procedure for Heat-up and Blank (thermal) Experiments ......... 71 3.5 Experimental Procedure for Generating Data for Kinetic Modeling ............... 71 3.6 Calculation of Yields, Conversions and H2 Uptake ......................................... 72 3.6.1 Residue and Toluene-insoluble Organic Residue (TIOR) Conversions ...... 74 3.6.2 Asphaltene Content Measurement ............................................................... 75 3.7 Instrumental Analyses Used for Catalyst and Coke Characterization ............. 76 3.7.1 Thermogravimetric Analysis (TGA) ............................................................ 76 3.7.2 Elemental (CHNS) Analysis ........................................................................ 77 3.7.3 Diffuse Reflectance Infrared Fourier Transform (DRIFT) Spectroscopy ... 77 3.7.4 X-ray Diffraction (XRD) Analysis .............................................................. 79 3.7.5 Solid State 13C Nuclear Magnetic Resonance (NMR) ................................. 80 3.7.6 Scanning Electron Microscopy (SEM) ........................................................ 81 3.7.7 Transmission Electron Microscopy (TEM) and High Resolution TEM (HRTEM) ..................................................................................................... 82 3.7.8 High Temperature Simulated Distillation (SIMDIS) ................................... 83 3.7.9 X-ray Photoelectron Spectroscopy (XPS) ................................................... 84  ix 3.7.10 Surface Area Analysis by the Brunauer, Emmett and Teller (BET) Method 84 3.7.11 Particle Size Analysis using Dynamic Light Scattering (DLS) ................... 86 Chapter 4 ................................................................................................................................. 87 Preliminary Catalyst Screening ............................................................................................... 87 4.1 Introduction ...................................................................................................... 87 4.2 Catalyst Activity Measurements ...................................................................... 89 4.2.1 Preliminary Screening of Catalysts .............................................................. 89 4.2.2 Comparison of Mo-micelle and AHM precursors measured in the Batch and Semi-batch Reactors .................................................................................... 95 4.3 Preliminary Coke-catalyst Characterization ................................................... 111 4.4 Conclusion ...................................................................................................... 122 Chapter 5 ............................................................................................................................... 124 Comparison of MoS2 Catalysts Prepared from Mo-micelle and Mo-octoate Precursors for Hydroconversion of Cold Lake Vacuum Residue: Catalyst Activity, Coke Properties and Catalyst Recycle .................................................................................................................... 124 5.1 Introduction .................................................................................................... 124 5.2 Results and Discussions ................................................................................. 126 5.2.1 Effect of Catalyst Precursor and Concentration ......................................... 126 5.2.2 Effect of Temperature and Heat-up Period ................................................ 138  x 5.2.3 Coke Properties .......................................................................................... 140 5.2.4 Preliminary Catalyst Recycle Experiments ............................................... 152 5.3 Conclusions .................................................................................................... 154 Chapter 6 ............................................................................................................................... 156 MoS2 Catalyst Recycle in Slurry-phase Residue Hydroconversion ..................................... 156 6.1 Introduction .................................................................................................... 156 6.2 Results and Discussions ................................................................................. 157 6.2.1 Activity of Recycled Catalyst .................................................................... 157 6.2.2 Optimum Coke Recycle Ratio Calculations in a Commercial Plant ......... 173 6.2.3 Significance of Solid Loading ................................................................... 174 6.2.4 Catalyst Deactivation ................................................................................. 180 6.3 Conclusions .................................................................................................... 186 Chapter 7 ............................................................................................................................... 188 Conceptual Deactivation Model of Recycled MoS2 Catalyst ............................................... 188 7.1 Introduction .................................................................................................... 188 7.2 Results and Discussions ................................................................................. 190 7.2.1 Activity of recycled catalyst ...................................................................... 190 7.2.2 TEM Analysis of the Fresh and Recycled Coke ........................................ 191  xi 7.2.3 XRD Analysis of the Fresh and Recycled Coke ........................................ 193 7.2.4 CHN Analysis of the Fresh and Recycle Coke .......................................... 194 7.2.5 EDX and XPS Analysis of the Fresh and Recycled Coke ......................... 198 7.2.6 BET Surface Area and BJH Pore Volume Analysis of the Fresh and Recycled Coke ........................................................................................... 202 7.2.7 13C Solid-State NMR Analysis of the Fresh and Recycled Coke .............. 203 7.2.8 TGA Analysis of the Fresh and Recycled Coke ........................................ 207 7.2.9 Activity and Characterization of Ex-situ Aged Coke-catalyst Mixture ..... 209 7.2.10 Conceptual model of catalyst deactivation ................................................ 215 7.3 Conclusion ...................................................................................................... 220 Chapter 8 ............................................................................................................................... 221 Kinetic Study of Residue Upgrading in the Semi-batch Reactor ......................................... 221 8.1 Introduction .................................................................................................... 221 8.2 Results and Discussions ................................................................................. 223 8.2.1 Catalyst Activity Test at Different Temperatures ...................................... 223 8.2.2 Developing the Kinetic Model ................................................................... 226 8.2.3 Data Normalization .................................................................................... 226 8.2.4 Kinetic Model Development ...................................................................... 228  xii 8.2.5 Parameter Estimation and Statistical Analysis Methodology .................... 232 8.2.6 Estimated Parameters and Fit of the Model ............................................... 233 8.3 Conclusions .................................................................................................... 241 Chapter 9 ............................................................................................................................... 243 Conclusions and Recommendations ..................................................................................... 243 9.1 Conclusions .................................................................................................... 243 9.2 Recommendations .......................................................................................... 246 9.2.1 Investigation of Hydroconversion in a Continuous Bubble-column Reactor 246 9.2.2 Quantifying the Onset of Catalyst Deactivation ........................................ 246 9.2.3 Recycle of Sludge-coke-catalyst instead of Coke-catalyst ........................ 247 9.2.4 Development of More Sophisticated Kinetic Model ................................. 247 Bibliography ......................................................................................................................... 248 Appendices ............................................................................................................................ 261 Appendix A Thermal Cracking Processes (Carbon Rejection Processes) ............ 262 A.1 Thermal Viscosity Reduction (Visbreaking) ............................................. 262 A.2 Coking Processes ....................................................................................... 262 Delayed Coking ..................................................................................... 263  xiii Fluid Coking .......................................................................................... 265 Flexicoking ............................................................................................ 267 Appendix B Geometry and Mixing Pattern of the Batch Reactor ......................... 269 Appendix C H2S Yield Sample Calculations ........................................................ 270 Appendix D Product Work-up of Batch Reactor Experiments ............................. 272 Appendix E Product Work-up of Semi-batch Reactor Experiments .................... 273 Appendix F Detailed Operating Procedure of Semi-batch HCR Experiments ..... 274 Appendix G Summary of Experimental Results of Batch Reactor Experiments .. 279 Appendix H Summary of Experimental Results of Semi-Batch Reactor Experiments ...................................................................................... 284 Appendix I Temperature and Pressure Profile of a Typical HCR Experiment .... 304 Appendix J Kubelka-Munk Transformation ........................................................ 305 Appendix K Molecular Weigh Calculations for the Gas, Liquid and Solid Products 306 Appendix L MATLAB M-files for Objective Function Minimization and Parameter Estimation ......................................................................................... 311 Appendix M Statistical Analysis of the Kinetic Model ......................................... 320 M.1 Significance and Calculation of p-vaue ..................................................... 320 M.2 Calculation of Standard Deviation and Confidence Interval ..................... 320 Appendix N Anova analysis and Error Calculation for Experimental and Model-predicted Values of Kinetic Lumps .................................................. 322  xiv List of Tables Table  2.1 Typical properties of different petroleums [6]. (Copyright ? 2006, Taylor & Francis Group LLC., Reproduced with permission) ............................................... 10 Table  2.2 Comparison of light and heavy crudes [5]. (Copyright ? 1994, Taylor & Francis Group LLC., Reproduced with permission) ........................................................... 13 Table  2.3 Different slurry-phase processes developed in the last few decades. ..................... 22 Table  2.4 Bond dissociation energies of different bonds. Data from [89]. (Copyright ? 1982, Annual Reviews, Reproduced with permission) ..................................................... 41 Table  3.1 Summary of microemulsions prepared with single metals (Fe, Mo and Co) and bi-metals (Fe-Mo, Co-Mo and Fe-Co). ....................................................................... 61 Table  3.2 Summary of DLS particle size estimates of reverse micelles present in catalyst precursor. ................................................................................................................ 62 Table  3.3 Properties of CLVR used as reactant in the present study. ..................................... 65 Table  3.4 Summary of operating conditions of experiments done for kinetic study. ............. 72 Table  4.1 Summary of the experimental conditions in the batch and semi-batch reactors. ... 95 Table  4.2 Result of CLVR hydroconversion experiments in the batch reactor using dispersed catalysts compared to literature data. .................................................................... 102 Table  4.3 Results of CLVR hydroconversion experiments using dispersed catalysts in the semi-batch reactor. ................................................................................................ 110  xv Table  4.4 Mo recovery from the coke samples recovered from the batch reactor experiments after reaction. ........................................................................................................ 112 Table  5.1 Coke yield, TIOR and asphaltene conversion, and liquid-phase S conversion in the heat-up experiments and the experiments using different reaction temperatures. 140 Table  5.2 BET surface area and total pore volume of coke recovered from non-catalytic and catalytic CLVR hydroconversion experiments. .................................................... 141 Table  5.3 TGA of coke recovered from hydroconversion reactions using different types of catalysts at different concentrations. ..................................................................... 149 Table  5.4 EDX and XPS analysis of coke generated from CLVR hydroconversion experiments in the semi-batch reactor. ................................................................. 151 Table  5.5 Results from CLVR hydroconversion using different ratios of fresh and recycled catalyst. ................................................................................................................. 153 Table  6.1 Comparison of repeated recycle experiments using 300 ppm Mo in the form of Mo-micelle precursor ................................................................................................... 158 Table  6.2 Experimental repeatability in hydroconversion experiments a ............................. 159 Table  7.1. EDX analysis of coke samples recovered after different stages of recycling experiments using 600 ppm Mo added in the form of Mo-micelle precursor. ..... 199 Table  7.2 XPS analysis of coke samples recovered after different stages of recycling experiments using 600 ppm Mo added in the form of Mo-micelle precursor. ..... 201  xvi Table  7.3. EDX and EPX analysis of coke samples recovered after different stages of recycling 600 ppm Mo added in the form of Mo-octoate precursor. .................... 201 Table  7.4. BET surface area and BJH total pore volume of coke samples recovered from fresh and 5th recycle of 600 ppm Mo in the form of Mo-micelle and Mo-octoate and a thermal experiment. ..................................................................................... 202 Table  7.5. TGA analysis of coke samples recovered from heat-up, fresh hydroconversion experiment and 5th recycle of 600 ppm and 1800 ppm Mo added to the reactor in the form of Mo-micelle and Mo-octoate. .............................................................. 208 Table  7.6. BET surface area, peak integral intensity ratio of aromatic to aliphatic carbon (CAr/CAl) and H/C atom ratio of fresh coke before and after thermal aging. ........ 211 Table  8.1 Summary of the different yields and H2 uptake in the hydroconversion experiments using 600 ppm Mo in the form of Mo-octoate ...................................................... 225 Table  8.2 Summary of normalized kinetic data (weight-base) ............................................. 227 Table  8.3 Summary of normalized kinetic data (mole-base) ................................................ 229 Table  8.4 Kinetic parameters of reaction calculated by the model. ...................................... 239 Table  C.1 Saample values for H2S yield calculation in semi-batch reactor experiments. .... 270 Table  G.1 Summary of experimental results of batch reactor experiments. ......................... 279 Table  H.1 Summary of experimental results of semi-batch reactor experiments. ................ 284 Table  K.1 Summary of sample calculations for calculating average molecular weight of gas mixture. ................................................................................................................. 307  xvii Table  K.2 Molecular weight of different HC cuts at different hydroconversion experiments. .............................................................................................................................. 309 Table  N.1 Anova analysis and error calculation for coke at 415 ?C ..................................... 322 Table  N.2 Anova analysis and error calculation for residue at 415 ?C ................................. 323 Table  N.3 Anova analysis and error calculation for HGO at 415 ?C ................................... 324 Table  N.4 Anova analysis and error calculation for LGO at 415 ?C .................................... 325 Table  N.5 Anova analysis and error calculation for naphtha at 415 ?C ............................... 326 Table  N.6 Anova analysis and error calculation for gas at 415 ?C ....................................... 327 Table  N.7 Anova analysis and error calculation for coke at 430 ?C ..................................... 328 Table  N.8 Anova analysis and error calculation for residue at 430 ?C ................................. 329 Table  N.9 Anova analysis and error calculation for HGO at 430 ?C ................................... 330 Table  N.10 Anova analysis and error calculation for LGO at 430 ?C .................................. 331 Table  N.11 Anova analysis and error calculation for naphtha at 430 ?C ............................. 332 Table  N.12 Anova analysis and error calculation for gas at 430 ?C ..................................... 333 Table  N.13 Anova analysis and error calculation for coke at 445 ?C................................... 334 Table  N.14 Anova analysis and error calculation for residue at 445 ?C............................... 335 Table  N.15 Anova analysis and error calculation for HGO at 445 ?C ................................. 336 Table  N.16 Anova analysis and error calculation for LGO at 445 ?C .................................. 337 Table  N.17 Anova analysis and error calculation for naphtha at 445 ?C ............................. 338 Table  N.18 Anova analysis and error calculation for gas at 445 ?C ..................................... 339  xviii Table  N.19 Anova analysis and error calculation for coke at 460 ?C................................... 340 Table  N.20 Anova analysis and error calculation for residue at 460 ?C............................... 341 Table  N.21 Anova analysis and error calculation for HGO at 460 ?C ................................. 342 Table  N.22 Anova analysis and error calculation for LGO at 460 ?C .................................. 343 Table  N.23 Anova analysis and error calculation for naphtha at 460 ?C ............................. 344 Table  N.24 Anova analysis and error calculation for gas at 460 ?C ..................................... 345            xix List of Figures Figure  2.1 Schematic of bunker reactor technology developed by Shell [26]. (Copyright ? 1998, Elsevier, with permission) ............................................................................ 18 Figure  2.2 A schematic of H-Oil prcess [6]. (Copyright ? 2006, Taylor & Francis Group LLC., with permission) ........................................................................................... 20 Figure  2.3 Hydrogenation/dehydrogenation cycle of naphthalene [5]. (Copyright ? 1994, Taylor & Francis Group LLC., Reproduced with permission) ............................... 45 Figure  2.4 Hydrogenation/dehydrogenation cycle of phenathrene [5]. (Copyright ? 1994, Taylor & Francis Group LLC., Reproduced with permission) ............................... 45 Figure  2.5 Schematic of hydrodesulfurization of dibenzothiophene (DBT). Reproduced from [92]. (Copyright ? 1994, Taylor & Francis Group LLC., Reproduced with permission) .............................................................................................................. 46 Figure  2.6 Hydrodenitrogenation (HDN) mechanism of pyrrole [5]. (Copyright ? 1994, Taylor & Francis Group LLC., Reproduced with permission) ............................... 47 Figure  2.7 Schematic of the phase-separation mechanism for coke formation in thermal conversion of residue [97]. (Copyright ? 2007, American Chemical Society, with permission) .............................................................................................................. 49 Figure  2.8 Optical micrograph of toluene-insoluble coke from thermolysis of the Cold Lake vacuum residue [95]. (Copyright ? 1993, American Chemical Society, with permission) .............................................................................................................. 51  xx Figure  2.9 Extended chain reaction mechanism for liquid-phase cracking, including catalytic reactions [87] (Copyright ? 2002, American Chemical Society, Reproduced with permission) .............................................................................................................. 52 Figure  2.10 Proposed kinetic model by Sanchez et al. [107]. (Copyright ? 2005, American Chemical Society, Reproduced with permission) ................................................... 55 Figure  3.1 Micelle size distribution in the emulsion analyzed by DLS. Solvent : surfactant = 3.1:1, equivalent Mo concentration in the feed = 600 ppm. ................................... 60 Figure  3.2 HRTEM micrographs of Mo (top right), Fe (top left), Mo-Co (bottom right) and Mo-Fe (bottom left) micelle catalysts prepared in reversed micelles and reduced in LiBH4.  The bars on each micrograph correspond to 20 nm. ................................. 62 Figure  3.3 Schematic of batch reaction system. ..................................................................... 66 Figure  3.4 Schematic of the semi-batch reaction system. ....................................................... 67 Figure  3.5 SEM micrograph of 4 different coke samples prior to recycling. ......................... 70 Figure  3.6 SEM micrographs of three different coke samples before DRIFT analysis. (A) 3 wt % FeS used as the catalyst, (B) 5th recycle of 600 ppm Mo using Mo-micelle precursor and (C) a thermal experiment. All the images have the same scale as (A). ................................................................................................................................ 79 Figure  4.1 CLVR hydroconversion using Mo (Mo-M) and Fe (Fe-M) catalysts prepared in reversed micelles compared to AHM (Mo-W) and iron sulphate (Fe-S) catalysts. T = 415?C, 5.5 MPa initial H2 pressure and 1 hour reaction time. ............................ 91  xxi Figure  4.2 CLVR hydroconversion using Mo-Co (MoCo-M) and Mo-Fe (MoFe-M) catalysts prepared in reversed micelles compared to Mo-M and Fe-M catalysts. T = 415?C, 5.5 MPa initial H2 pressure and 1 hour reaction time. ............................................ 92 Figure  4.3 CLVR hydroconversion using Mo (MoM) and Fe (FeM) catalysts prepared in reversed micelles over a range of metal concentrations in the CLVR.  Data for catalysts prepared in situ from AHM (MoW) and iron sulphate (FeS) as well as the results from a thermal experiment are included for comparison. ........................... 94 Figure  4.4 Comparison of H2 conversion and consumption in a series of experiments in the batch reactor using Mo-micelle precursor with different Mo concentrations. T = 415 oC, PH2, initial ~ 5.5 MPa and reaction time = 1h. ............................................... 96 Figure  4.5 Comparison of H2 conversion and consumption in two experiments in the batch reactor using 600 ppm Mo in the form of micelle and water-soluble AHM precursors. T = 415 oC, PH2, initial ~ 5.5 MPa and reaction time = 1h. ..................... 98 Figure  4.6 Coke yield in a series of experiments in the batch reactor using different concentrations of Mo-micelle precursor and 600 ppm of Mo using AHM precursor. T = 415 oC, PH2, initial ~ 5.5 MPa and reaction time = 1h. .................................. 100 Figure  4.7 Comparison of residue conversion in the experiments done in the batch reactor using different Mo precursors and concentrations with thermal experiment (M: Micelle precursor). T = 415 oC, PH2, initial ~ 5.5 MPa and reaction time = 1h. ...... 103 Figure  4.8 Comparison of coke and liquid yields, S conversion and residue conversion in the batch and semi-batch reactors using Mo micelle and catalyst prepared in situ using  xxii AHM. Reaction conditions: batch- T = 415 oC, PH2, initial ~ 5.5 MPa and 1 h reaction time; semi-batch- T = 415 oC, PH2 ~ 13.8 MPa at 900 mL(STP)/min and 1 h reaction. ................................................................................................................. 105 Figure  4.9 Comparison of different hydrocarbon cuts in liquid products from experiments in the batch and semi-batch reactors using Mo micelle and catalyst prepared in situ using AHM. Reaction conditions: batch- T = 415 oC, PH2, initial ~ 5.5 MPa and 1 h reaction time; semi-batch- T = 415 oC, PH2 ~ 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. ..................................................................................................... 106 Figure  4.10 Comparison of H2 consumption in experiments using different concentrations and types of catalyst in the semi-batch reactor. T = 445 oC, PH2 ~ 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. .................................................................... 108 Figure  4.11 Coke yield comparison between experiments using different catalyst types and concentrations in the semi-batch reactor. T = 445 oC, PH2 ~ 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. .................................................................... 109 Figure  4.12 Comparison of Mo content in different coke samples measured by EDX and calculated amount assuming all Mo added to the CLVR is associated with the coke. .............................................................................................................................. 113 Figure  4.13 SEM micrographs of different coke samples from hydroconversion reactions. a) and b): Batch reactor, 600 ppm Mo using AHM, T = 415 oC, PH2, initial ~ 5.5 MPa and 1 h reaction time. c) and d): Semi-batch reactor, thermal experiment, T = 445  xxiii oC, PH2 ~ 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. e): Batch reactor, 600 ppm Mo micelle, T = 415 oC, PH2, initial ~ 5.5 MPa and 1 h reaction time. ..... 115 Figure  4.14 X-ray diffractogram of coke samples from different experiments in the semi-batch reactor before (a) and after (b) washing with water. T = 445 oC, PH2 ~ 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. ................................................. 117 Figure  4.15 Elemental mapping of coke sample generated in an experiment using 1800 ppm Mo micelle in the semi-batch reactor. T = 445 oC, PH2 ~ 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. .................................................................... 119 Figure  4.16 HRTEM images of coke sample generated in an experiment using 1800 ppm Mo micelle in the semi-batch reactor. T = 445 oC, PH2 ~ 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. ............................................................................................ 121 Figure  5.1 Coke and liquid yields from CLVR hydroconversion as a function of Mo concentration for Mo-micelle and Mo-octoate catalyst precursors. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. ...................................................................................................................... 127 Figure  5.2 H2 uptake during CLVR hydroconversion as a function of Mo concentration for Mo-micelle and Mo-octoate catalyst precursors. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. ............................... 129 Figure  5.3 H/C atom ratio and CAr/CAl ratio of solid (coke) and liquid products from CLVR hydroconversion; (?) H/C atom ratio of liquid products, (?) H/C atom ratio of coke, (?) H/C atom ratio of coke and (?) CAr/CAl ratio of coke as a function of Mo  xxiv concentration for Mo-micelle and Mo-octoate catalyst precursors. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time .............................................................................................................................. 132 Figure  5.4 Coke yield and Residue (TIOR) conversion during CLVR hydroconversion as a function of Mo concentration for Mo-micelle and Mo-octoate catalyst precursors. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. ........................................................................................................ 135 Figure  5.5 Liquid product distribution from CLVR hydroconversion as a function of Mo concentration for Mo-micelle (A) and Mo-octoate (B) catalyst precursors. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. ........................................................................................................ 136 Figure  5.6 TIOR conversion to coke yield ratio from CLVR hydroconversion as a function of Mo concentration for Mo-micelle and Mo-octoate catalyst precursors. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time .............................................................................................................................. 138 Figure  5.7 XRD diffractogram of coke recovered from CLVR hydroconversion using 1800 ppm Mo from Mo-micelle and Mo-octoate catalyst precursors. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. ...................................................................................................................... 142 Figure  5.8 TEM micrographs of coke samples recovered from CLVR hydroconversion using different Mo concentrations using Mo-octoate precursor. (A) Non-catalytic  xxv experiment, (B) 600 ppm Mo using Mo-octoate precursor and (C) 1800 ppm Mo using Mo-octoate precursor. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time .............................................................. 144 Figure  5.9 DRIFT spectra of coke samples recovered from CLVR hydroconversion using different concentrations of Mo added in the form of Mo-octoate. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. Reaction time for heat-up experiment was zero and Mo concentration was 600 ppm using Mo-octoate precursor. .................................................................. 146 Figure  6.1 Coke yield versus recycle number from experiments using different catalyst precursors and Mo concentrations: 100 ppm Mo using Mo-micelle precursor (?), 300 ppm Mo using Mo-micelle precursor (?), 600 ppm Mo using Mo-micelle precursor (?) and 600 ppm Mo using Mo-octoate precursor (?). ....................... 160 Figure  6.2 H2 uptake versus the recycle number in recycle experiments starting with different Mo concentrations using Mo-micelle and Mo-octoate precursors: 100 ppm Mo using Mo-micelle (A), 300 ppm Mo using Mo-micelle (B), 600 ppm Mo using Mo-micelle (C) and 600 ppm Mo using Mo-octoate (D). ........................................... 164 Figure  6.3 Coke yield versus H2 uptake in the recycle experiments. ................................... 165 Figure  6.4 TIOR conversions in different recycle experiments starting with different Mo concentrations and precursors. (A) 100 ppm Mo using Mo-micelle precursor, (B) 300 ppm Mo using Mo-micelle precursor, (C) 600 ppm Mo using Mo-micelle precursor and (D) 600 ppm Mo using Mo-octoate precursor. .............................. 167  xxvi Figure  6.5 Selectivity change toward different hydrocarbon cuts in the recycle experiments starting with 100 ppm Mo using Mo-micelle precursor (A), 300 ppm Mo using Mo-micelle precursor (B), 600 ppm Mo using Mo-micelle precursor (C) and 600 ppm Mo using Mo-octoate precursor (D). (?) < 204 ?C, (?) 204 ? 348 ?C, (?) 348 ? 524 ?C and (?) > 524 ?C. ................................................................................. 170 Figure  6.6 Asphaltene conversion in recycle experiments using different initial catalyst concentrations and precursors: 100 ppm Mo using Mo-micelle precursor (?), 300 ppm Mo using Mo-micelle precursor (?), 600 ppm Mo using Mo-micelle precursor (?), 600 ppm Mo using Mo-octoate precursor (?) and a thermal experiment (?). .............................................................................................................................. 172 Figure  6.7 Coke added (?), generated (?) and total coke in the reactor (?) in a series of recycle experiments starting with 600 ppm Mo added to the reactor using Mo-micelle precursor. .................................................................................................. 175 Figure  6.8 Coke yield versus solid loading in the reactor using thermal coke and 600 ppm Mo added to the reactor in the form of Mo-octoate. Straight line shows linear decrease of the coke yield with solid loading in the reactor at solid loading range of zero to 7.75 wt % of coke in the feed. .............................................................................. 177 Figure  6.9 Comparison of coke yields in two experiments using 6 g of thermal coke and pure graphite along with 600 ppm Mo using Mo-octoate precursor in the reactor. Reported values are average of two repeat experiments. ...................................... 179  xxvii Figure  6.10 Coke yield versus Mo concentration in the coke-catalyst mixture used as the catalyst in different recycle experiments using different initial concentrations of Mo added to the reactor in the form of Mo-micelle and Mo-octoate precursors. Vertical line shows the Mo concentration in the coke above which the coke yield will not change with Mo concentration in the recycled coke. .............................. 182 Figure  6.11 Coke yield versus Mo concentration in the coke-catalyst mixture used as the catalyst in several series of recycle experiments using different initial concentrations of Mo added to the reactor in the form of Mo-micelle and Mo-octoate precursors. ................................................................................................ 184 Figure  6.12 XRD diffractogram of coke samples recovered after different hydroconversion experiments in a series of recycle experiments using initial Mo concentration of 1800 ppm added to the reactor in the form of Mo-octoate precursor. .................. 185 Figure  7.1. TEM micrograph of coke samples recovered from different hydroconversion experiments. (A) 1800 ppm Mo using Mo-octoate precursor, (B) 600 ppm Mo using fresh Mo-micelle precursor, (C) 1st recycle of 600 ppm Mo using Mo-micelle precursor, (D) 2nd recycle of 600 ppm Mo using Mo-micelle precursor, (E) 3rd recycle of 600 ppm Mo using Mo-micelle precursor. T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. ......................................................... 193 Figure  7.2. H/C atom ratio change of the coke samples recovered from different steps of recycle experiments in several series of recycle experiments using different Mo concentrations and precursors. T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. ............................................................................................ 196  xxviii Figure  7.3 N content change of coke sample during different steps of recycling using 1800 ppm Mo using Mo-octoate precursor (A) and 600 ppm Mo using Mo-micelle precursor (B). ........................................................................................................ 197 Figure  7.4. 13C NMR analysis and change in the peak integral intensity ratio of aromatic to aliphatic carbons (CAr/CAl) for coke samples recovered from different recycling steps in hydroconversion experiments using 600 ppm Mo in the form of Mo-micelle precursor (A) and 1800 ppm Mo using Mo-octoate precursor (B). ......... 206 Figure  7.5. X-ray diffractogram of coke sample recovered from the hydroconversion experiment using 1800 ppm Mo in the form of Mo-micelle. ............................... 212 Figure  7.6. Coke yields of hydroconversion experiments in which fresh Mo using Mo-micelle (A), recycled coke without thermal aging (B) and thermally-aged recycled coke (C) were used as the catalyst. ...................................................................................... 213 Figure  7.7. Schematic of catalyst and coke interaction as the number of recycles increases (A ? E). Highly hydrogenated and amorphous coke (gray area), heavily carbonaceous and graphitic (refractory) coke, i.e. hard coke, (black area) and catalyst particles (mostly single layers of MoS2, small white spots) are shown in a sample coke particle in the picture. ........................................................................................... 217 Figure  8.1 Proposed reaction mechanism for the residue conversion .................................. 230 Figure  8.2 Differential equations for the rate of consumption of different lumps. ............... 231  xxix Figure  8.3 Parity plots of moles of gas, naphtha, LGO, HGO, residue, coke and all the lumps. .............................................................................................................................. 235 Figure  8.4 Fit of the model for all the kinetic lumps at all 4 reaction temperatures of 415 ?C (A), 430 ?C (B), 445 ?C (C) and 460 ?C (D). The lines represent the model calculated values and the circles represent the experimental values. ................... 237 Figure  8.5 Arrhenius plot for different reactions. ................................................................. 238 Figure  A.1 Schematic of a delayed coker. ............................................................................ 265 Figure A.2 Schematic of a fluid coker???????????????..???.??267 Figure A.3 Schematic of a flexi coking process ????????????????...268 Figure  B.1 Schematic (LHS) and geometry (RHS) of batch reactor body. (Mechanical drawing by Autoclave Engineers). ........................................................................ 269 Figure  B.2 Mixing pattern of batch reactor equipped with straight blade mixer. (Mechanical drawing by Autoclave Engineers) ......................................................................... 269 Figure  D.1 Summary of product work-up in the batch reactor experiments. ....................... 272 Figure  E.1 Summary of product work-up in the semi-batch reactor experiments. ............... 273 Figure  I.1 A sample temperature and pressure profile of a typical hydroconversion experiment. ........................................................................................................... 304 Figure  K.1 Relative H/C atom ratio and molecular weight of refinery feedstocks. (Reproduced from Figure 14.18 of [1], Copyright ? 2006, Taylor & Francis Group LLC., with permission) ......................................................................................... 310   xxx Nomenclature AC  Activated carbon AHM  Ammonium heptamolybdate API  American petroleum institute AR  Atmospheric residue ASTM  American society for testing and materials bbl/day Barrel(s) per day BPR  Back-pressure regulator BET  Brunauer-Emmett-Teller BJH   Barrett-Joyner-Halenda CAr  Aromatic carbon CAl  Aliphatic carbon CCR  Conradson carbon residue CLVR  Cold Lake vacuum residue CSTR  Continuous stirred-tank reactor DAO  Deasphalted oil  xxxi df  Degree of freedom DLS  Dynamic light scattering DRIFT  Diffusice-reflective infrared Fourier transform EDX  Energy-dispersive X-ray spectroscopy EST  Eni slurry technology FCC  Fluid catalytic cracking FID  Flame ionization detector HCR  Hydroconversion HDN  Hydrodenitrogenation HDS  Hydrodesulfurization HGO  Heavy gas oil HRTEM High resolution transmission electron microscopy HTR  Hydrotrating HTSD  High temperature simulated distillation HYD  Hydrogenation LLKC   LaMarca-Libanti-Klein-Cronauer MEABP Mean average boiling point  xxxii mL  Milliliter MS  Relevant mean-square value NMR  Nuclear magnetic resonance ODE  Ordinary differential equation PAH  Polynuclear aromatic hydrocarbon PE4LE  polyoxyethylene-4-lauryl-ether PHP   Potassium hydrogen phthalate ppm  Part per million sccm  Standard cubic centimeter per minute SDA  Solvent deasphalting SEM  Scanning electron microscopy SIMDIS Simulated distillation SS  Sum of squared deviates STP  Standard temperature and pressure conditions TCD  Thermal conductivity detector TEM  Transmission electron microscopy TGA  Thermogravimetric analysis  xxxiii TIOR  Toluene insoluble organic residue t/d  ton(s) per day VABP  Volume average boiling point VGO  Vacuum gas oil Vol%  Volume percent VR  Vacuum residue Wt %  Weight percent XPS  X-ray photoelectron microscopy XRD  X-ray diffraction oC  Degree Celcius         xxxiv Acknowledgements I would like to express my deepest gratitude to my PhD supervisor, Professor Kevin. J. Smith for his scientific generosity and moral support throughout this PhD research. His time dedication, constructive guidance and patience were always the source of peace of mind for me. It has been an honor for me to work with Professor Smith and I always appreciate my learnings under his supervision throughout the course of this PhD study. I would like to sincerely thank Professor A. Paul Watkinson and Dr. Keng Chou for their time and being a part of examination committee and for their valuable advice regarding my PhD research. My deep appreciation goes to the Department of the Chemical abd Biological Engineering (CHBE) at UBC for giving me the opportunity to be a part of this unique department. I would like to thank all the great CHBE staff, Helsa Leung, Lori Tanaka, Amber Lee, Joan Dean, Alex Thng, Doug Yuen and all the workshop staff for their help and support. I would like to especially thank Mr. Richard Ryoo, store manager of CHBE, for all of his support and patience during my PhD. I would like to thank my good friend, Dr. Babak Derakhshandeh and all my group members (Dr. Xuebin Liu, Dr. Zhiming Fan, Dr. Sharif Fakhruz Zaman, Dr. Shahrzad Jooya Ardakani, Dr Rui Wang, Dr. Farnaz Sotoodeh, Dr. Shahin Goodarznia, Vickie Whiffen, Ross Kukard, Pooneh Ghasvareh, Mina Alyani, Ali Alzaid and Rahman Gholami Shahrestani) who helped me in the course of my PhD study.  xxxv I would like to express my deepest appreciation for my parents who have always been my moral support throughout my long academic journey. It was surely not possible without their support. And at last but not the least, I would like to express my sincere gratitude to my love and joy of life, my beautiful wife, Fahimeh. Her understanding, kindness, support and encouragement have always been heart warming for me.              xxxvi Dedication    To my loving wife and parents    xxxvii Chapter 1  Introduction 1.1 Background In recent decades, the price of conventional oils has increased significantly (by more than 300% since 2000). The increase is mainly due to political instabilities in the regions of the world where sweet and conventional crudes are still available [1]. In industrial countries, the abundant reserves of heavy crudes, both onshore and offshore, as well as oilsands, are being considered as replacements for conventional crude. In North and Latin America, several mega projects for the utilization of these resources are now under construction, while others are in various phases of preparation [2]. The demand for high-value petroleum products such as middle distillate and gasoline is increasing, while the demand for low value products such as residue based products and fuel is decreasing. These changes in demand for fossil fuels makes maximizing the liquid product yield from different processes and higher conversion levels of residue oil, an immediate concern to refiners. In addition to the changes in the demand for fossil fuels, increased environmental concerns have resulted in more rigorous specifications for petroleum products, including fuel oils. These restrictions show the need for utilizing and improving processes which convert the residue fraction of crude oil into lighter and more valuable products [1,3]. Nowadays, most of the heavy oil and extra-heavy oil in North America is imported into the United States from Canada, Mexico, and Venezuela. This is particularly true for Alberta, 1  Canada, which is connected by pipelines to many refineries in the United States. The oil sands of Canada consist mostly of bitumen, sand and water. The bitumen, a viscous mix of polyaromatic hydrocarbons, with an APIo <10 and high sulfur level (4-5 wt %), is separated from the sand and needs to be upgraded to a synthetic crude oil before it can be refined into commercial products such as gasoline and diesel [4].  The goal of residue oil upgrading is to increase the H/C atom ratio and the API? gravity of the product oil, while also producing lower boiling point distillates with higher commercial value than the residue oil feedstock [5]. Two categories of residue oil upgrading processes have been developed to achieve this goal: 1) carbon rejection (or coking) processes that decrease C content of the feed by producing coke, which has a very low H/C atom ratio, and 2) hydrogen addition (or hydroconversion) processes that increase H content of the feed by means of hydroconverion/hydrogenolysis reactions that reduce coke yield in favor of liquid products. An essential difference between coking (thermal decomposition in the absence of H2 and catalyst) and hydroconversion (mainly hydrogenolysis reactions in the presence of H2 and catalyst) is that in coking the production of light products is always associated with the production of polymerized heavier products, such as coke. The polymerization reactions may be partially or even entirely prevented during hydroconversion, which usually results in increased distillate yield [1,6]. At present, coking is the most widely practiced commercial upgrading technology for heavy feeds since it is well proven, reliable, and a relatively low cost technology. However, a significant fraction (up to 40 wt % depending on the feed properties) of the feed is converted 2  to gas and coke at the expense of potentially more valuable liquid products, and the liquid products are of poor quality [5,7]. As well, coking produces large quantities of high-sulfur, heteroatom-rich coke, creating a significant impact on the environment [8]. The production of a large amount of coke has a negative impact on coking processes, unless the generated coke can be used. Petroleum coke can be used in the anodes used for the production of aluminum, and a variety of carbon or graphite products, such as brushes for electrical equipment. These applications, however, require the coke to be low in mineral matter and sulfur, which dictates further processing and purification of the coke generated in coking processes [5]. The presence of high pressure H2 and a hydrogenation catalyst benefit processes converting residue oils. The hydrogen suppresses free radical addition reactions and dehydrogenation. The hydrogen removes and converts heteroaromatic species to hydrogen sulfide, water, and ammonia. All hydroconversion processes involve a complex set of reactions that occur in series and parallel, including cracking, hydrogenation, sulfur removal and demetallization. Hydroconversion processes use a catalyst or additive to control coke formation. The catalyst also serves as a surface for deposition of metals and enhances hydrogenation reactions. Due to the complexity of the reactions, directing the hydroconversion reactions toward desirable product characteristics is a challenge. The initial focus for residue hydroconversion was the removal of sulfur, but many recent processes have concentrated on the objective of achieving high conversion of the residue fraction of the feed. The emphasis on sulfur removal, in these process designs, shifts from the primary hydroconversion step to a secondary hydroprocessing reactor [5]. 3  In the past two decades, considerable attention has been paid to hydrogen addition technologies for residue upgrading. At present, most commercial hydroconverion processes are conventional fixed-bed technologies. These processes are limited by the choice of feedstock and the severity of operation, because carbonaceous and metal deposits decrease the activity of the catalysts, which are usually sulfides of transition metals like Mo, Ni, Co, W, Cr, V, Fe, Cu and Zn [8]. Ebullating-bed hydroconverion technology can process heavy feeds at high severity, and the catalyst can be added and withdrawn without shutting down the reactor [9]. Despite the advantages of this technology, the reactors are mechanically complex and use large amounts of expensive catalysts, the disposal of which is complicated by environmental and economic concerns. The problems facing existing technology in coking and hydroconverion processes have led to the development of slurry-phase hydroconverion technology that uses highly dispersed, unsupported catalysts. The use of dispersed catalysts offers several advantages. In particular, since the catalysts are used once-through, deactivation is reduced when compared to supported catalysts. Dispersed catalysts can be introduced into the feed as finely divided inorganic powders, water-soluble or oil-soluble salts. The presence of a well dispersed catalyst favours the rapid up-take of H2, preventing coke formation. The high activity of the dispersed catalyst increases the conversion of feed into light products while the coke yield and the asphaltene content of the liquid product is lower compared to fixed-bed processes. Many aspects of dispersed catalyst hydroconverion processes have been explored. The development of slurry-phase catalysts, their hydroconversion mechanisms and the processes that are close to commercialization have been reviewed by Del Bianco et al. [10,11]. A 4  number of slurry-phase processes have been developed and tested in bench-scale or pilot-scale process units in the last two decades. However, none has yet been commercialized because of the high cost of catalyst when used once-through. One approach to resolve this issue is to recover and recycle the used catalyst, thereby decreasing the catalyst consumption and cost. As discussed earlier, the drawback of the methods employing dispersed/dissolved catalysts is the uncertainty regarding the catalyst recovery for reuse. For example, a plant processing about 75,000 bbl/d (~12,000 t/d) of a heavy feed requiring the continuous addition of ~1000 ppm of metal catalyst, would require the addition of about 12 tons per day of metal. Therefore, it is crucial that most of the metal be recovered for reuse. In practice, dispersed catalysts end up in the residue fraction of the products. In an ideal case, part of the vacuum residue (VR) including the catalyst metal, may be recycled and blended with the feed. However, this would possibly require conditions ensuring reactivation of the catalytically active metal phases at the entrance of the reaction zone unless the recycled catalyst still possesses adequate activity. For example, a sufficiently high H2 pressure would ensure that the coke deposition on the catalyst particles would be low. In other words, most of the original activity of the catalyst would be retained [2]. A significant number of studies have reported on the effect of catalyst precursor type, catalyst loading, H-donor compounds in the feed and operating conditions on the hydroconversion reactions using fresh precursors in slurry-phase systems [12-24]. On the other hand, very few studies are available that address hydroconversion using recycled (both 5  supported and dispersed) catalysts. To address this apparent knowledge gap within this field, more detailed studies on the activity of recycled catalysts in slurry-phase hydroconversion processes is required. Furthermore, a detailed study of the catalyst deactivation mechanism in slurry-phase processes is needed so that a better understanding of the catalyst-coke interaction in slurry-phase processes can be developed. Results from such studies may ultimately lead to modifications to existing, near-commercial slurry-phase hydroconversion processes to make these processes economically viable. 1.2 Objectives of the Thesis The objectives of the present thesis are: 1) To investigate the catalytic activity of recycled dispersed catalysts for the hydroconversion of residue oil under high residue conversion conditions. 2) To study the potential deactivation of recycled catalyst in slurry-phase hydroconversion, and propose a conceptual deactivation mechanism for the catalyst.  1.3 Approach of the Thesis The approach to achieve the objectives of the thesis combined experimental studies to assess the activity of different catalyst precursors (both fresh and recycled) in the hydroconversion of a selected residue oil (CLVR), with careful and detailed characterization of the coke-catalyst mixture recovered after the hydroconversion experiments. 