Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Synthesis, characterization and hydroprocessing activity of modified transition metal phosphides Abu, Ibrahim Inamah 2007

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

Item Metadata

Download

Media
831-ubc_2007-266670.pdf [ 11.97MB ]
Metadata
JSON: 831-1.0058993.json
JSON-LD: 831-1.0058993-ld.json
RDF/XML (Pretty): 831-1.0058993-rdf.xml
RDF/JSON: 831-1.0058993-rdf.json
Turtle: 831-1.0058993-turtle.txt
N-Triples: 831-1.0058993-rdf-ntriples.txt
Original Record: 831-1.0058993-source.json
Full Text
831-1.0058993-fulltext.txt
Citation
831-1.0058993.ris

Full Text

SYNTHESIS, CHARACTERIZATION A N D HYDROPROCESSING ACTIVITY OF MODIFIED TRANSITION M E T A L PHOSPHIDES  by  IBRAHIM INAMAH A B U  B.Sc. (Hons)(Chem. Eng.), The University of Sci. & Tech, Ghana, 1984 M . Sc. (Chem. Eng.), University o f Petr. & M i n . Saudi Arabia, 1991  A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in  THE FACULTY OF GRADUATE STUDIES (Chemical & Biological Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA February, 2007  ©  Ibrahim Inamah A b u , 2007  Abstract In the present study, modified transition metal phosphides (bulk and supported catalysts) were prepared by the temperature programmed reduction (TPR) o f the corresponding phosphate precursors. The catalysts were characterized by x-ray diffraction ( X R D ) , B E T surface area, X ray  Photoelectron  Spectroscopy  (XPS),  n-propyl amine  chemisorption, C O uptake  and  Transmission Electron Microscopy ( T E M ) . The activity o f the prepared catalysts were then tested using 4,6-dimethyldibenzothiophene ( 4 , 6 - D M D B T ) for hydrodesulftrrisation ( H D S ) and carbazole for hydrodenitrogenation ( H D N ) . A s well, selected catalysts were tested using Light Gas O i l ( L G O ) derived from Athabasca bitumen. The H D S o f 4 , 6 - D M D B T showed that small amounts o f C o used to modify M2P a n d  -  M o P produced high selectivity for the direct desulfurization ( D D S ) product dimethylbiphenyl ( D M B P ) . Coo.4Ni2P/Al203 showed high conversion with little cracked products.  Supported  Coo.4Ni2P/MCM showed almost complete conversion o f 4 , 6 - D M D B T with cracked products. Fluorination marginally increased the conversion o f C o o . 4 N i P / A l 0 3 with insignificant changes 2  2  in the product distribution and the addition o f Pt enhanced hydrogenation. The N i M o P (0 < x < 1.1) bulk series and supported Nio.33MoP were tested for the x  hydrodenitrogenation ( H D N ) o f carbazole at 523-583 K , 3 M P a and a range o f space velocity o f 8.6-61 x 1 0 mol/h gcat. Nio.ovMoP showed the highest bicyclohexyl ( B C H X ) selectivity among 2  the N i M o P . The improved selectivity was attributed to the enhanced C O uptake and acidity that " x  resulted i n increased hydrogenation o f carbazole to tetrehydrocarbazole which readily undergoes C - N bond cleavage on acid sites to produce B C H X . The H D S conversion obtained over the Nio.33MoP/Ab_03 using the L G O showed higher conversion than a commercial sulfided MM0/AI2O3. The results from the present study therefore indicate that addition o f small amounts o f C o to metal phosphides is beneficial as enhanced  selectivity to the hydrogenolysis route o f 4 , 6 - D M D B T is promoted. Similarly, addition o f small amounts o f N i is beneficial for the H D N o f carbazole. However i n this case, the presence o f N i increased the hydrogenation o f carbazole making it easier for the hydrogenolysis to occur.  