Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

A study of metal phosphides for the hydrodeoxygenation of phenols and pyrolysis oil Whiffen, Victoria 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_whiffen_victoria.pdf [ 3.42MB ]
Metadata
JSON: 24-1.0073967.json
JSON-LD: 24-1.0073967-ld.json
RDF/XML (Pretty): 24-1.0073967-rdf.xml
RDF/JSON: 24-1.0073967-rdf.json
Turtle: 24-1.0073967-turtle.txt
N-Triples: 24-1.0073967-rdf-ntriples.txt
Original Record: 24-1.0073967-source.json
Full Text
24-1.0073967-fulltext.txt
Citation
24-1.0073967.ris

Full Text

A Study of Metal Phosphides for the Hydrodeoxygenation of Phenols and Pyrolysis Oil by Victoria Whiffen B.Eng., Dalhousie University, 2007  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) July 2013 © Victoria Whiffen, 2013  Abstract This dissertation addresses the hydrodeoxygenation (HDO) of the model compound 4methylphenol and pyrolysis oil, over alternative, non-sulfided catalysts. The HDO of 4methylphenol was studied over unsupported, low surface area MoS2, MoO2, MoO3, and MoP catalysts. The initial turn over frequency (TOF) for the HDO of 4-methylphenol decreased in the order MoP > MoS2 > MoO2 > MoO3. Among the catalysts examined, MoP had the highest hydrogenating selectivity, lowest activation energy, and per site activity (TOF) for the HDO of 4-methylphenol. However, the observed conversion over MoP was limited by its low surface area and CO uptake. Addition of citric acid (CA) improved the properties of unsupported MoP. CA acted as a structural promoter and formed a metal citrate during the catalyst preparation, which increased the surface area and CO uptake of the MoP. High surface area Ni2P catalysts were prepared similarly and based on initial TOFs, Ni2P was 6 times more active than MoP for the HDO of 4-methylphenol. The HDO of 4-methylphenol was found to be structure insensitive over both MoP and Ni2P. However, the Ni2P catalysts deactivated due to C deposition on the catalyst surface. A kinetic model of the direct deoxygenation and hydrogenation reaction pathways for the HDO reaction over MoP showed the former to have a higher barrier energy (Ea = 106 kJ/mol) than the latter (Ea = 85 kJ/mol). Finally, to validate the use of model compounds to screen catalysts, the HDO of 4methylphenol was compared to the HDO of pyrolysis oil over sulfide, oxide, and phosphide catalysts. MoP was found to have the highest yield of O-free liquid and the lowest coke yield, followed by Ni2P, NiMoS/Al2O3, MoS2, and MoO3 for the HDO of pyrolysis oil. Those catalysts displaying high hydrogenating abilities had a high degree of O free liquid and a low ii  yield of coke (MoP), while those catalysts displaying high isomerization abilities (MoO3) had a high coke yield. Overall, this thesis identified phosphide catalysts as a new class of catalysts for HDO reactions with strong hydrogenating abilities, and their activity was superior to commercial NiMoS/Al2O3 for the HDO of pyrolysis oil.  iii  Preface This Ph.D. thesis consists of eight chapters. Chapters 4-6 have been published previously in the literature. The Ph.D. study was conducted by Victoria Whiffen under the direct supervision of Professor Kevin Smith in the Department of Chemical and Biological Engineering at UBC. The reactor set-up, catalyst preparation, catalyst characterization, catalyst testing, data collection, data analysis, data interpretation, kinetic modeling, literature review, and dissertation preparation was done by Victoria Whiffen under the supervision of Professor Kevin Smith. The list of the manuscripts included in this dissertation are given below:  1. V.M.L. Whiffen, and K. J. Smith, “Hydrodeoxygenation of 4-Methylphenol over Unsupported MoP, MoS2, and MoOx Catalysts,” Energy & Fuels, 24 (9), 2010, pp. 47284737. A version of this manuscript is contained in Chapter 4. Victoria Whiffen undertook the set-up and modification of the batch reactor to measure the hydrodeoxygenation reaction activity. Victoria Whiffen performed the MoP catalyst preparation, catalyst characterization, hydrodeoxygenation reactions, kinetic modeling, as well as data analysis and interpretation. In addition the manuscript was prepared and written by Victoria Whiffen under the supervision and approval of Professor Kevin Smith. XPS measurements were performed by Dr. Ken Wong from the Interfacial Analysis and Reactivity Laboratory at UBC.  iv  2. V.M.L. Whiffen, and K.J. Smith, “The Effect of Calcination Temperature on the Properties of MoP for the Hydrodeoxygenation of 4-Methylphenol,” Nanocatalysis for Fuels and Chemicals – ACS Symposium Series, Vol. 1092; Ed. A. Dalai; Chapter 5; Washington, DC, 2012, pp. 61-73. A version of this manuscript is contained in Chapter 5. Victoria Whiffen performed the MoP catalyst preparation, catalyst characterization, hydrodeoxygenation reactions, kinetic modeling, as well as data analysis and interpretation. In addition, the manuscript was prepared and written by Victoria Whiffen under the supervision and approval of Professor Kevin Smith. TEM measurements were done by Bradford Ross in the Department of Botany at UBC.  3. V.M.L. Whiffen, K.J. Smith, and S. Straus, “The Influence of Citric Acid on the Synthesis and Activity of High Surface Area MoP for the Hydrodeoxygenation of 4-Methylphenol,” Applied Catalysis A, 419-420, 2012, pp. 111-125. A version of this manuscript is contained in Chapter 5. Victoria Whiffen performed the MoP catalyst preparation, catalyst characterization, hydrodeoxygenation reactions, kinetic modeling, as well as data analysis and interpretation. The mathematical code for the kinetic analysis was written by Victoria Whiffen and adapted by Professor Kevin Smith. The kinetic model was based on independent reactions of the product species over the catalysts tested. In addition, the manuscript was prepared and written by Victoria Whiffen under the supervision and approval of Professor Kevin Smith. NMR measurements were performed by Dr. Suzana Straus from the Department of Chemistry at UBC and TEM measurements were done by Bradford Ross in the Department of Botany at UBC.  v  4. V.M.L. Whiffen, and K.J. Smith, “A Comparative Study of 4-Methylphenol Hydrodeoxygenation Over High Surface Area MoP and Ni2P,” Topics in Catalysis, 55, 2012, pp. 981-990. A version of this manuscript is contained in Chapter 6. Victoria Whiffen performed the MoP and Ni2P catalyst preparation, catalyst characterization, hydrodeoxygenation reactions, kinetic modeling, as well as data analysis and interpretation. The effect of catalyst deactivation was also investigated by Victoria Whiffen by intentionally oxidizing the catalyst surfaces. In addition the manuscript was prepared and written by Victoria Whiffen under the supervision and approval of Professor Kevin Smith. TEM measurements were done by Bradford Ross in the Department of Botany at UBC.  5. V.M.L. Whiffen, and K.J. Smith, “The Effect of Calcination Temperature on the Properties and Hydrodeoxygenation Activity of Ni2P Catalysts Prepared Using Citric Acid,” Novel Materials for Catalysis and Fuels Processing – ACS Symposium Series (accepted November 2012). A version of this manuscript is contained in Chapter 6. Victoria Whiffen performed the Ni2P catalyst preparation, catalyst characterization, hydrodeoxygenation reactions, kinetic modeling, as well as data analysis and interpretation. In addition, the manuscript was prepared and written by Victoria Whiffen under the supervision and approval of Professor Kevin Smith. TEM measurements were done by Bradford Ross in the Department of Botany at UBC.  vi  Table of Contents Abstract .......................................................................................................................... ii Preface ........................................................................................................................... iv Table of Contents ........................................................................................................ vii List of Tables................................................................................................................ xv List of Figures .............................................................................................................. xx Nomenclature ............................................................................................................ xxvi Acronyms ................................................................................................................... xxx Acknowledgements ................................................................................................. xxxiii Dedication ................................................................................................................ xxxv Chapter 1 - Introduction............................................................................................... 1 1.1  Background ............................................................................................... 1  1.2  Objectives of the Thesis ............................................................................ 4  1.3  Work Plan ................................................................................................. 4  1.4  Thesis Layout ............................................................................................ 7  Chapter 2 - Literature Review ..................................................................................... 8 2.1 2.1.1 2.2  Lignocellulosic Fast Pyrolysis Oil ............................................................ 8 Phenolic Composition of Pyrolysis Oil................................................... 10 HDO of Phenolic Compounds ................................................................ 12 vii  2.3 2.3.1 2.4  Metal Sulfides for HDO .......................................................................... 14 Metal Sulfide Active Sites ...................................................................... 16 Metal Phosphides for HDO ..................................................................... 18  2.4.1  Preparation of High Surface Area Metal Phosphides Using Citric Acid.22  2.4.2  Metal Phosphide Active Sites ................................................................. 24  2.5  Pyrolysis Oil HDO .................................................................................. 25  2.6  Emerging Commercial HDO Processes .................................................. 30  2.7  Concluding Remarks ............................................................................... 33  Chapter 3 - Experimental Methods ........................................................................... 35 3.1  Catalyst Preparation ................................................................................ 35  3.1.1  Molybdenum Phosphide ......................................................................... 35  3.1.2  Nickel Phosphide .................................................................................... 37  3.2  Catalyst Characterization ........................................................................ 38  3.2.1  Elemental Analysis ................................................................................. 38  3.2.2  TGA ........................................................................................................ 39  3.2.3  DRIFTS ................................................................................................... 39  3.2.4  13  C NMR ................................................................................................. 40 viii  3.2.5  XRD ........................................................................................................ 41  3.2.6  XPS ......................................................................................................... 42  3.2.7  BET ......................................................................................................... 42  3.2.8  TPR ......................................................................................................... 42  3.2.9  TPO ......................................................................................................... 43  3.2.10  CO Chemisorption .................................................................................. 44  3.2.11  Acid Site Titration................................................................................... 45  3.2.12  SEM and TEM ........................................................................................ 45  3.2.13  Precursor Structures ................................................................................ 46  3.3  Catalyst Activity Tests ............................................................................ 46  3.3.1  4-Methylphenol HDO ............................................................................. 46  3.3.2  4-Methylphenol Kinetic Analysis ........................................................... 50  3.4  Pyrolysis Oil HDO .................................................................................. 51  Chapter 4 - Catalyst Screening for the HDO of 4-Methylphenol ........................... 54 4.1  Introduction ............................................................................................. 54  4.2  Results and Discussion ........................................................................... 55  4.2.1  Catalyst Characterization ........................................................................ 56 ix  4.2.2  Catalyst Activity ..................................................................................... 69  4.2.3  Product Distribution ................................................................................ 73  4.2.4  Catalyst Active Sites ............................................................................... 78  4.3  Conclusions ............................................................................................. 80  Chapter 5 - The Preparation of High Surface Area MoP for HDO ....................... 81 5.1  Introduction ............................................................................................. 81  5.2  Results and Discussion ........................................................................... 83  5.2.1  Characterization of the Dried Precursors ................................................ 83  5.2.2  Characterization of Calcined Precursors ................................................. 90  5.2.3  MoP Characterization ............................................................................. 99  5.2.4  Catalyst Activity and Product Distribution ........................................... 106  5.2.5  Reaction Kinetics .................................................................................. 109  5.3  Discussion ............................................................................................. 112  5.3.1  Catalyst Characterization ...................................................................... 112  5.3.2  Titration of MoP Active Sites ............................................................... 117  5.3.3  Reaction Sensitivity to Metal Phosphide Structure .............................. 118  5.3.4  Product Distribution .............................................................................. 121 x  5.4  Conclusions ........................................................................................... 122  Chapter 6 - The Preparation of High Surface Area Ni2P for HDO ...................... 124 6.1  Introduction ........................................................................................... 124  6.2  Results and Discussion ......................................................................... 125  6.2.1  Characterization of the Dried Precursors .............................................. 126  6.2.2  Characterization of Calcined Precursors ............................................... 130  6.2.3  Ni2P Characterization............................................................................ 132  6.2.4  Catalyst Activity and Product Distribution ........................................... 137  6.3  Discussion ............................................................................................. 145  6.3.1  Catalyst Characterization ...................................................................... 145  6.3.2  Ni2P Deactivation.................................................................................. 146  6.4  Conclusions ........................................................................................... 154  Chapter 7 - Relationship Between Pyrolysis Oil and 4-Methylphenol HDO ....... 156 7.1  Introduction ........................................................................................... 156  7.2  Results and Discussion ......................................................................... 156  7.2.1  Pyrolysis Oil Characterization .............................................................. 157  7.2.2  Pyrolysis Oil HDO ................................................................................ 157 xi  7.2.3  Comparison of Pyrolysis Oil HDO and 4-Methylphenol HDO ............ 163  7.2.4  Pressure and Temperature Effects ........................................................ 166  7.3  Conclusions ........................................................................................... 167  Chapter 8 - Conclusions and Recommendations .................................................... 168 8.1  Conclusions ........................................................................................... 168  8.2  Recommendations ................................................................................. 171  8.2.1  Preparation of High Surface Area Ni2P-CA ......................................... 171  8.2.2  Preparation of Bi-Metal and Sulfided Metal Phosphides ..................... 172  8.2.3  Ni2P Deactivation.................................................................................. 172  8.2.4  Catalyst Durability ................................................................................ 173  8.2.5  The HDO of Furanic, Lignin Derivatives, and O Mixtures .................. 173  8.2.6  The HDO of Pyrolysis Oil .................................................................... 173  8.2.7  Coke Recycle ........................................................................................ 174  8.2.8  DFT Modeling of Catalyst Surfaces ..................................................... 175  Bibliography .............................................................................................................. 176 Appendices ................................................................................................................. 189 Appendix A Sample Calculations ......................................................................... 190 xii  A.1  Economic Analysis for BINGO Process ................................................... 190  A.2  MoP Preparation ....................................................................................... 191  A.3  Catalyst Concentration .............................................................................. 191  A.4  BET ........................................................................................................... 192  A.5  XRD Lattice Parameters ........................................................................... 193  A.6  Kinetic Analysis ........................................................................................ 194  A.7  Metal Site Density..................................................................................... 196  Appendix B Error Analysis and Repeatability ..................................................... 198 B.1  Catalyst Characterization Repeatability .................................................... 198  B.2  Sampling Repeatability ............................................................................. 200  B.3  Reaction Repeatability .............................................................................. 202  B.4  Carbon Recovery ...................................................................................... 204  B.5  Pyrolysis Oil HDO .................................................................................... 205  Appendix C Gas Chromatography ....................................................................... 206 Appendix D Mass Transfer Effects ...................................................................... 209 Appendix E Kinetic Model Code ......................................................................... 215 E.1  Estimated Kinetic Parameter Error ........................................................... 215 xiii  E.2  Main Body Code ....................................................................................... 217  E.3  ModelMulti Code ...................................................................................... 220  E.4  Ordinary Differential Equation Code........................................................ 221  E.5  Calculation of Jacobian Matrix ................................................................. 221  E.6  Statistical Analysis of Kinetic Model ....................................................... 222  Appendix F Independent Reaction Data for Kinetic Mechanism ........................ 229 Appendix G Additional Experiments ................................................................... 232  xiv  List of Tables Table 2.1. Comparison between pyrolysis oil and petroleum properties .................................. 9 Table 2.2. Relative reactivities of oxygen compounds and groups for HDO over commercial CoMoS/Al2O3 ......................................................................................................... 11 Table 2.3. Activation energies of phenol derivatives estimated in a batch reactor over sulfided CoMo/Al2O3.............................................................................................. 13 Table 2.4. Model compound structures................................................................................... 22 Table 2.5. HDO of pyrolysed biomass before and after HDO treatment ............................... 29 Table 2.6. Overview of catalysts investigated for catalytic upgrading of pyrolysis oil .......... 30 Table 2.7. Reaction conditions and product analysis for Stage 1 (UBA). .............................. 31 Table 2.8. Reaction conditions and product analysis for Stage 2 (UBB). .............................. 32 Table 4.1. Main diffraction peaks and crystallite sizes of fresh and used Mo catalysts. ........ 59 Table 4.2. Surface area, CO uptake, and total acidity of Mo catalysts. .................................. 61 Table 4.3. H2 consumption of various catalysts using TPR technique. .................................. 62 Table 4.4. Oxygen consumption of used MoO3 for each TPO peak....................................... 64 Table 4.5. Mass increase for the re-oxidation of MoO3 via TGA........................................... 65 Table 4.6. XPS analysis of molybdenum catalysts ................................................................. 68 xv  Table 4.7. Kinetic parameters, Initial TOFs, and activation energies for the decomposition of 4-methylphenol over Mo catalysts at various reaction conditions. ........................ 71 Table 4.8. Activation energies for the DDO and HYD of 4-methylphenol. ........................... 78 Table 5.1. Chemical properties of the dried precursor (Dry-MoP-CA), the calcined precursors (Cal-MoP-CA-ttt K), and reduced/passivated MoP catalysts (MoP-CA-ttt K, MoPnoCA)...................................................................................................................... 84 Table 5.2. Mo (3d) and P (2p) XPS analysis of calcined precursors (Cal-MoP-CA-ttt K). ... 97 Table 5.3. Physical and chemical properties of MoP prepared with and without CA. ......... 100 Table 5.4. MoP lattice parameters and crystallite sizes determined by XRD....................... 104 Table 5.5. Conversion, product selectivity, and kinetic model parameter estimates for the hydrodeoxygenation of 4-methylphenol over MoP catalysts after 5 h reaction at 623 K and 4.4 MPa. .............................................................................................. 108 Table 5.6. MoP site density based on BET, XRD, and TEM size versus CO uptake........... 118 Table 6.1. Chemical properties of the dried precursor (Dry-Ni2P-CA), the calcined precursors (Cal-Ni2P-CA-ttt K), and reduced/passivated Ni2P catalysts (Ni2P-CA-ttt K, Ni2PnoCA).................................................................................................................... 126 Table 6.2. Physical and chemical properties of Ni2P prepared with and without CA. ......... 134  xvi  Table 6.3. Initial rate, initial TOF, kinetic parameter, deactivation parameter, and C deposition following the 5 h reaction for the hydrodeoxygenation of 4methylphenol over Ni2P catalysts at 623 K and 4.4 MPa. .................................... 140 Table 6.4. Physical and chemical properties of fresh and used MoP and Ni2P catalysts following 5 h hydrodeoxygenation reaction of 4-methylphenol at 623 K and 4.4 MPa. ...................................................................................................................... 148 Table 6.5. Conversion and product selectivities following 5 h hydrodeoxygenation of 4methylphenol over Ni2P-CA-773 K at 623 K and various pressures as well as the used catalyst C deposition and CO uptake............................................................ 149 Table 6.6. Initial rate and kinetic parameter for the hydrodeoxygenation of 4-methylphenol at 623 K and 4.4 MPa over oxidized catalysts. ......................................................... 151 Table 7.1. C, H, and O content of HDO product. ................................................................. 161 Table 7.2. Gas analysis of HDO product (mol.%). ............................................................... 162 Table 7.3. Mass and C balance for HDO reactions............................................................... 163 Table 7.4. The effect of temperature and pressure for the HDO of pyrolysis oil over MoPCA-823 K following the 1 h reaction. .................................................................. 166 Table B.1. Error associated with the surface area measurements of MoP-CA-773 K……. 198 Table B.2. Error associated with the CO uptake measurements of MoP-CA-823 K……... 199 Table B.3. Error associated with the CHN measurements of Cal-MoP-CA-773 K………. 199 xvii  Table B.4. Error associated with XRD analysis of MoP-CA-823 K……………………… 200 Table B.5. Error associated with GC injections – quantification of 4-methylphenol concentration……………………………………………………………………. 201 Table B.6. Kinetic and deactivation kinetic parameters for the HDO of 4-methylphenol over Ni2P-CA-773 K at 623 and 4.4 MPa………………………………………….... 203 Table B.7. Reactant and product concentrations for reaction of 4-methylphenol over MoPCA-823 K at 623 K (4.4 MPa), as well as carbon recovery……………………. 204 Table B.8. Error observed for the HDO of pyrolysis oil over MoS2 at 523 K……………. 205 Table B.9. Error in CH measurements of solid product over MoS2 for the HDO of pyrolysis oil at 523 K…........................................................................................................ 205 Table C.1. GC/FID temperature profile…………………………………………………… 206 Table C.2. GC/MS operational parameters………………………………………………... 208 Table D.1. Properties of decalin. .......................................................................................... 211 Table D.2. Reactor geometric properties. ............................................................................. 211 Table D.3. Mass transfer coefficient for stirred batch reactor. ............................................. 211 Table D.4. MoS2 catalyst properties. .................................................................................... 212 Table D.5. MoS2 catalyst Thiele modulus and effectiveness factors. ................................... 214  xviii  Table E.1. Experimental and model 4-methylphenol concentration data for the HDO reaction over MoP-823 K at 623 K and 4.4 MPa………………………..………………. 224 Table E.2. ANOVA analysis of the 4-methylphenol concentration data for the HDO reaction over MoP-823 K at 623 K and 4.4 MPa……………………………..…………. 225 Table E.3. Experimental and model toluene concentration data for the HDO reaction over MoP-823 K at 623 K and 4.4 MPa……………………………………………... 226 Table E.4. ANOVA analysis of the toluene concentration data for the HDO reaction over MoP-823 K at 623 K and 4.4 MPa……………………………………………... 226 Table E.5. Experimental and model methylcyclohexane concentration data for the HDO reaction over MoP-823 K at 623 K and 4.4……………………………………. 227 Table E.6. ANOVA analysis of methylcyclohexane concentration data for the HDO reaction over MoP-823 K at 623 K and 4.4……………………………………..……….. 228 Table F.1. Toluene conversion over MoS2 at 623 K and 4.4 MPa H2…………………….. 229 Table F.2. Methylcyclohexane conversion over MoS2 at 623 K and 4.4 MPa H2………... 230 Table F.3. 4-Methylcyclohexene conversion over MoS2 at 623 K and 4.4 MPa H2……… 230 Table F.4. 1-Methylcyclohexene conversion over MoS2 at 623 K and 4.4 MPa H2……… 231 Table G.1. Product selectivities over Mo and 5% Pd/SiO2 after 20% reactant conversion at 598 K (4.1 MPa)………………………………………………………………... 233  xix  List of Figures Figure 2.1. Hydrodeoxygenation path of phenol .................................................................... 12 Figure 2.2. Proposed catalytic active site model for NiMo/Al2O3 catalyst ............................ 17 Figure 2.3. Examples of reactions associated with catalytic pyrolysis oil upgrading ............ 26 Figure 3.1. 300 mL batch reactor used for HDO studies. ....................................................... 48 Figure 4.1. X-ray diffractogram of fresh and used MoP, and fresh and used MoS2............... 56 Figure 4.2. X-ray diffractogram of fresh and used MoO2, and fresh and used MoO3. MoO2 (■), Mo4O11 (□), and MoO3 (*)............................................................................... 58 Figure 4.3. TPO of fresh MoO2 and used MoO3..................................................................... 63 Figure 4.4. TGA re-oxidation of used MoO3. ......................................................................... 65 Figure 4.5. Mo (3d) XPS spectra for: (a) – fresh MoS2; (b) – fresh MoO2; (c) – fresh MoO3; (d) – used MoO3. ..................................................................................................... 67 Figure 4.6. XPS spectra of O 1s energy level for: (a) – fresh MoO2; (b) – fresh MoO3; (c) – used MoO3. ............................................................................................................. 69 Figure 4.7. 4-Methylphenol conversion over Mo catalysts at various reaction conditions. (■) MoO3, (◊) MoP, (▲) MoO2, (○) MoS2. Solid lines represent first order kinetic fit, 𝑥 = (1 − 𝑒 −𝑘𝐶cat 𝑡 ). (a) – 598 K, 4.1 MPa; (b) – 623 K, 4.4 MPa; (c) – 648 K, 4.8 MPa. ........................................................................................................................ 70  xx  Figure 4.8. 4-Methylphenol product selectivity over Mo catalysts at various reaction conditions. Toluene selectivity (■), hydrogenated product selectivity (◊), isomerization product selectivity (▲). (a) – 598 K, 4.1 MPa; (b) – 623 K, 4.4 MPa; (c) – 648 K, 4.8 MPa............................................................................................... 74 Figure 5.1. DRIFTS spectra of CA and the dried precursors Dry-MoP-CA and Dry-MoP-noCA. ..................................................................................................... 86 Figure  5.2.  Structure  of  (a)  -  citric  acid;  (b)  -  [Mo4(C6H5O7)O11]4-;  (c)  -  [Mo4(C5H5O5CN)2O11]4-; (d) - Mo(C4H5O3CN)O4; (e) - Mo(C4H5O3CN)O2•HPO4. Structures were inferred from spectroscopic data and literature references and calculated by DFT. .................................................................................................. 88 Figure 5.3.  13  C CPMAS spectra of CA, the dried precursor Dry-MoP-CA, and the calcined  precursor Cal-MoP-CA-773 K. (*) indicates spinning side bands. ....................... 90 Figure 5.4. DRIFTS spectra of the calcined precursors (Cal-MoP-CA-ttt K) and the calcined MoP-noCA precursor. ............................................................................................. 92 Figure 5.5. XRD of the calcined MoP-CA precursors (Cal-MoP-CA-ttt K) and the calcined MoP-noCA precursor. MoOPO4 (○); Mo2P2O11 (■)............................................... 93 Figure 5.6. XPS narrow scan spectra of the calcined MoP-CA precursors (Cal-MoP-CA-ttt K) and the calcined MoP-noCA precursor. (a) - Mo (3d); (b) - P (2p3/2); (c) - Mo (3d) fit; (d) - P (2p3/2) fit. ........................................................................................ 95  xxi  Figure 5.7. TPR-MS profiles of gas effluent from the calcined MoP precursors reduced in 9.5% H2 in Ar to 1173 K at 10 K/min. (a) - Cal-MoP-CA-773 K; (b) - Cal-MoPCA-973 K. ............................................................................................................... 98 Figure 5.8. SEM images of reduced/passivated MoP: (a) - MoP-noCA and (b) - MoP-CA-773 K............................................................................................................................ 101 Figure 5.9. XRD diffractograms of reduced/passivated MoP-CA and MoP-noCA catalysts. MoPO4(OH)3 (▼). ................................................................................................ 102 Figure 5.10. TEM images of reduced/passivated catalysts: (a) - MoP–noCA with passivation layer; (b) - MoP-CA-823 K with insert of lattice fringe d-spacing estimated at 2.1 Å for the (100) plane of MoP; (c) - MoP-CA-923 K, (d) nanoparticles encapsulated in an amorphous layer (MoP-CA-973 K); (e) - agglomerated particles (MoP-CA973 K) with insert of lattice fringe d-spacing of large particles estimated at 2.7 Å for the (110) plane of Mo oxide. ........................................................................... 105 Figure 5.11. Experimental and model concentration data versus reaction time: 4methylphenol (□), toluene (◊), methylcyclohexane (○), 1,3-dimethylcyclopentane (Δ), kinetic model fit (--). ..................................................................................... 107 Figure 5.12. HDO 1st-order kinetic parameter versus carbon content of reduced MoP-CA catalysts. ................................................................................................................ 110 Figure 5.13. Initial HDO rate of 4-methylphenol consumption, and toluene and methylcyclohexane production rates versus CO uptake of MoP-CA catalysts: HDO (■), DDO (□), HYD (○). ....................................................................................... 120 xxii  Figure 6.1. DRIFTS spectra of CA and the dried precursors Dry-Ni2P-CA and Dry-Ni2P-noCA. ................................................................................................... 128 Figure 6.2.  13  C CPMAS spectra of the dried precursor Dry-Ni2P-CA, and the calcined  precursor Cal-Ni2P-CA-773 K. (*) indicates spinning side bands. ..................... 129 Figure 6.3. DRIFTS spectra of the calcined precursors (Cal-Ni2P-CA-ttt K) and the calcined Ni2P-noCA precursor. ........................................................................................... 131 Figure 6.4. XRD of the calcined Ni2P-CA precursors (Cal-Ni2P-CA-ttt K) and the calcined Ni2P-noCA precursor. ........................................................................................... 132 Figure 6.5. XRD diffractograms of reduced and passivated Ni2P catalysts collected using a Co Kα X-ray source. Ni12P5 (*). ........................................................................... 135 Figure 6.6. TEM images of reduced and passivated Ni2P catalysts: (a) Ni2P-CA-773 K; (b) Ni2P-CA-823 K; (c) Ni2P-CA-973 K; (d) Ni2P-noCA; (e) lattice fringe d-spacing of Ni2P-CA-773 K at 1.7 and 2.5 Å for the (300) and (200) plane of Ni2P. ............. 136 Figure 6.7. Lognormal plot of the 4-methylphenol concentration versus time for HDO reaction at 623 K and 4.4 MPa over Ni2P catalysts. Ni2P-CA-773 K (▼), Ni2P-CA823 K (▲), Ni2P-CA-973 K (●), Ni2P-noCA (■), guideline shown for illustration (-). ........................................................................................................................... 138 Figure 6.8. Hydrodeoxygenation rate versus time for the reaction at 623 K and 4.4 MPa over Ni2P catalysts. Ni2P-CA-773 K (▼), Ni2P-CA-823 K (▲), Ni2P-CA-973 K (●), Ni2P-noCA (■), exponential decay law model fit (--). ......................................... 139  xxiii  Figure 6.9. Initial rate of hydrodeoxygenation versus CO uptake over Ni2P-CA catalysts...141 Figure 6.10. DDO product selectivity of toluene (a) and HYD product selectivity of methylcyclohexane (b) versus time for the reaction at 623 K and 4.4 MPa over Ni2P catalysts. Ni2P-CA-773 K (▼), Ni2P-CA-823 K (▲), Ni2P-CA-973 K (●), Ni2P-noCA (■), guideline shown for illustration (--) ........................................... 144 Figure 6.11. Lognormal of the 4-methylphenol concentration versus time for HDO reaction at 623 K and 4.4 MPa. Cal-MoP-CA-773 K (▼), passivated-MoP-CA-773 K (▲), passivated-Ni2P-CA-773 K (●), first order kinetic fit (--). ................................... 151 Figure 6.12. DDO (a) and HYD (b) product selectivity versus time for the reaction at 623 K and 4.4 MPa. Cal-MoP-CA-773 K (▼), passivated MoP-CA-773 K (▲), passivated Ni2P-CA-773 K (●), guideline shown for illustration (--). ................. 153 Figure 7.1. Pyrolysis oil product yield following HDO........................................................ 159 Figure 7.2. Pyrolysis oil product yield versus 4-methylphenol HYD selectivity (50% conversion). Liquid (●), coke (□), guideline shown for illustration (--)............... 164 Figure 7.3. Pyrolysis oil product yield versus 4-methylphenol DDO selectivity (50% conversion). Liquid (●), coke (□), guideline shown for illustration (--)............... 164 Figure 7.4. Pyrolysis oil product yield versus 4-methylphenol ISOM selectivity (50% conversion). Liquid (●), coke (□), guideline shown for illustration (--)............... 165 Figure A.1. Cost analysis of BINGO process ....................................................................... 190  xxiv  Figure B.1. HDO rate versus time for the 4-methylphenol decomposition over Ni2P-CA-773 K at 623 K and 4.4 MPa (3 trials)………………….…………………………… 202 Figure B.2. DDO and HYD product selectivities versus time for reaction at 623 K (4.4 MPa) over Ni2P-CA-773 K (3 trials)……………….……………………….………… 203  xxv  Nomenclature ao  =  XRD lattice parameter for hexagonal MoP (nm)  A  =  pre-exponential factor  A*  =  matrix for the estimated kinetic parameter/rate constant error  Ao  =  GC/FID area  Β  =  XRD peak width (radian)  co  =  XRD lattice parameter for hexagonal MoP (nm)  C  =  BET constant  Ca  =  concentration of 4-methylphenol (mol/mL)  Cao  =  initial concentration of 4-methylphenol (mol/mL)  Ccat  =  concentration of catalyst in reactor at ambient conditions (g/mL)  COV(k*)  =  covariance matrix  d  =  reactor diameter (m)  DAB  =  diffusivity of A in B (m2/s)  dc  =  crystallite size (nm)  Deff  =  effective diffusivity (m2/s)  dfBG  =  ANOVA degrees of freedom between groups  dfWG  =  ANOVA degrees of freedom within groups  dfT  =  ANOVA total degrees of freedom  dhkl  =  interplanar spacing for the (hkl) plane (nm)  dp(BET)  =  particle size determined by BET (nm)  dST  =  reactor stirrer shaft diameter (m)  dTEM  =  TEM particle size (nm) xxvi  Ea  =  activation energy (kJ/mol)  F  =  F-test distribution  FANOVA  =  ANOVA F value  G  =  Jacobian matrix  hL  =  liquid height in reactor (m)  hR  =  reactor height (m)  k  =  HDO kinetic parameter/rate constant (mL.min-1.g-1)  k4MP  =  HDO pseudo first order kinetic parameter/rate constant (s-1)  kd  =  deactivation kinetic parameter/rate constant (mL.min-1.g-1)  K  =  XRD shape factor, 0.9  kLa  =  volumetric liquid-side mass transfer coefficient (s-1)  m  =  number of data points  MSBG  =  ANOVA mean square value between groups  MSWG  =  ANOVA mean square value within group  n  =  number of dependent variables  N  =  reactor stirrer speed (s-1)  NA  =  Avogadro constant, 6.022 E+22 atoms/mol  P  =  number of parameters  P0  =  BET saturation vapor pressure (mmHg)  Q  =  weighting matrix  R  =  Universal gas constant, 8.314 J/(mol.K)  Re  =  Reynold’s number (dimensionless)  RF  =  GC response factor  xxvii  Rpar  =  particle radius (nm)  SBET  =  BET surface area (m2/g)  Sc  =  Schmidt number (dimensionless)  Sh  =  Sherwood number (dimensionless)  S(k*)  =  objective function evaluated at k values  SSBW  =  ANOVA sum of squares between groups  SSWG  =  ANOVA sum of squares within group  SST  =  ANOVA total sum of squares  t  =  HDO reaction time (min)  tα  =  tα-test distribution  T  =  temperature (K)  v  =  degrees of freedom for tα-test (m-1)  V  =  total BET volume adsorbed (cm3)  VG  =  gas volume in reactor (mL)  VL  =  liquid volume in reactor (mL)  Vm  =  BET volume adsorbed at monolayer coverage (cm3)  VT  =  total volume of reactor (mL)  We  =  Weber number (dimensionless)  x  =  conversion (%)  xi  =  response obtained from proposed model  Xavg  =  average value of sample  Xi  =  experimental sample value  z  =  total number of data sets  xxviii  γ  =  surface tension (N/m)  εp  =  particle porosity (dimensionless)  η  =  internal effectiveness factor (dimensionless)  θ  =  XRD angle of diffraction (radian)  θhkl  =  Bragg angle for (hkl) plane (degree)  θT  =  tortuosity factor (dimensionless)  λ  =  is the XRD wavelength of radiation (nm)  μ  =  viscosity (Pa.s)  µv  =  population mean for tα-test  π  =  pi bond  ρ  =  density (kg/m3 or g/cm3)  σ  =  sigma bond  σε2  =  variance for covariance matrix  σki  =  standard error for covariance matrix  σc  =  constriction factor (dimensionless)  σv  =  variance  Φ1  =  Thiele modulus (dimensionless)  Ω  =  overall effectiveness factor (dimensionless)  xxix  Acronyms 4MP  =  4-methylphenol  AFWA  =  aqueous fraction water addition  AHM  =  ammonium heptamolybdate tetrahydrate ((NH4)6Mo7O24.4H2O)  AHP  =  dimammonium hydrogen phosphate ((NH4)2HPO4)  ANOVA  =  analysis of variance  BE  =  binding energy  BET  =  Brunauer-Emmett-Teller  BDt  =  bone dry tonne  BINGO  =  biomass into gasoil  13  C-CPMAS =  13  C cross-polarization magic angle sample spinning  CA  =  citric acid (C6H8O7)  CMAS  =  Canadian Microanalytical Service Ltd.  