6  In the first step (catalyst screening, discussed in Chapter 4), different catalyst types and precursors were selected and tested under relatively mild hydroconversion experiments. These experiments, which were carried out in a batch reactor, were done under mild conditions to allow for better observation of differences between the different catalyst precursor activities. In catalyst screening, laboratory-synthesized mono-metallic and bi-metallic catalysts prepared in reversed micelles were compared to finely-powdered iron sulphate and water-soluble ammonium heptamolybdate (AHM). Chapter 5 compares the activity of more active catalysts in the batch and semi-batch reactor. Selected catalysts with high activity in terms of coke suppression were investigated further in a semi-batch hydroconversion reactor under more severe residue conversion conditions. In the semi-batch reactor, high pressure (13.8 MPa) and continuous flow of H2 gas (900 mL(STP)/min) through the reactor eliminated the possible effect of H2 starvation observed in the batch reactor. Also, utilizing a higher reaction temperature compared to the batch reactor (445 ?C in the semi-batch reactor vs. 415 ?C in the batch reactor experiments), catalyst precursors were studied under hydroconversion conditions that closely simulated industrial conditions. Chapter 6 focuses on the activity of recycled catalyst in the semi-batch reactor under high (> 80 wt %) residue conversion conditions. The most active catalyst precursor in terms of coke suppression under high residue conversion conditions identified in the semi-batch experiments (mono-metallic Mo-micelle precursor) was further investigated to assess the catalyst activity when recycled. The activity of fresh and recycled MoS2 catalyst, originating 7  from the Mo prepared in the reversed micelle (Mo-micelle) was compared to the activity of molybdenum octoate, a commercial oil-soluble Mo precursor. Chapter 7 presents a detailed study on the properties of the recycled coke-catalyst mixture. Deactivation of the recycled catalyst was monitored and studied by detailed analysis and characterization of the recycled coke-catalyst mixture at each recycle step. Characterization results were used to develop and propose a deactivation mechanism based on the evolution in properties of the coke-catalyst mixture observed after each recycle experiment. In Chapter 8, kinetic modeling of residue hydroconversion in the semi-batch reactor was completed, using a proposed kinetic network. The calculated values of the kinetic model are fitted and compared to the experimental values measured in a series of hydroconversion experiments. Using MATLAB programming and a parameter estimation algorithm, rate constants and activation energies of different reactions in the proposed mechanism were calculated. A statistical analysis of the calculated values and estimated parameters is also presented.       8  Chapter 2  Literature Review 2.1 Petroleum, Heavy Oil, Residue and Synthetic Crude Oil The terms crude oil and petroleum are used to describe a mixture of hydrocarbons and other compounds which resemble a wide variety of liquid phase fuels. Crude oils are widely different in their chemical (variable amounts of S, N, and O content) and physical (volatility, specific gravity, and viscosity) properties, which are mostly determined by the origin of the oil. Crude oils usually contain V and Ni metals (100?s of ppm), constituents of which are usually present in more viscous crude oils. The presence of these non-organic elements can cause serious difficulties in refineries in which these feedstocks are processed [6]. Due to an extremely wide variation in the properties of residue, the proportions in which the different constituents occur vary with origin. This variation is much less among the conventional crude oils compared to heavy-oils and residues (Table  2.1). Some crude oils consist of a larger fraction of the lower boiling point components and others (such as heavy oil and bitumen) consist of larger fractions of higher boiling point components (asphaltic components and residuum) [6]. Residue, or residuum, is operationally defined as the fraction of petroleum, heavy oil, or bitumen that does not distill under vacuum. Residues have atmospheric equivalent boiling 9  points over 525? C, and in a refinery they are produced as the bottom product from a vacuum distillation column [5]. Table  2.1 Typical properties of different petroleums [6]. (Copyright ? 2006, Taylor & Francis Group LLC., Reproduced with permission) Petroleum Specific Gravity API Gravity Residuum > 538 ?C (% v/v) U.S. Domestic    California 0.858 33.4 23.0 Oklahoma 0.816 41.9 20.0 Pennsylvania 0.800 45.4 2.0 Texas 0.827 39.6 15.0 Texas 0.864 32.3 27.9 Other Countries    Bahrain 0.861 32.8 26.4 Iran 0.836 37.8 20.8 Iraq 0.844 36.2 23.8 Kuwait 0.860 33.0 31.9 Saudi Arabia 0.840 37.0 27.5 Venezuela 0.850 17.4 33.6  While conventional crude oils have a residue content of 10 ? 30 wt %, heavy oils have a residue content > 40 wt %, which makes it necessary to upgrade these types of oils so that 10  they can be refined in conventional refineries [5]. Properties of light crude oil and heavy crude oil are compared in Table  2.2. Heavy oil derived from different sources can be upgraded into a marketable and transportable product. These products, although varied in nature, are a mixture of hydrocarbons that resemble conventional crude oil and hence are termed synthetic crude oil [6]. The synthetic crude oil that is produced from upgrading processes can be refined in conventional refinery processes. 2.2 Definition of Primary and Secondary Upgrading Briefly, primary upgrading of heavy-oil and residue refers to processes that convert high molecular weight compounds to distillate [5]. The overall objectives of primary upgrading can be categorized as: 1) Conversion of high boiling point compounds (mainly the residue fraction with boiling point higher than 524 ?C) into distillate (i.e. compounds with boling points below 524 ?C). 2) Boosting the H/C ratio of distillate products to an acceptable level for transportation fuels. To achieve these goals, the heavy-oil and residue are required to undergo a number of different reactions, some thermal and some catalytic. While primary upgrading mainly targets the breakage of the macromolecules in the residue that have high boiling points to yield distillate, secondary upgrading (or 11  hydrotreating) processes are designed to selectively remove heteroatoms from the feed to an acceptable level [5]. Among different heteroatoms present in the distillate, S and N are of most interest for transportation fuel applications. Although primary and secondary upgrading share a common underlying chemistry, the process designs for these processes are very different. The main differences in primary and secondary upgrading are in the severity of the process (mainly temperature and reaction time) and selectivity toward different reactions. Under severe primary upgrading conditions, thermal C-C bond cleavage predominates over catalytic reactions (mainly hydrogenation), whereas at the mild conditions of secondary upgrading (hydrotreatring), catalytic reactions can be directed toward the selective reactions of interest [5].    12  Table  2.2 Comparison of light and heavy crudes [5]. (Copyright ? 1994, Taylor & Francis Group LLC., Reproduced with permission)  Light crude Cold Lake Athabasca Morichal Heavy Arabian Hono Maya Gach Saran API Gravity 38 10 9 4.1 12.6 13.4 9.4 15.6 Sulfur, wt % 0.5 4.4 4.9 4.9 4.2 5.1 4.4 2.6 Nitrogen, wt % 0.1 0.4 0.5 0.8 0.26 0.7 0.5 0.4 Metals, wppm1 22 220 280 863 115 372 496 144 Viscosity, m2/s ? 106 at 40 ?C 5 5000 7000 N/A N/A N/A N/A N/A Vacuum Resid (525 ?C+)  Liquid, Vol % 11 52 52 80 51 46 59 50 1- wppm: part per million by weight    13  2.3 Heavy-oil and Residue Primary Upgrading Processes 2.3.1 Thermal Cracking (Carbon Rejection) Processes Thermal cracking is one of the earliest processes used for petroleum conversion. In thermal cracking, high boiling point compounds are converted into lower boiling point compounds. The process was developed in the early 1900s for gasoline production from the low-value and higher-boiling products of the distillation process. Later on, it was found that the carbon rejection process also produced a wide range of products. These products varied from very volatile gases to high-boiling, nonvolatile petroleum coke. The heavy-oils produced by carbon rejection process include light gas oils, heavy gas oils, as well as a residual oil that can be used as fuel oil [6]. Further details of different thermal cracking processes can be found in  Appendix A. 2.3.2 Catalytic Hydroconversion (Hydrogen Addition) Processes At present, coking is the most widely used upgrading process for heavy crude oils since it is a well proven, reliable, and relatively low cost technology. An essential difference between coking (thermal decomposition in the absence of H2 and catalyst) and hydroconversion (mainly hydrogenolysis reactions in the presence of H2 and catalyst) is that in coking, the production of light products is always associated with the production of polymerized heavier products, such as coke. The polymerization reactions may be partially or even entirely prevented during catalytic hydroconversion, which usually results in increased distillate yield [6]. 14  Hydroconversion facilities add flexibility of handling different feedstocks in refinery processing. This is mainly because hydroconversion makes having a lower molecular weight product with higher H content and a lower yield of coke in the refining processes feasible. Hydroconversion utilizes more severe conditions than hydrotreating (higher temperature and H2 pressure). The severe conditions allow the refiners to convert the feedstock to lower-boiling point products rather than to treat the feedstock for heteroatoms and metals removal only [2]. Hydrogen addition processes are operated at temperatures ranging from 400 to 460 oC and H2 pressures > 7-14 MPa. In primary upgrading by hydrogen addition, the thermal mode of cracking dominates the process. Therefore, it is often termed hydrocracking. Hydroconversion processes are normally carried out in the presence of a catalyst with the goal to mitigate coke formation as much as possible. This maximizes residue conversion to products lighter than coke. Higher liquid yields are obtained from hydroconversion processes, approximately 85 ? 95 wt % compared to about 60 - 70 wt % for coking. Further, better liquid product quality is obtained, with lower CCR and heteroatom content compared to coking. Fixed-bed and ebullated-bed reactors are two common commercialized hydroconversion technologies used in upgrading heavy oil. Slurry-phase reactor technology is a relatively modern process in this category, which is still under development [5,25]. Based on the type of reactor bed employed, commercial catalytic reactors using a granular catalyst can be divided into the three main categories: (1) Fixed-bed processes; (2) Moving-bed processes; (3) Ebullated-bed processes [26-28]. Fixed-bed reactors were 15  originally used for hydroprocessing of light feedstocks. Later, fixed-bed reactors were modified to be able to process heavier feedstocks. Although fixed-bed reactors can operate reliably on atmospheric residues, processing residue oil cannot be easily achieved in these systems. A major concern in fixed-bed processes is the excessive number of catalyst replacements required due to catalyst fouling and deactivation. In contrast, the moving- and ebullated-bed units have demonstrated reliable operations with vacuum residues. In fixed-bed and ebullated-bed processes, the high cost of the hydroprocessing catalyst because of catalyst replacement and disposal, has led to the development of slurry-phase processes. The slurry-phase process has captured a lot of interest in the last two decades. This process employs unsupported catalysts, which are added to the feed oil in different forms [2,6,28,29]. Based on the discussion above, one can conclude that the choice between different available processes for residue and heavy-oil upgrading is dependant on several criteria. Process operation complexity, catalyst cost, process stability and potential of a process to be able to process a specific feedstock, are some of items that refiners need to consider before selecting a process for hydroprocessing of heavy feedstocks [30]. In the following sections, a brief discussion of the hydrogen addition processes, with an example of each that is at commercial or near-commercial scale operation, is presented. 2.3.2.1 Fixed-bed Processes As discussed earlier (Section  2.3.2), fixed-bed processes were originally developed to process feeds containing no metals or asphaltenes. This type of reactor was later adapted to process heavier feedstocks. Fixed-bed reactors showed that they can easily process 16  atmospheric distillates as well as vacuum gas oil (VGO) and heavy gas oil (HGO). Both flow patterns of upflow and downflow can be used in fixed-bed reactors. The latter, so-called trickle-bed mode, is the most common form of flow pattern used in fixed-bed reactors [31]. In fixed-bed reactors, a single, stationary layer of catalyst of the same catalyst size and shape or layers of different catalysts can be used. Different layers may vary in shape, size and/or chemical composition [29,31]. Fixed-bed residue hydroprocessing units are usually limited to processing residues containing less than 250 ppm metals (Ni + V), in order to achieve a reasonable life for the catalyst before it must be replaced. Longer operating times require replacing the catalyst on-line [2]. Two processes which are available for hydroprocessing of feeds with higher metal content are: 1) Chevron?s OCR Process (on-line catalyst replacement), which allows processing of feeds with up to 400 ppm metal (Ni + V) and 2) Bunker reactor technology (moving-bed hydroprocessing), developed by Shell, which allows the catalyst to be removed from the bottom of the reactor and fresh catalyst added at the top of the bed. The bed of catalyst gradually moves down through the reactor [26]. A schematic of the bunker reactor technology is shown in Figure  2.1. The Unibon, atmospheric residue desulfurization (ARDS) and HYVAHL heavy-oil hydroprocessing processes are based on fixed-bed and moving-bed reactor technologies, details of which can be found elsewhere [2,26,31].  17   Figure  2.1 Schematic of bunker reactor technology developed by Shell [26]. (Copyright ? 1998, Elsevier, with permission)  2.3.2.2 Ebullated-bed Processes Metal catalysts supported on alumina (active metals of Co/Mo or Ni/Mo sulfides) can be used to process feeds that contain solids if the catalyst is fluidized. This prevents bed plugging by fine solids and fresh catalyst, while the latter is added regularly to maintain constant activity. In an ebullated-bed reactor, the catalyst is fluidized and remains suspended by the liquid phase in the reactor. The catalyst pellet diameter is usually < 1 mm. The ebullated-bed reactor is suitable for processing heavy oil at severe conditions (high temperature of > 450 ?C). By utilizing ebullated-bed technology, no catalyst bed plugging 18  problems are encountered and the liquid-recycle promotes good mixing within the reactor. H-Oil and LC-Fining are two licensed processes based on ebullated-bed technology [5]. H-Oil and LC-fining have very similar reactor configuration. In both processes, the H2 gas and residue feed enter the reactor at the bottom and flow upward [31]. These processes are often operated commercially at about 85% residue conversion [6]. Since the two processes have very similar configuration in reactor design, they are not discussed separately. A brief review of the H-Oil process is presented below. The H-Oil process uses a single-stage, two-stage or three-stage ebullated bed reactor. The process, which is designed to upgrade heavy-oil and residue feedstocks, can produce distillate products as well as desulfurized and demetallized residue as feed for coking processes. The feedstock is mixed with the recycle residue recovered from fractionation of products and is mixed with H2 gas flow. The mixture is fed into the bottom of the reactor, and the upward flow of the gas-liquid mixture expands the catalyst bed [6]. The expanded bed is maintained about 35 % above the settled level of catalyst in the reactor. This is achieved by controlling the gas-liquid mixture flowrate to the reactor [31]. The upgrading products as well as unreacted H2 gas are cooled by a heat exchanger, and the vapor is separated from the liquid. Unreacted H2, after scrubbing in an oil absorber, is recycled to the reactor inlet [6]. A schematic of the H-Oil process is shown in Figure  2.2.  19   Figure  2.2 A schematic of H-Oil prcess [6]. (Copyright ? 2006, Taylor & Francis Group LLC., with permission)  2.3.2.3 Slurry-phase Processes The problems facing existing technology in coking and hydroconversion processes (fixed-bed and ebullated-bed) have led to the development of slurry-phase hydroconversion technology that uses highly dispersed, unsupported catalysts. Since the dispersed catalyst is used once-through (in case of low cost catalytic materials), catalyst deactivation is much less of a concern compared to the processes in which supported catalysts are used. [5,6,32]. Dispersed catalysts can be introduced to the feed as finely divided inorganic powders, water-soluble and oil-soluble salts. The presence of a well dispersed catalyst favours the rapid up-take of H2, preventing coke formation. The high activity of the dispersed catalyst 20  increases the conversion of feed into light products. The coke amount and asphaltene content are lower in the products of slurry-phase processes compared to fixed-bed processes [2]. The development of slurry phase catalysts, their hydroconversion mechanisms and the main processes close to commercialization, have been reviewed [1,9-11,33,34] and several slurry-phase hydroconversion processes aimed at residue oil upgrading have been developed over the last few decades as summarized in Table  2.3. Examples include the M-coke process developed by Exxon in 1981 [35], the CANMET process [36-40], the EST process (Eni Slurry Technology) [41-44], VCC (VEBA Combi-Cracking) [45-47], SOC (Super Oil Cracking) [48], Intevep HDH Process [49,50] and the (HC)3 process [51,52].  21  Table  2.3 Different slurry-phase processes developed in the last few decades. a: Values vere converted assuming a feed density of 950 kg/m3.  Process Company, Country Catalyst Operating conditions Residue Conversion (wt %)  Reference T (?C) P (MPa) H2 consumption (Nm3/ton of feed) LHSV (h-1) HFC Nation Research Institute for Pollution and Resource, Japan Spent VHDS catalyst 450 3~10 50~177 0.35~ 0.91 65~75 [53] VCC Veba Oel, Germany Red-muds (Iron-based, Fe2O3) 430 ~ 490 15~25 380~414 -- 80~95 [46,47] CANMET Petro-Canada & Energy, Mine and Resources, Canada Iron sulphate (FeSO4) Pulverized coal 435 ~ 470 15 222~277 0.5~1 >90 [36,37,54] SOC (Super Oil Cracking) Asahi Chemical Industrial Co., Ltd & Chiyoda Chemical Engineering and Construction, Japan Dispersed catalyst 470 ~ 490 20~22 210~1263a -- 87~90 [48] HDH INTEVEP, Venezuela Pulverized ore (Iron based) -- 13 260 0.47 92 [49,50] TSSBH University of Petroleum, China Mo-dispersed catalyst 430 ~ 450 12 -- 1.0 90 [8,55,56] M-coke Exxon, USA Dispersed Mo 438 17.2 -- -- 90 [35] (HC)3 Hydrocarbon Conversion Technologies, Inc, USA Oil-soluble (Dispersed Fe & Mo) 440 ~ 490 17-25 173~276a 0.4-1.4 (L/h) 95 [51,57] EST ENI Technologies, Italy Mo-based 400 ~ 425 16 -- -- >98 [42] -- Kobe steel Ltd, Japan & Syncrude Canada Limonite ore Iron based 440 ~ 450 10 159a 1 -- [20] 22  The difference between these processes is mostly in the type of catalyst used. Numerous studies have reported on the performance of the catalysts, both in terms of the active metal and in terms of the form of the catalyst added to the slurry-phase reactor [8,12-16,20,22-24,56,58-66]. Although many studies have been completed on the development of slurry-phase processes, the problems associated with the development of the slurry processes are not completely solved, and none are presently operated on a commercial scale. Two of the slurry-phase processes, developed during the last 3 decades and relevant to the present study are discussed briefly below. 2.3.2.3.1 Exxon M-coke Process In 1981, Bearden and Aldridge [35] from Exxon Research and Development Laboratory reported a new concept in heavy feed upgrading based on catalytic control of free-radical reactions. In this process, the catalyst is formed in situ in heavy feeds from oil soluble or dispersible metal compounds under hydroconversion conditions, and is made up of micron sized particles, which contain a metal component and a carbonaceous component. These catalysts, which are called M-coke (micrometallic coke), are effective for control of coke forming reactions at quite low concentration, as little as 0.1g/kg metal based on oil feed suffices. Aqueous phosphomolybdic acid, molybdenum naphthenate and a carbonyl compound are examples of M-coke precursors which are thermally decomposed and sulfided into MoS2 catalyst in situ. Moreover, throughout the reaction, greater than 90% of Ni and V contaminants in the feed, were transformed into oil insoluble components, likely metal sulfides. These compounds, which deposit on the M-coke catalyst, may act as supplementary 23  catalyst materials. As well, the micron-sized highly dispersed M-coke particles serve as nucleation sites in this process; hence, a small amount of coke is formed in the hydroconversion reaction, and this helps to prevent the fouling of reactor surfaces. Three months of experimental results from a small continuous pilot plant in once-through operation were also reported by these authors [67]. The heavy feeds that were tested included Athabasca oilsand bitumen, Cold Lake crude and a light Arabian vacuum residuum. Molybdenum M-coke (MoS2) was used as the catalyst in all the tests. Process conditions included temperatures from 400 ~ 443?C, and pressure between 6.9 ~ 17.2 MPa. The process provided a 90 % conversion of 566?C+ fraction and substantial removal of heteroatom impurities. According to the authors, there is no feedstock limitation with respect to asphaltene, metals, or the presence of particulate matter such as clay. Although Exxon has continued to investigate this technology [67], no scale up of the plant has been reported. 2.3.2.3.2 Eni Slurry Technology (EST) Eni Slurry Technology (EST) is based on the slurry hydroconversion of the feedstock (heavy residues or oil sands bitumen) at a relatively low temperature in the presence of H2 and high concentrations (3,000 ppm) of molybdenum-based dispersed catalyst. The reaction product is distilled to recover the lighter fractions, while the residue is sent to a solvent deasphalting (SDA) unit to separate the desaphalted oil (DAO) that may represent an interesting feedstock for cracking processes such as FCC or hydrocarcking. The asphaltenes from SDA, containing most of the dispersed catalyst, are recovered, mixed with fresh feedstock and reprocessed in the slurry hydrotreatment reactor. The technical feasibility of 24  this configuration has been demonstrated by operating a continuous pilot plant (2.5 kg/h of liquid feed). The slurry reactor is operated under relatively mild conditions (400 - 425?C and 16 MPa). Very high conversion of CCR (> 90 %) and good product upgrading were obtained prolonging the run for several weeks. As well, the EST process exhibits excellent feedstock flexibility and product quality. Furthermore, electron microscopy analyses confirmed that the morphology of the catalyst was unchanged at the end of the pilot run. The MoS2 particles appeared to be well dispersed, and crystallites showed a few layers of stacking, thus demonstrating that no significant modifications of the catalyst occurred after prolonged use [42].   In a recent report [68] on the improvements made to the EST process, Bellussi et al. showed that by coupling the nano-dispersed MoS2 with a cracking catalyst (NiMo/?Al2O3), the MoS2 protected the cracking catalyst from poisoning caused by metals and coke deposition. The authors compared the hydroconversion of vacuum residue (boiling point > 550 ?C) in an ebullated-bed reactor and a CSTR. The reaction temperature of 430 ?C and H2 pressure of 160 bar were selected as the reaction conditions. In both reactors, the unreacted residue was recycled to the reactor for further cracking. Comparing experiments using NiMo/?Al2O3 catalyst with an experiment in which NiMo/?Al2O3 was coupled with nano-size MoS2 catalyst, the authors found that metal and coke deposition on the supported catalyst (NiMo/?Al2O3) decreased significantly. 25  2.4 Catalysts Used for Hydroconversion In general, catalysts used for hydroprocessing (hydroconversion and hydrotreating) can be categorized as follows: 1) conventional transition-metal sulfide catalysts (Mo, Ni, Co, W, Cr, V, Fe, Cu and Zn) supported on Al2O3 or SiO2, which are widely used in commercial fixed-bed and ebullated-bed hydroprocessing processes and 2) non-conventional hydroprocessing catalysts that consist of soluble catalysts and finely dispersed catalysts, which are used in slurry-phase reaction systems [2,6,69]. A brief review of these catalysts is presented below. 2.4.1 Conventional Catalysts Used in Hydroprocessing The conventional catalysts used for residue hydroprocessing are mostly sulfides of Co, Ni, W and Mo, supported on Al2O3 or SiO2, with a range of pore structures and active metal dispersions [1]. Although other forms of metals (such as carbides and nitrides) also have noticeable activity in activating H2 and carrying out hydrogenation reactions, their activity is not as good as sulfided catalysts in catalyzing hydrogenation reactions [2,70]. Generally, hydroprocessing catalysts are made of two parts: 1) Hydrogenation catalysts used for hydrotreating reactions (metal sulfides) and 2) Hydroconversion catalysts used for facilitating cracking reactions (catalysts with acidic active sites) [2]. Hydrotreating catalysts are used to remove heteroatoms (mainly S and N) since these compounds have a deleterious effect on acidic hydroconversion catalysts. The hydrotreating 26  catalysts also hydrogenate aromatics, which cannot be cracked directly by the hydroconversion catalysts or by thermal reactions during hydroprocessing. The hydrogenation function is provided by metal crystallites. The hydrogenation activity of different metals decreases in the following order on metallic catalysts: Noble metal > Sulfided non-noble metal > Sulfided noble metal where noble metals are the platinum group metals (Pt, Pd, Ir and Os) and the non-noble metals are transition metals such as Mo, Co and Ni. Noble metal catalysts are commonly used in low-sulfur or sulfur-free environments because they are poisoned by even low levels of S (several tens of ppm). Non-noble metals are usually used for industrial feedstocks because they are active in sulfided form. Non-noble metals are usually selected from two groups in the periodic table, group VI and VIII [2]. For hydrotreating distillates that contain a substantial amount of nitrogen compounds, Ni/Mo sulfide on ?-alumina is the preferred catalyst while for feedstocks with low nitrogen content, Co/Mo sulfide is the commonly selected catalyst [5]. Catalysts based on tungsten (e.g. Ni/W sulfide on ?-alumina) are also active, but are less commonly used due to the high price of W [71]. Under hydrotreating conditions, the active metals are in the sulfide form, with Ni acting as a promoter of crystallites of MoS2. Molybdenum contents are typically 10 ? 14 wt % of the catalyst (on a metal basis), while the promoter Ni or Co is typically added to a content of 2 ? 4 wt % [5,72,73]. Common additives in distillate hydrotreating catalysts are P, 27  which tends to enhance N removal at the expense of some loss in HDS activity, and F to enhance the acidity [71]. The hydroconversion catalyst, on the other hand, must have both metallic sites and acidic sites in order to promote the required reactions. The metallic sites provide the catalyst with a hydrogenation function as in a hydrotreating catalyst, whereas the acidic sites provide cracking and isomerization functions. There should be a rapid molecular transfer between the metallic sites and acidic sites in order to avoid undesirable secondary reactions and coke formation. Therefore, the balance of the metallic sites and acidic sites on the catalyst surface are very important to obtain desirable performance. The acidic sites are usually provided by zeolite and/or amorphous silica-alumina or ?-alumina [2,74]. 2.4.2 Non-conventional Catalysts for Hydroprocessing One of the main objectives of heavy-oil and residue upgrading has been the improvement in pumping ability for transportation by pipeline. In this regard, the down-hole upgrading (using microorganisms) and/or upgrading on the site of the production well, have been investigated. Hydrogen addition processes accounted for most of the novel methods, which have been investigated for upgrading heavy feeds to distillate fractions. These processes utilize catalysts either in a dissolved form or in a finely dispersed form.  Apparently, the methods such as down-hole upgrading, bio-catalysis and the use of dispersed catalyst are still in various stages of development. Thus, additional studies and investigations need to be done before commercial use of these catalysts and proceses on the site of refineries can occur. In this regard, novel processes must be cost-effective in 28  comparison with several carbon-rejection processes (coking and thermal cracking) and hydroconversion processes, which have been used commercially for upgrading the most problematic feeds for decades. They must also be competitive with conventional hydroprocessing. It is believed that the scale of operation may be an important factor influencing commercial viability of these novel processes [2]. Catalyst precursors added to the reactor in slurry-phase hydroconversion processes are divided into two main categories: 1) soluble precursors and 2) finely powdered precursors. 2.4.2.1 Soluble Catalysts Once dissolved in a liquid solution, the catalyst precursor can be blended with heavy feed. Depending on the type of catalyst, both aqueous and hydrocarbon phases can be used for catalyst dissolution. Under hydroprocessing conditions, the catalyst precursors are decomposed and are subsequently converted to a catalytically active phase via reactions with H2S, which is released during HDS. The catalyst made in situ is typically nanometer in size. This ensures a high dispersion and contact with the reactant molecules in the heavy oil feed. Moreover, in such a state, most of the accessible catalyst surface will be utilized; however, some difficulties and lower activity are usually observed in the case of water-soluble catalyst precursors compared to oil-soluble catalyst precursors [2,26]. 2.4.2.1.1 Oil-soluble Precursors There are a number of organometallic compounds, which are soluble in hydrocarbon liquids, e.g., metal salts of organic acids (naphthenic, acetic, oxalic, octoic), organic amines 29  and organometallics. The carbonyl compounds of transition metals (e.g., Fe(CO)5) and ferocenes are soluble in oil and as such are potential catalyst precursors as well [75]. Several examples found in the scientific literature may be used to assess the potential of oil-soluble catalysts for hydroprocessing of heavy feeds. Sato et al. [14] compared an ultra fine MoS2 with MoS2 made in situ from oil-soluble Mo-dithiocarbamate during hydroprocessing of Kuwait AR. An extensive micronization of the former was required in order to achieve activity, which would approach that of the in situ made MoS2. Shi et al. [19] noticed that catalytic activity of the dissolved Mo catalyst prepared from Mo-dithiocarbamate during upgrading of VR was significantly enhanced in the presence of a H donor such as tetralin. Apparently, this was due to the synergy between the catalyst and H donor compound used in their study (tetralin). Kennepohl et al. [64] investigated hydroprocessing of Athabasca bitumen using MoS2 formed in situ. They found that in concentrations lower than 800 ppm Mo, the MoS2 formed in situ from the dissolved Mo-naphthenate suppressed coke formation during hydroprocessing. However, when a basket with a conventional CoMoS/Al2O3 catalyst was present in the same system to achieve additional upgrading, the beneficial effect of the dispersed MoS2 was not observed. Moreover, activity of the conventional catalyst was affected by the dispersed catalyst. Most likely, the dispersed catalyst deposited on the external surface of the catalyst particles and as such had a harmful effect on the diffusion of reactant molecules into the interior of the particles. 30  According to Del Bianco et al. [21], the MoS2 produced in situ from a Mo-naphthenate precursor, facilitated active H required for stabilization of radicals produced by thermal cracking. They used a VR (containing V+Ni of 300 ppm) derived from Belaym crude oil. The experiments were conducted between 410 and 450 ?C at 9 MPa H2 using a Mo concentration ranging from 200 to 5000 ppm. While using the Safania VR (~200 ppm V+Ni), the Mo-naphthenate was a much better MoS2 precursor than phosphomolybdic acid [41]. With respect to HDS, the combination of Mo+Co was better than Mo+Ni, in agreement with Lee et al. [76]. These authors used the oil-soluble compounds of Mo, W, Ni and Co as the precursors for the corresponding dispersed metal sulfide catalysts. For single metals, the best performance was observed for the Mo catalyst. The combination of Co+Mo was the best for HDS, whereas Ni+Mo was best for HCR. Although the catalysts used in slurry-phase technology, for example, those derived from oil-soluble Fe(CO)5, Ni-naphthenate, Co-naphthenate, and Mo-naphthenate, are effective for coke suppression, they are difficult to control with respect to their solubility, dispersion in the residue oil, and size of the final metal sulfide catalyst. The new candidate catalysts for upgrading heavy oils are colloidal metal sulfide catalysts in the nm size range prepared as microemulsions from water-soluble metal salts. Better control of particle size and dispersion of catalyst in the oil is expected in this type of catalyst. Duangchan and Smith [77] investigated the performance of Mo, Fe and Ni micelle catalysts prepared by different combinations of solvent and surfactant. The size of catalyst was also investigated and catalyst size in the range of 5-8 nm with decreasing order of Mo > Fe > Ni, was reported. Performance of the catalysts prepared in microemulsions was studied in a batch reactor 31  operated at 430 oC and reaction time of 1 h. Catalysts were sulfided in situ by H2S gas ( 5 % H2S / 95 % H2) added to the reactor with an initial pressure of 3.5 MPa. The best catalyst performance in terms of coke yield was reported for the Mo catalyst followed by Fe and Co (5.8 %, 6.2 % and 7.1 % of coke yield respectively), and a synergistic effect of CoMo and NiMo catalyst was observed. 2.4.2.1.2 Water-soluble Precursors Water-soluble catalyst precursors have attracted less attention than oil-soluble precursors. It is believed that some problems would be encountered while adding an aqueous solution of precursor to heavy oil [2]. Agglomeration of the salts dissolved in the water during rapid heat-up and water removal would result in a poor dispersion of catalyst in the oil. Several water-soluble compounds have been used as precursors for in situ catalyst formation, e.g., thiomolybdates, phosphomolybdates, nickel and cobalt nitrates [78]. Similar to the oil-soluble precursors, under hydroprocessing conditions, these compounds decompose and are converted to catalytically active sulfides (e.g., MoS2), which are in a highly dispersed form. A high dispersion of the in situ-made catalyst in heavy oil ensures perhaps the most efficient contact with the reactants in the oil. Liu et al. [79] added water-soluble catalysts containing Mo, Ni and Fe in the amount of ~1000 ppm of metal to VGO derived from a Chinese crude. At 435 ?C and 10 MPa, the following catalyst activity order was established for the removal of asphaltenes + resins: Mo>Ni >Fe. The content of aromatics in the products decreased in the opposite order. Under 32  these conditions, the submicron size of the sulfides of Mo, Ni and Fe formed in situ, were the catalytically active phase. The catalytic activity of water-soluble catalysts is influenced by the method used for their co-slurrying with the heavy oil feed. Slow water removal from the catalyst precursor has been shown to have a positive effect on the activity of water-soluble AHM precursor in AR upgrading [59]. These authors reported that removing the water from the mixture of oil and water soluble precursor at 100 oC ? 180 oC while purging the mixture with nitrogen gas was an optimum condition in their study to maximize the catalytic activity of MoS2 [59]. In another study, the water-soluble bi-metallic catalyst containing MoS2 and Fe(1?x)S was prepared from (NH4)6Mo7O24?4H2O (AHM) and Fe(NO3)3?9H2O, respectively, using a high dispersion method (slow water removal at low temperature of 80 oC and overnight mixing) in the AR derived from a Chinese crude [80]. The most active combination comprised 1100?1300 ?g/g of MoS2 and 25 ?g/g of Fe(1?x)S. With this combination, the highest conversion to distillates and lowest coke formation was observed. About 200 ?g/g of Mo added in the form of the soluble Mo compound gave a high conversion of the VR to distillates [80]. 2.4.2.1.3 Finely-powdered Precursors Dispersing a mineral form of the finely divided transition metal compound catalyst with heavy feeds has been examined as well. Under conditions applied during hydroprocessing, as well as in the presence of H2 and H2S, these compounds are converted to catalytically active metal sulfides (e.g., MoS2) which are finely dispersed in the feed. In this regard, the active phase is either in micron or less than micron size depending on the extent 33  of micronization of the solid catalysts. Among the type of inorganic compounds used, salts, oxides, sulfides and alloys have been successfully dispersed in heavy feeds [81]. For example, according to several studies, the unsupported Mo sulfide prepared by mechanical milling slowed down coke formation via quenching radical reactions during hydroprocessing of heavy feeds [82-84]. 2.5 Recovery of Dispersed/ Dissolved Catalyst The drawback of the methods employing dispersed/dissolved catalysts is the uncertainty regarding the catalyst recovery for reuse. For example, a plant processing about 75,000 bbl/d (~12,000 t/d) of a heavy oil feed requiring continuous addition of ~1000 ppm of the metal catalyst, would require the addition of about 12 tons of metal per day. Therefore, it is crucial that most of the metal be recovered for reuse if slurry-phase processes are to be economically viable. The dispersed catalysts added to the slurry-phase reactors mostly end up in the heavy products fraction (i.e. residue, sludge or coke). Recycling of these metal-containing fractions to the reactor inlet and mixing with the feed along with some make-up catalyst seems a practical way to re-use the catalyst in slurry-phase reactors. To utilize such procedures in slurry-phase reactors, the activity of the recycled catalyst should be assessed and quantified to make sure the recycle catalyst still has a substantial catalytic activity. For example, a sufficiently high H2 pressure would ensure that the coke deposition on the catalyst particles would be low. In other words, most of the original activity of the catalyst would be retained [2]. 34  There are some indications of attempts to extract the catalyst from VR by a solvent for subsequent reuse. Without extraction, the final separation of catalyst depends on the residue utilization option. Thus, the metals of interest will end up in the ash and/or slag in a concentrated form providing that the final residue utilization involves combustion and/or gasification. In this case, conventional methods (e.g., leaching, extraction, etc.) are available for the separation of the metal from the ash. However, a high concentration of metals (V, Ni and Fe) in the ash suggests that the latter may be utilized directly. Apparently, the catalyst recovery for reuse requires additional attention before the processes employing dispersed catalysts can be used on a commercial scale. In any case, such a process would have to operate in a continuous and/or semi-continuous mode near or on the site of a petroleum refinery. With respect to the metal recovery for reuse, the study of Lee et al. [76] deserves attention. These authors used oil-soluble compounds of Mo, W, Ni and Co as the precursors for dispersed metal sulfide catalysts. In this study, a fixed bed of extrudates made either from a microporous AC or ?-Al2O3 was placed downstream of the reaction zone with the aim of removing metals from the product streams. A high efficiency of metal removal was achieved using the AC extrudates. In this study, the AR containing ~26 ppm V+Ni was used as the feed and 300 ppm concentration of bi-metallic catalyst was used as the dispersed catalyst. Results presented by the authors showed that the concentration of V, Ni and catalyst metal (Mo, Ni, Co, W) in the liquid product dropped to 11.8-15.9 ppm during 20 h of reaction. Previous investigations [18] have also shown that most (>95 wt %) of the metallic catalysts are captured by the coke and Kouzu et al. [85] have shown that the metallic catalysts 35  captured by the coke are still catalytically active. The fact that carbonaceous materials in the reactor (e.g. generated coke during hydroconversion or added coke to the reactor at the beginning) can absorb metal particles in the heavy oil during the hydroconversion reaction implies that the coke has the potential to be recycled to the reactor as the catalyst. However, few studies are available on catalyst recycle which are presented in Section  2.6. 2.6 Catalyst Recycle in Heavy-oil and Residue Upgrading While activity of different catalyst types in slurry-phase reactors as well as effect of different parameters on the activity of the catalyst (H-donor compounds, catalyst loading and operating conditions) have been studied [12-24], very few studies have reported on the activity of recycled catalysts in hydroconversion of reasidue and heavy-oils. Kouzu et al. [85] investigated the catalytic activity of a carbon-supported NiMoS catalyst in a batch autoclave and compared the catalyst activity to that of an alumina-supported catalyst in the hydroconversion of an atmospheric residue of a Middle Eastern crude at an initial H2 pressure of 5 MPa and temperature 350 ? 450 ?C. The authors recovered the solid toluene-insoluble fraction from the 1st hydroconversion experiment and recycled the whole recovered solid (catalyst + carbonaceous materials) in the next hydroconversion experiment. The authors showed that the activity of the carbon-supported catalyst remained almost constant when recycled twice, while the activity of the alumina-supported catalyst decreased in terms of suppression of toluene insoluble (coke) product. The carbon-supported catalyst had lower toluene insoluble yield in the 1st and 2nd recycle compared to the experiment using the fresh catalyst. The S content, N content and selectivity toward different hydrocarbons in the 36  products did not change significantly in the 1st and 2nd recycle for the carbon-supported catalyst. Using the alumina-supported catalyst, while the N content of liquid in the 1st and 2nd recycle remained constant, the S content and coke yield increased linearly with recycle number. The authors did not report on catalyst deactivation since the catalyst was not recycled more than twice. Dunn et al. [67] investigated the catalytic activity of two V and Ni-based carbonaceous solids with low (12 wt %) and high (50 wt %) metal loading in the hydroconversion of bitumen (Cold Lake atmospheric residue and gas oil to produce a product with the same API? and viscosity as Cold Lake bitumen). The solid with low (Ni+V) loading was recovered from an ExxonMobil flexicoker (Venturi fines) while the solid with high metal content was a mild-burnt Venturi fine mixed with the gasifier bed coke. A mixture of the solid ash and the feed oil were loaded in a semi-batch reactor and heated at 365 ?C for 20 mins for in-situ sulfidation of the vanadium and nickel oxides. Then the reactor was heated to 413 ?C and the reaction proceeded for 2 hours. The catalytic performance was reported in terms of the API? of the recovered liquid product from the hydroconversion experiments. The solid catalyst was recovered by hot filtration of products under N2 gas for recycle in a subsequent hydroconversion experiment. The study showed that the activity of the recycled catalyst was not the same as the activity of the catalyst used once-through. A 10 % reduction in the API? of the liquid in the 1st recycle and marginal decrease thereafter was observed. Although the performance of the recycled catalyst in their studies yielded a hydroconverted product of decreased stability and lower API?, the overall reduction in product quality was small in comparison to that produced using fresh catalyst. To circumvent catalyst deactivation, the 37  authors added 10 wt % fresh catalyst to the recycled solid but product analysis of the hydroconversion experiments showed that the addition of fresh catalyst did not improve product quality. The authors also studied the effect of mild calcination of the recycled catalyst and showed that partial removal of contaminants from the catalyst increased the activity of the recycled catalyst but did not restore the activity to the level of the fresh sulfided solid.  Another relevant study was reported by Del Bianco et al. [43]. Using an autoclave reactor operated at 410 ?C, the authors investigated the recyclability of the asphaltene-catalyst mixture recovered from hydroconversion experiments. Using a high Mo concentration of 3000 ppm prepared from a molybdenum naphthenate precursor, the asphaltene-catalyst mixture from the 1st hydroconversion experiment was recycled 10 times. Although one of the first studies that detailed the recycling of dispersed catalysts, the reaction temperature (410 ?C) was such that low conversion occurred (as is also the case for the results reported by Dunn et al. [67]). The results presented by Del Bianco et al., although very valuable for investigating the activity of recycled catalyst at low conversion, cannot be extrapolated to the high residue conversion conditions (high temperature of ~ 450 ?C) relevant in commercial operations. 2.7 Chemistry of Upgrading Reactions Different upgrading processes share common reaction chemistry and thermodynamics for converting residue (524 ?C+ fraction). Each process selects conditions of severity and catalytic activity to obtain the desired products and to control undesirable side reactions. 38  Given the number of apparent process alternatives, it is easy to lose sight of the common reactions in these processes. In this section, a brief introduction of the thermal and catalytic reactions occurring during the hydroconversion process is presented. 2.7.1 Thermal Reactions The term cracking is referred to the cracking of chemicals bonds of different compounds in the petroleum which is driven by high temperatures (>350 ?C), during which the higher molecular-weight molecules of crude are cracked and converted to lower molecular-weight molecules. Cracking reactions (carbon?carbon bond cleavage) require high temperature (to be thermodynamically favored) and do not have a significant rate at temperatures below 350 ?C [5,6]. Other than C-C bond breakage, C-S and C-N bond breakage are also of interest during upgrading [5]. Thermal or free-radical reactions occurring during the thermal conversion are not catalytic reactions. A free radical is an atom or group of atoms having an unpaired electron in their structure. Free-radicals are highly reactive and free-radical reactions determine the product distribution during thermal cracking [6]. Conversion and properties of the products are mainly determined by the severity of thermal processing. Severity of thermal treatment processes for processing the residue ranges from mild treatment (reduction of viscosity) to ultrapyrolysis for complete conversion to olefins and light oils. The operating temperatures for production of distillate products, range from 410 ?C to 550 ?C in thermal cracking [5]. The higher the temperature, the shorter the time required to achieve a given conversion. 39  2.7.1.1 Carbon-Carbon Bond Breakage At a high enough temperature, thermal reactions are noncatalytic and occur spontaneously in organic mixtures. As discussed earlier, the primary objective of any upgrading process is to break chemical bonds (mainly C-C, C-S, C-N and carbon-hydrogen) to produce a lighter product. Bond dissociation energies of the most common bonds present in petroleum are presented in Table  2.4. A high value of the dissociation energy is an indication of the difficulty to break these bonds. The carbon-hydrogen bond in the aromatic compounds has a higher dissociation energy than the other carbon-hydrogen bonds. Resonance stabilization in the aromatic compounds is one of the resons for this high stability. The C-S bond dissociation energy is the lowest among different bond energies. This means that the C-S bond will break at a lower temperature in upgrading processes compared to other bonds. This bond breakage and hydrogenation of the sulfur in the reactor produces H2S gas in the reactor. The H2S is usually the sulfidation agent of the catalyst in catalytic processes [2]. Although the ultimate goal of upgrading is C-C bond breakage to produce lower boiling point products, the mechanism of this bond breakage is still unclear. Experimental results by Blanchard and Gray [86] showed that the C-C bond breakage does not simply occur due to direct homolytic scission reactions followed by radical stabilization or recombination. A more recent study from Gray and McCaffrey [87] showed the importance and role of the chain reaction mechanism proposed by LaMarca et al. [88] and these authors have elaborated on the proposed model by LaMarca et al. for liquid-phase upgrading processes. By investigating the dissociation energy of different bonds in model compounds, 40  the authors concluded that the direct C-C bond breakage without chain reactions is far too slow. Gray et al. showed that the theoretical rate of reaction (assuming hemolytic scission reactions are solely responsible for C-C bond breakage) was not consistent with the observed rate of conversion [87]. Table  2.4 Bond dissociation energies of different bonds. Data from [89]. (Copyright ? 1982, Annual Reviews, Reproduced with permission) Chemical bond Representative bond Dissociation energy, KJ/mol C-C (aliphatic) C2H5-nC3H7 344 ? 4 C-H primary C2H5-H 411 ? 4 C-H secondary iC3H7-H 398 ? 4 C-H aromatic C6H5-H 464 ? 8 C-S CH3S-C2H5 307 ? 8 C-N C2H5-NH2 342 ? 8 C-O C2H5O-C2H5 344 ? 4  2.7.1.2 Reaction of Mixtures of Hydrocarbons The cracking of a mixture of hydrocarbons such as bitumen and residue oil is determined by the reactivity of its constituents. Free radical intermediates generated during different cracking reactions are being shared in different reactions during hydroconversion. The various compounds will interact during the degradation process and sharing of free radical intermediates in different reactions adds to the complexity of the reactions occuring in 41  the hydroconversion process [5]. The crackability of different hydrocarbon lumps can be described as follows, from the most reactive to the least reactive compounds: Paraffins > Linear olefins > Naphthenes > Cyclic olefins >  Aromatics > Bridged aromatics [5] Molecules with high molecular weight or boiling point tend to crack more readily than the hydrocarbons with small molecules and low boiling point. The higher cracking tendency can be explained by an increase in probability of presence of bonds with low dissociation energy which are more susceptible to cracking at elevated temperatures [6,32]. 2.7.1.3 Reactions of Sulfur, Nitrogen and Oxygen Compounds Among the different S compounds present in petroleum, thiophenic sulfur compounds remain unaffected by thermal reactions during upgrading, as with carbon-based aromatic compounds. On the other hand, the disulfides, thiols and thioethers are quite reactive under thermal process conditions. These sulfur compounds account for about 50 % of the total sulfur in bitumens and asphalts [5,90]. Due to the very low dissociation energy of the C?S bond compared to other aliphatic bonds (Table  2.4), the thermal reactions of sulfur are very favorable. For example: ? ? ? ? ?? + ??? ?  ? ? ?+ ?? ? ?+ ???               ( 2.1) (From M. R. Gray [5], Copyright ? 2002, American Chemical Society, Reproduced with permission) Disulfide bonds (R-S-S-R) are even weaker than thioethers, and are the most likely bonds to dissociate and serve as initiators for the free-radical chain reaction. 42  Thermal processing of bitumen (even at the relatively low temperature of 250 ?C,) is accompanied by S removal and evolution of hydrogen sulfide. The easy cleavage of the sulfide bonds is believed to be a major mechanism for cracking of the high molecular weight components of heavy-oil and residue. The extent of this conversion can reach up to 30 ? 50 wt % without any use of catalyst. The N compounds are present in bitumen and residue oil as heteroaromatic species. The nitrogen containing heteroaromatics remain unaffected by thermal reactions and catalysts with selective activity toward nitrogen removal are needed to remove these compounds from the oil. Removal of Oxygen species during bitumens and residue upgrading has not been studied as intensively as S and N containing compounds. While ethers and furans have the same cracking tendency as the similar sulfur compounds, carboxylic acids and ketones are relatively reactive under hydroconversion conditions [5]. 2.7.2 Catalytic Reactions The formation of light products from residue oil and heavy oil feedstocks is primarily driven by thermal reactions. The main role of catalysts in primary upgrading is to boost H transfer during the upgrading process and prevent condensation and coking reactions by transferring H to the highly reactive free-radicals and hydrogenation of olefins [5,87,91]. A wide range of transition metals are active for hydrogenation reactions such as Fe, Mo, W, Sn, Ni and Co. These metals are active to facile hydrogenation reactions in both zero valent metal state and in the sulfided state [2]. Under the reaction conditions, high sulfur content of the residue feed and release of H2S will sulfide the metals inside the reactor. This is the 43  reason why very active hydrogenation catalysts such as Pt cannot be used as the catalyst in primary upgrading. The metal used, surface area of the metal as well as the use of promoters, are the factors determining the activity of a catalyst in the upgrading reactor [5]. Supporting the metal sulfides on a high surface area support, using soluble organometallic compounds as the catalyst precursor (Section  2.4.2.1) or finely-powdered metal precursors (usually iron compounds) are normal practices to achieve a high surface area catalyst. 