Table of Contents Abstract Table of contents  iii  List of Tables  x  List of Figures  xiii  List of Abbreviations  xviii  Acknowledgement  xxii  Chapter 1 Introduction 1.1  Terminology  2  1.2  Some aspects o f a fixed bed reactor  6  1.2.1  Mass transfer i n a fixed bed reactor  6  1.2.2  Heat transfer i n a fixed bed reactor  7  1.3  Methods o f formulation to impove the hydoprocessing catalyst activity  8  1.4  Knowledge gap  11  1.5  Motivation  12  1.6  Objectives  13  Literature Review  14  Chapter 2 2.1  Sulfur containing compounds and the hydrodesulfurization ( H D S ) process  15  2.1.1  Types o f common sulfur compounds i n liquid fuels  15  2.1.2  Reactivity o f sulfur containing compounds  17  iii  2.1.3  Reaction pathways o f hydrodesulfiirization o f 4,6-dimethyldibenzothiophene  2.2  2.3  2.4  2.5  Nitrogen containing compounds and the H D N process  19 21  2.2.1  Nitrogen containing compounds in crude o i l  22  2.2.2  Difficulties o f H D N  27  Reaction networks and mechanism o f hydrodenitrogenation  29  2.3.1  Quinoline  28  2.3.2  Carbazole  30  Catalysts used for hydroprocessing  32  2.4.1  U s e o f transition-metal sulfides  33  2.4.2  U s e o f transition metal nitride/carbides  33  2.4.3  Use o f transition-metal phosphides  35  Developing a new phosphide catalyst for enhanced hydroprocessing  44  2.5.1  Promotional effect o f a second metal  47  2.5.2  Promotional effect o f a third metal component  48  2.5.3  Effects o f support  49  2.5.3.1  Use o f alumina  49  2.5.3.2  Use o f acidic support  49  2.5.3.3  Effect o f addition o f fluorine to alumina  50  2.6  Effect o f adding platinum  51  2.7  Effect o f process variables  52  iv  Chapter 3 3.1  Experimental  54  Preparation o f metal phosphides for hydrodesulfurization  54  3.1.1  Preparation o f bulk metal phosphides o f Co Ni2P and C o M o P  3.1.2  Preparation o f supported metal phosphide catalysts for  x  x  hydrodesulfurization o f 4,6-dimethyldibenzothiophene  55  56  3.2  Preparation o f N i x M o P as hydrodenitrogenation catalysts  58  3.3  Catalyst Characterization  59  3.3.1  Temperature Programmed Reduction (TPR)  3.3.2  Temperature Programmed Reduction using Tapered Element Oscillating Microbalance ( T E O M )  60  3.3.3  Powder X - r a y Diffraction ( X R D )  61  3.3.4  Brunnauer-Emmett-Teller ( B E T ) Surface Area  62  3.3.5  X - r a y Photoelectron Spectroscopy ( X P S )  62  3.3.6  Carbon monoxide ( C O ) uptake  62  3.3.7  n-Propyl amine (n-PA) chemisorption  64  3.3.8  Scanning Electron Microscopy-Energy Dispersive X - R a y  3.3.9 3.4  Emission ( S E M - E D X )  65  Transmission Electron Microscopy ( T E M )  65  Catalyst activity measurements for hydrodesulfurization o f 4,6-dimethyldibenzothiophene and hydrodenitrogenation o f carbazole  Chapter 4  4.1  59  65  Hydrodesulfurization over bulk and supported phosphides using 4,6-dimethyldibenzothiophene  68  Characterization o f bulk metal phosphides  68  v  4.1.1  X R D o f the prepared catalysts  69  4.1.2  T P R o f precursors  73  4.1.3  T E M o f prepared catalysts  77  4.1.4  Other catalyst properties  80  4.1.5  X P S o f prepared catalysts  80  4.2  Catalyst activity  86  4.3  Discussion on bulk phosphides used for H D S o f 4 , 6 - D M D B T  93  4.4  Properties and hydrodesulfurization o f 4,6-dimethyldibenzothiophene over supported catalysts  97  