CUS  =  coordinatively unsaturated sites  DDO  =  direct deoxygenation  DFT  =  Density Functional Theory  DRIFTS  =  diffuse reflectance infrared Fourier transform spectroscopy  EI  =  electron impact  FID  =  flame ionization detector  FTIR  =  Fourier transform infrared spectroscopy  FWHM  =  full width at half maximum  GC  =  gas chromatography  GHG  =  greenhouse gas xxx  HDN  =  hydrodenitrogenation  HDO  =  hydrodeoxygenation  HDS  =  hydrodesulfurization  HHV  =  high heating value (MJ/kg)  HYD  =  hydrogenation  ICP-AES  =  inductively coupled plasma atomic emission spectroscopy  ISOM  =  isomerization  LHSV  =  liquid hourly space velocity (h-1) (volume feed flow/volume of catalyst)  LUMO  =  lowest unoccupied molecular orbital  KUSY  =  potassium ion exchanged zeolite  MS  =  mass spectrometry  NMR  =  nuclear magnetic resonance  n-PA  =  n-propyl amine  NSERC  =  Natural Sciences and Engineering Council of Canada  OFWA  =  organic fraction water addition  PDF  =  powder diffraction file  RDS  =  rate determining step  SD  =  standard deviation  SEM  =  scanning electron microscopy  TCD  =  thermal conductivity detector  TEM  =  transmission electron microscopy  TGA  =  thermal gravimetric analysis  xxxi  TOF  =  turn over frequency (s-1)  TON  =  turn over number  TPD  =  tonne per day  TPO  =  temperature programmed oxidation  TPR  =  temperature programmed reduction  UBA  =  liquid product name following Stage 1 of Dynamotive BINGO process  UBB  =  liquid product name following Stage 2 of Dynamotive BINGO process  WHSV  =  weight hourly space velocity (h-1) (mass feed flow/mass of catalyst)  XPS  =  x-ray photoelectron spectroscopy  XRD  =  x-ray diffraction  xxxii  Acknowledgements There are many individuals that I am indebted to and who have provided so much support and insight to me while undertaking my doctoral degree. First and foremost, I would like to express my sincere thanks and appreciation to my doctoral supervisor, Professor Kevin J. Smith from the Department of Chemical and Biological Engineering at UBC. He has been a key influence in my successes during my Ph.D. appointment. His extensive knowledge, encouragement, guidance, and feedback have been invaluable to me. I have been able to grow and develop as a researcher due to Professor Smith’s outstanding supervision and mentorship. I have learned so much from him over the years and I strive to live by his example. I would next like to thank my committee members Dr. Sheldon Duff from the Department of Chemical and Biological Engineering and Dr. Michael Wolf from the Department of Chemistry at UBC for their advice and guidance while writing my dissertation. I would like to thank the past and present UBC catalysis group members and visiting scholars who have been helpful and supportive friends during my Ph.D. appointment: Farnaz Sotoodeh, Alex Imbault, Pooneh Ghasvareh, Shahin Goodarznia, Rahman Gholami, Ali Alzaid, Mina Alyani, Xuebin Liu, Liang Zhao, Benjamin Huber, Zaman Sharif, Hooman Rezaei, Shahrzad Jooya, Ross Kukard, Xu Zhao, Zhuangzhi Wu, Siying Bian, and Rui Wang. I would also like to thank Professor Mark Bussell for his valuable advice and guidance during his time at UBC. I would like to express my sincerest gratitude to Helsa Leong, Lori Tanaka, Amber Lee, and Anna Jamroz. I would also like to thank Richard Ryoo, Richard Zhang, and Ken xxxiii  Wong, as well as CHBE workshop personnel David Roberts, Doug Yuen, Graham Liebelt, Gordon Cheng, and Charles Cheung for their assistance and support. In addition I would like to acknowledge the Bioimaging Facility in the Department of Botany at UBC for TEM measurements done by Bradford Ross, as well as Mary Fletcher in the Department of Materials Engineering at UBC for helping me with SEM and XRD measurements. I would like to acknowledge the assistance and direction of Jenny Lai from the Department of Earth and Ocean Sciences at UBC for her help with XRD measurements. I would also like to thank Dr. Paul Xia from and Dr. Suzana Straus from the Department of Chemistry at UBC for NMR measurements, as well as Dr. Ken Wong from the Interfacial Analysis and Reactivity Laboratory at UBC for XPS measurements. In addition I would like to thank Canadian Microanalytical Service Ltd. (CMAS) for providing elemental analysis. I am also extremely thankful to Natural Sciences and Engineering Council of Canada (NSERC) for providing financial support over the years that allowed me to finish my Ph.D. without undertaking a financial burden. Finally I would like to express my gratitude to all of my friends near and far who always made themselves available to listen. I would also like to thank my mother for her encouragement and motivation to undertake my doctoral degree, as well as my sister and brother. In addition, I thank my family-in-law for all of their support over the years. Last but definitely not least, I would like to express my sincere thanks to my loving partner who has made this journey with me from the very beginning following high school. You have always been a constant support in my life and I thank you greatly for always being there for me.  xxxiv  Dedication  To my beloved and supportive partner  and  my mother  xxxv  Chapter 1 Introduction 1.1  Background  The increased awareness of global pollution and the depletion of natural oil resources have made the need for alternate energy sources a priority. Canada is heavily dependent on fossil fuels. In 2001, approximately 72% of the energy consumed in Canada was derived from nonrenewable fossil fuels. Fossil fuel combustion is responsible for two thirds of worldwide CO2 emissions [1]. Use of renewable biomass as a second generation energy source has the potential to reduce Canada’s reliance on fossil fuels and decrease CO2 emissions. Regrown plants use the CO2 generated upon combustion of biomass fuels and this neutralizes the C emissions [1]. As early as the 1940s, experimental data from Berl [2] supported the concept of oil production from biomass in water using an alkali catalyst. Since the oil crisis in the mid-1970s, considerable efforts have been made to convert woody biomass to liquid fuels. Pyrolysis oil derived from pyrolysed biomass has several environmental advantages over fossil fuels. No SOx emissions are generated upon combustion of pyrolysis oil, and pyrolysis oil fuels generate less than half the NOx emissions of diesel oil in gas turbines [3]. However, to upgrade woody biomass to a fuel comparable and compatible with petroleum, two processing steps are required. The first process involves upgrading the  1  solid biomass into a liquid pyrolysis oil product via fast pyrolysis [4, 5]. The products are ~65% liquid pyrolysis oil, ~15% solid char, and ~20% gases. The second step involves O removal from the pyrolysis oil. Pyrolysis oils derived from the fast pyrolysis of biomass have high O content (4550 wt.%) and moisture content (25 wt.%) that contributes to the low heating value, high viscosity, and minimal stability of these oils [5]. Therefore, in order to make these fuels comparable and compatible with petroleum fuels the O components in the oil must be removed. Water removal from pyrolysis oil has been attempted by (a) distillation, (b) dehydration, (c) solvent extraction, and (d) hydrodeoxygenation (HDO). However, pyrolysis oils are heat sensitive and subject to repolymerization, so they cannot be fractionated by distillation. Pyrolysis oils can be partially dehydrated with Na2SO4, but this results in the generation of SOx upon pyrolysis oil combustion. In addition water can be partially removed from pyrolysis oil via water or chloroform solvent extraction, but the produced oil phase still contains a significant amount of O. Hydrodeoxygenation involves the catalytic reaction of pyrolysis oil with H2 in the presence of a catalyst at moderate temperatures and pressures. The O is selectively removed from the organic molecule in the form of water. The H used in the HDO process is involved in the direct deoxygenation (DDO) of C-O bonds or hydrogenation (HYD) leading to aromatic ring saturation. Thus, it is important to understand the mechanism of the HDO reactions and the effect that catalyst structure has on product selectivity. The availability of high pressure H2 is an obstacle encountered in HDO, therefore the cost of hydrogen is a limiting factor with this technology. Depending on the final desired pyrolysis oil product, the HDO catalyst should possess a certain degree of HYD 2  activity. For example, diesel fuel and heating oil are composed of approximately 75-90 vol.% saturated hydrocarbons, and 10-25 vol.% aromatic hydrocarbons. Therefore, for diesel products, the catalyst should possess a high degree of hydrogenating activity. Gasoline consists of hydrocarbons with 5-12 carbons. Up to 20-50 vol.% of the compounds in gasoline are aromatic, while 50-80 vol.% are saturated aliphatic hydrocarbons. Therefore, catalysts for gasoline products should possess hydrogenolysis, HYD, and cracking abilities to crack the large hydrocarbon chains. Much research for the HDO of pyrolysis oil is available on supported bimetal sulfided catalysts such as NiMo/Al2O3 and CoMo/Al2O3 operated at high pressures. These catalysts have been shown to have excellent activity and selectivity in crude oil hydrodesulfurization (HDS) [6, 7]. However, unlike crude oils, pyrolysis oils do not contain S compounds to readily resulfide and reactivate the catalyst surface. Therefore, there is a need to investigate alternative non-sulfided catalysts that can deoxygenate pyrolysis oils at low pressures and in slurry phase. Unsupported catalysts used for the slurry phase HDO of pyrolysis oil are analogous to the use of unsupported, high surface area MoS2 catalysts for the slurry phase hydroconversion of residue oils [8]. Indeed, the use of an unsupported Fe catalyst for slurry phase pyrolysis oil hydroconversion has been described recently [9]. Slurry phase reactions are beneficial because the kinetics of the reaction can be determined from a single experiment that measures the reactant and product concentrations as a function of time, as opposed to data from a fixed bed reactor which must be operated at different residence times to obtain the data required for kinetic analysis. In addition, slurry phase reactions are real world systems therefore mass transfer  3  and deactivation can be investigated. Consequently, the reactions in this study have been done using a slurry phase reactor.  1.2  Objectives of the Thesis  The objectives of this study include identifying and improving the properties of unsupported, non-sulfided catalysts for HDO in slurry phase, with a particular focus on metal phosphides. The relationship between the structural properties of the catalysts and the conversion, product selectivity, and HDO mechanism will be determined. A refractory model compound, 4-methylphenol, was used to analyze the kinetics of the HDO reaction and to understand the O removal from phenols. Finally, the relationship between the HDO of the model compound versus the HDO of fast pyrolysis oil was examined. Results from the experiments were used to assess whether model compounds are useful to screen catalysts for HDO.  1.3  Work Plan  To address the objectives the following research tasks were undertaken:   The HDO of a model compound, 4-methylphenol, was used to assess activity and DDO/HYD selectivity of unsupported low surface area MoP, MoS2, and Mo oxide catalysts.    The use of citric acid (CA) in the preparation of high surface area MoP was investigated. The effect of preparation conditions on the activity of the catalyst was determined.  4    The HDO of high surface area Ni2P for the HDO of 4-methylphenol was evaluated.    The HDO of raw pyrolysis oil over the same catalysts used for the model compounds was examined. The relationship between the HDO of 4-methylphenol and raw fast pyrolysis oil was determined.  The study employed both experimental techniques and kinetic modeling to meet the stated objectives. The model compound 4-methylphenol was chosen to represent the phenolic content of pyrolysis oil as it is a refractory compound and has a reaction pathway that can measure both the DDO and HYD properties of the catalysts. The HDO of 4-methylphenol was carried out over commercial, unsupported, low surface area MoS2, MoO2, and MoO3 catalysts as well as experimentally prepared MoP and Ni2P (prepared with and without citric acid (CA) at various calcination temperatures). The catalysts were characterized using elemental analysis, BET, TPR, TPO, CO pulse chemisorption, acid titration, DRIFTS, SEM, TEM,  13  C NMR, XRD, and XPS. The  procedures for these characterizations are given in Section 3.2. A 300 mL autoclave batch reactor was used for the HDO reactions. The reaction pressure, temperature, and stirrer speed were monitored continuously throughout each experiment. The HDO reactions were carried out over several catalysts, at varying temperatures, and pressures using 4-methylphenol in an inert decalin solvent. The solvent was used to dissolve the solid 4-methylphenol. Liquid samples were withdrawn from the reactor at approximately 30 minute intervals for a total time of 5 h. The reaction start time was taken as zero following heat-up of the reactor from room temperature to the reaction temperature (approximately 30 minutes heat-up). The analysis of the model 5  compound liquid samples was done using a gas chromatograph (GC) and GC/mass spectrometer (MS). Three separate reaction temperatures were used to calculate the activation energy for 4-methylphenol HDO over each of the catalysts. Deactivation was observed over the Ni2P catalysts. In order to investigate this deactivation, deliberately oxidized phosphide catalysts and their phosphate precursors (samples prior to reduction) were tested for the HDO of 4-methylphenol to determine if the observed deactivation was a consequence of surface oxidation. The observed deactivation over the Ni2P catalysts was modeled using an exponential decay law. Kinetic modeling in Matlab allowed the determination of the kinetic parameters (rate constants) for the DDO and HYD reaction pathways for the HDO of 4-methylphenol, assuming first order kinetics. To determine the reaction mechanism, the product species were used as reactants to understand their reaction route. The HDO of fast pyrolysis oil was performed in a batch reactor using a decalin diluent over various catalysts. The total time for reaction was 1 h as the fast pyrolysis oil reacted much faster than the model compound. The reactant pyrolysis oil and the solid, liquid, and gas products were characterized by elemental analysis and H2O content via Karl Fischer Titration. The product gas was analyzed via GC-thermal conductivity detection (TCD). Finally, relationships between the HDO of the model compound 4methylphenol and the HDO of raw pyrolysis oil were determined. A relationship between the model compound product selectivity and the pyrolysis oil liquid yield was established.  6  1.4  Thesis Layout  The chapters that follow include a literature review of the HDO of model compounds over metal sulfide and phosphide catalysts, as well as a review of the HDO of pyrolysis oil (Chapter 2). In Chapter 3 the experimental methods employed in this study are presented. The first phase of the study, reported in Chapter 4, investigated the catalyst screening of unsupported MoOx, MoS2, and MoP for the HDO of 4-methylphenol. From this chapter, MoP was determined to be the best catalyst. Therefore, the use of a chelating agent to increase the MoP surface area and number of active sites was examined in Chapter 5. Similarly, in Chapter 6, Ni2P prepared using CA was investigated for the HDO of 4-methylphenol and compared to MoP. Finally, the HDO of fast pyrolysis oil was measured over MoP, Ni2P, MoOx, and MoS2 and the results are discussed in Chapter 7. Parallels between the HDO of 4-methylphenol and pyrolysis oil were established. The Appendices contain sample calculations (A), experimental repeatability (B), statistical analysis (B), information on the GC program used (C), mass transfer analysis (D), information on the kinetic model (E), independent reactions (F), and additional experiments (G).  7  Chapter 2 Literature Review 2.1  Lignocellulosic Fast Pyrolysis Oil  Fast pyrolysis oil can be made from a variety of forest and agricultural biomass wastes. Fast pyrolysis is a process carried out at atmospheric pressure by which small biomass particles are heated in the absence of oxygen, vaporized, and condensed into liquid. The residence time of the reaction is approximately one second [10-12]. Pyrolysis densifies the biomass feedstock to reduce storage space and transport costs. The process typically yields 65-70 wt.% liquid pyrolysis oil (on a dry feed basis), 15-20 wt.% char, and noncondensable gases [11-13]. However, the product yields depend on the process temperature, residence time, feedstock type, and feedstock pre-treatment. The high heating value (HHV) of pyrolysis oil is approximately 16–19 MJ/kg and it has an O content of 35-45 wt.%. In contrast, a heavy fuel oil typically has an O content of ∼1 wt.% and HHV of 40 MJ/kg [14]. The HDO of fast pyrolysis oil was focused on in this study fast pyrolysis produces the highest yield of liquid product compared to other pyrolysis systems. Pyrolysis oils derived from the pyrolysis of lignocellulosic biomass have high O contents due to the presence of compounds such as phenols, furans, carboxylic acids, ethers, aromatic alcohols, and free moisture [3, 5]. These oxygenated compounds are responsible for high viscosity, poor stability, non-volatility, immiscibility with fossil fuel, and low heating value of the pyrolysis oil [4]. Due to the reactivity of these oxygen 8  containing compounds, there are problems associated with the thermal stability and storage of the pyrolysis oil. During prolonged storage times, condensation reactions occur, which induce the formation of heavier compounds. This in turn causes a phase separation in the pyrolysis oil [5]. Therefore, in order for the pyrolysis to be both competitive and comparable to petroleum oil the oxygen in pyrolysis oil must be removed. The presence of O leads to the differences observed in the properties and behaviors of petroleum fuels compared to pyrolysis oils. The produced pyrolysis oil contains approximately 25 wt.% water and 50 wt.% O (on a wet basis) [10]. As shown in Table 2.1 the high O content of pyrolysis oil derived from a fast pyrolysis process yields an oil with a heating value of less than half that of fuel oil [15, 16]. Pyrolysis oil is found to be miscible with alcohols such as ethanol and methanol, but immiscible with hydrocarbons.  Table 2.1. Comparison between pyrolysis oil and petroleum properties [15, 16]. Reprinted and modified with permission from Elsevier, Copyright (2005) and Copyright (1994).  Parameter  Unit  Fast Pyrolysis Oil  Heating Oil #2  Heavy Fuel Oil  High Heating Value  MJ/kg  19.3  45.5  42.5  Density (288 K)  kg/L  1.21  0.865  0.986  C  wt.%  56.4  86.4  85.7  H  wt.%  6.2  12.7  10.5  N  wt.%  < 0.3  0.006  0.18  S  wt.%  < 0.01  0.2-0.7  < 2.8  O  wt.%  37.1  0.04  0.38  Pour Point  K  250  267  --  9  Within Canada in 2012, 28 million bone dry tonne (BDt) of biomass was available, partly due to the pine beetle destruction in British Columbia. This indicates the need to utilize these Canadian resources locally and take advantage of the biomass to reduce Canada’s dependence on foreign fuel as well as develop an industry for renewable fuel production. If 28 million BDt of Canadian biomass were converted to pyrolysis oil this would equate to approximately 18.2 million tonne of pyrolysis oil that could then be upgraded to produce 6.8 million tonne of O free oil with a HHV of 45 MJ/kg that is miscible with hydrocarbons. This could potentially displace 22% of Canada’s annual gasoline consumption. In particular, Canada has 310.1 million hectares of forest area. Annually, 0.9 million hectares are harvested. In British Columbia it is estimated that by 2018 the pine beetle will destroy 78% of all lodgepole pine stands which equates to approximately one billion m3 of timber [17]. Assuming an oven dried density of 400 kg/m3, this equates to 400 billion BDt of biomass that has potential to be upgraded to 96 billion tonne of O free liquid.  2.1.1  Phenolic Composition of Pyrolysis Oil  Pyrolysis oils derived from the pyrolysis of forest residues have a very complex structure. The exact composition of pyrolysis oil will vary depending on feedstock composition, feedstock location, and pyrolysis conditions such as temperature, heating rate, and residence time [5]. Thus, from a fundamental research standpoint, it is not practical to test the hydrodeoxygenation of pyrolysis oil for catalyst screening. This is because pyrolysis oil contains a large quantity of individual oxygenated compounds that would be difficult 10  to identify and quantify. Overall, pyrolysis oil contains approximately 3-10 wt.% phenolic compounds [5]. Phenols are also the most refractory compounds in pyrolysis oil and, thus, have been used as model compounds in many research based studies to represent pyrolysis oil feedstocks in catalyst assessments [2, 18-25]. Phenols consist of a hydroxyl group (-OH) attached to an aromatic hydrocarbon group. Phenols are very refractory compounds with high activation energies for the HDO reaction over commercial CoMoS/Al2O3, as shown in Table 2.2 and thus, are ideal model chemicals for HDO catalyst testing since they are the most difficult to deoxygenate.  Table 2.2. Relative reactivities of oxygen compounds and groups for HDO over commercial CoMoS/Al2O3 [26]. Reprinted and modified with permission from Elsevier, Copyright (1996).  Group  Ea (kJ/mol)  Ketone  50  Carboxylic Acid  109  Methoxyphenol  113  4-Methylphenol  142  2-Ethylphenol  151  Dibenzofuran  142  11  Structure  2.2  HDO of Phenolic Compounds  The HDO of phenolic compounds is believed to occur via two parallel paths as shown in Figure 2.1 for phenol. In the presence of H2, the first path involves the direct cleavage or hydrogenolysis of the C-OH σ bond to yield benzene and water (DDO). The second route involves the saturation of the aromatic ring via HYD, followed by HDO to yield cyclohexane and water.  Figure 2.1. Hydrodeoxygenation path of phenol [21]. Reprinted with permission from American Chemical Society, Copyright (2006).  As early as the 1970s, the effect of substitution on the overall HDO conversion of phenol was investigated by Rollman [27]. Furimsky et al. [24] found that alkyl substituted phenols (cresols) had a higher resistance to HYD over sulfided NiMo catalysts. Similarly, Massoth et al. [21] found that methyl substitution over sulfided Mo catalysts led to an increase in the selectivity of aromatic products and was independent of the methyl-group position. Odebunmi and Ollis [28] examined the activity of various substituted phenols (2,3, or 4-methylphenol) over sulfided CoMo/Al2O3 in a fixed-bed mode at 498-673 K and pressures between 3-12 MPa. The activation energies of the 12  various substituted phenols are shown in Table 2.3. As seen by these results, 4methylphenol had the highest activation energy. However, the reactivity of the phenols to HDO was found to decrease in the following order 3-methylphenol > 4-methylphenol > 2-methylphenol.  Table 2.3. Activation energies of phenol derivatives estimated in a batch reactor over sulfided CoMo/Al2O3 [28]. Reprinted and modified with permission from Elsevier, Copyright (1983).  Compound  Ea  Structure  (kJ/mol) 2-methylphenol  96  3-methylphenol  113  4-methylphenol  156  Besides the limitation of reaction rates by adsorption due to geometric effects (i.e. steric hindrance), the HYD of aromatic rings will be dependent on the electronic properties of the phenol that can be changed with methyl substitution [29]. Phenol adsorbs to the catalyst surface in regions of the phenol that have high electron density. This includes the heteroatomic C-O σ bond as well as the C=C π bonds, in the aromatic ring. Adsorption of phenol can occur in a flat manner involving C=C π bonds or perpendicular through the oxygen atom’s lone pair of electrons [21]. The electrostatic properties of phenols can be changed by ring substitution. Alkyl substituted phenols have a higher electron density on the aromatic ring than unsubstituted phenols. This will affect the ring interaction with the negatively charged ions from a catalyst such as MoS2. 13  Therefore, the activity of methylphenols depends on the catalyst surface, as well as the phenol structure since steric effects of alkyl substitution are important regardless of type of catalyst employed. This is evidenced by the fact that similar reactivity of phenols was observed over sulfided CoMo/Al2O3, NiCr, and Ni/SiO2 [28, 30].  2.3  Metal Sulfides for HDO  Much research on the removal of O from hydrocarbons by HDO is available on conventional supported metal sulfide catalysts, as well as supported noble metal catalysts, that are known to be active for sulfur removal from crude oils using high pressure H2 [15, 20, 21, 31-35]. HDO is closely related to HDS as both use H2 for the removal of a heteroatom forming H2O or H2S, respectively. However, noble metals are expensive and have high selectivity for hydrogenation reactions that consume H2. Although sulfided metal catalysts are very active for oxygen removal, S loss and oxidation of the active catalyst phase can occur during the hydrotreating of pyrolysis oil because of the oil’s high O content. Therefore, addition of a sulfiding agent such H2S or CS2 is required to maintain the catalyst sulfur content because pyrolysis oils are S free [19]. This is undesirable as S will poison the gas phase by producing H2S, which requires post-scrubbing. Hence, there is a need to move from traditional metal sulfide catalysts to alternative catalysts for HDO. Gevert et al. [23] performed the HDO of mono- and dimethyl substituted phenols in the presence of sulfided CoMo/Al2O3 in a batch reactor at 573 K and 5 MPa. The authors found that the reaction was first order and proceeded via two parallel reaction pathways, one leading to aromatic products and the other to naphthenic products. Overall, 14  toluene was found to be the main product with a selectivity of 80%, with 15% methylcyclohexane, and 5% methylcyclohexene selectivity from the HDO of 4methylphenol. Hydrogenolysis of the desired C-OH bond (DDO path) was predominant with a calculated kinetic parameter of 3.9 mL.min.-1gcat-1 and the HYD kinetic parameter was 0.8 mL.min.-1gcat-1. Similarly, Furimsky et al. [24] investigated the HDO of substituted phenols in hexadecane using oxidic and sulfided CoMo/Al2O3 in a continuous reactor. After a period of 3 hours at 623 K and 10 MPa over 50 g of catalyst, the final conversion of methylphenol was 95% over both the oxidic catalyst and the sulfided catalyst. The yield of parent products (O free cycloalkanes and aromatics) over the oxide catalyst was 97% compared to 82% over the sulfided catalyst. Of these parent products 24% were hydrogenated oxygen free products over the oxidic catalyst and 86% were hydrogenated parent products over the sulfided catalyst. Thus, the sulfided catalyst displayed a higher degree of HYD and thus, H2 consumption. A study by Yang et al. [20] investigated the hydrodeoxygenation of 4methylphenol over unsupported MoS2 catalysts. In this study, bulk-crystalline powder MoS2, exfoliated MoS2, and MoS2 derived from the in-situ decomposition of ammonium heptamolybdate tetrahydrate (AHM) were studied in a slurry reactor. From the collected data it was determined that the decomposition of 4-methylphenol followed first order kinetics and the activation energy, Ea, for 4-methylphenol decomposition was 134.7 kJ/mol over the AHM-derived catalyst. Over the AHM catalyst hydrogenolysis of the methyl bond (C-CH3) was observed to yield phenol at 39% selectivity. However the main product was toluene at 50% selectivity. The pseudo first order kinetic parameter was 15  determined to be 2.2 mL.min.-1gcat-1 over the AHM catalyst at 623 K and 2.8 MPa with 600 ppmw Mo. Massoth et al. [21] also studied the HDO of methyl substituted phenols at 573 K and 2.85 MPa H2 over sulfided CoMo/Al2O3. The products included benzene, cyclohexane, and water. From their analysis and product yields the authors suggested two reaction pathways: one leading to aromatic products (DDO) and one leading to partially or completely hydrogenated cyclohexanes (HYD). From the above summary it can be observed that metal sulfides, that are known to be active and selective for HDS are also active and selective for the HDO of model compounds.  2.3.1  Metal Sulfide Active Sites  To optimize HDO reactions it is important to understand the structure-activity relationships between the catalytic surface and the reacting species. If this relationship is understood, the catalyst structure and morphology can be tailored to increase the number of active sites. Under high temperatures and in the presence of H2, metals are reduced to lower oxidation states. Upon their reduction, coordinatively unsaturated sites (CUS), also known as Lewis acid sites, are created. In MoS2 catalysts the CUS are located at the edge planes and are believed to be the sites for catalytic reaction. These sites can either operate in DDO or in aromatic ring HYD [25, 36, 37]. In MoS2, the Mo edge of the catalyst has been proposed to be active for HYD, while S vacancies adsorb heteroatoms that lead to DDO in HDO. A study by Laurent and 16  Delmon [33] found that an increase in H2S partial pressure reduced the yield of aromatics but not the yield of HYD, which suggested separate sites for HYD and DDO. Conversely, Bunch and Ozkan [32] suggested that HYD sites are associated with metal CUS, whereas hydrogenolysis sites are believed to be Brønsted acid centers associated with the adsorption and dissociation of H2S; or similarly H2O since it is conceivable that OHgroups operate in a similar function to SH- groups. Bunch and Ozkan’s [32] proposed mechanism is given in Figure 2.2.  (a)  Mo HYD site, (b) Ni HYD site, (c) Mo Hydrogenolysis site  Figure 2.2. Proposed catalytic active site model for NiMo/Al2O3 catalyst [32]. Reprinted with permission from Elsevier, Copyright (2002).  Gevert et al. [23] and Furimsky et al. [24] also suggested separate sites for DDO and HYD over sulfided catalysts. However, a dual site that adsorbs both the aromatic ring and heteroatom of phenol at the same time is possible [25]. For example, two separate 17  studies by Moreau et al. [36, 38] proposed a dual site that allows adsorption of the π and σ bonds by adjacent electron accepting and donating sites. Massoth et al. [21] found that hydrogenolysis of the C-OH bond in phenol decreased as the concentration of H2S increased, which was a result of competitive adsorption of H2S to the catalysts’ S vacancies. From these results Massoth et al. [21] proposed that hydrogenolysis sites consist of CUS associated with Mo and adjacent HYD sites have S or SH groups. The H+ involved in DDO is primarily coming from a Brønsted acid site of the catalyst generated by the adsorption and dissociation of H2S, H2O, or H2. A clear understanding of HYD and DDO active sites is not well established since conflicting mechanisms are given throughout the literature for sulfided Mo-based catalysts. Although sulfided catalysts are active for HDO they lose S during the reaction. Therefore, sulfiding agents such as CS2 and H2S must be added to the feed stream to maintain sulfidation. This is undesirable and thus alternative non-sulfided catalysts should be investigated for HDO.  2.4  Metal Phosphides for HDO  Sulfided Mo catalysts that are known to be active and selective for C-S scission in HDS [7] have also demonstrated high activity and selectivity for C-O scission in HDO [20]. Similarly, other catalysts that are known to be active for HDS are candidates for HDO, such as metal phosphides. Recently, several authors have shown that metal phosphides are active and selective for HDS [39-41]. Studies have shown that MoP/SiO2 has 4 times the activity (on a mass basis) of MoS2/Al2O3 for the HDS of thiophene [42]. Stinner et al. [43] 18  showed that MoP has a 6 times higher turn over frequency (TOF) for the hydrodenitrogenation (HDN) of orthopropylaniline than MoS2/Al2O3, based on geometric estimates of surface site density. Several other studies have also reported that metal phosphides are more active and selective than sulfided metals for HDS [39, 44, 45] and HDN [46]. Therefore, the metal phosphide catalysts are also candidates for HDO [47]. In the past two years several authors have investigated the HDO of model compounds over metal phosphides. These studies were reported after the present study had been initiated and they followed the initial reports on metal phosphides for HDO made by the author of this dissertation (details contained in Chapters 4-6) [48-50]. Zhao et al. [51] investigated the HDO of guaiacol (shown in Table 2.4) at 573 K and 1.5 MPa in a fixed bed reactor over supported MoP/SiO2, Ni2P/SiO2, Co2P/SiO2, WP/SiO2, Fe2P/SiO2, and sulfided CoMo/Al2O3 for comparison. The sulfided CoMo/Al2O3 catalyst displayed low HDO activity and exhibited deactivation due to S loss. Overall MoP/SiO2 displayed the highest TOF. However, Ni2P/SiO2 displayed the highest selectivity to the fully deoxygenated product, benzene [47]. Li et al. [52, 53] investigated the deoxygenation of ethanol over Ni2P/SiO2 and the HDO of anisole (shown in Table 2.4) over monometallic Ni2P/SiO2, MoP/SiO2, and bimetallic NixMoyP/SiO2 catalysts with varying Ni and Mo contents. The HDO of anisole was measured in a continuous flow reactor at 573 K and 1.5 MPa, and 4 wt.% anisole in n-octane was fed to the reactor. Ni2P/SiO2 displayed ten times the TOF of MoP/SiO2. The MoP/SiO2 catalyst yielded a high selectivity towards phenol, whereas, Ni2P/SiO2 yielded a high selectivity toward the fully deoxygenated product cyclohexane, indicating successful HDO over this catalyst. Both catalysts displayed deactivation following a 12 h 19  reaction time. Ni2P/SiO2 deactivated by 29% and MoP/SiO2 lost 58% of its HDO activity. The authors suggested that the deactivation was due to oxidation of the catalyst surface that was more prominent over MoP/SiO2 compared to Ni2P/SiO2. Cho et al. [54] investigated the HDO of 2-methyltetrahydrofuran (shown in Table 2.4) in n-heptane over bimetallic NiFe phosphide with varying Ni:Fe ratios, as well as phase pure Ni2P and FeP, all supported on K ion exchanged y-zeolite (KUSY). The reaction was carried out in a fixed bed reactor at 0.5 MPa and temperatures ranging from 523 to 598 K. Ni2P/KUSY showed the highest conversion of 2-methyltetrahydrofuran whereas the NiFe phosphide activity decreased with Fe content.  Low 2-  methyltetrahydrofuran conversions (3%) produced mostly n-pentane and n-butane for Ni2P/KUSY, whereas the Fe containing catalysts produced 1-pentanol. At high conversions (70%) all the catalysts produced n-pentane and n-butane. Bui et al. [55] also studied the HDO of 2-methyltetrahydrofuran in heptane at 573 K and 0.1 MPa over a series of metal phosphides and found the activity to decrease in the following order Ni2P > WP > MoP > CoP > FeP > Pd/Al2O3. The activity was evaluated in a packed-bed reactor on the basis of an equal number of CO chemisorption sites (30 µmol). The product yield over Ni2P and CoP were pentane and butane, whereas MoP and WP produced mostly pentenes and pentadienes. Pd/Al2O3 and FeP/SiO2 produced mostly pentenes and a mixture of C4s. Bowker et al. [56] investigated the HDO of furan (shown in Table 2.4) over Ru phosphide catalysts at 673 K. The HDO of furan (normalized on a mass basis) followed the trend of decreasing activity of Ru2P/SiO2 > RuP/SiO2 ≥ Ru/SiO2 > > CoMo/Al2O3. Ru2P/SiO2 was three times more active than Ru/SiO2. The Ru phosphide catalysts 20  displayed a strong selectivity towards C4 products whereas the Ru catalyst favored C3 products. Olarte [57] studied the HDO of syringaldehyde (shown in Table 2.4). A 50 mL reactor pressurized to 6.9 MPa H2 and heated to 573 K was used. Nickel based catalysts (nickel phosphide, nickel oxide and nickel phosphate) as well as precious metals (Pt and Pd) with and without Al2O3 were tested as HDO catalysts. Of the three O-containing functional groups of syringaldehyde, the aldehydic group was found to be the most susceptible to O removal. The average TOF for syringaldehyde conversion by Ni12P5/Al2O3 was higher compared to the supported noble metal catalysts. Experiments catalyzed by NiO showed almost complete conversion of syringaldehyde (comparable to the supported catalyst), but the product varied. Ni12P5 eventually gave complete conversion after 240 minutes while the unreduced bulk Ni3(PO4)2 only reached ~80% conversion even after 240 minutes.  21  Table 2.4. Model compound structures.  Compound  Structure  Guaiacol Anisole  2-methyltetrahydrofuran Furan Syringaldehyde  The above summary shows that metal phosphides are a promising class of catalysts for HDO reactions with strong hydrogenating abilities. In the present study, Mo and Ni based catalysts were the focus due to their strong HDS and HDN activity and their cheaper price compared to noble metals.  2.4.1  Preparation of High Surface Area Metal Phosphides Using Citric Acid  Metal oxide supports such as Al2O3, TiO2, ZnO, and SiO2 are prone to coking and fouling by polar compounds [5, 47, 58, 59]. In addition, support interaction effects are common amongst supported metal phosphide catalysts. Phosphate interacts strongly with Al2O3 to form AlPO4 following calcination. In turn, excess P must be added to the preparation to maintain an adequate metal:P ratio. Additionally, the reduction of AlPO4 requires high temperature. To limit support interaction effects observed during reduction of the  22  supported metal phosphide precursor [47], several authors have investigated the use of organic chelating agents that act as structural promoters for the preparation of unsupported, metal phosphides. Wang and Smith [60] have shown that both the surface area and CO uptake of MoP can be increased from 5 m2/g and < 1 μmol/g to 139 m2/g and 42.4 μmol/g, respectively, by preparing the catalysts in the presence of CA. This was done by adding CA to aqueous solutions of ammonium heptamolybdate (AHM; (NH4)6Mo7O24.4H2O) and diammonium hydrogen phosphate (AHP; (NH4)2HPO4) with a 1:1 ratio of Mo:P and a CA:Mo ratio of 2:1, followed by drying (397 K) and calcination (773 K) in stagnant air, and reduction in H2 (923 K). Similarly, the increase in MoP surface area corresponded to an increase in 4,6-dimethyldibenzothiophene HDS conversion from 54.5% to 74.7% [60]. Cheng et al. [61] also observed hydrazine decomposition to increase from 55% to 85% for MoP prepared without and with CA, corresponding to MoP surface areas of 8.2 and 122 m2/g, respectively. Wang and Smith [60] also investigated the preparation of unsupported Ni2P using CA. CA was added to an aqueous solution of nickel nitrate (Ni(NO3)2.6H2O) and AHP with a 1:2 ratio of Ni:P and CA:Ni ratio of 2:1, followed by drying (397 K) and calcination (773 K) in stagnant air, and reduction in H2 (923 K). Excess P was added for preparation because phosphine (PH3) was produced during the reduction of the Ni phosphate precursor that led to P losses in the catalyst. The surface area of Ni2P was increased from 2 to 216 m2/g and the CO uptake from <1 to 30.1 µmol/g by adding CA. Furthermore, these authors showed that at 483 K and 3 MPa H2, the conversion of 4,6-dimethyldibenzothiophene increased from 74.7% for Ni2P prepared without CA to 87.7% for Ni2P prepared with CA. The TOF for HDS of 4,6-dimethyldibenzothiophene 23  over the Ni2P prepared with CA was twice that of the MoP. In addition, unsupported high surface area Ni2P (130 m2/g) was prepared by Yang et al. [40] by adding a polymer surfactant (Triton X-114) and ethylene glycol to the aqueous solutions of Ni(NO3)2 and (NH4)2HPO4 followed by drying and reduction in H2 at 500oC. Ibeh et al. [62] also performed the preparation of WP in the presence of CA. The calcination of WP-CA with W:CA = 1:4 was performed in a flow of air. However, the addition of CA to WP increased the surface area from 4.1 to only 18 m2/g. This was due to complete combustion of the CA in the flowing air. Therefore, C must be present in the catalyst as a citrate to promote increases in surface area. Consequently, the C content and citrate formation of catalysts prepared using CA should be investigated.  2.4.2  Metal Phosphide Active Sites  It has been suggested that both DDO and HYD reactions occur on the same site that chemisorbs CO, such as CUS [49, 63, 64]. Senol et al. [34] have suggested that for MoS2 catalysts, hydrogenation reactions occur on CUS that are more saturated compared to sites for hydrogenolysis. The CUS for hydrogenolysis are more electrophilic (i.e., more positively charged) than those for hydrogenation [32, 33, 65, 66]. This suggests that HYD occurs on both highly reduced sites and those having a lower coordination number. MoP displays bifunctional properties with strong hydrogenating capabilities similar to Pd and Pt on acidic supports [22, 48]. The source of acid sites (PO-H) over the MoP catalyst is a consequence of the incomplete reduction of phosphate species and CUS that dissociate H2 and H2O [41]. Both of these sites can result in Brønsted acidity that carry out protonation for hydrogenation reactions [21]. The latter Brønsted acid sites have also 24  been shown to be active centers for both hydrogenolysis and/or aromatic ring saturation [21, 24, 32, 36].  2.5  Pyrolysis Oil HDO  Pyrolysis oil can be upgraded to produce a product similar to petroleum oil by HDO of the pyrolysis oil. The H2 consumption for HDO upgrading has been found to be on the scale of 27–31 mol/L of pyrolysis oil processed [67]. Once upgraded, hydrodeoxygenated pyrolysis oil has an energy content of 39–43 MJ/kg [14]. The hydrotreatment of pyrolysis oil is very complex. Several reactions occur such as  hydrogenation,  hydrogenolysis,  decarboxylation,  decarbonylation,  cracking,  polymerization, and coke formation [59, 68] as shown in Figure 2.3. When pyrolysis oil is heated to temperatures above 333-373 K it becomes hard because of thermally-induced polymerization [5, 58, 69], especially over alumina based catalysts [5, 58, 59]. This is due to condensation reactions of unsaturated double bonds such as olefins, aldehydes, and ketones [26]. In particular phenolic precursors are also active coke precursors [70, 71]. Therefore strong hydrogenating catalysts are desirable for the HDO of pyrolysis oils to prevent the condensation of these unsaturated double bonds [71].  25  Figure 2.3. Examples of reactions associated with catalytic pyrolysis oil upgrading [59, 68, 72]. Reprinted with permission from Elsevier, Copyright (2011).  With pyrolysis oil a two stage HDO process is used to promote hydrogenation and prevent charring and coke formation. The first stage involves stabilization of the pyrolysis oils at 573 K to convert polymerizing O compounds that can form free radicals or condensation reactions, and saturate double bonds that can start chain propagation. The second stage involves “deep” HDO at temperatures above 623 K [71]. Following HDO of pyrolysis oil two liquid product phases are formed: an organic phase and an aqueous phase. Sometimes a second organic phase with a density lighter than water can be produced. H2 is supplied for the HDO reactions at high pressure, 7.5-30 MPa [68]. This ensures high solubility of H2 in the oil. Typical reaction times of 3-4 h in batch reactors and of 0.1-1.5 h-1 in continuous flow reactors (liquid hourly space velocity (LHSV)) and temperatures of 523-723 K, are used [68].  