2.7.2.1 Reactions of Aromatics When in contact with high-pressure H2 and a hydrogenation catalyst, aromatic groups can be hydrogenated to produce hydroaromatics and naphthenes. This reaction is more thermodynamically favorable when the number of rings in the aromatic cluster increases. For example, benzene rings are the least reactive aromatic compounds. One of the reasons for this high stability is the resonance structure of benzene [5].  Polyaromatics with higher number of benzene rings in their molecular structure compared to benzene (such as naphthalene and phenanthrene) are more easily hydrogenated than benzene. In the presence of H2, a hydrogenation catalyst, and a H acceptor, the hydrogenation reaction is reversible and these compounds can undergo cycles of hydrogenation and dehydrogenation as shown in Figure  2.3.  44   Figure  2.3 Hydrogenation/dehydrogenation cycle of naphthalene [5]. (Copyright ? 1994, Taylor & Francis Group LLC., Reproduced with permission)  Similar hydrogenation/dehydrogenation reaction can be written for phenanthrene:  Figure  2.4 Hydrogenation/dehydrogenation cycle of phenathrene [5]. (Copyright ? 1994, Taylor & Francis Group LLC., Reproduced with permission) Naphthalene 1,2,3,4-tetrahydronaphthalene4 R-HH24 R.++ +45  As shown in Figure  2.3 and Figure  2.4, these reactions are a mechanism to shuttle H from the gas phase to reactive free radicals in the liquid phase. These H transfer reactions are very important in terms of free radical stabilization and prevention of polymerization and condensation reactions. Polymerization and condensation reactions are of major mechanisms of coke formation in primary upgrading reactions (Sections  2.7.3 and  2.7.4) [5,87]. 2.7.2.2 Reactions of Sulfur and Nitrogen Compounds As mentioned earlier (Section  2.7.1.3), S containing compounds in heavy-oil and bitumen have different reactivity based on their structure. While aliphatic sulfur containing compounds are converted easily by thermal cracking, thiophenic compounds need a catalyst for hydrodesulfurization [5]. Catalytic hydrodesulfurization usually occurs by one of two paths presented in Figure  2.5 [92-94].  Figure  2.5 Schematic of hydrodesulfurization of dibenzothiophene (DBT). Reproduced from [92]. (Copyright ? 1994, Taylor & Francis Group LLC., Reproduced with permission) (Path 1)(Path 2)2 H23 H22 H2SS46  The first path which involves bond breakage by H is referred to as hydrogenolysis whereas the second path is refered to as hydrogenation. The importance of each of these paths in the overall S removal from the bitumen or residue oil is a function of reaction mixture, H2 pressure and the type of catalyst used in the process [5,72]. Compared to sulfur-containing compounds, nitrogen-containing compounds have much less reactivity. This is in part because in HDN reactions, only the hydrogenation path for N removal is observed [5,69]. This is illustrated by the hydrodenitrogenation of pyrrole in Figure  2.6.  Figure  2.6 Hydrodenitrogenation (HDN) mechanism of pyrrole [5]. (Copyright ? 1994, Taylor & Francis Group LLC., Reproduced with permission)  2.7.3 The Mechanism of Coke Formation during Upgrading The formation of solid sediments, or coke, during hydroconversion of bitumen and petroleum residue fraction can be a major limitation on reactor performance and yield. For example, conversion of 95% of the residue fraction of Athabasca bitumen by the H-Oil process yields 50 wt % solids in the 520+ ?C product [5]. At its heart, the formation of solid materials during upgrading is the result of a liquid-liquid phase transition [95]. 2 H2+2 H2 N H3HNHN47  During this process, a liquid feed (usually with solids content of < 1 wt %) reacts and forms new solid materials that were not present initially, therefore, liquid components are being converted into insoluble species. Understanding the mechanism of these reactions is a very important step in understanding the mechanism of coke formation during hydroconversion. Although, due to the complexity of the constituent molecules of crude oil and residue, details of these mechanisms are not clear. The other important factor to recognize is that solids at the temperature of sampling may not be solid at reactor conditions, in excess of 350 ?C. In high-pressure reactors, the phase compositions are even further away from materials that can be easily sampled and related to ambient conditions [5]. The formation of coke requires that a portion of the oil becomes insoluble, eventually leading to the formation of solids (solids at room temperature). Thermal cracking of petroleum materials can give insolubility to a portion of the liquid by the following mechanisms: 1) increasing aromatic C content, driven by cracking of side chains and dehydrogenation reactions, and 2) increasing molecular weight, driven by polymerization reactions. Either of these reactions, or both at the same time, can lead to insolubility. Most of the literature on the molecular mechanisms and models of coke formation deal with light olefins in cracking of naphtha, where the species are all in the gas phase. These compounds are either too volatile or too simple to properly study the liquid-phase reaction occurring during the heavy-oil and residue upgrading. In contrast, coking in conversion of bitumen is usually a liquid-phase phenomenon, which is much more favorable for polymerization [4,87,96]. 48  In the 1970s, the common view was that coke formation during residue thermolysis was caused by a sequence of polymerization and condensation steps from the lightest to the heaviest fractions as follows: Oils ? Resins ? Asphaltenes ? Carbenes ? Coke In 1993, Wiehe [95] proposed a mechanism for coke formation based on the liquid-liquid phase separation, which is the most commonly accepted mechanism of coke formation in the thermal conversion of residue. A summary of this mechanism is presented in Figure  2.7.  Figure  2.7 Schematic of the phase-separation mechanism for coke formation in thermal conversion of residue [97]. (Copyright ? 2007, American Chemical Society, with permission) 49  Based on the mechanism proposed by Wiehe [4,95], asphaltenes are assumed to consist of a polynuclear aromatic core which is thermally stable. Other pendant groups, such as aromatics and saturates, are connected to the polynuclear aromatics by thermally unstable bonds. When this molecule is exposed to thermal cracking temperatures, the unstable bonds break and form free radicals on the cracked asphaltene molecule body. Removal of the side chains from the asphaltene molecules decreases the solubility of these molecules. However, as long as these molecules are dispersed in the rest of the oil (resin and hydrogenated aromatic compounds, which are H donors), H abstraction from the solvent phase terminates the free radicals and prevents coke formation. With further cracking of side chains in the asphaltene molecule during the thermal conversion, asphaltene solubility in the oil becomes less and less, and eventually the asphaltene core becomes insoluble in the oil. Insoluble asphaltene cores undergo a liquid-liquid phase separation under reaction conditions. The phase separation severely decreases the H abstraction mechanism from the H donor compounds (such as resins and hydroaromatics, which are in a separate liquid phase than the asphaltenes). Since the asphaltene-rich phase has very little hydrogen to terminate free radicals, free radicals in asphaltene cores react and form a high molecular weight molecule which is insoluble and is called coke. Before the coke molecules are formed, asphaltene cores (polynuclear aromatic hydrocarbon) tend to orient parallel to each other. Due to parallel orientation of the coke precursor molecules (asphaltene), the molecules can be detected in an optical microscope using cross-polarized light. Under cross-polarized light, ordered structures (also called meso-phase) are bright while amorphous and unordered structures are 50  dark [95,97-100]. An example of the ordered structures observed under normal and polarized lights is presented in Figure  2.8.  Figure  2.8 Optical micrograph of toluene-insoluble coke from thermolysis of the Cold Lake vacuum residue [95]. (Copyright ? 1993, American Chemical Society, with permission)  2.7.4 The Mechanism of Catalytic Coke Suppression The mechanism of coke suppression in catalytic residue upgrading has been well studied [87,95,101,102]. Although conversion and cracking of macromolecules and polynuclear aromatic hydrocarbons (PAHs) to smaller molecules is thermally driven, the catalyst transfers H from the vapor phase to the liquid mixture. While the H transfer by catalysts to the heavy components has been recognized as an important step, the mechanism of this phenomenon is not well established. Although some studies propose that H atoms from catalytic centers hydrogenate radicals generated by homogeneous cracking reactions 51  [65,103], no experimental result proving the direct H transfer from the gas phase to free radicals compared to alternative mechanisms is available to date [87].  Initiation:                                                        R-R'  kin?  R.+R'.      (1) Propagation:                                      Hydrogen abstraction                            ?j.+ RjH kij?  ?jH + Rj.     (2) ?-scission                                              R.-C-C-?j  ki?  R-C=C + ?j.        (3) Radical addition (polymerization)        Ri.+ C=C-Rk ka?  Ri-C-C.-Rk    (4) Radical rearrangement                          R.-C-C-?jks?  Ri.      (5) Termination:                                                  Ri.+ Rj. kt?  Ri-Rj      (6) Olefin removal: Donor solvent                                        R-C=C + R,H2 kd?  R-C-C + R,    (7) Catalytic                                                R-C=C + H2(g) kh?  R-C-C    (8) Aromatic hydrogenation:                               R,+ H2 ka?  R,H2     (9)  Figure  2.9 Extended chain reaction mechanism for liquid-phase cracking, including catalytic reactions [87] (Copyright ? 2002, American Chemical Society, Reproduced with permission)  According to the mechanism proposed by LLKC and Gray et al. [87,88] formation of coke and insoluble phase in the reactor is a result of a series of reactions similar to polymerization (radical addition) reactions. Reactions include initiation (reaction 1 in Figure  2.9), propagation (reactions 2 to 5 in Figure  2.9) and termination (reaction 6 in 52  Figure  2.9) steps. If the termination reaction is not prevented by hydrogenation of the produced olefin in reaction 3 of Figure  2.9 (through reaction 8 in Figure  2.9), two aromatic radicals of ??.  and ??.  will undergo a termination reaction (reaction 6 in Figure  2.9). The polymerization and condensation will increase the aromatic content of the molecule and eventually will result in formation of a separate liquid phase (so called mesophase) which leads to coke formation. The role of the catalyst in hydroconversion of crude oil and residue is to facilitate and accelerate H transfer from the vapor phase to the liquid phase during the reaction. In the context of the mechanism proposed by Gray et al. (presented in Figure  2.9), one may consider two important contributions for the catalyst in upgrading of residue and coke prevention through the chain reaction mechanism: 1) hydrogenation of olefins, mainly produced by ?-scission of free-radicals (reaction 8 in Figure  2.9) [88] and 2) partial hydrogenation of PAHs which have acted as H donors and become dehydrogenated (reaction 9 in Figure  2.9) [87]. The overall outcome of these hydrogenation reactions is an increase in concentration of saturates and partially hydrogenated PAHs which act as hydrogen donor compounds during hydroconversion reactions. Due to the complexity of the molecules in residue oils, such catalytic hydrogenation and H transfer reactions can potentially be promoted by solids present in the feedstock and by the metal walls of the reactor vessels. The latter has a noticeable catalytic activity in small-scale experiments [104]. 53  2.8 Lumped Kinetic Modeling of Residue Hydroconversion A lumped kinetic modeling approach, in which a group of compounds are considered as a kinetic lump instead of dealing with each component independently, was developed to tackle the complexity of the kinetic modeling of reactions related to heavy-oil and residue processing. This method of kinetic modeling eliminates the complexities related to the large number of differential equations and limitations of computational capacity to solve these equations when dealing with complex feedstock like residue and heavy-oil. At the same time, lumped kinetic modeling provides very valuable information for the refineries since the lumps are selected so that they resemble the real fractions of products in normal refineries. Lumped kinetic modeling of residue and heavy-oil conversion has been reviewed [105] and studied by many researchers [106-110] mostly at low temperatures of < 430 ?C. Sanchez et al. [107] studied the hydroconversion of a heavy oil in a fixed-bed reactor. Using a commercial Ni/Mo catalyst in a 1-in diameter reactor, the authors investigated the activity of catalyst in the reaction temperature range of 380 ?C ? 420 ?C. Due to low reaction temperature, the authors neglected the coke formation and yield of solid product was excluded from the proposed kinetic model. Using a 5-lump kinetic model proposed by the authors (presented in Figure  2.10) and assuming first-order reactions, the model was applied to measured experimental data at three reaction temperatures of 380 ?C, 400 ?C and 420 ?C. The authors found that most of the gas production was from heavier compounds (i.e. residue) and no gas was produced from the reaction of lighter hydrocarbons (i.e. distillate and naphtha) even at the highest temperature of their experiments (420 ?C). They also found that 54  at moderate temperatures, the reaction selectivity was mainly toward the production of distillate and VGO and the naphtha production was very small. Although the authors did not provide any explanation for this phenomenon, it seems that the low reaction temperature of reaction cannot overcome the high activation energy of cracking reactions (due to small molecule sizes) to form lower boiling point compounds.  Figure  2.10 Proposed kinetic model by Sanchez et al. [107]. (Copyright ? 2005, American Chemical Society, Reproduced with permission)  In another very similar study by Sanchez et al. [111], the authors investigated the effect of pressure on the kinetics of hydroconversion at moderate conditions (reaction temperatures of 380-420 ?C and H2 pressure of 6.9-9.8 MPa). This was done by addition of a pressure effect term to the Arrhenius equation for k-value calculations as follow: ? = ?. ??? (????)(???)?       ( 2.2) 55  Where A is the pre-exponential factor, Ea is the activation energy of reaction, R is the gas constant, T is the reaction temperature, ?0 is the reference pressure, p is the reaction pressure and ? is a constant. The authors reported that an increase in H2 pressure only increased the residue conversion at high temperature (420 ?C) and H2 pressures up to 8.2 MPa. At lower reaction temperatures or H2 pressures higher than 8.2 MPa, an increase in the reaction pressure did not increase the residue conversion. They also reported that in the range of temperatures studied, no naphtha cracking to smaller molecules (i.e. gas) was observed and that gas and naphtha production was exclusively formed from cracking of residue and vacuum gas oil (HGO). 2.9 Summary The literature review presented in this chapter, highlights some advantages of slurry-phase reactors compared to fixed-bed reactors for residue oil hydroconversion. It is concluded that using more active and expensive metals (i.e. Mo) increases the yield of liquid product in the upgrading process and suppresses the formation of less valuable coke. However, commercial viability of slurry-phase upgrading using more active and expensive metals compared to cheap additives is dependent on the effective recovery and recycle of valuable metals added to the reactor as the catalyst. Having a good understanding of the activity of recycled, unsupported catalysts in a slurry-phase upgrading reactor, as well as the potential deactivation mechanism of unsupported catalysts in the slurry-phase reactor, is the key to develop commercial slurry-phase upgraders which use more valuable catalysts. 56  Although many studies are available on the mechanism of supported-catalyst deactivation in hydroconversion reactors, no study on the catalytic activity and deactivation mechanism of recycled catalysts in slurry-phase upgrading is known to the author. Consequently, assessment of catalyst activity of recycled and unsupported catalysts in hydroconversion of residue oil under high residue conversion conditions (T = 445 oC and H2 pressure of 13.8 MPa) as well as identification of the potential catalyst deactivation in the process of recycling, is the focus of the present study.           57  Chapter 3  Experimental Materials, Apparatus and Procedures 3.1 Catalyst Precursor Preparation 3.1.1 Micelle Precursor The ultimate goal in preparing micelle precursors, was to prepare nanoparticles of Mo and Fe using the reversed micelle technique, with the requirement that the metal content of the micelle be as high as possible such that the volume of the micelle to be added to the liquid oil was minimized.  A second goal was to prepare bi-metallic nanoparticles (Mo-Fe, Mo-Co and Fe-Co), to establish if any synergistic effects were present in such systems. Molybdenum (V) chloride (MoCl5) (Sigma-Aldrich, 99%), iron (II) nitrate (Fe(NO3)2.9H2O) and cobalt (II) nitrate (Co(NO3)2.6H2O) were used as precursors for Mo, Fe and Co metals in preparing the micelle precursors. n-Hexane (Acros Organics, 95%) and polyoxyethylene-4-lauryl-ether (PE4LE) (Sigma-Aldrich) were used as solvent and surfactant, respectively. A 2.0 M solution of lithium borohydride in tetrahydrofuran (LiBH4) (Sigma-Aldrich) was used as reducing agent. To prepare the micelle catalyst precursor, a salt of the desired metal was dissolved in a pool of water and then n-hexane was added to the solution followed by a surfactant. The mixture was then placed in an ultra-sonic shaker for 1-2 minutes to yield a transparent micelle dispersion. The metal salt was subsequently reduced by adding 5 times excess LiBH4 58  prior to addition of the catalyst to the reactor. All the micelle preparation steps were done at ambient temperature. As an example, to obtain 600 ppm Mo in 80 g of the feed oil, the catalyst precursor was prepared from 6 mL of n-hexane and 1.92 mL of PE4LE added to 0.1366 g of MoCl5. To obtain different Mo concentrations in the feed oil, the amount of each component of the emulsion was adjusted proportionally. Different ratios of solvent (n-hexane) to surfactant (PE4LE) were tested to both minimize the volume of the catalyst precursor for a certain metal concentration and also to ensure a stable metal particle was prepared in the reverse micelle. Table  3.1 summarizes the microemulsions prepared with single metals (Fe, Mo and Co) and bi-metals (Fe-Mo, Co-Mo and Fe-Co). The results shown in Table  3.1 demonstrate that by adjusting the amount of salt, solvent and surfactant, the required volume of the microemulsion needed to achieve 600 ppm of metal in the residue oil was reduced to less than 10 ml for the single metal catalysts and about 15 ml for the bi-metals. Results from dynamic light scattering (DLS) analysis (Section  3.7.11), summarized in Table  3.2, demonstrate that, except for the Fe sample prepared with high solvent volume (Fe-1), the micelle diameter was < 10 nm. The micelle diameter was in good agreement with TEM observations, illustrated in Figure  3.2. Hence, it was concluded that highly concentrated reversed micelles for both single and bi-metal systems, were successfully prepared and upon reduction, the resulting nanoparticles were < 10 nm in diameter.  In all future experiments, the Mo and Fe micelle catalysts were prepared according to Mo-2 and Fe-2 of Table  3.1, respectively. The use of a reversed micelle as the catalyst precursor requires that the metal content of the micelle be as high as possible so that the volume of the microemulsion to be added to the 59  liquid oil is minimized. By adjusting the amount of metal salt, solvent and surfactant, a solvent/surfactant ratio of 3.1:1 was shown to be optimum to obtain stable micelles while also decreasing the microemulsion volume needed to achieve 600 ppm of Mo (48 mg of Mo) in the residue oil to less than 10 mL. The Mo micelle prepared using this ratio of solvent/surfactant was analyzed by DLS and Figure  3.1 shows the Mo micelle size distribution in the emulsion with a micelle average diameter of 3.5 nm. Repeat measurements of this analysis for different Mo micelle preparations demonstrated good control of particle size by this procedure.  Figure  3.1 Micelle size distribution in the emulsion analyzed by DLS. Solvent : surfactant = 3.1:1, equivalent Mo concentration in the feed = 600 ppm. 60   Table  3.1 Summary of microemulsions prepared with single metals (Fe, Mo and Co) and bi-metals (Fe-Mo, Co-Mo and Fe-Co).ID Metals (g, ml) Metal salt Hexane (ml) PE4LE (ml) Reducing agent (ml) Catalyst Volume(1) (ml) Observations Mo-0 0.02, - MoCl5 50 2.5 0.6 (LiBH4) >100 Black after reduction, stable Mo-1 0.024, - MoCl5 6 0.96 0.6 (LiBH4) ~14 Transparent and stable before reduction. After reduction, black and nearly no sediment Mo-2 0.024, - MoCl5 3 0.96 - ~8 Fe-0 0.02, 0.14 Fe(NO3)2.9H2O 50 2.5 0.11 (N2H4) >100 Brown after reduction, stable Fe-1 0.024, 0.20 Fe(NO3)2.9H2O 6 0.96 0.12 (N2H4) ~14 Transparent and stable before reduction. After reduction, there is a little black sediment Fe-2 0.024, 0.20 Fe(NO3)2.9H2O 3 0.96 - ~8 Co-0 0.02, 0.11 Co(NO3)2.6H2O 50 2.5 0.11 (N2H4) >100 Yellow-Orange, stable Co-1 0.024, 0.10 Co(NO3)2.6H2O 6 0.96 0.12 (N2H4) ~14 Transparent and stable before reduction. After reduction, obvious cinnamon sediment Co-2 0.024, 0.10 Co(NO3)2.6H2O 3 0.96 - ~8 Mo-Fe-0 0.04, 0.14 MoCl5 + Fe(NO3)2.9H2O 100 5 1.2 (LiBH4) >100 - Mo-Fe-1 0.048, 0.2 MoCl5 + Fe(NO3)2.9H2O 12 1.92 1.2 (LiBH4) ~15 Little black sediment Mo-Co-0 0.04, 0.11 MoCl5+ Co(NO3)2.6H2O 100 5 1.2 (LiBH4) >100 - Mo-Co-1 0.048, 0.10 MoCl5 + Co(NO3)2.6H2O 12 1.92 1.2 (LiBH4) ~15 A little dark blue sediment Fe-Co-0 0.04, 0.25 Fe(NO3)2.9H2O + Co(NO3)2.6H2O 100 5 0.22 (N2H4) >100 - Fe-Co-1 0.048, 0.3 Fe(NO3)2.9H2O + Co(NO3)2.6H2O 12 1.92 0.24 (N2H4) ~15 Cinnamon and obvious floc 61  Table  3.2 Summary of DLS particle size estimates of reverse micelles present in catalyst precursor. Sample Radius (nm) Mo-2 5.3 Fe-2 7.5 Fe-1 12.1 ? 0.6 Mo-Fe-1 7.3 ? 0.4 Mo-Co-1 5.5 ? 1.5 Fe-Co-1 8.0 ? 0.4            Figure  3.2 HRTEM micrographs of Mo (top right), Fe (top left), Mo-Co (bottom right) and Mo-Fe (bottom left) micelle catalysts prepared in reversed micelles and reduced in LiBH4.  The bars on each micrograph correspond to 20 nm. 62  3.1.2 Commercial Oil-soluble (Mo-octoate) Precursor A Mo-octoate precursor, with 15.5 wt % Mo in molybdenum 2-ethylhexanoate, was provided by the Shepherd Chemical Company. The precursor was weighed and added to the residue oil so that the desired concentration of Mo in the residue feed was obtained. The active phase of catalyst (MoS2) was formed by in situ thermal decomposition followed by sulfidation of the metal in the reactor prior to the start of the reaction. 3.1.3 Water-soluble Precursor A water soluble catalyst precursor was prepared by dissolving appropriate amounts (8.9167 g) of AHM (Alfa Aesar, 99%) in 100 mL of de-ionized water. 1 mL of this solution provided 600 ppm of Mo metal in 80 g of feed. Similar to the oil-soluble precursor (Mo-octoate), the active phase of catalyst was generated in situ by thermal decomposition and sulfidation of the molybdenum salt. 3.1.4 Finely-powdered precursor The only finely-powdered catalyst precursor used in the hydroconversion experiment was FeSO4.H2O (mesh 200 provided by UOP LLC.). The required mass of solid was added to the reactor to yield 3 wt % of FeS in ~ 80 g of CLVR. Finely-powdered catalyst was used in only one concentration in both batch and semi-batch experiments in this study.   63  3.2 Batch Reactor Experimental Procedure A 300 mL stirred autoclave reactor (Autoclave Engineers) was used to assess the different catalysts in a batch mode at a reaction temperature of 415 ?C, initial H2 pressure of 5.5 MPa and a reaction time of 1 h. The geometry and mixing pattern on the batch reactor is shown in  Appendix B. In all experiments done in the batch reactor, 80 g of CLVR was used as feed. Properties of the CLVR can be found in Table  3.3. After loading the reactor with the 80 g of CLVR and the desired catalyst precursor, the reactor was flushed in N2 and then pressurized with pure H2 to an initial gas pressure of 5.5 MPa before heating to the desired reaction temperature (415 ?C). The reaction temperature and pressure were continuously monitored during the reaction. Depending on the catalyst type and concentration, the final reactor pressure at reaction temperature (415 ?C for experiments in the batch reactor) varied. After 1 h, the reaction was stopped by rapidly quenching the reactor using the cooling coil placed in the reactor. Solid and liquid products of the reaction were recovered and separated using a Beckman Coulter Allegra 25R refrigerated centrifuge, operated at 12,000 rpm for 30 mins. The liquid recovered from the centrifuged samples was tested to make sure that toluene-insoluble content of the liquid was less than 0.5 wt %. The recovered solid was washed on a 0.22 ?m membrane filter with toluene to remove any toluene-soluble material. Coke was defined as toluene-insoluble solids in all experiments. The gas remaining in the reactor (H2, hydrocarbons and H2S) was analyzed by a Varian Star 3400 CX gas chromatograph equipped with a 30m capillary column (i.d. of 0.53 mm) and a thermal 64  conductivity detector. Gas analysis data were used to calculate the H2 conversion and uptake, and the H2S and total gas yields. A schematic of the batch reaction system is presented in Figure  3.3. Table  3.3 Properties of CLVR used as reactant in the present study. Analyses and properties CLVR SARA analysis Saturates, wt % 11.0 Aromatics, wt % 38.7 Resins, wt % 32.6 Asphaltene, wt % 17.7 Elemental analysis C, wt % 81.6 H, wt % 9.7 S, wt % 6.0 N, wt % 0.5 H/C atom ratio 1.43 HTSD < 204 ?C, wt % 0.0 204 ? 348 ?C, wt % 0.0 348 ? 524 ?C, wt % 25.2 524+ ?C, wt % 74.8 Ash, wt % 1.4 Molecular weight 765 Specific gravity 1.04  65   Figure  3.3 Schematic of batch reaction system.  3.3 Semi-batch Reactor Experimental Procedure A 250 mL stirred semi-batch reactor (provided by Parr Instrument Company) was used to investigate the effects of higher temperature and pressure on catalyst performance as well as the performance of the recycled catalyst. Reaction temperatures of 415 oC, 430 ?C and 445 oC, a H2 pressure of 13.8 MPa, a H2 flowrate of 900 mL(STP)/min and a reaction time of 1 h were investigated. The inlet H2 gas flow and the exit gas were measured and monitored with a mass flow controller and mass flow meter, respectively. The total gas inlet and outlet volume was totalized using a Brooks Instrument 0154 flow controller for calculation of gas 66  yield and H2 uptake of each hydroconversion experiment. To capture the light hydrocarbons carried out of the reactor by the continuous H2 flow, a double-stage (2?150 mL) cold separator placed in an ice bath was used in the high pressure zone (after the reactor and before the back-pressure regulator) to separate and collect C5+ hydrocarbons leaving the reactor in the exit gas. A H2S scrubber using 400 mL of 1 N solution of NaOH was placed after the back-pressure regulator (BPR) to remove any H2S gas from the exit gas stream. The gas leaving the reactor during the 1h operation was collected in a Tedlar gas sampling bag for analysis. A schematic of the semi-batch reactor system is shown in Figure  3.4.  Figure  3.4 Schematic of the semi-batch reaction system.  67  The reactor was loaded with the desired amount of catalyst together with 80 g of the residue oil. After sealing, the reactor was purged in a 300 mL (STP)/min flow of N2 and then H2, each for 5 minutes. Using a back-pressure regulator (26-1700 series high pressure back-pressure regulator from Tescom) installed after the ice-bath condenser, held at 0 ?C, the reaction system was pressurized with H2 to a reaction pressure of 13.8 MPa for all experiments. The reactor was then heated to the desired reaction temperature and after 1 h, the reactor was rapidly cooled to ~ 15 ?C using an internal cooling coil and cold water as the coolant. Before depressurizing the reactor, a gas sample was collected for analysis by GC. Solid and liquid products of the reaction were also recovered and separated using a Beckman Coulter Allegra 25R refrigerated centrifuge, operated at 12,000 rpm and 15?C for 30 minutes. Random selections of the liquid samples recovered from the centrifuge were tested to ensure that the toluene-insoluble content of the liquid was less than 0.5 wt %. The recovered solid was mixed with the solid recovered from washing the reactor wall and internal fittings. The recovered solids were filtered (using a 0.22 ?m membrane filter) and washed with toluene to remove toluene-soluble materials, followed by drying in a vacuum oven (5 inHg) at 120 ?C for 3 - 4 hours. The coke was defined as the toluene-insoluble solids recovered in each experiment. Both liquids recovered from the centrifugation and also liquid recovered from the cold condenser (light ends) were weighed separately, and the total mass of the liquids were considered as the total recovered liquid of a hydroconversion experiment. The continuously collected gas sample from the exit line, as well as the gas sample collected from the reactor after completion of the experiment, were analyzed by an HP 5890A gas chromatograph. Light gases (C1?C4) and their isomers were separated using a 5 m 68  temperature programmed Porapak Q packed column and quantified with a flame ionization detector (FID). A block diagram of product work-up in the semi-batch reactor experiments is presented in  Appendix E. 3.3.1 Experimental Procedure for Recycle Experiments For the catalyst recycle experiments, the coke-catalyst mixture recovered from a hydroconversion experiment was mixed with about 50 mL of toluene at room temperature and shaken on a rotary shaker overnight. The slurry was then placed in an ultrasonic shaker for 15 minutes and filtered using a 0.22 ?m membrane filter. The solid was subsequently washed with fresh toluene to remove any oil-soluble compounds from the coke. The filtered coke was dried under 5 inHg vacuum at 120 ?C for 4 hours. The dried coke was then ground to a fine powder using an agate pestle and mortar and used as the recycle catalyst. SEM analysis of the coke prior to recycling (Figure  3.5) showed uniform particles with a size distribution from 1 ?m to 30 ?m and confirmed that grinding yields a consistent coke particle size. Great care was taken to ensure that no coke-catalyst mixture was lost during the recovery from the reactor, washing, filtration, drying and grinding steps. Except for a few milligrams of the coke sample which was taken for characterization, the rest of the coke-catalyst mixture was used as the catalyst in the next hydroconversion reaction. In the recycle experiments, the coke-catalyst mixture (prepared as described above) added to the reactor was mixed with the residue oil overnight (for at least 8 hours) at 12.4 MPa of N2 gas and 160 ?C. The reactor was depressurized the next day, purged and then 69  pressurized with H2 gas to the reaction pressure of 13.8 MPa to start the hydroconversion reaction using the recycled catalyst. The coke yield calculation for the recycle experiments was based on the net weight of coke generated (total solid recovered after reaction - the weight of solid coke added to the reactor) divided by the weight of the CLVR feed.       Figure  3.5 SEM micrograph of 4 different coke samples prior to recycling.  70  3.4 Experimental Procedure for Heat-up and Blank (thermal) Experiments A heat-up experiment in the present study refers to an experiment in which the reactor was quenched immediately after the reaction reached the desired reaction temperature. The reaction time for the heat-up experiments is taken as zero compared to the 1-h reaction time in normal hydroconversion experiments. A ?blank? or ?thermal? experiment refers to an experiment in which no catalyst was added to the reactor, but the reaction was carried out under high pressure H2 gas. All other experimental procedures of the thermal experiments were the same as those for the catalytic hydroconversion experiments. 3.5 Experimental Procedure for Generating Data for Kinetic Modeling To generate data for the kinetic modeling of residue -hydroconversion presented in  Chapter 8, 16 different experiments were done. Reaction temperatures of 415 ?C, 430 ?C, 445 ?C and 460 ?C were investigated and at each temperature, 4 different hydroconversion experiments with different reaction times of 0 (heat-up experiment), 30 min, 60 min and 120 min were completed. The product work-up was carried out in the same way as already described for the semi-batch reactor (Section  3.3). Since the sampling of the products from the reactor at different times was not possible (due to high temperature and pressure of the reactor), to generate the kinetic data at different reaction times, separate hydroconversion experiments with variable reaction times of 0, 30, 60 and 120 mins have been done independently. A summary of the operating conditions of the hydroconversion experiments are presented in Table  3.4. 71  Table  3.4 Summary of operating conditions of experiments done for kinetic study. Experiment # Temperature, K Reaction time, min H2 presure, MPa 1 733 0a 13.8 2 30 13.8 3 60 13.8 4 120 13.8     5 718 0a 13.8 6 30 13.8 7 60 13.8 8 120 13.8     9 703 0a 13.8 10 30 13.8 11 60 13.8 12 120 13.8     13 688 0a 13.8 14 30 13.8 15 60 13.8 16 120 13.8 a Reaction time of zero means a heat-up hydroconversion experiment (refer to Section  3.4 for definition of a heat-up experiment). 3.6 Calculation of Yields, Conversions and H2 Uptake Coke and liquid yields were similarly defined as the weight of coke (after washing and drying) and liquid recovered after each hydroconversion experiment divided by the CLVR 72  feed weight, respectively (wt % of feed). Gas yields from both reactors were calculated as the weight of hydrocarbon gas produced during the reaction divided by the CLVR feed weight. H2 conversion and uptake was calculated as follows: ?? ??????????, % =?????????????????????? ??????????? ???                    ( 3.1) ?? ??????, % =?????????????????????? ?????? ???                            ( 3.2) In the both experiments, (hydrogen)in and (hydrogen)out are the moles of H2 in the reactor before and after the reaction. In the semi-batch experiments, (hydrogen)in and (hydrogen)out are the total weight of H2 entering the reactor and total weight of H2 leaving the reactor respectively during the reaction time. Wfeed is the weight of the residue feed in the experiment. The H2S yields for the hydroconversion experiments done in the batch reactor were calculated by dividing the weight of generated H2S (measured by product-gas analysis using TCD) to the weight of CLVR (wt % of feed). As mentioned in Section  3.3, a NaOH-filled scrubber was used to remove the H2S gas from the exit gas flow of the semi-batch reactor. The amount of H2S from the exit gas stream that reacted with the NaOH was measured using a titration method. 0.4 ? 0.5 g of PHP (Sigma-Aldrich, 99.95 %) was dissolved in about 100 mL of de-ionized water and 1-2 drops of phenolphthalein indicator was added to the solution. 10 mL of the NaOH solution collected from the scrubber after the reaction was diluted to a total volume of 100 mL using 73  de-ionized water. A 50 mL burette was filled with the diluted NaOH solution and was used to titrate the standard PHP solution until a color change to pink was observed. The NaOH solution after the reaction was calculated based on the initial NaOH concentration (1 N), PHP solution concentration and the volume of basic solution needed to neutralize the acidic solution of PHP. The Yield of H2S gas was then calculated assuming that the reacted NaOH was used to neutralize the H2S in the exit gas stream. The titration was done in duplicate, and an average of the yield was reported as the H2S yield of the experiment. A sample set of calculations for the H2S yield calcultions is presented in  Appendix C. Liquid phase S removal was calculated by the S content of the feed (residue oil) and the liquid product measured by elemental analysis. 3.6.1 Residue and Toluene-insoluble Organic Residue (TIOR) Conversions The distribution of the different hydrocarbon cuts in the liquid product was determined by high temperature simulated distillation (HTSD) using the ASTM D7169 methodology. The residue in this study was defined as all products having boiling points above 524 ?C. Liquid phase residue conversion was determined by comparison of the residue content of the feed and the residue content of the liquid products as: ?????? ????? ??????? ?????????? % =  (??,????,?)??,?? ???   ( 3.3)  74  To calculate toluene-insoluble residue conversion, the generated coke in the hydroconversion experiment was included in the residue fraction and TIOR conversion was defined as: ???? ?????????? % =  (??,?????,?+???)??,?? ???          ( 3.4)  where WR,F is the weight of residue in the feed, WR,L is the weight of residue in the liquid product, and WC is the net weight of solid coke generated in the experiment. WR,F and WR,L were calculated by multiplying the wt% of the residue fraction in the feed or liquid (determined by the simulated distillation analysis) by the weight of the feed or product liquid from each experiment, respectively. 3.6.2 Asphaltene Content Measurement To measure the asphaltene content (n-C5 asphaltene) of the feed and liquid product, 2 g of sample was dissolved in ~ 2 mL of toluene. Asphaltenes were subsequently precipitated by introducing an excess amount (80 mL) of n-C5 to the solution. The slurry was placed on a rotary shaker and shaken for 5 minutes followed by 30 minutes of resting. The mixing/resting cycle was repeated 5 times. The slurry was then filtered with a 0.22 ?m membrane filter and washed with excess n-C5. Solid asphaltene was dried overnight in an oven at 120 ?C. Asphaltene conversion was then calculated by comparison of the asphaltene content of the feed and liquid product in each experiment. 75  3.7 Instrumental Analyses Used for Catalyst and Coke Characterization Prior to analysis, coke samples were washed with toluene and filtered using a 0.22 ?m membrane filter to remove any toluene-soluble materials. The coke was then dried in a vacuum oven and ground to form a powder as described in Section  3.3.1. 3.7.1 Thermogravimetric Analysis (TGA) TGA is used to quantify the weight loss (in the range of milligrams) of a solid or liquid sample vs. temperature over a wide range of temperatures (50 ?C? 1200 ?C depending on the instrument). The temperature ramp rate of the furnace is usually in the range 10 ? 50 ?C/min, depending on the sample type and application. Experiments can be carried out in inert gas, such as N2, as well as oxidizing gases, such as air. TGA analysis can also be done under constant temperature (isothermal mode) to investigate the rate of decomposition and weight loss in a specific sample [112]. In this study, TGA analysis was done under N2 gas at a flow of 50 mL (STP)/min using a TGA-50 thermogravimetric analyzer (Shimadzu, Japan) and 5?15 mg of coke. The coke was initially heated to 120 ?C at a rate of 5 ?C/min and held for 60 min to remove moisture, followed by heating to 900 ?C at a rate of 5 ?C/min; the final temperature being held for a further 120 min. The weight change of the sample during the experiment was recorded online by running TA-60WS Collection Monitor software. The experimental data were further processed for calculation of weight loss, etc. by TA60 software provided by Shimadzu. 76  3.7.2 Elemental (CHNS) Analysis Elemental analysis was done using a 2400 Perkin Elmer CHNS/O analyzer. After calibration of the unit with the standard solid sample, approximately 1.5 - 2.5 mg of coke was used for the elemental analysis. A combustion and reduction temperature of 975 ?C and 500 ?C, respectively were used in the combustion and reduction columns of the unit. Combustion gases from the combustion tube were well mixed in a mixing chamber before entering a packed column. The separated gases were then detected using a thermal conductivity detector (TCD) using He as reference gas. 3.7.3 Diffuse Reflectance Infrared Fourier Transform (DRIFT) Spectroscopy Infrared spectroscopy is a non-destructive technique that can provide useful information on the chemical bonding in materials. DRIFTS is usually used to study powders (such as catalysts) and reactions on powders and rough surface materials. DRIFT spectrometers measure light diffused by a sample dispersed in a diluent such as KBr or KCl salts using several mirrors. Comparing the diffuse reflection obtained from KBr or KCl with the diffuse reflection of diluent plus the solid sample, yields a graph resembling the transmission spectrum [112]. To improve the spectrum, Kubelka-Munk?s transformation is usually used. Details on Kubelka-Munk?s transformation are presented in  Appendix J. In the present study, DRIFT spectra were measured and recorded using a Nicolet 4700 FT-IR unit (Thermo Electron Corporation), OMNIC 7.2 software, Harrick DRIFT accessories and a liquid-nitrogen cooled mercury cadmium telluride (MCT) photoconductive detector at ambient temperature. All the coke samples were scanned in the range of 650 ~ 77  4000 cm-1 at a resolution of 4 cm-1 after 50 scans. The spectra were plotted assuming the model of Kubelka-Munk for diffuse reflectance, with the Kubelka-Munk function being analogous to absorbance. To compare different spectra in the present study, several spectra are shown in one plot by adding a constant to the values of Kubelka-Munk function. In some graphs, the stacking was done automatically by the scientific graphic and data analysis software.  To prepare coke samples for DRIFT analysis, the coke was diluted to 3-5 wt % in KBr and then placed in the electric muller to be ground for 60 seconds. This standard procedure was used for all coke samples to make sure the diluted, and ground coke samples had similar particle size prior to analysis. Three scanning electron microscope (SEM) pictures from three different coke samples prepared for DRIFT analysis shown in Figure  3.6 demonstrates that all the coke samples had very similar particle sizes.     78          Figure  3.6 SEM micrographs of three different coke samples before DRIFT analysis. (A) 3 wt % FeS used as the catalyst, (B) 5th recycle of 600 ppm Mo using Mo-micelle precursor and (C) a thermal experiment. All the images have the same scale as (A).  3.7.4 X-ray Diffraction (XRD) Analysis X-ray diffraction (XRD) is a non-destructive technique used to identify the crystalline phases present in materials. In addition to the crystalline phase, XRD is capable of measuring the structural properties of crystalline phases (crystalline size, phase composition, preferred orientation, etc.) of these phases [113]. 79  Sample preparation for XRD analysis was the same as that described in Section  3.3.1. Solid product (coke) was characterized by X-ray diffraction (XRD) after each recycle step. The XRD pattern of the recovered coke was obtained with a Rigaku Multiflex diffractometer using Cu K? radiation (? = 1.5406 ?, 40 kV and 20 mA), a scan range of 2? from 10? to 90? with a step size of 0.04? and a scan rate of 2?/ min. 3.7.5 Solid State 13C Nuclear Magnetic Resonance (NMR) Solid-state NMR, which is a relatively new compared to other spectroscopic techniques, is a non-destructive technique mainly used for qualitative and quantitative analysis of crystalline phases. NMR is a selective technique and can be done for any specific element independently. High resolution NMR was originally developed for analysis of solutions. However, recent advances and development made the technique a powerful technique for solid analysis [112]. Due to these advantages, solid-state NMR is increasingly used in the structural analysis of polymers, ceramics and glasses, composites, catalysts, coal and surfaces [114].  In this study, 13C cross-polarization magic angle sample spinning (13C CPMAS) spectra were recorded on a VarianInova 400, as well as a BrukerAvance 500 MHz spectrometer in the Department of Chemistry at the University of British Columbia. The units were operated at 13C frequencies of 100.521 MHz and 125.691 MHz, respectively. All samples were spun at a spinning frequency of 9000?0.002 kHz. In order to obtain the quantitative data, single pulse excitement with a delay time of d1 = 5 s was used. The number 80  of scans for samples was either 2048 or 8192. The recorded spectra were analyzed using DMFIT 2011 software [115]. 3.7.6 Scanning Electron Microscopy (SEM) Due to reasonable cost and a wide range of information that can be obtained from SEM, it is usually the preferred starting tool for investigating the surface of materials. The SEM provides a highly magnified image of the surface of a material that is very similar to what one would expect if one could actually observe the surface visually. The difference between SEM and an optical microscope is in the excitation source used in the instrument. Unlike the optical microscope which uses a light source, SEM uses secondary electrons to form the image of the sample. The resolution of SEM can be as high as few nm and it can operate at variable and adjustable magnifications of about 10 x ~ 300,000 x. Since in most cases the conductivity of the sample is beneficial for sample analysis, preparation of the samples is relatively easy in SEM analysis. If the sample is not conductive, a sample coating with conductive materials is required [116]. In this study, a 120 keV Hitachi S-3000N SEM equipped with a light element Energy-dispersive X-ray (EDX) detector was used for SEM analysis. EDX and EDX-mapping were used to measure the local and area semi-surface (detection depth of 1~2 ?m) composition at a point or small area and also the distribution of certain elements within the coke sample. The unit is capable of detecting elements with atomic numbers in the range of 4 to 98. However, the analytical precision of the unit decreases when very low concentrations (< 0.5 wt %) of elements are present in the sample. 81  3.7.7 Transmission Electron Microscopy (TEM) and High Resolution TEM (HRTEM) Due to certain advantages of TEM analysis (high point-to-point resolution of 0.2 nm in some instruments), TEM has become a popular technique in the last five decades to analyze the physical structure of different solid materials. In TEM, a focused electron beam passes through a very thinly-prepared sample (less than 200 nm) and the undeflected and deflected electrons are collected using a series of lenses and detectors (or a camera in modern TEM instruments). The resolution of TEM can vary from as little as 50 times to as much as a factor of 106. This high magnification range is generated by the small wavelength of the incident electrons, and is the key to the unique capabilities associated with TEM analysis [117]. HRTEM is a mode of TEM analysis in which high magnification and resolution allow study of crystalline structure of a sample. Displacing measurement of different crystalline planes makes qualitative analysis of different crystalline structures feasible in a sample. In the present study, a Tecnai G2 TEM with field emission of 200 keV and point/lattice resolution of 0.24/0.14 nm was used for high resolution TEM analysis of coke samples. To prepare the coke samples for TEM analysis, washed and dried coke (refer to Section  3.3.1) was ground to a fine powder using an agate mortar and pestle. The solid was then dispersed in ethanol using an ultrasonic shaker. A droplet of the emulsion was then placed on a 200 mesh copper grid coated with formvar carbon and was left to completely dry before the analysis. 82  3.7.8 High Temperature Simulated Distillation (SIMDIS) The determination of boiling point distribution of crude oils and vacuum residues provides important information for refinery operations as to the potential mass percent yield of products. This information is widely used to optimize refining processes. Simulated distillation, which is a technique designed based on gas chromatography through hardware and software technology, simulates the actual physical distillation of different petroleum fractions. A non-polar chromatographic column used in simulated distillation separates the hydrocarbons in order of their boiling points. The retention times of different hydrocarbons are correlated with the carbon number of the hydrocarbons. This is done by comparing the retention times to the retention times of different known hydrocarbons in a standard mixture of n-alkanes (i.e. polywax), covering the boiling range expected in the sample. Results are reported as a correlation between the boiling points and the percentages of the sample eluted from the column [112,118]. In the present study, a Shimadzu GC-2010 SIMDIS analyzer equipped with required accessories to perform ASTM D-7169, which can determine the boiling point of compounds up to 720 ?C, was used. A 5 m MTX?- 1HT SIMDIS column (provided by Restek) with 0.52 mm ID and 0.20 ?m phase thickness was used. The initial column temperature of -20 ?C, heating ramp of 15 ?C/min and holding time of 5.33 mins at 425 ?C was used as the column temperature program. An FID detector operated at 430 ?C detected the eluted hydrocarbons from the column. An on-column injector with variable pressure and temperature (initial temperature of 65 ?C, heating rate of 15 ?C and final temperature of 425 ?C held for 11 mins) 83  was used to introduce the injected sample to the column. An injection volume of 0.2 ?m was used for all the injections, and CS2 (Sigma-Aldrich) was used as solvent to dilute the oil samples. A mixture of Polywax 655 (provided by AccuStandard) and ASTD D2887 calibration mixture (provided by Restek) were used as calibration mixtures to generate the calibration curve. 3.7.9 X-ray Photoelectron Spectroscopy (XPS) Selectivity of the XPS analysis toward different elements as well as the quantitative nature of the technique has made XPS one of the most widely used techniques in the surface analysis field. XPS can detect all elements except hydrogen and He in gaseous, liquid or solid states. However, the technique is mainly used for solid analysis. The achievable depth of analysis in the sample varies from the top 2 atomic layers to 15-20 layers. Among all electron or ion spectroscopy techniques, XPS is the least destructive technique [114].  In this study, all the XPS analyses were done using a Leybold Max200 X-ray photoelectron spectrometer with an Al K? X-ray source generated at 15 kV and 20 mA. Pass energy of 192 eV for the survey scan, and 48 eV for the narrow scan were used for analysis. Deconvolution of the XPS profiles was performed using XPSPEAK 4.1 software. 3.7.10 Surface Area Analysis by the Brunauer, Emmett and Teller (BET) Method  The BET method measures the amount of a gas adsorbed on the surface of a solid (v) in units of cm3 per gram of the solid sample. Since the amount of adsorbed gas is usually measured at a constant temperature (boiling point of a cryogenic liquid such as liquid N2) as 84  a function of pressure p (in mm Hg), the measurement is called the adsorption isotherm. The amount of adsorbed gas on the surface of a sample is usually calculated by monitoring the pressure drop in the sample cell. This is calculated using the BET equation: ??(????)=????+ ???????????      ( 3.5) In the BET equation, v is the volume of gas adsorbed in cm3 (STP) per gram of solid at a pressure p and ?0 is the saturation pressure of the adsorbate at the adsorption temperature. The constant c is related to the heat of adsorption, and vm is the monolayer volume. From the plot, one determines the slope and the intercept, from which one can calculate vm and c. Knowing vm and the cross-section area of the adsorbate gas molecule, one can calculate the surface area of a porous solid. It should be noted that when isotherms are measured close to the boiling point of the adsorbed gas, e.g., for N2 adsorbed at 77 K, the range of validity of the BET equation is in the relative pressure (p/p0) range between 0.05 and 0.35. In general, for best accuracy, at least four points should be measured in this region. In the present study, the BET surface areas of the recovered coke-catalyst were determined from N2 adsorption-desorption isotherms measured at 77 K using a Micromeritics ASAP 2020 analyzer. Samples were degassed at 523 K under vacuum (500 ?m Hg) for 8 h before being analyzed. Eight N2 uptake measurements, made in the relative pressure range of 0.06 - 0.20, were used to calculate the BET surface area. 85  3.7.11 Particle Size Analysis using Dynamic Light Scattering (DLS) Dynamic Light Scattering is also known as Photon Correlation Spectroscopy. This technique is one of the most popular methods used to determine the size of particles. Using a monochromatic light beam, such as a laser, onto a solution with spherical particles in Brownian motion, the DLS technique determines the size of particles in motion by measuring the change in the wavelength of the light before and after colliding with the particle. DLS can also provide information on the size distribution of particles in a sample [119]. In this study, the diameter of the prepared reversed micelles was determined using dynamic light scattering (DLS). A Scitech Instruments ST-100 variable angle light scattering system using the 514 nm line of an Ar ion laser was utilized to determine the reversed micelle size distribution in the emulsion.        