4.4.1  X R D o f modified AI2O3 supported phosphides  4.4.2  X R D o f Pt-Coo.4Ni P/Al 03 and P t - C o o . N i P / M 0 3 - W A  101  4.4.3  Properties o f Coo.4Ni P/AI2O3 prepared on different supports  102  4.4.4  Properties o f Pt-Coo.4Ni2P/Al 03 prepared on different supports  104  2  2  4  2  99 2  2  2  4.5  Supported catalysts activity using 4 , 6 - D M D B T  105  4.6  Kinetics o f H D S  110  4.7  Discussion o f H D S on supported metal phosphides  115  4.8  Used catalyst properties  118  4.9  Catalyst regeneration  121  Chapter 5  Hydrodenitrogenation over bulk and supported metal phosphides using carbazole  122  5.1.  Bulk N i M o P  123  5.2.  Catalyst characterization o f prepared metal phosphides  123  x  5.2.1  T P R o f bulk N i x M o P  123  vi  5.2.2  X R D o f bulk N i M o P  124  5.2.3  X P S o f bulk N i M o P  127  x  x  5.3  Catalyst activity o f bulk N i M o P  131  5.4  Discussion on H D N o f carbazole using bulk phosphides  134  5.5  Properties and hydrodenitrogenation study on Nio.33MoP  x  supported catalysts  138  5.5.1  X R D o f modified Nio.3 MoP/Al 03  140  5.5.2  Properties o f prepared supported metal phosphides for H D N  141  3  2  5.6  Activity o f N i o ^ M o P on different supports  142  5.7  Kinetics o f the hydrodenitrogenation o f carbazole  144  5.8  Discussion on hydrodenitrogenation o f supported phosphides  150  Hydroprocessing using Light Gas O i l over supported catalysts  154  6.1  Results and discussion using Light Gas O i l  155  6.2  Kinetics o f hydrodesulfurization and hydrodenitrogenation over selected  Chapter 6  catalysts  158  Conclusions and Recommendations  168  7.1  Conclusions  168  7.2  Recommendations  170  Chapter 7  References Appendix  172 A . l Establishing validity of plug flow in experimental set-up  190  A . 2 Parameters for testing plug flow  190  vii  A . 3 Mass transfer limitation  192  A . 4 Reactor isothermal operation  194  A . 5 Determination o f saturation vapor pressure  195  A . 6 Determination o f vapor and saturated pressures  196  Appendix  B  Examples of calculation for characterization data  197  B. 1  Catalysts preparation: Calculation o f required chemicals  197  B.2  T P R calibration  200  B.3  Degree o f reduction  201  B.4  Temperature programmed reduction o f transition metal phosphide precursors i n H 2 using T E O M  202  B.5  X R D o f precursors o f calcined metal phosphides  204  B.6  Determination o f crystallite sizes  205  B.7  Calculation o f lattice parameters  208  B.8  X P S Survey Scan  209  B.9  Determination and repeatability o f C O chemisorption  212  B.10  Repeatability o f n-propylamine titration o f acid sites on metal phosphides  214  Determination o f the specific consumption o f 4 , 6 - D M D B T and carbazole  216  C  Hydrodesulfurization experiments  220  C. 1  Summary o f hydrodesulurization experiments  220  C.2  Response Factor and sample calculation o f H D S activity  C.3  Example o f repeatability o f hydrodesulfurization experiments  B. l 1  Appendix  viii  ,  221 224  C.4  Calculation o f standard error (S.E.)  227  C. 5  Data for hydrodesulfurization experiments.  