26  Based on the activation energies of species reported in Table 2.2 it can concluded that ketones can be deoxygenated at lower temperatures compared to phenols and furans. The H2 consumption will depend on the chemical composition of the pyrolysis oil, i.e. phenols will require more H2 for saturation of the aromatic ring. Elliott [4] and Elliott et al. [73] have reported H2 consumption and O removal from the first stage of HDO to be 0.15 m3/m3 and 7%, respectively, using various catalysts and a pressure of 13.8 MPa. Venderbosch et al. [74] removed 5% O as water with a H2 consumption of 0.078 m3/m3 at 448 K and 20.7 MPa. A study by Wildschut et al. [59] investigated the HDO of pyrolysis oil using a variety of noble metal catalysts: Ru/C, Ru/TiO2, Ru/Al2O3, Pt/C, and Pd/C and sulfided NiMo/Al2O3 and CoMo/Al2O3. The hydroprocessing reaction was done at temperatures between 523-623 K and pressures of 10 and 20 MPa. The authors loaded the batch reactor with 25 g of oil and 1.25 g catalyst and used a heating rate of 16 K/min with a continuous gas phase of H2 for a reaction time of 4 h. Overall, the Ru/C catalyst produced the highest yield of liquid (60 wt.%) and extent of deoxygenation (90 wt.%). The authors stated that the reaction at 523 K and 10 MPa represented “mild” HDO while the reaction at 623 K and 20 MPa represented “deep” HDO. Under “mild” HDO conditions a single oil product was obtained with 18-27 wt.% O and 21-55 wt.% liquid yield (dry basis). Deoxygenation was found to be higher over the noble metal catalysts compared to the conventional sulfided catalysts. Pd/C had the potential to produce higher oil yields than Ru/C but the oil had higher O content and the catalyst consumed more H2 due to the formation of CH4. Upon analysis it was shown that the oil contained phenols, aromatics, and hydrocarbons. This indicates that O in the form of phenol is difficult to remove. 27  De Mercader et al. [75] also studied the HDO of pyrolysis oil. First the authors pretreated the oil by mixing the oil with water at a 2:1 weight ratio to obtain a heavy organic rich oil fraction water addition (OFWA) and an aqueous fraction water addition (AFWA). They treated these fractions and the whole oil at temperatures of 493, 543, and 583 K and 19 MPa using 5 wt.% Ru/C and a residence time of 4 h. 250 g of oil was loaded to the reactor, then 5 wt.% catalyst was added (on a wet basis). The heat-up rate was 7-9 K/min. Hydrogenation of olefins, aldehydes and ketones at 473 K was done for stabilization. The recovery of C in the oil phase increased when the temperature was increased from 493 to 583 K (16.3 to 38.5 wt.%) from AFWA. However, H2 consumption from this phase was high due to CH4 formation. The H2 consumption of OFWA was lower than the whole oil HDO, however the oil had lower H/C ratio compared to AFWA. Overall, it does not appear that pretreating the oil with water adds significant benefit for HDO and the energy required for purification of the aqueous phase would make the process uneconomical. Zhang et al. [15] studied the HDO of pyrolysis oil over a supported sulfided catalyst. The authors used tetralin and tar oil as a diluent. Tetralin displayed the highest conversion and liquid yield. The sulfided catalyst used was composed of 24 wt.% MoO3, 4 wt.% CoO, 2.8 wt.% P, and 69.3 wt.% Al2O3 with a surface area of 160 m2/g. The HDO treatment was performed in a 500 mL autoclave at 2 MPa H2 and 663 K with a heat-up rate of 10 K/min. After 60 minutes, the HDO conversion of the pyrolysis oil reached 90%. The activation energy of the pyrolysis oil over the catalyst was determined to be 91.4 kJ/mol. The oil composition was determined by Fourier transform infrared spectroscopy (FTIR). As shown from the results in Table 2.5, after HDO treatment the 28  density of the oil decreased to a value similar to heavy oil and the heating value doubled. The solubility of the oil with aromatic toluene also increased to 100% indicating that the upgraded oil was now miscible with petroleum oil. Thus, HDO is a desirable process to convert highly acidic pyrolysis oil into a compatible high energy oil that is comparable and competitive with petroleum oil.  Table 2.5. HDO of pyrolysed biomass before and after HDO treatment [15]. Reprinted with permission from Elsevier, Copyright (2005).  Properties  Raw Pyrolysis oil  Upgraded Pyrolysis oil  Density , kg/L  1.12  0.93  Element Analysis C, wt.%  60.4  87.7  Element Analysis H, wt.%  6.9  8.9  Element Analysis O, wt.%  41.8  3  Element Analysis N, wt.%  0.9  0.4  HHV, MJ/kg  21.3  41.4  Solubility (methanol), wt.%  99  --  Solubility (toluene), wt.%  Little  100  Karimi et al. investigated the upgrading of 25 g of hemp seed pyrolysis oil at 623638 K and 5.5 MPa H2 (cold pressure) using 5 g of FexOy/SiO2/TiO2 (red mud) in slurry phase. The upgraded oil contained an aqueous and organic phase. The upgraded oil contained fewer oxygenated compounds and more saturated hydrocarbons than the crude pyrolysis oil [9]. A detailed summary of the hydroprocessing of pyrolysis oil was reported by Mortensen et al. [68] and is summarized in Table 2.6. From the HDO summary it can be  29  observed that all reaction systems employ very high temperatures and pressures, and generally long reaction times.  Table 2.6. Overview of catalysts investigated for catalytic upgrading of pyrolysis oil [68]. Reprinted and modified with permission from Elsevier, Copyright (2011).  Catalyst  Setup  Time  P  T (K)  (h)  (MPa)  HDO  O/C H/C  (%)  Yieldoil  Ref.  (wt.%)  Co-MoS2/Al2O3  Batch  4  20  623  81  0.8  1.3  26  [59]  Co-MoS2/Al2O3  Continuous  4a  30  673  100  0.0  1.8  33  [76]  Ni-MoS2/Al2O3  Batch  4  20  623  74  0.1  1.5  28  [59]  Pd/C  Batch  4  20  623  85  0.7  1.6  65  [59]  Pd/C  Continuous  4b  14  613  64  0.1  1.5  48  [73]  Ru/Al2O3  Batch  4  20  623  78  0.4  1.2  36  [59]  Ru/C  Continuous  0.2a  23  623-673  73  0.1  1.5  38  [74]  Ru/C  Batch  4  20  623  86  0.8  1.5  53  [59]  Ru/TiO2  Batch  4  20  623  77  1.0  1.7  67  [59]  a. b.  Calculated as the inverse of the weight hourly space velocity (WHSV). Calculated as the inverse of the LHSV.  2.6  Emerging Commercial HDO Processes  The overall goal for the commercialization of pyrolysis oil HDO is to have modular pyrolysis units close to the biomass source. This will densify the biomass for shipment to a centralized facility for upgrading. Current petroleum infrastructure utilizing HDS catalysts can be used. Dynamotive Energy Systems Corporation has developed a two-stage process that produces fuel from biomass at a cost of $2 per gallon, assuming the production accounts for 50% of the total cost this oil would sell for $4 per gallon which is less than the current 30  price of petroleum diesel at approximately $5 per gallon. The first stage is hydroreforming of their fast pyrolysis oil (BioOil) that decreases the O content of the oil to 10 wt.%, while increasing its HHV. Table 2.7 lists the reaction details and product analysis for UBA following treatment at 573 K, 12.4 MPa, and 1.5 g BioOil/(g-catalysth) [77]. UBA is the name given by Dynamotive for the hydrocarbon product following the initial HDO treatment.  Table 2.7. Reaction conditions and product analysis for Stage 1 (UBA) [77].  Component  Yield (wt.%)  UBA  45  Total Gas (~80% CO2 + 20% CH4)  8.1  Water Content of Aqueous Phase (81% H2O)  37.9  Organic Dissolved in Aqueous Phase (~38% methanol + 62% acetic acid)  8.9  Water Content  0.84  O Content  10  HHV (MJ/kg)  39.5  The final upgrading stage involves hydroprocessing over a commercial catalyst in the presence of hydrogen. The liquid product from stage 2 is designated as UBB. Table 2.8 lists the reaction details and product analysis for UBB following treatment at 623 K, 11.7 MPa, and 2.5 g UBA/(g-catalyst-h) [77].  31  Table 2.8. Reaction conditions and product analysis for Stage 2 (UBB) [77].  Component  Yield (wt.%)  UBB  83  Total Gas (~43% CO2 + 27% CH4 + 30% CO)  9.6  Water Content of Aqueous Phase (95% H2O)  8.6  Water Content  0.05  O Content  1  HHV (MJ/kg)  45  As evident from the table above, the water phase is relatively clean so acetic acid (worth on the order of $700/T-$1,500/T depending on purity) can be recovered using conventional technologies. This has been demonstrated in the laboratory using liquidliquid extraction. Since the HHV of BioOil is ~16 MJ/kg while that of hydrogen is ~121 MJ/kg, it may be seen from the data above that reaction is approximately thermoneutral so that the energy content of the BioOil has been effectively concentrated in UBA. This implies that the energy efficiency of this step is over ~ 80%. The HHV of UBB at ~45 MJ/kg is comparable to that of diesel. The overall yield of UBB from BioOil was 38%, while deoxygenation of BioOil was ~98% [77]. A 200 tonne per day (TPD) biomass pyrolysis plant will yield 130 TPD BioOil + 20 TPD char. This can be converted to 58.5 TPD UBA (0.45 x 130 TPD). H2 can be sourced from petroleum refineries for less than $3/kg for small scale deliveries out of pipeline range. The H2 consumption for UBA is 14 kg H2/(T BioOil). Conversion to UBB required 9 kg H2/(T BioOil). Therefore, the final stage of upgrading should be carried out at an oil refinery where hydrogen is cheaper and co-processed with petroleum. Acetic acid was assumed to be worth the low range of $800/T, where daily production would be 32  7.2 T. If 90% of the acetic acid was recovered, $5,180 per day of sales could offset the H2 cost. In addition char is valued at $150/tonne. Therefore, the total UBB production cost was estimated at $2/G [77]. The details for the economic analysis are given in Appendix A.1.  2.7   Concluding Remarks Pyrolysis oils are very complex systems. Therefore, in order to assess catalysts for HDO, a model compound should be used to represent the pyrolysis oil. Overall, methyl substituted phenols (mainly 4-methylphenol) are refractory and suitable compounds that can be used to model the HDO of pyrolysed biomass through both HYD and DDO paths. Therefore, these model compounds will be employed in the present study.    From the literature it was found that the decomposition of 4-methylphenol followed pseudo first order kinetic behavior. Conflicting reaction mechanisms for the decomposition of 4-methylphenol over sulfided Mo based catalysts suggest that there are both separate and dual sites available for HYD and hydrogenolysis reactions. Therefore, a goal of this study will be to determine the reaction path mechanism and the catalyst surface mechanism for the decomposition of 4-methylphenol over the catalysts tested.    Unlike petroleum oil, pyrolysis oils are S free so that the pyrolysis oil feedstreams do not readily activate Mo catalysts to a sulfided form. It is also undesirable to add sulfiding agents, such as H2S, to pyrolysis oil systems to reactivate catalyst surfaces. Thus, there is a need to move to alternative non-sulfided catalysts for HDO reactions. 33    HDS and HDO reactions are very similar. HDS involves the hydrogenolysis of a C-S bond, while HDO involves the hydrogenolysis of a C-O bond. Sulfided Mo catalysts, known to be active and selective for HDS [7] have demonstrated similar hydrogenolysis activity and selectivity in HDO [20]. Hence, other typical HDS catalysts could be examined for use in HDO for hydrogenolysis. Abu and Smith [41] have shown that phosphide catalysts are active and selective for HDS, and for that reason, phosphide catalysts will be investigated in the present study.    Recently, several authors [47, 51, 55, 56] have also found that phosphide catalysts are active for HDO reactions. However, to eliminate support interaction effects unsupported metal phosphides should be investigated. Unsupported high surface area metal phosphides can be prepared with the use of a chelating agent, such as CA [60, 61, 78]    For the HDO of pyrolysis oils, strong hydrogenating catalysts are required to prevent condensation reactions of unsaturated bonds [71]. Metal phosphides have strong hydrogenating abilities similar to supported noble metals [22, 48].    Overall, H2 for HDO should not be consumed to produce CH4. Therefore noncracking catalysts should be investigated.  34  Chapter 3 Experimental Methods This chapter will address the experimental methods that were employed in the current study. This includes the catalyst preparation and catalyst characterization, and the catalyst activity measurements for the HDO of 4-methylphenol and the HDO of pyrolysis oil.  3.1  Catalyst Preparation  Unsupported low surface area molybdenum disulfide (99%), molybdenum dioxide (99%), and molybdenum trioxide (99.5+%) were purchased from Sigma Aldrich, sieved to a particle size of dP < 53 µm and used as catalysts without further treatment.  3.1.1  Molybdenum Phosphide  Unsupported low and high surface MoP were prepared following the method of Stinner and Prins [43] and Wang and Smith [60] using precursor salts dissolved in water. Accordingly, for both MoP catalysts a P:Mo molar ratio of 1:1 was obtained by dissolving 4 g of AHM ((NH4)6Mo7O24.4H2O) and a corresponding amount of AHP ((NH4)2HPO4) in 30 mL of de-ionized water, with the addition of CA to the salt solution to give a 2:1 CA:Mo molar ratio for the high surface area catalyst [60, 61]. Appendix A.2 details the calculation for the required MoP precursor salt amounts. The precursor solution was aged for 24 h in a covered beaker held at 363 K in a water-bath and dried in an oven at 397 K for 24 h. The dried samples were calcined to 773, 823, 873, 923, or 973 K in air by heating to the desired temperature at 5 K/min, with the final temperature held 35  for 5 hours. The calcined catalyst precursors were ground to a powder (dP < 53 m) and converted to MoP by temperature-programmed reduction (TPR) in H2 at a flow rate of 160 mL(STP)/min and a heating rate of 5 K/min to 573 K, followed by a heating rate of 1 K/min to 923 K. The final temperature was held for 2.5 h. The samples were then cooled to room temperature in a He flow and passivated in a flow of 1 mol.% O2/He for 3 h prior to removal from the quartz u-tube reactor for characterization purposes. The catalysts used for activity measurements were transferred directly from the quartz u-tube used for reduction, under a He flow (25 mL(STP)/min) into ~15 mL of decalin. They were then transferred to the reactor for activity measurements, without exposure to air. In the present study, the solution of AHM and AHP, with or without CA, that was aged in a water bath at 363 K for 24 h and then dried in an oven at 397 K for 24 h, is referred to as the dried MoP precursor and designated as Dry-MoP-CA or Dry-MoPnoCA, respectively. The calcined MoP precursor refers to the dried precursors that were calcined in air to 773, 823, 873, 923 or 973 K. They are designated as Cal-MoP-CA-ttt K, where ttt is the calcination temperature (K). The sample prepared without CA was calcined at 773 K only and is designated as Cal-MoP-noCA. The reduced catalysts are identified as MoP-CA-ttt K, where ttt is the precursor calcination temperature (K). The reduced catalyst prepared without CA is designated as MoP-noCA. For one preparation, the CA ratio was increased to 4:1 and the sample was calcined to 773 K. This sample is denoted as MoP-4CA-773 K. Those catalysts that were passivated and tested for the HDO reaction without pre-reduction are designated as passivated MoP-CA-ttt K  36  3.1.2  Nickel Phosphide  Low and high surface area Ni2P catalysts were prepared with a P:Ni molar ratio of 1:1. Accordingly, 4 g of nickel nitrate (Ni(NO3)2.6H2O) with a corresponding amount of AHP ((NH4)2HPO4) were dissolved in 30 mL of de-ionized water [40, 60]. CA was added to the salt solution to give a 2:1 CA:Ni molar ratio. Although a P:Ni ratio of 2:1 for Ni2P prepared with CA has shown improved physical properties and higher activity for the HDS of 4,6-dimethyldibenzothiophene compared to Ni2P prepared with a P:Ni ratio of 1:1 [78], in the present study a P:Ni ratio of 1:1 was used so as to reduce the generation of toxic PH3 and avoid pyrophoric P sublimation during preparation. Following the addition of the salts and CA to water, the precursor solutions were aged for 24 h in a covered beaker held at 363 K in a water-bath and dried in an oven at 397 K for 24 h. The dried samples were calcined to 773, 823, and 973 K in stagnant air by heating at 5 K/min, with the final temperature held for 5 h. The calcined catalyst precursors were ground to a powder (dP < 53 m) and converted to Ni2P by TPR in H2 at a flow rate of 160 mL(STP)/min and a heating rate of 5 K/min to 573 K, followed by a heating rate of 1 K/min to 923 K. The final temperature was held for 2.5 h. The samples were then cooled to room temperature in a He flow and passivated in a flow of 1 mol.% O2/He for 3 h prior to removal from the quartz u-tube reactor for characterization. Other catalysts used for activity measurements were transferred directly from the quartz u-tube used for reduction, under a He flow (25 mL(STP)/min) into ~15 mL of decalin. They were then transferred to the reactor for activity measurements, without exposure to air. The solution of nickel nitrate and AHP, with or without CA, that was aged in a water bath at 363 K for 24 h and then dried in an oven at 397 K for 24 h, is referred to as 37  the dried Ni2P precursor and designated as Dry-Ni2P-CA or Dry-Ni2P-noCA, respectively. The calcined Ni2P precursors refer to the dried precursors that were calcined in air to 773, 823, or 973 K. They are designated as Cal-Ni2P-CA-ttt K, where ttt is the calcination temperature (K). The sample prepared without CA was calcined at 773 K only and is designated as Cal-Ni2P-noCA. The reduced catalysts are identified as Ni2P-CA-ttt K, where ttt is the precursor calcination temperature (K). Those catalysts that were passivated and tested for the HDO reaction without pre-reduction are designated as passivated Ni2P-CA-773 K.  3.2  Catalyst Characterization  Catalyst characterization was performed on the samples directly after preparation and these samples are referred to herein as the “fresh” catalysts. Characterization was also performed on the catalyst samples following the HDO reaction for 5 h at 623 K and 4.4 MPa, and these samples are referred to as “used” catalysts. The used catalysts were recovered from the reactor by decanting and filtering the reactor product solution. The recovered solids were washed with acetone followed by vacuum filtration to remove excess liquid. Appendix B.1 details the repeatability of the catalyst characterization.  3.2.1  Elemental Analysis  Elemental C, H, and N analysis was performed on the dried MoP and Ni2P precursors, the calcined MoP and Ni2P precursors, the reduced/passivated MoP and Ni2P, and selected used MoP and Ni2P catalysts following the 5 h HDO reaction, using a Perkin-Elmer 2400 Series II CHNS/O analyzer operated in the CHN mode. Prior to analysis all catalysts 38  were dried to 393 K to remove residual moisture. P analysis of selected samples was carried out using a colorimetric method, with direct comparison to a standard [79]. Mo and Ni concentrations of selected samples were determined by inductively coupled plasma atomic emission spectroscopy (ICP-AES). The P and metal analysis was performed by the Canadian Microanalytical Service Ltd (CMAS). O content was determined by the difference from 100 of the sum of all other elements.  3.2.2  TGA  Thermal gravimetric analysis (TGA) was also performed in air on the used MoO3 catalyst using approximately 10 mg of sample and a Perkin-Elmer TGA to monitor the change in mass during re-oxidation. Prior to analysis, the sample was treated in N2 (50 (STP)mL/min) at 393 K for 1.5 h. The gas flow was then changed to air at 50 mL(STP)/min and the temperature was ramped to 873 K at 10 K/min and held for 3 h. The TGA profiles were differentiated using Origin 8 software. TGA of ~10 mg of the dried MoP and Ni2P precursors was performed in a 50 mL(STP)/min flow of air while heating at 5 K/min to 773, 823, 873, 923, or 973 K. TGA of the precursor salts was performed similarly by heating to 973 K, with the final temperature held for 1 h.  3.2.3  DRIFTS  Diffuse reflectance infrared Fourier transform spectroscopy (DRIFTS) was performed on CA, the dried MoP and Ni2P precursors, the calcined MoP and Ni2P precursors, and the reduced/passivated MoP-CA and Ni2P-CA catalysts using a Nicolet FT-IR 5700 39  spectrophotometer equipped with a KBr beamsplitter, IR source, and MCT/B detector. The samples were tested in the range of 650-4000 cm-1 using 35 scans and a resolution of 2 cm-1. The calcined MoP precursors were ground to a fine powder (dp < 53 m), diluted with KBr and tested under a N2 atmosphere.  3.2.4  13C  NMR  Solid state nuclear magnetic resonance (NMR) experiments were performed on CA, the dried MoP-CA, the calcined MoP-CA precursors and the reduced/passivated MoP-CA catalysts. 13C cross-polarization magic angle sample spinning (13C-CPMAS) spectra were recorded on a Varian Inova 400, as well as a Bruker Avance 500 MHz instrument, operating at  13  C frequencies of 100.521 MHz and 125.691 MHz, respectively. The  samples were spun at a spinning frequency of 6,000 ± 0.002 or 10,000 ± 0.002 Hz. The 1  H 90º pulse was set to 2.5 μs and a ramped CP [80] with a contact time of 2 ms.  Decoupling was achieved using continuous wave or SPINAL-64 [81] decoupling, with a field strength of 66-100 kHz. The spectra were acquired using 16k or 32k scans and referenced to external adamantane,  CH 2 = 38.56 ppm and  CH = 29.50 ppm relative to TMS [82]. The spectra were analyzed using XWINNMR or DMFIT [83]. This analysis was performed by Dr. Suzana Straus from the Department of Chemistry at UBC. The solid state proton decoupled  13  C NMR data for the dried and calcined Ni2P-  CA precursors were acquired on a Bruker 400 MHz Avance spectrometer running with xwinnmr 2.6. The MAS speed was set at 6 kHz. The contact time to establish cross polarization was set to be 2 ms. Total number of scans was approximately 60,000 with a  40  recycle delay of 5 seconds. This analysis was performed by Dr. Paul Xia from the Department of Chemistry at UBC.  3.2.5  XRD  Powder X-ray diffraction (XRD) spectra were collected on both the fresh and used low surface area catalysts, MoS2, MoO2, MoO3, as well as the calcined MoP precursors, reduced/passivated MoP, and selected used MoP catalysts, using a Rigaku diffractometer with a CuKα X-ray source of wavelength of 1.54 Å. The analysis was performed using a 40 kV and 20 mA source, a scan range of 10–80o with a step size of 2/min. XRD patterns of the calcined Ni2P precursors, reduced/passivated Ni2P catalysts, and selected used Ni2P-CA were collected using a Bruker D8 Focus (LynxEye detector) with a Co Kα X-ray source of wavelength 1.7902 Å. The phase identification was carried out after subtraction of the background using standard software and powder diffraction files (PDF) for reference. Crystallite size (dc) estimates were made using the Scherrer equation,  𝑑𝑐 =  𝐾𝜆 𝛽cos𝜃  (3.1)  where the constant K was taken to be 0.9, λ is the wavelength of radiation, β is the peak width in radians, and θ is the angle of diffraction.  41  3.2.6  XPS  A Leybold Max200 X-ray photoelectron spectrometer (XPS) was used for surface analysis of the low surface area catalysts of MoS2, MoO2, MoO3, and MoP-noCA as well as analysis of the MoP precursors. Al Kα was used as the photon source generated at 15 kV and 20 mA. The pass energy was set at 192 eV for the survey scan and 48 eV for the narrow scan. The used and fresh catalysts were analysed. All XPS spectra were corrected to the C 1s peak at 285.0 eV. Deconvolution of the XPS profiles was done utilizing XPS Fit software. XPS was performed by Dr. Ken Wong from the Interfacial Analysis and Reactivity Laboratory at UBC.  3.2.7  BET  Brunauer-Emmett-Teller (BET) analysis was performed on the used and fresh low surface area catalysts MoS2, MoO2, and MoO3 as well as the fresh MoP and Ni2P catalysts and their calcined precursors. Surface areas were measured by N2 adsorption at 77 K using a volumetric unit (Micromeritics Flowsorb II 2300). Prior to N2 adsorption the samples were degassed at 423 K in 30 mol.% N2/He (15 mL(STP)/min) for 2 h to remove moisture. Similarly, the surface area of select used MoP and Ni2P catalysts was measured following the 5 h HDO reaction.  3.2.8  TPR  Temperature programmed reduction (TPR) of the fresh and used low surface area Mo oxide samples was performed in H2. TPR was also performed on the MoP-noCA calcined catalyst precursor and the passivated MoP-noCA. TPR experiments were carried out 42  using a Micromeritics AutoChem II 2920 automated catalyst characterization flow system. Prior to the TPR, approximately 0.1 g of catalyst was pretreated in Ar (50 mL(STP)/min) at 723 K for 1 h and then cooled to room temperature. Subsequently, the Ar flow was switched to 9.5 mol.% H2/Ar (50 mL(STP)/min) and the sample was heated from room temperature to 923 K at 5 K/min in accordance with the MoP preparation method. TPR was also performed on the fresh and used MoO3, and the fresh MoO2 at a heat up rate of 5 K/min to 998 K. The H2 consumption was measured via TCD. The TPR profiles were differentiated using Origin 8 software. Additionally, TPR of selected calcined MoP precursors was performed using a utube reactor placed in a temperature programmable muffle furnace with a quadrupole MS (SRC Residual Gas Analyzer, 200 amu) connected to the reactor effluent. The reactor was heated to 1173 K at 10 K/min in H2 at 50 mL(STP)/min and the reactor effluent passed directly to the MS without passing through a cold trap.  3.2.9  TPO  The average oxidation state of molybdenum in the oxide catalysts was determined by performing temperature programmed oxidation (TPO) on the used and fresh Mo oxide catalysts. Re-oxidation of the used catalyst was performed using a Micromeritics Autochem II 2920 unit. Prior to the TPO experiment, the catalyst was pretreated in He (50 mL(STP)/min) at 723 K for 1 h, and then cooled to room temperature. Subsequently, the He flow was switched to 10 mol.% O2/He (50 mL(STP)/min) and the sample was heated to 873 K for 2 h at a rate of 5 K/min. The O2 consumption was measured by a TCD. The TPO profiles were differentiated using Origin 8 software. 43  3.2.10 CO Chemisorption The CO uptake of the fresh and used MoS2, MoO2, and MoO3 catalysts was measured by a Micromeritics AutoChem II 2920 unit using pulsed chemisorption. The catalysts were pre-treated in Ar (50 mL(STP)/min) at 723 K for 1 h, and then cooled to room temperature. After this treatment, 0.5 mL pulses of CO were injected into a flow of He (50 mL(STP)/min) and the CO uptake was measured using a TCD. CO pulses were repeatedly injected until no further CO uptake was observed after consecutive injections. The CO uptake of the reduced MoP and Ni2P catalysts was measured by pulsed chemisorption using a Micromeritics AutoChem II 2920 unit. The reduced MoP and Ni2P samples were prepared from the calcined MoP and Ni2P precursors by in-situ reduction in 9.5 mol.% H2/Ar (50 mL(STP)/min) while heating at 5 K/min to 573 K followed by a ramp of 1 K/min to 923 K with the final temperature held for 1.5 h (replicating the standard reduction procedure of the calcined MoP and Ni2P precursors). The sample was then cooled in 50 mL(STP)/min He to room temperature prior to injecting pulses of CO. Similarly, the CO uptake of selected used MoP and Ni2P samples was determined by recovering the samples following the HDO reaction. These samples were re-reduced insitu with 9.5 mol.% H2/Ar (50 mL(STP)/min) while heating at 10 K/min to 773 K for 2 h. The samples were subsequently cooled in 50 mL(STP)/min He to room temperature prior to injecting pulses of CO. The CO uptake of the passivated MoP-CA, passivated Ni2PCA, and Cal-MoP-CA were tested without re-reduction by first flowing 50 mL(STP)/min He to 393 K for 2 h followed by cooling to room temperature and injection of pulses of CO.  44  3.2.11 Acid Site Titration The acid sites of the low surface area catalysts MoS2, MoO2, MoO3, and MoP-noCA were titrated using n-propyl amine (n-PA). The acidity analyses were performed using a Micromeritics AutoChem II 2920 unit. The catalyst was pre-treated in the same way as that used for the CO uptake measurement. A flow of He (50 mL(STP)/min) saturated at room temperature with n-propyl amine (Aldrich, 99.8%) was then injected onto the pretreated catalyst in 0.5 mL pulses. The adsorption was performed at 298 K and a thermal conductivity detector was used to quantify the amount of n-PA adsorbed. Pulses of n-PA were repeatedly injected until no further uptake was observed after consecutive injections.  3.2.12 SEM and TEM Scanning electron microscopy (SEM) images of selected, reduced/passivated MoP were generated at 2000x magnification using a Hitachi S-2300 SEM. Transmission electron microscopy (TEM) images of selected samples, deposited on C-coated Cu grids, were taken using a 120 kV Hitachi H7600 with a tungsten filament. The reduced/passivated catalyst samples were dispersed in ethanol and deposited on C-coated Cu grids. Higher resolution TEM images of the fresh MoP and Ni2P and used MoP and Ni2P were also generated using a FEI Tecnai TEM operating at 200 kV with a LaB6 filament. These images were used to determine the d-spacing of the crystallites. Lognormal particle size distributions were obtained by editing the images in Pixcavator 4.0 Image Analysis software using several images. The number of particles counted was > 50 for the CA samples, whereas for the larger noCA samples, the number of particles counted was > 10. 45  It is important to note that the particle sizes are based on the particle diameter and/or width dimension. TEM analysis was performed by Bradford Ross from the Bioimaging Facility at UBC.  3.2.13 Precursor Structures The geometry of the proposed MoP-CA catalyst precursor states were optimized and drawn using Dmol3 V4.1 software (Accelrys Inc). The double-numeric quality basis set with the gradient corrected GGA functional and PW91 were used in the optimization. For the numerical integration, a Fermi smearing of 5E-3 Ha (1 Ha = 27.21 eV), unrestricted spin, and the MEDIUM quality mesh size of the program (using about 1000 grid points for each atom in the calculation) were used. The tolerances of energy, gradient, and displacement convergence were 2E-5 Ha, 4E-3 Ha/Å, and 5E-3 Å, respectively.  3.3  Catalyst Activity Tests  The activity of the catalysts was tested for the HDO of 4-methylphenol and fast pyrolysis oil.  3.3.1  4-Methylphenol HDO  The HDO of 4-methylphenol was measured in a 300 mL stirred-batch autoclave reactor (shown in Figure 3.1) operated at 598 K, 623 K, and 648 K using low surface area MoS2, MoO2, MoO3, and MoP-noCA. Initially 100 mL of 2.96 wt.% 4-methylphenol (Aldrich, 97%) in decalin (Sigma-Aldrich, 98%) was added to the reactor and slurried with the catalyst to yield an equivalent Mo concentration of 8,000 ppmw. The 4-methylphenol 46  reactant was also used in past studies by Yang et al. [20], Gevert et al. [23], and Laurent and Delmon [33] to represent the phenolic fraction of pyrolysis oil. The reactant concentration employed was similar to that used by Odebunmi and Ollis [28] for HDO studies. To validate the initial 4-methylphenol concentration, several initial concentration samples were injected into the GC and the repeatability analysis is given in Appendix B.2. Decalin was used as a solvent as it is a fully saturated compound and it was found to be stable at the reaction conditions by running it in the presence of catalyst and absence of reactant at the reaction conditions. The reactor was first purged in 55 mL(STP)/min N2 for 1 h. Finally the reactor was pressurized with ultra high purity (UHP) H2 to 2.4 MPa and then heated to the desired temperature at 10.8 K/min and a stir rate of 1000 rpm. Subsequently the H2 pressure increased to 4.1 MPa, 4.4 MPa, or 4.8 MPa following heatup. During the heat up phase the catalyst was activated and minimal reaction occurred. Reaction during the heat-up phase was accounted for by analyzing the liquid in the reactor once reactor temperature was reached and setting this measured concentration as the concentration at time = 0 minutes. The H2 pressure, stirrer speed, and temperature were continuously monitored during the experiment. Small volumes (< 0.5 mL) of liquid sample were periodically removed from the reactor for analysis. The samples were collected from a liquid sampling line that was initially purged by withdrawing 2 mL of sample prior to collecting the 0.5 mL sample for analysis. The liquid samples were analysed by GC using a 14-A Shimadzu gas chromatograph equipped with a flame ionization detector (FID) and an AT-5 25 M × 0.53 mm capillary column. The product distribution was confirmed by GC/MS analysis using a Shimadzu QP-2010S GC/MS and  47  a Restek RTX5 30 M × 0.25 mm capillary column. Details of the set-up are provided in Appendix C. For all the activity data reported herein, the C balance across the reactor was > 90% and a number of the experiments were repeated to ensure repeatability of the data. The decomposition of 4-methylphenol was modeled assuming a pseudo 1st-order reaction given in Eq. 3.2.  ln(1 − 𝑥) = −𝑘𝐶cat 𝑡  (3.2)  where t is the HDO reaction time (min), Ccat is the concentration of the catalyst in the reactor at ambient conditions (g/mL), x is the conversion, and k is the kinetic parameter (mL.min-1.g-1).  Figure 3.1. 300 mL batch reactor used for HDO studies.  48  Similarly, HDO reactions were carried out in the stirred-batch reactor operated at 623 K and 4.4 MPa H2 using 3,900 ppmw MoP or Ni2P catalyst, assuming pure MoP/Ni2P and the density of the solvent at ambient conditions. The MoP catalyst mass required for this concentration is given in the sample calculations of Appendix A.2. Less catalyst was used in these tests due to the limitation of the amount of reduced metal phosphide produced from the u-tube reduction. Another test involved carrying out the HDO reaction at 623 K and 4.4, 5.1, or 6.3 MPa H2 using Ni2P-CA-773 K to investigate the effect of pressure on the catalyst deactivation encountered over these catalysts. Additionally, a metal phosphide catalyst concentration of 3,900 ppmw of passivated MoP-CA, passivated MoP-noCA, passivated Ni2P-CA, and Cal-MoP-CA were also tested for the HDO of 4methylphenol at 623 K and 4.4 MPa without re-reduction to observe the effect of surface oxidation. The deactivation observed over the Ni2P catalysts was modeled using the exponential decay law given in Eq. 3.3.  Rate = 𝑘𝐶a e−𝑘d𝐶cat 𝑡  (3.3)  where kd is the deactivation kinetic parameter (mL.min-1.g-1), t is the HDO reaction time (min), Ccat is the concentration of the catalyst in the reactor at ambient conditions (g/mL), and Ca is the concentration of 4-methylphenol (mol/mL). The C balance over the Ni2P catalysts was lower than that of the other catalysts. However, once the C deposition on the catalyst surface was accounted for, the carbon balance was > 95% for these experiments as well. 49  Following the HDO reaction that proceeded in the batch reactor for up to 5 h, selected MoP and Ni2P catalysts were recovered, washed in acetone, dried, and characterized by CHNS, M:P ratio, BET, CO chemisorption, XRD, and TEM using the methods already described in Section 3.2. The absence of internal and external mass transfer effects on the measured reaction kinetics was confirmed experimentally and by theoretical analysis. The volumetric liquid side mass transfer coefficient was estimated according to Dietrich et al. [84] for stirred tank bench top reactors operated in slurry phase. The Sherwood number was calculated using a correlation by Albal et al. [85]. The mass transfer coefficient was at least 1000 times higher than the observed reaction rate. The details of which are given in Appendix D. Furthermore, repeating the reactions at stirrer speeds of 600 and 1000 rpm showed that the measured reaction rates were within 10% of the average values. Internal mass transfer effects were minimal because of the small catalyst particle size and this was confirmed by operating the reactor with catalysts of particle size 53 µm and 180 µm. In both cases the measured reaction rates were within 10% of the average values. Overall, the C balance across the reactor was shown to be > 93%. In addition, all product selectivities for the HDO of 4-methylphenol are given in units of mol.%. Examples of the reaction repeatability and C balance for the HDO of 4methylphenol are given in Appendices B.3 and B.4, respectively.  3.3.2  4-Methylphenol Kinetic Analysis  A kinetic reaction network for 4-methylphenol HDO was also proposed from the experimental data of the present study. It was assumed that all reactions were 1st-order in 50  the component liquid concentrations and zero-order in H2 because of the large excess of H2 in the reactor. The kinetic parameters of the kinetic model were estimated by minimizing an objective function, defined as the sum of squares of the deviation between the measured component concentrations and those calculated by the model, using a Levenberg-Marquardt non-linear regression. At each iteration, the concentration of each compound as a function of time was calculated by numerical integration of the ordinary differential equations that described the time rate-of-change of the product and reactant concentrations. The ordinary differential equations were solved using the initial concentration values in a Runge-Kutta scheme of 5th-order with variable step size. In addition, the mole balance constraint was used to calculate the reactant concentration at each time step. The model was solved simultaneously in Matlab 7.1. The standard deviations (SD) of the kinetic parameters were calculated from the square root of the diagonal of the kinetic parameter covariance matrix [86]. Details of the kinetic model code are given in Appendix E.  3.4  Pyrolysis Oil HDO  Fast pyrolysis oil was obtained from Dynamotive. The oil was kept refrigerated at approximately 275 K.  Prior to use the oil was mixed vigorously to inhibit phase  separation. The pyrolysis oil C, H, N, S, and O (by difference) content was measured using a Perkin-Elmer 2400 Series II CHNS/O analyzer operated in the CHNS mode. The moisture content of the oil was measured using Karl Fischer using a 794 Basic Titrino from Brinkmann/Metrohm with aqualine. The HHV was measured using a Parr bomb calorimeter that combusted the pyrolysis oil sample. 51  The HDO of the pyrolysis oil was done at 523 K and 2.9 MPa (cold pressure) using a 300 mL stirred-batch autoclave reactor loaded with 10 mL of pyrolysis oil, 90 mL decalin diluent (inert), and 0.5 g of catalyst (MoP-CA-823 K, Ni2P-CA-773 K, MoS2, MoO3, and NiMoS/Al2O3). This catalyst loading was based on that used by previous studies [15, 59, 75]. The reactor was operated in a batch mode in both the liquid and gas phase. The reactor was heated to the desired temperature at 10.8 K/min and a stir rate of 1000 rpm. This heating rate was based on that used by previous studies [59, 75]. The reaction time was 1 h. A diluent was used because of pressure limitations encountered with the batch phase reactor based on the fact that the pyrolysis oil contained 24.6 wt.% O on a dry basis. 10 mL of oil yielded a mass of 12.2 g of oil. Therefore, 0.188 moles of H2 would be required to deoxygenate 10 mL of oil at a cold pressure of 2.3 MPa. When pure pyrolysis oil (100 mL) was reacted, H2 starvation occurred and the oil formed a solid mass due to the fact that 23 MPa was required. Please note that this, 23 MPa, pressure requirement was in excess of the safety limit of the reactor employed in this study. Therefore, the pyrolysis oil was diluted with inert decalin. Zhang et al. [15] also studied the HDO of pyrolysis oil using a tetralin diluent with a 10:1 diluent:pyrolysis oil ratio. Following cool down the gas phase was extracted using a gas bag and measured via GC-TCD (GC-2014 Shimadzu). The liquid phase from the reaction was decanted and the aqueous phase was separated from the organic phase (containing the decalin diluent) using a syringe. The organic phase collected changed from a clear liquid (decalin) to a yellow or orange liquid with varying transparency depending on the catalyst used. All aqueous phases were clear in color but turned a light brown following exposure to air. The solid product (containing the catalyst) was extracted from the reactor walls using a 52  brass brush drill. When the solid was washed with acetone an acetone-soluble dark viscous oil and solid was obtained. All liquid and solid product C, H, N, S, and O (by difference) contents were measured using a Perkin-Elmer 2400 Series II CHNS/O analyzer operated in the CHNS mode. The pyrolysis oil HDO reaction repeatability and product analysis repeatability are given in Appendix B.5.  53  Chapter 4 Catalyst Screening for the HDO of 4-Methylphenol1 4.1  Introduction  The O in pyrolysis oils can be removed from the hydrocarbon species via HDO. Much research on the removal of oxygen by HDO is available on conventional supported metal sulfide catalysts, as well as on supported noble metal catalysts, that are known to be active for sulfur removal from crude oils using high pressure hydrogen [15, 20, 31-35]. Noble metals are expensive and have high selectivity for hydrogenation reactions that consume H2. Although sulfided metal catalysts are very active for oxygen removal, oxidation of the active catalyst phase can occur during the hydrotreating of pyrolysis oil because of the oil’s high O content. The addition of a sulfiding agent such H2S or CS2 is required to maintain the catalyst S content, but this is undesirable [19]. Hence, there is a need to move from traditional metal sulfide catalysts to alternative catalysts for HDO processes. Sulfided Mo catalysts that are known to be active and selective for C-S scission in HDS [7] have also demonstrated high activity and selectivity for C-O scission in HDO [20]. Similarly, other catalysts that are known to be active for HDS are candidates for  1  Work in this chapter was reported previously in: V.M.L. Whiffen and K. J. Smith, “Hydrodeoxygenation of 4-Methylphenol over Unsupported MoP, MoS2, and MoOx Catalysts,” Energy & Fuels, 24 (9), 2010, pp. 4728-4737.  54  HDO, such as metal phosphides. Recently, several authors have shown that metal phosphides are active and selective in HDS [39-41]. Furthermore, due to the presence of high oxygen concentrations in pyrolysis oil, oxidation of the active catalyst phase may occur during the hydroprocessing reaction and the oxidized catalyst may also play a role in the HDO. Metal oxides can be reduced to form surface defects such as CUS that can catalyze hydrogenolysis and hydrogenation reactions at typical HDO conditions [21, 24]. Hence metal oxides are also candidates as alternative HDO catalysts. Furans and phenols account for 3 – 10 wt.