86  Chapter 4  Preliminary Catalyst Screening1 4.1 Introduction The sulfided, unsupported metal catalysts used in residue oil hydroconversion are typically generated in situ, with the catalyst precursor added to the residue oil as a water-soluble salt, oil-soluble metal complex or as a finely powdered solid. Sulfided transition metals (Mo, Ni, Co, W, Cr, V, Fe, Cu and Zn) are active for hydroconversion among which MoS2 is the most commonly used catalyst [6]. Under hydroconversion conditions, thermal decomposition and sulfidation of the catalyst precursors yields the active (sulfided) catalyst.  Although the performance of different metal sulfides introduced in the form of oil-soluble precursors for hydroconversion of heavy feeds has been investigated [14,19,21,75,76], there are fewer reports on the use of water-soluble precursors.   Most studies suggest a better dispersion and smaller particle size in the case of oil-soluble catalyst precursors compared to water-soluble precursors [2]. Liu et al. [79] used different water-soluble precursors for residue hydroconversion and characterized the catalysts recovered after the reaction. Due to the complexity of the characterization and the presence of large amounts of coke, a relatively 1 A version of this chapter has been published previously: Hooman Rezaei, Xuebin Liu, Shahrzad Jooya Ardakani, Kevin J. Smith and Maureen Bricker, A study of Cold Lake Vacuum Residue hydroconversion in batch and semi-batch reactors using unsupported MoS2 catalysts, Catalysis Today, 150 (2010), 244-254. 87                                                   light feed was used to minimize coke formation. The authors showed that at the reaction conditions, Mo, Ni and Fe were in sulfided form. XRD analysis of the recovered catalysts showed that the catalysts were completely crystalline.  An alternative approach to using water-soluble catalyst precursors is to prepare the water-soluble precursor in reversed micelles such that the water pool containing the metal salt solution is stabilized in an organic solvent, such as n-hexane, using an appropriate surfactant. This approach provides for the possibility of better control of the catalyst size and better dispersion of the catalyst precursor in the residue oil, compared to direct addition of the water-soluble precursor to the oil.  Although using the catalyst once through has the advantage of minimizing deactivation effects during slurry phase hydroconversion, the cost of catalyst, especially in the case of more active and expensive metals like Mo, is prohibitive (Section  2.5).  Clearly, recovery, regeneration and recycle of the spent catalyst would be needed to make slurry-phase processes economically viable. An understanding of the state of the catalyst after the hydroconversion reaction and the interaction between the catalyst and the coke is needed before an appropriate catalyst recovery and recycle strategy can be implemented.      In the present chapter, different types of catalysts were assessed in the batch reactor to identify the most active catalyst for further studies under high residue conversion conditions in the semi-batch reactor. Furthermore, the effectiveness of preparing water-soluble catalyst precursors in reversed micelles for use in hydroconversion of Cold Lake residue oil is demonstrated. After catalyst screening in the batch reactor, catalyst activity measurements 88  are compared using both batch and semi-batch reactors at low temperature (415 ?C). A report on the catalyst after reaction and its interaction with the generated coke is also presented as a preliminary study of coke-catalyst mixture recyclability. 4.2 Catalyst Activity Measurements 4.2.1 Preliminary Screening of Catalysts The initial catalyst testing was done in the batch autoclave reactor. To ensure a better control of the reaction and better observation of differences between different catalyst precursors and concentrations, these experiments were conducted at a relatively low temperature (415 ?C) that resulted in low residue conversions.  Nevertheless, the results were useful in an initial screening of potential catalysts to be used in later experiments using recycled catalysts. These initial experiments were also used to establish a consistent product work up procedure and a repeatable experimental methodology. Accordingly, the coke analysis procedure was modified from a purely filtration method to a centrifuge/filtration method, as detailed in Section  3.2 and  Appendix D. Although repeat experiments were done to confirm the repeatability of the catalyst screening experiments (presented in  Appendix G), no detailed statistical analysis of the repeated experiments was performed. This was because the results obtained from the catalyst screening experiments were considered preliminary and were mainly used to identify the most active catalyst precursor for further study in the semi-batch reactor and under high residue conversion conditions (high reaction temperature and pressure). 89  Initial experiments were aimed at a comparison of different dispersed catalysts and thermal conversion at the batch reactor process conditions. In all cases, the same CLVR feed was used, the properties of which are given in Table  3.3. The metal catalysts were prepared either in micelles or introduced directly to the oil as AHM or iron sulphate (FeSO4.6H2O) (Refer to Section  3.1 for more details on catalyst preparation).  In the latter case, an overnight thermal treatment at 160 ?C under N2 was used to pre-age the catalyst. Catalyst concentration in all the experiments reported here is presented as ppm of Mo in the CLVR, although Mo precursors yield the active form of the catalyst in the reactor (MoS2) by in situ thermal decomposition and sulfidation. Figure  4.1 compares the results obtained for a series of single metal catalysts (600 ppm of metal concentration), whereas Figure  4.2 compares the bimetal catalysts. The blank experiment refers to the experiment in which no catalyst was added to the reactor. Detailed results for each of the experiments are provided in  Appendix G including data from repeat experiments that showed good repeatability for the measured H2 uptake (0.91 ? 0.05 for experiments using 600 ppm in the form of micelle) or conversion, the coke yield and the residue conversion.  Note that for the entire batch reactor data reported in this section, the residue conversion was reported without accounting for the residue converted to toluene insoluble material. Figure  4.1 shows that the micelle prepared Mo (Mo-M) was significantly more active than the Mo introduced as AHM (Mo-W) and similarly, the micelle prepared Fe (Fe-M) was significantly more active than the iron sulphate (Fe-S).  In both cases the micelle catalysts had higher H2 conversion and lower coke yield than the Mo and Fe catalysts prepared directly from the metal salts. 90   Figure  4.1 CLVR hydroconversion using Mo (Mo-M) and Fe (Fe-M) catalysts prepared in reversed micelles compared to AHM (Mo-W) and iron sulphate (Fe-S) catalysts. T = 415?C, 5.5 MPa initial H2 pressure and 1 hour reaction time.  Figure  4.2 shows that at the batch reactor test conditions, no benefit was obtained by using bimetal micelle catalysts, and consequently, these catalysts were not investigated further. The low activity of the bi-metallic catalysts compared to the mono-metallic Mo-micelle precursor could be due to the fact that, as reported by others [2], Mo has the highest activity in residue hydroconversion followed by Ni and Fe. The Mo concentration in a bi-metallic precursor was half that of the mono-metallic Mo-micelle precursor and this is probably why the beneficial effect of promoter was not observed when a bi-mettallic precursor was used. Blank Mo-M Mo-W Fe-M Fe-S010203040506070 H2 conv  Residue conv  Coke yield  Gas yieldwt% 91   Figure  4.2 CLVR hydroconversion using Mo-Co (MoCo-M) and Mo-Fe (MoFe-M) catalysts prepared in reversed micelles compared to Mo-M and Fe-M catalysts. T = 415?C, 5.5 MPa initial H2 pressure and 1 hour reaction time.  A comparison of Fe and Mo catalysts over a range of concentrations was also made and the results obtained are summarized in Figure  4.3. The effect of catalyst concentration in the feed oil is demonstrated by the H2 conversion data of Figure  4.3-A, showing a clear increase as the Fe or Mo concentration in the oil increased, with significantly higher conversion achieved with the Mo-micelle catalyst compared to the Fe-micelle. The fact that the H2 conversion did not increase linearly with catalyst concentration reflects the fact that at high H2 conversion, H2 supply was limited in the batch reactor. Coke yield decreased initially with increased metal content but then increased as the Mo or Fe content increased above Mo-M (MoCo)-M (MoFe)-M Fe-M010203040506070wt%  H2 conv  Residue conv  Coke yield  Gas yield92  about 600 ppm metal. The data of Figure  4.3 also report results from one experiment in which 3 wt % Fe was added as iron sulphate, to mimic conditions relevant to the CANMET slurry hydroconversion process [120,121]. Clearly, at the low conversion conditions of the batch reactor tests, this catalyst was not as effective as the Mo catalyst.  The results from these preliminary, low conversion experiments demonstrated that Mo-micelle and Fe-micelle precursors had lower coke yield and higher H2 uptake than Mo and Fe catalysts derived from the dispersed metal salts. Furthermore, the bi-metal (Mo-Co and Mo-Fe) micelle precursors did not show significant benefit compared to the single metal catalysts. Consequently, it was decided to focus on the Mo catalysts in further work. 93           Figure  4.3 CLVR hydroconversion using Mo (MoM) and Fe (FeM) catalysts prepared in reversed micelles over a range of metal concentrations in the CLVR.  Data for catalysts prepared in situ from AHM (MoW) and iron sulphate (FeS) as well as the results from a thermal experiment are included for comparison.  0 500 1000 1500 3000020304050607080   Thermal MoM MoWFeM  FeSH2 Conversion (%)Fe or Mo, ppmA0 500 1000 1500 3000002468101214  ThermalMoM  MoWFeM  FeSCoke Yield (wt%)Fe or Mo, ppmB0 500 1000 1500 30000404550556065ThermalMoM  MoWFeM  FeS  Residue Conversion (wt%)Fe or Mo, ppmC94  4.2.2 Comparison of Mo-micelle and AHM precursors measured in the Batch and Semi-batch Reactors There were three categories of experiments done in this section to assess the catalysts: (i) experiments done in the batch reactor at relatively low temperature and pressure, (ii) experiments done in the semi-batch reactor utilizing low temperature and high H2 pressure and (iii) experiments done in the semi-batch reactor at high temperature and pressure. Table  4.1 summarizes the conditions of the experiments investigated in the present study. Table  4.1 Summary of the experimental conditions in the batch and semi-batch reactors. Catalyst precursor Catalyst concentration, ppm Batch reactor a Semi-batch reactor a PH2, initial = 5.5 MPa PH2 = 13.8 MPa b T = 415 oC T = 415 oC T = 445 oC Mo micelle 100 ? -- ? 300 ? -- ? 600 ? ? ? 900 ? -- -- 1800 -- -- ? AHM 600 ? ? ? 1800 -- -- ? Thermal experiment 0 ? -- ? a Reaction time for experiments in both reactors was 1 h. b H2 flowrate: 900 mL(STP)/min.  95  Batch reactor data measured at low temperature (415 oC) and pressure (5.5 MPa) are discussed first. The low temperature data emphasize differences in catalyst activities since the contribution from thermal cracking is low at 415 oC (as evidenced by residue conversions < 60 wt %). Figure  4.4 shows the H2 conversion and consumption as a function of the Mo concentration for catalysts prepared from Mo-micelle precursors.  Figure  4.4 Comparison of H2 conversion and consumption in a series of experiments in the batch reactor using Mo-micelle precursor with different Mo concentrations. T = 415 oC, PH2, initial ~ 5.5 MPa and reaction time = 1h.  96  H2 conversion and consumption both increased with increasing Mo concentration in the oil. Note, however, that the H2 conversion did not increase linearly with catalyst concentration because at high H2 conversion, the H2 supply was limited in the batch reactor. At 100 ppm Mo, about 55 wt % of the H2 in the reactor was converted after 1h. This rapid H2 consumption decreased both the H2 concentration in the gas phase in contact with the residue oil in the reactor and the system total pressure. The reduced pressure decreases H2 solubility in the residue oil and the mass transfer rate between the gas and liquid phase, both resulting in reduced H2 conversion. Another explanation for the trend shown in Figure  4.4 is that at 415 oC, a relatively low temperature for hydroconversion and thermal cracking, the rate of reaction is limited by the concentration of hydrocarbon free radicals, so that further increases in H radical generation (due to increased Mo concentration) does not result in a significant increase in H2 consumption. In this case, increasing the reaction temperature to increase the rate of hydrocarbon free radical generation, while catalyst concentration remains fixed at a relatively high value (i.e. 600 ppm Mo), will result in increased H2 consumption. Although not done in the present study, Tye and Smith [18] reported a 32.4 % increase in the H2 consumption as the reaction temperature increased from 415 oC to 430 oC at the conditions of their study (initial H2 pressure of 3.5 MPa, reaction time of 1 h and 600 ppm Mo using exfoliated MoS2).  A comparison of the H2 conversion and H2 consumption using MoS2 prepared from the reversed micelles and AHM precursors of the present study showed that at 600 ppm Mo, the catalyst prepared from the reversed micelle precursors had higher H2 conversion and consumption than the water-soluble AHM catalyst precursor (Figure  4.5). Note that although 97  both catalysts were tested under the same operating conditions (T = 415 oC, PH2, initial ~ 5.5 MPa and reaction time = 1 h), the reactor pressure was higher with the AHM precursor than the Mo micelle at all reaction times due to lower H2 conversion. This suggests that even at lower H2 pressures, the Mo micelle-based catalyst had better performance than the AHM in terms of H2 consumption.  Figure  4.5 Comparison of H2 conversion and consumption in two experiments in the batch reactor using 600 ppm Mo in the form of micelle and water-soluble AHM precursors. T = 415 oC, PH2, initial ~ 5.5 MPa and reaction time = 1h.  Since H capping of free radicals generated by thermal hydrocarbon cracking reactions is thought to be the main mechanism of coke prevention in hydroconversion by some investigators [21], the higher H2 conversion achieved using the Mo-micelle catalyst precursor 98  versus the AHM precursor would suggest that a significant difference in the coke yields from these two catalysts should also be observed. Figure  4.6 shows that the coke suppression in the case of AHM catalyst precursor was very poor. At a concentration of 600 ppm Mo in the feed, the coke yield was much higher with the AHM catalyst precursor than the Mo-micelle precursor. Changes in coke yield over a range of Mo concentrations in the case of Mo-micelle catalyst were very small for experiments in the batch reactor at low temperature (415 oC). Since both catalysts were prepared using water-soluble precursors, the noticeable difference in catalyst activity shows the distinct advantage of using the reversed micelles to disperse the precursor. One likely reason for the poor activity of the AHM precursor compared to the Mo-micelle catalyst precursor is agglomeration of Mo particles due to rapid evaporation of water from poorly dispersed AHM during heat up of the reactor. The well-dispersed Mo salt, stabilized in the reversed micelle, was much less likely to agglomerate during the heat up period. Also note that in the present study, the AHM catalyst may not have been utilized to its maximum potential. Catalytic properties of ammonium molybdates can be significantly increased by pre-sulfidation of AHM using H2S at low, moderate and high temperatures [122]. 99   Figure  4.6 Coke yield in a series of experiments in the batch reactor using different concentrations of Mo-micelle precursor and 600 ppm of Mo using AHM precursor. T = 415 oC, PH2, initial ~ 5.5 MPa and reaction time = 1h.  The increased coke yield at a Mo concentration of 900 ppm in the feed, shown in Figure  4.6, is consistent with results from experiments done in the semi-batch reactor at high concentration of Mo in the feed (1800 ppm) described below. Furthermore, similar increases in coke yield were observed in the experiments presented in  Chapter 5 and were also reported by Tye and Smith [18] and Del Bianco et al. [21]. Using exfoliated MoS2 and Mo naphthenate, respectively, these authors reported a small increase in coke yield as the Mo concentration increased from 600 ppm to 900 ppm [18] and from 200 ppm to 5000 ppm [21]. Table  4.2 compares results from the present study, using the batch reactor, to literature data reported at operating conditions similar to those of the present study. Although Table  4.2 100  shows higher activity of MoS2 prepared from a water-soluble precursor prepared in reversed micelles, compared to exfoliated MoS2, water-soluble AHM and oil-soluble Mo naphthenate catalyst precursors, not all studies used the same residue oil as reactant. The apparent advantage of the Mo-micelle catalyst precursor over the exfoliated MoS2 is more significant considering that the Mo-micelle precursor was evaluated using a feed with a low H/C atomic ratio (1.43) that has a much higher tendency for coke formation, compared to the feed used by Tye and Smith [18], which had a H/C atomic ratio of 1.53. Table  4.2 also allows comparison of results from experiments using 900 ppm Mo catalyst prepared from the Mo-micelle precursor (this work) and a Mo naphthenate precursor [21]. The data suggest that the Mo-micelle precursor had almost the same H2 consumption but higher coke yield than the Mo naphthenate at a Mo concentration of 1000 ppm in the feed. However, note that Del Bianco et al. [21] utilized a much higher H2 pressure (9 MPa) and a lower temperature (410 oC) compared to that used in the batch experiments of the present study (5.5 MPa and 415 oC), further evidence of the efficiency of the Mo-micelle catalyst precursor. 101  Table  4.2 Result of CLVR hydroconversion experiments in the batch reactor using dispersed catalysts compared to literature data. Catalyst type, Mo concentration H2 conversion H2 consumption Coke yield Liquid yield Gas yield Sulfur removal Reference ppm wt % wt % on feed wt % wt % wt % wt %  Thermal a        0 22.0 0.3 8.6 84.3 3.0 35.0 This study Mo micelle a        100 54.8 0.7 1.2 96.7 2.2 35.0 This study 300 64.0 0.8 1.4 97.1 2.1 31.7 This study 600 69.4 0.9 1.2 98.9 1.9 38.3 This study 900 73.6 1.0 1.6 101.4 2.0 41.7 This study AHM a        600 28.2 0.4 8.2 87.1 3.2 26.7 This study Exfoliated MoS2 b       360 27.0 N/A 1.5 95.1 3.4 18.8 [18] 600 56.8 N/A 1.1 96.8 2.1 28.1 [18] 900 73.4 N/A 1.8 96.4 1.8 26.9 [18] Mo naphthenate c       1000 N/A 1.2 0.8 N/A N/A N/A [42] a H2 Pressure = 5.5 MPa, temperature = 415 oC, reaction time = 1 hr, atomic H/C ratio of the feed = 1.43. b H2 Pressure = 3.5 MPa, temperature = 415 oC, reaction time = 1 hr, atomic H/C ratio of the feed = 1.58. c H2 Pressure = 9.0 MPa, temperature = 410 oC, reaction time = 1 hr, atomic H/C ratio of the feed = 1.41.     102  Residue conversion using the Mo-micelle catalyst precursors over a range of Mo concentrations showed no improvement with increased catalyst concentration (Figure  4.7). In addition, the Mo micelle at the highest Mo concentration had a lower residue conversion compared to the thermal experiment and the experiment using 600 ppm Mo derived from (AHM). Noting that the latter two experiments had the highest coke yield among all the experiments, it is concluded that the main reason for the high residue conversion in these systems is a higher rejection of residue in the form of coke and a higher gas yield, at the expense of more valuable liquid product fractions (Table  4.2).  Figure  4.7 Comparison of residue conversion in the experiments done in the batch reactor using different Mo precursors and concentrations with thermal experiment (M: Micelle precursor). T = 415 oC, PH2, initial ~ 5.5 MPa and reaction time = 1h. 103  The batch reactor experiments of the present study were limited by H2 supply and low temperature and consequently, residue conversions were low (< 60 wt %). To investigate the effect of H2 supply and reaction temperature, experiments were done in a semi-batch reaction system at more severe conditions (Table  4.1). To investigate the effect of H2 supply and pressure on the catalytic activity, two experiments using 600 ppm Mo prepared via reversed micelles and water-soluble AHM precursors, were done at 13.8 MPa H2 and a H2 flowrate of 900 mL(STP)/min. Other operating conditions were unchanged. Residue conversion in the experiments using Mo micelle and AHM precursors both increased in the semi-batch reactor compared to the batch reactor. The residue conversion increased from 54.4 wt % to 72.7 wt % in the case of the AHM precursor, due to both increased coke yield and H2 pressure. In the case of the Mo-micelle precursor, the residue conversion increased from 44.5 wt % to 55.6 wt %, a consequence of increased H2 pressure, since the coke yield decreased from 1.2 wt % to 1.05 wt %. Sulfur conversion also significantly increased at the higher H2 pressure in the semi-batch reactor compared to the experiments at lower H2 pressure in the batch reactor. These results are summarized in Figure  4.8. 104   Figure  4.8 Comparison of coke and liquid yields, S conversion and residue conversion in the batch and semi-batch reactors using Mo micelle and catalyst prepared in situ using AHM. Reaction conditions: batch- T = 415 oC, PH2, initial ~ 5.5 MPa and 1 h reaction time; semi-batch- T = 415 oC, PH2 ~ 13.8 MPa at 900 mL(STP)/min and 1 h reaction.  Figure  4.9 summarizes the HTSD results from the experiments using Mo micelle and AHM as catalyst precursors in the batch and semi-batch reactors. The results reveal that the product distribution in the liquid in the case of the AHM was shifted to lighter products compared to the micelle. The results also showed reasonable agreement between the product quality in the batch and semi-batch hydroconversion of CLVR, although differences ascribed 105  to different heat up rates (due to different reactor geometries and heater types) were identified.   Figure  4.9 Comparison of different hydrocarbon cuts in liquid products from experiments in the batch and semi-batch reactors using Mo micelle and catalyst prepared in situ using AHM. Reaction conditions: batch- T = 415 oC, PH2, initial ~ 5.5 MPa and 1 h reaction time; semi-batch- T = 415 oC, PH2 ~ 13.8 MPa at 900 mL(STP)/min and 1 h reaction time.  An investigation of the effect of temperature on the catalyst activity and hydroconversion products was completed by conducting a series of experiments in the semi-batch reactor utilizing both high temperature (445 oC) and excess H2 (13.8 MPa and 900 mL(STP)/min). Mo was added at concentrations of 600 and 1800 ppm Mo in the feed using 106  the AHM precursor and 100, 300, 600, 1800 ppm Mo in the feed using the reversed micelle precursor.  A thermal experiment in the presence of H2 was also completed for comparison. At higher temperatures, the rate of cracking reactions, which are mostly thermal and sensitive to temperature [123], would be expected to increase significantly, resulting in more hydrocarbon free radicals being produced. A comparison of the H2 consumption in these experiments using different concentrations of Mo is presented in Figure  4.10. In the case of the Mo-micelle catalyst precursor, H2 consumption increased with Mo concentration much more significantly than was seen in the batch reactor (Figure  4.4). In the semi-batch reactor, the presence of an abundant amount of H2 at high pressure and rapid capping of free radicals at the higher reaction temperature likely occurred [21]. Consistent with the trend reported in the batch reactor, the AHM had a much lower H2 consumption compared to the Mo-micelle precursor even at high concentrations. Since the active form of the catalyst in both cases was MoS2, the poor catalyst activity in terms of H2 consumption when using AHM precursor is ascribed to low surface area, a result of agglomeration of catalyst particles and poor dispersion of MoS2 in the reactor. Although H2 consumption with the AHM precursor was much less than the Mo-micelle precursor, increased temperature and pressure in the semi-batch reactor increased the H2 consumption of both AHM and Mo-micelle precursors compared to the batch reactor.  107   Figure  4.10 Comparison of H2 consumption in experiments using different concentrations and types of catalyst in the semi-batch reactor. T = 445 oC, PH2 ~ 13.8 MPa at 900 mL(STP)/min and 1 h reaction time.  Figure  4.11 shows that the coke yield decreased in the semi-batch reactor with increased Mo concentration using AHM or Mo-micelle as the precursor. However, consistent with the results from the batch and semi-batch reactors at 415 oC (Table  4.2), the AHM catalyst precursor showed much lower activity in terms of coke suppression compared to the Mo-micelle catalyst (Figure  4.11). Although the coke yield in the case of the Mo-micelle precursor was higher compared to experiments at 415 oC, the catalyst showed good activity for coke suppression except at a Mo concentration of 100 ppm. Clearly, with a significant increase in the rate of cracking due to increased temperature, 100 ppm of well dispersed Mo catalyst did not transfer sufficient hydrogen to hydrogenate all the free radicals. With increased Mo concentration from 100 ppm to 300 ppm and 600 ppm, the hydrogenation 108  reaction rate surpassed the rate of condensation and coke formation reactions, and this decreased the coke yield from 4.74 wt % to 1.81 wt % and 1.83 wt %, respectively. Similar to the results in the batch reactor, with further increases in Mo-micelle catalyst concentration (to 1800 ppm), a slight increase in coke yield was observed.   Figure  4.11 Coke yield comparison between experiments using different catalyst types and concentrations in the semi-batch reactor. T = 445 oC, PH2 ~ 13.8 MPa at 900 mL(STP)/min and 1 h reaction time.  Using 600 ppm Mo in the semi-batch reactor, the liquid yield increased from 68.6 wt % in the case of AHM-based catalyst to 90.0 wt % by using the micelle-based catalyst. At a Mo concentration of 1800 ppm the increase in liquid yield was from 73.1 wt % to 88.8 wt %. This was a consequence of higher coke and gas yields in the case of the AHM precursor compared to the Mo micelle. The results presented in Table  4.3, show that 600 ppm Mo 109  derived from the micelle catalyst precursor gave the best results for hydroconversion of CLVR. Although H2 consumption, coke yield and gas yield using 300 ppm and 600 ppm Mo prepared from the reversed micelle precursors were very similar, 600 ppm Mo gave higher liquid yield and particularly higher S conversion. Table  4.3 Results of CLVR hydroconversion experiments using dispersed catalysts in the semi-batch reactor. Catalyst type, Mo concentration H2 conversion H2 consumption Coke yield Liquid yield Gas yield Liquid phase S removal ppm wt % wt % on feed wt % wt % wt % wt % Thermal a 0 12.8 1.6 21.2 64.2 9.1 N/A Mo micelle a       100 N/A N/A 4.8 81.4 7.4 55.2 300 15.6 2.0 1.8 87.5 5.0 53.5 600 16.2 2.1 1.8 90.0 5.6 67.3 1800 18.8 2.4 2.9 88.8 5.9 N/A AHM a       600 13.0 1.6 16.6 68.6 7.6 N/A 1800 13.7 1.7 13.5 73.1 7.7 N/A Experimental conditions in batch reactors: a H2 Pressure = 13.8 MPa, temperature = 445 oC, reaction time = 1 h, atomic H/C ratio of the feed = 1.43.  110  4.3 Preliminary Coke-catalyst Characterization The main goal of characterizing the coke generated during hydroconversion was to investigate the potential for coke-catalyst recycle. To achieve this goal, three main issues need to be addressed: (i) the metal added to the reactor must be recovered in the solid product, otherwise a more detailed study is required to determine the rate of catalyst make-up. If no catalyst is present in the coke, the coke cannot be recycled as the hydroconversion catalyst to the reactor inlet (ii) the metal must be well dispersed within the coke. High agglomeration of catalyst particles in the coke-catalyst mixture will result in a very low catalytic surface area and very poor catalytic activity of the recycled catalyst. In this case, treatments to improve the dispersion of catalyst particles would be required before each recycling step and (iii) since the metal sulfides are the active form of hydroconversion catalysts, the state of the metal before recycling should be investigated to make sure that metal is in its active phase. If not, some pretreatment of the metals (sulfidation) present in the coke may be required prior to coke-catalyst mixture recycle. Elemental and EDX analyses were done on selected coke samples to determine the amount of catalyst captured by the coke at the end of the reaction. Results from three experiments (Table  4.4) revealed that the generated coke captured most of the Mo catalyst. The Mo recovery varied from 85 wt % in the experiment using 300 ppm Mo in the feed to 93 wt % in the experiment using 600 ppm Mo and 91 wt % in the experiment in which 900 ppm Mo was used as the catalyst. These results are consistent with results reported by others. Tye and Smith [18] reported that about 95 wt % of Mo using different Mo precursors ended up in 111  the coke generated during hydroconversion. Lee et al. [76] also reported that the metal concentration decreased in the liquid phase from 326 ppm to 12 ? 16 ppm when using a fixed bed of extrudates, made of microporous activated carbon to capture the catalyst in the hydrodesulfurization of residue oil. In the latter study, high metal adsorption was not surprising due to the high porosity and surface area of the activated carbon. Table  4.4 Mo recovery from the coke samples recovered from the batch reactor experiments after reaction. Mo (using Mo-micelle) concentration Mo added to the reactor Mo concentration in the coke Mo recovered from coke sample Mo recovery ppm mg mg/g coke mg wt % 300 24.0 31.3 a 20.4 84.8 600 48.0 25.0 a 44.8 93.2 900 72.0 47.7 b,c 65.8 91.4 Experimental conditions:  H2 Pressure = 5.5 MPa, temperature = 415 oC, reaction time = 1 h, atomic H/C ratio of the feed = 1.43. a Chemical analysis by Inductively Coupled Plasma Spectrometry (ICP) method. Analysis was done by an analytical services company outside UBC. b EDX analysis done by variable pressure 120 keV Hitachi S-3000N SEM. c An average of analysis results of 8 different areas.  Figure  4.12 shows the calculated Mo concentration in the coke recovered after reaction versus the coke concentration measured by EDX for a number of experiments done in both the batch and semi-batch reactors. The calculated amount of Mo in the coke samples was 112  determined based on the assumption that all the Mo added to the CLVR was associated with the coke at the end of the reaction. Good agreement between the calculated and measured concentrations confirmed the results presented in Table  4.4.  Figure  4.12 Comparison of Mo content in different coke samples measured by EDX and calculated amount assuming all Mo added to the CLVR is associated with the coke.  Selected coke samples from the batch and semi-batch reactor experiments were analyzed by SEM to investigate potential differences in the morphology of the coke generated in the thermal and catalytic experiments at 415 oC and 445 oC. Although there were some differences in the general morphology of the coke samples, the coke samples had very similar features as observed by SEM. For example, Figure  4.13-a and Figure  4.13-b show the coke samples from an experiment in the batch reactor with a Mo concentration of 0123456789100 1 2 3 4 5 6 7 8 9 10Measured Mo content of coke, wt%Theoretical Mo content of coke, wt %113  600 ppm added as AHM. Three main features were apparent in all the coke samples: a continuum phase of coke with some cavities typical of porous materials (Figure  4.13-a), a relatively smooth surface of carbonaceous material (Figure  4.13-b), and smaller precipitated and attached particles with size < 20 ?m on the surface (Figure  4.13-b). It has been reported that treatment of solid carbonaceous materials with different solvents affects the morphology of the solid [124]. Since Figure  4.13-a and Figure  4.13-b show different morphology of coke in the same sample, with the same reaction temperature and solvent used for solid recovery, the differences observed are ascribed to different stages of coke formation. Small particles on the surface (Figure  4.13-b) may depict nucleation and growth of the coke particles which result in a continuum phase with some porosity in the structure (Figure  4.13-a). The porous structure with further coke generation in its pores produced a smooth surface of carbonaceous material (Figure  4.13-a). The same morphology can be seen in coke samples generated in thermal experiments in the absence of catalyst (Figure  4.13-c and Figure  4.13-d). These observations suggest that the use of the catalyst and the different reaction temperatures, had little effect on the general morphology of the generated coke. The three steps: (1) coke nucleation and deposition on the surface, (2) agglomeration of small deposited particles and development of a continuum porous phase and (3) production of a smooth surface of coke are also shown in Figure  4.13-e. For TEM analysis, the coke generated in an experiment using 1800 ppm Mo, added in the form of a reversed micelle to the reactor, was used so as to ensure a high Mo concentration in the recovered coke that would allow the MoS2 to be observed by this technique. 114                Figure  4.13 SEM micrographs of different coke samples from hydroconversion reactions. a) and b): Batch reactor, 600 ppm Mo using AHM, T = 415 oC, PH2, initial ~ 5.5 MPa and 1 h reaction time. c) and d): Semi-batch reactor, thermal experiment, T = 445 oC, PH2 ~ 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. e): Batch reactor, 600 ppm Mo micelle, T = 415 oC, PH2, initial ~ 5.5 MPa and 1 h reaction time. 115  The coke generated by thermal reaction and by reaction with 600 ppm and 1800 ppm Mo, added via the micelle precursor, were examined by XRD (Figure  4.14). MoS2 has three characteristic peaks corresponding to the (002), (100) and (110) planes at 2? ~ 14o, 2? ~ 33.3o and 2? ~ 59.1o. Although the coke samples from the latter two experiments had a theoretical Mo concentration of about 3.1 wt % and 5.7 wt %, respectively, no peak corresponding to the (002) plane of MoS2 was observed in the diffractograms. Broader peaks corresponding to the (100) and (110) planes, which had some overlap with peaks from graphite and other metal sulfides, were identified in the diffractogram of the coke sample of the experiment using 1800 ppm Mo micelle. These peaks were not observed clearly in the XRD pattern of the coke sample recovered from an experiment using 600 ppm Mo micelle, mainly due to the low Mo concentration in the generated coke. The lack of a reflection corresponding to the (002) plane suggested the presence of highly dispersed crystallites with only a few atomic layers, that have higher hydrogenation activity than crystalline MoS2 that displays a prominent (002) peak [125]. Nickel and iron sulfide, present at a lower concentration than the MoS2, were also identified in the X-ray diffractograms. Since Fe and Ni were present in the feed, the intensity of these peaks was the same in all diffractograms from all coke samples and was independent of catalyst concentration. Agglomeration of these metal sulfides, compared to the highly dispersed MoS2, suggests that the coke-catalyst mixture recovered after the reaction has a high potential of remaining active in terms of residue conversion, hydrogenation and coke suppression.  116   (A)  (B) Figure  4.14 X-ray diffractogram of coke samples from different experiments in the semi-batch reactor before (a) and after (b) washing with water. T = 445 oC, PH2 ~ 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. Figure  4.14-A also suggests a change in the crystallinity of the coke with increased Mo concentration. In the coke sample generated in the thermal experiment, a sharp peak at 2? ~ 25.68o characteristic of graphitic carbon, was observed and the intensity of this peak 117  decreased significantly with increased Mo. This suggests the generation of more amorphous coke when the catalyst was used in the hydroconversion reaction. Note that the sharp peaks at 23.3?, 33?, 40.5? and 58.2? in the XRD diffractogram of coke sample recovered from the experiment using 1800 ppm Mo in the form of Mo-micelle (Figure  4.14-A) are ascribed to NH4Cl, a side-product of the micelle catalyst preparation procedure. These peaks disappeared when the coke was washed with de-ionized water (Figure  4.14-B). The dispersion of different metals on the surface of the coke samples generated in the experiment using 1800 ppm Mo micelle was investigated using SEM-EDX elemental mapping. Figure  4.15 confirmed the conclusion drawn from the XRD analysis: a high dispersion of C and S, major elements in the coke samples, were observed in the elemental maps. Fe and Ni were dispersed within the coke sample but in some locations, agglomeration of these metals (especially Fe which had a high concentration probably due to metal particles dislodging from the reactor wall during the reactor cleaning) was observed. Comparison of the map of the Fe metal with the S map shows that agglomeration of the Fe occurs at the same location as the S. This observation in the case of Fe, which showed good crystallinity in the XRD analysis (2? ~ 29.9 o, 33.9o, 43.6 o and 53.1 o), confirmed that the metals in the coke were present as metal sulfides.  118                                Figure  4.15 Elemental mapping of coke sample generated in an experiment using 1800 ppm Mo micelle in the semi-batch reactor. T = 445 oC, PH2 ~ 13.8 MPa at 900 mL(STP)/min and 1 h reaction time.  119  Further insight into the structure and morphology of the coke-catalyst mixture was obtained by HRTEM analysis. Two morphologies were apparent in the coke HRTEM micrographs. In the first micrograph (Figure  4.16-a), a tangled structure within all the sample was observed, but an interlayer space of 0.357 nm was also identified, in agreement with the d002 spacing of amorphous carbonaceous material [126]. Amorphous edges are well defined in the images and this structure seems to be the dominant structure within the sample. In agreement with the XRD results, stacking of (002) planes (with 0.62 nm interplanar spacing [124]) was not observed in any of the micrographs, indicating a high dispersion of MoS2 within the coke sample. No nickel sulfide or iron sulfide was observed either, most probably due to the low concentration of these metals in the sample. The second morphology identified by HRTEM was much more ordered with a smaller interlayer spacing (Figure  4.16-b) than that identified for the amorphous layers. These features are ascribed to graphitic carbon, which was also observed in the XRD diffractograms. Alkane-induced disruption is believed to be the main cause of the difference in the two types of structures observed in the HRTEM micrographs [126]. 120    Figure  4.16 HRTEM images of coke sample generated in an experiment using 1800 ppm Mo micelle in the semi-batch reactor. T = 445 oC, PH2 ~ 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. 121  4.4 Conclusion Catalyst tests done in a batch reactor at low temperature (415 ?C) resulted in low residue conversion (40 - 50 wt %) and showed that the Mo-micelle catalysts had lower coke yield and higher H2 consumption than the Mo catalyst derived from AHM added directly to the oil. Furthermore, the Mo-micelle catalysts showed increased H2 uptake with increased Mo concentration (0 - 900 ppm), although the effect on coke yield was small. The batch reactor catalyst tests (done under low conversion conditions) also showed that among the micelle prepared catalysts, Mo had better performance (higher H2 uptake and lower coke yield) than Fe, whereas both the Mo and Fe micelle prepared catalysts performed better than Mo derived from AHM and Fe derived from iron sulphate. The bimetal catalysts showed no beneficial effect of the addition of the second metal to the Mo catalysts. Two types of catalyst both prepared from water-soluble precursors were further examined and their activities were compared in batch and semi-batch slurry reactors. In both reactors, based on higher H2 consumption, higher liquid yield, lower coke yield and lower gas yields, the Mo-micelle catalyst precursor was shown to give higher activity than the water-soluble AHM catalyst precursor. Operating the catalysts at a higher temperature showed that the AHM catalyst precursor was not capable of rapid hydrogenation and hence coke suppression, whereas the Mo-micelle catalyst precursor had very good coke suppression (coke yield < 2 wt %).  Chemical and EDX analysis of selected coke-catalyst samples recovered from the reactors revealed that most of the metal added to the reactor at the beginning of the reaction 122  was captured by the coke recovered from the reaction products. The presence of MoS2, highly dispersed in the coke in a few atomic layers was confirmed by XRD analysis. The presence of other metal sulfides in the coke was shown by XRD and EDX-mapping techniques, but these metal sulfides were more agglomerated in the coke compared to the MoS2. Two main morphologies were observed in the HRTEM images of the coke, one an amorphous structure and the second a more ordered crystalline structure ascribed to graphitic carbon. No obvious evidence of MoS2 stacking in the coke sample was observed. The characterization results indicated that most of the catalyst was highly dispersed in the generated coke, suggesting a good potential for the recovered coke to be recycled and used as the hydroconversion catalyst.         123  Chapter 5  Comparison of MoS2 Catalysts Prepared from Mo-micelle and Mo-octoate Precursors for Hydroconversion of Cold Lake Vacuum Residue: Catalyst Activity, Coke Properties and Catalyst Recycle2 5.1 Introduction Several studies have reported on the performance of different metal sulfides for residue oil hydroconversion in which the catalyst was introduced to the reactor in the form of an oil-soluble precursor [14,19,21,76]. The oil-soluble catalysts have superior activity compared to the activity of the catalysts added to the reactor as water-soluble salts or finely divided powders. For example, Sato et al. [14] compared an ultra-fine MoS2 with MoS2 prepared in-situ from oil-soluble Mo-dithiocarbamate during hydroprocessing of Kuwait AR. The authors showed that in order to achieve an activity close to the activity of the in-situ prepared catalyst, an extensive micronization of the MoS2 was required. Unlike the oil-soluble precursors, water-soluble precursors have captured much less interest because of the problems associated with introducing aqueous solutions to the residue oil [2]. Fast 2 A version of this chapter has been published previously: Hooman Rezaei, Shahrzad Jooya Ardakani and Kevin J. Smith, Comparison of MoS2 catalysts prepared from Mo-micelle and Mo-octoate precursors for hydroconversion of Cold Lake Vacuum Residue: Catalyst activity, coke properties and catalyst recycle, Energy&Fuel, 2012, 26, 2768-2778. 124                                                   evaporation of water and agglomeration of precursor salts results in the production of large catalyst particles with low activity. In the previous chapter, it was shown that the catalytic activity of a water-soluble precursor (AHM) could be significantly increased by transforming the water-soluble precursor into an oil-soluble precursor using a reversed micelle. The state of the catalyst in the catalyst-coke mixture recovered from the hydroconversion reaction was also investigated to determine the potential for catalyst-coke recycle. Results of that study indicated that the recovered catalyst-coke could potentially be re-used in subsequent hydroconversion reactions. In the present chapter, the performance of the MoS2 catalyst prepared from the reversed micelle precursor is compared to that of MoS2 prepared from an oil-soluble precursor. Mo-octoate was chosen as the precursor since it represents a low-cost, oil-soluble alternative to the Mo-micelle precursor. Since the generated coke acts as a support for the MoS2 particles in the recovered coke-catalyst, an understanding of the properties of the generated coke and changes in its properties with catalyst type and concentration, as well as with operating conditions, is necessary. Coke-catalyst characterization results are presented, and differences in the coke properties are linked to the catalyst type, concentration and operating conditions in the hydroconversion experiments. The recovered coke-catalyst was also recycled and the activity of the recycled coke-catalyst under high residue conversion conditions is compared to the fresh catalyst. 125  5.2 Results and Discussions 5.2.1 Effect of Catalyst Precursor and Concentration In this chapter, results of experiments in which catalyst precursors were added to the reactor are compared to non-catalytic experiments. Thus possible effects of the reactor wall are minimized.  Furthermore, to ensure that sufficient H2 was available during reaction even at high residue conversions, a semi-batch reactor with continuous H2 gas flow was used. The H2 flow removes volatile compounds present in the reactor and increases the viscosity of the remaining liquid in the reactor. Removal of light compounds (which act as a solvent for the asphaltenes in the oil mixture) in the reactor decreases the asphaltene solubility, promoting liquid-liquid phase separation, which eventually leads to coke formation [95,127]. Consequently, the coke yields reported herein are likely higher than what would be expected from a continuous industrial hydroconversion reactor operating at the same severity. Figure  5.1 reports the coke and liquid yield as a function of Mo concentration for both the Mo-micelle and Mo-octoate catalyst precursors. The coke yield decreased from 21.2 wt % in the non-catalytic experiment to 4.8 wt % with 100 ppm Mo using the Mo-micelle precursor. Clearly, the decrease in coke formation is a consequence of the MoS2 catalyst. The coke suppression is likely due to direct olefin hydrogenation, with the olefins produced by ?-scission and partial hydrogenation of PAHs to form H donor species that prevent polymerization and condensation reactions [87]. The catalytic hydrogenation reactions cause a noticeable increase in the H2 uptake, even when the catalyst is added to the reactor in small concentrations, as shown in Figure  5.2. In the non-catalytic experiment, despite the presence 126  of a high H2 pressure in the reactor, the H2 was not effectively transferred to the liquid phase. Consequently, the olefins and free radicals undergo polymerization and termination reactions and produce a significant amount of an insoluble aromatic-rich liquid phase which eventually produces coke.  Figure  5.1 Coke and liquid yields from CLVR hydroconversion as a function of Mo concentration for Mo-micelle and Mo-octoate catalyst precursors. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time.  For both catalyst precursors, an increase in Mo concentration from 100 to 300 ppm decreased the coke yield from 4.8 wt % to 1.9 wt %. Further increases in the catalyst 0 500 1000 1500 2000010206080100  Coke or liquid yield, wt%Mo Concentration, ppm Coke yield, Mo-Micelle Coke yield, Mo-Octoate Liquid yield, Mo-Micelle Liquid yield, Mo-Octoate127  concentration to 600 ppm Mo had no significant effect on the coke yield. When the Mo concentration increased to 1800 ppm, the coke yield (2.2 wt %) increased marginally compared to the coke yield with 600 ppm Mo (1.8 wt %) for both the Mo-micelle and Mo-octoate catalyst precursors. These small changes in coke yield with Mo concentration could be a result of two competing mechanisms. On the one hand, the hydrogenation rate is dependent on catalyst concentration and dispersion, while on the other, the catalyst particles (similar to other solid particles in the reactor) may act as nucleation sites for the agglomeration of small solid coke particles at the reaction conditions [2]. Consequently, there is a maximum catalyst concentration that minimizes coke yield and in the present study, 600 ppm Mo appears to be the optimum concentration in terms of coke suppression. Similar observations have been reported by others [17,21,35]. Panariti et al. [17] showed that the competing effects occur over a wide range of hydroconversion temperatures from 400 ?C to 440 ?C. In addition to the mechanism described above, Panariti et al. [17] claimed that excess hydrogenation of the liquid feed that occurs at high Mo concentrations, reduces asphaltene stability, which eventually causes more coke formation. This mechanism has also been proposed by others [95,128].  128   Figure  5.2 H2 uptake during CLVR hydroconversion as a function of Mo concentration for Mo-micelle and Mo-octoate catalyst precursors. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time.  The goal of hydroconversion is to maximize liquid yield and minimize coke and gas yield at high residue conversions. Figure  5.1 shows that the liquid yield increased in the catalytic experiments compared to the non-catalytic experiment. In the non-catalytic experiment, the liquid yield was very low (59.0 wt %) due to the high gas (9.1 wt %) and coke (21.2 wt %) yields. By increasing the catalyst concentration, the liquid yield increased significantly to ~ 82 wt % with 100 ppm Mo and to ~ 88.0 wt % with 300 ppm Mo for both 0 200 400 600 800 1000 1200 1400 1600 1800 20001.21.41.61.82.02.22.4   Mo-octoate Mo-micelleHydrogen uptake, wt%Mo concentration, ppm129  the Mo-micelle and Mo-octoate precursors. A very slow increase in liquid yield with further increases in Mo concentration to 1800 ppm was observed, consistent with the trend observed for the coke yield.  Figure  5.1 also shows that the Mo-octoate precursor had almost the same activity as the Mo-micelle precursor in terms of coke and liquid yields. This indicated that although preparation of an oil-soluble Mo-micelle precursor from a water-soluble Mo salt increased the catalytic activity of MoS2 compared to the activity of the water-soluble precursor alone ( Chapter 4), the Mo-micelle precursor had the same performance as the oil-soluble Mo-octoate precursor. The small differences in liquid yield from the Mo-micelle and Mo-octoate catalyst precursors is believed to be due to the presence of n-hexane used as solvent in the preparation of the Mo-micelle precursor. A fraction of the n-hexane was recovered in the condenser after reaction and this increased the liquid yield in experiments in which the Mo-micelle precursor was used. The H2 uptake (Figure  5.2) also showed a significant increase when MoS2 was added to the reactor and the H2 uptake for both the Mo-micelle and Mo-octoate catalyst precursors was almost the same. However, unlike the coke and liquid yields, the H2 uptake increased continuously with Mo concentration up to 1800 ppm, suggesting that at high Mo concentrations (600 and 1800 ppm) although the H2 was being consumed, it was not being effectively transferred to the unstable, cracked molecules to suppress coke formation.  According to the chain reaction mechanism for liquid-phase cracking proposed by Gray and McCaffrey [87], which is an extended form of the LaMarca-Libanti-Klein-Cronauer (LLKC) 130  free-radical chain reaction mechanism [88], the olefin and aromatic hydrogenation reactions are catalytic. Hence, the high catalyst concentration yields a more hydrogenated product. The continued increase in H2 consumption with Mo concentration, reported in the present study, also suggests that no matter how fast the rate of H transfer from the gas phase to the free radicals and PAHs, polymerization and condensation reactions occur and cannot be completely prevented. Hence some products have significantly higher aromatic content than the feed, which eventually causes liquid phase separation and coke formation [102]. Once phase separation occurs, the formation of toluene-insoluble material (coke) is extremely fast [102]. Consequently, zero coke formation during hydroconversion of residue oil under hydroconversion conditions which yields significant residue conversion has never been reported. In summary, the H2 consumption that occurred with increased catalyst concentration resulted in the hydrogenation of olefins and aromatics and the production of more hydrogenated products (Figure  5.3), not additional coke suppression. This result is consistent with a marginally increased coke yield due to excess hydrogenation of the liquid phase and instability of the asphaltene phase in the reactor when a very high Mo concentration (1800 ppm Mo) was used (Figure  5.1). 131   Figure  5.3 H/C atom ratio and CAr/CAl ratio of solid (coke) and liquid products from CLVR hydroconversion; (?) H/C atom ratio of liquid products, (?) H/C atom ratio of coke, (?) H/C atom ratio of coke and (?) CAr/CAl ratio of coke as a function of Mo concentration for Mo-micelle and Mo-octoate catalyst precursors. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time  Although the H2 uptake increased when the Mo concentration increased from 600 ppm to 1800 ppm, the rate of this increase was much lower than when the Mo concentration increased from 0 to 600 ppm. This could be explained by the fact that there are potentially two sources of H2 consumption in hydroconversion reactions. The first is hydrogen transfer by H-donor compounds to the free radicals generated during cracking, which is likely 0 300 600 900 1200 1500 18000.60.91.21.523456  H/C atomic ratio or CAr/CAl ratioMo concentration, ppmMo-micelle precursorMo-octoate precursor132  responsible for most of the H2 consumption since the numerous side chains of the complex molecules present in residue oil would generate a substantial number of free radicals. The second source of H2 consumption is hydrogenation of compounds with unsaturated bonds (such as olefins) and aromatic rings of PAHs present in the residue oil. Using very high Mo concentrations (1800 ppm in the present study) significantly increases the number of hydrogenation sites but the rate of H2 transfer to the liquid is limited. Hence the H2 uptake does not increase proportionally with the catalyst concentration at high catalyst concentrations.  Figure  5.3 shows that unlike the coke yield, the H/C atom ratio in the products increased linearly with increased Mo concentration in the feed. Hydrogenation of PAHs reduced the amount of aromatics in the products, as shown in Figure  5.3, where the aromatic to aliphatic carbon ratio (CAr/CAl) in the product coke from experiments using the Mo-micelle precursor, decreased rapidly with increased catalyst concentration. The H/C atom ratios from two experiments, shown by the circle in Figure  5.3, were slightly off-set from the line showing the correlation between the H/C atom ratio of the liquid product and the Mo concentration. This is because the H/C atom ratio of the liquid product depends on two factors, whereas the H/C atom ratio of the coke is mainly dependent on the H content of the macromolecular coke precursors which are more hydrogenated as catalyst concentration increases and hence produce more hydrogenated coke. In the liquid product on the other hand, the H/C atom ratio depends on both the extent of carbon removal (coke formation) and the degree of hydrogenation of the liquid phase. For the non-catalytic experiment, although the H2 uptake was low (Figure  5.2), a very high coke yield (Figure  5.1) resulted in a high 133  H/C atom ratio of the liquid product. In the experiment with 100 ppm Mo using the Mo-micelle as the catalyst precursor, although the coke yield was lower than the coke yield from the non-catalytic experiment (Figure  5.1), a higher H2 uptake increased the H/C atom ratio to the same extent as the non-catalytic experiment. Figure  5.4 reports the TIOR conversion and coke yield as a function of the Mo concentration used in the hydroconversion of CLVR. The non-catalytic experiment had a much lower TIOR conversion compared to the catalytic experiments, due to a much lower H2 uptake in the absence of the catalysts (Figure  5.2). Low H2 uptake increases the coke formation significantly. When the Mo-micelle or Mo-octoate catalyst precursors were added to the reactor (even at very low concentrations of 100 ppm of Mo), the TIOR conversion increased significantly from 64 wt %  in the non-catalytic experiment to 85 ? 90 wt % in the catalytic experiments. The TIOR conversion remained almost constant as the catalyst concentration increased above 100 ppm Mo for both Mo precursors. This is because the residue conversion is a consequence of thermal cracking, followed by H transfer to the olefins produced by ?-scission of the free radicals produced during cracking. These reactions occur in series starting with C-C bond cleavage followed by H transfer to unsaturated bonds and free radicals to prevent polymerization and condensation reactions. Since the cracking reaction is mainly controlled by temperature [123,129-131], the cracking reaction will be rate limiting during residue conversion and excess catalyst does not affect the extent of residue conversion. 134   Figure  5.4 Coke yield and Residue (TIOR) conversion during CLVR hydroconversion as a function of Mo concentration for Mo-micelle and Mo-octoate catalyst precursors. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time.  The HTSD data of Figure  5.5 reports the liquid product distribution according to standard BP ranges for naphtha  (< 204 ?C), light gas oil (204 ? 348 ?C) heavy gas oil (348 - 524 ?C) and residue (or pitch, > 524 ?C). The data show that higher catalyst concentration resulted in a small shift in the distribution of liquid products from naphtha toward light and heavy gas oils, both of which are more favorable products for refineries. Both the 100 ppm Mo and the non-catalytic experiment produced lighter liquid products with higher naphtha yield compared to experiments using higher catalyst concentrations. This shift in liquid product distribution was at the expense of a lower total liquid yield (including > 524?C 0 500 1000 1500 20000102060708090  Coke yield or TIOR conversion, wt%Mo concentration, ppm Coke yield, Mo-Micelle Coke yield, Mo-Octoate TIOR conversion, Mo-Micelle TIOR conversion, Mo-Octoate135  fraction) and higher coke yield (Figure  5.1).  The Mo-micelle and Mo-octoate precursors also showed very similar liquid product distributions, confirming that the catalyst precursors had no impact on the MoS2 catalyst activity.  Figure  5.5 Liquid product distribution from CLVR hydroconversion as a function of Mo concentration for Mo-micelle (A) and Mo-octoate (B) catalyst precursors. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. -20020406080100(A)0 1800600300100   524 ?C +   348 ?C - 524 ?C   204 ?C - 348 ?C   < 204 ?CHydrocarbon yields, wt% of feedMo concentration, ppm-2002040608010018006003000 100  Hydrocarbon yields, wt% of feedMo concentration, ppm 524 ?C +   348 ?C - 524 ?C   204 ?C - 348 ?C   < 204 ?C(B)136  Based on the results reported herein, it may be concluded that increasing catalyst concentration above a certain value (600 ppm) does not have a significant impact on either the quality of the products or the coke suppression. To estimate the optimum concentration of catalyst in the semi-batch reactor under the hydroconversion conditions used in the present study, the ratio of the TIOR conversion to the coke yield was evaluated. Maximizing this ratio reflects the ultimate goal of high residue conversion while coke production is minimized. Figure  5.6 shows the change in the TIOR conversion to coke yield ratio for different experiments of the present study using both Mo-micelle and Mo-octoate precursors. Figure  5.6 shows that the ratio increased significantly when the catalyst was added to the reactor, up to 600 ppm Mo concentration in the reactor. At 1800 ppm Mo concentration the ratio decreased marginally. Although the two precursors showed very similar activity in terms of hydroconversion of the residue feed, the Mo-octoate precursor would be preferred. This is due to the simplicity of using an oil-soluble precursor compared to the Mo-micelle precursor which requires several synthesis steps.  In addition, the lower cost of the commercially available Mo-octoate precursor, compared to the Mo-micelle that requires the use of expensive surfactants, would be a significant advantage.   137   Figure  5.6 TIOR conversion to coke yield ratio from CLVR hydroconversion as a function of Mo concentration for Mo-micelle and Mo-octoate catalyst precursors. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time  5.2.2 Effect of Temperature and Heat-up Period Heck et al. [130] showed that part of the residue fraction in residue oil has a tendency to crack very rapidly. Since hydroconversion reactions occur at temperatures above 350 ?C [2] and the standard reaction temperature of the present study was 445 ?C, the hydroconversion reaction must have also occurred during heat-up of the reactor. For the semi-batch slurry-phase reactor used in the present study, the heat-up period from 350 ?C to 445 ?C was approximately 20 mins.  Consequently, it was important to determine the effect 0 200 400 600 800 1000 1200 1400 1600 1800 200001020304050   Mo-Micelle Mo-OctoateTIOR conversion to coke yield ratioMo concentration, ppm138  of the heat-up period and reaction temperature on the residue oil hydroconversion. Since the Mo-micelle and Mo-octoate precursors showed very similar activity, these experiments were done using the Mo-micelle precursor.  Table  5.1 summarizes the reaction times and reaction temperatures investigated to determine the effect of the heat-up period. In experiments B and D, following heat-up of the reactor to 430 ?C and 445 ?C, respectively, the reactor was quenched immediately and the products recovered and analyzed. Experiments A, C and E were standard hydroconversion experiments in which a 1-h reaction time followed the heat-up period, each done at different reaction temperatures. Comparison of experiments B and D with experiments C and E clearly show that at the higher temperature, the heat-up period had a much more significant effect on the coke yield, asphaltene and TIOR conversions and liquid-phase S conversion, compared to the experiments done at lower temperatures. Heat-up to 445 ?C resulted in 62 % residue conversion and 1.9 wt % coke yield. The data of Table  5.1 show that at 430 ?C, 43 % of the total coke produced was formed during the heat-up period and this increased to 65 % at 445 ?C (comparing experiments B with C and D with E). Large aliphatic fragments attached to the polyaromatic core in residue macromolecules have a tendency to crack very rapidly [130]. Due to the very high rate of cracking when the concentration of these fragments is high (during the heat-up period), the rate of H transfer to the cracked fragments is slower than the rate of condensation to coke particles.  Comparing the data of experiments A, C and E in Table  5.1 also shows that increasing the reaction temperature from 415 ?C to 430 ?C and then to 445 ?C increased the coke yield and the TIOR, asphaltene and S conversions. 139  Table  5.1 Coke yield, TIOR and asphaltene conversion, and liquid-phase S conversion in the heat-up experiments and the experiments using different reaction temperatures. Experiment identifier A B C D E Reaction temperature, ?C 415 430 430 445 445 Reaction time*, h 1 0 1 0 1 Coke yield, wt % 0.68 1.08 2.52 1.88 2.90 TIOR conversion, wt % 52.09 47.90 69.82 61.80 83.74 Asphaltene conversion, wt % 43.33 38.32 66.20 50.01 75.13 S conversion, wt % 47.67 35.00 59.33 43.33 67.29 * Reaction time after heat-up to the desired reaction temperature Experimental conditions: PH2 = 13.8 MPa at 900 mL(STP)/min. Catalyst used for all experiments was 600 ppm Mo-micelle. 5.2.3 Coke Properties The coke recovered from the experiments of the present study was characterized to investigate the effect of catalyst on the coke properties. Since after reaction most of the catalyst added to the reactor was associated with the product coke (Section  4.3), understanding the produced coke morphology and the coke-catalyst interaction is necessary if the catalyst is to be recycled in slurry phase hydroconversion.  The BET surface area of the coke generated at different operating conditions is presented in Table  5.2 and compared to the BET surface area of pure graphite (Aldrich, mesh 140  325, 99.999 wt %). The coke recovered from a non-catalytic experiment had a low BET surface area (4.9 m2/g), comparable to that of graphite (3.9 m2/g), whereas the BET area of the coke generated by catalytic hydroconversion was significantly higher. The coke recovered from the two hydroconversion experiments using 600 ppm Mo-micelle and Mo-octoate precursors had BET areas of 13.5 and 18.5 m2/g, respectively. The BJH total pore volume of coke samples had similar trends. Although the surface area and pore columes are low, even for the coke samples recovered from the catalytic experiments, the results indicate that the coke was more amorphous and porous in the presence of the MoS2 catalyst, compared to the coke recovered from the non-catalytic experiment. This is probably due to the higher H content of the coke generated during catalytic hydroconversion (Figure  5.3) which reduced the crystallinity of the carbonaceous material and increased the surface area. These observations are consistent with XRD analysis of the coke reported previously (Section  4.3) that showed an intense graphite peak for coke recovered from a non-catalytic experiment that became less intense and broader as catalyst concentration increased. Table  5.2 BET surface area and total pore volume of coke recovered from non-catalytic and catalytic CLVR hydroconversion experiments.    Experiment for coke source Sample Commercial Graphite  Non-catalytic  600 ppm Mo-micelle  600 ppm Mo-octoate  BET surface area, m2/g 3.9  4.9 13.5 18.5 BJH total pore volume, cm3/g 0.018  0.028 0.063 0.075 Reaction conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. 141  The type of catalyst precursor seemed not to have a significant effect on the crystalline structure of the carbon and the MoS2 in the coke-catalyst mixture recovered after reaction. Figure  5.7 shows the XRD diffractogram of the coke recovered from two hydroconversion experiments with 1800 ppm of Mo added as Mo-micelle and Mo-octoate precursors. Iron and nickel sulfides were observed as very sharp peaks with similar intensities for both samples. Due to possible washout of metal particles from the reactor wall during coke recovery, comparison of the iron and nickel sulfide peaks between different coke samples is not meaningful. A noticeable amount of ammonium chloride (NH4Cl) was detected when the Mo-micelle precursor was used.  Figure  5.7 XRD diffractogram of coke recovered from CLVR hydroconversion using 1800 ppm Mo from Mo-micelle and Mo-octoate catalyst precursors. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. 10 20 30 40 50 60 70 80 90Intensity, (a.u)2?, degree1800 ppm Mo-octoate1800 ppm Mo-micelleNH4ClNH4ClNH4ClGraphite-2HMoS2 (110)Fe(1-x)SFe(1-x)S Ni0.96SFe(1-x)S Ni0.96SMoS2 (100)142  Figure  5.7 shows a relatively broad peak of MoS2 at ~ 59o and the absence of a sharp peak from the (002) plane of MoS2 at ~14o, indicative of a high dispersion of MoS2 with a very small number of crystalline layers [125]. Such a structure was confirmed by TEM analysis (Figure  5.8). In Figure  5.8-B and Figure  5.8-C, single layers of MoS2 can be observed within the coke samples. The number of these single layers is clearly higher in Figure  5.8-C, corresponding to the coke recovered when using 1800 ppm Mo prepared from the Mo-octoate precursor compared to the 600 ppm Mo of Figure 8-B. Although the concentration of MoS2 in the coke recovered from an experiment using 1800 ppm Mo using the Mo-octoate precursor was higher than the MoS2 concentration in the coke recovered from the experiment using 600 ppm Mo-octoate precursor, the MoS2 catalyst remained single layered and highly dispersed within the coke. Furthermore, there was no sign of catalyst agglomeration or stacking of the MoS2 layers with increased catalyst concentration. Figure  5.8 also shows the difference in morphology of the coke recovered from non-catalytic and catalytic experiments. The TEM micrographs of the coke recovered from the non-catalytic experiments clearly showed the presence of graphitic structures (Figure  5.8-A, where a graphitic particle of diameter ~40 nm is clearly visible) whereas no such features were present in the coke samples recovered from the catalytic experiments.    143         Figure  5.8 TEM micrographs of coke samples recovered from CLVR hydroconversion using different Mo concentrations using Mo-octoate precursor. (A) Non-catalytic experiment, (B) 600 ppm Mo using Mo-octoate precursor and (C) 1800 ppm Mo using Mo-octoate precursor. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time  144  DRIFTS spectra of the coke-catalyst are presented in Figure  5.9.  Three main frequency regions can be distinguished in which bands occur (2800 ? 3000 cm-1, 1300 ? 1500 cm-1 and 700 ? 1300 cm-1) together with an intense and relatively broad band at 1600 cm-1. These bands are in agreement with literature on coke deposition on catalysts [132-138]. In the range of the C-H stretching vibration in alkanes (region of 2800 ? 3000 cm-1 or paraffinic region) three main bands were observed at 2862, 2917 and 2962 cm-1. These are assigned to the symmetric vibrations of CH2 and CH3, the asymmetric vibration of CH2, and the asymmetric vibration of CH3, respectively [137]. The bands are distinguishable in all spectra of Figure  5.9. In the coke samples recovered after 1-h reaction with catalyst concentrations of 100, 300 and 1800 ppm Mo using the Mo-octoate precursor, the ratio of the band intensities at 2917 and 2962 cm-1 was almost constant, implying that the length of the aliphatic branches in the coke samples were similar.  The coke recovered from the non-catalytic experiment had a lower ratio indicative of shorter aliphatic chains. On the other hand, in the coke recovered after heat-up to 445 oC (top spectrum in Figure  5.9), the band at 2917 cm-1 was of much higher intensity compared to the band intensity at 2962 cm-1. Clearly after 1-h reaction, a portion of the aliphatic chains of the residue oil is cracked and separated from the macromolecules to produce gas or liquid, depending on the length of the chain. This makes the aliphatic chain length of the coke precursors shorter after 1-h reaction compared to after the heat-up period. 145   Figure  5.9 DRIFT spectra of coke samples recovered from CLVR hydroconversion using different concentrations of Mo added in the form of Mo-octoate. Operating conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. Reaction time for heat-up experiment was zero and Mo concentration was 600 ppm using Mo-octoate precursor.  The band at 1600 cm-1 in the DRIFTS spectra is attributed to the stretching vibration of C = C in microcrystalline graphitic structures, which are present in polycyclic aromatic compounds [134,137,139]. The 3050 cm-1 band is probably due to the C ? H stretching vibration in aromatics [140,141]. Although the band at 1600 cm-1 had almost the same intensity in all the coke samples, there was a decreasing trend in the intensity of the aromatic 4000 3600 3200 2800 1600 1400 1200 1000 800 6000.000.250.500.751.001442 1378 132530502962291728621600866 811Heat-up1800 ppm300 ppm100 ppm  Kubelka-MunkWave number (cm-1)Non-catalytic751146  band at 3050 cm-1 with increasing Mo concentration. Hence it may be concluded that with a decrease in catalyst concentration, the concentration of aromatic rings in the recovered coke increased. In the range of the C ? H bending vibration (region of 1300 - 1500 cm-1) in paraffinic or olefinic compounds, the band at 1442 cm-1 can be assigned to the asymmetric bending vibration of aliphatic groups attached to aromatic rings. This band usually replaces the in-plane bending vibration of CH2 at ~1465 cm-1 when aromatics are formed [137,141]. The band at 1378 cm-1 originated from the symmetric bending vibration of CH3 and the band at 1325 cm-1 is an unknown absorption. These bands are distinguishable in the DRIFTS spectra of the coke and Figure  5.9 shows that they are almost identical for all the coke samples analyzed. In the region of 700 - 1000 cm-1 three main absorption bands are distinguished at 751 cm-1, 811 cm-1 and 866 cm-1. These likely originate from the in-phase out-of-plane wagging vibrations of hydrogen in substituted aromatic rings [140,141]. The peaks were very intense in the coke recovered from the non-catalytic experiment. As the catalyst concentration increased to 1800 ppm Mo, using the Mo-octoate precursor, the intensity of the peaks decreased and finally disappeared. The DRIFTS spectrum of the coke recovered from the heat-up experiment did not show any distinguishable peaks at these wave numbers. Although the DRIFTS spectra showed a clear change in the region 1000 - 1300 cm-1 (peak intensity increased as the catalyst concentration increased), no conclusion could be made regarding these bands, mainly because the peaks were not well resolved. 147  TGA analysis of the coke recovered from the non-catalytic and catalytic experiments using Mo concentrations 100, 300 and 600 ppm of Mo are presented in Table  5.3. The coke was recovered from experiments in which both Mo-micelle and Mo-octoate precursors were used as the catalyst precursor. The results clearly show differences between the coke recovered from non-catalytic and catalytic experiments. Using both Mo-octoate and Mo-micelle precursors, the coke loses much more weight compared to the coke recovered from the non-catalytic experiment. Due to the high temperature reached during the TGA analysis in N2, cracking reactions occur and carryover of the cracked products causes the weight loss. Hence the results indicate that the coke  recovered from the catalytic experiments undergo cracking reactions much more readily than the coke from non-catalytic experiments, probably due to the more graphitic nature of the coke recovered from the non-catalytic experiments. The weight loss increased marginally as the catalyst concentration increased, with the total weight loss of 26.9 % with 300 ppm Mo increasing to 30.8 % with 600 ppm Mo.    148  Table  5.3 TGA of coke recovered from hydroconversion reactions using different types of catalysts at different concentrations.  Mo-micelle precursor  Mo-octoate precursor  Weight loss, %  Weight loss, % Mo, ppm 120-550 ?C 550-750 ?C 750-900 ?C 120-900 ?C  120-550 ?C 550-750 ?C 750-900 ?C 120-900 ?C 0 4.31 5.20 3.36 12.87  4.31 5.20 3.36 12.87 100 N/A N/A N/A N/A  9.85 6.80 6.76 23.41 300 10.50 5.62 7.38 26.89  11.35 6.23 6.86 24.44 600 16.03 5.36 9.44 30.83  N/A N/A N/A N/A ?????? ???? =  ?? ?????? 100 [%] Wi : Weight of the coke at initial temperature Wf : Weight of the coke at final temperature 149  XPS and EDX characterization data of the coke is summarized in Table  5.4.  EDX analysis showed that the Mo concentration in the coke increased from 1.28 to 2.82 wt % as the catalyst concentration increased from 300 ppm to 600 ppm using the Mo-micelle catalyst precursor. The Mo/C weight ratio in the coke was 0.02 for the coke recovered from the experiment using 300 ppm Mo and this ratio doubled as the catalyst concentration doubled. Although the same Mo concentration (600 ppm) was used in both heat-up experiments B and D, the Mo/C weight ratio increased to 0.09 and 0.06 for the heat-up experiment to 430 oC and 445 oC, respectively. This could be because the coke samples recovered from the heat-up experiments had a higher H/C atom ratio (Figure 10) than the coke samples recovered after 1 h reaction.  More likely, however, is the fact that less coke was generated in the heat-up experiments compared to after 1-h reaction, yielding a higher Mo/C weight ratio. The decreased Mo/C weight ratio as the reaction temperature increased from 430 to 445 oC is also consistent with more coke generation and a lower H/C ratio of the recovered coke at higher temperatures because of more cracking and removal of lighter and more hydrogenated parts of the residue. XPS analysis of the coke followed the same trend as the EDX data. The Mo/C weight ratio of the coke from the two experiments using 600 ppm Mo-micelle and Mo-octoate precursors were the same (0.002). The ratio, similar to the EDX data, increased to 0.007 and 0.005 for the heat-up experiments to 430 oC and 445 oC, respectively. The XPS data of Table  5.4 show Mo/C ratios about 10x?s lower than those obtained by EDX. EDX has an analysis depth of a few micrometers whereas analysis by XPS is confined to a few nanometers below the surface. Considering these differences and recalling that coke 150  recovered from the hydroconversion reaction had relatively low surface area, the EDX and XPS data suggest that adsorption of the MoS2 to the coke surface is followed by further coke formation and growth of the coke particle that eventually occludes the MoS2 from the particle surface. The data suggest that the MoS2 was not adsorbed after all the coke had formed but rather the MoS2 was continuously adsorbed on the coke surface and then covered by more coke during further reaction. The high coverage of MoS2 by solid coke suggests that the MoS2 may act as a nucleation site for coke formation. Table  5.4 EDX and XPS analysis of coke generated from CLVR hydroconversion experiments in the semi-batch reactor.    EDX Analysisa XPS Analysis Coke sample Coke yield, wt % Mo, wt %b Mo, wt % Mo/C Mo, wt % Mo/C 300 ppm MoM 1.91 1.57 1.28 0.02 N/A N/A 600 ppm MoM 1.83 3.27 2.82 0.04 0.22 0.002 600 ppm MoO 1.76 3.41 3.22 0.04 0.25 0.002 Heat-up Bc 1.08 5.56 4.70 0.09 0.54 0.007 Heat-up Dc 1.88 3.19 4.92 0.06 0.46 0.005 a- EDX analysis data presented are average of at least 6 different data points. b- Theoretical wt % of Mo in the coke assuming all the coke added to the reactor will end up in the recovered solid coke. c- Heat-up experiment letters are based on letters in Table  5.1. In both heat-up experiments, 600 ppm Mo-micelle was used as catalyst precursor. MoM : Mo-micelle used as catalyst precursor. MoO : Mo-octoate used as catalyst precursor. 151  5.2.4 Preliminary Catalyst Recycle Experiments Although unsupported MoS2 catalysts have shown very high activity in terms of coke suppression and residue conversion, using these catalysts once through in the upgrading process results in high catalyst costs. One approach to decreasing these costs is to recycle the catalyst. In the present study, the solid coke-catalyst recovered from an experiment in which 600 ppm Mo, using the Mo-micelle precursor, was selected for recycle. Since previous work demonstrated that all the Mo added to the residue feed was present in the recovered coke (Section  4.3), a Mo concentration of 600 ppm in the residue feed would result if all the recovered coke was added to the reactor in a subsequent hydroconversion experiment to simulate catalyst recycle. In the present study, the recovered coke was divided into 3 portions and mixed with Mo-micelle to yield a fixed total Mo concentration of 300ppm, made up of 0, 33, 67 and 100% recycled Mo. Results of residue oil hydroconversion using the different portions of recycled catalyst are compared in Table  5.5. The data show that the recycled catalyst had very similar activity to that of the fresh Mo precursor in terms of coke suppression, H2 uptake, gas yield and H2S yield. Increasing the fraction of recycled catalyst in the total catalyst added to the reactor did not have any significant effect on the catalyst activity. The coke yield from all four experiments was < 2.5 % although, in the experiment done with 100% recycled catalyst (no fresh Mo precursor addition) the coke yield was slightly higher compared to the experiments in which some fresh catalyst precursor was used. High H2 uptake of ~2.2 wt % was also observed. S removal from the liquid feed (H2S yield) and H/C atom ratio of the liquid 152  products, were almost identical for all four experiments. The liquid yield was almost constant in all cases, with an average value of 86.6 ? 1.7 wt % (Table  5.5). More importantly, TIOR conversion also remained constant (84.7 ? 0.7 wt %)  and there was no significant difference between TIOR conversions measured with fresh catalyst, catalyst mixed with recycled coke-catalyst and completely recycled catalyst. Table  5.5 Results from CLVR hydroconversion using different ratios of fresh and recycled catalyst.  Exp. 1 Exp. 2 Exp. 3 Exp. 4 Mo concentration, ppm Fresh Recycle  300 0  200 100  100 200  0 300 H2 uptake, wt % 2.21 2.27 2.19 2.42 Gas yield, wt % 5.83 6.01 5.87 6.08 Coke yield, wt % 1.91 1.77 1.64 2.22 H2S yield, wt % 1.12 1.11 1.04 1.43 Liquid H/C atom ratio 1.53 1.53 1.53 1.53 Liquid yield, wt % 88.03 87.05 87.71 83.77 Naphtha 21.41 20.84 21.32 23.47 Light gas oil 32.10 32.33 32.20 30.69 Heavy gas oil 24.16 24.23 24.40 21.31 Residue 10.36 9.65 9.79 8.29 TIOR conversion 83.60 84.74 84.73 85.55 Reaction conditions: T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time. 153  Table  5.5 also compares the selectivity toward different hydrocarbon cuts in the liquid products for the recycled catalyst. All the catalysts used in the recycle experiments had almost the same selectivity except in the experiment with 100% recycled catalyst. In this case a slight shift of the products toward naphtha (< 204 ?C) and light gas oil (204 ?C - 348 ?C) fractions was observed, probably due to a slightly higher coke yield compared to the coke yields of the other 3 experiments (Table  5.5). The lighter liquid product was at the cost of lower liquid yield compared to the experiments in which a mixture of the fresh catalyst precursor and the recycle catalyst and also the experiment in which only fresh catalyst precursor, was used. 5.3 Conclusions MoS2 catalysts prepared using a water-soluble precursor dispersed in reversed micelles had the same activity for residue oil hydroconversion as MoS2 prepared from a Mo-octoate precursor. At 445 ?C and 13.8 MPa H2, the coke yield decreased from 22 wt % in the absence of a catalyst, to 4.8 wt % in the presence of 100 ppm Mo, and 600 ppm Mo was found to be optimum in terms of maximizing the residue conversion (84 wt %) and minimizing the coke yield (2.9 wt %). Considering the complexity of the preparation of the Mo-micelle precursor and the cost of synthesis, commercial Mo-octoate catalyst is a more suitable catalyst for residue hydroconversion compared to the Mo-micelle precursor. Results from heat-up experiments showed that coke produced during heat-up was aged during the 1 h reaction, generating a less porous coke with lower H/C atom ratio and increased aromatic content. XPS and EDX analysis of the recovered coke showed that the 154  coke-catalyst interaction was a physical interaction in which small catalyst particles act as nucleation sites for coke molecules during condensation. The coke-catalyst recovered from different hydroconversion experiments was successfully recycled. Comparison between the catalytic activity of the recycled catalyst and the fresh catalyst showed very similar activities.              155  Chapter 6  MoS2 Catalyst Recycle in Slurry-phase Residue Hydroconversion3 6.1 Introduction Slurry-phase processes have been studied extensively in the last 3 decades, due to their high flexibility in processing different quality feeds, effective mass transfer in the reactor, low pressure drop and high H2 utilization, which translates into lower coke and gas formation compared to fixed-bed hydroconversion processes. The drawback of slurry-phase processes using unsupported catalysts is the uncertainty regarding catalyst recovery and reuse. Catalyst recovery and reuse is not an issue when using low cost metal catalysts such as Fe, but usually this is at the expense of producing low quality liquid and a substantial amount of low value coke. In processes using more expensive metals such as Mo, at metal concentrations above ~300 ppm in the feed, catalyst recovery and reuse is critical in making these processes economically viable [2]. In this chapter, the results of catalyst recycle under high residue conversion conditions in the semi-batch, slurry-phase hydroconversion unit are presented. Catalyst recycle was 3 A version of this chapter has been published previously: Hooman Rezaei, Shahrzad Jooya Ardakani and Kevin J. Smith, Study of MoS2 Catalyst Recycle in Slurry-phase Residue Hydroconversion, Energy&Fuel, 2012, 26, 6540-6550. 156                                                   simulated by recovering the solid coke-catalyst mixture from each hydroconversion experiment. After washing in toluene, drying and crushing, the coke-catalyst mixture was added to the residue oil of a subsequent hydroconversion experiment without any further addition of catalyst to the reactor. The effect of initial catalyst concentration as well as the catalyst precursor on the number of possible recycles before deactivation was investigated. The effect of solids loading in the reactor was also studied and a preliminary analysis of the coke-catalyst mixture is presented. Finally, a possible mechanism of catalyst deactivation during hydroconversion is described. 6.2 Results and Discussions 6.2.1 Activity of Recycled Catalyst It was previously shown (Section  4.3) that more than 90 % of the catalyst initially added to the reactor was present in the solid product (coke) recovered after the hydroconversion reaction. Characterization of the coke-catalyst mixture showed that the metal particles in the coke-catalyst mixture were both sulfided and well-dispersed within the coke matrix. To investigate the effect of initial catalyst concentration on the number of catalyst recycles before catalyst deactivation was observed, both Mo-micelle and Mo-octoate precursors were used to provide a range of Mo concentrations. The activity of the recycled catalyst was assessed in terms of H2 uptake, coke yield, residue and asphaltene conversion and selectivity of the catalyst toward different hydrocarbon cuts in the liquid products. 157  The errors associated with the experimental measurements in the semi-batch reactor are summarized in Table  6.1 and Table  6.2. Table  6.1 reports the measured data form a series of repeated recycle experiments using 300 ppm Mo in the form of the Mo-micelle precursor and shows good repeatability of the recycle experiments. Table  6.1 Comparison of repeated recycle experiments using 300 ppm Mo in the form of Mo-micelle precursor  Fresh experiment  1st Recycle  2nd Recycle  1 2  1 2  1 2 Feed, g 80.23 80.22  80.67 79.85  79.94 79.25 H2 uptake (wt %) 2.05 1.99  2.42 1.98  1.62 2.01 H2S yield (wt %) 1.58 1.29  1.43 1.47  1.03 1.14 Gas yield (wt %) 3.51 4.98  6.08 4.90  7.88 6.40 Liquid yield (wt %) 88.48 87.48  83.77 89.84  73.64 79.22 wt % < 204 ?C 24.92 24.17  28.02 N/A  33.67 28.99 204 ?C?348 ?C 34.55 35.48  36.64 N/A  37.92 37.56 348 ?C?524 ?C 26.85 26.87  25.44 N/A  19.53 21.44 524 ?C+ 13.68 13.48  9.90 N/A  8.88 12.01 Coke yield (wt %) 1.95 1.81  2.52 2.87  11.45 8.11 Mass Recovery (wt %) 93.11 93.18  92.47 96.83  93.48 93.87 Asphaltene (wt %) 60.96 65.14  59.38 53.28  62.20 55.87 TIOR Conversion (wt %) 81.22 81.82  85.55 N/A  75.96 76.45  158  Table  6.2 reports the standard errors of the measured values estimated from a series of 10 repeated experiments conducted at 3 different sets of operating conditions. In addition the C and H recovery percentage from these 10 experiments are reported. Table  6.2 Experimental repeatability in hydroconversion experiments a Parameter Std. Error as a % of the mean H2 uptake, error ? 2.90 % Gas yield, error ? 8.81 % Liquid yield, error ? 1.04 % wt % < 204 ?C ? 3.88 % 204 ?C?348 ?C ? 1.54 % 348 ?C?524 ?C ? 1.22 % 524 ?C+ ? 4.79 % Coke yield, error ? 5.38 % TIOR Conversion, error ? 1.38 %  Mean Value ? STD DEV b Mass Recovery (wt %) 93.63 ? 0.71 Carbon Recovery (wt %) 94.36 ? 2.02 Hydrogen Recovery (wt %) 95.54 ? 4.06  a: Values are calculated based on 10 different repeat experiments using different catalyst concentrations and precursors. b: STD DEV: Standard deviation  159  Mo concentrations of 100, 300 and 600 ppm using the Mo-micelle precursor, and 600 and 1800 ppm using the Mo-octoate precursors were investigated. Figure  6.1 shows the coke yield versus the number of recycles in this series of experiments measured at 445 oC, 13.8 MPa H2 and 1 h reaction time.   Figure  6.1 Coke yield versus recycle number from experiments using different catalyst precursors and Mo concentrations: 100 ppm Mo using Mo-micelle precursor (?), 300 ppm Mo using Mo-micelle precursor (?), 600 ppm Mo using Mo-micelle precursor (?) and 600 ppm Mo using Mo-octoate precursor (?). 0 1 2 3 4 502468101214  Coke yield (wt%)Recycle number160  Figure  6.1 shows that with 100 ppm Mo derived from the Mo-micelle precursor, the hydroconversion of CLVR had a coke yield of 4.8 wt %. In the 1st recycle of the coke-catalyst mixture recovered from this experiment, the coke yield increased to 13 wt %, a clear indication of rapid catalyst deactivation. An increase in the initial Mo concentration to 300 ppm, using the same Mo-micelle precursor, decreased the coke yield to 1.9 wt % compared to the experiment with 100 ppm Mo. The recovered coke, when recycled, had a coke yield of 2.5 wt % but this rapidly increased to 11.4 wt % in the 2nd recycle, indicating catalyst deactivation after the 1st recycle. Figure  6.1 also shows that with 600 ppm Mo using both Mo-micelle and Mo-octoate precursors, a significant increase in the number of recycles was possible before a loss in the catalyst activity and increased coke yield was observed.  With 600 ppm Mo derived from the Mo-micelle and Mo-octoate precursors, the coke-catalyst mixture was recycled 4- and 3-times, respectively, while the catalyst maintained its activity in terms of coke suppression. After the 4th recycle, both catalysts were deactivated and the coke yield increased to more than 8 wt %. In  Chapter 5 it was shown that Mo-micelle and Mo-octoate precursors, when added to the reactor in the form of fresh precursors, have very similar activity in terms of coke suppression, residue conversion, H2 uptake and liquid yield. The differences in performance of these two precursors in the 4th recycle experiment are believed to be due to differences in the chemical and morphological properties of the generated coke. The properties of the generated coke during recycle are described and discussed in more detail in  Chapter 7. 161  One effect that should be noted in Figure  6.1 is that in the experiments with high catalyst concentrations (600 ppm Mo using Mo-micelle and Mo-octoate), the coke yield reached a minimum in the 1st recycle. This is probably because with a high initial catalyst concentration, a substantial amount of catalyst is available for H transfer to the free radicals generated during thermal cracking. Consequently, coke formation is suppressed and coke yields similar to the coke yield from experiments using the fresh precursor are observed. Previous studies [43] have shown that the asphaltene (which is the main precursor of toluene insoluble coke) recovered from the liquid product of a hydroconversion reaction can be further cracked when recycled. Similarly, a portion of the toluene insoluble coke, which consists of both reactive amorphous coke and more refractory coke, may react and crack during recycle. The coke yield from the recycle experiments was calculated from the total mass of coke recovered after the hydroconversion reaction minus the mass of coke added to the reactor before the recycle experiment. If the recycled catalyst is very active and generation of new coke in the 1st recycle is the same as the coke generated with fresh catalyst precursor, the consumption of the recycled coke in the recycle experiment will result in a reduced coke yield. As the number of recycles increases, consumption of the recycled coke continues, but higher rates of coke formation due to the gradual deactivation of the catalyst, increases the coke yield. The H2 uptake during hydroconversion is an important indicator of catalyst hydrogenation activity. Upgrading of heavy-oil and residue starts with a thermally driven C ? C bond cleavage (when no catalyst with strong acidic sites is available) [87]. Subsequently, H2 transfer from the gas phase to the thermally produced free radicals and olefins stabilizes 162  and hydrogenates the cracked compounds. The transfer of H2, only a small portion of which is direct H2 transfer to highly reactive free radicals, prevents polymerization and condensation reactions of large molecules and PAHs [87] yielding toluene insolubles. In the presence of excess free radical and olefinic compounds (which are generated during cracking and ?-scission reactions), H2 consumption is expected to be proportional to the extent of coke suppression achieved during hydroconversion. Figure  6.2 shows the changes in the H2 uptake from different recycle experiments in which different initial Mo concentrations and Mo precursors were used. In all cases, H2 uptake showed a noticeable decrease with recycle number. With 100 ppm Mo using the Mo-micelle precursor initially, the decrease occurred in the 1st recycle. For experiments with 300 and 600 ppm Mo using Mo-micelle or Mo-octoate precursors initially, the decrease occurred on the 2nd and 5th recycle respectively. H2 uptake using the recycled catalysts at Mo concentrations of 300 and 600 ppm showed a maximum. The maximum is possibly because in the 1st or 2nd recycle the catalyst remains active, whereas cracking of the recycled coke produces significantly more molecules with unsaturated bonds and free radicals which are a potential source of H2 consumption. 163   Figure  6.2 H2 uptake versus the recycle number in recycle experiments starting with different Mo concentrations using Mo-micelle and Mo-octoate precursors: 100 ppm Mo using Mo-micelle (A), 300 ppm Mo using Mo-micelle (B), 600 ppm Mo using Mo-micelle (C) and 600 ppm Mo using Mo-octoate (D).  Figure  6.3 shows the correlation between the coke yield and H2 uptake for all the recycle experiments done in the present study. Clearly, an increase in H2 uptake is correlated 0 1 2 3 4 51.701.801.902.002.101.601.802.002.201.601.802.002.202.401.801.952.102.25  Recycle numberAB  H2 uptake (wt%)C D  164  with a decrease in coke yield. H2 uptakes higher than a certain value in the experiments of this study (~ 2.1 wt %, shown by the vertical line in the Figure  6.3) appear not to have very much effect on coke suppression (data points shown in the circle in Figure  6.3). Generally higher H2 consumption is an indication of increased coke suppression provided there are sufficient cracked molecules in the reactor that need to be hydrogenated. However, if the cracking rate is not very high (such as at low temperature or with low carbon content of the feed oil) while the rate of hydrogenation is high due to the excess catalyst in the reactor, then H2 is used for hydrogenation that increases the H/C atom ratio of the liquid and solid products (Figure  5.3). Also, excessive hydrogenation of the liquid in the reactor reduces asphaltene stability, which eventually causes more coke formation. Under these conditions, an increase in H2 uptake has little effect on the extent of coke suppression (Section  5.2.1 and [17,95]).  Figure  6.3 Coke yield versus H2 uptake in the recycle experiments. 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.80246810121416Coke yield (wt%)Hydrogen uptake (wt%)165  In the TIOR conversion calculation (equation 3.4), the generated coke is considered as part of the residue product of the reaction, based on its high boiling point (above 524 ?C). TIOR conversions from the recycle experiments using Mo-micelle and Mo-octoate as catalyst precursors are shown in Figure  6.4. Consistent with the coke yields presented in Figure  6.1, a significant decrease in TIOR conversion was observed at the point when the catalyst deactivated (the last recycle experiment for each Mo concentration). The TIOR conversions at different Mo concentrations using fresh precursor (first data points with recycle number zero) are all in the range of 81 ? 85 wt %. The high level of TIOR conversion in the first experiment was maintained during recycling provided the catalyst was still active. With a Mo concentration of 100 ppm (Mo-micelle precursor), the TIOR conversion decreased to 74 wt % in the 1st recycle (coke yield from this experiment also increased as presented in Figure  6.1). The deactivation was observed after the 2nd and 4th recycle experiments with 300 ppm and 600 ppm Mo in the form of Mo-micelle, respectively. Recycle experiments using the Mo-micelle precursor had a slightly higher TIOR conversion compared to the recycle experiments using the Mo-octoate precursor. Comparing the TIOR conversion data presented in Figure  6.4 with the coke yield data shown in Figure  6.1, it can be concluded that catalyst deactivation decreased the TIOR conversion significantly. The decrease in TIOR conversion was because of the substantial amount of coke generated in the experiments in which the catalyst was deactivated. 166   Figure  6.4 TIOR conversions in different recycle experiments starting with different Mo concentrations and precursors. (A) 100 ppm Mo using Mo-micelle precursor, (B) 300 ppm Mo using Mo-micelle precursor, (C) 600 ppm Mo using Mo-micelle precursor and (D) 600 ppm Mo using Mo-octoate precursor.  Figure  6.5-A, Figure  6.5-B, Figure  6.5-C and Figure  6.5-D present the naphtha (< 204 ?C boiling point), light gas oil (204 - 348 ?C boiling point), heavy gas oil (348 - 524 ?C 0 1 2 3 4 5 672768084727680847276808472768084 TIOR conversion (wt%)Recycle numberAB C D  167  boiling point) and residue (> 524 ?C) yields in the liquid products from the recycle experiments with Mo concentrations of 100, 300 and 600 ppm using the Mo-micelle precursor and 600 ppm using the Mo-octoate precursors. Starting with 100 ppm Mo using the Mo-micelle precursor (Figure  6.5-A), selectivity toward different hydrocarbon products did not change very much in the recycle experiment. This is possibly because the experiment in which 100 ppm Mo using Mo-micelle precursor was used had a high (~ 5 wt %) coke yield and catalyst concentration was not high enough to suppress coking. At higher initial Mo concentration (300 and 600 ppm), the coke yield of the 1st experiment (experiment in which fresh precursor was added to the reactor) was significantly lower than the coke yield of the experiment in which the catalyst was deactivated (last experiment of each recycle series). This significant change in the coke yield had a clearer and more visible impact on the selectivity change of the hydroconversion reaction toward different hydrocarbon cuts in the liquid products. With an increase in the number of recycles, the yield of fractions heavier than naphtha (light gas oil, heavy gas oil and residue) decreased in the liquid products. This shift is especially noticeable in the last experiment of the recycle series with a Mo concentration of 600 ppm using both Mo-micelle (Figure  6.5-C) and Mo-octoate (Figure  6.5-D) precursors. The decrease is accompanied by a shift toward lighter products (the yield of naphtha increased) due to a high coke yield (Figure  6.1) and carbon rejection in the hydroconversion reactor when the catalyst is deactivated in the last recycle experiment. The results are in agreement with previous work where a less active catalyst precursor resulted in high coke yield and produced lighter liquid product than the experiment in which less coke was produced ( Chapter 4). Similarly, Del Bianco et al. [43] investigated recycling of Mo-168  based dispersed catalysts in a slurry-phase batch reactor. Unlike the present study, they used very mild hydroconversion conditions (410 ?C) and an initial Mo concentration of 3000 ppm. Furthermore, instead of recycling the solid coke-catalyst, they recycled the n-C5 asphaltene-catalyst mixture, but reported a similar trend with a decreased heavy gas oil and residue fraction of the liquid product with increased recycling. 169   Figure  6.5 Selectivity change toward different hydrocarbon cuts in the recycle experiments starting with 100 ppm Mo using Mo-micelle precursor (A), 300 ppm Mo using Mo-micelle precursor (B), 600 ppm Mo using Mo-micelle precursor (C) and 600 ppm Mo using Mo-octoate precursor (D). (?) < 204 ?C, (?) 204 ? 348 ?C, (?) 348 ? 524 ?C and (?) > 524 ?C. 0 1 2 3 4 50102030010203001020300102030 Recycle numberAB C D  Hydrocarbon yield (wt%)170  Transformation of asphaltene molecules is known to be the main mechanism of coke (toluene insoluble hydrocarbons) formation during hydroconversion. The transformation occurs through the formation of a mesophase, an aromatic rich liquid phase, and liquid-liquid phase separation in the hydroconversion reactor [5,87,96]. It is also known that this transformation is irreversible and fast [95] but it can be partially prevented if an active hydrogenation catalyst is present in the hydroconversion reactor to suppress the polymerization of the polyaromatic cores of the asphaltenes. Hydrogenation results in the production of hydrocarbons with fewer aromatic rings in their structure and prevents mesophase formation and liquid-liquid phase separation. Figure  6.6 shows the asphaltene conversion measured during catalyst recycle experiments. Similar to the coke yield from these experiments (Figure  6.1), when the catalyst deactivated (the last experiment in each recycle series), the asphaltene conversion decreased noticeably. The decrease is due to a low rate of H2 transfer to the liquid for conversion of the asphaltene molecules into lighter liquid products. This mechanism is also supported by the high asphaltene conversion level from experiments in which fresh precursors were used (recycle number zero) when the initial Mo concentration increased from 100 ppm to 300 ppm and 600 ppm. The effect of catalyst concentration on the asphaltene conversion is much clearer when asphaltene conversion in a thermal experiment (in which no catalyst was used; the hollow circle data points in Figure  6.6) is compared with the asphaltene conversions of any of the catalytic experiments. 171   Figure  6.6 Asphaltene conversion in recycle experiments using different initial catalyst concentrations and precursors: 100 ppm Mo using Mo-micelle precursor (?), 300 ppm Mo using Mo-micelle precursor (?), 600 ppm Mo using Mo-micelle precursor (?), 600 ppm Mo using Mo-octoate precursor (?) and a thermal experiment (?).  The asphaltene conversion data for 600 ppm Mo, using the Mo-micelle and Mo-octoate precursors, shown in Figure  6.6, suggest that the Mo using Mo-micelle precursor had a higher asphaltene conversion compared to the Mo-octoate precursor and this trend is reversed when the number of recycles is greater than 3. However, taking into account the error associated with the asphaltene conversion data of these experiments (73.55 ? 9.39 wt % and 0 1 2 3 4 5405060708090  Asphaltene conversion (wt %)Recycle number172  65.80 ? 7.05 wt % respectively for experiments using 600 ppm Mo in the form of fresh Mo-micelle and Mo-octoate precursors measured in 8 different repeat experiments) it is concluded that there is minimal difference in the asphaltene conversion between the two precursors. 6.2.2 Optimum Coke Recycle Ratio Calculations in a Commercial Plant To calculate the optimum recycle ratio of hydroconversion coke, hydroconversion experiments should be done at the reaction conditions of the commercial plant to identify the number of recycles at which the coke is deactivated. Assuming the same deactivation behavior observed in the present study (600 ppm Mo using Mo-micelle precursor was deactivated after 4 successful recycles in the 5th recycle), the following methodology can be used to determine the minimum and maximum recycle ratio of hydroconversion coke to be mixed with fresh catalyst at the reactor entrance. Assuming the desired catalyst concentration in the reactor is ? and ? is the recycle ratio of the coke (0< ?<1), ?. ? will be the concentration of the catalyst in the coke being recycled for the 1st time and (? ? ?. ?) is the required fresh catalyst to be added to the reactor. Assuming 200 ppm of fresh Mo is the maximum Mo concentration that can be added to the reactor at each cycle to make the hydroconversion plant commercially viable (?) and assuming 600 ppm of Mo as the desired Mo concentration in the reactor, The minimum recycle ratio constraint is determined as follows: (? ? ?.?) ? ?      ?       (? ? ?) ???    ?     (? ? ?) ???   ?    ? ? ?.?? 173  Based on the reaction conditions of the hydroconversion unit, there is a minimum amount of catalyst that is needed to have the minimum and acceptable coke yield. For example in the present study, as presented in Figure  5.1, 300 ppm of Mo is the minimum Mo concentration required in the slurry-phase reactor to have a minimum coke yield (?). The Mo concentration in the coke being recycled from the experiment in which fresh Mo was used (assuming ? as the desired Mo concentration in the reactor and ? as the recycle ratio of the coke) will be equal to ?. ? and ?. ?2,?. ?3,?. ?4, ?. ?5 and ?. ?6 from the 1st recycle, 2nd recycle, 3rd recycle, 4th recycle and 5th recycle experiments, respectively. Knowing that the recycle coke from the 4th recycle in the present study (which was used in the 5th recycle and showed poor catalytic activity) is deactivated, the following inequality can be written: ?? ? ?.??? ? ?  ?    ?? ? ??? ???  ?    ?? ? ?? ????    ?    ?? ? ?.?   ?    ? ? ?.?? So the coke recycle ratio assuming 600 ppm Mo concentration in the reactor, catalyst deactivation after the 4th recycle, a maximum make-up Mo concentration of 200 ppm should be between 66 ? 87 wt % of the generated coke. 6.2.3 Significance of Solid Loading In all of the recycle experiments reported herein, the coke-catalyst mixture recovered from the hydroconversion experiment was washed, dried and ground to a powder and then used as the catalyst for the subsequent hydroconversion experiment.  In this way, catalyst recycle was simulated and the activity of the recycled catalyst determined. The coke yield from the hydroconversion experiments with recycle catalyst was calculated based on the net 174  amount of coke produced in the hydroconversion experiment. To ensure that all the Mo catalyst initially added to the reactor was used in the subsequent recycle experiments, all of the solid coke-catalyst recovered from the hydroconversion experiment was used in the next recycle hydroconversion experiment. Consequently, as the number of recycle experiments increased, the solids loading in the reactor increased. Figure  6.7 compares the amount of coke added to the reactor, to that generated and the total coke in the reactor after recycling for a series of recycle experiments, starting with 600 ppm Mo using the Mo-micelle precursor.  Figure  6.7 Coke added (?), generated (?) and total coke in the reactor (?) in a series of recycle experiments starting with 600 ppm Mo added to the reactor using Mo-micelle precursor. 