228  Appendix  D  Hydrodenitrogenation experiments  249  D. 1  Summary o f hydrodenitrogenation experiments  249  D.2  Response Factor and sample calculation o f H D N activity  250  D.3  Example o f repeatability o f hydrodenitrogenation experiments  254  D.4  Data for hydrodenitrogenation experiments  256  Appendix E Program for the Gaussian Newton-Raphson Parameter Estimation  ix  269  List of Tables 2.1  Representative sulfur compounds in liquid fuels  16  2.2  Comparison o f H D S conversion o f individual polyaromatic S-containing compounds using different solvents over sulfided C0M0/AI2O3  18  2.3  Representative nitrogen compounds in liquid fuels  24  2.4  Bond Energies between carbon heteroatoms in poly-atomic molecule  26  2.5  Summary of hydroprocessing activity o f metal phosphides  39  2.6  Summary o f activity o f metal phosphides promoted with C o or N i  41  4.1  Lattice parameters estimated from P X R D o f reduced catalysts  71  4.2  Summary of temperature programmed reduction data  4.3  Properties o f the prepared metal phosphides  4.4  Activities o f bulk metal phosphides for the H D S o f 4 , 6 - D M D B T measured at 583 K and 3.0 M P a H  4.5  „78 81  89  2  Comparison o f sulfur speciation in liquid product from H D S o f 4,6D M D B T measured at 583 K and 3.0 M P a H over N i P and Coo.o Ni P 2  2  8  2  91  4.6 A  Properties o f supports  103  4.6B  Properties o f the prepared Coo.4Ni P/Al 03 phosphide on different supports  103  4.7  Properties of the prepared Pt-Coo.4Ni P/Al 03 phosphide on different support  105  4.8  Activities o f supported metal phosphides for the H D S o f 4 , 6 - D M D B T measured  2  2  2  at 583 K and 3.0 M P a H  2  107  2  4.9  Estimated 1 order rate constant for hydrodesulfurization o f 4 , 6 - D M B T  114  5.1  Lattice parameters estimated from X R D o f reduced catalysts  126  5.2  Physiochemical properties of prepared Coo.4Ni P on different supported  st  2  metal phosphides  128 x  5.3  Activities o f bulk metal phosphides for the H D N o f carbazole measured at 583 K and 3.0 M P a H  132  2  5.4  Properties o f prepared Nio.33MoP/Al 03 and Nio.33MoP/MCM catalysts  5.5  Activities o f Nio 33M0P supported metal phosphides for the H D N o f  2  carbazole measured at 583 K and 3.0 M P a H 5.6  141  144  2  Estimated 1 order rate constants for the hydrodenitrogenation o f st  carbazole  148  6.1  Characteristics o f Light Gas O i l derived from Athabasca bitumen  155  6.2  Apparent kinetic parameters for H D S and H D N o f light gas o i l at different temperatures  6.3  159  Comparative apparent activation energy for H D S and H D N o f L G O over Nio.33MoP/y-Al 03, Coo.4Ni P/Al 03, Pt-Coo.4Ni P/Al 03 and commercial 2  2  2  2  2  sulfided catalysts  162  6.4  Measured catalyst atom ratios before and after reaction i n L G O  163  6.5  A t o m ratios determined by X P S o f supported metal phosphide catalysts before and after reaction in L G O  164  Al  Parameters o f catalyst and reactor  190  B1  Data for the crystallite sizes o f N i M o P  B2  Results o f the repeatability o f C O uptake on Coo.o8Ni P  213  B3  Results o f the n - P A repeatability using Coo.o8Ni P  215  CI  Summary o f experiments for hydrodesulfurization o f 4,6-dimethyl-  208  x  2  2  dibenzothiophene ( 4 , 6 - D M D B T - 3 0 0 0 ppm) over transition metal phosphides at 3.0 M P a , H pressure for 12 h time on stream 2  xi  220  C2  Results o f hydrodesulfurization o f 4 , 6 - D M D B T over Co .o8Ni P  222  C3  Repeatability o f hydrodesulfurization o f 4 , 6 - D M D B T using C o o . o s ^ P  224  Dl  Summary o f experiments for hydrodenitrogenation o f carbazole  0  2  (3000 ppm) over transition metal phosphides at 3.0 M P a , H 2 pressure for 12 h time on stream  249  