% of all the compounds in pyrolysis oil [5] and since these are the most refractory species present in the pyrolysis oil, they have been used as model compounds to represent pyrolysis oil feedstocks in past studies of catalytic hydrotreatment [15, 21, 32, 87, 88]. In the present Chapter, a systematic study of the activity and selectivity of different Mo catalysts for the HDO of 4-methylphenol is reported. Unsupported MoS2, MoP, MoO2, and MoO3 catalysts, used to eliminate support interaction effects, as well as internal mass transfer limitations, have been examined for the HDO of 4-methylphenol.  4.2  Results and Discussion  The catalysts were characterized using the methods described in Chapter 3.2. In addition, the catalysts were tested for the HDO of 4-methylphenol using the methods described in Chapter 3.3.  55  4.2.1  Catalyst Characterization  The X-ray diffractogram of the fresh MoP after passivation, presented in Figure 4.1, confirmed that phase pure MoP had been successfully prepared using the methods described in Chapter 3 and based on the powder diffraction file (PDF#065-6487) for MoP. Following the 5 h reaction at 623 K and 4.4 MPa the structural integrity of the MoP was maintained. Table 4.1 also shows that the crystallite size of the fresh and used MoP, estimated by XRD line broadening, were of similar dimensions (20-30 nm, Table 4.1).  Used MoP  Intensity  Fresh MoP  Used MoS2  Fresh MoS2 10  20  30  40  50  60  70  80  Angle, 2  Figure 4.1. X-ray diffractogram of fresh and used MoP, and fresh and used MoS2.  Figure 4.1 shows similar results for the used and fresh MoS2 with comparable crystallite sizes in the range of 35-130 nm (Table 4.1). The diffractograms are indicative of MoS2 (PDF#065-0160). 56  The X-ray diffractogram of the fresh MoO2 (Figure 4.2) showed the presence of monoclinic MoO2, Mo4O11, and MoO3 phases, suggesting that air exposure resulted in some oxidation of the MoO2. The proportion of MoO3 (PDF#047-1320) in the catalyst was minimal (< 5%), as evidenced by the low intensity features at 2θ = 23.1o, 27.6o, 49.5o, and 55.9o, compared to those of the dioxide. The monoclinic Mo4O11 (PDF#0720447) phase also displayed low intensity features at 2θ = 24.7o and 28.2o. Following the HDO reaction for 5 h at 623 K and 4.4 MPa, the MoO2 composition changed. Figure 4.2 shows that the trioxide and Mo4O11 phases were no longer visible, suggesting that these phases were reduced to monoclinic MoO2 (PDF#032-0671) under H2 at 4.4 MPa and 623 K. The X-ray diffractogram of the used and fresh MoO3 is also given in Figure 4.2. The fresh MoO3 features are consistent with monoclinic MoO3, however, following reaction, a significant phase transformation occurred resulting in a less crystalline solid recovered from the reactor. The color of the MoO3 changed from light green (MoO3) to dark blue after reaction, indicative of Mo oxidation states (V) and (VI) [89]. Additionally, the phases present in the catalyst changed from crystalline MoO3 with an intense peak at 2θ = 25.8o (PDF#047-1320), to a mixture of MoO2 and Mo4O11, where the main peak angle was 2θ = 26.2o. The peak at 2θ = 26.2o is the third most prominent characteristic peak of Mo4O11. However, the first and second largest peaks for Mo4O11 at 2θ = 22.6o and 24.7o (PDF#072-0447) were not present in the X-ray diffractogram, which was due to the amorphous nature of the sample. Figure 4.2 and Table 4.1 show that the main peaks of the fresh MoO3 catalyst shifted to MoO2 and Mo4O11. It is possible that other phases are present, however the amorphous nature of the used catalyst does not allow for 57  distinct identification and quantification. The crystallite size estimated by XRD line broadening also decreased by a factor > 10.  Used MoO2  Fresh MoO2  Intensity  *  *  *  * Used MoO3  *  10  *  *  *  *  20  Fresh MoO3  * * * *  *  30  40  * * * * * * * ** * * 50  60  ** *  * 70  * 80  Angle, 2  Figure 4.2. X-ray diffractogram of fresh and used MoO2, and fresh and used MoO3. MoO2 (■), Mo4O11 (□), and MoO3 (*).  Based on the work of Matsuda et al. [90], it was assumed that under the reaction conditions, the in-situ reduction of the fresh crystalline MoO3 began preferentially at the shear planes. As reduction proceeded at these planes, the metal lattice contracted and fractured the crystal decreasing its size and creating micropores due to oxygen removal [90].  58  Table 4.1. Main diffraction peaks and crystallite sizes of fresh and used Mo catalysts.  Angle (plane)  Crystallite size  Angle (plane)  Crystallite size (nm)  (nm) Fresh MoP  Used MoP  42.9o (101)  22.6  43.0o (101)  23.3  32.0o (100)  22.5  32.0o (100)  24.2  27.9o (110)  26.3  27.9o (110)  28.3  Fresh MoS2  Used MoS2  14.4o (002)  139.1  14.4o (002)  129.6  39.6o (103)  46.4  39.6o (103)  45.9  49.8o (105)  39.1  49.8o (105)  36.6  Fresh MoO2  Used MoO2  26.1o (-111)  74.7  26.0o (-111)  73.4  37.0o (-211)  26.4  37.0o (-211)  24.7  53.5o (-312)  27.5  53.5o (-312)  26.1  Fresh MoO3  Used MoO3  25.7o (040)  91.1  26.2o (-111)a (211)b  38.7  39.0o (103)  87.6  36.7o (200)a (1000)b  15.9  27.4o (021)  200.7  53.6o (-222)a (1311)b  17.8  a. b.  Plane for monoclinic MoO2 Plane for monoclinic Mo4O11  The measured physicochemical properties of the catalysts are compared in Table 4.2. The BET surface area, CO chemisorption, and total acidity measured by n-PA adsorption, were expectedly low on all the bulk catalysts. The BET surface area, CO chemisorption, and n-PA adsorption properties of the MoP, MoS2, and MoO2 were unchanged following recovery of the catalyst after 5 h reaction at 623 K and 4.4 MPa. Consequently, the properties of the used catalysts are not reported separately in Table 4.2. The surface area of the fresh MoO3 was found to be very low, but increased by a factor of 59  10 following reaction. Similarly, Matsuda et al. [91] reported that bulk MoO3, with a surface area of 5 m2/g, was transformed into microporous MoxOy with an increased surface area after reduction in H2 at 623 K. As noted above, reduction of the crystalline trioxide generates micropores due to the shearing of the oxide planes, which in turn causes the surface area to increase. This result is consistent with the crystallite size measurements by XRD of the fresh and used MoO3 (Table 4.1). The surface acidity was found to be lowest over the phosphide catalyst and highest over the oxide catalysts. Following reaction and recovery of the catalyst, the surface acidity of the MoO3 catalyst decreased. This was because the surface acidity was associated with the oxide layer of the catalyst. As in-situ reduction occurred over the trioxide, the amount of oxide species decreased. The CO adsorption data of Table 4.2 correspond to very low metal dispersions for each catalyst (<0.02%), whereas the MoO3 CO uptake was approximately zero because of its insulating properties. Following reaction and in-situ reduction of MoO3, the total Mo available on the surface increased significantly. The partial reduction of the trioxide catalyst that occurs in this case creates fairly well reduced metal sites and the formation of CUS that aid in the dissociation of the CO [64].  60  Table 4.2. Surface area, CO uptake, and total acidity of Mo catalysts.  Sample  SBET  CO Uptake  Total Acidity  (m2/g)  (µmol/gcat)  (µmol/gcat)  MoP  8.80  0.5  3.6  MoS2  4.30  0.9  9.8  MoO2  4.80  1.5  33.3  Fresh MoO3  0.30  < 0.5  121.5  Used MoO3  3.50  11.0  77.8  The TPR profile of the MoP calcined catalyst precursor has been reported previously by Wang and Smith [60], and Abu and Smith [41]. As noted previously, the degree of reduction of the MoP calcined precursor is difficult to quantify by TPR because of an unknown reaction stoichiometry. However, assuming a stoichiometry of (2MoO3)∙P2O5 + 11H2  2MoP + 11H2O, yielded a 62% reduction of the calcined MoP precursor. Following preparation by TPR, the MoP was passivated in 1 mol.% O2/He. TPR of the passivated MoP, given in Table 4.3, showed a H2 uptake of 1.8 mmol/g, equivalent to ~5% of the H2 needed to reduce the completely oxidized precursor. TPR of the fresh MoO3 resulted in a H2 uptake of 5.0 mmol/g, equivalent to a reduction of 24% of the MoO3 to Mo, assuming the stoichiometry for direct reduction from Mo6+ to Mo0. The fresh MoO2 consumed < 15% (or 2.2 mmol/g) of the H2 needed for complete reduction of MoO2, assuming the stoichiometry for direct reduction from Mo4+ to Mo0. After reaction, the used MoO3 consumed 4.1 mmol/g, which represents an uptake between that of the MoO3 and MoO2, suggesting that the used MoO3 had prominent phase with an oxidation state between Mo6+ and Mo4+, consistent with the XRD of the used MoO3 that showed the presence of MoO2, and Mo4O11. Ressler et al. [92] have 61  observed that the reduction of MoO3 to MoO2 is a direct reaction with no intermediates at temperatures between 623 K and 698 K (1 atm), whereas at temperatures above 698 K (1 atm) the molybdenum oxide (Mo4O11) can be formed as a result of a parallel reaction between MoO2 and MoO3. However, Burch [93] and Sloczyn´ski and Bobin´ski [94] suggested that Mo4O11 is an intermediate product of the reduction process within the temperature range of 723-823 K. Sloczyn´ski [95] has also proposed that the dissociative adsorption of hydrogen is the rate-determining step of the reduction. In the present study the use of high pressure H2 (4.1-4.8 MPa) at 623-648 K could accelerate the formation of Mo4O11 at the lower temperature due to the dependence of reduction severity on the hydrogen pressure.  Table 4.3. H2 consumption of various catalysts using TPR technique.  Sample  H2 Uptake (mmol/gcat)  Passivated MoPa  1.8  Fresh MoO3b  5.0  Used MoO3b  4.1  Fresh MoO2b  2.2  a. b.  reduced to 923 K at 5 K/min. reduced to 998 K at 5 K/min.  TPO of the used MoO3 is compared to that of MoO2 in Figure 4.3. The used MoO3 required 1.2 mmol/g O2 for re-oxidation, whereas the fresh MoO2 required 2.65 mmol/g for oxidation. This suggests that the fresh MoO2 catalyst had a lower oxidation state (Mo4+) than the used MoO3. Three main peaks were observed in the TPO of the used MoO3 (Table 4.4). The first peak was attributed to the oxidation of Mo0 to Mo4+, the 62  second peak was attributed to the oxidation of Mo4+ to Mo6+, and the final peak was attributed to the oxidation of Mo4O11 to Mo6+. Deconvolution of the three peaks in Figure 4.3 found that peak A accounted for 7% of the oxygen uptake, peak B for 29% and peak C for 64%. Based on these results the used catalyst composition was calculated to be 1 wt.% Mo, 10 wt.% MoO2, and 89 wt.% Mo4O11. These results are consistent with the XRD data. However, TPO analysis provides reliable quantification of the phase  O2 Consumption  compositions. Therefore, the used MoO3 catalyst is referred to as MoxOy.  A  B  C  Fresh MoO2 Used MoO3  300  400  500  600  700  800  Temperature (K)  Figure 4.3. TPO of fresh MoO2 and used MoO3.  63  900  Table 4.4. Oxygen consumption of used MoO3 for each TPO peak.  Peak  Temperature  Assumed Path  (K)  O2 Consumption  MoOx Species  (μmol/g)  (wt.%)  A  630  Mo + O2  MoO2  87  1  B  688  2MoO2 + O2  2MoO3  360  10  C  773  Mo4O11 + ½O2  4MoO3  793  89  Re-oxidation of the used MoO3 was also performed by TGA and the differential TGA profile is displayed in Figure 4.4. The total mass gain of the sample during oxidation was 51.4 μg/mg. Three main peaks were observed for the re-oxidation of the used MoO3 (Table 4.5). The first peak was attributed to the oxidation of Mo0 to Mo4+, the second peak was attributed to the oxidation of Mo4+ to Mo6+, and the third peak was attributed to the oxidation of Mo4O11 to Mo6+. The shift in the oxidation peaks in Figure 4.4 compared to Figure 4.3 was due to different heat up rates used for TPO (5 K/min) and the TGA (20 K/min). Deconvolution of the three peaks in Figure 4.4 found that peak A accounted for 9% of the mass increase, peak B for 31% and peak C accounts for 60%, and based on these results the used catalyst composition was calculated as 1 wt.% Mo, 11 wt.% MoO2, and 88 wt.% Mo4O11. These results are in very good agreement with those obtained from the TPO analysis.  64  -4  8.0x10  A  B  C  -4  Weight Gain (mg/s)  6.0x10  -4  4.0x10  -4  2.0x10  0.0 500  550  600  650  700  750  800  850  900  Temperature (K)  Figure 4.4. TGA re-oxidation of used MoO3.  Table 4.5. Mass increase for the re-oxidation of MoO3 via TGA.  Peak Temperature  Assumed Path  (K)  Mass Increase  MoOx Species  (μg/mg)  (wt.%)  A  668  Mo + O2  MoO2  4.6  1  B  725  2MoO2 + O2  2MoO3  16.2  11  C  796  Mo4O11 + ½O2  4MoO3  30.6  88  The MoP XPS spectra of the fresh catalyst were reported previously by Wang and Smith [60] and are summarized in Table 4.6. The measured P/Mo ratio shows an excess of P at the surface of the MoP. The XPS Mo (3d) region of the other Mo catalysts is shown in Figure 4.5 and the corresponding binding energies (BEs) are also reported in Table 4.6. The Mo (3d) XPS regions were fit using XPS Fit software and fixing the BEs, the Mo 3d5/2 and 3d3/2 energy split, and the full width at half maximum (FWHM) 65  according to literature values. The error associated with the fit and the chemical composition was approximately ± 12%. Fresh MoS2 (Figure 4.5 (a)) displayed two well resolved spectral lines at BEs of 229.9 and 233.0 eV, assigned to the Mo 3d5/2 and 3d3/2 spin-orbit components of Mo in MoS2. The fresh MoO2 (Figure 4.5 (b)) was found to contain 45 mol.% Mo4+ (BEs of 230.1 and 233.2 eV), 30 mol.% Mo5+ (BEs of 231.4 and 234.5 eV), and 25 mol.% Mo6+ (BEs of 233.3 and 236.5 eV) [96, 97]. The MoO2 also displayed excess oxygen at the surface. These results indicated that the MoO2 was oxidized at the surface by air exposure, leading to the presence of MoO3 and Mo5+. XRD also confirmed the presence of MoO3 and Mo4O11 in the fresh MoO2 sample. Fresh MoO3 showed two well resolved spectral lines at a BE of 233.2 and 236.3 eV, assigned to the Mo 3d5/2 and 3d3/2 spin-orbit components (Figure 4.5 (c)). The MoO3 recovered after the 5 h HDO reaction at 623 K and 4.4 MPa (Figure 4.5 (d)) had a surface composition of 52 mol.% MoO3 (BEs = 233.3 and 236.4 eV) and 48 mol.% Moδ+ species with a binding energy higher than that of Mo5+ (BEs = 232.4 and 235.2 eV), but lower than Mo6+. This suggested a charge of 5 < δ < 6 for the Mo species on the surface and is consistent with the observations made from XRD, TPR, and TPO.  66  a  Intensity  b  c  d  244  242  240  238  236  234  232  230  228  226  224  222  Binding Energy (eV)  Figure 4.5. Mo (3d) XPS spectra for: (a) – fresh MoS2; (b) – fresh MoO2; (c) – fresh MoO3; (d) – used MoO3.  The fresh MoO3 displayed an O/Mo atomic ratio of 3, which decreased to 2.7 following the 5 h reaction at 623 K and 4.4 MPa (Table 4.6). These results also indicate that the MoO3 was partially reduced to an oxidation state between Mo5+ and Mo6+. Molybdenum dioxide was not detected in the Mo (3d) spectrum of the used MoO3. This was because XPS is a surface technique and the used catalyst surface was likely oxidized following reaction and recovery from the reactor.  67  Table 4.6. XPS analysis of molybdenum catalysts.  Sample  Binding Energy (eV)  Composition (mol.%)  X/Mo  Mo 3d5/2  MoP  (atom ratio)  Moδ +  Mo4+  Mo5+  Mo6+  Moδ +  Mo4+  Mo5+  Mo6+  228.4a,  -  -  -  54a  -  -  -  1.2  232.3b  46b  MoS2  -  229.9  -  -  -  100  -  -  2.0  MoO2  -  230.1  231.3  233.3  -  45  30  25  2.1  MoO3  -  -  -  233.1  -  -  -  100  3.0  Used  232.4b  -  -  233.3  48b  -  -  52  2.7  MoO3 0 < δ < 1. 5 < δ < 6. X - was either P, S or O. a.  b.  In Figure 4.6 the spectra for the O 1s energy level is given for the fresh MoO2 (a), fresh MoO3 (b), and used MoO3 (c). The main oxygen spectral line at 531.1 eV was observed, as well as a shoulder at 532.1 eV. The low BE was attributed to the presence of Brønsted Mo-OH acidic group(s) on the surface of the fresh MoO2 and used MoO3 [98]. These acidic groups were absent over the fresh MoO3. This could be due to the fact that the formation of the Brønsted acid sites is due to the dissociation of H2 and H2O over CUS. The fresh MoO3 was not reduced and therefore did not possess these CUS for adsorption and dissociation.  68  Intensity  a  b  c  538  536  534  532  530  528  526  524  Binding Energy (eV)  Figure 4.6. XPS spectra of O 1s energy level for: (a) – fresh MoO2; (b) – fresh MoO3; (c) – used MoO3.  4.2.2  Catalyst Activity  The hydrodeoxygenation of 4-methylphenol was performed at 598 K (4.1 MPa), 623 K (4.4 MPa.), and 648 K (4.8 MPa) over MoS2, MoO2, MoO3, and MoP. Figure 4.7 shows the 4-methylphenol conversions as a function of time at each reaction temperature for all four catalysts. The reaction of 4-methylphenol followed 1st-order kinetics as shown in Figure 4.7.  69  a  b  c  4-Methylphenol Conversion (%)  100  80  60  40  20  0 0  60 120 180 240 300 0  60 120 180 240 300 0  60 120 180 240 300  Reaction Time (min)  Figure 4.7. 4-Methylphenol conversion over Mo catalysts at various reaction conditions. (■) MoO 3, (◊) MoP, (▲) MoO2, (○) MoS2. Solid lines represent first order kinetic fit, 𝒙 = (𝟏 − 𝒆−𝒌𝑪𝐜𝐚𝐭 𝒕 ). (a) – 598 K, 4.1 MPa; (b) – 623 K, 4.4 MPa; (c) – 648 K, 4.8 MPa.  The 1st-order kinetic parameters, k, are given in Table 4.7. A study by Yang et al. [20] found the 1st-order kinetic parameter for the HDO of 4-methylphenol over AHMderived MoS2 to be 2.2 mL.min-1.gcat-1 at 623 K and 2.8 MPa. Another study by Gevert et al. [23] found the kinetic parameters for the DDO and HYD pathways for the HDO of 4methylphenol to be 3.9 mL.min-1.gcat-1 and 0.8 mL.min-1.gcat-1, respectively, for the reaction at 573 K and 5 MPa over CoMoS/Al2O3. The values of these kinetic parameters are higher than those calculated for the catalysts used in the present study. However, the kinetic parameters are normalized in terms of the catalyst loading and they do not reflect the activity in terms of the number of active sites. The MoO2 and MoS2 had similar conversions at the reaction temperatures examined. Overall, the conversion was lowest  70  over the MoO2 and MoS2. The MoP displayed high conversions at all reaction temperatures and reached 98% conversion after 5 h at 648 K (4.8 MPa). The MoO3 displayed the highest conversion among the catalysts tested at each reaction temperature. Complete conversion over the MoO3 catalyst occurred after 2.3 h at 623 K (4.4 MPa), and 1.7 h at 648 K (4.8 MPa).  Table 4.7. Kinetic parameters, Initial TOFs, and activation energies for the decomposition of 4methylphenol over Mo catalysts at various reaction conditions.  Sample  k (mL.min-1.gcat-1) 598 K  623 K  648 K  MoO3  0.40  1.84  2.44  MoO2  0.12  0.34  MoS2  0.07  MoP  0.25  a.  Initial TOFa (s-1) 598 K  Ea (kJ/mol)  623 K  648 K  0.19  0.90  1.25  122 ± 2  0.65  0.37  1.10  2.14  113 ± 1  0.18  0.41  0.51  1.30  2.92  110 ± 1  0.82  1.13  2.29  7.10  10.45  98 ± 1  TOF calculated from active metal site data obtained from CO chemisorption.  The higher conversions over the MoO3 catalyst were a result of small crystallites and the creation of CUS that expose the metal and result in high CO uptakes (Table 4.2). The CUS possess a net positive charge that leads to strong acidic properties of the catalyst. The sites have been shown to be active for the hydrodeoxygenation of oxygenated compounds by dissociating H2 for hydrogenolysis or hydrogenation reactions [21]. The vacancy sites also dissociate H2 and H2O to produce Brønsted acid centers, leading to the high n-PA adsorption of the used MoO3 in comparison to the other catalysts. The Brønsted acid sites have also been shown to be active for HDO reactions  71  and are active centers for hydrogenolysis and/or aromatic ring saturation [23-25, 32, 33, 36, 37]. The CO chemisorption data were used to estimate the initial TOFs for the 4methylphenol decomposition over MoS2, MoO2, MoO3, and MoP. The initial TOF for each catalyst at the three reaction conditions are reported in Table 4.7 together with the apparent activation energies, Ea. The activation energy was calculated using the Arrhenius Equation given in Eq. 4.1 by plotting the natural lognormal (ln) of the kinetic parameter versus the inverse of the reaction temperature.  𝑘4MP = 𝐴e−𝐸𝑎/(𝑅𝑇)  (4.1)  where A is the pre-exponential factor, R is the Universal gas constant, and T is the temperature. The initial HDO TOFs decreased in the order: MoP > MoS2 > MoO2 > MoO3. Similarly, the activation energies were found to increase in the order of MoP < MoS2 < MoO2 < MoO3. The high initial TOF of the MoP was likely due to the electronic properties of this catalyst. The Mo (3d) of the MoP had the highest electron density (lowest BE, Table 4.6) among all catalysts tested. Since the lowest unoccupied molecular orbital (LUMO) of C-O is antibonding, surfaces that are able to transfer electron density to this orbital facilitate the dissociation of the C-O bond. Therefore, the increased electron density of the Mo may account for the higher initial TOF experienced over the MoP catalyst. The electron density of the catalysts decreased in the order MoP > MoS 2 > MoO2 > MoO3, which is in agreement with initial TOF and activation energy trends among these catalysts. As a comparison, previous work found the activation energy for 72  the HDO of 4-methylphenol to be 142 kJ/mol over CoMoS/Al2O3 [26], 156 kJ/mol over CoMoS/Al2O3, and 134.7 kJ/mol over AHM-derived MoS2 [20]. These activation energies are much higher than that observed over MoP for the HDO of 4-methylphenol at 98 kJ/mol. This demonstrates the superior activity of MoP compared to traditional metal sulfides.  4.2.3  Product Distribution  Several products were detected during the HDO of 4-methylphenol. Toluene was the main reaction product observed over MoO2, MoS2, and MoO3 catalysts. Hydrogenation products, such as methylcyclohexane, 1-methylcyclohexene, and 4-methylcyclohexene were also produced. Isomerization (ISOM) products, such as ethylidenecyclopentane, ethylcyclopentane, 1,3-dimethylcyclopentane, and 4,4-dimethylcyclopentene were also detected. These products suggest that ring opening reactions occurred. At high temperatures, very small amounts of cracked products, such as dimethylbenzene and methylcyclopentane were also observed, indicative of CH4 formation. These products were also referred to as “isomerization” products. All products detected were oxygen free, thus confirming that HDO was successful. The product distribution from the HDO of 4-methylphenol over MoO3 was found to be a strong function of time and temperature (Figure 4.8). As the temperature increased the degree of hydrogenation and cracking increased over MoO3. Although nearly 100% conversion of 4-methylpenol was reached at 623 K and 648 K, consecutive hydrogenation and isomerization reactions of methylcyclohexenes continued. The hydrogenation reactions increased as the temperature increased. These partially saturated 73  products were consumed and underwent isomerization to pentane derivatives, indicative of the high catalyst acidity. After 1.7 h at 648 K, nearly 80% of the hydrogenated products observed over MoO3 were fully saturated to methylcyclohexane.  a  b  c  MoO3  80 60 40 20 0  MoO2  80  Product Selectivity (%)  60 40 20 0  MoS2  80 60 40 20 0  MoP  80 60 40 20 0 0  70  140  210  280  0  70  140  210  280  0  70  140  210  280  Reaction Time (min)  Figure 4.8. 4-Methylphenol product selectivity over Mo catalysts at various reaction conditions. Toluene selectivity (■), hydrogenated product selectivity (◊), isomerization product selectivity (▲). (a) – 598 K, 4.1 MPa; (b) – 623 K, 4.4 MPa; (c) – 648 K, 4.8 MPa.  74  Both MoO2 and MoS2 catalysts displayed a similar degree of hydrogenolysis, but differed in their hydrogenation and isomerization selectivity - with MoO2 having a higher selectivity for complete hydrogenation to methylcyclohexane, while MoS2 had a high selectivity for partially saturated products - where 36% of the hydrogenated product was methylcyclohexane, 32% was 4-methylcyclohexene, and 32% was 1-methylcyclohexene after 5 h at 648 K. Furimsky et al. [24] found conflicting results over CoMo/Al2O3 oxide and sulfide, reporting that the sulfided catalysts produced a higher degree of hydrogenated products in comparison to the oxide catalyst. Gevert et al. [23] observed similar selectivities for the decomposition of 4-methylphenol over sulfided CoMo/Al2O3 at 573 K and 5 MPa as the MoS2 catalyst of the present Chapter. The MoO2 catalyst, with the higher surface acidity, also displayed a higher selectivity towards isomerization products. At 648 K the MoO2 product selectivity towards completely hydrogenated product was similar to that of the MoO3, which could indicate the formation of CUS over the MoO2 catalyst at the higher temperature. After the 5 h reaction at 648 K the selectivity towards hydrogenated products over MoO2 increased, where nearly 80% of the hydrogenated product was methylcyclohexane as opposed to only 38% and 16% methylcyclohexane at 623 K and 598 K, which could indicate the formation of CUS. Two parallel reactions for the decomposition of 4-methylphenol were observed over the phosphide catalyst to produce toluene and hydrogenated products. Of all catalysts tested, MoP displayed the highest selectivity towards hydrogenated products where 100% of the hydrogenated product was completely saturated to methylcyclohexane at all reaction temperatures. No methylcyclohexenes were detected over the MoP. Small amounts of the isomerization product 1,3-dimethylcyclopentane were detected at 623 and 75  648 K, indicative of the presence of an acidic phosphate phase. As the temperature over the MoP catalyst increased the selectivity towards toluene increased, thereby displaying the high activation energy for the formation of this product. Based on these results, hydrogenolysis to produce toluene and the coupled ring saturation/rapid dehydration to produce 4-methylcyclohexene were found to be the primary reactions for the HDO of 4-methylphenol. Toluene was found to be a stable product, while 4-methylcyclohexene was hydrogenated and cracked to produce saturated and isomerization products. This was determined by independent reactions of the products and intermediates over the catalysts. The results of which are contained in Appendix F. The reaction scheme is similar to that proposed by Laurent and Delmon [33] and suggests that the rate determining step (RDS) over the MoO3, MoO2, and MoS2 was the saturation of the 4-methylphenol ring to produce 4-methylcyclohexanol, which was then rapidly dehydrated to produce 4-methylcyclohexene. The proposed reaction pathway for the HDO of 4-methylphenol is given in Scheme 4.1. This scheme was determined by using the product species as reactants over the catalysts as described above and shown in Appendix F. Overall, the scheme for the HDO of 4-methylphenol over MoP is a simplified version of Scheme 4.1 and where 4methylcyclohexene is either rapidly converted to methylcyclohexane or isomerization product, 1,3-dimethylcyclopentane. Overall, no methylcyclohexenes were detected for the HDO of 4-methylphenol over MoP.  76  DDO  HYD  Scheme 4.1. Reaction mechanism for the HDO of 4-methylphenol.  Based on the reaction mechanism in Schemes 4.1, the kinetic parameters for the DDO  of  4-methylphenol  to  toluene  and  the  HYD  of  4-methylphenol  to  methylcyclohexane were determined. Based on these kinetic parameters the activation energies for the two pathways over all catalysts was determined (Table 4.8). For each case the activation energy for DDO was always greater than the activation energy for HYD. Overall, MoP displayed the lowest activation energy for both pathways followed by MoS2, MoO2, and MoxOy. As a comparison, previous work by Li et al. [99] found that the activation energies, for a similar model compound, 1-naphthol, over sulfided NiMo/Al2O3, to be 139 kJ/mol for the DDO path and 100 kJ/mol for the HYD path. These values are comparable to the activation energies calculated for the MoS2 catalyst.  77  Table 4.8. Activation energies for the DDO and HYD of 4-methylphenol.  Catalyst  4.2.4  EDDO  EHYD  (kJ/mol)  (kJ/mol)  MoxOy  145 ± 16  140 ± 11  MoO2  142 ± 13  137 ± 8  MoS2  123 ± 11  96 ± 8  MoP  106 ± 10  85 ± 7  Catalyst Active Sites  Coordinatively unsaturated sites on the oxide and sulfided catalysts display Lewis acid behavior and these sites (CUS) catalyze the reactions in hydroprocessing. On MoS2, coordinatively unsaturated sites are located at the edge planes and are believed to be the sites for catalytic hydrogenolysis reactions, while the rim planes (top edge planes) are active for hydrogenation [100, 101]. In contrast to the rim–edge model, Hensen et al. [102] reported that increasing MoS2 layer stacking increases hydrogenation because of a less hampered planar adsorption of reactants. Bunch and Ozkan [32] suggest that hydrogenation sites are associated with metallic Lewis acid sites, whereas hydrogenolysis sites are believed to be Brønsted acid centers associated with the adsorption and dissociation of H2S, or similarly H2O, since it is likely that OH- groups operate in a similar manner to SH- groups. The O2-, OH- and H+ groups present on the surface of oxidic catalysts exhibit Brønsted acid character and are formed by dissociative adsorption of H2O on the oxygen vacancies, or similarly S2-, and SH- groups over sulfided catalysts. The dissociation of H2 also converts O2- or S2- groups into OH- or SH- groups [34].  78  In the present study, it is suggested that the H+ involved in nucleophilic substitution reaction for the hydrogenolysis of C-O is primarily associated with Brønsted acid sites of the catalyst generated by the adsorption and dissociation of H2O or H2 to produce OH- or SH- groups on the oxide or sulfide catalyst, respectively. The OH- or SHgroups also provide Brønsted acidity for acid-catalyzed reactions. Therefore, the MoO3 catalyst displaying the highest Brønsted acidity, displays the highest hydrodeoxygenation (per unit mass of catalyst). The primary reaction for the production of 4methylcyclohexene from 4-methylcyclohexanol may be by an elimination reaction, which causes the dehydration of the protonated alcohol under acidic conditions to produce a π bond. Senol et al. [34] suggest that hydrogenation reactions occur on CUS and the degree of saturation of the CUS for the hydrogenation reactions are different from those for the hydrogenolysis reaction. Therefore, the CUS for hydrogenolysis reactions are more electrophilic (i.e. more positively charged) than those for hydrogenation reactions [32, 33, 65, 66]. Thus, in the present study the hydrogenation of 4-methylphenol to 4methylcyclohexanol is limited over the MoO3, MoO2, and MoS2 catalysts due to the abundance of very electrophilic CUS that cause a high selectivity towards hydrogenolysis products. The source of the acid sites over the MoP catalyst is likely a consequence of the incomplete reduction of phosphate species, which can result in Brønsted acidity that carries out protonation for hydrogenation reactions (PO-H). However, the measured surface acidity of the phosphide catalyst was low. Therefore, any acidic sites present on the phosphide catalyst were less electrophilic than those of any other catalyst, which accounted for its high hydrogenation abilities. The high selectivity towards 79  hydrogenation observed over MoP was also a consequence of the noble-metal like properties of the transition metal phosphide [103]. Thus, MoP displayed bifunctional acidic and metallic properties with strong hydrogenating capabilities similar to Pd and Pt on acidic supports [22].  4.3  Conclusions  The results of the present chapter demonstrate that unsupported low surface area MoO3, MoO2, MoS2, and MoP catalysts are active for the hydrodeoxygenation of 4methylphenol. The catalysts were stable at the reaction conditions, with the exception of MoO3, which underwent reduction to a mixed oxide containing Mo4O11, MoO2, and Mo phases. This partially reduced Mo oxide was found to have high activity for the decomposition of 4-methylphenol due to the formation of Lewis and Brønsted surface acidity. Both MoO2 and MoS2 displayed similar conversions, selectivities, and activation energies for the decomposition of 4-methylphenol. The initial TOF was found to be highest over the MoP catalyst, which also displayed the lowest activation energy and highest selectivity towards completely hydrogenated products. Therefore, potential lies in producing hydrogenated O free diesel and/or gasoline species with this catalyst. However, the conversion observed over MoP was limited by its low surface area and low CO uptake. Thus, methods to improve the MoP properties should be investigated.  80  Chapter 5 The Preparation of High Surface Area MoP for HDO2 5.1  Introduction  Supported metal sulfide catalysts such as MoS2/Al2O3 promoted with Ni [20, 32], as used in the HDS of petroleum oils, are also potential catalysts for the HDO of pyrolysis oils. However, because pyrolysis oils are S free, a sulfiding agent such as CS2 or H2S must be fed to the HDO reactor to maintain adequate sulfidation of the catalyst, and this is undesirable [19]. Hence, there is an interest in non-sulfided catalysts for the HDO of pyrolysis oil. Catalysts such as metal phosphides, that are active and selective for HDS, are also candidates for HDO [39-41]. Studies have shown that MoP/SiO2 has 4 times the activity (on a mass basis) of MoS2/Al2O3 for the HDS of thiophene [42]. Stinner et al. [43] showed that MoP has a 6 times higher TOF for the HDN of orthopropylaniline than MoS2/Al2O3, based on geometric estimates of surface site density. Several other studies  Work in this chapter was reported previously in: V.M.L. Whiffen, K.J. Smith and S. Straus, “The Influence of Citric Acid on the Synthesis and Activity of High Surface Area MoP for the Hydrodeoxygenation of 4-Methylphenol,” Applied Catalysis A, 419-420, 2012, pp. 111-125 and V.M.L. Whiffen, and K.J. Smith, “The Effect of Calcination Temperature on the Properties of MoP for the Hydrodeoxygenation of 4-Methylphenol,” Nanocatalysis for Fuels and Chemicals – ACS Symposium Series, Vol. 1092; Ed. A. Dalai; Chapter 5; Washington, DC, 2012, pp. 61-73. 2  81  have also reported that metal phosphides are more active and selective than sulfided metals for HDS [39, 44, 45] and HDN [46]. Chapter 4 demonstrated that unsupported, low surface area MoP has a lower activation energy and higher conversion for the HDO of 4-methylphenol compared to low surface area MoS2, but the MoP activity was limited by low metal dispersion. Therefore, unsupported MoP is a potential candidate for slurry phase HDO of pyrolysis oil, provided the MoP surface area can be increased. Wang and Smith [60] have shown that both the surface area and CO uptake of MoP can be increased from 5 m2/g and < 1 μmol/g to 139 m2/g and 42.4 μmol/g, respectively, by preparing the catalysts in the presence of CA. The increase in MoP surface area corresponded to an increase in 4,6-dimethyldibenzothiophene HDS conversion from 54.5% to 74.7% [60]. Similarly, hydrazine decomposition was found to increase from 55% to 85% for MoP prepared without and with CA, corresponding to MoP surface areas of 8.2 and 122 m2/g, respectively [61]. In the present chapter, MoP prepared using CA, has been investigated for the HDO of 4-methylphenol, a refractory model compound present in pyrolysis oils. The objective of the study was to understand the genesis of the high surface area, unsupported MoP catalysts using CA. The residual C content of the catalysts is shown to play an important role in determining the catalyst morphology and the HDO activity.  82  5.2  Results and Discussion  The prepared catalysts were characterized using the methods described in Chapter 3. In particular the characterization of the dried, calcined, and reduced MoP catalysts was performed to observe the citrate structure and C content of the precursors and reduced catalysts.  5.2.1  Characterization of the Dried Precursors  CHN analysis confirmed that the dried precursor, prepared without CA (Dry-MoP-noCA) was free of C and N, whereas the dried precursor prepared in the presence of CA (DryMoP-CA) contained 22.1 wt.% C and 8.3 wt.% N. The complete chemical composition of the Dry-MoP-CA sample is given in Table 5.1.  83  Table 5.1. Chemical properties of the dried precursor (Dry-MoP-CA), the calcined precursors (Cal-MoP-CA-ttt K), and reduced/passivated MoP catalysts (MoP-CA-ttt K, MoP-noCA).  Sample  Ca  Na  Ha  Mob  Pc  Od  C:Mo  (wt.%)  N:Mo  P:Mo  O:Mo  (molar ratio)  Dry-MoP-CA  22.1  8.3  3.2  19.4  6.8  40.2  9.0  3.0  1.1  12.3  Cal-MoP-CA-773 K  20.4  6.6  0.7  31.7  10.1  30.4  5.2  1.4  1.0  5.8  Cal-MoP-CA-823 K  15.4  4.0  0.2  --  --  --  --  --  1.2e  --  Cal-MoP-CA-873 K  12.6  3.6  0.4  --  --  --  --  --  1.2e  --  Cal-MoP-CA-923 K  9.1  2.8  0.3  --  --  --  --  --  1.2e  --  Cal-MoP-CA-973 K  6.2  1.4  0.1  44.0  14.6  33.8  1.1  0.2  1.0  4.6  MoP-CA-773 K  12.8  0.8  0.5  53.5  18.3  14.1  2.0  0.1  1.1  1.6  MoP-CA-823 K  9.4  0.5  0.6  --  --  --  --  --  --  --  MoP-CA-873 K  6.2  0.5  0.5  --  --  --  --  --  --  MoP-CA-923 K  5.1  0.3  0.4  --  --  --  --  --  --  --  MoP-CA-973 K  4.0  0.2  0.3  63.5  21.9  10.1  0.5  0.0  1.1  1.0  MoP-noCA  0.0  0.0  0.4  67.3  23.1  9.2  0.0  0.0  1.1  0.8  a.  average measured by CHN and GC combustion analysis. measured by ICP-AES. c. measured by colorimetric method. d. calculated by difference. e. measured by XPS. b.  84  To better understand the effect of the thermal treatment on the catalyst precursors, TGA of the individual precursor salts was completed. For AHP, two weight loss peaks at 433 and 472 K corresponded to the formation of H2O and NH3. At 472 K, AHP is transformed into polyphosphoric acid (PPA), HO(PO2OH)xH [104]. Above 773 K, PPA dehydrates to a phosphorus oxide such as P4O10, which can sublime [104]. The reaction stoichiometry is shown in Eq. (5.2) and (5.3). Three well defined mass loss peaks were observed for the AHM at 393, 549, and 662 K. The first mass loss corresponded to the production of (NH4)2O•2.5MoO3, NH3 and H2O as shown in Eq. (5.4) [105]. The final mass losses were attributed to the production of (NH4)2O•4MoO3, MoO3, NH3, and H2O as shown in Eq. (5.5) and (5.6) [105]. CA melts at 426 K and dehydrates to aconitic acid (C6H6O6) at 448 K and combusts at approximately 478 K (Eq. (5.7)) [106].  2(NH4)2HPO4  4NH3 + 2H3PO4  (5.2)  2H3PO4 3H2O + P2O5  (5.3)  2.5[(NH4)6Mo7O24•4H2O]  7[(NH4)2O•2.5MoO3] + NH3 + 10.5H2O  (5.4)  7[(NH4)2O•2.5MoO3]  4.375[(NH4)2O•4MoO3] + 5.25NH3 + 2.625H2O  (5.5)  4.375[(NH4)2O•4MoO3]  17.5MoO3 + 8.75NH3 + 4.375H2O  (5.6)  C6H8O7 + 4.5O2  6CO2 + 4H2O  (5.7)  TGA of the Dry-MoP-CA sample showed mass losses of 39.5, 48.6, 50.6, 54.6 to 59.2 wt.% at temperatures of 773, 823, 873, 923 to 973 K, respectively. Assuming the stoichiometry shown in equations (5.2)-(5.6), and the ratio of Mo:P and CA:Mo for the calcination of the Dry-MoP-CA precursor, the theoretical mass loss is 68.9 wt.%, 85  significantly greater than that obtained from the TGA analysis. Mixing of the precursor salts produced intermediates that did not allow removal of C and N as easily as from the AHP, AHM and CA alone, suggesting that more stable Mo-C, Mo-N, or C-N intermediates formed during the drying process. DRIFTS of the Dry-MoP-CA precursor is compared to CA and the Dry-MoPnoCA in Figure 5.1. The Dry-MoP-CA displayed a band at 1030 cm-1, characteristic of PO-Mo stretching and a band at 1420 cm-1, characteristic of PO43- (Figure 5.1) [61, 62].  CA 976 1030 1420  Intensity  1720  860  1200 1360 1620  Dry-MoP-CA 1100  890  734  Dry-MoP-noCA  2500  2000  1500  1000 -1  Wavenumber (cm )  Figure 5.1. DRIFTS spectra of CA and the dried precursors Dry-MoP-CA and Dry-MoP-noCA.  The band at 976 cm-1 in Figure 5.1 is associated with the symmetric stretching mode of P-OH [107]. Mo=O stretching was also observed at 860 cm-1[6]. Bands at 1620 and 1360 cm-1 are characteristic of asymmetric and symmetric stretching of Mo citrate 86  carboxylic ions (COO-), as shown in Figure 5.2 (b) (C6 and C1) for [Mo4(C6H5O7)2O11]4[6, 108]. These ions originate from the carboxylic acid present in CA (Figure 5.2 (a)), in which the carboxylic groups are coordinated to the Mo in a monodentate (C6) and bridged fashion (C1), respectively (Figure 5.2 (b)) [6, 62, 109-112]. The bands at 1720 and 1200 cm-1 are associated with the non-bonded and non-dissociated C=O stretch and C-O stretch of the carboxylic group of Mo citrate, as shown in Figure 5.2 (b) (C5) [6]. Dry-MoP-noCA displayed additional bands at 734 and 890 cm-1, characteristic of valence oscillations of Mo-O-Mo and Mo=O as observed in Figure 5.2 (b) for the bridged and terminal Mo species [6, 107, 111]. An additional band at 1100 cm-1 was assigned to PO43[113].  87  a  b H O C5  C6 C3  C1  C4  C5  N  C4  C  C1  C3 C2  C2  P Mo  C6  c  C2  C5 C4  C5 C3 C2  C6  C6 C3  C1 C4  H O N  C 5  C 4  C  C  C  P  1  3  M  C 6  c  o  d  e  C 5 C  C  C  C  4  3 C 2  1  6 Figure 5.2. Structure of (a) - citric acid; (b) - [Mo4(C6H5O7)O11]4-; (c) - [Mo4(C5H5O5CN)2O11]4-; (d) - Mo(C4H5O3CN)O4; (e) - Mo(C4H5O3CN)O2•HPO4. Structures were inferred from spectroscopic data and literature references and calculated by DFT.  88  Taken together, it can be concluded that the DRIFTS data for the Dry-MoP-CA sample showed that it was composed of Mo citrate and phosphomolybdate complexes. This conclusion is further supported by the  13  C CPMAS data of Figure 5.3, which  compares the spectra of CA and Dry-MoP-CA. In the CA spectrum, three shifts at 179.5, 176.3 and 174.5 ppm arise from the C6, C1, C5 (Figure 5.2 (a)) carboxylic acid carbons. The shift at 72.4 ppm was assigned to the central C of the CA molecule (C3 of Figure 5.2 (a)). The shifts at 44.2 and 43.8 ppm are characteristic of C4 and C2 CH2 moieties (Figure 5.2 (a)) [6, 114-116]. The broadening of the carboxylic, methylene, and alcohol regions in the Dry-MoP-CA spectrum were a consequence of both heat treatment and citrate formation. The downfield shifts of the bands for citrate formation were also observed in previous studies due to metal substitution effects [6, 114]. The peaks within the range of 200-174 ppm correspond to monodentate bonded carboxyl species (C6), the non-bonded carboxylic group (C5), and the carboxyl group bridged between two Mo atoms (C1) in Figure 5.2 (b), respectively. The peaks at 86.6 ppm and 53.4 ppm corresponded to the central C in the citrate structure (C3) and methylene (C4 and C2) in Figure 5.2 (b).  89  72.5 176.3 179.5 174.5  43.8 44.2  Intensity  CA 192.2  86.6 188.6  53.4  *  Dry-MoP-CA *  *  * 129.7  Cal-MoP-CA-773 K 206.7  250  43.8  200  150  100  50  0  Chemical Shift (ppm)  Figure 5.3. 13C CPMAS spectra of CA, the dried precursor Dry-MoP-CA, and the calcined precursor Cal-MoP-CA-773 K. (*) indicates spinning side bands.  5.2.2  Characterization of Calcined Precursors  The amount of C and N remaining after calcination of the Dry-MoP-CA precursor was dependent upon the calcination temperature, as shown in Table 5.1. The presence of N in the calcined precursors prepared with CA, suggests that the C in the CA samples binds to both Mo and N. Furthermore, the low surface area of all the calcined precursors (< 1 m2/g) suggested that the residual C present after calcination was non-porous and did not contribute to the precursor surface area. The elemental analysis via ICP-AES, colorimetric method, and XPS of the calcined precursors showed no P loss as the calcinations temperature increased from 773 to 973 K (Table 5.1).  