0 1 2 3 4 502468101214161820  Coke (wt% of liquid feed)Recycle number175  In all the recycle experiments, the maximum initial solids loading in the reactor was < 7.5 wt %. To investigate any potential solids loading effect on the coke yield of the recycle experiments, several experiments with variable amounts of coke added to the reactor along with 600 ppm Mo using the Mo-octoate precursor were conducted. In this series of experiments, the coke added to the reactor originated from a thermal hydroconversion experiment in which no catalyst was used. The standard washing, drying and grinding procedure was performed on these coke samples prior to recycling. Figure  6.8 shows the coke yield versus solid coke loading in the reactor for these experiments. With an increase in solids loading up to 7.5 wt %, the coke yield decreased linearly. However, when the solids loading was increased further to 10 and 12.5 wt %, the coke yield from the hydroconversion experiment increased rapidly. The linear decrease in coke yield with increased solids loading is due to cracking of the added coke, a portion of which yielded liquid or gas. Assuming a constant generation of coke in the reactor for all of the experiments (600 ppm Mo was used in these experiments) and assuming that a portion of the added coke cracks under hydroconversion conditions, a linear decrease in coke yield with coke loading would be expected. 176   Figure  6.8 Coke yield versus solid loading in the reactor using thermal coke and 600 ppm Mo added to the reactor in the form of Mo-octoate. Straight line shows linear decrease of the coke yield with solid loading in the reactor at solid loading range of zero to 7.75 wt % of coke in the feed.  Another very important observation from this series of experiments was the net zero coke yield when 7.5 wt % of coke was added to the reactor with 600 ppm Mo using the Mo-octoate precursor (Figure  6.8). This observation, the validity of which was confirmed by repeated experiments (presented in Figure  6.8), suggests that the coke generated during hydroconversion equals the amount of solid coke that is converted to liquid or gas at a coke loading of 7.5 wt %. This observation can be explained based on the concept of the presence of two types of coke in the recovered solid, one of which has an amorphous structure and is 0 2 4 6 8 10 12 14-0.50.00.51.01.52.02.53.03.54.0  Coke yield (wt%)Coke loading (wt%)177  soluble in quinolone and the other which is more calcined and graphitic in nature. The graphitic portion will have a high resistance to cracking when recycled while the amorphous coke can be easily cracked. The amorphous portion of the recovered coke is most likely responsible for the decreased coke yield observed in the experiments when the solid loading in the reactor increased. Note that a coke yield of 1.9 wt % was obtained with 600 ppm Mo added to the reactor in the form of Mo-octoate (Figure  6.1). If one assumes a consistent amount of coke converted to gas or liquid by cracking of the added coke in the recycle experiment, an approximate calculation suggests that ~ 25 wt % of the added coke (originally from a thermal experiment) was converted to liquid and gas. The results presented in Figure  6.8 are also consistent with the results presented for coke recycle starting with 600 ppm Mo added to the reactor in the form of both Mo-micelle and Mo-octoate precursors (Figure  6.1 and Figure  6.7). The coke yield (from the 1st recycle of the coke-catalyst mixture) was lower than the coke yield from the experiment in which fresh precursor was used. This observation can now be explained based on the findings from the solid loading hydroconversion experiments. Since in the 1st recycle experiment, recycle coke cracked and the coke generation was low due to a highly active and available catalyst, the net coke yield from the 1st recycle experiment was lower than the coke yield of the experiment in which no coke was added to the reactor and fresh precursor was used. The proposed mechanism of coke consumption in the 1st recycle, although observed in other experiments (Figure  6.1), was confirmed by using pure graphite (graphite is much less susceptible to cracking compared to the coke generated by CLVR hydroconversion) instead of the thermal coke. In this experiment, pure graphite (7.5 wt %) along with 600 ppm of Mo 178  using the Mo-octoate precursor were added to the residue oil and the hydroconversion reaction was carried out under the same conditions as before.  Figure  6.9 shows a comparison of the coke yields from these two experiments. Clearly, the graphite resulted in an increased coke yield to ~ 4 wt % compared to the average coke yield of -0.1 wt % (average of two experiments) using the thermal coke. This is because the graphite does not crack to any great extent compared to the coke from the CLVR hydroconversion.  Figure  6.9 Comparison of coke yields in two experiments using 6 g of thermal coke and pure graphite along with 600 ppm Mo using Mo-octoate precursor in the reactor. Reported values are average of two repeat experiments.  Thermal coke -- Pure graphite-0.10.03.03.54.0  Coke yield (wt%)Origin of solid179  The results presented in Figure  6.8, suggest that continuous recycle of catalyst in hydroconversion experiments with zero net coke yield is feasible. However, as discussed in Section  6.2.4, coke aging changes the chemical properties of the coke, making it more graphitic and hence less susceptible to cracking. This results in less and less cracking of the recycled coke as the number of recycles increases and consequently, an increase in the net coke yield would eventually occur. When the solid loading was increased to 10 or 12.5 wt % in the reactor, the coke yield increased significantly to above 3 wt % (Figure  6.8). Although no supporting experimental data are available, it is believed that this increase was a consequence of highly disturbed hydrodynamics in the reactor due to the very high solids loading. 6.2.4 Catalyst Deactivation The data presented in Figure  6.1 showed that the initial Mo concentration in the reactor had an effect on the number of recycles before deactivation was observed and that this number increased as the initial catalyst concentration increased. With an increase in the initial catalyst concentration, more catalyst is available at the surface and pores of the recycled catalyst-coke mixture. Hence hydrogenation reactions and transfer of H to unsaturated bonds and free radicals produced during cracking occurs which eventually stabilizes the highly reactive chemical species in the reactor. Starting with a very low catalyst concentration (for example, 100 ppm Mo), the catalyst concentration in the recovered coke-catalyst mixture was very low, due to the low initial Mo concentration and a relatively high coke yield. Consequently, very few active sites would be available on the surface of the coke 180  or within the pores with large enough diameter for macromolecules present in the feed to diffuse and react with H2. When the initial catalyst concentration in the reactor increased, although part of the catalyst was totally covered with coke and therefore would not be available for hydrogenation reactions, there remains sufficient catalyst to participate in hydrogenation reactions and transfer H to free radicals and olefins and thereby suppress coke formation. According to the above, there would be an optimum Mo concentration in the recovered coke-catalyst mixture. Mo concentrations above the optimum level would be in excess having the same performance in terms of coke suppression. On the other hand, when the Mo concentration in the coke-catalyst mixture drops below the optimum, the catalyst activity would drop due to a lack of active sites for hydrogenation. To confirm this mechanism, the coke yield from all recycle experiments versus the Mo concentration in the coke-catalyst mixture was plotted as shown in Figure  6.10. The Mo concentration of the coke was calculated by dividing the mass of Mo added to the reactor by the total mass of coke recovered at each recycling step. Since in each recycle experiment some coke was generated, the weight of coke continuously increased as the number of recycles increased (Figure  6.7). This yields a continuous decrease of Mo concentration in the coke-catalyst mixture as the number of recycles increases. As shown in Figure  6.10, at low concentrations of Mo in the recycle coke, the coke yield decreased significantly with an increase in Mo concentration. When the recycled coke-catalyst mixture had a Mo concentration of ~1.2 wt % and higher (shown by the vertical line 181  in Figure  6.10), the coke yield remained low (1.5 - 2.5 wt %). Hence a Mo concentration of more than 1.2 wt % in the coke provides excess hydrogenation capability in the reactor. On the other hand, when the Mo concentration dropped below 1.0 wt %, the coke yield increased significantly, probably due to a lack of active sites for hydrogenation.  Figure  6.10 Coke yield versus Mo concentration in the coke-catalyst mixture used as the catalyst in different recycle experiments using different initial concentrations of Mo added to the reactor in the form of Mo-micelle and Mo-octoate precursors. Vertical line shows the Mo concentration in the coke above which the coke yield will not change with Mo concentration in the recycled coke.  Based on the results presented in Figure  6.10, the question arises as to whether or not Mo concentration in the recycled coke is the only variable affecting the recycled catalyst 0 1 2 3 4 5 602468101214  Coke yield (wt%)Mo concentration in the coke (wt%)182  activity. If so, then increasing the Mo concentration in the feed would increase the number of recycles before catalyst deactivation would be observed. Hence, a series of recycle experiments using 1800 ppm Mo derived from Mo-octoate as the catalyst precursor were completed and the results are shown in Figure  6.11. With 1800 ppm Mo, the catalyst was recycled 4 times without significant loss in activity in terms of coke suppression but, surprisingly, the catalyst deactivated in the 5th recycle (coke yield ~ 6 wt %). This was while some of the coke-catalyst mixture with lower Mo concentrations in the coke in comparison to the Mo concentration of coke recovered from 5th recycle of 1800 ppm Mo using Mo-octoate precursor (with ~ 2.5 wt % of Mo in the coke sample) had far lower coke yield (data points shown in the square in Figure  6.11). Although these coke samples had Mo in the coke, they were recovered from experiments in which the number of recycles was low (1st, 2nd and/or 3rd recycle). Hence, although the initial catalyst concentration affects the number of possible recycles, coke aging also appears to decrease the recycled catalyst-coke activity. Coke aging during several recycle experiments could change the morphology and chemical properties of the coke. These changes reduce the likelihood that the coke will crack and be consumed (transformed into liquid or gas phase) during recycle. Furthermore, a change in morphology can potentially increase the resistance to diffusion of macromolecules into the coke pores to reach the catalyst hydrogenation sites and may also result in encapsulation of the MoS2. 183   Figure  6.11 Coke yield versus Mo concentration in the coke-catalyst mixture used as the catalyst in several series of recycle experiments using different initial concentrations of Mo added to the reactor in the form of Mo-micelle and Mo-octoate precursors.  Figure  6.12 shows the XRD diffractograms of coke samples recovered after each hydroconversion experiment and before recycle. There are several features in Figure  6.12 which give very useful insight toward identifying the deactivation mechanism of the recycled catalyst. Two peaks at 59? and 33? are assigned to the (110) and (100) crystal planes of MoS2. As seen in Figure  6.12, the peak intensities decreased as the number of recycles increased because increased recycle increased the total amount of coke in the reactor (Figure  6.7) and this decreased the Mo concentration and consequently, the MoS2 concentration in the coke 0 1 2 3 4 5 6 7 8024681012  Coke yield (wt%)Mo concentration in the coke (wt%) 1800 ppm Mo-octoate184  sample. A noticeable feature of the MoS2 XRD peaks is that they are broad. This is an indication that during recycling there was no significant MoS2 crystal size change or agglomeration of MoS2. The absence of the (002) reflection of the MoS2 at 14? is also an indication of highly dispersed MoS2 with few layers [125].  Figure  6.12 XRD diffractogram of coke samples recovered after different hydroconversion experiments in a series of recycle experiments using initial Mo concentration of 1800 ppm added to the reactor in the form of Mo-octoate precursor.  0 20 40 60 80 1005th recycle4th recycle3rd recycle2nd recycle1st recycleGraphite-2HGraphite-2HNi0.96SNi0.96S   &Fe1-XSFe1-XSV3S5  Intensity (a.u.)Angle (2?)MoS2 (110)Fe1-XSFresh coke32 34 36Intensity (a.u.)Angle (2?)185  All the peaks that are not labeled in Figure  6.12 were assigned to Ni, V and Fe sulfides. These peaks had almost the same intensity as the number of recycles increased since the metal sulfides originated from the residue oil and their concentration would be constant in the recovered coke from different experiments. Another important observation from Figure  6.12 is the change in the shape of the peak at ~ 26? assigned to graphite. This peak is an indication of the presence of graphitic carbon in the coke samples. As shown in Figure  6.12, the peak becomes larger and sharper as the number of recycles increased. This indicates that the recovered coke becomes more graphitic with increased recycling at high pressure and temperature. This transformation makes the coke less susceptible to cracking and dissociation under reaction conditions and consequently, will result in more MoS2 trapped in the highly stable matrices of the graphitic coke. 6.3 Conclusions The coke-catalyst mixture recovered from hydroconversion experiments using CLVR was successfully recycled several times in subsequent hydroconversion experiments. Results from the recycle experiments showed that with a high initial Mo concentration in the feed (> 600 ppm), the recovered coke-catalyst mixture could be recycled ? 4 times. The concentration of Mo, as well as the coke morphology, were identified as the main factors that determined the potential for the catalyst recycle in the hydroconversion of CLVR. Experiments done to investigate the effect of solid loading in the reactor showed that for the range of operating conditions investigated, solid loading did not have a significant effect on 186  the measured catalyst activity. However, at high solid loading (> 8 wt %), a significant decrease in efficiency of the MoS2 catalyst for coke suppression was observed.               187  Chapter 7  Conceptual Deactivation Model of Recycled MoS2 Catalyst 7.1 Introduction Catalyst deactivation during residue oil upgrading and hydrotreating is a major problem in the petroleum refinery industry. Loss of catalytic activity results in either unit shut down for catalyst regeneration (if possible) or utilization of costly alternatives such as parallel reactors or continuous regeneration of catalyst [142]. Among several factors that may cause catalyst deactivation, the formation of highly carbonaceous deposits on the catalyst surface is believed to be one of the main reasons for catalyst deactivation in residue oil hydroprocessing [143,144]. The problem of carbonaceous deposit formation becomes more serious when the process operates at high residue conversions (> 60 %) [101]. The mechanism of formation of the carbonaceous deposits is not very well understood, mainly because of the complexity of the feed, the numerous reactions occurring simultaneously and the number of factors (operating conditions) affecting the process [145,146]. In the case of supported catalysts used in the hydroconversion and hydrotreating of crude oils and residue oil, the number of studies done on carbon-supported catalysts are far fewer than those done on ?-Al2O3-supported or zeolite-supported catalysts [146]. Although the deactivation mechanism is possibly the same for all of these supports (coke formation on the active site deposited on the support), slower coke formation on carbon-supported catalysts compared to ?-Al2O3-supported catalysts during hydroprocessing is believed to be 188  the reason for a slower deactivation rate observed on carbon-supported catalysts compared to ?-Al2O3-supported catalysts. The slower coke formation is attributed to the neutral surface of the carbon-supported catalysts compared to the acidic ?-Al2O3 support [146]. Catalyst deactivation in slurry-phase hydroconversion, in which unsupported catalysts are used, is likely different to the deactivation observed on supported catalysts. Clearly, the chemical properties of the support are not a factor in this case. Although numerous studies have investigated the mechanisms of catalyst deactivation and have developed new techniques in characterization of deactivated catalysts used in hydroprocessing [106,132,137,142,143,145,147-153], no study on the deactivation mechanism of unsupported catalysts used in slurry-phase hydroconversion is known to the author. This is likely because slurry-phase processes are based on once-through catalyst use and catalyst deactivation is not significant due to the very short residence time (~ 1 h) of the catalyst in the slurry-phase reactor. Commercialization of slurry-phase processes requires catalyst recycle to decrease the high catalyst consumption and cost [2,5,6,154] and under recycle conditions, catalyst deactivation may be important. In  Chapter 4, it was shown that the coke recovered from residue oil hydroconversion experiments done in batch and semi-batch reactors, contained almost all of the unsupported MoS2 catalyst added to the reactor initially. Furthermore, it was shown ( Chapter 6) that the catalyst captured by the toluene-insoluble coke (namely the coke-catalyst mixture) could be recycled several times in the semi-batch reactor at the hydroconversion conditions of this 189  study before catalyst deactivation was observed, provided that the metal concentration in the coke was above 1.2 wt % under hydroconversion conditions of the present study. In this chapter, the characterization of the coke-catalyst mixture recovered at different stages of recycle, combined with recycle catalyst activity results, are reported. Based on the characterization results, a mechanism of catalyst deactivation during residue oil hydroconversion in a slurry-phase reactor is proposed. 7.2 Results and Discussions 7.2.1 Activity of recycled catalyst The catalytic performance of the recycled coke-catalyst mixture in the hydroconversion of residue oil in a slurry-phase semi-batch reactor under high residue conversion conditions was reported in  Chapter 6. The study showed that catalyst deactivation (as reflected in increased coke yield) was determined in part by the concentration of catalyst in the reactor. Figure  6.1 demonstrates the correlation generated from the recycle hydroconversion experimental data. Note that although the catalyst concentration is reported in terms of Mo, the active phase of the catalyst is in the form of MoS2 [2,5]. Observations from Figure  6.1 and Figure  6.11 and the discussion presented in Section  6.2.4 suggest that the initial Mo concentration in the feed cannot be solely responsible for loss in catalytic activity of the recovered coke-catalyst mixture. An increase in the initial Mo concentration in the feed can only increase the number of successful recycle 190  experiments to a limited extent. Hence it is concluded that the age of the coke-catalyst mixture also plays a role in determining the activity of the recycled catalyst. 7.2.2 TEM Analysis of the Fresh and Recycled Coke Previously in  Chapter 4, it was shown that the recovered coke-catalyst mixture after each hydroconversion experiment contained all the catalyst added to the reactor. Catalyst particle migration on the support surface (in the case of the present study, the coke matrix) and agglomeration during hydroconversion could result in an increase in MoS2 domain size, decreased catalyst surface area and consequently, the number of available active sites. The dispersion and possible agglomeration of MoS2 particles within the coke matrix was therefore studied. Using 1800 ppm Mo in the reactor, the MoS2 could be easily observed in the recovered coke-catalyst mixture using TEM (Figure  7.1-A). However, for Mo concentrations of ? 600 ppm, observation of the MoS2 layers within the coke-catalyst mixture was not always possible (Figure  7.1-B to Figure  7.1-E). This was due to the low concentration of MoS2 in the recovered coke-catalyst mixture after the hydroconversion experiment. As the number of recycles increased with 600 ppm Mo using the Mo-micelle precursor, the presence of MoS2 in the TEM images could not be readily identified (Figure  7.1-C and Figure  7.1-E). TEM images of the coke samples recovered from the recycle experiments (Figure  7.1-C, Figure  7.1-D and Figure  7.1-E) suggests that the MoS2 layers did not agglomerate when the number of recycles increased. 191        192   Figure  7.1. TEM micrograph of coke samples recovered from different hydroconversion experiments. (A) 1800 ppm Mo using Mo-octoate precursor, (B) 600 ppm Mo using fresh Mo-micelle precursor, (C) 1st recycle of 600 ppm Mo using Mo-micelle precursor, (D) 2nd recycle of 600 ppm Mo using Mo-micelle precursor, (E) 3rd recycle of 600 ppm Mo using Mo-micelle precursor. T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time.  7.2.3 XRD Analysis of the Fresh and Recycled Coke TEM results were consistent with the XRD analysis of the coke samples recovered from different recycling steps. To identify MoS2 by XRD, coke samples recovered from different recycling steps using 1800 ppm Mo (derived from the Mo-octoate precursor) were examined and are shown in Figure  6.12. Fresh coke in Figure  6.12 refers to the coke sample recovered from the hydroconversion experiment in which fresh Mo-octoate precursor was used. The (110) plane of MoS2 shows a peak at 59?. This peak had a significant intensity in the coke sample recovered from the experiment in which fresh catalyst precursor was used. 193  As the number of recycles increased, the MoS2 peak intensity decreased. This simply reflects a decreased MoS2 concentration in the coke as the amount of coke generated after each recycle increases whereas the amount of MoS2 is fixed corresponding to the amount added initially as fresh catalyst. Although the intensity of the MoS2 peak decreased with recycle number, there was no indication of peak sharpening. This indicates that the crystallite size of the MoS2 did not increase significantly as the number of recycles increased.  On the other hand, the intensity of the peaks assigned to iron sulfide (44?, 53? and 71?), nickel sulfide (30? and 44?) and vanadium sulfide (45.5?) did not change when the number of recycles increased. Since these metals were not added to the reactor externally but originated from the feed, their concentrations in the generated coke remained constant after each recycle. Peaks at 25.8? and 34.9? are assigned to graphitic carbon. Note that the intensity of these peaks increased as the number of recycles increased. The peaks were also much sharper when comparing the peaks in the coke sample recovered from the 5th recycle with the coke sample recovered from the experiment in which fresh catalyst precursor was used. This is an indication of coke aging and transformation of coke formed in the first experiment to a more refractory, graphitic coke during several recycle experiments under high pressure and high temperature conditions. 7.2.4 CHN Analysis of the Fresh and Recycle Coke Figure  7.2 shows the H/C atom ratio of the coke samples recovered from different recycles using different initial Mo concentrations. The H/C atom ratio of the coke samples 194  from experiments in which fresh catalyst precursor was used varied from 0.71 to 0.75, depending on the catalyst concentration and precursor type. When a higher concentration of catalyst was used in the reactor, a more hydrogenated coke was produced due to increased rate of the hydrogenation reactions ( Chapter 5). Figure  7.2 shows that the H/C atom ratio of the coke sample recovered from the experiment in which 600 ppm Mo (derived from the Mo-micelle precursor) was used, was high and similar to the H/C atom ratio of the coke sample recovered from the experiment in which 1800 ppm Mo (derived from  Mo-octoate) was used. This is due to the formation of NH4Cl during the synthesis of the Mo-micelle precursor. NH4Cl is a highly hydrogenated compound, which boosts the H/C atom ratio of the recovered coke sample. As shown in Figure  7.2, all the coke samples recovered from experiments in which fresh catalyst precursor was used had a very high H/C atom ratio (> 0.7). As the number of recycles increased, the H/C atom ratio decreased significantly to about 0.59. This indicates that the generated coke became more carbonaceous with increased time-on-stream (increased number of recycles). An interesting observation from Figure  7.2 is that the large difference between the H/C atom ratio of the coke samples recovered from the fresh experiment decreased with recycling, and after 5 recycles all the recovered coke samples had the same H/C atom ratio.  195   Figure  7.2. H/C atom ratio change of the coke samples recovered from different steps of recycle experiments in several series of recycle experiments using different Mo concentrations and precursors. T = 445 ?C, PH2 = 13.8 MPa at 900 mL(STP)/min and 1 h reaction time.  The N content of the coke samples recovered from different recycles showed a minimum as the number of recycles increased. As an example, the N content of the coke samples recovered from recycle experiments using 600 ppm Mo using the Mo-micelle precursor and 1800 ppm Mo using the Mo-octoate precursor are shown in Figure  7.3.  0.560.580.600.620.640.660.680.700.720.740.760.78   600 ppm Mo-micelle 600 ppm Mo-octoate 1800 ppm Mo-octoateH/C atom ratioRecycle #Fresh 1st 2nd 3rd 4th 5th196   Figure  7.3 N content change of coke sample during different steps of recycling using 1800 ppm Mo using Mo-octoate precursor (A) and 600 ppm Mo using Mo-micelle precursor (B).  A decrease in the N content is an indication of conversion of N-containing entities. This suggests that the coke generated in the early stages of recycle, was still reactive under cracking conditions. However, during the latter stages, the N content of the coke increased, an indication of the formation of more refractory coke in the reactor which was less susceptible to cracking. This effect has been reported by others in a continuous 1.801.952.102.252.402.002.202.402.602.803.00 N (wt%)Recycle #AFresh 1st 2nd 3rd 4th 5thB  N (wt%)197  hydroconversion unit [142]. These authors used a NiMo/Al2O3 catalyst for upgrading Maya heavy crude and observed a decrease in the N content of the coke deposited on the catalyst at low time-on-streams (TOS). During the latter stages, the N content of the coke increased linearly with TOS as the coke became more refractory. The cracking of the recycled coke in hydroconversion experiments has been described in detail previously ( Chapter 6). 7.2.5 EDX and XPS Analysis of the Fresh and Recycled Coke EDX and XPS analysis of the coke samples recovered after recycle using 600 ppm Mo in the form of both Mo-micelle and Mo-octoate are reported in Table  7.1, Table  7.2 and Table  7.3. Table  7.1 shows that the C content of the recovered coke increased with an increase in the number of recycles, an indication of the formation of a more carbonaceous coke as the number of recycles increased. On the other hand, the S and Mo concentrations in the coke samples decreased with increased recycle. Since almost all the Mo in the recovered coke originated from the catalyst added to the reactor (the CLVR feed had a Mo content of <12 ppm), it can be concluded that the catalyst concentration in the coke-catalyst mixture decreased with increased recycle. This is because of coke accumulation with an increase in the recycle number while the weight of Mo metal in the total coke remained constant. 198  Table  7.1. EDX analysis of coke samples recovered after different stages of recycling experiments using 600 ppm Mo added in the form of Mo-micelle precursor.       Note: Average values reported in the table are average of at least 6 different measurements within the coke sample.    After heat-up  After 1 h reaction  After 1st Recycle  After 2nd Recycle After 5th Recycle   ? ? ?/?  ? ? ?/?  ? ? ?/?  ? ? ?/?  ? ? ?/?  wt % wt % %  wt % wt % %  wt % wt % %  wt % wt % %  wt % wt % % C 70.24 1.98 3  78.38 2.81 4  83.27 0.93 1  85.99 0.53 1  91.21 1.13 1 S 9.35 1.08 12  7.32 1.23 17  6.45 0.59 9  5.21 0.12 2  2.16 0.11 5 Mo 3.85 0.59 15  3.10 0.51 16  2.00 0.26 13  1.13 0.09 8  0.41 0.05 12 199  The S content of the coke samples reported in Table  7.1 show the same trend as the Mo content and decrease with increased recycle. If the S atoms in the coke are not associated with Mo in the form of MoS2 but are present in the form of organic S containing compounds, the S concentration should be independent of the Mo concentration in the coke. Since this is not observed in the data reported in Table  7.1, it can be concluded that most of the S in the coke is associated with Mo in the form of MoS2. The coke recovered after the heat-up experiment had a Mo content of 3.85 wt %. Multiplying by the weight of generated coke, this was equivalent to the amount of Mo added to the CLVR at the beginning of the experiment. XPS analysis of the coke samples recovered from different recycle experiments with 600 ppm Mo (using Mo-micelle precursor) are presented in Table  7.2. They follow the same trend as the EDX analysis results presented in Table  7.1. Although the same trends are observed (C content increase and Mo and S content decrease as the number of recycles increased), the S and Mo concentrations in Table  7.2 are much smaller than those reported in Table  7.1. Recalling that the depth of analysis for EDX is in the range of a few micrometers, while the depth of analysis for XPS is in the order of few atomic layers, it can be concluded that there was minimal accumulation of MoS2 catalyst on the outer surface of the coke particles. This means that the catalyst particles are not physically adsorbed on the surface of the coke after the reaction and during quenching, although they may act as nucleation sites as the coke particles solidify during quenching after the reaction. EDX and XPS analyses of the coke generated in hydroconversion experiments using the Mo-octoate precursor are reported in Table  7.3.  The observed trends were very similar to those reported in Table  7.1 and Table  7.2.  200  Table  7.2 XPS analysis of coke samples recovered after different stages of recycling experiments using 600 ppm Mo added in the form of Mo-micelle precursor.   After heat-up  After 1 h reaction  After 3rd recycle  After 5th recycle  wt %  wt %  wt %  wt % C 91.38  92.59  94.69  96.25 S 1.31  0.69  0.59  0.56 Mo 0.06  0.02  0.01  0.00  Table  7.3. EDX and EPX analysis of coke samples recovered after different stages of recycling 600 ppm Mo added in the form of Mo-octoate precursor.  EDX*   XPS  After 1 h reaction  After 3rd recycle  After 1 h reaction  After 3rd recycle  ? ? ?/?  ? ? ?/?  ?  ?  wt % wt % %  wt % wt % %  wt %  wt % C 77.29 1.71 2  80.95 7.91 10  91.59  95.77 S 5.95 1.24 21  7.20 2.89 40  0.67  0.33 Mo 3.22 0.99 31  1.73 0.98 57  0.03  0.01 *: Average values reported for EDX analysis are average of at least 6 different measurements within the coke sample.  Based on the EDX and XPS results presented in Table  7.1, Table  7.2 and Table  7.3, it can be concluded that the formation of coke around the MoS2 catalyst particles and/or adsorption of MoS2 particles on the surface of the condensed coke, is followed by further 201  coke formation, which eventually covers most of the MoS2 particles. This limits the presence of MoS2 on the surface of the coke-catalyst mixture. 7.2.6 BET Surface Area and BJH Pore Volume Analysis of the Fresh and Recycled Coke The BET surface area and the total pore volume of the coke recovered from the fresh precursor and the 5th recycle using 600 ppm Mo in the form of Mo-micelle and Mo-octoate as well as a coke sample recovered from a thermal experiment are reported in Table  7.4. Due to the destructive nature of the degassing process, the BET measurement could not be done on every coke sample recovered at different stages of recycling. Table  7.4. BET surface area and BJH total pore volume of coke samples recovered from fresh and 5th recycle of 600 ppm Mo in the form of Mo-micelle and Mo-octoate and a thermal experiment. Catalyst Recycle number BET surface area, m2/g Pore volume, cm3/g 600 ppm Mo-micelle Fresh 13.5 0.063 5th recycle 0.9 0.003 600 ppm Mo-octoate Fresh 18.5 0.075 5th recycle 0.6 0.005 Thermal experiment Fresh 4.9 0.028  202  The BET surface area of the coke recovered from an experiment that used 600 ppm Mo (derived from the Mo-micelle precursor) was 13.5 m2/g. After the 5th recycle the coke-catalyst surface area had decreased to 0.9 m2/g. The same trend was observed when 600 ppm Mo derived from the Mo-octoate precursor was used as the catalyst. The pore volume of the coke samples, although very low, also showed a significant decrease when the coke was recycled (using both Mo precursors). These results suggest that coke aging due to recycling results in further cracking of side chains in the coke precursor macromolecules, as the coke is converted to a more refractory, dense coke. In the hydroconversion experiments with high catalyst concentration, high hydrogenation activity is present and hydroconversion yields more hydrogenated and amorphous coke compared to the experiments in which less active catalyst was used ( Chapter 5). Comparing the BET surface area and pore volume of the coke samples recovered after 5 recycles with the coke sample recovered from a thermal experiment (Table  7.4) shows that the recycled coke had significantly less surface area and pore volume compared to the thermal coke. The much lower surface area of coke sample recovered from a 5th recycle experiment compared to the surface area of coke sample recovered from a thermal experiment (Table  7.4) shows the greater effect of aging compared to catalyst concentration in the coke on the chemical and morphological properties of the coke. 7.2.7 13C Solid-State NMR Analysis of the Fresh and Recycled Coke Coke samples recovered from the fresh, the 3rd and 5th recycle experiments using 600 ppm Mo (Mo-micelle precursor) and 1800 ppm Mo (Mo-octoate precursor) were analyzed 203  using solid-state 13C NMR (Figure  7.4-A and Figure  7.4-B). All the recovered coke samples using both Mo-micelle and Mo-octoate precursors exhibit two peaks with a chemical shift in the range of 110-150 ppm, corresponding to non-substituted aromatic carbons (120-130 ppm), a carbon bridge between aromatic rings (130-140 ppm) and substituted aromatic carbons (140-150) [137,139,155,156]. The peak, which is centered at 120 ppm, corresponds to aromatic C-H bonds [142,157]. Another feature of the spectra is the peak within the range of 10-100 ppm, which corresponds to aliphatic carbons in the coke [142]. For both coke samples, the integral intensity ratio of the aromatic to aliphatic peaks increased linearly with recycle number (see inserts of Figure  7.4-A and Figure  7.4-B). From these results, it can be concluded that the nature of the coke changes with an increase in the number of recycles because of competing hydrogenation-cracking reactions. A decrease in the aliphatic content of the coke recovered in the 3rd and 5th recycle compared to the sample recovered from the first experiment without any recycle is in agreement with the decrease in the H content of these coke samples presented in Figure  7.2. These results also imply that part of the coke generated in the first experiment using the fresh catalyst, is cracked and solubilized when placed back into the reactor during the recycle experiment. The portion of coke that is cracked and solubilized is mainly aliphatic. The reversibility of coke formation by hydroconversion has been reported previously [142,149,158-160] but few studies have identified the nature of the part of the coke that undergoes this reverse reaction. In one study, Callejas et al. [142] also concluded that the portion of the coke that was hydrocracked was mostly aliphatic. 204       300 200 100 0 -100 -2000501000501000100200 IntensityppmFreshA Intensity3rd recycle  Intensity 5th recycle0 1 2 3 4 51.52.02.53.03.54.0Ar/Al peakintegral intensity ratioRecycle number205   Figure  7.4. 13C NMR analysis and change in the peak integral intensity ratio of aromatic to aliphatic carbons (CAr/CAl) for coke samples recovered from different recycling steps in hydroconversion experiments using 600 ppm Mo in the form of Mo-micelle precursor (A) and 1800 ppm Mo using Mo-octoate precursor (B).   300 200 100 0 -100 -200025507502550050100B IntensityppmFresh Intensity3rd recycle  Intensity 5th recycle0 1 2 3 4 53.03.54.04.5Ar/Al peakintegral intensity ratioRecycle number206  7.2.8 TGA Analysis of the Fresh and Recycled Coke TGA analysis of the coke recovered from experiments using fresh and 5th recycle 600 ppm Mo catalyst, derived from the Mo-micelle, and 1800 ppm Mo derived from Mo-octoate, are presented in Table  7.5. TGA analysis of the coke generated in a heat-up experiment using 1800 ppm Mo (derived from the Mo-octoate) is also compared to the aged coke samples. As shown in Table  7.5, the coke generated when using 600 ppm Mo in the form of a Mo-micelle,  lost ~31 % of its original weight when heated to 900 oC under N2 gas flow. This value decreased to about 13.5 % for the coke sample recovered after the 5th recycle. In the case of the coke sample recovered from the 5th recycle of 600 ppm Mo (using Mo-micelle precursor), a major portion of the weight loss originated from volatile compounds in the range of 120-550 oC whereas in the temperature range of 750-900 oC the weight loss was not very significant. The coke recovered from the experiment using fresh catalyst had a significant weight loss in the high-temperature region 750-900 oC. This indicates higher thermal stability of the coke recovered from the 5th recycle compared to the coke sample which had not yet been recycled. With 1800 ppm Mo added in the form of Mo-octoate precursor, the coke recovered after heat-up had a very high weight loss of 37.3 % in the temperature range 120-900 oC. Recalling that in the heat-up experiment the reactor was quenched immediately after reaching the reaction temperature, the coke recovered from this experiment had not been aged in the reactor under hydroconversion conditions. 207  Table  7.5. TGA analysis of coke samples recovered from heat-up, fresh hydroconversion experiment and 5th recycle of 600 ppm and 1800 ppm Mo added to the reactor in the form of Mo-micelle and Mo-octoate.   ?????? ???? =  ?? ?????? 100 [%] Wi : Weight of the coke at initial temperature Wf : Weight of the coke at final temperature a: Average of 3 different measurements b: Average of 4 different measurements  600 ppm Mo using Mo-micelle precursor a  1800 ppm Mo using Mo-octoate precursor  b  Weight loss, %  Weight loss, % Experiment 120-550 ?C 550-750 ?C 750-900 ?C 120-900 ?C  120-550 ?C 550-750 ?C 750-900 ?C 120-900 ?C Heat-up N/A N/A N/A N/A  21.82 8.40 7.11 37.33 Fresh catalyst 16.05 5.36 9.44 30.85  13.47 3.20 2.79 19.46 5th recycle of catalyst 9.28 2.49 1.74 13.51  10.79 3.66 3.11 17.56 208  On the other hand, coke samples recovered from the 1-h hydroconversion experiment and the 5th recycle had significantly less weight loss in the same temperature range of 120-900 oC. The results are a clear indication of changes in the chemical properties and thermal behavior of the coke samples when they are aged inside the reactor during recycling. The changes make the recycled and aged coke less susceptible to reactions occurring at high temperatures such as cracking of short chains in aromatics, crystallization, dehydrogenation and skeleton rearrangement [161]. 7.2.9 Activity and Characterization of Ex-situ Aged Coke-catalyst Mixture The coke characterization results showed dramatic changes in the morphological, thermal and chemical properties of the coke when recycled and aged under hydroconversion conditions. The characterization results showed that as the number of recycles increased, the coke became more carbonaceous and the H content of the coke decreased significantly (Figure  7.2). At the same time, a shift in the type of carbon in the coke was observed. The aliphatic portion of the coke was reduced during the early stages of recycling (Figure  7.4) and a noticeable shift toward the formation of graphitic carbon was observed with further recycling (Figure  6.12). These results suggest that the morphological and chemical changes of the coke may be an important factor in catalyst deactivation and that the catalyst concentration in the coke was not the only factor that determined the activity of the recycled catalyst in terms of coke yield. To confirm the effect of coke aging on catalyst activity, a series of experiments were done in which the coke was aged ex-situ. Two coke samples, recovered from experiments 209  that used 600 ppm and 1800 ppm Mo, derived from the Mo-micelle precursor, were thermally aged in a muffle furnace under 100 sccm of He gas at 700 ?C for 15 h. These conditions were selected to ensure that a measurable change in the chemical and morphological properties of the coke before and after aging was observed. The aged coke samples were characterized and used as the catalyst in a subsequent hydroconversion experiment. Results were compared to the experiments with fresh catalyst and a recycled catalyst without ex-situ aging. Table  7.6 presents the BET surface area, the peak integral intensity ratio of aromatic to aliphatic carbon (CAr/CAl) and the H/C atom ratio of the coke samples, before and after thermal aging. As shown in Table  7.6 for both coke samples recovered from experiments using 600 and 1800 ppm Mo, the H/C atom ratio decreased significantly during the ex-situ thermal aging. On the other hand, the ratio of the aromatic to aliphatic carbon increased noticeably. These results suggest that during aging, the coke loses a major fraction of the highly aliphatic side chains and is converted to a more carbonaceous material with high aromatic content.  The BET surface area of the coke significantly increased after thermal aging for both coke samples recovered from experiments using 600 ppm and 1800 ppm Mo in the form of Mo-micelle precursor. In contrast, the BET surface area of the recycled coke after the 5th recycle was significantly less than the surface area of the fresh coke (Table  7.4). Recalling the TGA analysis results (Table  7.5), there was a significant weight loss when the coke samples were heated under N2 in the range of 120-750 ?C. Cracking of the volatile 210  compounds during the ex-situ thermal aging of the coke at high temperature and removal of light cracking products by the continuous flow of He, yields a more porous coke and increases BET surface area during the aging process. This mechanism was confirmed by observing the weight loss of the coke samples during the aging process. Table  7.6. BET surface area, peak integral intensity ratio of aromatic to aliphatic carbon (CAr/CAl) and H/C atom ratio of fresh coke before and after thermal aging.  600 ppm Mo using Mo-micelle  1800 ppm Mo using Mo-micelle  Before aging After aging  Before aging After aging H/C atom ratio 0.76 0.22  0.63 0.17 CAr/CAl 1.96 11.13  3.13 14.31 SBET, m2/g 13.5 45.2  17.5 55.5  The coke sample recovered from the experiment using 1800 ppm Mo in the form of Mo-micelle was also analyzed by XRD before and after thermal aging and the results are presented in Figure  7.5. The XRD diffractogram of the coke before and after thermal aging showed a very significant change in the crystal structure during the ex-situ aging process. For the coke sample recovered from the experiment using 1800 ppm Mo, derived from the Mo-micelle, two broad peaks for graphitic carbon were observed at about 25.8? and 34.8?. These peaks transformed into two very sharp and intense peaks after the thermal aging which is an indication of graphitic coke formation with larger crystallite size compared to the coke sample before aging. The average crystallite size of the graphite particles for the coke sample before aging was ~8.4 nm. This number increased to 58.7 nm after thermal aging. This result 211  is consistent with the changes in the crystalline structure of coke when recycled several times (shown in Figure  6.12). The intensity of the other peaks (assigned to different metal sulfides) in Figure  7.5 increased mainly due to an increase in the concentration of these compounds in the aged coke (since part of the coke cracked during the TGA and hence the weight of the coke decreased after aging).  Figure  7.5. X-ray diffractogram of coke sample recovered from the hydroconversion experiment using 1800 ppm Mo in the form of Mo-micelle.  0 20 40 60 80 100After aging  Intensity (a.u.)Angle (2?)Before agingGraphite-2HMoS2 (110)212  Characterization of two coke samples before and after thermal aging showed that the ex-situ thermal aging simulated the changes in chemical properties of the coke that occurred during recycling. If the conclusions made based on the characterization of the recycled coke are correct i.e. that the changes in chemical properties of the coke during recycling are the main mechanism for catalyst deactivation, then the activity of the ex-situ aged coke when recycled should be very low in terms of coke suppression. To confirm the mechanism of deactivation, the activity of the aged coke-catalyst samples was determined and the results compared to the first recycle experiments in which the coke samples were recycled without thermal aging. These results are presented in Figure  7.6.  Figure  7.6. Coke yields of hydroconversion experiments in which fresh Mo using Mo-micelle (A), recycled coke without thermal aging (B) and thermally-aged recycled coke (C) were used as the catalyst. A B C024681012  Coke yield (wt %)Experiment 600 ppm Mo 1800 ppm Mo213  As shown in Figure  7.6, using 600 and 1800 ppm Mo derived from the Mo-micelle resulted in coke yields of < 3 wt %. When the coke-catalyst mixture recovered from these experiments was recycled without any thermal aging of the coke (experiment B in Figure  7.6), the recycle experiments had a lower coke yield than the experiments in which fresh catalyst was used. This is an indication of coke consumption during the early stages of recycle when the catalyst remained highly active, as has been discussed previously in  Chapter 6. The results are also consistent with the results presented in Section  7.2.7 of the present study. When the coke-catalyst mixture, recovered from experiment A, was thermally aged and then recycled, the coke yield increased significantly (9.1 wt % and 11.5 wt % using 600 ppm and 1800 ppm Mo, respectively). The significant change in the activity of the recycled coke-catalyst mixture can be attributed to the changes in the chemical properties of the coke during the thermal aging process. These results are consistent with the characterization results presented before and clarify the role of changes in chemical properties of the coke in the deactivation of recycled coke-catalyst. These results also explain why, when using a high concentration of Mo (1800 ppm), the number of possible recycles compared to the experiments in which 600 ppm Mo was used (Figure  6.11), did not increase.  Among the changes in morphological properties of the coke during ex-situ aging, the change in BET surface area was different to the trend observed in the recycled coke (comparing Table  7.4 and Table  7.6). The BET surface area of the ex situ aged coke increased after the aging process whereas the BET surface area of coke samples recovered after the 5th recycle decreased. Despite the increase in BET surface area of the coke during the ex-situ aging, the catalytic activity of the aged coke in terms of coke suppression was 214  lower than the coke not aged. This implies that morphological changes (i.e. changes in BET surface area and pore volume of the coke) cannot be a major factor determining catalyst deactivation. Under hydroconversion conditions, (high temperature and high H2 pressure), a portion of the solid coke will likely re-dissolve in the oil and therefore not act as a support for the adsorbed MoS2 catalyst. Hence, high or low surface area of the coke at ambient conditions cannot be a basis for determining high or low activity of the recycled catalyst. 7.2.10 Conceptual model of catalyst deactivation Due to the thermodynamic complexity of the multi-phase, multi-component reaction mixture during hydroconversion, the impact of the phase behavior of different compounds on the measured catalyst performance during hydroconversion is difficult to determine. However, in-situ observation of the mesophase formed during reaction can give insight into the effect of operating conditions on coke formation. Bagheri et al. [99] have recently studied the effect of catalyst and mixing during hydroconversion of Athabasca vacuum residue on mesophase formation. The authors showed that fluid mixing in the presence of a catalyst reduced the size of the mesophase compared to experiments in which no catalyst and no mixing was used. The authors proposed 2 different roles (chemical and physical) of the catalyst in the suppression of coke formation during the hydroconversion. The chemical role of the catalyst was to facilitate H transfer as suggested in the literature [87]. In assessing the physical role of the catalyst (10,000 ppm of an iron-based catalyst) on mesophase formation, the authors analyzed the coke formed after hydroconversion using SEM and EDX. Since some catalyst particles were observed on the surface of the coke and the mesophase, the 215  authors concluded that the catalyst particles in the reactor were on the outer surface of the mesophase and the hindering effect of the catalyst particles prevented coalescence of small mesophase domains to form macrometer-scale and bulk mesophase particles.  In the present research, XPS and EDX analyses was used to determine the distribution of the MoS2 catalyst within the coke matrix after reaction. As discussed in Section  7.2.5, the results showed that there was a relatively low catalyst concentration on the outer surface of the coke recovered after reaction, compared to the bulk of the coke-catalyst particle. The difference in the observations made in the present study compared to that reported by Bagheri et al. [99] may be partly due to differences in the type of catalyst used. Bagheri et al. used a relatively high concentration of supported Fe-based catalyst with particles in the micrometer size range, whereas in the present study the oil-soluble precursors yielded nano-scale MoS2 catalyst particles [2]. The XPS and EDX results suggest a structure of the coke-catalyst mixture recovered from a hydroconversion experiment as shown in Figure  7.7-A.  Based on the catalyst activity results using recycle and fresh catalyst, and the characterization of coke-catalyst mixtures recovered from the hydroconversion experiments, a conceptual mechanism of coke aging in the hydroconversion experiments, as shown in Figure  7.7 (A to E), is proposed. The coke recovered from the experiment in which fresh catalyst precursor was used (Figure  7.7-A) is compared to the coke structure under hydroconversion conditions inside the reactor for the 1st and 2nd recycle experiments (Figure  7.7-B and Figure  7.7?D)  and the coke recovered from the hydroconversion reactor after quenching the 1st and the 2nd recycle experiments (Figure  7.7-C and Figure  7.7-E).216                                                                                                                                         Figure  7.7. Schematic of catalyst and coke interaction as the number of recycles increases (A ? E). Highly hydrogenated and amorphous coke (light gray area), heavily carbonaceous and graphitic (refractory) coke, i.e. hard coke, (black area), catalyst particles (mostly single layers of MoS2, small white spots) and liquid phase in the reactor (dark gray area in Figure  7.7?B and Figure  7.7-C) are shown in the picture. Coke-catalyst mixture after recovey at ambient conditions, after 1st recycle Coke-catalyst mixture after recovey at ambient conditions, zero recycle Dispersed coke-catalyst mixture at reaction conditions, 1st recycle Soft coke is cracked and/or re-dissolved in the liquid under reaction conditions and releases the catalyst in the feed  The graphitic and refractory coke remains intact under hydroconversion conditions and blocks the access of the reactants to the encapsulated catalyst particles  Fresh residue Fresh residue  Dispersed coke-catalyst mixture at reaction conditions, 2nd recycle Coke-catalyst mixture after recovey at ambient conditions, after 2nd recycle 217  Based on the results of the coke-catalyst activity tests presented in Section  7.2.1 and the characterization results, the following steps are proposed for the observed coke-catalyst deactivation as the catalyst is recycled: (i) the coke recovered from the experiment in which fresh catalyst precursor was used, i.e. fresh coke, had a high H/C atom ratio and aliphatic carbon content (Figure  7.7-A).  Based on the EDX and XPS results presented in Table  7.1, Table  7.2 and Table  7.3 it was concluded that few catalyst particles were present on the outer surface of the coke recovered from the hydroconversion experiments. This was shown to be also the case for the dispersion and encapsulation of MoS2 layers within the carbonaceous materials in the M-Coke process [162]. Fresh coke consisted of small fractions of graphitic coke (dark black area in Figure  7.7-A) but the majority of the coke was susceptible to further cracking. This conclusion is consistent with the review of Chianelli et al. [162] on the use of single-layer MoS2 particles in heavy-oil processing in the Microcat Technology (M-Coke) developed by Exxon. The authors indicated that a major fraction of the toluene-insoluble material formed during the hydroconversion reaction is not pyrolytic coke, due to relatively low reaction temperature and time in the hydroconversion reactor. Similar results were obtained in the present study, however, pyrolytic coke was formed as the catalyst-coke was recycled. This part of the coke that is not graphitic and refractory (light gray area in Figure  7.7-A), was either cracked and/or re-dissolved in the liquid phase of the reactor when recycled for the 1st time. Consequently, the catalyst particles would be released from the coke during recycle. (ii) The portion of the coke that was not cracked or was not dissolved back into the liquid phase, i.e. the hard graphitic coke (black area in Figure  7.7-B), remained almost intact during the reaction and subsequent recycle experiments. This portion of the 218  coke would not release the catalyst particles (mostly single layers of MoS2, white spots in the black areas in Figure  7.7-B) encapsulated within it back into the oil phase. Hence the catalyst particles would not be available for H2 transfer to olefins and free radicals. The hard coke consists mainly of polyaromatic compounds with very few aliphatic side chains and a significant amount of graphitic carbon. During the 1st recycle and reactor quenching after the reaction, some fresh coke is generated, condensed and deposited on the hard coke particles (gray area in Figure  7.7-C). Due to aging of the coke during recycle, the fraction of hard coke in the coke-catalyst mixture recovered from the 1st recycle experiment increased (black area in Figure  7.7-C) both through growth of the recycled graphitic particles and formation of new graphitic structures. At the same time, after the 1st recycle, some new amorphous and soft coke is formed. (iii) As recycle continues, the cycle of amorphous and highly hydrogenated coke cracking and phase change to liquid inside the reactor continues, while at the same time, thermal aging under hydroconversion conditions converts more and more of the highly hydrogenated and amorphous coke to highly carbonaceous and graphitic coke (Figure  7.7-D and Figure  7.7?E). The latter encapsulates more of the catalyst particles, such that they can no longer participate in the hydroconversion reaction (iv) At a certain point during the recycle, the recovered coke will consist of a significant amount of refractory, hard coke that is not susceptible to cracking when recycled. Consequently, the amount of catalyst in the coke-catalyst mixture that can be released when the coke-catalyst mixture is recycled, decreases to the point where efficient coke suppression in the reactor is no longer possible.  Since these changes in the chemical properties of the coke are independent of the catalyst concentration in the coke and only depend on the severity of the hydroconversion 219  conditions, concentrations higher than a certain concentration of Mo (600 ppm in the present study) will not have any effect on the number of recycles that are possible before coke suppression ceases. This is because, although there will be more catalyst in the reactor for H transfer to the unsaturated olefins and free radicals generated by cracking in the reactor, the catalyst particles are not available due to their encapsulation in the coke matrix. This is the reason why when using a very high concentration of Mo (1800 ppm) the number of possible recycles did not increase compared to the recycle experiments done using 600 ppm Mo (Figure  6.11). 7.3 Conclusion The initial catalyst concentration in the coke as well as the changes in the chemical properties of the coke during recycling are important factors in determining the rate of deactivation of the catalyst during CLVR hydroconversion. Above a certain limit, an increase in the catalyst concentration does not affect the number of possible recycles. This is because the changes in the chemical properties of the coke are almost independent of the catalyst concentration in the reactor and depend on the severity of the hydroconversion conditions.     220  Chapter 8  Kinetic Study of Residue Upgrading in the Semi-batch Reactor 8.1 Introduction As discussed in Sections  2.3.2.1 and  2.3.2.2, trickle bed reactors using a fixed bed of catalyst as well as ebbulated-bed processes, are the most commonly used hydroconversion processes in industry. On the other hand, slurry-phase processes are relatively new technologies that are in the development phase. An important factor in developing slurry-phase hydroconversion technology is design of the reactor. Trickle-bed reactors, ebbulated-bed reactors and slurry-phase reactors are complex due to the presence of three different phases (gas, solid and liquid) and complications due to mass, heat and momentum transfer effects in each phase [105]. There are two different approaches in designing a reaction system. The first approach is using highly empirical correlations, which are obtained through commercial and experimental experience. The correlations are usually proprietary. The second approach is to develop a kinetic model and understand the reaction mechanism using mathematical modeling. Once the kinetic model for a certain reaction is developed, the model can be used for design, simulation and optimization of a reactor [105]. When dealing with the reaction of a complex mixture of compounds, there are different approaches to kinetic modeling. The first is to consider each individual compound and all 221  possible reactions that may be present. However, this approach is not feasible when dealing with residue and heavy-oils, which consist of numerous different compounds and thousands of possible individual reactions. This complexity leads to the generation of a complex set of equations which are very difficult to solve [105]. Another approach is to categorize the many compounds in the residue and heavy-oils into different fractions or lumps, usually based on their boiling point range. These lumps are treated as singular compounds when the ordinary differential equations and rate of reactions are developed for a proposed reaction mechanism. In this way, the number of reactions, as well as the parameters required to fit a kinetic model, is reduced significantly. To make the information obtained from the lumped kinetic modeling useful for process simulation, the lumps are defined so that they resemble a real fraction of refinery products. For example, residue (boiling point of > 524 ?C), HGO (boiling point of 348 ?C to 524 ?C), LGO (boiling point of 204 ?C to 348 ?C), naphtha (boiling point of up to 204 ?C) and gas (C1 ? C4) are common lumps used in lumped kinetic modeling of heavy-oil and residue. Although many studies [105,107-111,163,164] have been done on the kinetic modeling of heavy-oils and residue using different numbers of lumps, none have reported the production and consumption rates at elevated temperatures (i.e. 445 ?C and 460 ?C in this study). Furthermore, due to the low hydroconversion reaction temperature and very low coke production, none of these studies included coke as a lump in the kinetic modeling. In the present study, however, as shown in  Chapter 6 and  Chapter 7, coke consumption at 445 ?C was observed. Hence, in the present study, coke has been included in the kinetic model as a separate lump with both coke generation and consumption included in the model. 222  As presented in Section  3.5, a total of 16 experiments were done to generate the data needed to study the kinetics of CLVR hydroconversion. To eliminate the effect of the heat-up period (presented in Section  5.2.2), a heat-up experiment at each reaction temperature was done and the yields of different products obtained from the heat-up experiment were considered as the initial conditions of the feed at that temperature. Due to this assumption, although there was no naphtha and LGO in the CLVR used as the feed (Table  3.3), the liquid considered as the feed for the kinetic study had all 4 hydrocarbon cuts of naphtha, LGO, HGO and residue in its composition. A summary of the operating conditions of the experiments done to generate the kinetic modeling data is presented in Table  3.4. As described in Section  3.5, since sampling from the reactor was not feasible due to high reaction temperature and pressure, each of the 16 experiments had to be done independently to generate a set of data at a specific reaction temperature and reaction time. This methodology has been used by other researchers as well [165]. 8.2 Results and Discussions 8.2.1 Catalyst Activity Test at Different Temperatures Table  8.1 presents the activity of the 600 ppm Mo added to the reactor as Mo-octoate in terms of coke suppression and selectivity toward different product lumps at different temperatures. Note that at all the reaction temperatures, increasing the reaction time shifted the product toward production of lighter hydrocarbons (naphtha and LGO) and gas. This is simply due to further cracking of large molecules of HGO and residue into smaller molecules with a lower boiling point.  223  Increasing the reaction temperature had a similar effect to reaction time. At each reaction time, comparing different product yields showed a shift toward the formation of lighter products (Naphtha, LGO and gas) and the consumption of heavier compounds (HGO and residue) in the reactor. An important observation from the results presented in Table  8.1 is the coke generation behavior. Although at low temperatures (415 ?C and 430 ?C) the coke yield continuously increased with reaction time, this behavior was different at high temperatures (445 ?C and 460 ?C). After reaching a maximum, the coke yield decreased. This can be explained by conversion of soft and highly hydrogenated coke generated at the early stages of hydroconversion at high temperatures into liquid products (discussed in  Chapter 6 and  Chapter 7).    224   Table  8.1 Summary of the different yields and H2 uptake in the hydroconversion experiments using 600 ppm Mo in the form of Mo-octoate   Weight percent of different cuts    Ta, ?C tb, min Naphtha LGO HGO Residue Gas yield, wt % Coke yield, wt % H2 uptake, wt % 460 0c 15.80 21.40 35.80 27.00 3.01 1.31 N/A 30 29.28 33.73 25.63 11.37 5.19 2.39 1.49 60 37.62 36.58 18.01 7.79 7.53 1.96 2.20 120 43.10 36.97 14.40 5.53 11.12 1.65 2.92          445 0c 9.50 14.00 40.80 35.70 1.79 0.84 N/A 30 18.90 26.30 34.10 20.80 3.86 1.96 1.28 60 23.71 31.60 28.54 16.15 5.05 1.83 2.12 120 29.40 38.51 23.75 8.34 8.91 1.41 2.55          430 0c 4.62 9.91 44.38 41.09 1.19 0.69 N/A 30 12.85 19.76 38.46 29.19 2.53 1.44 0.96 60 18.90 23.80 34.23 23.07 3.12 1.56 1.50 120 25.63 26.20 29.36 18.81 4.59 2.18 1.97          415 0c 2.94 6.97 45.96 44.13 0.66 0.55 N/A 30 9.89 16.24 37.84 36.04 1.66 0.97 1.02 60 12.70 17.70 36.10 33.50 2.29 1.59 1.16 120 14.40 20.90 34.80 29.90 3.88 1.82 1.69 a T: Reaction temperature b  t: Reaction time c Experiment with zero reaction time is a heat-up experiment (Section  3.4)225  8.2.2 Developing the Kinetic Model 8.2.3 Data Normalization To ensure that all the kinetic data were calculated on the same mass base, the data presented in Table  8.1 was normalized to ensure a closed mass balance i.e. the sum of the mass of the 6 kinetic lumps (gas, naphtha, LGO, HGO, residue and coke) added to 80 g (the weight of the residue feed in each experiment). The normalized sets of data are presented in Table  8.2. Due to the negligible mass of hydrogen added to the products (hydrogen uptake), this mass was not included in the data normalization. The most common practice used by many researchers [107-111,163-165] to express the concentration of different lumps in the rate equations is to use wt % in the reactor. This method limits the development of reaction mechanisms in which 2 products are generated from a single reaction. As an example, considering the reaction:  ??? ? ???+ ???       ( 8.1) If all the concentrations are expressed in wt %, the differential mole balance equation for each species cannot be solved unless a function that describes the mass yield from each reaction is known. However, if the concentrations are expressed on a molar basis, reaction stoichiometry can be used to develop the differential mole balance equations.  226   Table  8.2 Summary of normalized kinetic data (weight-base)   Yield of different cuts, g    Ta, ?C tb, min Naphtha LGO HGO Residue  Gas yield, g Coke yield, g 460 0c 12.03 16.29 27.25 20.55  2.71 1.18 30 21.42 24.67 18.75 8.32  4.69 2.16 60 26.87 26.13 12.86 5.56  6.80 1.77 120 29.93 25.68 10.00 3.84  9.19 1.36          445 0c 7.40 10.90 31.77 27.79  1.46 0.68 30 14.25 19.83 25.71 15.68  3.15 1.53 60 17.56 23.41 21.14 11.96  4.12 1.58 120 20.84 27.29 16.83 5.91  7.26 1.25          430 0c 3.63 7.78 34.83 32.25  0.96 0.55 30 9.67 15.19 29.57 22.44  2.00 1.14 60 14.39 18.12 26.06 17.57  2.57 1.29 120 19.20 19.63 21.99 14.09  3.45 1.64          415 0c 2.32 5.51 36.31 34.86  0.55 0.46 30 7.70 12.63 29.44 28.04  1.38 0.81 60 9.74 13.58 27.69 25.69  1.95 1.35 120 10.84 15.73 26.19 22.50  3.23 1.51 a T: Reaction temperature b  t: Reaction time c Experiment with zero reaction time is a heat-up experiment (Section  3.4)227  To calculate the molar concentrations, the molecular weight of the different lumps was calculated using the gas analysis, the simulated distillation analysis of the liquid products and an empirical correlation presented by Speight [6]. Details of the calculations with sample calculations are presented in  Appendix K. The moles of the different lumps in the reactor were calculated by dividing the weight of each lump by the molecular weight as presented in Table  8.3. Since in the absence of acidic catalysts the cracking reactions are driven thermally, the rates of reaction and kinetic parameters are expressed independently of the catalyst weight in the reactor. It was assumed that in all the experiments done for kinetic modeling, enough catalyst was present to provide hydrogen for the hydrogenation reactions. 8.2.4 Kinetic Model Development Many studies [105,109,164] reported that reactions involved in residue hydroconversion mostly follow first-order kinetics. In the present work, many different kinetic reaction networks were developed based on different mechanisms proposed in the literature [105,107-111,164] assuming first-order kinetics for the conversion of heavier lumps to lighter lumps. In all the developed reaction models, a coke generation/consumption term was added to the reaction mechanism in accordance with the observation that the coke concentration increased and decreased for reactions done at high temperatures (445 ?C and 460 ?C, see Table  8.1). Gas formation from naphtha cracking has been reported to be unlikely and the rate of this reaction was found to be negligible [107,111]. Hence the gas production from naphtha has not been included in the proposed reaction network. 228  Table  8.3 Summary of normalized kinetic data (mole-base)   Mole of different cuts   Ta, ?C tb, min Naphtha LGO HGO Residue Gas Coke 460 0c 9.383E-2 7.763E-2 7.203E-2 3.637E-2 1.102E-1 2.236E-4 30 1.671E-1 1.176E-1 4.956E-2 1.472E-2 1.908E-1 4.096E-4 60 2.097E-1 1.245E-1 3.400E-3 9.848E-3 2.770E-1 3.361E-4 120 2.335E-1 1.346E-1 2.643E-3 6.797E-3 3.741E-1 2.588E-4         445 0c 5.771E-2 5.195E-2 8.396E-2 4.919E-2 5.939E-2 1.300E-4 30 1.112E-1 9.450E-2 6.795E-2 2.776E-2 1.281E-1 2.897E-4 60 1.370E-1 1.115E-1 5.587E-2 2.117E-2 1.676E-1 2.995E-4 120 1.626E-1 1.301E-1 2.117E-2 1.046E-2 2.956E-1 2.369E-4         430 0c 2.829E-2 3.707E-2 9.207E-2 5.708E-2 3.890E-2 1.052E-4 30 7.545E-2 7.240E-2 7.815E-2 3.972E-2 8.124E-2 2.156E-4 60 1.123E-1 8.636E-2 6.889E-2 3.109E-2 1.047E-1 2.441E-4 120 1.498E-1 9.354E-2 5.812E-2 2.494E-2 1.406E-1 3.113E-4         415 0c 1.812E-2 2.624E-2 9.596E-2 6.170E-2 2.232E-2 8.673E-5 30 6.005E-2 6.021E-2 7.782E-2 4.963E-2 5.616E-2 1.530E-4 60 7.600E-2 6.470E-2 7.318E-2 4.548E-2 7.935E-2 2.569E-4 120 8.456E-2 7.496E-2 6.369E-2 3.983E-2 1.314E-1 2.874E-4 a T: Reaction temperature b  t: Reaction time c Experiment with zero reaction time is a heat-up experiment (Section  3.4)229  Comparison between the different kinetic models was made based on the following factors: 1) Model prediction should satisfy the observed trend in the experimental data, 2) Estimated parameters (i.e. k values) had to be positive with satisfactory trends at different temperatures and 3) the calculated p-value of different lumps should have an acceptable value close to 1.0. Note that p-value (ranging fron 0 to 1) calculated from one-way ANOVA analysis is an indication of how well the model describes the measured data. After evaluating 5 different kinetic schemes for residue conversion, the following reaction scheme was selected and kinetic parameters were determined:   Figure  8.1 Proposed reaction mechanism for the residue conversion    230  The rate of reaction for the consumption of each lump was developed as follows:  Figure  8.2 Differential equations for the rate of consumption of different lumps.  Since the liquid in the reactor is in batch mode (while the gas flow was continuous), at each reaction temperature (415 ?C, 430 ?C, 445 ?C and 460 ?C) the following mass balance for each lump in the isothermal batch reactor was solved using MATLAB v7.1 software (ode45 function of MATLAB was used which uses Runge-Katta order 4 for solving the ODE set): ?????= ??         ( 8.2) Gas production in the model was assumed to be only through residue and HGO cracking reactions (reactions 2 and 3 in Figure  8.1). While in the residue fraction cracking of side chains mainly generates gas, it was assumed that in cracking 1 mole of HGO, cleavage of aromatic rings produces more gas compared to the cracking of 1 mole of residue. 231  Parameter b in the model was used to account for this excessive gas formation from HGO cracking.  8.2.5 Parameter Estimation and Statistical Analysis Methodology Estimation of the parameters of the kinetic model was done using an objective function minimization. The objective function (OBJ) was defined as follows: ??? = ? ??(??,??? ? ??,?????)^???=?        ( 8.3) where the variables are defined as follows: i:  Indicator of each kinetic lump (i.e. naphtha, HGO, LGO etc.) m: Total number of lumps (m is 6 in the present study) ??,??? : Mole of lump i measured from the experiment (Table  8.3) ??,????? : Mole of lump i calculated by the model ??: Weight of the square of the error for each lump The objective function minimization was done using MATLAB v7.1 and a Nelder-Mead simplex method. MATLAB m-files needed to minimize the objective function, solving the ODEs and estimating the parameters are presented in  Appendix L. 232  One-way ANOVA analysis of the results was done using Originpro V8.6 software. Details of the one-way ANOVA analysis and standard deviation calculations are presented in  Appendix M. Two constraints were applied to the parameters estimated by the model so that in the Arrhenius equation both activation energy (Ea) and pre-exponential constant (?0) should be positive. The Arrhenius equation is defined as follows: ? = ??. ?(?????)         ( 8.4) where the variables are defined as follow: k: Reaction rate constant ?0: Pre-exponential constant (factor) ??: Activation energy of reaction R: Gas constant T: Reaction temperature 8.2.6 Estimated Parameters and Fit of the Model The moles of each lump predicted by the kinetic model were compared to the experimental values for all kinetic lumps. The parity plots presented in Figure  8.3 show that the model results show a good fit of the experimental data. 233         0.0 5.0E-2 1.0E-1 1.5E-1 2.0E-1 2.5E-1 3.0E-1 3.5E-1 4.0E-10.05.0E-21.0E-11.5E-12.0E-12.5E-13.0E-13.5E-14.0E-1Model calculated, moleExperimental, mole(Gas)0.0 5.0E-2 1.0E-1 1.5E-1 2.0E-1 2.5E-1 3.0E-10.05.0E-21.0E-11.5E-12.0E-12.5E-13.0E-1(Naphtha)Model calculated, moleExperimental, mole2.0E-2 4.0E-2 6.0E-2 8.0E-2 1.0E-1 1.2E-1 1.4E-12.0E-24.0E-26.0E-28.0E-21.0E-11.2E-11.4E-1(LGO)Model calculated, moleExperimental, mole3.00E-2 4.50E-2 6.00E-2 7.50E-2 9.00E-2 1.05E-13.00E-24.50E-26.00E-27.50E-29.00E-21.05E-1(HGO)Model calculated, moleExperimental, mole0.0 1.0E-2 2.0E-2 3.0E-2 4.0E-2 5.0E-2 6.0E-2 7.0E-20.01.0E-22.0E-23.0E-24.0E-25.0E-26.0E-27.0E-2(Resdiue)Model calculated, moleExperimental, mole1.0E-4 2.0E-4 3.0E-4 4.0E-45.0E-51.0E-41.5E-42.0E-42.5E-43.0E-43.5E-44.0E-44.5E-4(Coke)Model calculated, moleExperimental, mole234   Figure  8.3 Parity plots of moles of gas, naphtha, LGO, HGO, residue, coke and all the lumps.  Figure  8.4 (A-D) also shows the fit of the model for different kinetic lumps at all 4 reaction temperatures as a function of reaction time.  0.00 7.50E-2 1.50E-1 2.25E-1 3.00E-1 3.75E-10.05.0E-21.0E-11.5E-12.0E-12.5E-13.0E-13.5E-14.0E-1(All)Model calculated, moleExperimental, mole235   (A)  (B) 236   (C)  (D) Figure  8.4 Fit of the model for all the kinetic lumps at all 4 reaction temperatures of 415 ?C (A), 430 ?C (B), 445 ?C (C) and 460 ?C (D). The lines represent the model calculated values and the circles represent the experimental values. 237  Table  8.4 presents the model-calculated k-values for the reactions of Figure  8.1. These values are very similar to the values reported by other researchers [106,107] at low temperatures (< 430 ?C).The activation energy (Ea) of the reactions as well as the pre-exponetial factor (k0) were calculated using the Arrhenius equation (? = ?0exp (??? ??? ). This was done by fitting the rate of reaction at different reaction temperatures to the Arrhenius equation and plotting ln k vs 1/T as presented in Figure  8.5.  Figure  8.5 Arrhenius plot for different reactions.  1.36E-3 1.38E-3 1.40E-3 1.42E-3 1.44E-3 1.46E-3-10-9-8-7-6-5-4-3-2 k-1   k1   k2   k3   k4Ln(k) (min-1)1/T (K-1)238  Table  8.4 Kinetic parameters of reaction calculated by the model.  k-values, 1/min  Activation energies, kJ/mol Ta k-1 k1 k2 k3 k4 b Ea,-1 Ea,1 Ea,2 Ea,3 Ea,4 415 ?C 3.98E-5 ? 1.63E-6 0 4.02E-3 ? 2.51E-6 6.85E-3 ? 1.96E-6 4.44E-4 ? 3.14E-6 1.36E0 ? 5.60E-4 205.34 103.96 213.94 69.11 76.34 430 ?C 5.58E-5 ? 2.32E-6 0 8.03E-3 ? 3.91E-6 8.77E-3 ? 2.33E-6 6.72E-4 ? 2.66E-6 9.80E-1 ? 4.16E-4 445 ?C 1.90E-4 ? 1.32E-5 6.52E-2 ? 1.54E-3 1.22E-2 ? 1.18E-5 1.16E-2 ? 5.38E-6 8.91E-4 ? 3.82E-6 2.17E0 ? 1.13E-3 460 ?C 3.25E-4 ? 1.02E-4 9.31E-2 ? 2.97E-3 4.51E-2 ? 1.13E-4 1.42E-2 ? 1.30E-5 1.00E-3 ? 5.68E-6 2.49E0 ? 1.92E-3 a T: Reaction temperature 239  By comparing the results presented in Table  8.4 and noting Figure  8.1, it can be seen that the reaction of residue and HGO into HGO and LGO, respectively are the fastest reactions. The reaction of coke to form residue also has a very high k value (k1). The rate of reaction to form naphtha from LGO is significantly slower compared to the rate of reaction for LGO and HGO formation. As reported by others [107], the higher stability of the LGO molecule compared to residue and HGO molecules is the reason why the rate of this reaction is low. The rate constant for coke formation from the residue (k-1) is low compared to the k1, k2 and k3. This can be explained by the fact that formation of coke from the residue fraction proposed here is not an elementary reaction. Coke formation from the residue consists of a cracking reaction and recombination of polynuclear aromatic hydrocarbons (PAH) to form a separate liquid phase and eventually insoluble coke. Since the isolation and measurement of the PAHs was not feasible in the present study, these reactions were combined in the proposed reaction mechanism. The values reported for different rate constants by others for thermal hydroconversion (without any catalyst) of residue and heavy-oil were similar to the values obtained in the present study. As an example, Sanchez et al. [107] reported 203 kJ/mol, 165 kJ/mol and 225 kJ/mol for Ea,2, Ea,3 and Ea,4 respectively. Lower activated energies observed in the present study for cracking of HGO and LGO into LGO and naphtha (Ea,3 and Ea,4) could be due to the  presence of catalyst in the experiments of the present study. Among the 5 different reactions proposed in the mechanism presented in Figure  8.1, reactions of residue cracking to form coke and HGO have the highest activation energy and have consequently, the highest sensitivity to temperature change. This means with increasing temperature, coke formation is very fast. This finding can be used to minimize the coke 240  formation in hydroconversion by a controlled two-stage hydroconversion at a low temperature followed by a higher temperature reaction. This finding is also consistent with the findings of other researchers [130] that proposed a two-stage hydroconversion to prevent fast coke formation. Heck found that doing the hydroconversion at a low temperature followed by a higher temperature hydroconversion reaction significantly decreases the coke yield of a hydroconversion reaction. The author explained the phenomena by considering fast reactive and slow reactive components in the residue and concluded that the reaction of fast reactive portion of the residue can be controlled toward formation of liquid products at a low temperature and this can be followed by a high temperature hydroconversion reaction to maximize the residue conversion. The value of parameter b was very similar for the case of 415 ?C and 430 ?C, while it significantly increased when the reaction temperature increased to 445 ?C and 460 ?C. This was probably due to the further cracking of the more stable aromatic rings and generation of more gas at elevated temperatures of 445 ?C and 460 ?C. 8.3 Conclusions Most of the lumped kinetic models of residue hydroconversion have been done at low and moderate temperatures (< 430 ?C) to minimize coke and sludge formation and consequently these studies ignore this portion of the product in the kinetic models. The present study shows that a kinetic model that includes the coke fraction as a separate lump describes the overall production distribution, including the coke yield, as a function of reaction time. The model confirms that coke formation is partially reversible at high 241  temperatures (445 ?C and 460 ?C) and this was successfully implemented in the proposed kinetic model for residue hydroconversion in the slurry-phase reactor of the present study.               242  Chapter 9  Conclusions and Recommendations 9.1 Conclusions Hydroconversion of a vacuum residue in a slurry-phase reactor under both low (40 ? 50 wt %) and high residue conversion conditions (> 85 wt %, T = 445 oC and H2 pressure of 13.8 MPa) have been studied. The focus of the research was to assess the activity of recycled catalyst in hydroconversion including coke suppression, as well as to identify the potential deactivation mechanisms associated with the recycled catalyst. Catalyst screening studies showed that the MoS2 prepared in a reverse micelle had superior activity in terms of coke suppression and liquid yield compared to Fe-micelle, bi-metallic micelle precursors (Mo-Co, Mo-Fe and Mo-Ni) and water soluble AHM. Preparation of MoS2 in reversed micelles (oil-soluble) using a water-soluble salt significantly increased the activity of the catalyst in terms of coke suppression. This is believed to be because of the small size of the MoS2 particles prepared in the reversed micelle as well as the oil-solubility of the reversed micelles. Using an oil-soluble precursor results in a better dispersion of active metal in the residue feed compared to a water-soluble precursor (i.e. AHM). Low concentrations of Mo (100 ppm) in the reactor decreased the coke production significantly compared to thermal hydrocnversion, while increased Mo concentration from 100 ? 900 ppm had a small effect on the coke yield. 243  The state of the catalyst in the coke samples recovered from batch hydroconversion experiments, determined by XRD, EDX-mapping and HRTEM, showed that the MoS2 was well-dispersed within the coke matrix with no indication of catalyst particle agglomeration. Elemental analysis of the coke-catalyst mixture also showed that more than 90 % of the metal added to the reactor at the beginning of each experiment was captured by coke particles generated during hydroconversion. Based on these characterization results, it was concluded that the recovered coke-catalyst mixture had the potential to be recycled as a catalyst in subsequent hydroconversion experiments. Selected catalysts from the batch experiments were studied further in the semi-batch reactor under high residue conversion conditions. Although the laboratory-synthesized Mo-micelle had higher activity in terms of coke suppression and liquid yield compared to the water-soluble AHM, its activity did not show any improvement when compared with a commercial Mo-octoate as Mo precursor. The two precursors had very similar activity in terms of coke yield, H2 uptake and residue conversion. Experimental results from CLVR hydroconversion experiments using the recycled coke-catalyst mixture showed that the coke-catalyst mixture could be successfully recycled several times without activity loss. The coke-catalyst mixtures recovered from experiments with a high initial Mo concentration (600 ppm in the present study) were successfully recycled ? 4 times. Although the coke builds up in the reactor during the recycle experiments, a study of solid loading effects showed that solid loading did not have any significant effect on the catalyst activity over the range of solid loadings encountered in the 244  present study. Initial catalyst concentration, as well as significant changes in the chemical and morphological properties of the coke, were identified as the main factors determining the activity of the recycled catalyst. Ex situ thermal aging of the coke-catalyst mixtures recovered from different hydroconversion experiments starting with different initial Mo concentrations confirmed that the thermal aging of the coke had a very significant effect on the activity of the coke-catalyst mixture. The catalyst deactivation was shown to be not due to agglomeration and increased catalyst particle size but due to significant changes in the morphology of the coke. Formation of graphitic structures in the hydroconversion coke, which are less susceptible to cracking at the reaction conditions, during the recycling, limits the accessibility of catalyst in the reactor. Growth of these graphitic structures during the recycling at some point makes the major part of the catalyst in the reactor inaccessible and results in poor catalytic activity.  A kinetic model that included coke formation and consumption was used to describe the product distributions. It was shown that coke as a separate lump can be defined in lumped kinetic modeling of residue and the distribution of different products can be successfully described through kinetic modeling. 245  9.2 Recommendations 9.2.1 Investigation of Hydroconversion in a Continuous Bubble-column Reactor As discussed in Section  5.2.2, hydroconversion begins during the heat-up period in the reactor and before reaching the reaction temperature in the reactor. In the reaction system used in the present study, the effect of the heat-up period could not be isolated from the calculation of coke yield, liquid yield, H2 uptake and other variables. To simulate the hydroconversion reaction conditions at industrial hydroconversion conditions (usually in a continuous reaction system in which both gas and liquid phases are continuous), utilizing a vertical continuous bubble-column reactor will mimic the industrial conditions more closely. Using a bubble-column reactor will also eliminate the effect of the heat-up period in the hydroconversion experiments. More importantly, using a continuous bubble-column reactor, would allow the complex hydrodynamics of this type of reactor to be studied together with a series of reactor simulation problems. 9.2.2  Quantifying the Onset of Catalyst Deactivation Quantification and identification of catalyst deactivation onset (detailed quantification of the changes in the physical and chemical properties of the coke can help to optimize the potential hydroconversion process using a recycled catalyst. This information can be used to maximize the ratio of coke-catalyst mixture recyle and decrease the metal (used in make-up catalyst) consumption. This quantification can be done by relating the changes in the physical and chemical properties of the coke (such as aromatic to aliphatic carbon ratio, formation and growth of graphitic structures etc.) with the hydroconversion conditions (reaction pressure 246  and time). Another effective technique in assessment of the growth of the graphitic structure and growth on the coke particle size is in-situ observation of coke formation using polarized light. Results from this technique will give insight toward a detailed understanding of the mechanism of coke formation as well as the effect of catalyst/additive on the formation and growth of coke particles. 9.2.3 Recycle of Sludge-coke-catalyst instead of Coke-catalyst In the present study, the coke-catalyst mixture recovered from a hydroconversion experiment was solvent washed, dried and ground prior to being used as a catalyst in a subsequent hydroconversion experiment. This procedure would be difficult to follow in a commercial upgrading unit. Assessing the activity of the recovered coke-catalyst mixture without complete sludge removal prior to the recycling would be useful to have a better understanding of the effect of the sludge associated with the coke on the activity of coke-catalyst mixture. 9.2.4 Development of More Sophisticated Kinetic Model Although the proposed kinetic model presented in  Chapter 8 was able to successfully describe the rate of reaction and mole concentration of different products in the reactor, more detail can be added to the model to make it more versatile. As an example, the reversibility of the coke formation reaction can be further studied by performing the hydroconversion experiment at the same reaction temperatures of the present study but with much longer reactions times (i.e. 180, 240, 300 mins). This way the potential plateau or peak for coke formation at lower temperatures of 415 ?C and 430 ?C can be observed.  247  Bibliography [1] MS Rana, V Samano, J Ancheyta, JAI Diaz. A review of recent advances on process technologies for upgrading of heavy oils and residua, Fuel. 86 (2007) 1216-1231.  [2] E Furimsky, Catalysts for Upgrading Heavy Petroleum Feeds, 1st ed., Elsevier, Amsterdam, 2007.  [3] H Rezaei, X Liu, SJ Ardakani, KJ Smith, M Bricker. A study of Cold Lake Vacuum Residue hydroconversion in batch and semi-batch reactors using unsupported MoS2 catalysts, Catalysis Today. 150 (2010) 244-254.  [4] IA Wiehe, Process Chemistry of Petroleum Macromolecules, CRC Press 2008.  [5] MR Gray, Upgrading Petroleum Residues and Heavy Oils, Marcel Dekker, New York, 1994.  [6] JG Speight, The Chemistry and Technology of Petroleum, 4th ed., CRC Press, New York, 2006.  [7] WL Nelson, Petroleum refinery engineering, 4th edition ed., McGraw-Hill, New York, 1958.  [8] B Shi, G Que. Chemical structure change of recycling vacuum bottom residue from slurry-bed hydrocracking of Karamay atmospheric residue in the pilot plant, Prepr.Symp.- Am.Chem.Soc., Div.Fuel Chem. 48 (2003) 718-721.  [9] FM Dautzenberg, JC De Deken. Reactor developments in hydrotreating and conversion of residues, Catal.Rev.- Sci.Eng. 26 (1984) 421-444.  [10] A Del Bianco, N Panariti, M Marchionna. Upgrading heavy oil using slurry processes, Chemtech. 25 (1995) 35-43.  [11] A Del Bianco, N Panariti, S Di Carlo, J Elmouchnino, B Fixari, P LePerchec. Thermocatalytic hydroconversion of heavy petroleum cuts with dispersed catalyst, Appl.Catal., A. 94 (1993) 1-16.  [12] A Matsumura, T Kondo, S Sato, I Saito, WF de Souza. Hydrocracking Brazilian Marlim vacuum residue with natural limonite. Part 1: catalytic activity of natural limonite, Fuel. 84 (2005) 411-416.  248  [13] A Matsumura, S Sato, T Kondo, I Saito, WF de Souza. Hydrocracking Marlim vacuum residue with natural limonite. Part 2: experimental cracking in a slurry-type continuous reactor, Fuel. 84 (2005) 417-421.  [14] K Sato, Y Iwata, K Honna, Y Araki, T Yoneda, Y Miki, et al. Proposal of new dispersed catalysts for the primary upgrading of residual oils, Stud.Surf.Sci.Catal. 121 (1999) 411-414.  [15] C Ovalles, E Filgueiras, A Morales, I Rojas, JC de Jesus, I Berrios. Use of a Dispersed Molybdenum Catalyst and Mechanistic Studies for Upgrading Extra-Heavy Crude Oil Using Methane as Source of Hydrogen, Energy Fuels. 12 (1998) 379-385.  [16] N Panariti, A Del Bianco, G Del Piero, M Marchionna. Petroleum residue upgrading with dispersed catalysts: Part 1. Catalysts activity and selectivity, Applied Catalysis A: General. 204 (2000) 203-213.  [17] N Panariti, A Del Bianco, G Del Piero, M Marchionna, P Carniti. Petroleum residue upgrading with dispersed catalysts: Part 2. Effect of operating conditions, Applied Catalysis A: General. 204 (2000) 215-222.  [18] CT Tye, KJ Smith. Cold Lake bitumen upgrading using exfoliated MoS2, Catalysis Letters. 95 (2004) 203-209.  [19] B Shi, D Lin, L Wang, G Que. Synergism between hydrogen donors and dispersed catalysts in residue hydrocracking, Prepr.- Am.Chem.Soc., Div.Pet.Chem. 46 (2001) 410-413.  [20] T Okui, M Yasumuro, M Tamura, T Shigehisa, S Yui. Convert oil sands into distillate cost-effectively, Hydrocarbon Process. 85 (2006) 79-81, 83-85.  [21] A Del Bianco, N Panariti, S Di Carlo, PL Beltrame, P Carniti. New Developments in Deep Hydroconversion of Heavy Oil Residues with Dispersed Catalysts. 2. Kinetic Aspects of Reaction, Energy Fuels. 8 (1994) 593-597.  [22] B Demirel, EN Givens. Hydroconversion of resids with dispersed molybdenum catalysts derived from phosphomolybdates, Fuel. 79 (2000) 1975-1980.  [23] D Liu, M Li, W Deng, G Que. Reactivity and Composition of Dispersed Ni Catalyst for Slurry-Phase Residue Hydrocracking, Energy Fuels. 24 (2010) 1958-1962.  [24] H Luo, W Deng, J Gao, W Fan, G Que. Dispersion of Water-Soluble Catalyst and Its Influence on the Slurry-Phase Hydrocracking of Residue, Energy Fuels. 25 (2011) 1161-1167.  249  [25] EC Sanford. Mechanism of coke prevention by hydrogen during residuum hydrocracking, Prepr.- Am.Chem.Soc., Div.Pet.Chem. 38 (1993) 413-416.  [26] E Furimsky. Selection of catalysts and reactors for hydroprocessing, Applied Catalysis A: General. 171 (1998) 177-206.  [27] X Zhang, The impact of multiphase behaviour on coke deposition in heavy oil hydroprocessing catalysts, ProQuest Dissertations and Theses. (2006) 164 p.  [28] JF Le Page, SG Chatila, M Davidson, Resid and Heavy Oil Processing, Editions Technip 1992.  [29] JG Speight, B Ozum, Petroleum Refining Processes, Marcel Dekker, Inc., New York, 2001.  [30] B Scheffer, MA van Koten, KW R?bschl?ger, FC de Boks. The shell residue hydroconversion process: development and achievements, Catalysis Today. 43 (1998) 217-224.  [31] M Marafi, A Stanislaus, E Furimsky, Handbook of Spent Hydroprocessing Catalysts: Regeneration, Rejuvenation, Reclamation, Environment and Safety, 1st edition ed., Elsevier, Oxford, 2010.  [32] CS Hsu, PR Robinson, Practical Advances in Petroleum Processing, Volume 1, 1st edition ed., Springer, New York, 2006, pp. 474.  [33] S Zhang, D Liu, W Deng, G Que. A Review of Slurry-Phase Hydrocracking Heavy Oil Technology, Energy Fuels. 21 (2007) 3057-3062.  [34] B Del A., A Zangirolami, C Di S. New developments in deep hydroconversion of heavy residues with dispersed catalysts. Part 2. Effect of severity on products yield and quality, Prepr.- Am.Chem.Soc., Div.Pet.Chem. 38 (1993) 407-12.  [35] R Bearden, CL Aldridge. Novel catalyst and process to upgrade heavy oils, Energy Prog. 1 (1981) 44-48.  [36] MA Menzies, AE Silva, JM Denis. Hydrocracking without catalysis upgrades heavy oil, Chem.Eng.(N.Y.). 88 (1981) 46-47.  [37] AE Silva, HK Rohrig, AR Dufresne. Canmet process going in to Montreal refinery, Oil Gas J. 82 (1984) 81-2, 84, 88.  [38] G Muir, BB Pruden, A Jain, Operation of the CANMET hydrocracking process in a conventional refinery, Periodical. (1989) AM-89-15.  250  [39] BB Pruden. Hydrocracking of bitumen and heavy oils at CANMET, Can.J.Chem.Eng. 56 (1978) 277-280.  [40] BB Pruden, NK Benham, The role of polar aromatics in residuum hydrocracking, Book of Abstracts, 213th ACS National Meeting, San Francisco, April 13-17. (1997) PETR-039.  [41] N Panariti, F Bazzano, A Delbianco, M Marchionna, R Montanari, S Rosi. Optimization of process conditions in slurry-phase upgrading of resids, Prepr.- Am.Chem.Soc., Div.Pet.Chem. 46 (2001) 78-81.  [42] N Panariti, A Delbianco, M Marchionna, R Montanari, S Rosi. Upgrading extra heavy residues and tar sands via EST: Eni Slurry Technology, Am.Inst.Chem.Eng., [Spring Natl.Meet.]. (2003) 664-669.  [43] B Del A., N Panariti, M Marchionna, Upgrading activity of recycled Mo-based dispersed catalysts, Book of Abstracts, 213th ACS National Meeting, San Francisco, April 13-17. (1997) PETR-132.  [44] R Montanari, S Rosi, M Marchionna, A Delbianco, N Panariti. Upgrading petroleum residues with EST process, Proc.World Pet.Congr. 17th (2002) 331-338.  [45] U Graeser, K Niemann. Proven hydrogenation processes for upgrading residua being revived in Germany, Oil Gas J. 80 (1982) 121-2, 125-7.  [46] W Doehler, DIK Kretschmar, L Merz, K Niemann. VEBA-Combi-Cracking - a technology for upgrading of heavy oils and bitumen, Prepr.- Am.Chem.Soc., Div.Pet.Chem. 32 (1987) 484-489.  [47] K Niemann, DIK Kretschmar, M Rupp, L Merz, Upgrading of Alberta Derived Bitumen Applying the VCC Technology, Periodical. 5 (1989) 225-242.  [48] M Seko, N Ohtake, K Kuri, K Kato, N Asakura. Super oil cracking (SOC) process for resid upgrading, AICHE Symp.Ser. 85 (1989) 43-48.  [49] G De Drago, J Guitian, J Krasuk, J Larrauri, R Marzin, F Silva, et al. The development of HDH process, a refiner's tool for residual upgrading, Prepr.- Am.Chem.Soc., Div.Pet.Chem. 35 (1990) 584-592.  [50] T Galiasso Roberto, B Solari, Upgrading Orinoco heavy oil with HDH-SHP technology, Am.Inst.Chem.Eng., [Spring Natl.Meet.]. (2003) 654-663.  [51] R Lott, LK Lee. Upgrading of heavy crude oils and residues with (HC)3TM hydrocracking technology, Am.Inst.Chem.Eng., [Spring Natl.Meet.]. (2003) 577-593.  251  [52] R Padamsey, RT Bailey, TJ Cyr, R Lott, (HC)3 Process - An Economic Process For Upgrading Alberta Bitumens, Periodical. (1994) 433.  [53] T Sakabe, T Yagi. Crack residue with spent HDS catalyst, Hydrocarbon Processing. 50 (1979) 103-107.  [54] NK Benham, BB Pruden, M Roy, Hydrocracking of heavy hydrocarbons with control of polar aromatics, PCT Int.Appl. (1997) 35.  [55] J Zhou, W Deng, G Que. Two-stage slurry bed hydrocracking of residue, Prepr.- Am.Chem.Soc., Div.Pet.Chem. 45 (2000) 614-617.  [56] B Shi, G Que. Hydrocracking of LiaoHe vacuum residue with oil-soluble Co-Ni bimetallic catalysts and hydrogen donor, Prepr.Symp.- Am.Chem.Soc., Div.Fuel Chem. 48 (2003) 722-724.  [57] B Qzum, L Lewkowicz, T Cyr, M Oguztoreli. Hydrodynamics of a bitumen upgrader, AICHE J. 35 (1989) 1032-1038.  [58] DK Lee, WL Yoon, SI Woo. Hydrotreatment of an atmospheric residual oil over the dispersed cobalt and molybdenum catalysts in a carbon expanded-bed reactor, Fuel. 75 (1996) 1186-1192.  [59] D Liu, X Kong, M Li, G Que. Study on a Water-Soluble Catalyst for Slurry-Phase Hydrocracking of an Atmospheric Residue, Energy Fuels. 23 (2009) 958-961.  [60] Y Iwata, K Sato, T Yoneda, Y Miki, Y Sugimoto, A Nishijima, et al. Catalytic functionality of unsupported molybdenum sulfide catalysts prepared with different methods, Catalysis Today. 45 (1998) 353-359.  [61] C Liu, J Zhou, G Que, W Liang, Y Zhu. Hydrocracking of Gudao residue with dispersed-phase Mo catalyst, Fuel. 73 (1994) 1544-1550.  [62] Y Iwata, Y Araki, K Honna, Y Miki, K Sato, H Shimada. Hydrogenation active sites of unsupported molybdenum sulfide catalysts for hydroprocessing heavy oils, Catalysis Today. 65 (2001) 335-341.  [63] B Fixari, S Peureux, J Elmouchnino, P Le Perchec, M Vrinat, F Morel. New Developments in Deep Hydroconversion of Heavy Oil Residues with Dispersed Catalysts. 1. Effect of Metals and Experimental Conditions, Energy Fuels. 8 (1994) 588-592.  252  [64] D Kennepohl, E Sanford. Conversion of Athabasca Bitumen with Dispersed and Supported Mo-Based Catalysts as a Function of Dispersed Catalyst Concentration, Energy Fuels. 10 (1996) 229-234.  [65] EC Sanford, JG Steer, K Muehlenbachs, MR Gray. Residuum Hydrocracking with Supported and Dispersed Catalysts: Stable Hydrogen and Carbon Isotope Studies on Hydrogenation and Catalyst Deactivation, Energy Fuels. 9 (1995) 928-935.  [66] R Ren, Z Wang, C Guan, B Shi. Study on the sulfurization of molybdate catalysts for slurry-bed hydroprocessing of residuum, Fuel Process Technol. 86 (2004) 169-178.  [67] JA Dunn, JB MacLeod, RD Myers, R Bearden. Recycle of Vanadium and Nickel-Based Catalysts in a Hydroconversion Process, Energy Fuels. 17 (2003) 38-45.  [68] G Bellussi, A Landoni, R Millini, D Molinari, P Pollesel, G Rispoli, New catalyst system for the conversion of oil residues, 15th International Congress on Catalysis. (2012).  [69] JG Speight, The Desulfurization of Heavy Oils and Residua, 2nd ed., Marcel Dekker, Inc., New York, 1999.  [70] E Furimsky. Metal carbides and nitrides as potential catalysts for hydroprocessing. Appl.Catal., A. 240 (2003) 1-28.  [71] O Weisser, S Landa, Sulphide catalysts: their properties and applications, Pergamon Press, Oxford, 1973.  [72] R Prins, De Beer, V. H. J., GA Somorjai. Structure and Function of the Catalyst and the Promoter in Co-Mo Hydrodesulfurization Catalysts, Catalysis Reviews. 31 (1989) 1-41.  [73] TC Ho. Hydrodenitrogenation Catalysis, Catalysis Reviews. 30 (1988) 117-160.  [74] R Krishna. Reactivity and product selectivity in FCC (fluid catalytic cracking) and hydrocracking. Influence of feedstock, Erdoel Kohle Erdgas Petrochem. 42 (1989) 194-199.  [75] GT Hager, XX Bi, FJ Derbyshire, PC Eklund, JM Stencel. Ultrafine iron carbide as liquefaction catalyst precursor, Prepr.Pap.- Am.Chem.Soc., Div.Fuel Chem. 36 (1991) 1900-1908.  [76] DK Lee, PS Koon, WL Yoon, IC Lee, SI Woo. Residual Oil Hydrodesulfurization Using Dispersed Catalysts in a Carbon-Packed Trickle Bed Flow Reactor, Energy Fuels. 9 (1995) 2-9.  253  [77] A Duangchan, Residue upgrading using dispersed catalysts prepared in reverse micelles, ProQuest Dissertations and Theses. (1998) 242 p.  [78] CE Snape, C Bolton, RG Dosch, HP Stephens. High liquid yields from bituminous coal via hydropyrolysis with dispersed catalysts. Energy Fuels. 3 (1989) 421-425.  [79] D Liu, W Guo, W Deng, J Zhou, G Que. Study on water-soluble dispersed catalyst of slurry phase hydrocracking, Prepr.- Am.Chem.Soc., Div.Pet.Chem. 50 (2005) 391-394.  [80] C Guan, Z Wang, S Yu, A Guo. Study on the bimetallic catalysts for residue suspended bed hydrocracking, Prepr.- Am.Chem.Soc., Div.Pet.Chem. 48 (2003) 150-153.  [81] D Garg, EN Givens. Relative activity of transition metal catalysts in coal liquefaction, Fuel Process Technol. 8 (1984) 123-134.  [82] C Song, AK Saini, Y Yoneyama. A new process for catalytic liquefaction of coal using dispersed MoS2 catalyst generated in situ with added H2O, Fuel. 79 (2000) 249-261.  [83] K Hirano, M Kouzu, T Okada, M Kobayashi, N Ikenaga, T Suzuki. Catalytic activity of iron compounds for coal liquefaction, Fuel. 78 (1999) 1867-1873.  [84] T Kaneko, K Tazawa, N Okuyama, M Tamura, K Shimasaki. Effect of highly dispersed iron catalyst on direct liquefaction of coal, Fuel. 79 (2000) 263-271.  [85] M Kouzu, Y Kuriki, K Uchida, K Sakanishi, Y Sugimoto, I Saito, et al. Catalytic Hydrocracking of Petroleum Residue over Carbon-Supported Nickel/Molybdenum Sulfides, Energy Fuels. 19 (2005) 725-730.  [86] CA Blanchard, MR Gray, Free radical chain reactions in thermal hydrocracking of residue, 213th ACS National Meeting, San Francisco, April 13-17. (1997) FUEL-054.  [87] MR Gray, WC McCaffrey. Role of Chain Reactions and Olefin Formation in Cracking, Hydroconversion, and Coking of Petroleum and Bitumen Fractions, Energy Fuels. 16 (2002) 756-766.  [88] C LaMarca, C Libanati, MT Klein, DC Cronauer. Enhancing chain transfer during coal liquefaction: a model system analysis, Energy Fuels. 7 (1993) 473-478.  [89] DF McMillen, DM Golden. Hydrocarbon Bond Dissociation Energies, Annu.Rev.Phys.Chem. 33 (1982) 493-532.  [90] JB Green, SKT Yu, CD Pearson, JW Reynolds. Analysis of sulfur compound types in asphalt, Energy Fuels. 7 (1993) 119-126.  254  [91] Y Miki, S Yamadaya, M Oba, Y Sugimoto. Role of catalyst in hydrocracking of heavy oil, Journal of Catalysis. 83 (1983) 371-383.  [92] P Michaud, JL Lemberton, G P?rot. Hydrodesulfurization of dibenzothiophene and 4,6-dimethyldibenzothiophene: Effect of an acid component on the activity of a sulfided NiMo on alumina catalyst, Applied Catalysis A: General. 169 (1998) 343-353.  [93] N Wei, S Ji, P Wu, Y Guo, H Liu, J Zhu, et al. Preparation of nickel phosphide/SBA-15/cordierite monolithic catalysts and catalytic activity for hydrodesulfurization of dibenzothiophene, Catalysis Today. 147, Supplement (2009) S66-S70.  [94] CT Tye, KJ Smith. Hydrodesulfurization of dibenzothiophene over exfoliated MoS2 catalyst, Catalysis Today. 116 (2006) 461-468.  [95] IA Wiehe. A phase-separation kinetic model for coke formation, Ind Eng Chem Res. 32 (1993) 2447-2454.  [96] IA Wiehe. A phase separation kinetic model for coke formation, Prepr.- Am.Chem.Soc., Div.Pet.Chem. 38 (1993) 428-33.  [97] KA Gould, IA Wiehe. Natural Hydrogen Donors in Petroleum Resids, Energy Fuels. 21 (2007) 1199-1204.  [98] SR Bagheri, A Bazyleva, MR Gray, WC McCaffrey, JM Shaw. Observation of Liquid Crystals in Heavy Petroleum Fractions, Energy Fuels. 24 (2010) 4327-4332.  [99] SR Bagheri, MR Gray, JM Shaw, WC McCaffrey. In Situ Observation of Mesophase Formation and Coalescence in Catalytic Hydroconversion of Vacuum Residue Using a Stirred Hot-Stage Reactor, Energy Fuels. 26 (2012) 3167-3178.  [100] SR Bagheri, B Masik, P Arboleda, Q Wen, KH Michaelian, JM Shaw. Physical Properties of Liquid Crystals in Athabasca Bitumen Fractions, Energy Fuels. (2012).  [101] A Stanislaus, A Hauser, M Marafi. Investigation of the mechanism of sediment formation in residual oil hydrocracking process through characterization of sediment deposits, Catalysis Today. 109 (2005) 167-177.  [102] I Wiehe. Mitigation of the fouling by popcorn coke, Petrol Sci Technol. 21 (2003) 673-680.  [103] EC Sanford. Conradson Carbon Residue Conversion during Hydrocracking of Athabasca Bitumen: Catalyst Mechanism and Deactivation, Energy Fuels. 9 (1995) 549-559.  255  [104] E Schmidt, C Song, HH Schobert. Hydrotreatment of 4-(1-Naphthylmethyl)bibenzyl in the Presence of Iron Catalysts and Sulfur, Energy Fuels. 10 (1996) 597-602.  [105] J Ancheyta, S Sanchez, MA Rodriguez. Kinetic modeling of hydrocracking of heavy oil fractions: A review, Catalysis Today; Hydroprocessing of Heavy Oil Fractions. 109 (2005) 76-92.  [106] B Sanchez, MS Gross, BD Costa, CA Querini. Coke analysis by temperature-programmed oxidation: Morphology characterization, Applied Catalysis A: General. 364 (2009) 35-41.  [107] S Sanchez, MA Rodriguez, J Ancheyta. Kinetic Model for Moderate Hydrocracking of Heavy Oils, Ind Eng Chem Res. 44 (2005) 9409-9413.  [108] RM de Almeida, R Guirardello. Hydroconversion kinetics of Marlim vacuum residue, Catalysis Today. 109 (2005) 104-111.  [109] S Sadighi, A Ahmad, M Rashidzadeh. 4-Lump kinetic model for vacuum gas oil hydrocracker involving hydrogen consumption, Korean J.Chem.Eng. 27 (2010) 1099-1108.  [110] H Loria, G Trujillo-Ferrer, C Sosa-Stull, P Pereira-Almao. Kinetic Modeling of Bitumen Hydroprocessing at In-Reservoir Conditions Employing Ultradispersed Catalysts, Energy Fuels. 25 (2011) 1364-1372.  [111] S Sanchez, J Ancheyta. Effect of pressure on the kinetics of moderate hydrocracking of Maya crude oil, Energy Fuels. 21 (2007) 653-661.  [112] Z Fan, Formation and characteristics of carbonaceous deposits from heavy hydrocarbon vapours, ProQuest Dissertations and Theses. (2006).  [113] MF Toney, S Brennan. Structural depth profiling of iron oxide thin films using grazing incidence asymmetric Bragg x-ray diffraction, J.Appl.Phys. 65 (1989) 4763-4768.  [114] CR Brundle, CA Evans, S Wilson, Encyclopedia of Materials Characterization: Surfaces, Interfaces, Thin Films, Butterworth-Heinemann 1992.  [115] D Massiot, F Fayon, M Capron, I King, S Le Calve, B Alonso, et al. Modelling one- and two-dimensional solid-state NMR spectra, Magn.Reson.Chem. 40 (2002) 70-76.  [116] DE Newbury, Advanced Scanning Electron Microscopy and X-Ray Microanalysis, Springer 1986.  256  [117] DC Joy, AD Romig, J Goldstein, Principles of Analytical Electron Microscopy, Springer 1986.  [118] JG Speight, Handbook of petroleum product analysis, Wiley-Interscience 2002.  [119] BJ Berne, R Pecora, Dynamic Light Scattering: With Applications to Chemistry, Biology, and Physics, Dover Publications 2000.  [120] K Fernie, G Muir, B Pruden, Residue upgrading by the CANMET Hydrocracking Process, Proc.- Int.Symp.Heavy Oil Residue Upgrading Util. (1992) 261-7.  [121] TF Yen Chilingarian G.V., Asphaltenes and asphalts, Vol. 2, 1st edition ed., Elsevier Science, New York, 2000, pp. 644.  [122] J Lopez, EA Pasek, Process for preparing heavy oil hydroprocessing slurry catalyst, U.S. (1987) 11 pp. Cont.-in-part of U.S. 4,552,821.  [123] F Khorasheh, HA Rangwala, MR Gray, IG Dalla Lana. Interactions between thermal and catalytic reactions in mild hydrocracking of gas oil, Energy Fuels. 3 (1989) 716-722.  [124] F Pedraza, J Cruz-Reyes, D Acosta, MJ Yanez, M Avalos-Borja, S Fuentes. The structure characterization of MoS2 and WS2 catalysts by HREM, Journal of Physics: Condensed Matter. 5 (1993) A219.  [125] M Soto-Puente, M Valle, E Flores-Aquino, M Avalos-Borja, S Fuentes, J Cruz-Reyes. Synthesis, characterization and cyclohexene hydrogenation activity of high surface area molybdenum disulfide catalysts, Catal.Lett. 113 (2007) 170-175.  [126] F Trejo, J Ancheyta, MS Rana. Structural Characterization of Asphaltenes Obtained from Hydroprocessed Crude Oils by SEM and TEM, Energy Fuels. 23 (2009) 429-439.  [127] SR Bagheri, MR Gray, WC McCaffrey. Influence of Depressurization and Cooling on the Formation and Development of Mesophase, Energy Fuels. 25 (2011) 5541-5548.  [128] EY Sheu, OC Mulins, Asphaltenes: Fundamentals and Applications, illustrated ed., Plenum Press, New York, United States of America, 1995.  [129] WI Beaton, RJ Bertolacini. Resid hydroprocessing at Amoco, Catal.Rev.- Sci.Eng. 33 (1991) 281-317.  [130] RH Heck, FT DiGuiseppi. Kinetic and Mechanistic Effects in Resid Hydrocracking, Energy Fuels. 