90  The DRIFTS spectra of the Cal-MoP-CA and Cal-MoP-noCA precursors (Figure 5.4) all displayed two bands at 1116 and 1008 cm-1, which were assigned to P=O and PO-Mo stretching of P-Mo heteropoly compounds [61, 117, 118]. The band observed at 750 cm-1, corresponding to Mo-O-Mo bridge species of the valence oscillations of MoO2, was also present in all samples. The presence of Mo bound to a terminal oxygen (Mo=O) was confirmed by the band at 890 cm-1 [107, 111]. For the Cal-MoP-CA precursors, the broad band between 1500 and 1700 cm-1 with a peak at 1620 cm-1, associated with the asymmetric and symmetric stretching vibration of the monodentate bound carboxylic ion (COO-) (Mo citrate), was still present and decreased in relative intensity with increased calcination temperature. This implies that although more C was combusted as the calcination temperature increased, calcination did not completely oxidize all of the Mo citrate [6, 61, 62, 108, 109, 112]. There was no evidence of the non-bonded carboxylic group C=O and C-O in the Mo citrate (at 1720 and 1200 cm-1) (C5 in Figure 5.2 (b)), which suggests complete conversion of the carboxylic group. The band observed at 2340 cm-1 was associated with adsorbed CO2 [119], whereas the band at 2225 cm-1 was only present for those CA precursors calcined at 773, 823, and 873 K (Figure 5.4). This band was assigned to the conjugated CN bond of the nitrile group [120, 121] and is consistent with the CHN and TGA analysis that suggested residual N binds with C from the CA precursor salt to form a stable intermediate. It is likely that the non-bonded carboxylic group of Mo citrate (C5 in Figure 5.2 (b)) in the Dry-MoP-CA reacted with NH3 and dehydrated twice to form the nitrile complex, [Mo4(NC6H5O5)2O11]4-, as shown in Figure 5.2 (c). The presence of the CN group was confirmed by the 13C CPMAS spectra of CalMoP-CA-773 K shown in Figure 5.3, with the presence of a prominent peak at 129.7 ppm 91  characteristic of a metalated nitrile (Figure 5.2 (c)) [122]. The spectrum of Cal-MoP-CA773 K was identical in functionality to all other calcined precursors and the reduced MoP-CA-773 K. The peaks at 206.7 and 43.8 ppm were characteristic of CO and CH 2 moieties present in Mo citrate [6, 114-116]. A fit of the CO and CN lines (at 190 and 129 ppm) in the  13  C CPMAS spectra of the Cal-MoP-CA precursors (data not shown) using  DMFIT [83] showed that as the calcination temperature increased the CO content decreased by ca. 10% and the CN content increased by the same amount. This change occurred as the temperature was raised from 823 to 873 K, confirming the conversion of the non-bonded carboxylic group (C5 Figure 5.2 (b)) to a nitrile (C5 Figure 5.2 (c)).  2340  2225  1116 1008 890 1620  750  2225  Intensity  Cal-MoP-CA-773 K  2340  Cal-MoP-CA-823 K Cal-MoP-CA-873 K Cal-MoP-CA-923 K Cal-MoP-CA-973 K Cal-MoP-noCA 3500  3000  2500  2000  1500  1000  -1  Wavenumber (cm )  Figure 5.4. DRIFTS spectra of the calcined precursors (Cal-MoP-CA-ttt K) and the calcined MoPnoCA precursor.  92  The X-ray diffractograms of the calcined MoP precursors are shown in Figure 5.5. The Cal-MoP-noCA diffractogram confirmed the formation of Mo2P2O11 (PDF#00-0150610) with peaks at 2θ = 23.3o, and 27.3o. The Cal-MoP-CA precursors heated to 973 and 923 K showed diffraction peaks at 25.1o and 28.9o. These peaks are characteristic of MoOPO4 (PDF#00-034-1276). All the other calcined precursors were amorphous.  Cal-MoP-CA-773 K  Intensity  Cal-MoP-CA-823 K Cal-MoP-CA-873 K Cal-MoP-CA-923 K Cal-MoP-CA-973 K Cal-MoP-noCA  10  20  30  40  50  60  70  80  Angle, 2  Figure 5.5. XRD of the calcined MoP-CA precursors (Cal-MoP-CA-ttt K) and the calcined MoPnoCA precursor. MoOPO4 (○); Mo2P2O11 (■).  The XPS spectra (Figure 5.6 (a) and (b)) of the Cal-MoP-CA and the Cal-MoPnoCA precursors showed that the binding energy (BE) of both Mo and P increased as the precursor C content decreased.  93  234 eV 237.2 eV  233 eV  134.6 eV  a  236.4 eV  134.2 eV  Cal-MoP-CA-773 K  Cal-MoP-CA-773 K  Cal-MoP-CA-823 K  Intensity  Intensity  b  Cal-MoP-CA-823 K Cal-MoP-CA-873 K  Cal-MoP-CA-873 K Cal-MoP-CA-923 K  Cal-MoP-CA-923 K  Cal-MoP-CA-973 K  Cal-MoP-CA-973 K Cal-MoP-noCA  Cal-MoP-noCA 246  244  242  240  238  236  234  232  230  228  226  142  Binding Energy (eV)  140  138  136  134  132  Binding Energy (eV)  94  130  128  126  Cal-MoP-CA-773 K  c  Intensity  Cal-MoP-CA-973 K  Cal-MoP-CA-773 K  Cal-MoP-CA-773 K  c  Cal-MoP-noCA  244  Cal-MoP-CA-973 K  Intensity  Intensity  Cal-MoP-CA-973 K  Cal-MoP-noCA  Cal-MoP-noCA  a  244  242  240  b  238  236  234  232  230  228  226  142  140  138  136  134  132  130  128  Binding Energy (eV)  Binding Energy (eV)  Figure 5.6. XPS narrow scan spectra of the calcined MoP-CA precursors (Cal-MoP-CA-ttt K) and the calcined MoP-noCA precursor. (a) - Mo (3d); (b) - P (2p3/2); (c) - Mo (3d) fit; (d) - P (2p3/2) fit.  95  242  240  238  236  234  232  Binding Energy (eV)  d  230  228  226  The Mo (3d5/2) BE assigned to Mo6+ shifted from 233.0 eV to 234.0 eV with increased calcination temperature [48]. The P (2p3/2) BE has been reported to be 129.5 eV for the metal phosphides [44, 123], 135.2–135.6 eV for P2O5 [124, 125], and 133.5-133.9 eV for P5+ of PO43- [60]. Hence the BE of 134.4 eV for P 2p3/2 was assigned to a mixture containing both P2O5 and PO43-. To confirm the identity and proportion of Mo and P species present in the calcined samples, the XPS spectra were deconvoluted as summarized in Table 5.2 and shown in Figure 5.6 (c) and (d) for selected calcined samples. From Table 5.2 it can be observed that the Cal-MoP-CA precursors contained both Mo6+ (BE 233.0-234.0 eV) and Moδ+, where 5 < δ < 6, with a BE of 232.3 eV [48], as well as both P2O5 and PO43-. Overall, as the calcination temperature increased the proportion of Mo6+ to Moδ+ and P2O5 to PO43- increased. The trend could be associated with the crystallized molybdenum phosphate species identified by XRD of the Cal-MoPnoCA and the Cal-MoP-CA precursors calcined at 923 and 973 K. In these samples Mo is present as Mo2P2O11 (Cal-MoP-noCA) and MoOPO4 (Cal-MoP-CA -923 K and CalMoP-CA -973 K), which changes the fractional electron occupancy and electrostatic field around Mo resulting in a binding energy shift. As the calcination temperature of the CA samples increased and C was removed, the binding energy of Mo and P shifted towards that of the crystalline no-CA sample.  96  Table 5.2. Mo (3d) and P (2p) XPS analysis of calcined precursors (Cal-MoP-CA-ttt K).  BE (eV) Sample  Species Distribution  Mo 3d5/2  P 2p3/2  Mo (mol.%)  P (mol.%)  Mo6+  Moδ+  PO43-  P2O5  Mo6+ Moδ+ PO43- P2O5  Cal-MoP-CA-773 K  233.4  232.2  134.2  135.2  60  40  81  19  Cal-MoP-CA-823 K  233.6  232.4  134.2  135.2  61  39  77  23  Cal-MoP-CA-873 K  233.6  232.3  134.2  135.2  60  40  83  17  Cal-MoP-CA-923 K  233.9  232.2  134.2  135.2  68  32  75  25  Cal-MoP-CA-973 K  233.9  232.4  134.2  135.2  73  27  61  39  Cal-MoP-noCA  233.9  232.7  134.2  135.2  95  5  40  60  where 5 < δ < 6.  The TPR gas production profiles, measured by MS and presented in Figure 5.7 (a) and (b), show that the reduction of the calcined samples is complicated by the production of CH4, CO, CO2, NH3 (but no PH3) and H2O. (Note that H2O production is assumed although it was not quantified by the MS analysis). The results of Figure 5.7 (a) show that for the reduction of the Cal-MoP-CA-773 K sample, there were two reduction steps, one starting at approximately 800 K in which CO2, CH4 and NH3 were produced, and a second starting at approximately 1000 K in which CO2, CH4, CO and NH3 were produced. For the reduction of the Cal-MoP-CA-973 K sample (Figure 5.7 (b)), however, only the high temperature features were present, indicating that even after calcination to 973 K and reduction to 923 K, some of the CA remained in the catalyst sample. In the absence of CA (and therefore C and N in the TPR product gas) Cal-MoP-noCA consumed 25.6 mmol/g assuming the following stoichiometry: Mo2P2O11 + 11H2  11H2O + 2MoP, as determined by XRD. This indicated approximately 66.4% reduction, 97  as was determined by Abu and Smith [41] (62%) for MoP-noCA assuming MoO3·P2O5 stoichiometry.  a NH3 CO2  CO  CH4 PH3 450  600  750  900  1050  Temperature (K)  b  NH3  CO2  CO  CH4 PH3  450  600  750  900  1050  Temperature (K)  Figure 5.7. TPR-MS profiles of gas effluent from the calcined MoP precursors reduced in 9.5% H2 in Ar to 1173 K at 10 K/min. (a) - Cal-MoP-CA-773 K; (b) - Cal-MoP-CA-973 K.  98  5.2.3  MoP Characterization  The physical and chemical properties of the reduced and passivated MoP catalysts, prepared with and without CA at various calcination temperatures, are summarized in Table 5.1 and Table 5.3, while the morphology of the MoP-noCA and the MoP-CA-773 K, as determined by SEM, is compared in Figure 5.8. The MoP-noCA had a rough surface (Figure 5.8 (a)) while the MoP-CA-773 K displayed a smooth dense surface. Cheng et al. [61] found similar morphologies via SEM analysis of MoP prepared with and without CA. MoP prepared in the absence of CA was found to have clusters of platelets whereas in the presence of CA the MoP was shown to be smooth and larger in size [61]. The MoP-noCA was free of both C and N whereas the CA catalysts contained significant amounts of C that decreased as the calcination temperature increased. Even upon calcination to 973 K and reduction to 923 K, 4.0 wt.% C remained in the MoP-CA-973 K catalyst. Overall, calcination temperature did not affect the P:Mo ratio of the reduced MoP-CA catalysts. This was confirmed by the P:Mo ratios given in Table 5.1 and the absence of PH3 in the TPR-MS profiles (Figure 5.7).  99  Table 5.3. Physical and chemical properties of MoP prepared with and without CA.  Catalyst  MoP-CA-  MoP-CA- MoP-CA- MoP-CA- MoP-CA-  MoP-  773 K  823 K  873 K  923 K  973 K  noCA  SBET (m2/gcat)  136  112  75  63  53  8  dp(BET) (nm)  6  7  11  13  15  100  7.0  9.0  5.0  3.0  5.0  4.0  Pore Size (nm)  2  3  3  2  4  20  Sµpores (m2/gcat)  46  34  10  19  12  2  2.0  1.4  0.4  0.8  0.5  --  Pore Sizeµ (nm)  1.7  1.7  1.4  1.7  1.6  --  d101 (nm)  16  19  20  18  19  23  5 (0.5)  6 (0.2)  --  8 (3)  9 (1)  55 (5)  139  119  --  93  86  15  54  60  34  22  19  <1  Pore V (cm3/gcat) x102  Pore Vµ (cm3/gcat) x102  dTEMa (SD) (nm) STEMb (m2/g) CO Uptake (μmol/gMoP) a. b.  average TEM particle size. surface area calculated from TEM particle size.  100  Figure 5.8. SEM images of reduced/passivated MoP: (a) - MoP-noCA and (b) - MoP-CA-773 K.  Table 5.3 also shows that the addition of CA to the catalyst precursors significantly increased the surface area of the MoP and this was a consequence of a reduced MoP particle size. Compared to the MoP-noCA calcined at 773 K, the surface area of the MoP-CA, calcined at the same temperature, increased by a factor of ~10. As the calcination temperature of the CA catalysts increased, the surface area decreased. Appendix A.4 contains the equations used to calculate the BET surface area of the catalysts. The catalyst particle size was estimated from the BET surface area using the equation dp(BET) = 6/(SBET.ρ), assuming cubic or spherical geometry, where dp(BET) is the particle size calculated from the BET surface area, and ρ is the MoP density (7.5 g/cm3) [123]. The particle size of the MoP catalyst prepared without CA was estimated at 100 nm, significantly larger than the crystallite size calculated from the XRD data using Scherrer’s equation (23 nm) for the (101) plane of MoP (Figure 5.9). This suggests significant agglomeration of the metal phosphide crystallites prepared without CA. In contrast, the calculated particle size of MoP-CA-773 K was 6 nm, smaller than the crystallite size obtained from the line-broadening 101  analysis of the XRD data (16 nm). The difference between the calculated particle size and XRD data was likely a consequence of MoP-CA nanoparticles (< 5nm) not detected by XRD line broadening [126]. Similar differences were observed by Wang and Smith [60] for MoPCA characterized by XRD and TEM. In the present Chapter, TEM analysis showed a particle size that increased from 5 to 9 nm for the MoP-CA catalysts, as the calcination temperature increased from 773 to 973 K. The increased particle size was likely due to the loss in C that resulted in agglomeration of the crystallites as the calcination temperature increased. This suggests that the C from the CA acted as a structural promoter that limited agglomeration of the MoP crystallites.  16  20  24  28  32  36  CA - 773 K  40  CA - 823 K  Intensity  CA - 873 K CA - 923 K CA - 973 K  noCA 10  20  30  40  50  60  70  80  Angle, 2  Figure 5.9. XRD diffractograms of reduced/passivated MoP-CA and MoP-noCA catalysts. MoPO4(OH)3 (▼). 102  The CO uptake of the MoP-CA-773 K (corrected for the amount of C and N present in the reduced catalysts) increased to 54 μmol/gMoP compared to < 1 μmol/gMoP for the MoPnoCA. MoP-CA calcined at 823 K displayed the highest CO uptake. However, as the calcination temperature increased above 823 K the CO uptake of the MoP-CA decreased, consistent with agglomeration of the MoP crystallites. The X-ray diffractograms, given in Figure 5.9, were used to estimate the crystallite size of the catalysts reported in Table 5.3. The XRD of all catalysts showed the (101) reflection at 2θ = 43.0o, characteristic of MoP (PDF#065-6487). Although the MoP-CA catalysts contained C and N, reflections due to Mo2C and Mo2N were not present in the XRD patterns. Interestingly, MoP-CA-973 K and MoP-CA-923 K displayed additional diffraction peaks at 2θ = 19.6o, 22.5o, and 37.6o, characteristic of MoPO4(OH)3 (PDF#00-011-0333). MoP-CA-973 K and MoP-CA-923 K included crystalline MoOPO4 precursors following calcination and this phase produced both MoP and unreduced MoPO4(OH)3 following reduction. The formation of the unreduced phase could be a consequence of hydration from the production of water. The quantity of this species was less than 2%, as determined by the XRD peak intensity ratios. Table 5.4 reports the lattice parameters, ao and co, and the crystallite sizes for the (100) and (001) planes of MoP. The lattice parameters were calculated using the equations contained in Appendix A.4. The (101) crystallite size is reported in Table 5.3. The lattice parameters were calculated from the (100) and (101) planes. MoP has a hexagonal WC-structure, with lattice parameters ao = 0.322 nm and co = 0.319 nm and the calculated lattice parameters (Table 5.4) were in good agreement with these values. As the calcination temperature of the MoP-CA catalysts increased, the lattice 103  parameter spacing remained unchanged, suggesting that the residual C in the CA catalysts was not dissolved within the MoP lattice. The XRD profiles of the reduced/passivated MoPCA catalysts showed no evidence of a graphitic or carbide phase and hence it was concluded that the C present in the reduced catalysts was amorphous.  Table 5.4. MoP lattice parameters and crystallite sizes determined by XRD.  2θ (o) Catalyst  Lattice Parameters  Crystallite Size  (nm)  (nm)  Phase (101) (100) (001)  ao  co  d100  d001  MoP-CA-773 K  MoP  43.0  32.0  27.9  0.322  0.319  17  18  MoP-CA-823 K  MoP  43.1  32.1  27.9  0.321  0.319  20  22  MoP-CA-873 K  MoP  43.2  32.2  28.0  0.321  0.319  20  23  MoP-CA-923 K  MoP  43.1  32.1  28.0  0.321  0.319  19  22  MoP-CA-973 K  MoP  43.2  32.3  28.1  0.318  0.318  21  23  MoP-noCA  MoP  43.0  32.0  27.9  0.322  0.319  23  26  PDF#065-6487  MoP  43.0  32.0  27.9  0.322  0.319  --  --  TEM micrographs of the reduced and passivated catalysts are presented in Figure 5.10. In the case of the MoP catalyst prepared without CA (MoP–noCA, Figure 5.10 (a)), the particles aggregated to form large nanoclusters with a size of 50–60 nm, surrounded by a 20 nm passivation layer, similar to the morphology described by Wang and Smith [60], and Cheng et al. [61]. Separate 5 nm particles were clearly evident for the MoP-CA calcined at 823 K (Figure 5.10 (b)) as observed by Wang and Smith [60], and Cheng et al. [61]. These separate nanoparticles were also observed for MoP-CA-973 K (Figure 5.10 (c)). Most of the MoP-CA catalysts had particle sizes between 5-9 nm, although those calcined at higher 104  temperature did display larger agglomerated particles, but still retained the phase of dispersed nanoparticles. The lognormal average size of these dispersed nanoparticles is given in Table 5.3 along with their associated SD.  a  a  b  Figure 5.10. TEM images of reduced/passivated catalysts: (a) - MoP–noCA with passivation layer; (b) MoP-CA-823 K with insert of lattice fringe d-spacing estimated at 2.1 Å for the (100) plane of MoP; (c) MoP-CA-923 K, (d) nanoparticles encapsulated in an amorphous layer (MoP-CA-973 K); (e) agglomerated particles (MoP-CA-973 K) with insert of lattice fringe d-spacing of large particles estimated at 2.7 Å for the (110) plane of Mo oxide.  Three different morphologies of the reduced and passivated MoP-CA catalysts were identified from the TEM micrographs: (1) separate nanoparticles that were highly dispersed in an amorphous layer (Figure 5.10 (b)), (2) concentrated nanoparticles encapsulated in an 105  oxygen or carbonaceous passivation layer (Figure 5.10 (d)), and (3) those that agglomerated to form larger particles (50 nm) with an amorphous exterior resembling MoP-noCA (Figure 5.10 (e)). The lattice fringes of the reduced/passivated MoP-CA particles contained within the dispersed nanoparticle region had a d-spacing of ~2.1 Å (Figure 5.10 (b) insert) indicative of the (101) plane of MoP with a d-spacing of 2.1 Å (PDF#065-6487) [42]. The d-spacing associated with the larger exterior MoP-CA particles was ~2.7 Å (Figure 5.10 (e) insert) indicative of Mo oxide with a d-spacing of 2.7 Å for the (110) plane (PDF#047-1320). The small particles of MoP, identified via TEM, contributed to the high surface area of the MoPCA catalysts. This was shown by both the small pore volume and the reasonable agreement between the measured BET area and the area (STEM) calculated from the TEM particle size of the MoP-CA catalysts (Table 5.3).  5.2.4  Catalyst Activity and Product Distribution  Toluene, methylcyclohexane, and small amounts of 1,3-dimethylcyclopentane were detected as products of the 4-methylphenol HDO reaction. The error observed in the reactant conversion was approximately ± 7% as determined from repeat experiments (See Appendix B.3 for error analysis.). The reactant and product concentrations as a function of reaction time, shown in Figure 5.11 for the MoP-CA catalysts, suggested three important reactions: hydrogenolysis of 4-methylphenol to produce toluene, hydrogenation to methylcyclohexane, and isomerization to yield 1,3-dimethylcyclopentane, a consequence of MoP surface acidity due to unreduced phosphate species [41]. No methylcyclohexenes were detected over the MoP catalysts. All products were oxygen-free, confirming that HDO was effective. 106  Figure 5.11. Experimental and model concentration data versus reaction time: 4-methylphenol (□), toluene (◊), methylcyclohexane (○), 1,3dimethylcyclopentane (Δ), kinetic model fit (--).  107  Table 5.5 reports the 4-methylphenol conversions and product selectivities after 5 h reaction at 623 K and 4.4 MPa over the MoP catalysts. The conversions are net of the uncatalyzed thermal reaction, which resulted in 12% conversion after 5 h at the reaction conditions. Table 5.5 also reports the kinetic parameters and their error accounting for t0.025 values (refer to Appendix E.1). The low surface area catalyst prepared in the absence of CA had a higher degree of hydrogenated products (50%) compared to MoP-CA-773 K (37%). As the calcination temperature of the MoP-CA catalysts increased to 823 K, the conversion and selectivity toward 4-methylcyclohexane increased. This was due to aggregation of the MoP crystallites with increased calcination temperature above 823 K.  Table 5.5. Conversion, product selectivity, and kinetic model parameter estimates for the hydrodeoxygenation of 4-methylphenol over MoP catalysts after 5 h reaction at 623 K and 4.4 MPa.  Catalyst  Kinetic Parameter (mL.min-1.gMoP-1)  4MP X Product Selectivity (%) (%)  DDO  HYD  ISOM  k1  k2  k3  MoP-CA-773 K  58  60  37  3  0.52  0.04  0.32  0.04  0.03  0.04  MoP-CA-823 K  71  51  47  2  0.60  0.10  0.55  0.11  0.03  0.08  MoP-CA-873 K  56  56  42  2  0.48  0.06  0.35  0.06  0.01  0.06  MoP-CA-923 K  40  56  43  1  0.26  0.04  0.18  0.04  0.01  0.04  MoP-CA-973 K  38  57  41  2  0.26  0.04  0.18  0.04  4.7E-2  0.04  MoP-noCA  45  49  50  1  0.25  0.01  0.27  0.01  0.02  0.01  108  5.2.5  Reaction Kinetics  The 1st-order kinetic parameter for the conversion of 4-methylphenol at 623 K over the MoPCA catalysts was calculated accounting for the thermal reaction, catalyst concentration, catalyst C and N content, and the solution volume. MoP-noCA had a total 4-methylphenol decomposition kinetic parameter of 0.52 ± 0.005 mL.min-1.gMoP-1. The MoP-CA-773 K had a kinetic parameter of 0.89 ± 0.04 mL.min-1.gMoP-1 and this reached a maximum at 1.08 ± 0.12 mL.min-1.gMoP-1 for the MoP-CA-823 K catalyst (Figure 5.12). The precursors calcined at lower temperatures had higher C content and lower kinetic parameters. However, at higher calcination temperatures (above 873 K) both the C content and the 1st-order kinetic parameter decreased compared to MoP-CA-773 K and MoP-CA-823 K, due to agglomeration of the MoP crystallites as C content decreased at high temperature. Overall, a maximum was found to exist between C content and observed activity of the MoP-CA catalysts. The maximum kinetic parameter occurred at approximately 9.4 wt.% C and a maximum in CO uptake for the MoP-CA catalysts occurred for the same catalyst (MoP-CA-823 K). Sample calculations for these kinetic parameters and rates are given in Appendix A.6.  109  -1 -1 MoP-CA Kinetic Parameter (mL.min gMoP )  1.2  0.8  0.4  0.0 0  2  4  6  8  10  12  14  16  18  Carbon Content (wt%)  Figure 5.12. HDO 1st-order kinetic parameter versus carbon content of reduced MoP-CA catalysts.  On the basis of the product distribution results, hydrogenolysis or DDO to produce toluene and the coupled ring saturation/rapid dehydration to produce 4-methylcyclohexene, were assumed to be the primary reactions for the HDO of 4-methylphenol. Separate HDO experiments using toluene and 4-methylcyclohexene as reactants showed that toluene and methylcyclohexane were unreactive at the chosen conditions while 4-methylcyclohexene was hydrogenated or slowly isomerized. The data for these experiments is reported in Appendix F. Hence, during HDO of 4-methylphenol, toluene did not undergo further reaction whereas 4-methylcyclohexene was hydrogenated (HYD) to methylcyclohexane or isomerized to 1,3dimethylcyclopentane (ISOM). The reactions are similar to those proposed by Laurent and Delmon [33] as shown in Scheme 5.1. 110  CH3 k1 H2  HO  k2  CH3  HO 3H2  CH3  CH3  CH3 H2  -H2O H2  H3C  k3  CH3  Scheme 5.1. Proposed reaction scheme for the hydrodeoxygenation of 4-methylphenol over MoP catalysts.  The 1st-order kinetic parameters estimated for the individual reactions are given in Table 5.5 along with their associated error. The kinetic parameter k1 is associated with the path for direct hydrogenolysis of 4-methylphenol to produce toluene (i.e. direct deoxygenation or DDO). The rate determining step for the production of methylcyclohexane (k2) was the hydrogenation of 4-methylcyclohexene (i.e. HYD). The reaction pathway was modeled using a pseudo steady-state approximation for the two reaction intermediates, 4methylcyclohexanol  and  4-methylcyclohexene.  Similarly,  the  production  of  1,3-  dimethylcyclopentane (k3) from 4-methylcyclohexene was modeled by making a pseudo steady-state approximation for the hexanol and hexene intermediates. This resulted in three differential equations that governed the concentration of toluene, methylcyclohexane, and 1,3-dimethylcyclopentane. The kinetic parameters and their standard deviations were corrected by accounting for the thermal reaction, catalyst concentration, catalyst C and N content and the solution volume. The model fit to the experimental concentration versus time data is presented in 111  Figure 5.11. All MoP-CA catalysts had a k2:k1 (HYD:DDO) ratio of approximately 0.78:1. MoP-noCA had a higher selectivity toward HYD products with the k2:k1 ratio equal to 1.01:1. For the MoP-CA catalysts, the k1 and k2 paths leading to DDO and HYD reached a maximum with calcination temperature, as was found with the total consumption kinetic parameter k (Figure 5.12). All the catalysts displayed similar kinetic parameters for k3 – isomerization to produce the pentane derivative. However, this path was much higher over the MoP-CA-773 K, consistent with a higher surface acidity of this catalyst because of the low calcination temperature and incomplete conversion to MoP due to C consuming H2 as CH4. The error associated with the kinetic parameter values, k1 and k2, was < 10% whereas the error in k3 was high due to the low concentration of isomerization products and the limited number of data points for this species over the entire time range. The error analysis for the fit between the experimental and kinetic model reactant and product concentration data is given in Appendix E.6.  5.3 5.3.1  Discussion Catalyst Characterization  (a) Dried MoP-CA Precursor CA forms two types of anionic complexes with Mo depending on the pH of the precursor solution. A type (i) complex of the form [Mo4(C6H5O7)2O11]4- forms over a pH range of 2-3 with a Mo:CA ratio of 2:1, whereas a type (ii) complex of the form shown in Eq. (5.8) occurs over a pH range of 4-8 with a Mo:CA ratio of 1:1.  112  pMoO42- + qcitH3- + rH+ ↔ (MoO42-)p(citH3-)q(H+)r  (5.8)  where (p,q,r) is (1,1,1)4-, (1,1,2)3-, or (1,1,3)2- [111], and cit = C2O4. Although the Mo:CA ratio used in this study was 1:2, the type (i) complex is assumed to form since the measured pH of the precursor solution containing AHM, AHP, and CA was 3. The type (i) anion is tetranuclear and contains -MoO- chains. The tetrameric anion contains two types of Mo6+ ions, each with a terminal oxygen atom (Mo=O), bridged oxygen (Mo-O-Mo), and carboxyl oxygen of three different types, depending on their coordination to Mo and C atoms of the citrate ligand, as shown in Figure 5.2 (b) [6, 111, 127]. One C atom is non-bonded, one is coordinated to Mo in a monodentate fashion, and the last is bridged to Mo [6, 128]. The Mo binds to the CA by replacing the H atom in the carboxyl ligand. This Mo=O, and three carboxyl groups were observed in the DRIFTS and NMR spectra (Figure 5.1 and Figure 5.3) of the Dry-MoP-CA sample. The tetrameric citrate anion [Mo4(C6H5O7)2O11]4- and [NH4]+ most likely produced a complex of the form [NH4]4[Mo4(C6H5O7)2O11] during drying of the catalyst precursor [129]. If it is assumed that citrate formation requires nitrogen in the form of NH4+, a 3:1 C:N ratio is expected for the complex [NH4]4[Mo4(C6H5O7)2O11] and this corresponds to the measured C:N ratio reported in Table 5.1. DRIFTS also confirmed the presence of P-O-Mo, PO43-, and P-OH groups in the Dry-MoP-CA sample and therefore it was concluded that the dried CA precursor contained both Mo citrate and phosphomolybdate complexes. In the absence of CA it is likely that the formation of a Mo polymer occurred for the Dry-MoP-noCA sample.  113  (b) Calcined MoP-CA Precursors Following calcination of the dried precursor, NMR (Figure 5.3) and DRIFTS (Figure 5.4) analysis confirmed the presence of a C  N group and that all non-bound C in the dried citrate complex was converted to C  N . Therefore the metalated citrate was terminally bound to a nitrile group which resulted from dehydration of the non-bound carboxylic group (C5 Figure 5.2 (c)), NH3 + R−COOH  R−CONH2 + H2O  R−CN + H2O. As the calcination temperature of the precursors increased, N was liberated, breaking the anioniccationic bond and enabling aggregation of the Mo species that produced the low surface area Cal-MoP-CA precursors. The CHN data of Table 5.2 and the DRIFTS data for which the carboxylic C bridged to Mo at 1360 cm-1 (C1 Figure 5.2 (c)) was absent, agreeing with this assertion, implying that the tetrameric citrate, [NH4]4[Mo4(C6H5O7)O11], decomposed during calcination to a Mo citrate monomer containing five carbons Mo(C4H5O3CN)O4 (Figure 5.2 (d)). The Mo oxide species then reacted with phosphate groups to produce a Mo phosphate complex of the form Mo(C4H5O3CN)O2•HPO4 (Figure 5.2 (e)). The molar ratios of C:Mo, N:Mo, P:Mo of Mo(C4H5O3CN)O2•HPO4 are comparable to the molar ratios measured for Cal-MoP-CA-773 K in Table 5.1. The molar ratio of O:Mo is higher for the proposed complex compared to the measured values. This may be a consequence of the formation of a mixed oxide or hydroxyl. The proposed reactions of these transformations are given in equations (5.9) and (5.10) for the 2:1 CA:Mo and 1:1 P:Mo precursor salts.  56C6H8O7 + 4[(NH4)6Mo7O24•4H2O] + 28(NH4)2HPO4 + 189O2  7[NH4]4[Mo4(C6H5O7)2O11] + 52NH3 + 203H2O + 252CO2 + 28H3PO4 114  (5.9)  7[NH4]4[Mo4(C6H5O7)2O11] + 28H3PO4 + 3.5O2  (5.10)  14[Mo(C4H5O3CN)O2•HPO4] + 14NH3 + 70H2O + 14CO2 + 14MoOPO4  Interaction between the Mo and P was confirmed by the P-O-Mo and P-Mo bonding identified by DRIFTS (Figure 5.4). XRD of the precursors calcined at 923 and 973 K (CalMoP-CA-923 and Cal-MoP-CA-973 K) confirmed the formation of MoOPO4 (Figure 5.5). Additionally, as the calcination temperature increased the Mo(C4H5O3CN)O4 decomposed and C was removed. This was evidenced by the DRIFTS data (Figure 5.4) in which the intensity of the peak at 1620 cm-1, associated with C6 in Figure 5.2 (d), decreased. Thus, agglomeration of the Mo crystallites resulted due to the destruction of the metalated citronitrile complex at calcination temperatures above 873 K which, in turn, produced low surface area MoP-CA-(873, 923, and 973 K) following reduction.  (c) Reduced MoP-CA catalysts The reduced MoP-CA catalysts calcined at low temperatures also contained a small amount of nitrile from the Mo nitrile complex, Mo(C4H5O3CN)O2•HPO4, shown in Figure 5.2 (e), likely because of incomplete reduction. However, the reduced MoP-CA catalysts had significantly less C and N than the corresponding Cal-MoP-CA precursors (Table 5.1). The TPR-MS profile of the MoP-CA-773 K, presented in Figure 5.7, shows the simultaneous evolution of CH4, CO2, NH3 (and H2O) at approximately 850 K, likely corresponding to the decomposition and reduction of Mo(NC5H5O3)O2•HPO4. The decrease in surface area of the MoP-CA catalysts with increased calcination temperature is likely due to the removal of C 115  that resulted in increased agglomeration of the MoP crystallites. Both the pore and micropore size of the MoP-CA catalysts remained unchanged as the calcination temperature increased (Table 5.3), suggesting that the loss in surface area was not due to pore collapse of porous MoP or C at high calcination temperature, but rather the small pore volumes were a consequence of the interparticle volume between the MoP nano-crystallites. The high surface areas of the reduced MoP-CA catalysts were not a consequence of the residual C content of the reduced catalysts either, since there was good agreement between the SBET and STEM data (Table 5.3) of the MoP-CA catalysts. If indeed residual C contributed significantly to the surface area, excess CA would increase the surface area due to a high residual C content. However, a MoP-CA catalyst, prepared with a CA:Mo ratio of 4:1 and calcined at 773 K, had a surface area of only 5 m2/g and a C content of 11.0 wt.%. Similarly, Cheng et al. [61] reported a surface area of 13 m2/g for a MoP-4CA-773 K catalyst. Wang and Smith [78] also found no increase in surface area when the CA:metal ratio was varied from 1.5:1 to 3:1 for Ni2P catalysts prepared in the presence of CA. The importance of the CA and residual C in limiting agglomeration of the metal phosphide crystallites is further illustrated by the results of Ibeh et al. [62] who performed the calcination of WP-CA with W:CA = 1:4 in a flow of air. The addition of CA to WP increased the surface area from 4.1 to only 18 m 2/g [62]. Although the C content of this WP-CA was not reported, our own experiments show that calcination in flowing air is much more effective in removing the C than when the calcination is done without air flow. Consequently, preventing agglomeration of the WP crystallites would be less effective in air flow, resulting in relatively low surface area WP upon reduction [62]. 116  Two important effects of CA have been suggested in the literature [130]. Firstly, CA has the ability to form complexes with Mo and crystallization of the formed molybdenum citrate during drying is minimal, thereby inhibiting aggregation of the metal species. Secondly, during calcination, the space occupied by CA is converted to mesopores that do not collapse after elimination of the CA, and are also likely composed of amorphous metal phosphide [60]. However, the pore volumes determined in the present Chapter for the MoPCA catalysts were low, which instead suggests that the high surface area was due to the occurrence of MoP nanoparticles that did not agglomerate.  5.3.2  Titration of MoP Active Sites  The MoP-noCA catalysts were not compared to the MoP-CA catalysts on an active site basis due to the very low CO uptake of the MoP-noCA catalyst (< 1 µmol/g), also reported by Wang and Smith [60], and Abu and Smith [41]. The total metal site density of the MoP-CA catalysts was calculated using the TEM, XRD, and SBET data, by assuming spherical MoP particles and a surface area to mass ratio of 6/(dp.ρ), where ρ is the bulk density of MoP and dp is the particle size. The MoP active site area, estimated at 9.71 atoms/nm2, was calculated from the crystal structure of MoP and the average atom exposure of the low index planes [43, 123]. The site density, calculated from the SBET and TEM data was well correlated with the measured CO uptake data (Table 5.6) and therefore, it was assumed that the CO uptake was an appropriate measure of the number of active sites on the MoP catalyst. The details for the calculation of the metal site densities are given in Appendix A.7. The XRD data did not correlate well with the CO uptake data, likely due to the fact that small crystallites are not 117  detectable by XRD [126] and the fact that the small MoP particles were readily oxidized to form amorphous oxide during passivation. As others have shown, the calculated MoP site densities were much higher than the measured CO uptake data due to coverage of the CO adsorption sites by phosphorous [41, 123, 131].  Table 5.6. MoP site density based on BET, XRD, and TEM size versus CO uptake.  CO Uptake Catalyst  Metal Site  Metal Site Density Metal Site Density  Density  from XRD  from TEM  from BET  (μmol/gcat)  (μmol/gcat)  (µmol/gcat)  (μmol/gcat)  MoP-CA-773 K  46  2239  693  2239  MoP-CA-823 K  55  1804  614  1916  MoP-CA-873 K  32  1208  618  --  MoP-CA-923 K  21  1015  673  1498  MoP-CA-973 K  18  854  638  1385  MoP-noCA  <1  129  553  234  5.3.3  Reaction Sensitivity to Metal Phosphide Structure  Sawhill et al. [44] reported a correlation between thiophene HDS activity and O2 uptake of supported Ni2P, and used this relationship to calculate the TOF. Clark and Oyama [123] measured the HDS of dibenzothiophene, and the HDN of quinoline over supported MoP/Al2O3 and used the CO uptake to calculate TOFs. Oyama et al. [132] also used CO uptake to quantify the number of active sites over Ni2P/SiO2. These authors reported that the HDS of dibenzothiophene was not affected by the catalyst structure, but the HDN of 118  quinoline required synergy involving initial isomerization of the large ringed molecules followed by denitrogenation over Ni2P/SiO2 catalysts [132]. Similarly HDN over MoP/SiO2 and WP/SiO2 required a dual site involving both acid and nucleophilic centers [133]. The initial HDO consumption rates per gramMoP of the MoP-CA catalysts prepared at different calcination temperatures, calculated from the initial reactant concentration and the 4-methylphenol decomposition 1st-order kinetic parameter, are plotted as a function of CO uptake in Figure 5.13. The activity correlated with the CO uptake and from the slope of the linear regression plot, an initial TOF of 0.079 ± 0.012 s-1 was calculated, assuming that one CO molecule adsorbs on one active site of the MoP catalyst [42]. These results show that the hydrodeoxygenation of 4-methylphenol over MoP-CA catalysts is not structure sensitive, at least over an MoP size range of 5-9 nm. Details for the calculation of the initial TOF are given in Appendix A.6.  119  -1  -1  Initial Production Rate (mmol.min .gMoP )  0.4  -1  TOFHDO = 0.079 ± 0.012 s -1 TOFDDO = 0.045 ± 0.011 s TOFHYD = 0.035 ± 0.009 s  0.3  -1  0.2  0.1  0.0 0  10  20  30  40  50  60  70  CO Uptake (mol/gMoP) Figure 5.13. Initial HDO rate of 4-methylphenol consumption, and toluene and methylcyclohexane production rates versus CO uptake of MoP-CA catalysts: HDO (■), DDO (□), HYD (○).  An alternative interpretation of the data in Figure 5.13 is that the increased CO uptake is not due to decreased particle size but rather an increase in unsaturated Mo sites that bind CO. However, the absence of PH3 during the reduction of the Cal-MoP-CA precursors and the fact that all the reduced catalysts had the same P:Mo ratio, suggests that differences in Mo coordination among the catalysts calcined at different temperatures was not important. Furthermore, one would expect that if P loss occurred, it would increase with increased calcination temperature. This in turn would result in a higher degree of CUS that would increase the CO uptake. However, MoP-CA-973 K, calcined at the highest temperature, had the lowest CO uptake. Hence it was concluded that the change in CO uptake reflects changes  120  in MoP particle size and hence, based on Figure 5.13, the hydrodeoxygenation of 4methylphenol over MoP-CA is not structure sensitive.  5.3.4  Product Distribution  The initial DDO and HYD reaction rates, calculated from the DDO (k1) and HYD (k2) 1storder kinetic parameters and the initial reactant concentration, were also found to correlate with the CO uptake of the MoP-CA catalysts, as shown in Figure 5.13. This suggests that DDO and HYD occur on the same site that chemisorbs CO, such as a CUS [63, 64]. The TOFDDO and TOFHYD, calculated from the correlations given in Figure 5.13, were 0.045 ± 0.011 s-1 for DDO and 0.035 ± 0.009 s-1 for HYD. No relationship was found to exist between the CO uptake and the rate of isomerization, suggesting that isomerization was a consequence of unreduced phosphate acidity. Senol et al. [34] have suggested that for MoS2 catalysts, hydrogenation reactions occur on CUS that are more saturated compared to sites for hydrogenolysis. The CUS for hydrogenolysis are more electrophilic (i.e., more positively charged) than those for hydrogenation [32, 33, 65, 66]. This suggests that HYD occurs on both highly reduced sites and those having a lower coordination number. The fraction of electrophilic sites is higher on Mo sulfides and oxides compared to MoP, and this causes their hydrogenation activity to be low compared to MoP [48]. The TOFHYD:TOFDDO ratio was 0.78:1 on the MoP-CA catalysts, corresponding to a higher selectivity for hydrogenation compared to MoS2 which was approximately 0.16:1 HYDselectivity:DDOselectivity at 61% conversion (T = 623 K, PH2 = 4.4  121  MPa) [48]. The TOFHYD:TOFDDO ratio was 1.06:1 for the MoP-noCA catalyst, likely a consequence of a more reduced catalyst and a higher ratio of weak electrophilic sites. MoP displays bifunctional properties with strong hydrogenating capabilities similar to Pd and Pt on acidic supports [22]. Appendix G contains data for the HDO of 4-methylphenol over Pd/SiO2. As observed from the data the MoP catalysts have strong hydrogenating properties. The source of the acid sites (PO-H) over the MoP catalyst is a consequence of the incomplete reduction of phosphate species and CUS that dissociate H2 and H2O [41]. Both of these sites can result in Brønsted acidity that carry out protonation for hydrogenation reactions [21]. The latter Brønsted acid sites have also been shown to be active centers for both hydrogenolysis and/or aromatic ring saturation [21, 24, 32, 36]. The primary reaction for the production of 4-methylcyclohexene from 4methylcyclohexanol was the dehydration of the protonated alcohol under acidic conditions to produce a π bond, which then underwent rapid hydrogenation to methylcyclohexane over MoP. The rate limiting reaction involves the hydrogenation of the initial 4-methylphenol aromatic ring. This hydrogenation pathway involves electron withdrawing character from the catalyst surface, in which the reactant is absorbed through the π system.  5.4  Conclusions  Addition of CA to the AHM and AHP solutions followed by calcination and reduction increased the surface area, which led to an increase in the number of active sites of the MoPCA catalyst compared to the MoP-noCA catalyst. However, 20.4 wt.% residual C was present in the Cal-MoP-CA precursor calcined at 773 K. The C, present as a citrate, acted as 122  a structural promoter of the calcined samples, limiting agglomeration of the metal crystallites during reduction, to produce highly dispersed MoP nanoparticles. A maximum initial hydrodeoxygenation rate and CO uptake was found for the MoP-CA catalyst containing 9.4 wt.% C. Calcination of MoP-CA at temperatures above 823 K resulted in crystallite agglomeration, which reduced the CO uptake and initial hydrodeoxygenation rate. The hydrodeoxygenation of 4-methylphenol was found to be structure insensitive over all MoPCA catalysts. The calculated initial TOF was approximately 0.079 s-1. The TOFHYD:TOFDDO ratio over MoP-noCA (1.06:1) was higher than MoP-CA (0.78:1). This suggested that MoPnoCA contained a higher ratio of weak electrophilic CUS, which suggested a more reduced metal phosphide is the MoP-noCA than MoP-CA.  123  Chapter 6 The Preparation of High Surface Area Ni2P for HDO3 6.1  Introduction  Previous work has shown that catalysts such as metal phosphides that are active and selective for HDS, are also candidates for HDO [39-41, 48, 51-53, 56]. In Chapter 4, unsupported low surface area MoP was reported to have a lower activation energy and higher conversion for the HDO of 4-methylphenol compared to unsupported low surface area MoS2. Addition of CA to the precursor salts used to prepare unsupported MoP, increased the catalyst surface area from 8 to 136 m2/gcat, decreased the particle size from 55 to 5 nm, and increased the CO uptake from <1 to 54 µmol/gMoP as reported in Chapter 5 . The improved properties were shown to be due to the formation of a Mo citrate that prevented agglomeration of the metal crystallites during the calcination and reduction phase of the catalyst preparation. The addition of CA increased the HDO conversion of 4-methylphenol from 45 to 58% at 623 K and 4.4 MPa.  Work in this chapter was previously reported in: V.M.L. Whiffen and K.J. Smith, “A Comparative Study of 4Methylphenol Hydrodeoxygenation Over High Surface Area MoP and Ni 2P,” Topics in Catalysis, 55, 2012, pp. 981-990 and V.M.L. Whiffen, and K.J. Smith, “The Effect of Calcination Temperature on the Properties and Hydrodeoxygenation Activity of Ni2P Catalysts Prepared Using Citric Acid,” Novel Materials for Catalysis and Fuels Processing – ACS Symposium Series (accepted November 2012). 3  124  Similar results were reported by Wang and Smith [60] for the preparation of unsupported Ni2P using CA. The surface area of Ni2P was increased from 2 to 216 m2/g and the CO uptake from <1 to 30.1 µmol/g by adding CA to the precursor salts used for Ni2P preparation. Furthermore, these authors showed that at 483 K and 3 MPa H2, the conversion of 4,6-dimethyldibenzothiophene increased from 74.7% for Ni2P prepared without CA to 87.7% for Ni2P prepared with CA. The HDS TOF of the Ni2P prepared with CA was twice that of the MoP. These results suggest that Ni2P prepared using CA is a potential candidate for the HDO of pyrolysis oil due to a higher activity for the HDS of 4,6dimethyldibenzothiophene compared to MoP. In addition, the lower cost of Ni compared to Mo is another incentive to use Ni based phosphide catalysts for HDO. In this Chapter, unsupported, high surface area Ni2P, prepared using CA, has been tested for the HDO of 4methylphenol, a refractory model compound present in pyrolysis oils. In particular, the stability of the Ni2P catalysts during the hydrodeoxygenation of 4-methylphenol is examined. The Ni2P catalysts were prepared using CA and the effect of calcination temperature on the properties and activity was examined. In particular the catalysts were calcined at temperatures of 773, 823, and 973 K as the greatest differences in the MoP-CA catalysts tested in Chapter 5 were observed at calcination temperatures of 823 and 973 K.  6.2  Results and Discussion  The prepared catalysts were characterized using the methods described in Chapter 3. In particular the characterization of the dried, calcined, and reduced Ni2P catalysts was  125  performed to observe the citrate structure and C content of the precursors and reduced catalysts.  6.2.1  Characterization of the Dried Precursors  CHN analysis confirmed that the dried precursor, prepared without CA (Dry-Ni2P-noCA) was free of C and N, whereas the dried precursor prepared in the presence of CA (Dry-Ni2PCA) contained 24.7 wt.% C and 5.2 wt.% N. This indicates that the Dry-Ni2P-CA precursor had a higher C content than the Dry-MoP-CA precursor prepared in Chapter 5, but a lower N content. The complete chemical composition of the Dry-Ni2P-CA sample is given in Table 6.1.  Table 6.1. Chemical properties of the dried precursor (Dry-Ni2P-CA), the calcined precursors (Cal-Ni2PCA-ttt K), and reduced/passivated Ni2P catalysts (Ni2P-CA-ttt K, Ni2P-noCA).  Sample  C:Ni  N:Ni  P:Ni  Dry-Ni2P-CA  10.2  2.1  1.0  Cal-Ni2P-CA-773 K  1.8  0.5  1.1  Cal-Ni2P-CA-823 K  0.8  0.4  1.1  Cal-Ni2P-CA-973 K  1.0  0.4  0.9  Ni2P-CA-773 K  0.6  0.0  2.3  Ni2P-CA-823 K  0.6  0.0  2.3  Ni2P-CA-973 K  0.6  0.0  2.3  Ni2P-noCA  0.0  0.0  2.1  126  To better understand the effect of the thermal treatment on the catalyst precursors, TGA of the individual precursor salts was completed. For AHP, two weight loss peaks at 433 and 472 K corresponded to the formation of H2O and NH3. Five well defined mass loss peaks were observed for the Ni nitrate decomposition at 353, 463, 496, 515, and 587 K. CA was completely combusted at 478 K. TGA of the Dry-Ni2P-CA sample showed mass losses of 52.0, 54.7, to 60.9 wt.% at temperatures of 773, 823, to 973 K, respectively. Overall the total mass loss difference between the calcination temperatures of 773 and 973 K was 8.9 wt.% for the Dry-Ni2P-CA precursor. As a comparison, the mass loss difference for the Dry-MoP-CA precursor calcined at 773 and 973 K was 19.7 wt.%. This indicates that the Dry-MoP-CA precursor forms a more stable citrate during calcination than the Dry-Ni2P-CA. The theoretical mass loss of the individual precursor salts for the Dry-Ni2P-CA was 76.2 wt.%, assuming the Ni:P and CA:Ni ratios used for preparation. The theoretical mass loss is significantly greater than that obtained from the TGA analysis. Therefore, mixing of the precursor salts produced intermediates that did not allow removal of C and N as easily as from the AHP, Ni nitrate and CA alone, suggesting that more Ni-C, Ni-N, or C-N intermediates formed during the drying process. These observations are similar to those observed for the synthesis of MoP using CA and discussed in Chapter 5. DRIFTS of the Dry-Ni2P-CA precursor is compared to CA and the Dry-Ni2P-noCA in Figure 6.1. The Dry-Ni2P-CA displayed bands at 1420 and 1100 cm-1, characteristic of PO43[61, 62, 113].  127  CA  Intensity  1720  Dry-Ni2P-CA  1620  1200 1100  1420  920 740  1356 1400 1048 1120  Dry-Ni2P-noCA 2500  826 920  2000  1500  740  1000 -1  Wavenumber (cm )  Figure 6.1. DRIFTS spectra of CA and the dried precursors Dry-Ni2P-CA and Dry-Ni2P-noCA.  The band at 1620 cm-1 is characteristic of asymmetric stretching of Ni citrate carboxylic ions (COO-) [6, 108]. As in the case of Mo citrate (Chapter 5), these ions originate from the carboxylic acid present in CA, in which the carboxylic groups are coordinated to Ni [6, 62, 109-112]. The bands at 1720 and 1200 cm-1 are associated with the non-bonded and non-dissociated C=O stretch and C-O stretch of the carboxylic group of Ni citrate [6]. DryNi2P-noCA displayed additional bands at 740 and 920 cm-1 that are characteristic of PO43and P-O-Ni [134, 135]. Additional peaks were observed in Cal-Ni2P-noCA at 1120 and 1048 cm-1 and were assigned to P=O [134, 136]. It can be concluded from the DRIFTS data that the Dry-Ni2P-CA sample was composed of Ni citrate and phosphomolybdate complexes. This conclusion is further 128  supported by the 13C CPMAS data of Figure 6.2, which displays the spectra of Dry-Ni2P-CA. The peak within the range of 200-174 ppm corresponds to the carboxyl group bridged between two Ni atoms. The peak at 53.4 ppm corresponds to the methylene species present in the metal citrate species [6, 114-116]. The broadening of the carboxylic, methylene, and alcohol regions in the Dry-Ni2P-CA spectrum were a consequence of both heat treatment and citrate formation. The downfield shifts of the bands for citrate formation were also observed in previous studies and in Chapter 5 due to metal substitution effects [6, 114].  177.2  Dry-Ni2P-CA  53.4  *  Intensity  *  125.6  300  250  *  *  Cal-Ni2P-CA-973 K  200  150  100  50  0  Chemical Shift (ppm)  Figure 6.2. 13C CPMAS spectra of the dried precursor Dry-Ni2P-CA, and the calcined precursor CalNi2P-CA-773 K. (*) indicates spinning side bands.  129  6.2.2  Characterization of Calcined Precursors  Calcination of the Dry-Ni2P-CA precursor significantly decreased the C and N content of the samples. Overall, calcination to 973 K caused small P loss. The C content (C:Ni ratio) was reduced from 10.2:1 to 1.8:1 following calcination of the dried precursor to 773 K. Additional calcination to 973 K reduced the C:Ni content to 1.0:1. In comparison, the C:Mo ratios of Cal-MoP-CA-773 K and Cal-MoP-CA-973 K were 5.2:1 and 1.1:1, respectively. Cal-Ni2P-CA-773 K had a N:Ni ratio of 0.5:1 whereas Cal-MoP-CA-773 K had a N:Mo ratio of 1.4:1. These results demonstrate that calcination of the Dry-Ni2P-CA precursor to 773 K removed nearly all the C and N present in the samples. Calcination to higher temperatures was required to remove the C and N from the Dry-MoP-CA precursor. DRIFTS of the Cal-Ni2P precursors are given in Figure 6.3. The bands observed at 1073 and 759 cm-1 were assigned to P=O and PO43-, respectively [134-136]. The presence of P-O-Ni was present by the band observed at 925 cm-1 [134]. For the Cal-Ni2P-CA precursors, the broad band between 1500 and 1700 cm-1 was associated with the asymmetric and symmetric stretching vibration of the monodentate bound carboxylic ion (COO-) (Ni citrate). Overall, the intensity of the citrate band was relatively the same over all Cal-Ni2P-CA samples. This is because calcination to 773 K removed nearly all the C present in the precursor. There was evidence of the non-bonded carboxylic C-OH in the Ni citrate (1200 cm-1) [6]. The band observed at 2340 cm-1 was associated with adsorbed CO2 [119]. Overall a band at 2225 cm-1 assigned to the conjugated CN bond of the nitrile group [120, 121] was absent in the Cal-Ni2P-CA precursors. This band was observed in the Cal-MoP-CA precursors. This indicates that the non-bonded carboxylic C-OH (1200 cm-1) was not 130  converted to a conjugated CN bond in the Cal-Ni2P-CA precursors. The absence of the CN group was confirmed by the 13C CPMAS spectra of Cal-Ni2P-CA-973 K shown in Figure 6.2, with the absence of a prominent peak at 129.7 ppm, characteristic of a metalated nitrile [122]. Instead a peak at 125.6 ppm was observed for the Cal-Ni2P-CA-973 K precursor which is characteristic of RHC=CHR stretching indicating the decomposition of citrate [137].  1200 1073 925 1578  2340  759  Cal-Ni2P-CA-773 K  Intensity  Cal-Ni2P-CA-823 K  Cal-Ni2P-CA-973 K  Cal-Ni2P-noCA 3500  3000  2500  2000  1500  1000  -1  Wavenumber (cm )  Figure 6.3. DRIFTS spectra of the calcined precursors (Cal-Ni2P-CA-ttt K) and the calcined Ni2P-noCA precursor.  The X-ray diffractograms of the calcined Ni2P precursors are shown in Figure 6.4. The Cal-Ni2P-CA-973 K diffractogram confirmed the formation of either Ni(H2PO4)2.2H2O (PDF#00-060-0336), NiP2O6 (PDF#04-010-2427), or Ni(PO3)2 (PDF# 00-028-0708) with 131  peaks at 2θ = 35.0o and 51.0o. The peaks observed on Cal-Ni2P-CA-773 and 823 K were identified as noise. The broad peaks show that the calcined precursors were mostly amorphous.  Intensity  Cal-Ni 2P-CA-773 K Cal-Ni 2P-CA-823 K Cal-Ni 2P-CA-973 K Cal-Ni 2P-noCA  10  20  30  40  50  60  70  80  Angle, 2   Figure 6.4. XRD of the calcined Ni2P-CA precursors (Cal-Ni2P-CA-ttt K) and the calcined Ni2P-noCA precursor.  6.2.3  Ni2P Characterization  The properties of the reduced and passivated Ni2P catalysts are summarized in Table 6.2. The Ni2P prepared in the absence of CA (Ni2P-noCA) was free of C. However, the catalysts prepared with CA contained a C:Ni ratio of 0.6:1 for all calcination temperatures. Chapter 5 reported that the calcination temperature used for the preparation of MoP-CA catalysts significantly affected their C content. The C:Mo ratio was high for MoP-CA-773 K at 2:1, whereas MoP-CA-973 K had a C:Mo of 0.5:1. This implies that C is more easily removed 132  from the Ni2P-CA-773 K catalysts at calcination temperatures below 973 K compared with that of the MoP-CA catalysts. The Ni:P ratio of the Ni2P-CA catalysts increased from 1:1 for the calcined Ni2P-CA-773 K and noCA precursors to 2.3:1 for the reduced and passivated Ni2P-CA catalysts and was 2.1:1 for the reduced and passivated Ni2P-noCA catalyst. The increase in metal content was due to P losses during reduction of the Ni2P calcined precursor, which led to PH3 generation as evidenced by the P sublimation on the sides of the quartz utube used for reduction following cool down. This P loss was not observed for the MoP catalysts prepared in Chapter 5. Table 6.2 also shows that the addition of CA to the catalyst precursors significantly increased the surface area of the reduced Ni2P catalysts. Compared to Ni2P-noCA calcined at 773 K, the surface area of the Ni2P-CA-773 K increased by a factor of ~10 due to the formation of a metal citrate [49]. As the calcination temperature of Ni2P-CA catalysts increased from 773 to 973 K, the surface area decreased from 101 to 51 m2/gcat. The decrease was caused by sintering of the Ni2P-CA catalysts at higher calcination temperatures that led to agglomeration of the Ni2P crystallites. Ni2P prepared in the absence of CA had a crystallite size of 57 nm estimated by Scherrer’s equation using the (210) plane of Ni2P, whereas its particle size from TEM imaging was determined to be 259 nm. This again indicates significant agglomeration of the metal crystallites in the Ni2P-noCA catalyst. Ni2P-CA-773 K had a crystallite size of 34 nm compared to 42 nm for the Ni2P-CA-823 K catalyst and 50 nm for the Ni2P-CA-973 K catalyst. Ni2P-noCA had a CO uptake of < 1 μmol/gNi2P compared to that of 20 μmol/gNi2Pfor the Ni2P-CA-773 K catalyst. The CO uptake of Ni2P-CA-773 K was also greater than both the Ni2P-CA-823 K and the Ni2P-CA-973 K catalysts that had CO 133  uptakes of 9 and 10 μmol/gNi2P, respectively. These results further indicate that nearly complete C removal and particle sintering occurred at calcination temperatures above 773 K, leading to inferior Ni2P-CA properties. Overall, MoP-CA catalysts displayed superior properties to Ni2P-CA catalysts that had higher surface areas and smaller particle sizes than the latter. This was due to the ability of the Cal-MoP-CA precursors to form more stable citrates than the Cal-Ni2P-CA precursors.  Table 6.2. Physical and chemical properties of Ni2P prepared with and without CA.  Ni2P  SBET  dXRD (hkl)  dTEM (SD)  CO Uptake  (m2/gcat)  (nm)  (nm)  (μmol/gNi2P)  CA-773 K  101  34 (210)  36 (11)  20  CA-823 K  75  42 (210)  54 (26)  10  CA-973 K  51  50 (210)  54 (15)  9  noCA  6  57 (210)  259 (44)  <1  The diffractograms of the Ni2P catalysts can be seen in Figure 6.5. The XRD of the Ni2P catalysts and the used Ni2P-CA-773 K showed the (210) reflection at 2θ = 47.6o, characteristic of Ni2P (PDF#00-003-0953). Although the Ni2P-CA catalysts contained C, reflections due to Ni3C were not present in the XRD patterns. Additional, low intensity diffraction peaks for the Ni2P-CA catalysts were displayed at 2θ = 44.8, 48.9, and 57.5o, characteristic of Ni12P5 (PDF#04-007-1003). The excess in Ni was also seen from the elemental analysis of the Ni:P ratio of the Ni2P-CA catalysts in Table 6.2. This result suggests that CA addition to the Ni2P catalysts leads to higher P loss [50]. The presence of Ni12P5 was not reported in previous work by Wang and Smith [78] and Yang et al. [40] who 134  reported phase pure Ni2P prepared with a P:Ni ratio of 0.5:1–3:1 and a CA:Ni ratio of 1:1– 3:1 [78]; and a P:Ni ratio of 1:1 with Triton (polymer surfactant)/ethylene glycol [40]. This difference may be due to minor differences in the thermal treatments of the samples and/or a consequence of the more sensitive XRD detector used in the present Chapter.  Ni2P-CA-773 K * *  *  *  *  *  *  *  *  Intesnity  Ni2P-CA-823 K  Ni2P-CA-973 K  Ni2P-noCA  10  20  30  40  50  60  70  80  Angle, 2  Figure 6.5. XRD diffractograms of reduced and passivated Ni2P catalysts collected using a Co Kα X-ray source. Ni12P5 (*).  TEM micrographs of the reduced and passivated Ni2P catalysts and used Ni2P-CA773 K are presented in Figure 6.6. Separate 36–55 nm particles were clearly evident for the Ni2P-CA catalysts (Figure 6.6 (a)–(c)). Significant agglomeration was found to occur over Ni2P-CA-823 K and Ni2P-CA-973 K that had crystallites sizes of 42 and 50 nm but particle sizes of 54 nm each. Some agglomeration was observed over Ni2P-CA that had a crystallite 135  size of 34 nm and a particle size of 36 nm. Ni2P-noCA had an average particle size of 259 nm (Figure 6.6 (d)). High resolution images of the Ni2P-CA-773 K catalyst further confirmed the formation of Ni2P with lattice d-spacings of 1.7 and 2.5 Ǻ for the (300) and (200) plane of Ni2P, respectively (Figure 6.6 (e)).  Figure 6.6. TEM images of reduced and passivated Ni2P catalysts: (a) Ni2P-CA-773 K; (b) Ni2P-CA-823 K; (c) Ni2P-CA-973 K; (d) Ni2P-noCA; (e) lattice fringe d-spacing of Ni2P-CA-773 K at 1.7 and 2.5 Å for the (300) and (200) plane of Ni2P. 136  6.2.4  Catalyst Activity and Product Distribution  Figure 6.7 displays the natural lognormal plot of the 4-methylphenol concentration versus time for the HDO reaction over the Ni2P catalysts at 623 K and 4.4 MPa. The data accounts for the thermal reaction of 12% conversion of the initial 4-methylphenol concentration following the 5 h reaction. From Figure 6.7 it can be observed that the data does not follow a first order kinetic trend and that deactivation of all Ni2P catalysts occurred during the HDO reaction. It was observed that CA addition did increase the conversion over the Ni2P-CA-773 K catalyst compared to Ni2P prepared without CA. Both Ni2P-CA-823 K and Ni2P-CA-973 K had similar activities indicating significant agglomeration of the Ni2P-CA catalyst at a calcination temperature of 823 K. Chapter 5 found an “optimal” calcination temperature of 823 K for the preparation of MoP-CA as this led to the highest rate of hydrodeoxygenation of 4-methylphenol. Temperatures below 823 K led to residual C in MoP-CA that blocked the active sites, while temperatures above 823 K destroyed the Mo citrate structure and led to sintering of the catalyst particles. Overall, the “optimum” calcination temperature for the preparation of Ni2P-CA over the temperature range tested was 773 K.  137  ln(4-Methylphenol Concentration)  -1.25  -1.50  -1.75  -2.00  -2.25  -2.50  -2.75 0  50  100  150  200  250  300  350  Time (min)  Figure 6.7. Lognormal plot of the 4-methylphenol concentration versus time for HDO reaction at 623 K and 4.4 MPa over Ni2P catalysts. Ni2P-CA-773 K (▼), Ni2P-CA-823 K (▲), Ni2P-CA-973 K (●), Ni2PnoCA (■), guideline shown for illustration (--).  Recent work by Li et al. [53], Zhao et al. [51], Bui et al. [55], and Cho et al. [54] observed minimal or no deactivation for the hydrodeoxygenation of pyrolysis oil model compounds over supported Ni2P catalysts. However, Li et al. [53] did encounter 29% deactivation over Ni2P/SiO2 following a 12 h reaction for the HDO of anisole. These results are contrary to that of the present work; however, this may be a consequence of the presence of a catalyst support or different reactor configurations used in those studies. In the present work, the unsupported catalysts were tested over a range of conversions in batch mode, whereas in fixed-bed studies, the reactor typically operates at a single pass conversion, with no recycle, determined by the chosen residence time. Coke precursors remain in the reactor 138  and have the time to form coke in the batch mode, whereas this is less likely in a fixed-bed reactor. In order to extract the kinetic parameters, the exponential decay law given in Eq. (3.3) was applied to the Ni2P concentration data. A plot of the Ni2P rate versus time is given in Figure 6.8.  0.4  -1  2  -1  HDO Rate (mmol.min .gNi P )  0.5  0.3  0.2  0.1  0.0 0  50  100  150  200  250  300  350  Time (min)  Figure 6.8. Hydrodeoxygenation rate versus time for the reaction at 623 K and 4.4 MPa over Ni 2P catalysts. Ni2P-CA-773 K (▼), Ni2P-CA-823 K (▲), Ni2P-CA-973 K (●), Ni2P-noCA (■), exponential decay law model fit (--).  The kinetic parameters (k and kd) can be found in Table 6.3 along with the initial rates and initial TOFs of the Ni2P catalysts. Please note that the kinetic parameter errors are given in scientific notation. Overall Ni2P-CA-773 K was 2.3 times more active than MoP-CA-773 139  K on a mass basis and 6 times more active on a site basis. Li et al. [53] also found Ni2P/SiO2 had ten times the TOF compared to MoP/SiO2 for the HDO of anisole. Similarly, Bui et al. [55] found that Ni2P had a higher TOF than MoP for the HDO of 2-methyltetrahydrofuran. Conversely, Zhao et al. [51] found that MoP/SiO2 had a higher TOF for the HDO of guaiacol compared to Ni2P/SiO2. All Ni2P-CA catalysts used in the present study had similar initial TOFs of 0.464 ± 0.006 s−1, normalized by CO uptake, indicating structure insensitivity for the HDO of 4-methylphenol over the Ni2P-CA catalysts (Figure 6.9). Structure insensitivity was also reported over MoP-CA catalysts for the HDO of 4-methylphenol in Chapter 5. The present result is not surprising because of the relatively large crystallite dimensions (34–50 nm) of the Ni2P-CA catalysts. The TOF of Ni2P-noCA was not calculated due to the low CO uptake measured over this catalyst (< 1µmol.g Ni2P).  Table 6.3. Initial rate, initial TOF, kinetic parameter, deactivation parameter, and C deposition following the 5 h reaction for the hydrodeoxygenation of 4-methylphenol over Ni2P catalysts at 623 K and 4.4 MPa.  Ni2P  Initial Rate  Initial TOF  (mmol.min−1.gNi2P −1)  (s−1)  kMP  kd  C  (mL.min−1.gNi2P−1) (mL.min−1.gNi2P-1) Deposition (wt.%)  CA-773 K  0.55 ± 1.7E-2  0.449 ± 1.8E-2  2.15 ± 9.0E-2  1.20 ± 4.9E-2  3.7  CA-823 K  0.28 ± 3.9E-2  0.469 ± 7.1E-2  1.07 ± 1.6E-1  1.29 ± 2.5E-2  2.6  CA-973 K  0.26 ± 2.0E-2  0.470 ± 3.7E-2  1.01 ± 6.8E-2  1.16 ± 9.2E-2  1.9  noCA  0.18 ± 8.2E-2  --  0.70 ± 2.3E-3  1.17 ± 5.0E-3  1.5  140  Initial TOFHDO = 0.464 ± 0.006 s  -1  -1  Initial Hydrodeoxygenation Rate (mmol.min .gNi2P )  0.8 -1  0.6  0.4  0.2  0.0 0  5  10  15  20  25  30  CO Uptake (mol/gNi P) 2  Figure 6.9. Initial rate of hydrodeoxygenation versus CO uptake over Ni 2P-CA catalysts.  All the Ni2P catalysts displayed deactivation and had a deactivation parameter in the range of 1.20 mL.min−1.gNi2P−1. This suggests that a similar mechanism of active site deactivation was present for all the Ni2P catalysts. Significant C deposition was found over all the Ni2P catalysts tested in this Chapter, as reported in Table 6.3. It was observed that the amount of C deposited on the Ni2P catalyst was correlated with the rate of reaction (k). It is suggested that the C deposition was due to the formation of isomerization product intermediates that polymerize and precipitate out of solution to form coke. The hydrodeoxygenation of 4-methylphenol over Ni2P proceeds through two pathways. The first pathway leads to the DDO product toluene. The second pathway involves the coupled ring saturation/rapid dehydration to produce 4-methylcyclohexene which is rapidly  hydrogenated  (HYD)  to  methylcyclohexane 141  [49].  Isomerization  of  4-  methylcyclohexene is also possible, however, isomerization products were not detected for the HDO of 4-methylphenol over Ni2P catalysts. Therefore, it is proposed that the isomerization intermediates are consumed to form coke in this case. The Ni2P DDO (toluene) selectivity versus time, reported in Figure 6.10 (a), was found to decrease slightly with reaction time. Ni2P-CA-823 K displayed the lowest selectivity towards DDO followed by Ni2P-CA-773 K and Ni2P-CA-973 K, whereas Ni2PnoCA had the highest selectivity towards DDO. The Ni2P HYD (methylcyclohexane) selectivity versus time, reported in Figure 6.10 (b), was found to increase with reaction time. Ni2P-CA-823 K displayed the highest degree of HYD followed by Ni2P-CA-773 K, Ni2PCA-973 K, and Ni2P-noCA. The increase in HYD and decrease in DDO selectivity versus time suggests either the further HYD of toluene to methylcyclohexane as a function of reaction time or the deactivation of the DDO active site. With the exception of Ni2P-CA-823 K, those catalysts displaying smaller particle sizes had a higher degree of hydrogenation. This variation in product selectivity with particle size may suggest structure sensitivity of the HYD and/or DDO routes over the Ni2P catalysts, even though the initial TOF for 4methylphenol consumption was not dependent on the particle size (Figure 6.9). The DDO selectivity remained relatively constant as a function of time over the MoP catalysts. The MoP-noCA had a lower DDO selectivity than the MoP-CA-773 K. The DDO selectivity as a function of time decreased marginally with time over the Ni2P catalysts and the Ni2P-CA-773 K had a lower DDO selectivity compared to the Ni2P-noCA. The HYD selectivity remained relatively constant as a function of time over the MoP catalysts. The MoP-noCA had a higher HYD selectivity than the MoP-CA-773 K that previously was ascribed to the residual C from 142  the CA that blocked the HYD active site. The HYD selectivity versus time over the Ni 2P catalysts was found to increase marginally with reaction time and Ni2P-CA-773 K had a higher degree of HYD than the Ni2P-noCA. These results indicate an opposite effect of CA addition on the product selectivity over MoP versus Ni2P. MoP-CA-773 K favored the DDO pathway while larger particles (MoP-noCA) favored the HYD pathway. Conversely, Ni2P-CA-773 K favored the HYD pathway while larger particles (Ni2P-noCA) favored the DDO pathway. Overall, MoP-noCA had a higher selectivity to HYD products compared to Ni2P-noCA. However, the addition of CA to Ni2P resulted in increased surface sites and a higher selectivity to HYD products compared to MoP-CA-773 K. This opposite behavior is explained by the fact that the C:M ratio of MoP-CA-773 K was 2:1, while over the Ni2P-CA-773 K it was 0.6:1, indicating a higher C density over high surface area MoP-CA-773 K catalyst. Thus, CA addition led to blockage of the HYD active sites of the MoP-CA-773 K but to a much lesser extent over the Ni2P-CA-773 K.  143  80  a  DDO Selectivity (%)  70  60  50  40 0  50  100  150  200  250  300  350  Time (min)  60  b  HYD Selectivity (%)  50  40  30  20 0  50  100  150  200  250  300  350  Time (min)  Figure 6.10. DDO product selectivity of toluene (a) and HYD product selectivity of methylcyclohexane (b) versus time for the reaction at 623 K and 4.4 MPa over Ni 2P catalysts. Ni2P-CA-773 K (▼), Ni2P-CA-823 K (▲), Ni2P-CA-973 K (●), Ni2P-noCA (■), guideline shown for illustration (--).  Previous work by Oyama et al. [138] found that the HDN of quinoline was structure sensitive over Ni2P/SiO2 catalysts. It was suggested that changes in the P levels on the 144  surface of Ni2P/SiO2 disrupted a dual site involving an acid site and a basic site. In the present work, changes in P levels were observed between the Ni2P catalysts prepared with and without CA. All Ni2P-CA catalysts had the same P content. Previous work by Whiffen and Smith [50] suggested that differences observed in the HYD selectivity over Ni2P-noCA and Ni2P-CA-773 K was due to the production of Ni12P5 in Ni2P-CA-773 K, resulting in a higher HYD selectivity compared to Ni2P-noCA, which was free of Ni12P5. This was based on previous work by Wang et al. [139] who found that Ni12P5/SiO2 has a higher C=C hydrogenation selectivity than Ni2P/SiO2 for the hydrogenation of cinnamaldehyde to hydrocinnamaldehyde.  6.3 6.3.1  Discussion Catalyst Characterization  The addition of CA to the precursor salt solutions used in the synthesis of MoP and Ni2P increased the surface area and CO uptake of the resulting metal phosphide catalysts [49, 78]. However, the physical properties such as surface area and particle size were found to be superior over the MoP-CA catalysts compared to the Ni2P-CA catalysts. The Ni2P-CA catalysts lost nearly all C at a calcination temperature of 773 K, thereby destroying the metal citrate that acted as a structural promoter to prevent agglomeration during reduction. The primary difference observed between the Ni and Mo citrate was the formation of a conjugated nitrile group. The nitrile group was present in the Cal-MoP-CA precursors calcined at temperatures below 873 K, and confirmed by elemental analysis, DRIFTS, and NMR data. The Cal-Ni2P-CA precursors had very low N:Ni contents indicating that nearly 145  all N was removed at 773 K. Similarly, the DRIFTS and NMR data confirmed the absence of the nitrile group in the Cal-Ni2P-CA precursors. Therefore, it is proposed that this nitrile group acted to stabilize the metal citrate in the MoP precursors from combusting during calcination. Both Ni2P-CA and MoP-CA were prepared using AHP ((NH4)2HPO4). Therefore, the N component of the metal citrate was not produced from AHP. The Ni in Ni2P-CA was supplied by Ni nitrate (Ni(NO3)2.6H2O) whereas Mo in MoP-CA was supplied by ((NH4)6Mo7O24.4H2O). The Ni source contained nitrate and the Mo source contained ammonium. This suggests that ammonium more readily forms a stable nitrile than the nitrate in the production and stabilization of the metal citrate. Thus, it may be beneficial to prepare Ni2P-CA using a Ni ammonium precursor.  6.3.2  Ni2P Deactivation  Chapters 5 and 6 have shown that both the high surface area MoP and Ni2P are active catalysts for the HDO of 4-methylphenol and that Ni2P has a higher activity than MoP for the reaction, both on a mass and site basis. The higher activity of Ni2P compared to MoP has also been reported by others for both HDS and HDO reactions [51, 53, 78]. Importantly, catalyst deactivation during the HDO of 4-methylphenol was observed over the Ni2P catalysts, whereas the MoP catalysts did not deactivate at the chosen reaction conditions (623 K and 4.4 MPa H2). Previous studies have shown that deactivation over metal phosphides can occur by P loss [140]. In addition, deactivation by C deposition and by oxidation of the catalysts may be possible. Deactivation of metal sulfide catalysts by oxidation during HDO 146  has been suggested previously [141]. In the present Chapter, the catalysts were likely very susceptible to deactivation by oxidation because of the use of a batch reactor that resulted in increased H2O concentration as the reaction proceeded. The data of Table 6.3 show that the rate of deactivation for both the Ni2P-noCA and the Ni2P-CA-773 K catalysts was approximately the same (~1.2 mL.min-1.gNi2P-1, Table 6.3), suggesting that the mechanism of deactivation was similar over the Ni2P-noCA and the Ni2P-CA-773 K catalysts. To address the source of the deactivation, the used catalysts (including MoP-CA-773 K from Chapter 5) were recovered from the reactor and their properties measured, including the M:P content, the amount of C deposited after reaction, the surface area, XRD crystallite size, TEM particle size, and CO uptake, as reported in Table 6.4. The data show that C deposition over the MoP-CA-773 K catalyst following the 5 h reaction was low (0.2 wt.%), corresponding to an absence of deactivation over this catalyst. Similarly, the high surface area and CO uptake, and small particle size of the MoP-CA-773 K catalyst were unchanged after use. The data of Table 6.4 also show that the used Ni2P-CA-773 K did not lose P during the reaction (the P:Ni ratio remained at 2.3:1), indicating that the deactivation observed over this catalyst was not due to P loss. However, the C deposition over the Ni2P-CA-773 K was high (3.7 wt.%) and both the surface area and CO uptake of the Ni2P-CA-773 K was reduced significantly to 13 m2/gcat and 6 μmol/gNi2P, respectively, following reaction. XRD of the used Ni2P-CA-773 K also indicated a small increase in crystallite size from 34 to 47 nm for the (210) reflection, suggesting that some sintering of the Ni2P crystallites also occurred. An increase in particle size was also observed from the TEM images and similar trends were observed for the used Ni2P-noCA. 147  Table 6.4. Physical and chemical properties of fresh and used MoP and Ni2P catalysts following 5 h hydrodeoxygenation reaction of 4-methylphenol at 623 K and 4.4 MPa.  Catalyst  M:P  C  SBET  dXRD  Ratio  deposited  (m2/gcat)  (hkl)  (wt.%)  dTEM (SD) CO Uptake (nm)  (μmol/gMxP)  (nm)  Fresh MoP-CA-773 K  1.1  -  136  16 (101)  5 (0.5)  54  Used MoP-CA-773 K  1.1  0.2  130  16 (101)  5 (1.2)  52a  Fresh Ni2P-CA-773 K  2.3  -  101  34 (210)  36 (11)  20  Used Ni2P-CA-773 K  2.3  3.7  13  47 (210)  61 (18)  6a  Fresh Ni2P-noCA  2.1  -  6  57 (210)  259 (44)  <1  Used Ni2P-noCA  2.1  1.5  --  --  --  <1a  Fresh Passivated-Ni2P-CA  2.3  -  101  34 (210)  36 (11)  <1b  Used Passivated-Ni2P-CA  --  5.1  --  --  --  <1a  a. b.  CO uptakes measured following pretreatment in H2 to 773 K. CO uptakes measured without H2 pretreatment to 773 K.  The effect of pressure on the HDO of 4-methylphenol was examined over the Ni2PCA in an effort to reduce the catalyst deactivation by C deposition. Ni2P-CA was tested at 623 K and elevated H2 pressures of 5.3 and 6.1 MPa. The results, shown in Table 6.5, indicate that increased pressure led to increased conversion and HYD product selectivity following the 5 h reaction. Importantly, there was no deactivation observed at the higher pressures. In addition, the Ni2P-CA catalysts recovered after the reaction at elevated pressure were found to be free of deposited C and they retained their CO uptake capacity following rereduction to 773 K (Table 6.5). These results suggest that increased H2 pressure reduced the catalyst deactivation, either by hydrogenation of the precursors that led to carbon deposition or by reduction of sites that were partially oxidized by the reactants or products (water) during reaction. Furthermore, it was concluded that the observed changes in the Ni2P 148  crystallite size had a minor effect on the observed deactivation since otherwise the same deactivation phenomena would be observed at all H2 pressures.  Table 6.5. Conversion and product selectivities following 5 h hydrodeoxygenation of 4-methylphenol over Ni2P-CA-773 K at 623 K and various pressures as well as the used catalyst C deposition and CO uptake.  Catalyst  Conversion  Product Selectivity  C Deposited CO Uptakea  (%)  DDO (%)  HYD (%)  (wt.%)  (μmol/gMxP)  Ni2P-CA (4.4 MPa)  62.6  54.7  45.3  3.7  6  Ni2P-CA (5.3 MPa)  71.4  48.2  51.8  0.0  --  Ni2P-CA (6.1 MPa)  79.2  45.7  54.3  0.0  19  a.  CO uptakes measured following pretreatment in H2 to 773 K.  To determine if catalyst oxidation contributed to the observed deactivation of the Ni2P catalysts, the activity of several catalysts that had been purposely exposed to air were examined. The passivated MoP-CA-773 K (CO uptake <1 μmol/gMoP measured without rereduction), the passivated Ni2P-CA-773 K (CO uptake <1 μmol/gNi2P measured without rereduction), and the Cal-MoP-CA-773 K (CO uptake <1 μmol/g measured without rereduction and surface area <1 m2/gcat) were tested for the HDO of 4-methylphenol at 623 K and 4.4 MPa H2. The lognormal plot of the 4-methylphenol concentration versus time over these catalysts is given in Figure 6.11. The data show that all samples had approximately the same activity and all followed 1st-order kinetics. The initial rates, as well as the 1st-order kinetic parameters, k, of the samples are reported in Table 6.6. The data show that the passivated MoP-CA-773 K and passivated Ni2P-CA-773 K resulted in a significantly lower activity compared to the reduced samples. The loss is no doubt a consequence of a partially 149  oxidized surface that results when the catalyst is passivated, leading to a lower number of surface metal phosphide sites, even though the high surface area of the catalysts was maintained. In particular, both the Cal-MoP-CA-773 K (Mo phosphate precursor) and the passivated MoP-CA-773 K had the same activity, indicative of the fact that passivation of the high surface area MoP-CA-773 K catalysts led to complete oxidation. An additional HDO test of the passivated, low surface area MoP-noCA catalyst resulted in the same activity as the MoP-noCA, indicating that the passivated low surface area catalyst with large particles was readily reduced to MoP under reaction conditions. One would expect that if catalyst deactivation occurred as a consequence of oxidation by the reactants/products present in the HDO reactor, then oxidation of both MoP-CA-773 K and Ni2P-CA-773 K would occur since they have similar oxygen affinities [132]. Furthermore, both the passivated MoP-CA-773 K and Ni2P-CA-773 K showed lower activities than their reduced counterparts. Consequently, since only the Ni2P was observed to deactivate during HDO, it was concluded that deactivation by oxidation was not the likely mechanism of deactivation during HDO at the conditions of the present study.  150  ln(4-Methylphenol Concentration)  -1.36  -1.38  -1.40  -1.42  -1.44  -1.46 0  50  100  150  200  250  300  350  Time (min)  Figure 6.11. Lognormal of the 4-methylphenol concentration versus time for HDO reaction at 623 K and 4.4 MPa. Cal-MoP-CA-773 K (▼), passivated-MoP-CA-773 K (▲), passivated-Ni2P-CA-773 K (●), first order kinetic fit (--).  Table 6.6. Initial rate and kinetic parameter for the hydrodeoxygenation of 4-methylphenol at 623 K and 4.4 MPa over oxidized catalysts.  Catalyst  Initial Rate (mmol.min-1.g-1)  k (mL.min-1.g-1)  Cal-MoP-CA-773 K  0.023 ± 2.0E-4  0.090 ± 9.9E-3  Passivated MoP-CA-773 K  0.024 ± 2.6E-4  0.094 ± 9.9E-3  Passivated Ni2P-CA-773 K  0.024 ± 7.5E-3  0.093 ± 2.9E-4  The products from the HDO of 4-methylphenol over the calcined precursor and the passivated  catalysts,  included  the  DDO  product 151  toluene,  the  HYD  products  methylcyclohexane, 1-methylcyclohexene, and 4-methylcyclohexene, as well as the ISOM products of ethylidenecyclopentane, ethylcyclopentane, 1,3-dimethylcyclopentane, and 4,4dimethylcyclopentene. The isomerization products were not observed over the reduced catalysts and are a result of surface acidity arising from the oxidized metal phosphides [48]. The DDO and HYD product selectivities over the oxidized samples are given in Figure 6.12 with the balance being the ISOM products. Cal-MoP-CA-773 K had the highest selectivity to DDO products whereas the passivated Ni2P-CA-773 K had the lowest selectivity to DDO products and the highest selectivity to HYD products. Of particular note is the fact that all the oxidized samples had a relatively high selectivity to isomerization products, ranging from 1020%, compared to the reduced samples. The passivated Ni2P-CA-773 K was recovered following the 5 h reaction and was found to contain 5.1 wt.% C (Table 6.4). Re-reduction of this sample to remove oxidized metal phosphide, resulted in a CO uptake of <1 μmol/gNi2P, confirming that the loss in surface sites was due to C deposition and not oxidation of the metal phosphide.  152  90  a  b  a  b  a  80  DDO Selectivity (%)  70 60 50 40 30 20 10 0  50  100  150  200  250  300  350  Time (min)  60  b  HYD Selectivity (%)  50  40  30  20  10  0 0  50  100  150  200  250  300  350  Time (min)  Figure 6.12. DDO (a) and HYD (b) product selectivity versus time for the reaction at 623 K and 4.4 MPa. Cal-MoP-CA-773 K (▼), passivated MoP-CA-773 K (▲), passivated Ni2P-CA-773 K (●), guideline shown for illustration (--).  Together these results show that oxidation of both MoP and Ni2P catalysts does indeed result in a loss in catalyst HDO activity.  However, there is also a change in  selectivity with increased isomerization products observed as a consequence of surface 153  acidity that arises from the oxidized metal phosphide. Overall, the oxidation results indicate that the observed deactivation of the Ni2P catalysts was not due to oxidation of the phosphide under the reaction conditions of the present Chapter. Oxidation of the Ni2P would have resulted in the production of ISOM products, which were absent over the Ni2P catalysts. The product selectivity over the Ni2P catalysts at 623 K and 4.4 MPa remained relatively constant as a function of reaction time indicating no preferential deactivation of one specific surface site. This leads to the conclusion that deactivation was the result of a fouling mechanism that led to the non-selective coverage of Ni2P surface sites by carbon/coke. Presumably, the coke precursors arise from the olefinic and isomerization intermediates of the HYD pathway of 4-methylphenol HDO that was favored over the Ni2P-CA-773 K catalyst compared to the MoP-CA-773 K catalyst (Scheme 5.1). Furthermore, the fact that ISOM products were not observed over the Ni2P-CA-773 K catalyst suggests that the coke formation was a consequence of a strong adsorption of these intermediates on the catalyst surface with subsequent conversion to coke. With increased hydrogen pressure from 4.4 to 5.3 and 6.1 MPa the coke precursors were likely hydrogenated, thereby preventing coke formation and catalyst deactivation at the higher reaction pressures.  6.4  Conclusions  The synthesis of unsupported high surface area Ni2P-CA catalysts was reported. Calcination at 773 K led to nearly complete removal of C from the catalysts. Further calcination did not affect the C content of the catalysts, but led to sintering that reduced the catalysts surface area, CO uptake, and increased Ni2P particle size. Unlike in the case of MoP-CA, where an 154  “optimum” calcination temperature of 823 K was observed (Chapter 5), the present Chapter found an “optimum” calcination temperature of 773 K for the preparation of Ni2P-CA in the temperature range tested. This was due to nearly complete destruction of the Ni citrate in the calcined Ni2P-CA precursor at 773 K that led to decreased C content. The destruction of the Cal-Ni2P-CA citrate was due to the absence of a stabilizing nitrile group in the structure. This nitrile group was present in the metal citrate of the Cal-MoP-CA precursors and led to superior physical properties, such as surface area and particle size following reduction of the MoP-CA catalysts. At 623 K and 4.4 MPa H2 the Ni2P-CA-773 K was found to be 2.3 times more active than MoP-CA-773 K on a mass basis and 6 times more active on a site basis for the HDO of 4-methylphenol. However, all Ni2P catalysts deactivated due to non-selective C deposition on the catalyst surface. The deactivation was well defined by an exponential decay law and was not observed over the MoP catalysts. Oxidation was excluded as a potential cause of deactivation over the Ni2P catalysts due to the absence of a high selectivity to ISOM products that were found to be produced over deliberately oxidized samples. In addition, increases in H2 pressure from 4.4 to 5.3 and 6.1 MPa at 623 K improved the Ni2P-CA-773 K HDO activity and HYD selectivity and prevented catalyst deactivation. The initial TOFs of all Ni2P-CA catalysts were comparable and independent of calcination temperature and particle size. This implies that the HDO of 4-methylphenol is structure insensitive for the Ni2P crystallite sizes between 34−50 nm.  155  Chapter 7 Relationship Between Pyrolysis Oil and 4-Methylphenol HDO 7.1  Introduction  The HDO of pyrolysis oil model compounds is well described in the literature [18, 21, 23, 24, 28, 32, 34, 51, 53, 55, 56]. Similarly, the HDO of fast pyrolysis oil is also well studied [15, 68, 71, 73, 75, 142]. However, few studies have investigated the relationship between pyrolysis oil HDO and model compound HDO over specified catalysts. The present chapter reports on a comparison of pyrolysis oil HDO and 4-methylphenol HDO over MoO3, MoS2, MoP-CA-823 K, and Ni2P-CA-773 K catalysts. The Ni2P-CA-773 K and MoP-CA-823 K catalysts were chosen based on their superior activities identified in Chapters 5 and 6. To observe the ligand effect and surface acidity effect MoO3 and MoS2 were also tested.  7.2  Results and Discussion  The HDO of pyrolysis oil was done using the methods described in Chapter 3.3. The pyrolysis oil HDO tests were compared to the HDO of 4-methylphenol at 623 K and 4.4 MPa operated using 3,900 ppmw catalyst. The overall product yield of the pyrolysis oil was compared to the product selectivity of the 4-methylphenol following 50% reactant conversion.  156  7.2.1  Pyrolysis Oil Characterization  Fast pyrolysis oil was obtained from an industrial source and stored at 275 K. Prior to use the oil was mixed vigorously to inhibit phase separation. The pyrolysis oil C, H, N, S, and O (by difference) content was measured using a Perkin-Elmer 2400 Series II CHNS/O analyzer operated in the CHN mode. The C content was determined to be 45.6 wt.%, the H content was 7.4 wt.