8 (1994) 557-560.  257  [131] CS Hsu, PR Robinson, Practical Advances in Petroleum Processing, Volume 2, 1st edition ed., Springer, New York, 2006, pp. 419.  [132] J Datka, Z Sarbak, RP Eischens. Infrared Study of Coke on Alumina and Zeolite, Journal of Catalysis. 145 (1994) 544-550.  [133] JV Ibarra, C Royo, A Monz?n, J Santamar?a. Fourier transform infrared spectroscopic study of coke deposits on a Cr2O3-Al2O3 catalyst, Vibrational Spectroscopy. 9 (1995) 191-196.  [134] HG Karge, W Nie?en, H Bludau. In-situ FTIR measurements of diffusion in coking zeolite catalysts, Applied Catalysis A: General. 146 (1996) 339-349.  [135] B Paweewan, PJ Barrie, LF Gladden. Coking during ethene conversion on ultrastable zeolite Y, Applied Catalysis A: General. 167 (1998) 353-362.  [136] Z Sarbak. Characterization and infrared study of the effect of Cr, Mo and W on carbon deposition on platinum/alumina, Applied Catalysis A: General. 177 (1999) 85-97.  [137] M Rozwadowski, M Lezanska, J Wloch, K Erdmann, R Golembiewski, J Kornatowski. Investigation of Coke Deposits on Al-MCM-41, Chemistry of Materials. 13 (2001) 1609-1616.  [138] PE Eberly. High-temperature infrared spectroscopy of olefins adsorbed on faujasites, J.Phys.Chem. 71 (1967) 1717-1722.  [139] Z Fan, AP Watkinson. Formation and Characteristics of Carbonaceous Deposits from Heavy Hydrocarbon Coking Vapors, Ind Eng Chem Res. 45 (2006) 6428-6435.  [140] RM Silverstein, FX Webster, DJ Kiemle, Spectrometric identification of rganic compounds, 7th ed., John Wiley & Sons, United States of America, 2005.  [141] N Colthup, L Daly, S Wiberley, Introduction to Infrared and Raman Spectroscopy, Academic Press, New York, United States of America, 1990.  [142] MA Callejas, MT Marti ?nez, T Blasco, E Sas      residue hydrotreating catalysts by solid-state 13C-NMR spectroscopy and temperature-programmed oxidation, Applied Catalysis A: General. 218 (2001) 181-188.  [143] M Absi-Halabi, A Stanislaus, DL Trimm. Coke formation on catalysts during the hydroprocessing of heavy oils, Applied Catalysis. 72 (1991) 193-215.  [144] E Furimsky, FE Massoth. Introduction, Catalysis Today. 17 (1993) 537-659.  258  [145] E Furimsky. Deactivation of molybdate catalyst during hydrodeoxygenation of tetrahydrofuran, Industrial & Engineering Chemistry Product Research and Development. 22 (1983) 34-38.  [146] E Furimsky, JJ Spivey, Carbons and Carbon-Supported Catalysts in Hydroprocessing, The Royal Society of Chemistry 2008.  [147] CH Bartholomew. Mechanisms of catalyst deactivation, Applied Catalysis A: General. 212 (2001) 17-60.  [148] HS Cerqueira, C Sievers, G Joly, P Magnoux, JA Lercher. Multitechnique Characterization of Coke Produced during Commercial Resid FCC Operation, Ind Eng Chem Res. 44 (2005) 2069-2077.  [149] A Fonseca, P Zeuthen, JB Nagy. 13C n.m.r. quantitative analysis of catalyst carbon deposits, Fuel. 75 (1996) 1363-1376.  [150] E Furimsky, FE Massoth. Deactivation of hydroprocessing catalysts, Catal.Today. 52 (1999) 381-495.  [151] M Millan, C Adell, C Hinojosa, AA Herod, R Kandiyoti. Mechanisms of catalytic activity in heavily coated hydrocracking catalysts, Oil Gas Sci.Technol. 63 (2008) 69-78.  [152] VA Munoz, SV Ghorpadkar, MR Gray. Characterization of Coke on Spent Hydroprocessing Catalysts by Optical Microscopy, Energy Fuels. 8 (1994) 426-434.  [153] CA Querini, SC Fung. Coke characterization by temperature programmed techniques, Catalysis Today. 37 (1997) 277-283.  [154] JG Speight. New approaches to hydroprocessing, Catalysis Today. 98 (2004) 55-60.  [155] AR Pradhan, T Lin, W Chen, S Jong, J Wu, K Chao, et al. EPR and NMR Studies of Coke Induced Selectivation over H-ZSM-5 Zeolite during Ethylbenzene Disproportionation Reaction, Journal of Catalysis. 184 (1999) 29-38.  [156] MS Solum, JM Veranth, Y Jiang, AM Orendt, AF Sarofim, RJ Pugmire. The Study of Anthracene Aerosols by Solid-State NMR and ESR, Energy Fuels. 17 (2003) 738-743.  [157] KD Bartle, WR Ladner, TG Martin, CE Snape, DF Williams. Structural analysis of supercritical-gas extracts of coals, Fuel. 58 (1979) 413-422.  [158] PW Tamm, HF Harnsberger, AG Bridge. Effects of feed metals on catalyst aging in hydroprocessing residuum, Ind.Eng.Chem.Proc.Des.Dev. 20 (1981) 262-273.  259  [159] CS Kim, FE Massoth. Deactivation of a Ni/Mo hydrotreating catalyst by vanadium deposits, Fuel Process Technol. 35 (1993) 289-302.  [160] A Fonseca, P Zeuthen, JB Nagy. Assignment of an average chemical structure to catalyst carbon deposits on the basis of quantitative 13C n.m.r. spectra, Fuel. 75 (1996) 1413-1423.  [161] Z Fan, AP Watkinson. Aging of Carbonaceous Deposits from Heavy Hydrocarbon Vapors, Ind Eng Chem Res. 45 (2006) 6104-6110.  [162] RR Chianelli, MH Siadati, lR De, G Berhault, JP Wilcoxon, R Bearden, et al. Catalytic Properties of Single Layers of Transition Metal Sulfide Catalytic Materials, Catalysis Reviews. 48 (2006) 1-41.  [163] I Elizalde, J Ancheyta. On the detailed solution and application of the continuous kinetic lumping modeling to hydrocracking of heavy oils, Fuel. 90 (2011) 3542-3550.  [164] S Sadighi, A Ahmad, SRS Mohaddecy. 6-Lump Kinetic Model for a Commercial Vacuum Gas Oil Hydrocracker, Int.J.Chem.React.Eng. 8 (2010) A1.  [165] FM Song, CG Liu, GS Zhou, GJ Yang. Thermal Hydrocracking Kinetics of Chinese Gudao Vacuum Residue, Pet.Sci.Technol. 22 (2004) 689-708.  [166] JJ McKetta, Petroleum Processing Handbook, M. Dekker, Ney York, 1992.  [167] RA Meyers, Handbook of Petroleum Refining Processes, 3rd edition ed., McGraw-Hill, New York, 2004.  [168] JB Maxwell, Data book on hydrocarbons: application to process engineering, Van Nostrand Reinhold Company, Florida, USA, 1950.  [169] P Englezos, N Kalogerakis, Applied Parameter Estimation for Chemical Engineers, M. Dekker, New York, USA, 2001.      260         Appendices       261  Appendix A Thermal Cracking Processes (Carbon Rejection Processes) Thermal cracking processes are categorized according to the severity of the operating conditions. Low pressure and high temperature of the reactor in these processes made these processes very attractive to the industry. Currently, visbreaking and delayed coking are the widest operated residue conversion processes on an industrial scale. These processes are still very attractive processes for refineries from an economic point of view [6]. A.1 Thermal Viscosity Reduction (Visbreaking) Visbreaking (viscosity reduction) is a thermal cracking operation under relatively mild conditions aimed at reducing the viscosity of residua to produce fuel so that it can be pumped easily and meets specifications [5,6,166]. To produce a fuel with acceptable viscosity (lower than the viscosity of residue), the residue oil needs to be mixed with fuels with lower viscosity. The low viscosity fuel is usually produced in a visbreaking process. Visbreaking is also used to reduce the viscosity of fuels with very high viscosity to make the trabsfer of these fuels in the pipelines feasible [6,29]. A.2 Coking Processes Coking is a thermal process for the continuous conversion of residua into lower boiling point products. Coking can process a wide range of feedstock like straight-run residua or cracked residua. Gases, naphtha, light gasoil, heavy gasoil, and coke are products of the coking processes [6]. The high amount of coke in these processes (> 20 wt %) is particularly attractive when the green coke produced can be sold for anode or graphitic carbon 262  manufacture or when there is no market for fuel oils, as in Western Canada [5]. For these purposes, the produced coke requires further purification by removing S and metallic compounds from the coke. Further, the increasing attention paid to reduce atmospheric pollution has also served to direct some attention to coking, since the process not only concentrates such pollutants as feedstock S in the coke, but usually yields products that can be conveniently subjected to desulfurization processes. Coking processes generally utilize longer reaction times (> 1 hours) than thermal cracking processes. To accomplish this, drums or chambers (reaction vessels) are employed, but it is necessary to use two or more such vessels, so that coke removal can be accomplished in those vessels not on-stream, without interrupting the semi-continuous nature of the process (Figure  A.1). Coking processes have the advantage of converting the residue fraction of the feed. This conversion is always associated with conversion of a substantial fraction of the feed into the solid coke. The amount of produced coke is highly influenced by the nature of the residue feed, especially the carbon residue content of the feed. The formation of large quantities of coke is a severe drawback of coking processes, unless the coke can be put to use [6]. Delayed Coking When it comes to the conversion of the bottom-of-the-barrel fraction of the refinery feeds, delayed coking has always been among the first choices of many refiners. This is mainly due to the ability of the delayed coking in processing very heavy feedstocks. High temperature of the reactor and relatively long residence time of the feed in the hot zone of the 263  process makes complete rejection of metals and also partial or complete conversion of the feed to the light products (naphtha and diesel) feasible in this process [6]. In a delayed coker (Figure  A.1), before entering the large coke drum, residue feed is heated to ~ 487 to 520 ?C. Cracking under high temperature generates vaporized products as well as solid coke. The generated vapors exit the drum and enter to the product fractionator while the generated coke stays in the reactor [131]. The process uses long reaction times (> 1hr) in the liquid phase to convert the residue fraction of the feed to gases, distillates and coke. The condensation reactions that give rise to the highly aromatic coke product also tend to retain S, N and metals. Consequently, produced coke is enriched in these elements relative to the feed [5]. The general goal of using delayed coking processes changed significantly in the last few decades. While these processes were solely being used to completely convert atmospheric residue to the fuels lighter than diesel, recently designed cokers have aimed to minimize coke formation and produce a heavy coker gas oil. The latter can be catalytically upgraded in other processes into lighter fuels [6]. 264   Figure  A.1 Schematic of a delayed coker [6]. (Copyright ? 2006, Taylor & Francis Group LLC., with permission).  Being a thermal cracking process and a more expensive process compared to the processes using solvent at ambient temperature for residue removal are disadvantages of delayed coking [6]. Fluid Coking Fluid coking is a continuous process that uses the fluidized solids (coke) to convert residua, including vacuum residua and cracked residua, to more valuable products with lower boiling points [167]. The process is useful for processing of residue (Section  2.1). A fluidized 265  bed of small hot and coke particles is continuously circulated in the reactor where the residuum feedstock is sprayed. Circulating solid in the reactor ensures a high temperature and low residence time of feedstock in the reactor. Fluid coking has a lower coke yield and higher yield of liquid products compared to the delayed coking process [6,29]. Two vessels constitute the main part of a fluid coking process, a coke burner and a reactor. Circulating coke particles between these two vessels ensures a efficient heat transfer. The required heat is generated by combusting a fraction of the generated coke in the reactor. The feed coming from the bottom at 260 ?C ? 370 ?C is fed directly to the reactor. The reactor temperature is maintained at 480 ?C ? 560 ?C. This high temperature partially vaporizes the feed and less volatile portion of the feed is deposited on the fluidized coke particles.The deposited materials on the surface of the hot coke cracks and vaporizes, leaving behind the very heavy portion of the feed which dries to coke. The vapor products pass through the condenser a part of the entrained coke is removed from the vapor products [6,29,32]. A simplified schematic of a fluid coker is shown in Figure A.2. .  266   Figure A.2 Schematic of a fluid coker [6]. (Copyright ? 2006, Taylor & Francis Group LLC., with permission).  Flexicoking Flexicoker is a very similar process to fluid coking for primary upgrading of residue. In fleixicoking, instead of removal of the excess coke from the heater, the excess coke is converted into gas in a gasifier to produce refinery fuel gas [6,167]. In the process, steam and air are used to combust excess coke into a low-heating value gas in a fluid-bed gasifier. High temperatures of 830 ?C to 1000 ?C in the burber drum are usually maintained by the supplied air to the gasifier. A typical gas product, after removal of hydrogen sulfide, contains carbon monoxide (CO, 18%), carbon dioxide (CO2, 10%), hydrogen (H2, 15%), nitrogen (N2, 51%), 267  water (H2O, 5%), and methane (CH4, 1%). Since the coking reactor in fexicoking is the same as the fluid coking, two processes have the same yield of liquid. The requirement for a large additional reactor in the gasifier section, especially if high conversion of the coke is required, is a main drawback of the flexicoking process. Nowadays, units are designed to gasify 60 % to 97 % of the coke from the reactor. Even with the gasifier, the product coke will contain significantly more sulfur than the feed, which limits the attractiveness of even the most advanced coking processes [6]. A schematic of flexicoking process is shown in Figure A.3.  Figure A.3 Schematic of a flexi coking process [6]. (Copyright ? 2006, Taylor & Francis Group LLC., with permission).   268  Appendix B Geometry and Mixing Pattern of the Batch Reactor                                Figure  B.1 Schematic (LHS) and geometry (RHS) of batch reactor body. (Mechanical drawing by Autoclave Engineers).  Figure  B.2 Mixing pattern of batch reactor equipped with straight blade mixer. (Mechanical drawing by Autoclave Engineers) 269  Appendix C H2S Yield Sample Calculations Table  C.1 Saample values for H2S yield calculation in semi-batch reactor experiments. Volume of Fresh NaOH solution in scrubber (ml) 400 Concentration of Fresh NaOH solution (mol/L) 1 Abstracted Volume of used NaOH solution (ml) 10 Volume of NaOH solution for Analysis (ml) 100 Titration Volume of NaOH solution in Analysis 1 (ml) 23 Titration Volume of NaOH solution in Analysis 2 (ml) 24.3 Weight of PHP in erlen 1 (g) 0.4217 Weight of PHP in erlen 2 (g) 0.4469 Weight of residue feed, g 80.60   ???? = 204.2 ?/???               ????? = 40 ?/???              ??2? = 34.08 ?/??? ????? ?? ??? ?? ????? 1 =  0.4217204.2= 0.0021 ??? Based on the stoichiometry of the reaction between NaOH and PHP, it is assumed that 23 mL of diluted NaOH solution has the same mole of NaOH as the moles of PHP in erlen 1. ????????????? ?? ??????? ???? ?? ????? 1 =  0.0021 ???23 ???1000 ??1 ?= 0.0913 ???/? ????????????? ?? ???? ?????? ???????? = 0.0913 ? 10 = 0.913 ???/? ????? ?? ???? ???? ?????? ??? ???????? = (1? .913)??? ? 0.4 ? = 0.0348 ??? 270  Moles of produced H2S during the reaction in equal to the moles of NaOH used in titration: ?????? ?? ???????? ?2? = 0.0348 ??? ? 34.08????= 1.19 ? ?2? ????? ????? ?? ????? 1 ????????? = ?1.1980.60? ? 100 = 1.48 % Calculating the H2S yield based on the titration of erlen 2: ?2? ????? ????? ?? ????? 2 ????????? = ?1.3580.60? ? 100 = 1.68 % ??????? ?2? ????? ?? ??? ?????????? =1.48 + 1.682= 1.58 %         271  Appendix D Product Work-up of Batch Reactor Experiments   Figure  D.1 Summary of product work-up in the batch reactor experiments.   272  Appendix E Product Work-up of Semi-batch Reactor Experiments   Figure  E.1 Summary of product work-up in the semi-batch reactor experiments.  273  Appendix F Detailed Operating Procedure of Semi-batch HCR Experiments 1- Thoroughly clean the reactor body and internal fittings (mixer, mixer shaft, thermocouple well and cooling coil) using methylene chloride and paper towel as well as some metal polish to make sure no residue is left from th previous experiment. 2- Using a multimeter measure the combined heater resistance. The value should be around 49 ohms. If you read a different value, measure the resistance of each of 6 heaters individually (each should have a resistance of close to 298 ohms) to identify the faulty heater and replace it. 3- Check the condenser and H2S scrubber to make sure they are both empty and no liquid is left from the previous experiment. 4- Weight about 80 g of CLVR and pour it in the reactor. This step should be done quickly since the CLVR will be hard to work with when gets cold. 5- Add the desired amount of the fresh or recycled catalyst to the reactor. 6- Place the graphite gasket in position on top of the reactor and raise the reactor so that the engraved ?5000? is facing front right. This position will proper wall thermocouple placement. 7- Position the head clamps (hook sections at sides) and lock. 8- Tight nuts (opposite clamping scheme) to following torques (for 2000psi): 10, 20, 30, 35, 40 lb.ft. Go around 2 or 3 times to ensure well tightened. 9- Place reactor wall thermocouple in place (slotted in with the tip over "5000") and Position and secure reactor and bottom insulations. Do not twist the reactor insulation when in place as this will move the wall thermocouple. 274  10- Check to make sure that main cooling water valve is open and that water is flowing to pressure transducer and stirrer. 11- Slowly open 3-way valve (on left after MFC) to nitrogen (facing right) and check valves along the product line. They should be clear to vent. 12- Open N2 cylinder. Use N2 regulator to slowly increase pressure until flow is ~ 700 ml/min 13- Switch on 40 and 42 in the main switch box of the lab. A green power light on the unit controller should turn on. Turn on power switches for control boxes, thermocouple and detectors. 14- Set MFC1 value on the MFC control box to V0T and set the value to ~ 85. This is equal to a H2 flow of 913 mL(STP)/min. 15- Ensure both black and green regulators on the gas booster are closed. 16- Turn off N2 gas (regulator then cylinder). 17- Switch the selector valve connected to the H2 and N2 cylinders to H2. 18- Ensure open H2 cylinder. 19- Open blue and yellow air valves and increase feed (H2) pressure on the regulator to ~ 600 psi. 20- Purge the reactor for around 3-5 mins. 21- Start CalGrafix software on the PC (both control boxes must be on). 22- Select "Add new instrument" (menu or button) and add Device 1 (on COM 3 communication port). 23- Tighten the BPR finger-tight then about 6 half turns with the spanner. 275  24- Wait for gas flow out to be registered on MFC2 channel on the MFC control box. 25- Tighten a little with the spanner, wait for flow, repeat to reach 2000psi pressure in the reactor. 26- As pressure rises, will need to adjust gas booster accordingly so that the outlet pressure of the booster is always 200 psi higher than the pressure in the reactor. This will ensure that the MFC will not be exposed to a high pressure difference and minimize the risk of MFC being damaged. 27- Measure 400ml of 1N NaOH solution. Ensure bottom valve of the scrubber is closed, but the vent and feed lines of the scrubber are completely open. 28- Hold the funnel in place and add the NaOH solution to the scrubber. 29- Close vent and feed lines. 30- Fill the bath around the condenser with ice and a little cold water and pile ice on top, around the condenser. 31- Drain the water and refill the ice during operation as necessary. 32- Right-click on the software screen on the PC and navigate to Properties -> Module 1 (Rxn T) -> Setpoint to 445 ?C. The state of the controller should be on PARK. 33- Mixer (displayed in RPM) is usually controlled manually (switch on the box). If automatic control is required, the switch on the control box should be changed to ?remote? mode and a ?PI? control mode should be used. Do NOT use ?PID? mode since there will be a lot of fluctuations in mixer RPM. 276  34- IMPORTANT: Do not turn on the mixer when using heavy-oil and residue until the reactor reached the temperature of 80 ? 100 ?C. At lower temperatures, very high viscosity of heavy-oil or residue will damage the mixer motor. 35- As the reactor pressure (read on reactor P gauge as is most accurate) nears 2000psi, reduce H2 flow on the control box to 85. 36- Create a chart by pressing the create chart botton on the menu bar of the software and choosing ?Programmer? and ?Input 1? in the following menu. 37- Important: Ensure that the safety valve on the water line is open a little. This is to release the pressure generated by the evaporation of the water trapped in the cooling coil at high temperatures. 38- When the pressure is reached and the exit flow is stable, start chart, stopwatch and heating at the same time. 39- During heating, record the wall T, reaction T and outlet flow at regular intervals. 40- When wall T reaches about 340?C, alternate the heater setting between PARK and PID, switching it off when it rises too much and on again when it starts to drop. This is to ensure the wall doesn't overheat and liquid have enough time to get to a temperature as close as possible to the reactor wall. 41- As soon as the reaction mixture reaches the desired temperature, i.e. 445?C, mark the time on the temperature record as start time and continue the reaction until the desired reaction time. 42- 1 min before desired reaction time has passed, set heater to ?PARK? mode. 43- Set MFC1 to ?V-T?. This will close the H2 flow. 277  44- Tighten the BPR 2 half turns (to stop the exit flow). 45- Close the inlet 3-way valve after the MFC. 46- Close the cooling water safety valve and open the cooling water inlet and outlet valves toward the reactor cooling coil. This will quench the reactor immediately. 47- Wear high-temperature gloves and safety glasses and remove both top and bottom insulations. 48- Decrease stirring to approximately 250 RPM. 49- Close H2 valve, regulator and cylinder. Close the exit flow of H2 from the gas booster by closing the green regulator on the booster. Close the air regulator (black regulator on the left side of the booster front panel) and close both inlet air valves. 50- Save the chart by navigating to File -> Export -> Chart to the desired location on the PC. 51- When the reactor reached the room temperature, recored the temperature and pressure of the reactor for futher mass balance calculations. 52- Slowly depressurize the system by "notching" open the BPR and listening for bubbles in the H2S scrubber. 53- One can also watch for the red light on the outlet MFC or the reactor pressure on the software screen. Depressurization should be done as slow as possible since fast depressurization will cause liquid carry over into the exit line and condenser. 54- When the reactor is totally depressurized, the head clamp can be opened. Note that you will need to hold the reactor with your other hand and lower the reactor slowly and place it on the reactor rack. 55- Follow with the product work up procedure described in  Appendix E. 278  Appendix G Summary of Experimental Results of Batch Reactor Experiments Table  G.1 Summary of experimental results of batch reactor experiments.  279     280     281     282     283  Appendix H Summary of Experimental Results of Semi-Batch Reactor Experiments Table  H.1 Summary of experimental results of semi-batch reactor experiments.  284     285    286    287    288    289    290    291    292    293    294   295   296   297   298   299   300    301    302   303  Appendix I Temperature and Pressure Profile of a Typical HCR Experiment   Figure  I.1 A sample temperature and pressure profile of a typical hydroconversion experiment.     0 20 40 60 80 100 12015020025030035040045018002000Furnace Temperature, oCReaction Temperature, oC  Reaction Temperature or PressureTime (min)Pressure, psi304  Appendix J Kubelka-Munk Transformation Kubelka-Munk?s transformation is a widely used technique to improve the spectrum. The diffuse reflectance spectrum of a diluted sample of "infinitely thick" layer (i.e., at least 3 mm) is usually calculated with reference to the diffuse reflectance of the pure diluent to yield the reflectance, R. The Kubelka-Munk's equation is written as: ?(?) =(???)???=??                 ( 9) where k is the absorption coefficient, and s is the scattering coefficient. Based on the Kubelka-Munk theory, there is a linear relationship between the molar absorption coefficient, k, and the peak values of f(R) for each band, provided s remains constant. Since s is dependent on particle size and range, these parameters should be made as consistent as possible if quantitative data are needed.        305  Appendix K Molecular Weigh Calculations for the Gas, Liquid and Solid Products Gas chromatography analysis of the product gases was used to calculate the average molecular weight of product gas generated in hydroconversion experiments as described follow: 1- To calculate the number of moles of each gas (??) in the gas mixture, weight of each individual gas (??) was divided to the molecular weight of that particular gas (??,?) as follow: ?? =????,?         ( K-1) 2- The total weight of hydrocarbons (??) was simply calculated by adding the weight of each individual gases (??) in the generated gas mixture as follow: ?? = ? ????=?          ( 91) where i and m are the indicators for each individual gas and total number of gases in the gas mixture respectively. Knowing the ?? and ??, the average molecular weight of the gas mixture (???) is calculated as follow: ??? =??? ????=?         ( K-2) A sample of the calculations is presented in Table  K.1.  306  Table  K.1 Summary of sample calculations for calculating average molecular weight of gas mixture. Gas Weight, g Mw Mole CH4 1.74 16.00 0.1086 C2H4 0.01 28.00 0.0003 C2H6 1.18 30.00 0.0395 C3H6 0.05 42.00 0.0011 C3H8 0.79 44.00 0.0179 i-C4H10 0.12 58.00 0.0020 C4H8 0.02 56.00 0.0004 n-C4H10 0.26 58.00 0.0044 Total weight = 4.16 g Total mole = 0.174 mole Average molecular weight = 23.88 g/mole  To calculate the average molecular weights of different hydrocarbon cuts in the liquid products, a methodology presented by Maxwell [168] was used. The first step in calculating the molecular weight is to determine the volume average boiling point (VABP) of each hydrocarbon cut using the simulated distillation results. VABP of a fraction is calculated as follow: ???? =??+?????+?????        ( K-3) where ?0, ?50 ?100 are the temperatures (?F) at which 0, 50 and 100 vol% of the liquid is out of the distillation column respectively. 307  The next stem is the conversion of VABP to mean average boiling point (MEABP). This is done by calculating the slope of the simulated distillation curve as follow: ?(?????) =?????????        ( K-4) where ?70 and ?10 are temperatures (?F) at which 10 and 70 % of the liquid is out of the distillation column respectively. Knowing the slope (S) and the VABP and using the Maxwell?s correlation data [168], a temperature differential (??) is calculated and then the MEABP is calculated as follow: ???? = ???? + ??         ( K-5) Knowing the MEABP and the average API density of the liquid, the molecular weight of the hydrocarbon fraction can be calculated using the Maxwell?s correlation data [168]. In the present study, molecular weights of different hydrocarbon cuts of the liquids produced in the hydroconversion reactions using different reaction times (0, 60 and 120 mins) were compared and very small differences were observed between different experiments. An average of molecular weights of the hydrocarbon cuts from various experiments was used as the molecular weights of different fractions in the liquid. These results are presented in Table  K.2.   308  Table  K.2 Molecular weight of different HC cuts at different hydroconversion experiments.  Reaction time   Molecular weight of fraction, g/mole 0 60 120  Average Naphtha 126.5 130 128  128.17 LGO 215 206 208.5  209.83 HGO 395 380 360  378.33 Residue 562 571 562  565.00  To estimate the Molecular weight of the coke generated in hydroconversion experiments, an empirical correlation presented by Speight [6] was used, which related the H/C atom ratio of hydrocarbons to the average molecular weight. The reproduced correlation is presented in Figure  K.1. In the present study, molecular weight of the coke generated in the experiment using 600 ppm Mo added to the reactor in the form of Mo-octoate and 1 h reaction time was used as the average molecular weight for the coke lump. This coke had an H/C atom ratio of 0.71 and its molecular weight was estimated to be 5267.33 g/mole. 309   Figure  K.1 Relative H/C atom ratio and molecular weight of refinery feedstocks. (Reproduced from Figure 14.18 of [6], Copyright ? 2006, Taylor & Francis Group LLC., with permission)        y = -0.319ln(x) + 3.4436R? = 0.992900.511.522.510 100 1000 10000H/C atom ratioAverage molecular weight, g/mole310  Appendix L MATLAB M-files for Objective Function Minimization and Parameter Estimation Main m-file codes for obejctive function minimization, parameter estimation and A* matrix calculations for statistical analysis: clear all;   global nvar nx nt x0 y0 y t T Y Ymodel dy  % Added by Hooman global CC CR CH CL CN CG          % For tracking concentrations global Model_Mass_Balance   h = 1e-10;   T = [0  % time of reactions 30 60 120]';   % Reading data from Excel file: Raw Kinetic Data.xlsx   %%%%%%%%%%%%%%%%%%%%%%%%%%%% 460 C %%%%%%%%%%%%%%%%%%%%%%   % % CCX: Coke concentration, mole % CCX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U59:X59'); %  % % CRX: Residue concentration, mole % CRX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U58:X58'); %  % % CHX: HGO concentration, mole % CHX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U57:X57'); %  % % CLX: LGO concentration, mole % CLX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U56:X56'); %  % % CNX: Naphtha concentration, mole % CNX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U55:X55'); %  % % CGX: Gas concentration, mole % CGX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U61:X61'); 311  %   %%%%%%%%%%%%%%%%%%%%%%%%%%%% 445 C %%%%%%%%%%%%%%%%%%%%%%   % CCX: Coke concentration, mole CCX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U71:X71');   % CRX: Residue concentration, mole CRX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U70:X70');   % CHX: HGO concentration, mole CHX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U69:X69');   % CLX: LGO concentration, mole CLX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U68:X68');   % CNX: Naphtha concentration, mole CNX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U67:X67');   % CGX: Gas concentration, mole CGX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U73:X73');   %%%%%%%%%%%%%%%%%%%%%%%%%%%% 430 C %%%%%%%%%%%%%%%%%%%%%%   % % CCX: Coke concentration, mole % CCX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U83:X83'); %  % % CRX: Residue concentration, mole % CRX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U82:X82'); %  % % CHX: HGO concentration, mole % CHX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U81:X81'); %  % % CLX: LGO concentration, mole % CLX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U80:X80'); %  % % CNX: Naphtha concentration, mole % CNX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U79:X79'); %  % % CGX: Gas concentration, mole 312  % CGX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U85:X85'); %     %%%%%%%%%%%%%%%%%%%%%%%%%%%% 415 C %%%%%%%%%%%%%%%%%%%%%%   % % CCX: Coke concentration, mole % CCX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U95:X95'); %  % % CRX: Residue concentration, mole % CRX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U94:X94'); %  % % CHX: HGO concentration, mole % CHX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U93:X93'); %  % % CLX: LGO concentration, mole % CLX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U92:X92'); %  % % CNX: Naphtha concentration, mole % CNX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U91:X91'); %  % % CGX: Gas concentration, mole % CGX =xlsread('Raw Kinetic Data.xlsx','Modified SIMDIS_ Normalized','U97:X97'); %   %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%    CCX=CCX'; CRX=CRX'; CHX=CHX'; CLX=CLX'; CNX=CNX'; CGX=CGX';   nvar=6; % Number of lumps in the proposed mechanism i.e. Coke, Residue, ...    t=T;  nt=length (t)  nx=nt-1   y = [CCX CRX CHX CLX CNX CGX];  % Matrix of measured concentrations    % inital guess for parameters, i.e. k values 313     k0=[1      0.01      0.01      0.01      0.01      0.01];    % fminsearch function to minimize the objective function  OPTIONS = optimset('display', 'iter','LargeScale', 'off',...                      'MaxIter', 10000, 'MaxFunEvals', 10000);     [x,FVAL,EXITFLAG,output] = fminsearch('ObjectiveFunc_mole', k0, ... OPTIONS, T,CCX,CRX,CHX,CLX,CNX,CGX);       %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% x0 = [CCX(1)      CRX(1)         CHX(1)      CLX(1)      CGX(1)];       tspan=0:1:120;     tic                      [t,yx1r]=ode45('ODEfunm_mole',tspan,x0,[],[x(1)+h x(2) x(3) x(4) x(5) x(6)]);     [t,yx2r]=ode45('ODEfunm_mole',tspan,x0,[],[x(1) x(2)+h x(3) x(4) x(5) x(6)]);     [t,yx3r]=ode45('ODEfunm_mole',tspan,x0,[],[x(1) x(2) x(3)+h x(4) x(5) x(6)]);     [t,yx4r]=ode45('ODEfunm_mole',tspan,x0,[],[x(1) x(2) x(3) x(4)+h x(5) x(6)]);     [t,yx5r]=ode45('ODEfunm_mole',tspan,x0,[],[x(1) x(2) x(3) x(4) x(5)+h x(6)]);     [t,yx6r]=ode45('ODEfunm_mole',tspan,x0,[],[x(1) x(2) x(3) x(4) x(5) x(6)+h]);          [t,yx1l]=ode45('ODEfunm_mole',tspan,x0,[],[x(1)-h x(2) x(3) x(4) x(5) x(6)]);     [t,yx2l]=ode45('ODEfunm_mole',tspan,x0,[],[x(1) x(2)-h x(3) x(4) x(5) x(6)]);     [t,yx3l]=ode45('ODEfunm_mole',tspan,x0,[],[x(1) x(2) x(3)-h x(4) x(5) x(6)]);     [t,yx4l]=ode45('ODEfunm_mole',tspan,x0,[],[x(1) x(2) x(3) x(4)-h x(5) x(6)]);     [t,yx5l]=ode45('ODEfunm_mole',tspan,x0,[],[x(1) x(2) x(3) x(4) x(5)-h x(6)]);     [t,yx6l]=ode45('ODEfunm_mole',tspan,x0,[],[x(1) x(2) x(3) x(4) x(5) x(6)-h]); 314    toc              %%%%%%%%% Calculating the differential (central) %%%%%%%%%%     df1dx1=(yx1r(:,1)-yx1l(:,1))/(2*h);    df1dx2=(yx2r(:,1)-yx2l(:,1))/(2*h);    df1dx3=(yx3r(:,1)-yx3l(:,1))/(2*h);    df1dx4=(yx4r(:,1)-yx4l(:,1))/(2*h);    df1dx5=(yx5r(:,1)-yx5l(:,1))/(2*h);    df1dx6=(yx6r(:,1)-yx6l(:,1))/(2*h);        df2dx1=(yx1r(:,2)-yx1l(:,2))/(2*h);    df2dx2=(yx2r(:,2)-yx2l(:,2))/(2*h);    df2dx3=(yx3r(:,2)-yx3l(:,2))/(2*h);    df2dx4=(yx4r(:,2)-yx4l(:,2))/(2*h);    df2dx5=(yx5r(:,2)-yx5l(:,2))/(2*h);    df2dx6=(yx6r(:,2)-yx6l(:,2))/(2*h);      df3dx1=(yx1r(:,3)-yx1l(:,3))/(2*h);    df3dx2=(yx2r(:,3)-yx2l(:,3))/(2*h);    df3dx3=(yx3r(:,3)-yx3l(:,3))/(2*h);    df3dx4=(yx4r(:,3)-yx4l(:,3))/(2*h);    df3dx5=(yx5r(:,3)-yx5l(:,3))/(2*h);    df3dx6=(yx6r(:,3)-yx6l(:,3))/(2*h);        df4dx1=(yx1r(:,4)-yx1l(:,4))/(2*h);    df4dx2=(yx2r(:,4)-yx2l(:,4))/(2*h);    df4dx3=(yx3r(:,4)-yx3l(:,4))/(2*h);    df4dx4=(yx4r(:,4)-yx4l(:,4))/(2*h);    df4dx5=(yx5r(:,4)-yx5l(:,4))/(2*h);    df4dx6=(yx6r(:,4)-yx6l(:,4))/(2*h);        df5dx1=(yx1r(:,5)-yx1l(:,5))/(2*h);    df5dx2=(yx2r(:,5)-yx2l(:,5))/(2*h);    df5dx3=(yx3r(:,5)-yx3l(:,5))/(2*h);    df5dx4=(yx4r(:,5)-yx4l(:,5))/(2*h);    df5dx5=(yx5r(:,5)-yx5l(:,5))/(2*h);    df5dx6=(yx6r(:,5)-yx6l(:,5))/(2*h);    Jac_Matrix = [df1dx1 df1dx2 df1dx3 df1dx4 df1dx5 df1dx6               df2dx1 df2dx2 df2dx3 df2dx4 df2dx5 df2dx6               df3dx1 df3dx2 df3dx3 df3dx4 df3dx5 df3dx6               df4dx1 df4dx2 df4dx3 df4dx4 df4dx5 df4dx6               df5dx1 df5dx2 df5dx3 df5dx4 df5dx5 df5dx6];      Astar_Matrix = Jac_Matrix'*Jac_Matrix;   315  Astar_Matrix_Inv = inv(Astar_Matrix);   %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%   % Solving ODEs after objective function minimization; Naphtha is calculated by forcing the mass balance     tspan=0:1:120;     tic        x0 = [CCX(1)      CRX(1)       CHX(1)      CLX(1)      CGX(1)];        [T,Ymodel]=ode45('ODEfunm_mole',tspan,x0,[],x);          toc    %%%%%%%%%%%%%%%     Naphtha iscalculated by forcing mass balance     CC  = Ymodel(:,1);   CR  = Ymodel(:,2);   CH  = Ymodel(:,3);   CL  = Ymodel(:,4);   CG  = Ymodel(:,5);   CN  = (80 - ((CC*700)+(CR*565)+(CH*378)+(CL*210)+(CG*24.56)))/128.17;   %%%%%%%%%%%      Mass balance check for the model  Model_Mass_Balance = (CC*700 + CR*565 + CH*378 + CL*210 + CN*128.17 + CG*24.56 );    %%%%%%%%%%% disp ('X-values:')     disp (t)  disp ('Y-values:')     disp (y)  disp ('Model calculated responses:')     disp (Y) disp ('Estimated parameter values are:')     disp (x) disp ('Number of iterations:')     disp (output)      subplot(3,3,1)                                                                 plot(t,CCX,'o',tspan,CC,'--')                title('Coke')               xlabel('Time (min)')               ylabel('Concentration (mole)') subplot(3,3,2) 316                plot(t,CRX,'o',tspan,CR,'--')               title('Residue')                xlabel('Time (min)')               ylabel('Concentration (mole)') subplot(3,3,3)               plot(t,CHX,'o',tspan,CH,'--')                title('HGO')                xlabel('Time (min)')               ylabel('Concentration (mole)') subplot(3,3,4)               plot(t,CLX,'o',tspan,CL,'--')                title('LGO')                xlabel('Time (min)')               ylabel('Concentration (mole)') subplot(3,3,5)               plot(t,CNX,'o',tspan,CN,'--')               title('Naphtha')                xlabel('Time (min)')               ylabel('Concentration (mole)') subplot(3,3,6)               plot(t,CGX,'o',tspan,CG,'--')               title('Gas')                xlabel('Time (min)')               ylabel('Concentration (mole)')                subplot(3,3,7)               plot(tspan,Model_Mass_Balance,'--')               title('Mass balance check')                xlabel('Time (min)')               ylabel('Total mass (g)')         317  M-file for objective function formation: function OBJ=ObjectiveFunc_mole(x,T,CCX,CRX,CHX,CLX,CNX,CGX)    global CC CR CH CL CN CG Y  global nt nvar y  global Conc_Average   x0 = [CCX(1)      CRX(1)        CHX(1)      CLX(1)      CGX(1)];  [T,Y]=ode45('ODEfunm_mole',T,x0,[],x);      CC = Y(:,1); CR = Y(:,2); CH = Y(:,3); CL = Y(:,4); CG = Y(:,5); CN = y(:,5) + (((y(:,1)-CC)*700+(y(:,2)-CR)*565+(y(:,3)-CH)*378+(y(:,4)-CL)*210+(y(:,6)-CG)*24.56)/128.17);    % Naphtha calculation using mass balance        OBJ=sum( (CC-CCX).^2  + (CR-CRX).^2+...          (CH-CHX).^2 + (CL-CLX).^2 + (CN-CNX).^2 + (CG-CGX).^2 );         318  M-file for ODE formation: function dy = ODEfunm_mole(t,y,flag,p)  format long global dy   dy = zeros(5,1);   % naphtha is calculated by forcing mass balance   %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%   b = p(1); k_1 = p(2); k1  = p(3); k2  = p(4); k3  = p(5); k4  = p(6);  %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%   % Mechanism 5: Naphtha calculations based on the mass balance.   dy(1) = k_1*y(2)-k1*y(1);              %production rate of coke dy(2) = k1*y(1)-(k_1+k2)*y(2);         %production rate of Resdiue dy(3) = k2*y(2)-k3*y(3);               %production rate of HGO dy(4) = k3*y(3)-k4*y(4);               %production rate of LGO dy(5) = k2*y(2)+b*k3*y(3);             %production rate of Gas   %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%        319  Appendix M Statistical Analysis of the Kinetic Model M.1 Significance and Calculation of p-vaue As described in Section  8.2.4, p-value is an indication of the proximity of the experimental values to the values calculated by the kinetic model. P-value (which ranges from 0 to 1) closer to 1 is an indication of a good fit of the model on the experimental values. To calculate the p-value for the different kinetic lumps, a one-way 1-way ANOVA analysis was done using the Originpro v.8.6 software. Detailed results of 1-way ANOVA are presented in  Appendix N. M.2 Calculation of Standard Deviation and Confidence Interval Procedure explained by Englezos [169] was used to calculate the standard deviation of the k-values calculated by the model. The covariance matrix of parameters (COV(k*)) is calculated as follow: COV(k*) = ??????        ( 9) where ??2 is the is the varianve of parameters and matrix A is calculated using the jacubian matrix and the inversion of the jacubian matrix of the equatiins calculated at k*. A good estimation (???2) of the variance (??2) can be calculated as follow: ???? =???(??)?.?         ( 9) 320  where SLS(??) is the least square of reaction rates calculated at k* (optimized parameters) and d.f is the degree of freedom which is expressed as follow: d.f = Nm-p = (number of dependent variables?number of datapoint ? number of variables) The standard error associated with each calculated parameter (??ki) iscalculated as followusing square root of the corresponding diagonal elements of the matrix A-1: ???? = ????{???}??        ( 9) Knowing the ??ki , the marginal confidence interval ((1-?)100%) of each estimated parameter (ki) can be expresses as follow: ??? ? ??/?? ??? ? ?? ? ??? + ??/?? ???       ( M-1) where t?/??  is obtained from the T-distribution table at with ? = N?? ? and ? is margine of error. In the present study, ? and ? were 0.05 and 19 respectively and t?/??  was 2.093 from the Student?s T-distribution table.     321  Appendix N Anova analysis and Error Calculation for Experimental and Model-predicted Values of Kinetic Lumps Table  N.1 Anova analysis and error calculation for coke at 415 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     8.67E-05 8.67E-05         0.000153 0.000156  SUMMARY       0.000257 0.000218  Groups Count Sum Average Variance   0.000287 0.00032  Experimental 4 7.84E-04 1.96E-04 8.62E-09      Model predicted 4 7.81E-04 1.95E-04 9.82E-09                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 1.19E-12 1 1.19E-12 0.000129 0.991 5.99    Within Groups 5.53E-08 6 9.22E-09                 Total 5.53E-08 7        322  Table  N.2 Anova analysis and error calculation for residue at 415 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.0617 0.0617         0.049635 0.054567  SUMMARY       0.045476 0.048259  Groups Count Sum Average Variance   0.039827 0.037746  Experimental 4 0.196638 0.04916 8.6E-05      Model predicted 4 0.202271 0.050568 0.000103                          1- way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 3.97E-06 1 3.97E-06 0.041917 0.845 5.99    Within Groups 0.000568 6 9.46E-05                 Total 0.000572 7         323  Table  N.3 Anova analysis and error calculation for HGO at 415 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.095963 0.095963         0.077818 0.084508  SUMMARY       0.073184 0.074442  Groups Count Sum Average Variance   0.063687 0.057812  Experimental 4 0.310652 0.077663 0.000183      Model predicted 4 0.312725 0.078181 0.000262                          1- way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 5.37E-07 1 5.37E-07 0.002413 0.962 5.99    Within Groups 0.001335 6 0.000223                 Total 0.001336 7         324  Table  N.4 Anova analysis and error calculation for LGO at 415 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.02624 0.02624         0.060208 0.044053  SUMMARY       0.064697 0.059612  Groups Count Sum Average Variance   0.07496 0.085024  Experimental 4 0.226105 0.056526 0.000446      Model predicted 4 0.214929 0.053732 0.000621                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 1.56E-05 1 1.56E-05 0.029268 0.870 5.99    Within Groups 0.003201 6 0.000533                 Total 0.003216 7         325  Table  N.5 Anova analysis and error calculation for naphtha at 415 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.018121 0.018336         0.060049 0.045191  SUMMARY       0.076 0.069087  Groups Count Sum Average Variance   0.084556 0.109368  Experimental 4 0.238726 0.059681 0.000871      Model predicted 4 0.241982 0.060496 0.001491                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 1.33E-06 1 1.33E-06 0.001122 0.974 5.99    Within Groups 0.007087 6 0.001181                 Total 0.007088 7         326  Table  N.6 Anova analysis and error calculation for gas at 415 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.022322 0.022322         0.056159 0.054496  SUMMARY       0.079346 0.082857  Groups Count Sum Average Variance   0.131404 0.129915  Experimental 4 0.289231 0.072308 0.0021      Model predicted 4 0.28959 0.072397 0.002082                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 1.61E-08 1 1.61E-08 7.68E-06 0.998 5.99    Within Groups 0.012547 6 0.002091                 Total 0.012547 7          327  Table  N.7 Anova analysis and error calculation for coke at 430 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.000791 0.000791         0.001622 0.001401  SUMMARY       0.001837 0.001879  Groups Count Sum Average Variance   0.002343 0.002548  4 0.006593 0.001648 4.18E-07 4      4 0.006619 0.001655 5.53E-07 4                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 8.34E-11 1 8.34E-11 0.000172 0.990 5.99    Within Groups 2.91E-06 6 4.85E-07                 Total 2.91E-06 7         328  Table  N.8 Anova analysis and error calculation for residue at 430 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.000105 0.000105         0.000216 0.00019  SUMMARY       0.000244 0.000257  Groups Count Sum Average Variance   0.000311 0.00035  Experimental 4 0.000876 0.000219 7.37E-09      Model predicted 4 0.000901 0.000225 1.07E-08                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 7.87E-11 1 7.87E-11 0.008704 0.929 5.99    Within Groups 5.42E-08 6 9.04E-09                 Total 5.43E-08 7         329  Table  N.9 Anova analysis and error calculation for HGO at 430 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.092073 0.092073         0.078153 0.08148  SUMMARY       0.068891 0.071075  Groups Count Sum Average Variance   0.058123 0.052364  Experimental 4 0.29724 0.07431 0.000207      Model predicted 4 0.296991 0.074248 0.000286                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 7.69E-09 1 7.69E-09 3.12E-05 0.996 5.99    Within Groups 0.001481 6 0.000247                 Total 0.001481 7         330  Table  N.10 Anova analysis and error calculation for LGO at 430 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.037069 0.037069         0.072399 0.058676  SUMMARY       0.086364 0.07729  Groups Count Sum Average Variance   0.09354 0.106224  Experimental 4 0.289373 0.072343 0.00063      Model predicted 4 0.279258 0.069815 0.000859                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 1.28E-05 1 1.28E-05 0.017174 0.900 5.99    Within Groups 0.004468 6 0.000745                 Total 0.004481 7         331  Table  N.11 Anova analysis and error calculation for naphtha at 430 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.028293 0.028484         0.075449 0.068754  SUMMARY       0.112284 0.1034  Groups Count Sum Average Variance   0.149799 0.158623  Experimental 4 0.365824 0.091456 0.002694      Model predicted 4 0.359261 0.089815 0.003041                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 5.38E-06 1 5.38E-06 0.001877 0.967 5.99    Within Groups 0.017208 6 0.002868                 Total 0.017213 7         332  Table  N.12 Anova analysis and error calculation for gas at 430 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.038902 0.038902         0.081244 0.072705  SUMMARY       0.104696 0.1013  Groups Count Sum Average Variance   0.140588 0.145442  Experimental 4 0.36543 0.091358 0.001818      Model predicted 4 0.358349 0.089587 0.002037                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 6.27E-06 1 6.27E-06 0.003252 0.956 5.99    Within Groups 0.011566 6 0.001928                 Total 0.011573 7         333  Table  N.13 Anova analysis and error calculation for coke at 445 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.00013 0.00013         0.00029 0.000301  SUMMARY       0.0003 0.000324  Groups Count Sum Average Variance   0.000227 0.000237  Experimental 4 0.000956 0.000239 6.04E-09      Model predicted 4 0.00083 0.000207 9.09E-09                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 1.99E-09 1 1.99E-09 0.263106 0.826 5.99    Within Groups 4.54E-08 6 7.56E-09                 Total 4.74E-08 7         334  Table  N.14 Anova analysis and error calculation for residue at 445 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.049194 0.049194         0.027755 0.031591  SUMMARY       0.021172 0.020736  Groups Count Sum Average Variance   0.010461 0.009464  Experimental 4 0.108582 0.027146 0.000267      Model predicted 4 0.110985 0.027746 0.000286                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 7.22E-07 1 7.22E-07 0.00261 0.961 5.99    Within Groups 0.001659 6 0.000276                 Total 0.00166 7         335  Table  N.15 Anova analysis and error calculation for HGO at 445 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.083962 0.083962         0.067953 0.073591  SUMMARY       0.055874 0.061374  Groups Count Sum Average Variance   0.044488 0.039386  Experimental 4 0.252277 0.063069 0.000286      Model predicted 4 0.258313 0.064578 0.000367                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 4.55E-06 1 4.55E-06 0.013943 0.910 5.99    Within Groups 0.001959 6 0.000327                 Total 0.001964 7         336  Table  N.16 Anova analysis and error calculation for LGO at 445 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.051946 0.051946         0.094496 0.079355  SUMMARY       0.111544 0.103076  Groups Count Sum Average Variance   0.130062 0.139148  Experimental 4 0.388047 0.097012 0.001114      Model predicted 4 0.373524 0.093381 0.001367                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 2.64E-05 1 2.64E-05 0.021254 0.889 5.99    Within Groups 0.007443 6 0.00124                 Total 0.007469 7         337  Table  N.17 Anova analysis and error calculation for naphtha at 445 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.057709 0.057858         0.111178 0.100512  SUMMARY       0.137022 0.132975  Groups Count Sum Average Variance   0.162564 0.17524  Experimental 4 0.468472 0.117118 0.002009      Model predicted 4 0.466586 0.116647 0.002472                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 4.45E-07 1 4.45E-07 0.000198 0.989 5.99    Within Groups 0.013442 6 0.00224                 Total 0.013443 7         338  Table  N.18 Anova analysis and error calculation for gas at 445 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.059393 0.059393         0.128077 0.133649  SUMMARY       0.167561 0.193669  Groups Count Sum Average Variance   0.295638 0.278473  Experimental 4 0.650669 0.162667 0.009856      Model predicted 4 0.665184 0.166296 0.008609                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 2.63E-05 1 2.63E-05 0.002852 0.959 5.99    Within Groups 0.055394 6 0.009232                 Total 0.055421 7         339  Table  N.19 Anova analysis and error calculation for coke at 460 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.000224 0.000224         0.00041 0.000364  SUMMARY       0.000336 0.000354  Groups Count Sum Average Variance   0.000259 0.000267  Experimental 4 0.001228 0.000307 6.89E-09      Model predicted 4 0.001208 0.000302 4.65E-09                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 5.02E-11 1 5.02E-11 0.008702 0.929 5.99    Within Groups 3.46E-08 6 5.77E-09                 Total 3.47E-08 7         340  Table  N.20 Anova analysis and error calculation for residue at 460 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.036374 0.036374         0.014721 0.018328  SUMMARY       0.009848 0.009942  Groups Count Sum Average Variance   0.006797 0.003525  Experimental 4 0.067741 0.016935 0.000179      Model predicted 4 0.068169 0.017042 0.000203                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 2.3E-08 1 2.3E-08 0.00012 0.992 5.99    Within Groups 0.001144 6 0.000191                 Total 0.001144 7         341  Table  N.21 Anova analysis and error calculation for HGO at 460 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.072026 0.072026         0.049557 0.052694  SUMMARY       0.034001 0.036008  Groups Count Sum Average Variance   0.026433 0.015537  Experimental 4 0.182017 0.045504 0.000405      Model predicted 4 0.176265 0.044066 0.000578                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 4.13E-06 1 4.13E-06 0.008408 0.930 5.99    Within Groups 0.002951 6 0.000492                 Total 0.002955 7         342  Table  N.22 Anova analysis and error calculation for LGO at 460 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.077628 0.077628         0.11758 0.107934  SUMMARY       0.124519 0.12624  Groups Count Sum Average Variance   0.134595 0.138463  Experimental 4 0.454321 0.11358 0.000623      Model predicted 4 0.450265 0.112566 0.0007                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 2.06E-06 1 2.06E-06 0.003108 0.957 5.99    Within Groups 0.00397 6 0.000662                 Total 0.003972 7         343  Table  N.23 Anova analysis and error calculation for naphtha at 460 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.093834 0.093918         0.167116 0.152438  SUMMARY       0.209657 0.196235  Groups Count Sum Average Variance   0.233542 0.257001  Experimental 4 0.704149 0.176037 0.003758      Model predicted 4 0.699591 0.174898 0.004753                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 2.6E-06 1 2.6E-06 0.00061 0.981 5.99    Within Groups 0.025532 6 0.004255                 Total 0.025535 7         344  Table  N.24 Anova analysis and error calculation for gas at 460 ?C Experimental Model predicted  Anova: Single Factor ?= 0.05     0.110176 0.110176         0.190772 0.215363  SUMMARY       0.276962 0.286501  Groups Count Sum Average Variance   0.374112 0.36378  Experimental 4 0.952022 0.238005 0.012871      Model predicted 4 0.97582 0.243955 0.011628                          1-way ANOVA Accept Null Hypothesis because p > 0.05 (Means are the same)     Source of Variation SS df MS F P-Value FCritical    Between Groups 7.08E-05 1 7.08E-05 0.005779 0.942 5.99    Within Groups 0.073496 6 0.012249                 Total 0.073567 7      345  

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.24.1-0073852/manifest

Comment

Related Items