%, and the O content (measured by difference) was 47 wt.%. The moisture content of the oil was measured using Karl Fischer titration and was determined to be 25.2 ± 1.1 wt.%. The HHV of the oil was measured at 17.0 MJ/kg. The pyrolysis oil contained 24.6 wt.% O on a dry basis. 10 mL of oil yielded a mass of 12.2 g of oil. Therefore, 0.188 moles of H2 was required to deoxygenate 10 mL of oil at a cold pressure of 2.3 MPa. Based on this calculation it was determined that pure pyrolysis oil (100 mL) could not be deoxygenated in the batch reactor employed in this study due to limitations on the safe operating pressure of the unit. Consequently, the pyrolysis oil HDO experiments were conducted by diluting the oil in a solvent so that the H2 required for complete HDO was significantly reduced.  7.2.2  Pyrolysis Oil HDO  The HDO reaction was performed for 1 h at 523 K and 2.9 MPa (cold pressure) in batch mode using 0.5 g of catalyst (MoO3, MoS2, MoP-CA-823 K, Ni2P-CA-773 K) with 10 mL of pyrolysis oil in 90 mL decalin diluent. The catalyst amount was based on the amount used in past studies (5 wt.%) [59]. Sulfided NiMo/Al2O3 was also tested for comparison as it is a commercially available catalyst with a high surface area. A HDO reaction temperature of 523 157  K was used to emulate “mild” HDO conditions with a heating rate of 10.8 K/min [15, 59, 71]. The results for the HDO reaction are given in (Figure 7.1). It should be noted that the aqueous phase that made up the mass balance is not included in Figure 7.1. The thermal reaction performed in the absence of catalyst yielded 93.4 wt.% coke and 5 wt.% gas. The reaction over MoO3 yielded 0 wt.% liquid but decreased the coke product compared to the thermal reaction. The reaction over MoS2 produced 2.6 wt.% liquid and 68.5 wt.% coke. The sulfided NiMo/Al2O3 yielded 14.6 wt.% liquid and 60.4 wt.% coke. MoP-CA-823 K and Ni2P-CA-773 K yielded 48.1 and 21.1. wt.% liquid, respectively and 30.1 and 48.0 wt.% coke, respectively. The data show that MoP-CA-823 K had the highest yield of O free liquid and the lowest yield of coke among all catalysts tested. It is likely that deactivation over Ni2P-CA-773 K led to the inferior product yields observed over this catalyst compared to MoP-CA-823 K. Previous work by Wildschut et al. [59] also utilized a batch reactor for the HDO of pyrolysis oil. Their study produced hydrocarbon liquid yields of 26, 28, 65, 36, 53, and 67 wt.% over Co-MoS2/Al2O3, Ni-MoS2/Al2O3, Pd/C, Ru/Al2O3, Ru/C, Ru/TiO2, catalysts respectively, at 623 K and 20 MPa (continuous H2) after a 4 hour reaction using 25 g of pyrolysis oil and 1.25 g of catalyst [59]. This demonstrates that the MoP-CA-823 K used in this study (for the 1 h reaction at a lower temperature and H2 pressure) had an activity and hydrocarbon liquid yield that was similar to the supported precious metals used by Wildschut et al. [59] as they displayed bifunctional acidic and metallic properties with strong hydrogenating capabilities [22]. A study by Zhang et al. [15] also studied the HDO of pyrolysis oil in a tetralin solvent over a supported sulfided MoO3.CO/Al2O3 catalyst 158  containing 2.8 wt% P at 663 K and 2MPa H2. The authors found 90% HDO conversion following a 1 h reaction.  100  Liquid Coke Gas  Product Yield (wt%)  80  60  40  20  0  Thermal  MoO3  MoS2 NiMoS/Al2O3 MoP-CA Ni2P-CA  Figure 7.1. Pyrolysis oil product yield following HDO.  The coke phase reported in Figure 7.1 was composed of two fractions. The first fraction was an acetone soluble phase that dried into a thick viscous oil following evaporation (referred to as thick oil), whereas the second fraction was a hard, solid product that had to be removed from the sides of the reactor using significant mechanical force. The mass of catalyst added to the reactor was subtracted from solid phase mass obtained. The aqueous product was separated from the hydrocarbon liquid using a syringe. Initially the aqueous phase was a clear liquid but turned dark brown following storage. The liquid hydrocarbon product that was free of O was light yellow in color and its mass percent was 159  determined by subtracting the initial mass of decalin diluent added. Decalin was found to be stable at the reaction conditions. Therefore, it was assumed it did not contribute to the product species. Table 7.1 displays the C, H, and O (by difference) content of the liquid, aqueous, solid, and thick oil product phases. All catalysts, with the exception of MoO3 and MoS2 produced O-free hydrocarbon liquid. It is important to note that this liquid contained decalin, therefore the C, H, and O analysis is significantly influenced by the decalin phase. Wildschut et al. found 81, 74, 85, 89, 86, and 70% O removal from the liquid hydrocarbon phase over Co-MoS2/Al2O3, Ni-MoS2/Al2O3, Pd/C, Ru/Al2O3, Ru/C, Ru/TiO2, catalysts respectively, at 623 K and 20 MPa (continuous H2) after a 4 hour reaction using 25 g of pyrolysis oil and 1.25 g of catalyst. Other studies found 100% O removal for a 33 wt.% hydrocarbon yield over Co-MoS2/Al2O3 at 673 K and 30 MPa [76], 64% O removal for a 48 wt.% hydrocarbon yield over Pd/C at 613 K and 14 MPa [73], and 73% O removal for a 38 wt.% hydrocarbon yield over Ru/C at 623-673 K and 20 MPa [74]. Please note that the continuous H2 pressures employed in these studies were much higher than that of the present work and yielded much less O removal thereby demonstrating the high activity of the MoP-CA-823 K catalyst for the HDO of pyrolysis oil. The aqueous liquid phase produced over each catalyst in the current study contained C, indicating the potential presence of acetic acid and alcohols in this phase. Both the thick oil and solid product still retained O, but its content was reduced from the initial pyrolysis oil O content. Overall, all catalyst product species had approximately the same C, H, and O content regardless of catalyst used. 160  Table 7.1. C, H, and O content of HDO product.  Liquid Phase  Aqueous Phase  Solid Phase  Thick Oil  C  H  O  C  H  O  C  H  (wt.%)  (wt.%)  (wt.%)  (wt.%)  (wt.%)  (wt.%)  (wt.%)  (wt.%)  MoO3  --  --  --  19.1  11.3  69.6  61.4  7.5  31.1  --  --  --  MoS2  86.3  13.1  0.6  14.5  10.5  75.0  57.6  5.3  37.1  62.7  7.3  30.0  NiMo/Al2O3  86.7  12.3  0.0  21.9  10.0  68.1  62.2  5.6  32.2  62.4  5.9  31.7  MoP-CA-823 K  86.1  13.9  0.0  21.3  10.8  67.9  56.6  4.5  38.9  64.8  6.8  28.6  Ni2P-CA-773 K  86.3  13.7  0.0  23.4  9.9  66.7  63.2  5.6  31.1  62.8  5.7  31.5  161  O  C  H  O  (wt.%) (wt.%) (wt.%) (wt.%)  Table 7.2 lists the analysis of the gas phase following the 1 h HDO reaction. All catalysts produced CO and CO2 following the HDO reaction. MoO3 and sulfided NiMo/Al2O3 produced the highest amount of CO2, which is likely related to the high surface acidity of these catalysts. CH4 formation was observed over Ni2P-CA-773 K, which is a consequence of the high HYD ability observed over this catalyst. De Mercader et al. [75] also observed CH4 formation for the HDO of pyrolysis oil over the strong hydrogenating catalyst, 5 wt.% Ru/C. In addition, the Ni2P-CA-773 K produced a high gas yield of CO2, whereas MoP-CA-823 K produced a high gas yield of CO. This decarboxylation and decarbonylation over the Ni and Mo based phosphide, respectively, could be related to the oxidation of the metal phosphides in the presence of pyrolysis oil that led to the formation of surface acid sites, thereby leading to gas formation. All catalysts tested produced approximately the same yield of gas, but differed in their gas composition.  Table 7.2. Gas analysis of HDO product (mol.%).  Catalyst  CO  CO2  CH4  C3H8  MoO3  27.3  68.2  0  4.5  MoS2  57.2  43.4  0  0  NiMoS/Al2O3  34.7  65.3  0  0  MoP-CA-823 K  71.8  28.2  0  0  Ni2P-CA-773 K  43.6  51.1  5.3  0  The mass and C balances for all HDO reactions are given in Table 7.3. The large mass loss encountered over MoO3 was due to losses of the aqueous phase. Condensation on the reactor walls caused difficulties in the collection of the aqueous phase in this case. 162  Table 7.3. Mass and C balance for HDO reactions.  Catalyst  7.2.3  Mass Balance  C Balance  (wt.%)  (wt.%)  MoO3  83.2  -0.4  MoS2  96.5  -5.9  NiMo/Al2O3  102.4  +4.4  MoP-CA-823 K  103.3  +4.1  Ni2P-CA-773 K  91.4  +1.8  Comparison of Pyrolysis Oil HDO and 4-Methylphenol HDO  To observe the effectiveness of using model compounds to screen catalysts the HDO of 4methylphenol was compared to the HDO of pyrolysis oil. To do this, the product selectivity for the HDO of 4-methylphenol at 623 K and 4.4 MPa following 50% conversion was compared to the product yield of pyrolysis oil following the 1 h reaction. Figures 7.2 and 7.3 display the relationship between the pyrolysis oil product yields of coke and oil versus the 4methylphenol HYD and DDO selectivity. Overall, the catalysts displaying a high selectivity for HYD of 4-methylphenol (MoP-CA-823 K and Ni2P-773 K) also displayed a high yield of O free liquid and a low yield of coke following the HDO of pyrolysis oil. Whereas, the catalysts displaying a high selectivity for DDO and low selectivity for HYD (MoS2 and MoO3) also had a high yield of coke and a low yield of O free liquid following the HDO of pyrolysis oil. Similar results were observed for those catalysts having a high degree of ISOM of 4-methylphenol (Figure 7.4).  163  100  MoS2  Product Yield (wt.%)  80  MoO3  60  MoP-CA 40  20  Ni2P-CA 0 0  10  20  30  40  50  4-Methylphenol HYD Selectivity (%)  Figure 7.2. Pyrolysis oil product yield versus 4-methylphenol HYD selectivity (50% conversion). Liquid (●), coke (□), guideline shown for illustration (--).  100  MoS2  80  Product Yield (wt.%)  MoO3 60  MoP-CA 40  20  Ni2P-CA 0 50  55  60  65  70  75  80  85  90  4-Methylphenol DDO Selectivity (%)  Figure 7.3. Pyrolysis oil product yield versus 4-methylphenol DDO selectivity (50% conversion). Liquid (●), coke (□), guideline shown for illustration (--). 164  100  MoO3  Product Yield (wt.%)  80  MoS2  60  MoP-CA 40  Ni2P-CA 20  0 0  2  4  6  8  10  12  4-Methylphenol ISOM Selectivity (%)  Figure 7.4. Pyrolysis oil product yield versus 4-methylphenol ISOM selectivity (50% conversion). Liquid (●), coke (□), guideline shown for illustration (--).  A high degree of coke was observed for the thermal reaction as well as the HDO reaction over MoO3, MoS2, and sulfided NiMo/Al2O3. This is because pyrolysis oil polymerizes at temperatures above 373 K [5, 58, 69], especially over acidic based catalysts to form coke [5, 58, 59]. The polymerization encountered over the MoO3, MoS2, and sulfided NiMo/Al2O3 catalysts was caused by the condensation of unsaturated double bonds such as olefins, aldehydes, and ketones present in the pyrolysis oil [26]. Less coking was observed over the metal phosphide catalysts due to their strong hydrogenating abilities that saturated these double bonds before condensation occurred. Overall, all solid coke had less O and H than the initial pyrolysis oil feed indicating that indeed condensation reactions occurred.  165  In addition the high yields of coke observed in this study compared to other studies [4, 15, 68, 71, 73] were likely a consequence of the limited H2 supply in the batch reactor.  7.2.4  Pressure and Temperature Effects  The effect of temperature and pressure on the HDO of pyrolysis oil was tested over the MoPCA-823 K. The results are given in Table 7.4. When the temperature of the reaction was increased from 523 to 548 K the yield of O free liquid decreased and the coke yield increased. The liquid yield was maintained and the coke yield was decreased when the pressure was increased from 2.9 to 4.3 MPa. The O free liquid yield result is surprising as an increase in H2 pressure should increase the amount of H2 supplied to the pyrolysis oil during the reaction. It is likely that the increase in pressure did not significantly affect the H2 solubility in the oil due to the H2 starvation encountered in these reactions because of the batch reactor configuration.  Table 7.4. The effect of temperature and pressure for the HDO of pyrolysis oil over MoP-CA-823 K following the 1 h reaction.  Conditions (T, cold P)  O Free Liquid  Coke Content  Gas Content  Content (wt.%)  (wt.%)  (wt.%)  (523 K, 2.9 MPa)  48.1  30.1  6.2  (548 K, 2.9 MPa)  26.6  43.5  5.0  (523 K, 4.3 MPa)  47.6  27.9  4.6  166  7.3  Conclusions  Of all catalysts tested for the HDO of pyrolysis oil MoP-CA-823 K was found to have the highest yield of O free liquid and lowest coke yield, followed by Ni2P-CA-773 K, sulfided NiMo/Al2O3, MoS2 and MoO3 for the HDO of pyrolysis oil. A relationship was found to exist between the product selectivity of model compound 4-methylphenol and the product yield of raw pyrolysis oil. Those catalysts displaying high hydrogenating abilities had a high degree of O free liquid and a low yield of coke (MoP-CA-823 K), while those catalysts displaying high isomerization abilities had a high yield of coke (MoO3). Overall, phosphide catalysts are successful catalysts for HDO reactions and are superior to commercial, sulfided NiMo/Al2O3. This is because they have strong HYD abilities that saturate double bonds thus preventing condensation reaction, chain propagation, and coke precipitation.  167  Chapter 8 Conclusions and Recommendations 8.1  Conclusions  The HDO of pyrolysis oil and pyrolysis oil model compounds is well cited in the literature over traditional Mo-based sulfided catalysts that are active and selective for HDS and HDN reactions. However, since pyrolysis oils do not contain S, the sulfided catalysts lose S during reaction without the addition of CS2 or H2S. The objective of the present study was to investigate the HDO of pyrolysis oil model compound 4-methylphenol over alternative nonsulfided catalysts. A catalyst screening study for the HDO of 4-methylphenol is reported in Chapter 4. It was demonstrated that unsupported low surface area MoO3, MoO2, MoS2, and MoP catalysts were active for the HDO of 4-methylphenol as all products produced were free of O. At the reaction conditions, MoO3 underwent reduction to Mo4O11, MoO2, and Mo. This mixed Mo oxide was found to have high activity for the decomposition of 4-methylphenol. MoO2 and MoS2 displayed similar conversions, selectivities, and activation energies for the decomposition of 4-methylphenol. The initial TOF was found to be highest over the MoP catalyst, which also displayed the lowest activation energy and highest selectivity towards the completely hydrogenated product, methylcyclohexane. However, the 4-methylphenol conversion over the MoP was limited by its low surface area and CO uptake. In order to increase the MoP surface area and CO uptake, the MoP was prepared using a chelating agent, CA. Chapter 5 demonstrated that the addition of CA to the AHM and 168  AHP solutions, followed by calcination and reduction, increased the surface area and dispersion of the MoP-CA compared to MoP-noCA. However, 20.4 wt.% residual C was present in the Cal-MoP-CA precursor calcined at 773 K. The C, present as a citrate, acted as a structural promoter of the calcined samples, limiting agglomeration of the metal crystallites during reduction, to produce highly dispersed MoP nanoparticles. A maximum initial hydrodeoxygenation rate and CO uptake was found for the MoP-CA catalyst containing 9.4 wt.% C. Calcination of MoP-CA at temperatures above 823 K resulted in crystallite agglomeration, which reduced the CO uptake and initial hydrodeoxygenation rate. The hydrodeoxygenation of 4-methylphenol was found to be structure insensitive over all MoPCA catalysts. The calculated initial TOF was approximately 0.079 s-1. The TOFHYD:TOFDDO ratio over MoP-noCA (1.06:1) was higher than MoP-CA (0.78:1). This suggested that MoPnoCA contained a higher ratio of weak electrophilic CUS, which suggested a more reduced metal phosphide in the MoP-noCA than MoP-CA. The synthesis of unsupported high surface area Ni2P-CA catalysts is reported in Chapter 6. Calcination at 773 K led to nearly complete removal of C from the catalysts. Further calcination did not affect the C content of the catalysts, but led to sintering that reduced the catalyst’s surface area, CO uptake, and increased Ni2P particle size. In the case of MoP, an “optimum” calcination temperature of 823 K was observed, whereas an “optimum” calcination temperature of 773 K for the preparation of Ni2P-CA was identified over the temperature range tested. The lower “optimum” calcination temperature was due to nearly complete destruction of the Ni citrate in the calcined Ni2P-CA precursor at 773 K that led to decreased C content. The destruction of the Cal-Ni2P-CA citrate was due to the 169  absence of a stabilizing nitrile group in the structure. The nitrile group was present in the Mo citrate precursors and led to superior physical properties, such as surface area and particle size, of the MoP-CA catalysts following reduction, compared to the Ni2P-CA catalysts. At 623 K and 4.4 MPa H2 the Ni2P-CA-773 K was initially found to be 2.3 times more active than MoP-CA-773 K on a mass basis and 6 times more active on a site basis for the HDO of 4-methylphenol. However, all Ni2P catalysts deactivated due to non-selective C deposition on the catalyst surface. The deactivation was well defined by an exponential decay law and was not observed over the MoP catalysts. Oxidation was excluded as a potential cause of deactivation over the Ni2P catalysts due to the absence of a high selectivity to ISOM products that were found to be produced over deliberately oxidized samples. In addition, increases in H2 pressure from 4.4 to 5.3 and 6.1 MPa at 623 K improved the Ni2P-CA-773 K HDO activity and HYD selectivity and prevented catalyst deactivation. The deactivation parameter was found to be approximately equal over all Ni2P catalysts, indicating a similar mechanism of deactivation. The initial TOFs of all Ni2P-CA catalysts were comparable and independent of calcination temperature and particle size. This implies that the HDO of 4-methylphenol was structure insensitive for the Ni2P-CA catalysts with crystallite sizes between 34−50 nm. Finally, the use of 4-methylphenol to screen catalysts for HDO was validated by comparing catalyst performance for 4-methylphenol HDO and pyrolysis oil HDO. Of all catalysts tested for pyrolysis oil HDO, MoP-CA-823 K was found to have the highest yield of O free liquid and lowest coke yield, followed by Ni2P-CA-773 K, sulfided NiMo/Al2O3, MoS2 and MoO3. A relationship was found to exist between the product selectivity of model 170  compound 4-methylphenol and the product yield of raw pyrolysis oil. Those catalysts displaying high hydrogenating abilities had a high degree of O free liquid and a low yield of coke (MoP-CA-823 K), while those catalysts displaying high isomerization abilities had a high yield of coke (MoO3). Overall, phosphide catalysts were identified as successful catalysts for HDO reactions and were superior to commercial, sulfided NiMo/Al2O3. This is because they had strong HYD abilities that saturated double bonds thus preventing condensation, chain propagation, and coke precipitation reactions.  8.2 8.2.1  Recommendations Preparation of High Surface Area Ni2P-CA  As discussed in Chapter 5, calcination of the Dry-Ni2P-CA precursor significantly destroyed the metallic citrate which led to inferior physical properties of the reduced Ni2P-CA-773 K catalyst compared to MoP-CA-773 K. It was proposed that the loss of citrate in the Cal-Ni2PCA precursor was due to the absence of the stabilizing nitrile group, which was present in the Cal-MoP-CA precursor. The nitrile group was supplied by the ammonium component of the Mo precursor salt, whereas the Ni precursor salt contained nitrate. To confirm the role of ammonium, it is recommended that Ni2P-CA be prepared using an ammonium salt such as ammonium nickel sulfate hexahydrate ((NH4)2Ni(SO4)2.6H2O). However, the S present in this precursor salt may lead to the inclusion of S in the catalyst structure. Additionally, it is recommended that an alternative chelating agent, with a maximized amount of carboxylic groups for the metal to bind to, should be investigated for the preparation of unsupported Ni2P, such as oxalic or malic acid. 171  8.2.2  Preparation of Bi-Metal and Sulfided Metal Phosphides  It is recommended to dope the MoP-CA catalyst with Ni. It was shown in Chapter 6 that Ni2P-CA had superior activity to MoP-CA. However, all Ni2P catalysts deactivated. In order to increase the activity of the MoP-CA it should be doped with Ni. In addition, the loading of the Ni should be adjusted to observe if and when deactivation of the Ni species occurs. Sulfided metal phosphides should be prepared and tested for the HDO of 4methylphenol and pyrolysis oil to observe if sulfidation increases the activity of the phosphide catalysts. This can be done through ex- or in-situ sulfidation of the metal phosphide catalysts using H2 and CS2. In addition, the characterization of these catalysts should be undertaken to determine the surface site concentration and chemistry of the active sites.  8.2.3  Ni2P Deactivation  It was suggested in Chapter 6 that the deactivation of the Ni2P catalysts for the HDO of 4methylphenol was due to C deposition on the catalyst surface. It was proposed that the C deposition caused from the formation of isomerization product intermediates that polymerized and precipitated out of solution to form coke. Therefore, these intermediates (methylcyclohexenes) should be used as reactants over Ni2P to observe if they lead to coking and deactivation of the catalyst surface.  172  8.2.4  Catalyst Durability  The durability of the catalysts should be investigated by recycling the catalysts for multiple HDO reactions. From this the turn over number (TON) and catalyst lifetime can be determined.  8.2.5  The HDO of Furanic, Lignin Derivatives, and O Mixtures  The present study investigated the HDO of phenolic compounds where the O was present as a side group. It is suggested to investigate the HDO of a compound where the O is contained within the aromatic ring of the hydrocarbon compound. Such a compound could be a furan derivative. This should be done to observe the reactivity of these compounds over the catalysts tested in this study. In addition, a more representative compound of pyrolysis oil, such as syringaldehyde, guaiacol, lignin derivatives, or a mixture of multiple oxygenated compounds (to model pyrolysis oil) should be used as a reactant mixture to observe if the phosphide catalysts are superior catalysts for the HDO of other O containing hydrocarbons such as furans, acids, and alcohols.  8.2.6  The HDO of Pyrolysis Oil  It is suggested to test the HDO of fast pyrolysis oil in the absence of decalin diluent. This can be done by using a semi-batch reactor with a continuous supply of H2 at elevated pressures. Either a new reactor should be employed or the current reactor should be modified. This would require the addition of mass flow controllers, mass flow meters, a compressor, condensers, and a back pressure regulator. The pyrolysis oil should be tested at the 173  temperature range employed in this study at an increased H2 pressure for 1 h. Additional temperatures, H2 pressures, agitation speeds, solvent concentrations, and catalyst concentrations should be tested to develop a clear understanding of these effects on the pyrolysis oil reaction. It is also recommended that supported metal phosphide catalysts be compared to supported sulfide catalysts for the HDO of pyrolysis oil. In addition, the HDO of pyrolysis oil should be done with pyrolysis oil supplied by a different industrial source. This can be done to observe the effect of C, H, and O content as well as pyrolysis oil aging on the HDO reaction. Further analysis should be done on the liquid hydrocarbon and aqueous product species including IR analysis, GC-MS, simulated distillation, HHV analysis, Karl Fischer titration, and density measurements. XRD and HHV analysis should also be done on the solid product. The decalin diluent used in the present study led to issues that complicated some of these analyses.  8.2.7  Coke Recycle  The catalytic HDO of pyrolysis oil in Chapter 7 led to the generation of significant amounts of solid coke product that encapsulated the catalysts. It is recommended to recycle the pyrolysis oil coke product containing the catalyst by drying and grinding it and re-using it as a catalyst in the batch reactor. This idea was proposed by Rezaei et al. [151] for the hydroconversion of heavy oils and residue in a slurry phase reactor. Rezaei et al. [151] showed that the coke product (containing MoS2 catalyst) could be recycled successfully many times. The solid product acted as a catalyst and reduced the solid yield and increased the liquid yield for subsequent hydroconversion reactions. 174  8.2.8  DFT Modeling Catalyst Surfaces  Denisty functional theory (DFT) is a modeling technique used to investigate the electronic structure of materials. This method should be employed to build MoP, Ni2P, and MoS2 surfaces to obtain the adsorption energies of the 4-methylphenol reactant and intermediate species (methylcyclohexenes) on the catalyst surface. This will determine if the C deposition encountered over the Ni2P catalyst was due to strong adsorption of the methylcyclohexene on the surface. This will also illustrate the binding of the reactant aromatic ring (as planar or point adsorption) and O heteroatom with the catalyst system.  175  Bibliography [1] D. Chiras, Environmental Science: Creating a Sustainable Future, Jones & Bartlett Learning, 2004. [2] E. Berl, Production of Oil from Plant Material, Science 99 (1944) 309-312. [3] D. Mohan, Pittman Charles U. and P.H. Steele, Pyrolysis of Wood/Biomass for Bio-oil: A Critical Review, Energy Fuels 20 (2006) 848-889. [4] D. Elliott, Historical Developments in Hydroprocessing Bio-oils, Energy Fuels 21 (2007) 1792-1815. [5] E. Furimsky, Catalytic Hydrodeoxygenation, Appl. Catal., A 199 (2000) 147-190. [6] O. Klimov, A. Pashigreva, M. Fedotov, D. Kochubey, Y. Chesalov, G. Bukhtiyarova and A. Noskov, Co-Mo Catalysts for Ultra-Deep HDS of Diesel Fuels Prepared via Synthesis of Bimetallic Surface Compounds, J. Mol. Catal. A: Chem. 322 (2010) 80-89. [7] C. Tye and K. Smith, Hydrodesulfurization of Dibenzothiophene over Exfoliated MoS2 Catalyst, Catal. Today 116 (2006) 461-468. [8] H. Rezaei, X. Liu, S. Ardakani, K. Smith and M. Bricker, A Study of Cold Lake Vacuum Residue Hydroconversion in Batch and Semi-Batch Reactors Using Unsupported MoS2 Catalysts, Catal. Today 150 (2010) 244-254. [9] E. Karimi, A. Gomez, S. Kycia and M. Schlaf, Thermal Decomposition of Acetic and Formic Acid Catalyzed by Red Mud-Implications for the Potential Use of Red Mud as a Pyrolysis Bio-Oil Upgrading Catalyst, Energy Fuels 24 (2010) 2747-2757. [10] R. Venderbosch and W. Prins, Fast Pyrolysis Technology Development, Biofuels, Bioprod. Bioref. 4 (2010) 178-208. [11] S. Jones, C. Valkenburg, C. Walton, D. Elliott, J. Holladay, D. Stevens, C. Kinchin and S. Czernik, Production of Gasoline and Diesel from Biomass via Fast Pyrolysis, Hydrotreating and Hydrocracking: a Design Case, Pacific Northwest National Laboratory PNNL-18284 (2009) . [12] M. Ringer, V. Putsche and J. Scahill, Large-Scale Pyrolysis Oil Production: A Technology Assessment and Economic Analysis, National Renewable Energy Laboratory NREL/TP-510-37779 (2006).  176  [13] D. Bradley, European Market Study for BioOil (Pyrolysis Oil), Climate Change Solutions, IEA Bioenergy Task 40- Bio-trade (2006). [14] S. Czernik and A. Bridgwater, Overview of Applications of Biomass Fast Pyrolysis Oil, Energy Fuels 18 (2004) 590-598. [15] S. Zhang, Y. Yan, T. Li and Z. Ren, Upgrading of Liquid Fuel from the Pyrolysis of Biomass, Bioresour. Technol. 96 (2005) 545-550. [16] A. Bridgwater, Catalysis in Thermal Biomass Conversion, Appl. Catal., A 116 (1994) 547. [17] United Nations Economic Commission for Europe, Canada’s Wood Supply Potential for Bioenergy Use (2008). [18] Y. Romero, F. Richard, Y. Reneme and S. Brunet, Hydrodeoxygenation of Benzofuran and its Oxygenated Derivatives (2,3-Dihydrobenzofuran and 2-Ethylphenol) over NiMoP/Al2O3 Catalyst, Appl. Catal., A 353 (2009) 46-53. [19] W. Wang, Y. Yang, J. Bao and H. Luo, Characterization and Catalytic Properties of NiMo-B Amorphous Catalysts for Phenol Hydrodeoxygenation, Catal. Commun. 11 (2009) 100-105. [20] Y. Yang, C. Tye and K. Smith, Influence of MoS2 Catalyst Morphology on the Hydrodeoxygenation of Phenols, Catal. Commun. 9 (2008) 1364-1368. [21] F. Massoth, P. Politzer, M. Concha, J. Murray, J. Jakowski and J. Simons, Hydrodeoxygenation of Methyl-Substituted Phenols: Correlations of Kinetic Parameters with Molecular Properties, J. Phys. Chem. B 110 (2006) 14283-14291. [22] A. Talukdar, K. Bhattacharyya and S. Sivasanker, Hydrogenation of Phenol over Supported Platinum and Palladium Catalysts, Appl. Catal., A 96 (1993) 229-39. [23] B. Gevert, J. Otterstedt and F. Massoth, Kinetics of the Hydrodeoxygenation of MethylSubstituted Phenols, Appl. Catal. 31 (1987) 119-31. [24] E. Furimsky, J. Mikhlin, D. Jones, T. Adley and H. Baikowitz, On the Mechanism of Hydrodeoxygenation of Ortho Substituted Phenols, Can. J. Chem. Eng. 64 (1986) 982-5. [25] H. Weigold, Behaviour of Co-Mo-Al2O3 Catalysts in the Hydrodeoxygenation of Phenols, Fuel 61 (1982) 1021-1026.  177  [26] P. Grange, E. Laurent, R. Maggi, A. Centeno and B. Delmon, Hydrotreatment of Pyrolysis Oils from Biomass: Reactivity of the Various Categories of Oxygenated Compounds and Preliminary Techno-Economical Study, Catal. Today 29 (1996) 297-301. [27] L. Rollmann, Catalytic Hydrogenation of Model Nitrogen, Sulfur, and Oxygen Compounds, J. Catal. 46 (1977) 243-252. [28] E. Odebunmi and D. Ollis, Catalytic Hydrodeoxygenation: I. Conversions of o-, p-, and m-Cresols, J. Catal. 80 (1983) 56-64. [29] N. Samchenko and N. Pavlenko, Reactivity of Alkylphenols in Liquid Phase Catalytic Hydrogenation, React. Kinet. Catal. Lett.18 (1982) 155-158. [30] E. Shin and M. Keane, Catalytic Hydrogen Treatment of Aromatic Alcohols, J. of Catal. 173 (1998) 450-459. [31] M. Bejblova, P. Zamostny, L. Cerveny and J. Cejka, Hydrodeoxygenation of Benzophenone on Pd Catalysts, Appl. Catal., A 296 (2005) 169-175. [32] A. Bunch and U. Ozkan, Investigation of the Reaction Network of Benzofuran Hydrodeoxygenation over Sulfided and Reduced Ni-Mo/Al2O3 Catalysts, J. Catal. 206 (2002) 177-187. [33] E. Laurent and B. Delmon, Influence of Oxygen-, Nitrogen-, and Sulfur-Containing Compounds on the Hydrodeoxygenation of Phenols over Sulfided Cobalt-Molybdenum/γAl2O3 and Nickel-Molybdenum/γ-Al2O3 Catalysts, Ind. Eng. Chem. Res. 32 (1993) 2516-24. [34] O. Senol, T. Viljava and A. Krause, Effect of Sulphiding Agents on the Hydrodeoxygenation of Aliphatic Esters on Sulphided Catalysts, Appl. Catal., A 326 (2007) 236-244. [35] T. Viljava, R. Komulainen and A. Krause, Effect of H2S on the Stability of CoMo/Al2O3 Catalysts During Hydrodeoxygenation, Catal. Today 60 (2000) 83-92. [36] C. Moreau, C. Aubert, R. Durand, N. Zmimita and P. Geneste, Structure-Activity Relationships in Hydroprocessing of Aromatic and Heteroaromatic Model Compounds over Sulfided Nickel Oxide-Molybdena/γ-Al2O3 and Nickel Oxide-Tungsten Trioxide/γ-Al2O3 Catalysts: Chemical Evidence for the Existence of Two Types of Catalytic Sites, Catal. Today 4 (1988) 117-31. [37] M. Vogelzang, C. Li, G. Schuit, B. Gates and L. Petrakis, Hydrodeoxygenation of 1Naphthol: Activities and Stabilities of Molybdena and Related Catalysts, J. Catal. 84 (1983) 170-177. 178  [38] C. Moreau, J. Joffre, C. Saenz and P. Geneste, Hydroprocessing of Substituted Benzenes over a Sulfided CoO-MoO3/γ-A12O3 Catalyst, J. Catal. 122 (1990) 448-451. [39] S. Oyama, X. Wang, F. Requejo, T. Sato and Y. Yoshimura, Hydrodesulfurization of Petroleum Feedstocks with a New Type of Nonsulfide Hydrotreating Catalyst, J. Catal. 209 (2002) 1-5. [40] S. Yang, C. Liang and R. Prins, Preparation and Hydrotreating Activity of Unsupported Nickel Phosphide with High Surface Area, J. Catal. 241 (2006) 465-469. [41] I. Abu and K. Smith, The Effect of Cobalt Addition to Bulk MoP and Ni2P Catalysts for the Hydrodesulfurization of 4,6-Dimethyldibenzothiophene, J. Catal. 241 (2006) 356-366. [42] D. Phillips, S. Sawhill, R. Self and M. Bussell, Synthesis, Characterization, and Hydrodesulfurization Properties of Silica-Supported Molybdenum Phosphide Catalysts, J. Catal. 207 (2002) 266-273. [43] C. Stinner, R. Prins and T. Weber, Formation, Structure, and HDN Activity of Unsupported Molybdenum Phosphide, J. Catal. 191 (2000) 438-444. [44] S. Sawhill, D. Phillips and M. Bussell, Thiophene Hydrodesulfurization over Supported Nickel Phosphide Catalysts, J. Catal. 215 (2003) 208-219. [45] X. Wang, P. Clark and S. Oyama, Synthesis, Characterization, and Hydrotreating Activity of Several Iron Group Transition Metal Phosphides, J. Catal. 208 (2002) 321-331. [46] V. Zuzaniuk and R. Prins, Synthesis and Characterization of Silica-Supported Transition-Metal Phosphides as HDN Catalysts, J. Catal. 219 (2003) 85-96. [47] R. Prins and M. Bussell, Metal Phosphides: Preparation, Characterization and Catalytic Reactivity, Catal. Lett. 142 (2012) 1413-1436. [48] V. Whiffen and K. Smith, Hydrodeoxygenation of 4-Methylphenol over Unsupported MoP, MoS2, and MoOx Catalysts, Energy Fuels 24 (2010) 4728-4737. [49] V. Whiffen, K. Smith and S. Straus, The Influence of Citric Acid on the Synthesis and Activity of High Surface Area MoP for the Hydrodeoxygenation of 4-Methylphenol, Appl. Catal., A 419-420 (2012) 111-125. [50] V. Whiffen and K. Smith, A Comparative Study of 4-Methylphenol Hydrodeoxygenation over High Surface Area MoP and Ni2P, Top. Catal. 55 (2012) 981-990.  179  [51] H. Zhao, D. Li, P. Bui and S. Oyama, Hydrodeoxygenation of Guaiacol as Model Compound for Pyrolysis Oil on Transition Metal Phosphide Hydroprocessing Catalysts, Appl. Catal., A 391 (2011) 305-310. [52] D. Li, P. Bui, H.Y. Zhao, S.T. Oyama, T. Dou and Z.H. Shen, Rake Mechanism for the Deoxygenation of Ethanol over a Supported Ni2P/SiO2 Catalyst, J. Catal. 290 (2012) 1-12. [53] K. Li, R. Wang and J. Chen, Hydrodeoxygenation of Anisole over Silica-Supported Ni2P, MoP, and NiMoP Catalysts, Energy Fuels 25 (2011) 854-863. [54] A. Cho, J. Shin, A. Takagaki, R. Kikuchi and S. Oyama, Ligand and Ensemble Effects in Bimetallic NiFe Phosphide Catalysts for the Hydrodeoxygenation of 2Methyltetrahydrofuran, Top. Catal. 55 (2012) 969-980. [55] P. Bui, J. Cecilia, S. Oyama, A. Takagaki, A. Infantes-Molina, H. Zhao, D. Li, E. Rodriguez-Castellon and A. Lopez, Studies of the Synthesis of Transition Metal Phosphides and Their Activity in the Hydrodeoxygenation of a Biofuel Model Compound, J. Catal. 294 (2012) 184-198. [56] R. Bowker, M. Smith, M. Pease, K. Slenkamp, L. Kovarik and M. Bussell, Synthesis and Hydrodeoxygenation Properties of Ruthenium Phosphide Catalysts, Catal. 1 (2011) 917922. [57] M. Olarte, Base-Catalyzed Depolymerization of Lignin and Hydrodeoxygenation of Lignin Model Compounds, Georgia Inistitute of Technology Thesis (2011). [58] E. Furimsky and F. Massoth, Deactivation of Hydroprocessing Catalysts, Catal. Today 52 (1999) 381-495. [59] J. Wildschut, F. Mahfud, R. Venderbosch and H. Heeres, Hydrotreatment of Fast Pyrolysis Oil Using Heterogeneous Noble-Metal Catalysts, Ind. Eng. Chem. Res. 48 (2009) 10324-10334. [60] R. Wang and K. Smith, Hydrodesulfurization of 4,6-Dimethyldibenzothiophene over High Surface Area Metal Phosphides, Appl. Catal., A 361 (2009) 18-25. [61] R. Cheng, Y. Shu, L. Li, M. Zheng, X. Wang, A. Wang and T. Zhang, Synthesis and Characterization of High Surface Area Molybdenum Phosphide, Appl. Catal., A 316 (2007) 160-168. [62] B. Ibeh, S. Zhang and J. Hill, Effect of Citric Acid on the Synthesis of Tungsten Phosphide Hydrotreating Catalysts, Appl. Catal., A 368 (2009) 127-131.  180  [63] S. Bouwens, J. Vissers, V. de Beer and R. Prins, Phosphorus Poisoning of Molybdenum Sulfide Hydrodesulfurization Catalysts Supported on Carbon and Alumina, J. Catal. 112 (1988) 401-10. [64] Z. Hong and J. Regalbuto, Nature of Adsorption Sites on Sulfided Mo Catalysts and Their Selectivity in Chemisorption of Probe Molecules, J. Phys. Chem. 99 (1995) 9452-7. [65] F. Bataille, J. Lemberton, P. Michaud, G. Perot, M. Vrinat, M. Lemaire, E. Schulz, M. Breysse and S. Kasztelan, Alkyldibenzothiophenes Hydrodesulfurization-Promoter Effect, Reactivity, and Reaction Mechanism, J. Catal. 191 (2000) 409-422. [66] A. Bunch, L. Zhang, G. Karakas and U. Ozkan, Reaction Network of Indole Hydrodenitrogenation over NiMoS/γ-Al2O3 Catalysts, Appl. Catal., A 190 (2000) 51-60. [67] R. Agrawal and N. Singh, Solar Energy to Biofuels, Annu. Rev. Chem. Biomol. Eng. 1 (2010) 343-364. [68] P. Mortensen, J. Grunwaldt, P. Jensen, K. Knudsen and A. Jensen, A Review of Catalytic Upgrading of Bio-oil to Engine Fuels, Appl. Catal., A 407 (2011) 1-19. [69] M. Samolada, W. Baldauf and I. Vasalos, Production of a Bio-Gasoline by Upgrading Biomass Flash Pyrolysis Liquids via Hydrogen Processing and Catalytic Cracking, Fuel 77 (1998) 1667-1675. [70] R. Sharma and N. Bakhshi, Conversion of non-Phenolic Fraction of Biomass-Derived Pyrolysis Oil to Hydrocarbon Fuels over HZSM-5 Using a Dual Reactor System, Bioresour. Technol. 45 (1993) 195-203. [71] N. Joshi and A. Lawal, Hydrodeoxygenation of Pyrolysis Oil in a Microreactor, Chem.Eng. Sci. 74 (2012) 1-8. [72] J. Adjaye and N. Bakhshi, Catalytic Conversion of a Biomass-Derived Oil to Fuels and Chemicals I: Model Compound Studies and Reaction Pathways, Biomass Bioenergy 8 (1995) 131-149. [73] D. Elliott, T. Hart, G. Neuenschwander, L. Rotness and A. Zacher, Catalytic Hydroprocessing of Biomass Fast Pyrolysis Bio-oil to Produce Hydrocarbon Products, Environ. Prog. Sustain. Energy 28 (2009) 441-449. [74] R. Venderbosch, A. Ardiyanti, J. Wildschut, A. Oasmaa and H. Heeres, Stabilization of Biomass-Derived Pyrolysis Oils, J. Chem. Technol. Biotechnol. 85 (2010) 674-686.  181  [75] F. Mercader, M. Groeneveld, S. Kersten, C. Geantet, G. Toussaint, N. Way, C. Schaverien and K. Hogendoorn, Hydrodeoxygenation of Pyrolysis Oil Fractions: Process Understanding and Quality Assessment Through co-Processing in Refinery Units, Energy Environ. Sci. 4 (2011) 985-997. [76] W. Baldauf, U. Balfanz and M. Rupp, Upgrading of Flash Pyrolysis Oil and Utilization in Refineries, Biomass Bioenergy 7 (1994) 237-244. [77] D. Radlein and T. Bouchard, A Preliminary Look at the Economics of a New Biomass Conversion Process by Dynamotive, Dynamotive Report (2009), accessed in 2010 from www.dynamotive.com. [78] R. Wang and K. Smith, The Effect of Preparation Conditions on the Properties of HighSurface Area Ni2P Catalysts, Appl. Catal., A 380 (2010) 149-164. [79] T. Ma and R. Rittner, Modern Organic Elemental Analysis, Marcel Dekker, Inc, 1979. [80] S. Hediger, B. Meier and R. Ernst, Adiabatic Passage Hartmann-Hahn Cross Polarization in NMR Under Magic Angle Sample Spinning, Chem. Phys. Lett. 240 (1995) 449-56. [81] B. Fung, A. Khitrin and K. Ermolaev, An Improved Broadband Decoupling Sequence for Liquid Crystals and Solids, J. Magn. Reson. 142 (2000) 97-101. [82] W. Earl and D. Vanderhart., Measurement of Carbon-13 Chemical Shifts in Solids, J. Magn. Reson. 48 (1982) 35-54. [83] D. Massiot, F. Fayon, M. Capron, I. King, C. Le Stephanie, B. Alonso, J. Durand, B. Bujoli, Z. Gan and G. Hoatson, Modeling One- and Two-Dimensional Solid-State NMR Spectra, Magn. Reson. Chem. 40 (2002) 70-76. [84] E. Dietrich, C. Mathieu, H. Delmas and J. Jenck, Raney-Nickel Catalyzed Hydrogenations: Gas-Liquid Mass Transfer in Gas-Induced Stirred Slurry Reactors, Chem. Eng. Sci. 47 (1992) 3597-3604. [85] R. Albal, Y. Shah, N. Carr and A. Bell, Mass Transfer Coefficients and Solubilities for Hydrogen and Carbon Monoxide Under Fischer-Tropsch Conditions, Chem. Eng. Sci. 39 (1984) 905-907. [86] F. Sotoodeh, L. Zhao and K. Smith, Kinetics of H2 Recovery from Dodecahydro-NEthylcarbazole over a Supported Pd Catalyst, Appl. Catal., A 362 (2009) 155-162. [87] A. Vuori, A. Helenius and J. Bredenberg, Influence of Sulphur Level on Hydrodeoxygenation, Appl. Catal. 52 (1989) 41-56. 182  [88] C. Lee and D. Ollis, Interactions Between Catalytic Hydrodeoxygenation of Benzofuran and Hydrodesulfurization of Dibenzothiophene, J. Catal. 87 (1984) 332-338. [89] F. Cotton and G. Wilkinson, Advanced Inorganic Chemistry (A Comprehensive Text). Ed. 3, Interscience Publishers, New York, NY, 1972. [90] T. Matsuda, Y. Hirata, H. Itoh, H. Sakagami and N. Takahashi, Effect of Reduction Temperature on the Transformation of MoO3 to MoOx with a Large Surface Area, Micropor. Mesopor. Mater. 42 (2001) 337-344. [91] T. Matsuda, Y. Hirata, F. Uchijima, H. Itoh and N. Takahashi, Characteristics of MoO3 Reduced with H2 at the Different Flow Rates of H2, Bull. Chem. Soc. Jpn. 73 (2000) 10291034. [92] T. Ressler, R. Jentoft, J. Wienold, M. Gunter and O. Timpe, In Situ XAS and XRD Studies on the Formation of Mo Suboxides during Reduction of MoO3, J. Phys. Chem. B 104 (2000) 6360-6370. [93] R. Burch, Preparation of High Surface Area Reduced Molybdenum Oxide Catalysts, J. Chem. Soc., Faraday Trans. 1 74 (1978) 2982-2990. [94] J. Sloczyn´ski and W. Bobin´ski, Autocatalytic Effect in the Processes of Metal Oxide Reduction. I. Kinetic Model of the Reduction, J. Solid State Chem. 92 (1991) 420-435. [95] J. Sloczyn´ski, Autocatalytic Effect in MoO3 Reduction in Hydrogen, React. Solids 7 (1989) 83-88. [96] J. Aigler, M. Houalla and D. Hercules, Surface Structure and Metathesis Activity of Photoreduced Allyl-Based Mo/SiO2 Catalysts, Top. Catal. 10 (2000) 123-126. [97] G. Ilangovan and K. Pillai, Electrochemical and XPS Characterization of Glassy Carbon Electrode Surface Effects on the Preparation of a Monomeric Molybdate(VI)-Modified Electrode, Langmuir 13 (1997) 566-575. [98] F. Al-Kharafi, H. Al-Kandari and A. Katrib, Different Catalytic Reactions of n-Hexane and 1-Hexene on Molybdenum Based Catalysts, Catal. Lett.123 (2008) 269-275. [99] C. Li, Z. Xu, Z. Cao, B. Gates and L. Petrakis, Hydrodeoxygenation of 1-Naphthol Catalyzed by Sulfided Nickel-Molybdenum/γ-Alumina: Reaction Network, AIChE J. 31 (1985) 170-174. [100] M. Daage and R. Chianelli, Structure-Function Relations in Molybdenum Sulfide Catalysts: The "Rim-Edge" Model, J. Catal. 149 (1994) 414-427. 183  [101] S. Kasztelan, H. Toulhoat, J. Grimblot and J.P. Bonnelle, A Geometrical Model of the Active Phase of Hydrotreating Catalysts, Appl. Catal. 13 (1984) 127-159. [102] E. Hensen, P. Kooyman, Y. van der Meer, A. van der Kraan, V. de Beer, J. van Veen and R. van Santen, The Relation Between Morphology and Hydrotreating Activity for Supported MoS2 Particles, J. Catal. 199 (2001) 224-235. [103] Y. Lee and S. Oyama, Bifunctional Nature of a SiO2-Supported Ni2P Catalyst for Hydrotreating: EXAFS and FTIR Studies, J.Catal. 239 (2006) 376-389. [104] D. Saihi, I. Vroman, S. Giraud and S. Bourbigot, Microencapsulation of Ammonium Phosphate with a Polyurethane Shell. Part II. Interfacial Polymerization Technique, React. Funct. Polym. 66 (2006) 1118-1125. [105] Z. Hanafi, M. Khilla and M. Askar, The Thermal Decomposition of Ammonium Heptamolybdate, Thermochim. Acta 45 (1981) 221-32. [106] M. Barbooti and D. Al-Sammerrai, Thermal Decomposition of Citric Acid, Thermochim. Acta 98 (1986) 119-26. [107] L. Grondahl, S. Suzuki and E. Wentrup-Byrne, Influence of a Diene Impurity on the Molecular Structure of Phosphate-Containing Polymers with Medical Applications, Chem. Commun. (Cambridge, U. K. ) (2008) 3314-3316. [108] M. Stefanescu, M. Stoia, O. Stefanescu and P. Barvinschi, Obtaining of Ni0.65Zn0.35Fe2O4 Nanoparticles at Low Temperature Starting from Metallic Nitrates and Polyols, J. Therm. Anal. Calorim. 99 (2010) 459-464. [109] J. Ryczkowski, Spectroscopic Evidences of EDTA Interaction with Inorganic Supports During the Preparation of Supported Metal Catalysts, Vib. Spectrosc. 43 (2007) 203-209. [110] J. Ryczkowski, IR studies of EDTA Alkaline Salts Interaction with the Surface of Inorganic Oxides, Appl. Surf. Sci. 252 (2005) 813-822. [111] A. Samotus, A. Kanas, M. Dudek, R. Grybos and E. Hodorowicz, 1:1 Molybdenum(VI) Citric Acid Complexes, Transition Met. Chem. (London) 16 (1991) 495-9. [112] Z. Zhou, H. Wan and K. Tsai, Syntheses and Spectroscopic and Structural Characterization of Molybdenum(VI) Citrato Monomeric Raceme and Dimer, K4[MoO3(cit)]·2H2O and K4[(MoO2)2O(Hcit)2]·4H2O, Inorg. Chem. 39 (2000) 59-64. [113] M. Aramendía, V. Borau, C. Jiménez, J. Marinas and F. Romero, Synthesis and Characterization of Magnesium Phosphates and Their Catalytic Properties in the Conversion of 2-Hexanol, J. Colloid Interface Sci. 217 (1999) 288-298. 184  [114] J. Fischer, L. Merwin and R. Nissan, NMR Investigation of the Thermolysis of Citric Acid, Appl. Spectrosc. 49 (1995) 120-6. [115] A. Silva, X. Kong and R. Hider, Determination of the pKa Value of the Hydroxyl Group in the α-Hydroxycarboxylates Citrate, Malate and Lactate by 13C NMR: Implications for Metal Coordination in Biological Systems, Biometals 22 (2009) 771-778. [116] S. Schantz, P. Hoppu and A. Juppo, A Solid-State NMR Study of Phase Structure, Molecular Interactions, and Mobility in Blends of Citric Acid and Paracetamol, J. Pharm. Sci. 98 (2009) 1862-1870. [117] P. Atanasova and T. Khalachev, Influence of Phosphorus Concentration on the Type and Structure of the Compounds Formed in the Oxide Form of Phosphorus-NickelMolybdenum/Alumina Catalysts for Hydrodesulfurization, Appl. Catal. 48 (1989) 295-306. [118] S. Damyanova, L. Gomez, M. Banares and J. Fierro, Thermal Behavior of 12Molybdophosphoric Acid Supported on Zirconium-Loaded Silica, Chem. Mater. 12 (2000) 501-510. [119] S. Dowker and J. Elliott, Infrared Study of Trapped Carbon Dioxide in Thermally Treated Apatites, J. Solid State Chem. 47 (1983) 164-73. [120] R. Mohareb, F. El-Omran and J. Ho, The Reaction of 2-Amino-3-Cyano-4,5,6,7Tetrahydrobenzo[b]-thiophene with Diethyl Malonate: Synthesis of Coumarin, Pyridine, and Thiazole Derivatives, Heteroat. Chem. 12 (2001) 168-175. [121] R. Mohareb, N. Louca, G. Elmegeed and H. Hana, Synthesis of Novel Tryptophan Derivatives of Potential Biological Activity, J. Chil. Chem. Soc. 54 (2009) 175-179. [122] A. Abbotto, S. Bradamante and G. Pagani, Charge Mapping in Carbanions. Weak Charge Demand of the Cyano Group as Assessed from a Carbon-13 NMR Study of Carbanions of α-Activated Acetonitriles and Phenylacetonitriles: Breakdown of a Myth, J. Org. Chem. 58 (1993) 449-55. [123] P. Clark and S. Oyama, Alumina-Supported Molybdenum Phosphide Hydroprocessing Catalysts, J. Catal. 218 (2003) 78-87. [124] L. Dake, D. Baer and D. Friedrich, Auger Parameter Measurements of Phosphorus Compounds for Characterization of Phosphazenes, J. Vac. Sci. Technol., A 7 (1989) 1634-8. [125] R. Gresch, W. Mueller-Warmuth and H. Dutz, X-Ray Photoelectron Spectroscopy of Sodium Phosphate Glasses, J. Non-Cryst. Solids 34 (1979) 127-36.  185  [126] B. Coughlan and M. Keane, An X-Ray Diffraction Line-Broadening Study of Copper Metal Supported on a Range of Y Zeolites, Zeolites 11 (1991) 854-7. [127] L. Nassimbeni, M. Niven, J. Cruywagen and J. Heyns, Complexation Between Molybdenum(VI) and Citrate: Crystal and Molecular Structure of [Mo4O11(citrate)2](Me3N(CH2)6NMe3)2.12H2O, J. Crystallogr. Spectrosc. Res. 17 (1987) 373-82. [128] A. Pashigreva, G. Bukhtiyarova, O. Klimov, Y. Chesalov, G. Litvak and A. Noskov, Activity and Sulfidation Behavior of the CoMo/Al2O3 Hydrotreating Catalyst: The Effect of Drying Conditions, Catal. Today 149 (2010) 19-27. [129] R. Zhang, Z. Zhou and H. Wan, Solid and Solution Structural Studies of Dimeric and Tetrameric Molybdenum(VI) Malate Complexes, Chin. J. Struct. Chem. 27 (2008) 919-926. [130] R. Takahashi, S. Sato, T. Sodesawa, M. Suzuki and N. Ichikuni, Ni/SiO2 Prepared by Sol-Gel Process Using Citric Acid, Micropor. Mesopor. Mater. 66 (2003) 197-208. [131] P. Clark, W. Li and S. Oyama, Synthesis and Activity of a New Catalyst for Hydroprocessing: Tungsten Phosphide, J. Catal. 200 (2001) 140-147. [132] S. Oyama, X. Wang, Y. Lee and W. Chun, Active Phase of Ni2P/SiO2 in Hydroprocessing Reactions, J. Catal. 221 (2004) 263-273. [133] P. Clark, X. Wang, P. Deck and S. Oyama, Push-Pull Mechanism of Hydrodenitrogenation over Silica-Supported MoP, WP, and MoS2 Hydroprocessing Catalysts, J. Catal. 210 (2002) 116-126. [134] Y. Zhao, Y. Zhao, H. Feng and J. Shen, Synthesis of Nickel Phosphide Nano-Particles in a Eutectic Mixture for Hydrotreating Reactions, J. Mater. Chem. 21 (2011) 8137-8145. [135] M. Rokita, M. Handke and W. Mozgawa, The AlPO4 Polymorphs Structure in the Light of Raman and IR Spectroscopy Studies, J. Mol. Struct. 555 (2000) 351-356. [136] K. Senevirathne, A. Burns, M. Bussell and S. Brock, Synthesis and Characterization of Discrete Nickel Phosphide Nanoparticles: Effect of Surface Ligation Chemistry on Catalytic Hydrodesulfurization of Thiophene, Adv. Funct. Mater. 17 (2007) 3933-3939. [137] A. Padwa, W. Dent, A. Schoffstall and P. Yeske, Silyl Group Transfer in the Cycloaddition Reactions of Silyl Iminium Salts Derived from Aryl-Substituted Oximes, J. Org. Chem. 54 (1989) 4430-4437.  186  [138] S. Oyama, X. Wang, Y. Lee, K. Bando and F.G. Requejo, Effect of Phosphorus Content in Nickel Phosphide Catalysts Studied by XAFS and Other Techniques, J. Catal. 210 (2002) 207-217. [139] H. Wang, Y. Shu, M. Zheng and T. Zhang, Selective Hydrogenation of Cinnamaldehyde to Hydrocinnamaldehyde over SiO2 Supported Nickel Phosphide Catalysts, Catal Lett. 124 (2008) 219-225. [140] Y. Lee, Y. Shu and S. Oyama, Active Phase of a Nickel Phosphide (Ni2P) Catalyst Supported on KUSY Zeolite for the Hydrodesulfurization of 4,6-DMDBT, Appl. Catal., A 322 (2007) 191-204. [141] Y. Yoshimura, T. Sato, H. Shimada, N. Matsubayashi and A. Nishijima, Influences of Oxygen-Containing Substances on Deactivation of Sulfided Molybdate Catalysts, Appl. Catal. 73 (1991) 55-63. [142] T. Choudhary and C. Phillips, Renewable Fuels via Catalytic Hydrodeoxygenation, Appl. Catal., A 397 (2011) 1-12. [143] Y. Miyake, A. Baylaucq, C. Zeberg-Mikkelsen, G. Galliero, H. Ushiki and C. Boned, Stereoisomeric Effects on Volumetric Properties Under Pressure for the System cis-+ transDecalin, Fluid Phase Equilib. 252 (2007) 79-87. [144] R. Walpole, R. Myers, S. Myers and K. Ye, Probability and Statistics for Engineers & Scientists. Ed. 7, Prentice Hall, Upper Saddle River, NJ, 2002. [145] R. Lemoine and B. Morsi, An Algorithm for Predicting the Hydrodynamic and Mass Transfer Parameters in Agitated Reactors, Chem. Eng. J. 114 (2005) 9-31. [146] R. Perry and D. Green, Perry’s Chemical Engineering Handbook. Ed. 7, Mc-Graw Hill, New York, NY, 1997. [147] National Institute of Standards and Technology (NIST), Chemistry Web Book, 2008. [148] J. Benitez, Principles and Modern Applications of Mass Transfer Operations, Wiley, 2009. [149] H. Fogler, Elements of Chemical Reaction Engineering, Prentice Hall, Upper Saddle River, NJ, 2006. [150] P. Englezos and N. Kalogerakis, Applied Parameter Estimation for Chemical Engineers, Marcel-Dekker, New York, NY, 2001.  187  [151] H. Rezaei, S. Ardakani, and K. Smith, Study of MoS2 Catalyst Recycle in Slurry-Phase Residue Hydroconversion, Energy Fuels 26 (2012) 6540-6550.  188  Appendices  189  Appendix A Sample Calculations  A.1  Economic Analysis for BINGO Process  The details for the economic analysis for Dynamotive’s biomass into gasoil (BINGO) process are given in Figure A.1 and were obtained from Radlein and Bouchard [77].  Figure A.1. Cost analysis of BINGO process [77]. 190  A.2  MoP Preparation  The MoP-CA catalysts were prepared with a 1:1 ratio of Mo:P and a 2:1 ratio of CA:Mo. 4 g of AHM ((NH4)6Mo7O24.4H2O) was mixed with a corresponding amount of AHP ((NH4)2HPO4) according to the calculation below:  molAHM 7 molMo molP molAHP 132 g AHP = 4 g AHM ( )( )( )( )( ) = 2.99 g AHP 1236 g AHM molAHM molMo molP molAHP  The amount of CA required for preparation was calculated based on the equation below:  molAHM 7 molMo 2 molCA 192 g CA = 4 g AHM ( )( )( )( ) = 8.70 g CA 1236 g AHM molAHM molMo molCA  A.3  Catalyst Concentration  Sample calculations for the amount of catalyst required for the reaction of 4-methylphenol over MoP-CA-773 K using 3,900 ppmw of catalyst are as follows:  Mass of 4-methylphenol reactant  = 2.77 g  Volume of reactant solvent decalin  = 100 mL  191  1. Required amount of MoP-CA-773 K catalyst per gram of solution:  = 3,900 ppmMoP =  3,900 g MoP 1,000,000 g solution  =  0.0039 gMoP gsolution  2. Solution mass:  = 2.77 g 4MP + 100 mLdecalin (𝜌DECALIN )  The density of decalin was determined from an equation of state from Miyake et al. [143] at ambient conditions. Therefore, the solution mass becomes:  0.896 g  = 2.77 g 4MP + 100 mL (  mL  ) = 92.37 g solution  3. Required amount of MoP:  =  A.4  0.039 gMo gsolution  (92.37 g solution ) = 0.36 g MoP  BET  The BET equation used for multi-point analysis is given in Eq. A.1.  𝑃⁄𝑃0  𝑃 𝑉(1− ) 𝑃0  =  𝐶−1  𝑃  ( )+  𝑉𝑚 𝐶 𝑃0  192  1 𝑉𝑚 𝐶  (A.1)  where C is a constant, V is the total volume adsorbed at pressure P, Vm is the volume adsorbed at monolayer coverage (STP), and P0 is the saturation vapor pressure of the adsorbate gas (760 mmHg). The left side of the equation was then plotted versus P/P0 (valid up to P/P0 ~ 0.4) to give a linear plot of slope (C-1)/VmC and intercept 1/VmC. The surface area was then calculated using Eq. A.2.  𝑆BET = 4.37𝑉𝑚  (A.2)  where the 4.37 constant is related the average molecular area of the adsorbate (N2) at 77 K. This equation is valid for Vm in units of cm3. The single point BET analysis was done using Eq. A.3.  𝑃  𝑉𝑚 = 𝑉 (1 −  A.5  𝑃0  )  (A.3)  XRD Lattice Parameters  The lattice parameters (ao and co for a hexagonal cell) in Table 5.4 were calculated using Eq. A.4.  1 𝑑hkl  2  =  4 (h2 + hk+k2 ) 3  𝑎o  2  +  193  1 𝑐o 2  (A.4)  where dhkl is the interplanar lattice spacing of the (hkl) plane and h, k, and l are the numerical values. The interplanar lattice spacings for the (hkl) planes (i.e. (101) and (100)) were then calculated using the Bragg’s Law in Eq. A.5.  𝑑hkl =  𝜆  (A.5)  2sin𝜃hkl  where θhkl is the Bragg angle for the (hkl) plane (degree) and 𝜆 is the XRD wavelength of radiation (nm).  A.6  Kinetic Analysis  From the plot of ln(1-x) versus time the pseudo first order kinetic parameter for the HDO of 4-methylphenol, k4MP, over MoP-CA-773 K at 623 K and 4.4 MPa, was determined to be 4.82E-5 s-1. Please note that this pseudo first order kinetic parameter accounts for the thermal blank. The kinetic parameter normalized by the catalyst concentration was then calculated using Eq. A.6:  𝑘=  𝑘4𝑀𝑃  (A.6)  𝐶𝑐𝑎𝑡  100 mL  𝑘 = (4.82E − 5 s−1 ) (  0.36 gcat  194  )  Accounting for the C content of the catalyst the kinetic parameter becomes:  −1 ) 𝑘 = (1.33E − 2 mL. s−1 . g cat (  g cat 60 s )( ) = 0.924 mL. s−1 . g −1 MoP 0.87 g MoP min  The initial rate for the HDO of 4-methylphenol was then calculated as:  Initial Rate = 𝑘𝐶𝑎𝑜  2.77 g4MP  Initial Rate = (0.924 mL. s −1 . g −1 MoP ) (  100 mL  (A.7)  mol  ) (108.1 g  1,000 mmol  4MP  )(  mol  )  Initial Rate = 0.24 mmol. min−1 . g −1 MoP  The initial TOF for the HDO of 4-methylphenol over MoP-CA-823 K at 623 K and 4.4 MPa was calculated as follows using the CO uptake and assuming a 1:1 binding of CO:metal, where the CO uptake = 54 µmol/gMoP:  Initial TOF =  Initial Rate  (A.8)  CO Uptake  g  1,000 µmol  MoP Initial TOF = (0.24 mmol. min−1 . g −1 MoP ) (54 µmol) (  Initial TOF = 0.073 s−1 195  mmol  min  ) (60 s)  A.7  Metal Site Density  The metal site density of the MoP-CA catalysts was determined, assuming a spherical geometry. The surface area is related to the volume, area, and density according to Eq. A.9 and Eq. A.10.  4π𝑟 2  Area  Surface Area = Volume.𝜌 = 4 3  π𝑟 3 𝜌  =  6  (A.9)  𝑑p 𝜌  so that, 𝑑p =  6  (A.10)  Surface Area.𝜌  Using these equations the surface area and/or particle size was calculated. The MoP active site area is 9.71 atoms/nm2 and the density is 7.5 g/cm3 [43, 123]. The metal site density of the MoP-CA catalysts was then calculated using their SBET, dXRD, and dTEM data. Shown below is the calculation for the MoP-CA-823 K metal site density based on the TEM particle size using Eq. A.11.  Surface Area =  6  g 6.1 nm.7.5 3 cm  m3  (10E+6 cm3 ) (  10E+9 nm  m2  m  gMoP  ) = 131  The surface area was then corrected for the amount of C in the MoP-CA-823 K catalyst so that the area was stated per gcat.  196  Surface Area = 131  m2 gMoP  (0.91  gMoP gcat  m2  ) = 119 g  cat  Metal Site Denity = Surface Area(Active Site Area)/𝑁A  = 119  (A.11)  m2 9.71 atoms mol 10E + 18 nm2 10E + 6 µmol ( ) ( ) ( )( ) g cat nm2 6.022E + 23 atoms m2 mol  = 1916  µmol g cat  197  Appendix B Error Analysis and Repeatability  B.1  Catalyst Characterization Repeatability  The standard deviation was calculated to observe the variability in the measured data. The standard deviation (SD) is given in Eq. B.1.  2 ∑𝑚 𝑖=1(𝑋𝑖 −𝑋𝑎𝑣𝑔 )  SD = √  (B.1)  𝑚−1  where Xi is the sample variable, Xavg is the average sample value, and m is the number of samples. The square of the SD is defined as the variance σv.  The repeatability experiments for the measured BET surface area (SBET) of MoP-CA-773 K catalysts are given in Table B.1. This data was based on both single and multi-point BET analysis.  Table B.1. Error associated with the surface area measurements of MoP-CA-773 K.  SBET (m2/gcat) 130 139 138 Xavg  136  SD  5 6.7  % SD/ Xavg 198  The repeatability experiments for the measured CO uptake data of the MoP-CA-823 K catalysts are given in Table B.2.  Table B.2. Error associated with the CO uptake measurements of MoP-CA-823 K.  CO Uptake (µmol/gMoP) 59 61 62 Xavg  60  SD  2 3.3  % SD/ Xavg  Table B.3 displays the error associated with the CHN analysis of Cal-MoP-CA-773 K.  Table B.3. Error associated with the CHN measurements of Cal-MoP-CA-773 K.  C (wt.%)  H (wt.%)  N (wt.%)  24.1  0.9  8.1  21.0  1.5  6.7  20.4  0.7  6.6  21.7  1.3  7.1  21.7  1.4  7.1  Xavg  21.8  1.2  7.1  SD  1.4  0.3  0.6  % SD/ Xavg  6.5  27.4  8.3  199  Table B.4 displays the error associated with the XRD and crystallite size analysis of MoPCA-823 K for the (101) and (110) planes.  Table B.4. Error associated with XRD analysis of MoP-CA-823 K.  B.2  d101 (nm)  d110 (nm)  20.9  16.1  19.1  16.1  18.0  16.1  Xavg  19.3  16.1  SD  1.5  0.0  % SD/ Xavg  7.77  0.0  Sampling Repeatability  To observe the error and accuracy associated with GC sampling, the initial concentration of 4-methylphenol for six separate samples was quantified by using the GC. It can be observed from the data in Table B.5 that the error is low.  200  Table B.5. Error associated with GC injections – quantification of 4-methylphenol concentration.  4MP Concentration (mol/L) 0.250 0.255 0.256 0.255 0.254 0.256 Xavg  0.254  SD  2.25E-3 0.98  % SD / Xavg  To determine if the data was accepted a tα-test was applied, where if the computed tα-value fell between –t0.05 and t0.05 the test was satisfied. The tα-value was calculated using Eq. B.2.  𝑡𝛼 =  𝑋𝑎𝑣𝑔−𝜇𝑣 (  𝑆𝐷 ) √𝑚  (B.2)  where µv is the population mean at 0.256 mol/L. The critical ± t0.05 values were obtained from Walpole et al. [144] using tα-table distributions where the degrees of freedom, v, was taken as m-1 so that t0.05= ± 2.015. Using Eq. B.2 the calculated tα-value of -1.814 was satisfied. Therefore, the data was accepted.  201  B.3  Reaction Repeatability  To observe the HDO experiment repeatability the reaction of 4-methylphenol over Ni2P-CA773 K at 623 K (4.4 MPa) was performed for three separate trials. The rate versus time data is shown below in Figure B.1.  0.5  -1  -1  HDO Rate (mmol/min .gNi2P )  0.4  0.3  0.2  0.1  0.0 0  50  100  150  200  250  300  350  Time (min)  Figure B.1. HDO rate versus time for the 4-methylphenol decomposition over Ni2P-CA-773 K at 623 K and 4.4 MPa (3 trials).  From the data it can be observed that the experiment is repeatable. The kinetic parameters obtained from Figure B.1 for the decomposition of 4-methylphenol are given in Table B.6. The standard deviation and calculated error are also provided in Table B.6.  202  Table B.6. Kinetic and deactivation parameters for the HDO of 4-methylphenol over Ni2P-CA-773 K at 623 and 4.4 MPa.  k  kd  (mL.min−1.gNi2P−1) (mL.min−1.gNi2P−1) 2.13  1.23  2.21  1.18  2.00  1.18  Xavg  2.11  1.20  SD  1.06E-1  2.89E-2  % SD/ Xavg  5.02  2.41  The product selectivity repeatability versus time for the DDO and HYD products over  60  60  50  50  DDO Product Selectivity (%)  DDO Product Selectivity (%)  Ni2P-CA-773 K at 623 K and 4.4 MPa are given in Figure B.2.  40  30  20  10  0  40  30  20  10  0 50  100  150  200  250  300  50  Time (min)  100  150  200  250  300  Time (min)  Figure B.2. DDO and HYD product selectivities versus time for reaction at 623 K (4.4 MPa) over Ni2PCA-773 K (3 trials).  203  B.4  Carbon Recovery  The C recovery for each experiment should be 100%. However, due to experimental and sampling errors the error in the carbon recovery could be significant. Therefore, the C recovery was investigated for the reaction of 4-methylphenol over MoP-CA-823 K at 623 K (4.4 MPa). Table B.7 displays the product and reactant concentrations over time for the reaction of 4-methylphenol over MoP-CA-823 K at 623 K (4.4 MPa). It can be observed from the data that the carbon recovery ranged from 96-98%. The loss in C was likely due to sampling and experimental errors. Since the initial 4-methylphenol concentration was low, small deviations in the recovered carbon showed large errors.  Table B.7. Reactant and product concentrations for reaction of 4-methylphenol over MoP-CA-823 K at 623 K (4.4 MPa), as well as carbon recovery.  Time  4-  (min)  Methylphenol  Toluene Methylcyclohexane (mol/L)  (mol/L)  (mol/L)  1,3-  C  Dimethylcyclopentane  Recovered  (mol/L)  (%)  0  0.256  0  0  0  --  63  0.163  4.36E-2  3.97E-2  0  96.1  113  0.160  4.49E-2  3.95E-2  1.46E-3  96.0  171  0.138  6.08E-2  5.45E-2  2.25E-3  99.6  229  0.113  6.71E-2  6.29E-2  3.03E-3  96.2  300  5.01E-2  9.22E-2  8.64E-2  4.16E-3  90.9  204  B.5  Pyrolysis Oil HDO  The error associated with the repeatability of the pyrolysis oil HDO reactions is given in Table B.8. Overall, the error was low with the exception of the gas yield.  Table B.8. Error observed for the HDO of pyrolysis oil over MoS2 at 523 K.  Coke  Viscous  Gas  Hydrocarbon  Aqueous Liquid  Yield (wt.%)  Oil (wt.%)  (wt.%)  Liquid (wt.%)  (wt.%)  43.7  25.1  10.1  2.6  18.5  45.6  30.4  4.3  2.6  17.1  36.9  31.4  6.2  3.1  22.5  Xavg  42.1  29.0  6.9  2.8  19.4  SD  4.6  3.4  3.0  0.3  2.8  % SD/ Xavg  10.9  11.6  43.1  9.9  14.5  Table B.9 displays the error associated with the CH analysis of the solid coke fraction obtained from the HDO of fast pyrolysis oil over MoS2 at 523 K.  Table B.9. Error in CH measurements of solid product over MoS2 for the HDO of pyrolysis oil at 523 K.  C (wt.%)  H (wt.%)  57.6  5.3  57.7  5.4  57.6  5.3  Xavg  57.6  5.3  SD  3.46E-2  7.00E-2  % SD/ Xavg  6.0E-2  1.3  205  Appendix C Gas Chromatography A 14-A Shimadzu GC/FID with a HELIX AT-5, (Alltech Part No. 16857) capillary column (Stationary phase 5% phenyl – 95% methylpolysiloxane) was used for sample analysis. The total sample injected was 0.1 µL. The injector temperature was set to 513 K and the detector temperature was set to 553 K. Both the polarity and range were set to “2.” The carrier gas was H2 and was set to 500 kPa for carrier P, carrier M was set to 50 kPa, H2 for combustion was set to 40 kPa, and air for combustion was set to 30 kPa [air:H2; 300 mL/min:30 mL/min; 10:1]. The temperature profile of the GC program is given in Table C.1.  Table C.1. GC/FID temperature profile.  Parameter  Value  Initial Temperature (K)  323  Hold Time (min)  5  Temperature Rate (K/min)  3  Temperature (K)  333  Hold Time (min)  5  Temperature Rate (K/min)  5  Temperature (K)  147  Hold Time (min)  1  Temperature Rate (K/min)  20  Temperature (K)  533  Hold Time (min)  1  206  The following parameters were used for the data processor; width = 5, slope = 70, drift = 0, min. area = 200, t.dbl = 0, stop time = 28 min, atten = 4, speed = 10, method = 0041, format = 0000, spl.wt = 100, is.wt = 1, window = 5 Known standards were injected into the GC/FID to determine their retention time and concentration versus area profile. From the 4-methylphenol reactant concentration (Ca), and the obtained area from GC/FID analysis (Ao), the response factor (RF) was calculated:  𝑅𝐹 =  𝐴𝑜  (C.1)  𝐶𝑎  Similarly, the response factor for the products toluene, methylcyclohexane, 1methylcyclohexene, 3-methylcyclohexene, and the isomerization products was determined from Eq. C.1. To confirm the compound analysis a Shimadzu QP-2010-S equipped with an electron impact (EI) mass spectrometer was used. The ramp method used to analyze the liquid reaction samples is given in Table C.2. The column used for analysis was a non-polar fused silica column (5% phenyl, 95% dimethylpolysiloxane- 30 m × 0.25 mm × ID 0.25 μm – Restek RTX5) with a temperature range of 213-573 K. Additional information about the parameters used for analysis are given in Table C.2.  207  Table C.2. GC/MS operational parameters.  Parameter  Value  Unit  Injection Temperature  513  K  Pressure  51.3  kPa  Total Flow Carrier (He)  258  mL/min  Column Flow  1.02  mL/min  Linear Velocity  36.5  cm/s  1  mL/min  Ion Source Temperature  473  K  Interface Temperature  573  K  Gas Chromatograph  Purge Flow Mass Spectrometer  208  Appendix D Mass Transfer Effects The volumetric liquid side mass transfer coefficient, kLa, was calculated using Eq. D.1. This equation was obtained from Dietrich et al. [84] for stirred tank bench top reactors operated in slurry phase; using the stirrer diameter as the characteristic length.  𝑘𝐿𝑎 =  𝑆ℎ.𝐷𝐴𝐵 2 𝑑𝑆𝑇  kLa  = volumetric liquid-side mass transfer coefficient, (s-1)  Sh  = Sherwood number, (dimensionless)  DAB  = diffusivity of A in B, (m2/s)  dST  = stirrer diameter, (m)  (D.1)  The Sherwood number was calculated using a correlation by Albal et al. [85].  𝑆ℎ = 0.0141𝑅𝑒 0.67 𝑆𝑐 0.5 𝑊𝑒 1.29  Sc  = Schmidt number, (dimensionless)  Re  = Reynold’s number, (dimensionless)  We  = Weber number, (dimensionless)  (D.2)  The Schmidt number is the ratio of the momentum diffusivity to the mass diffusivity.  209  𝑆𝑐 =  μ  = viscosity, (Pa.s)  ρ  = density, (kg/m3)  𝜇  (D.3)  𝜌𝐷𝐴𝐵  The Reynolds number is ratio of viscous forces to inertial forces  𝑅𝑒 =  . N  2 𝜌𝑁𝑑𝑆𝑇  𝜇  (D.4)  = stirrer speed, (s-1)  The Weber number is a dimensionless number used to correlate a measure of the fluid’s inertia to its surface tension [145].  𝑊𝑒 =  γ  3 𝜌𝑁 2 𝑑𝑆𝑇  𝛾  (D.5)  = surface tension, (N/m)  The diffusivity of H2 in decalin was obtained from Perry and Green [146], and the National Institute of Standards and Technology [147].  The diffusivity, as well as the solvent  properties are given in Table D.1 [148]. For simplicity, the system mass transfer effects were investigated at atmospheric conditions. 210  Table D.1. Properties of decalin [146-148].  Parameter  Nomenclature  Value  Unit  Density  ρ  896  kg/m3  Viscosity  µ  2.39E-3  Pa.s  Surface Tension  γ  3.10E-2  N/m  DAB  1.93E-9  m2/s  H2 Diffusion Coefficient  The reactor geometric properties are displayed in Table D.2.  Table D.2. Reactor geometric properties.  Parameter  Nomenclature  Value  Unit  Diameter  d  0.0445  m  Height  hR  0.2032  m  Stirrer Shaft Diameter  dST  0.0127  m  Total Volume  VT  315  mL  Liquid Volume  VL  100  mL  VG = VT - VL  215  mL  Liquid Height  hL = 4VL/(πd2)  0.1651  m  Stirrer Speed  N  17  s-1  Gas Volume  The calculated parameters are given in Table D.3 . Table D.3. Mass transfer coefficient for stirred batch reactor.  Weber Reynold’s Schmidt Sherwood  kLa  We  Re  Sc  Sh  (s-1)  17  1041  1364  3650  0.0455  211  Typical kLa numbers that are illustrative in industrial units are often of the 0.1 s-1 magnitude, and rarely exceed 0.3s-1. Thus, the calculated value of kLa = 0.0437 s-1 is fairly representative of a real system and will increase as the temperature is increased. The calculated mass transfer coefficient was much greater than the kinetic parameter k4MP = 2.33E-5 s-1 for crystalline MoS2. Therefore, external mass transfer is negligible. As an example, the properties of the crystalline MoS2 catalyst are given in Table D.4.  Table D.4. MoS2 catalyst properties.  Catalyst  Kinetic Parameter – k4MP  Particle Radius - Rpar  (s-1)  (nm)  2.33E-5  280  Crystalline MoS2  To determine the influence of internal diffusion the Thiele modulus, Φ1, was calculated for the first order reaction. The Thiele modulus is the surface reaction rate to the internal diffusion rate and is given by Eq. D.6  𝑘4𝑀𝑃  𝜙1 = 𝑅𝑝𝑎𝑟 ( 𝐷  Rpar  = radius of particle, (m)  Deff  = effective diffusivity, (m2/s)  𝑒𝑓𝑓  0.5  )  (D.6)  The effective diffusivity is defined in Eq. D.7 and is used to describe the average diffusivity: 212  𝐷𝑒𝑓𝑓 = 𝐷𝐴𝐵 (  εp  = particle porosity, (dimensionless)  σc  = constriction factor, (dimensionless)  θT  = tortuosity factor, (dimensionless)  𝜀𝑝 𝜎𝑐 𝜃𝑇  )  (D.7)  For simplicity it was assumed that the effective diffusivity was 0.2DAB [149]. The effectiveness factor, η, is the ratio of actual rate of the reaction to the rate of reaction that would result if the entire interior surface were exposed to the external pellet surface conditions and is given by Eq. D.8.  3  𝜂 = (𝛷2 ) (𝛷1 coth𝛷1 − 1) 1  (D.8)  The overall effectiveness factor, Ω, is the ratio of the overall rate of reaction to the rate that would results if the entire surface were exposed to the bulk concentration and is given by Eq. D.9.  𝛺=  𝜂 𝜂𝑘 (1+ 4𝑀𝑃 ) 𝑘𝐿𝑎  213  (D.9)  The calculated properties are displayed in Table D.5. As shown from the results, for MoS2 Φ1 < 0.4, and η = Ω = 1. Therefore, for this system there were negligible internal and external mass transfer resistances so that the reaction was kinetically controlled [149].  Table D.5. MoS2 catalyst Thiele modulus and effectiveness factors.  Catalyst  Thiele  Effectiveness  Overall  Modulus  Factor  Effectiveness  Φ1  η  Factor Ω  Crystalline MoS2  7.0E-4  1.00  214  0.99  Appendix E Kinetic Model Code  E.1  Estimated Kinetic Parameter Error  The standard error calculation for the kinetic parameters was done using the method of Englezos et al. [150]. The kinetic parameters are reported as k* ± 𝑡𝛼𝑣 𝜎𝑘𝑖 . 2  The covariance matrix (COV(k*)) distribution was calculated using Eq. E.1:  COV(𝑘 ∗ ) = σ2ε [𝐴∗ ]−1  (E.1)  A* was calculated at k* and was obtained by Eq. E.2.  𝑇 𝐴∗ = ∑𝑚 𝑖=1 𝐺𝑖 𝑄𝑖 𝐺𝑖  (E.2)  where Q is a weighting matrix and was taken as the unit matrix. G is given by Eq. E.3 for i number of rate functions and j number of kinetic parameters.  𝜕𝑓  𝐺𝑖𝑗 = (𝜕𝑘 ) 𝑘𝑗 𝑗  σε2 is the variance and was calculated using Eq. E.4:  215  (E.3)  𝑆 (𝑘 ∗ )  𝜎𝜀 2 = 𝑛𝑚−𝑃  (E.4)  S(k*) is the objective function evaluated at k values, n is the number of dependent variables, m is the number of data points and P is the number of parameters.  2 𝑆(𝑘 ∗ ) = ∑𝑚 𝑖=1(𝑋𝑖 − 𝑥𝑖 )  (E.5)  where m is the number of data points, Xi is the experimental response, and xi is the model response. The standard error associated with the estimated kinetic parameter is denoted as σki and was calculated using the diagonal elements of the A* matrix, defined as:  𝜎𝑘𝑖 = 𝜎𝜀 √{[𝐴∗ ]−1 }𝑖𝑖  (E.6)  The ki values were reported as:  𝑘𝑖∗ − 𝑡𝛼𝑣 𝜎𝑘𝑖 ≤ 𝑘𝑖 ≤ 𝑘𝑖∗ + 𝑡𝛼𝑣 𝜎𝑘𝑖 2  (E.7)  2  where the degrees of freedom was, v = (nm-P). The tα-values for each data set were obtained from Walpole et al. [144] for 95% confidence so that the t0.025 values were taken at ± 2.060 or ± 2.080 depending on the data set used.  216  E.2  Main Body Code  The following codes were used to model the hydrodeoxygenation of 4-methylphenol over MoP-CA-773 K at 623 K and 4.4 MPa H2. A modified leasqr function using LevenbergMarquardt nonlinear regression was used to calculate the kinetic parameter solutions and their errors. ===============================================================  clear all global nvar nx x0 y0 global verbose global n1 n2 n3 n4 H2 verbose(1:2) = 1; % x is the indep varaibale vector e.g. time measurements % y is matrix of responses % columns of y are responses y1, y2 (e.g. mol frac of component 1 and 2) % rows of y are y values at the value of the indep variable (time) in x % first row of y is initial value of response % the program uses the Levenberg-Marquardt method to estimate parameters % and calc statistics - done in leasqr and dfdp % the L-M requires the model to be calculated -this is done in modelmulti.m % and assumes the model is a series of ODEs, with the number of odes is equal % to the number of responses. The ODEs are calculated in ODEfunm. Note that % this function must use the correct model for each y % time data (min) T =[0 51 108 150 193 240 300] nt=length (T) x(1:nt-1)=T(2:nt) x nx=length(x) % 4-methylphenol concentration data (mol/L) CAX =[0.256 0.213 0.186 0.163 0.158 0.144 0.108]; 217  % toluene concentration data (mol/L) CBX = [0 0.027 0.042 0.057 0.059 0.070 0.089]; % methylcyclohexane concentration data (mol/L) CCX = [0 0.014 0.024 0.033 0.036 0.0414 0.055]; % isomerization concentration data (mol/L) CDX = [0 0.0016 0.0036 0.0026 0.0033 0.0039 0.0051]; for j=1:nt-1 y1(j)=CAX(j+1); y2(j)=CBX(j+1); y3(j)=CCX(j+1); y4(j)=CDX(j+1); end nvar=4; x0=0.; oldx=x; nx = length(x); y = [y1' y2' y3' y4'] %y = [y1' y2' y3'] newy=y(:); oldy=reshape(newy,nx,nvar); 218  x=x' newx=[x;x;x;x;] %newx=[x;x;x;] y01(1:nx)=CAX(1); y02(1:nx)=CBX(1); y03(1:nx)=CCX(1); y04(1:nx)=CDX(1); newy0=[y01';y02';y03'; y04']; %newy0=[y01';y02';y03']; %INPUT DATA NOW IN CORRECT COLUMN FORMAT y0=newy0 x=newx y=newy % provide initial parameter guesses theta=[0.0002 0.001 0.009] ; np=length(theta) pin=theta nxcheck=size(x) nycheck=size(y) % Begin calculation by calling L-M least squares routine [f,p,kvg,iter,corp,covp,covr,stdresid,Z,r2]=leasqr(x,y,pin,'modelmulti', 0.00001, 10000); disp('RESPONSE:') if kvg ==1 disp ('PROBELM CONVERGED') elseif kvg == 0 disp('PROBLEM DID NOT CONVERGE') end oldf=reshape(f,nx,nvar); oldr=reshape(y-f, nx, nvar); disp ('X-values:') disp (oldx') disp ('Y-values') disp(oldy) disp('f-values - i.e. model calculated responses') disp(oldf) disp('Residuals:') disp (oldr) % disp ('Standardized residuals:') % disp (stdresid) disp ('Estimated parameter values are;') disp (p) 219  disp ('Covariance of estimated parameters - sqrt of diagonal gives CL') disp (covp) disp('R2 values is:') disp (r2) plot (oldx,oldy,'d'), hold, plot (oldx,oldf) figure subplot(3,3,1) plot(oldx(:),oldy(:,1),'o',oldx(:),oldf(:,1),'--') title('4MP') xlabel('Time (min)') ylabel('Concentration (mol/L)') subplot(3,3,2) plot(oldx(:),oldy(:,2),'o',oldx(:),oldf(:,2),'--') title('TOLUENE') xlabel('Time (min)') ylabel('Concentration (mol/L)') subplot(3,3,3) plot(oldx(:),oldy(:,3),'o',oldx(:),oldf(:,3),'--') title(METHYLCYCLOHEXANE') xlabel('Time (min)') ylabel('Concentration (mol/L)') subplot(3,3,4) plot(oldx(:),oldy(:,4),'o',oldx(:),oldf(:,4),'--') title('ISOMERIZATION') xlabel('Time (min)') ylabel('Concentration (mol/L)') ===============================================================  E.3  ModelMulti Code  function f = modelmulti (x,pin) % Solve a simple system of ODE's - response variables % find the solution at sepcified x values - corresponding to measured data % first data point in x corresponds to initial condition global nvar nx x0 y0 global verbose nxx=length(x); yzero=reshape(y0,nx,nvar); for i = 1:nx xf = x(i); xoo=x0; 220  yzed=yzero(i,:); [xmodel,ymodel] = ode45 (@ODEfunm,[xoo,xf], yzed,[],pin); yfinal(i,:)=ymodel(end,:); end f = yfinal(:); ===============================================================  E.4  Ordinary Differential Equation Code  function yprime=ODEfunm(xatx,yatx,pin) global nvar nx x0 y0 xstep % disp('*****************************YPRIME') % disp (knt) % nx k1=pin(1); k2=pin(2); k3=pin(3); %ntest=(knt/nx);  yp(1)=-k1*yatx(1)-k2*yatx(1)-k3*yatx(1); yp(2)=k1*yatx(1); yp(3)=k2*yatx(1); yp(4)=k3*(yatx(1)); yprime =[yp(1)';yp(2)';yp(3)';yp(4)']; ===============================================================  E.5  Calculation of Jacobian Matrix  function prt=dfdp(x,f,p,dp,func) % numerical partial derivatives (Jacobian) df/dp for use with leasqr m=length(x);n=length(p); %dimensions ps=p; prt=zeros(m,n);del=zeros(n,1); % initialize Jacobian to Zero for j=1:n del(j)=dp(j) .*p(j); %cal delx=fract(dp)*param value(p) if p(j)==0 del(j)=dp(j); %if param=0 delx=fraction end p(j)=ps(j) + del(j); 221  if del(j)~=0, f1=feval(func,x,p); if dp(j) < 0, prt(:,j)=(f1-f)./del(j); else p(j)=ps(j)- del(j); prt(:,j)=(f1-feval(func,x,p))./(2 .*del(j)); end end p(j)=ps(j); %restore p(j) end return ===============================================================  E.6  Statistical Analysis of Kinetic Model  Both one way analysis of variance (ANOVA) and F-test analysis were done to observe the error between the kinetic model and the experimental data. In both cases the single factor was set to 0.05, i.e. 95% confidence interval. The sum of squares within groups (SSWG) for the ANOVA calculation was calculated using Eq. E.8.  2 𝑆𝑆𝑊𝐺 = ∑𝑚 𝑖=1 𝑋𝑖 −  (∑𝑚 𝑖=1 𝑋𝑖 )  2  𝑚  2 + ∑𝑚 𝑖=1 𝑥𝑖 −  (∑𝑚 𝑖=1 𝑥𝑖 )  2  𝑚  (E.8)  where Xi is the experimental response and xi is the response obtained from the proposed model.  The total sum of squares between groups (SST) was calculated using Eq. E.9.  𝑚 2 2 𝑆𝑆𝑇 = ∑𝑚 𝑖=1 𝑋𝑖 + ∑𝑖=1 𝑥𝑖 −  222  2  [∑𝑚 𝑖=1(𝑋𝑖 + 𝑥𝑖 )] 𝑚1 +𝑚2  (E.9)  The sum of squares between groups (SSBG) was calculated using Eq. E.10.  𝑆𝑆𝐵𝐺 = 𝑆𝑆𝑇 − 𝑆𝑆𝑊𝐺  (E.10)  The degrees of freedom (df) for the ANOVA analysis was calculated between groups (dfBG = z – 1)], where z is the total number of data sets. The degrees of freedom was also determined within groups dfWG = ∑𝑧𝑖=1(𝑚 − 1)𝑖 . The mean square value between groups and within groups was then calculated using Eq. E.11 and Eq. E.12, respectively.  𝑀𝑆𝐵𝐺 =  𝑀𝑆𝑊𝐺 =  𝑆𝑆𝐵𝐺 𝑑𝑓𝐵𝐺  𝑆𝑆𝑊𝐺 𝑑𝑓𝑊𝐺  (E.11)  (E.12)  The FANOVA value was calculated using Eq. E.13. The critical ± F0.05 values were obtained from Walpole et al. [144] using F-table distributions with degrees of freedom dfBG and dfWG.  𝐹ANOVA =  𝑀𝑆𝐵𝐺 𝑀𝑆𝑊𝐺  (E.13)  In addition, a simple F-test was performed using Eq. E.14, where σvi is the variance of each data set (i.e. experimental and model). 223  𝐹=  𝜎𝑣1  (E.14)  𝜎𝑣2  Shown in Table E.1 is the experimental and kinetic model 4-methylphenol concentration data for the HDO of 4-methylphenol over MoP-CA-823 K at 623 K and 4.4 MPa.  Table E.1. Experimental and model 4-methylphenol concentration data for the HDO reaction over MoP823 K at 623 K and 4.4 MPa.  Time Experimental  Model  (min)  (mol/L)  (mol/L)  0  0.256  0.256  63  0.173  0.202  113  0.170  0.167  171  0.139  0.134  229  0.123  0.107  300  7.35E-2  8.17E-2  0.156  0.158  3.74E-3  4.13E-3  Xavg σv  The ANOVA analysis of the experimental and model 4-methylphenol concentration data is given in Table E.2. As shown from the data the FANOVA value was < Fcrit, therefore the confidence interval was satisfied and the model was accepted.  224  Table E.2. ANOVA analysis of the 4-methylphenol concentration data for the HDO reaction over MoP823 K at 623 K and 4.4 MPa.  Source of Variation  SS  df  MS  FANOVA  Fcrit  Between Groups  1.511E-5  1  1.5E-5  3.84E-3  4.96  Within Groups  3.93E-2  10  3.94E-3  Total  3.94E-2  11  The F-test was also done using both the experimental and model 4-methylphenol data. The critical ± F0.05 values were obtained from Walpole et al. [144] using F-table distributions where the degrees of freedom, v1 and v2, were taken as m-1 for each data set so that Fcrit = ± 5.05. Using Eq. E.14 the calculated F-value of 0.91 was satisified so that F < Fcrit. Therefore, the model was accepted. ANOVA analysis was also applied to the toluene concentration data as shown in Table E.3.  225  Table E.3. Experimental and model toluene concentration data for the HDO reaction over MoP-823 K at 623 K and 4.4 MPa.  Time Experimental  Model  (min)  (mol/L)  (mol/L)  0  0  0  63  4.36E-2  2.76E-2  113  4.49E-2  4.56E-2  171  6.08E-2  6.25E-2  229  6.71E-2  7.61E-2  300  9.22E-2  8.90E-2  Xavg  5.14E-2  5.01E-2  σv  9.48E-4  1.08E-3  The ANOVA analysis of the experimental and model toluene concentration data is given in Table E.4. As shown from the data the FANOVA value was < Fcrit, therefore the confidence interval was satisfied and the model was accepted.  Table E.4. ANOVA analysis of the toluene concentration data for the HDO reaction over MoP-823 K at 623 K and 4.4 MPa.  Source of Variation  SS  df  MS  FANOVA  Fcrit  Between Groups  5.06E-6  1  5.06E-6  4.99E-3  4.96  Within Groups  1.01E-2  10  1.01E-3  Total  1.01E-2  11  The F-test was also done using both the experimental and model toluene concentration data. The critical ± F0.05 values were obtained from Walpole et al. [144] using F-table distributions 226  where the degrees of freedom, v1 and v2, were taken as m-1 for each data set so that Fcrit = ± 5.05. Using Eq. E.14 the calculated F-value of 0.88 was satisified so that F < Fcrit. Therefore, the model was accepted. ANOVA analysis was also applied to the methylcyclohexane concentration data as shown in Table E.5.  Table E.5. Experimental and model methylcyclohexane concentration data for the HDO reaction over MoP-823 K at 623 K and 4.4 MPa.  Time Experimental  Model  (min)  (mol/L)  (mol/L)  0  0  0  63  3.97E-2  2.57E-2  113  3.95E-2  4.20E-2  171  5.45E-2  5.82E-2  229  6.29E-2  7.09E-2  300  8.64E-2  8.30E-2  Xavg  4.72E-2  4.66E-2  σv  8.35E-4  9.36E-4  The ANOVA analysis of the experimental and model toluene concentration data is given in Table E.6. As shown from the data the FANOVA value was < Fcrit, therefore the confidence interval was satisfied and the model was accepted.  227  Table E.6. ANOVA analysis of the methylcyclohexane concentration data for the HDO reaction over MoP-823 K at 623 K and 4.4 MPa.  Source of Variation  SS  df  MS  FANOVA  Fcrit  Between Groups  8.6E-7  1  8.6E-7  9.70E-4  4.96  Within Groups  8.86E-3  10  0.00089  Total  8.86E-3  11  The F-test was also done using both the experimental and model toluene concentration data. The critical ± F0.05 values were obtained from Walpole et al. [144] using F-table distributions where the degrees of freedom, v1 and v2, were taken as m-1 for each data set so that Fcrit = ± 5.05. Using Eq. E.14 the calculated F-value of 0.89 was satisified so that F < Fcrit. Therefore, the model was accepted. The combination of these tests confirmed that the proposed kinetic model fit well to the experimental data of both the reactant and product concentrations. Therefore, the model was accepted with 95% confidence.  228  Appendix F Independent Reaction Data for Kinetic Mechanism The data for the reaction of toluene over MoS2 at 623 K and 4.4 MPa H2 is given in Table F.1. The initial concentration used was based on that obtained from the conversion of 4methylphenol to toluene. Based on the data obtained in Table F.1, toluene was found to slowly hydrogenate to methylcyclohexane at the reaction conditions. However, the concentration of methylcyclohexane produced was considered to be negligible. Therefore, it was assumed that toluene was a stable product at the reaction conditions.  Table F.1. Toluene conversion over MoS2 at 623 K and 4.4 MPa H2.  Time (min)  Toluene  Methylcyclohexane  (mol/L)  (mol/L)  0  5.20E-2  0.0  50  5.20E-2  3.31E-5  181  5.22E-2  1.88E-4  241  5.21E-2  4.36E-4  295  5.20E-2  5.73E-4  The data for the reaction of methylcyclohexane over MoS2 at 623 K and 4.4 MPa H2 is given in Table F.2. The initial concentration used was based on that obtained from the conversion of 4-methylphenol to methylcyclohexane. Based on the data obtained in Table F.2, methylcyclohexane was found to be a stable product at the reaction conditions. No other reaction products were observed for this reaction.  229  Table F.2. Methylcyclohexane conversion over MoS2 at 623 K and 4.4 MPa H2.  Time (min)  Methylcyclohexane (mol/L)  0  4.30E-2  44  4.30E-2  105  4.28E-2  126  4.29E-2  185  4.30E-2  261  4.30E-2  The data for the reaction of 4-methylcyclohexene over MoS2 at 623 K and 4.4 MPa H2 is given in Table F.3. The initial concentration used was based on that obtained from the conversion of 4-methylphenol to 4-methylcyclohexene. Based on the data obtained in Table F.3,  4-methylcyclohexene  was  found  to  convert  to  1-methylcyclohexene  and  methylcyclohexane.  Table F.3. 4-Methylcyclohexene conversion over MoS2 at 623 K and 4.4 MPa H2.  Time (min)  4-Methylcyclohexene  1-Methylcyclohexene  Methylcyclohexane  (mol/L)  (mol/L)  (mol/L)  0  4.39E-2  0  0  67  1.16E-2  1.86E-2  1.51E-2  116  6.54E-3  1.27E-2  2.51E-2  186  3.57E-3  7.36E-3  3.43E-2  254  0  4.75E-3  4.09E-2  230  The data for the reaction of 1-methylcyclohexene over MoS2 at 623 K and 4.4 MPa H2 is given in Table F.4. The initial concentration used was based on that obtained from the conversion of 4-methylphenol to 1-methylcyclohexene. Based on the data obtained in Table F.4, 1-methylcyclohexene was found to convert to 4-methylcyclohexene, methylcyclohexane, and isomerization products including dimethylcyclopentane and ethylidenecyclopentane.  Table F.4. 1-Methylcyclohexene conversion over MoS2 at 623 K and 4.4 MPa H2.  Time  1-Methylcyclohexene  4-Methylcyclohexene  Methylcyclohexane  Isomerization  (min)  (mol/L)  (mol/L)  (mol/L)  (mol/L)  0  3.95E-2  0  0  0  33  1.56E-2  9.26E-3  1.26E-2  0  106  1.24E-2  6.11E-3  1.92E-2  2.48E-3  194  6.00E-3  0  3.12E-2  1.90E-3  249  0  0  3.43E-2  4.04E-3  231  Appendix G Additional Experiments The activity and selectivity for the HDO of 4-methylphenol over 5% Pd/SiO2, 18% NiMoS/Al2O3, 10% MoOx/SiO2, and zero valent Mo metal was investigated at the lower temperature range of 598 K and 4.1 MPa. From the results (Table G.1) it was determined that the 5% Pd/SiO2 was a very active catalyst that displayed strong hydrogenating abilities. Both Mo and 10% MoOx/SiO2 were less active and displayed a high selectivity to toluene. The sulfided NiMo/Al2O3 was fairly active and displayed a similar activity to unsupported MoS2.  232  Table G.1. Product selectivities over Mo and 5% Pd/SiO2 after 20% reactant conversion at 598 K (4.1 MPa).  Catalyst  k  Toluene  Methylcyclohexane  1 & 4-  Isomerization  (mL.min-1.gcat-1)  Selectivity  Selectivity  Methylcyclohexene  Selectivity  (%)  (%)  Selectivity (%)  (%)  5% Pd/SiO2  1.83  33.4  61.5  5.1  --  18% NiMoS/Al2O3  0.77  63.7  34.4  --  1.9  10% MoOx/SiO2  0.15  82.9  10.3  2.9  3.9  Mo  8.31E-2  71.6  17.3  9.1  2.0  233  

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.24.1-0073967/manifest

Comment

Related Items