D2  Results o f hydrodenitrogenation o f carbazole over N10.07M0P  252  D3  Repeatability o f hydrodenitrogenation over N10.07M0P using carbazole  254  xii  List of Figures 1.1  Schematic layout o f a fully integrated refinery  4  1.2  Commercial approach using SynSat process and Criterion/Lummus hydrotreating reactor technology  5  2.1  Reaction pathway for 4 , 6 - D M D B T over sulfided catalysts  20  2.2  H D N network o f quinoline over sulfided N i - M o / A b O s catalysts  29  2.3  H D N reaction pathways for carbazole over sulfided, reduced and nitrided Mo/Al 0 2  31  3  2.4  Crystal structures o f some transition metal phosphides  36  2.5  Schematic diagram o f the preparation o f metal phosphides  38  2.6  Alternative catalytic route for conversion o f 4 , 6 - D M D B T  47  3.1  Schematic diagram o f the fixed bed reactor  67  4.1  X-ray diffractograms o f reduced C02P, C o P and C o N i 2 P catalysts x  (A - N i i P , * - C o P , T - N i C o P )  70  4.2  X - r a y diffractograms o f reduced M o P and C00.07M0P catalysts  73  4.3  T P R o f Ni2P and C0XM2P catalyst precursors measured i n 10 % H2  2  5  2  i n A r at a rate o f 60 ml(STP)/min 4.4  75  T P R o f C02P, C o P , M o P and C00.07M0P catalyst precursors measured in 10 % H i n A r at a rate o f 60 ml(STP)/min  76  2  4.5  T E M micrographs o f bulk metal phosphides: (i) M 2 P and (ii) C o o . o s ^ P . Estimated d-spacing o f lattice fringes shown are ( A ) 5.0 A (B) 3.4 A and (C) 2.8 A corresponding to (1 0 0), (0 0 1) and (0 1 1) planes o f N i P 2  4.6  X P S o f (a) C o 2p region (b) P 2p region, and (c) N i 2p region o f xiii  79  Co Ni2P catalysts after reduction and passivation  83  x  4.7  X P S o f P 2p region and M o 3d region o f M o P and C00.07M0P catalysts after reduction and passivation  4.8  85  Correlation o f P / M ratio determined by X P S and n - P A : C O uptake ratio determined by adsorption for C o N i P (•), C00.07M0P ( • ) , C o P (o), C o P (0) x  2  2  with M o P ( A ) and N i P ( D ) as indicated  87  2  4.9  Conversion o f 4 , 6 - D M D B T and selectivity to D M B P over various metal phosphide catalysts at 583 K and 3.0 M P a H , plotted as a function o f C O uptake: 2  C o N i P (•), C00.07M0P ( A ) , C o P (0) and C o P (o) with M o P ( A ) x  2  2  and N i P ( D ) as indicated  92  2  4.10  X R D o f prepared metal phosphides on different acidic supports  100  4.11  X R D o f Pt- C o o . N i P / A l 0  101  4.12  Plot o f In ( 1 - X ) versus space time at 3.0 M P a using P t - C o N i P / A l O  4.13  Simplified reaction scheme for the H D S o f 4 , 6 - D M D B T  4.14  Correlation o f the experimental data versus the predicted from the model for  4  2  2  and Pt- C o o . N i P / A l 0 - W A  3  4  2  2  3  A  0 4  2  2  3  109 Ill  conversion o f D M D B T (0) and yields o f products ( D M B P : » ; M C H T : A ; DMBCH:«) 4.15  112  Plots o f In k versus 1000/T for the hydrodesulfurization o f 4 , 6 - D M D B T using P t - C o N i P / A l O 0 4  4.16  2  2  115  3  Plots o f Bransted acidity as a function o f both conversion and D M B P selectivity on all supported catalysts: N i P / A l 0 (•), 2  2  3  C o N i P / A l O ( A ) , C o o . N i P / A l 0 - F ( • ) , C o N i P / M C M (•) 0 4  2  2  3  4  2  2  3  0 4  2  Pt- C o N i P / A l O (•), Pt- C o . N i P / A l O - W A (A ) 0 4  4.17  2  2  3  0  4  2  2  3  Comparison o f the X R D profiles o f the fresh (reduced) and spent xiv  117  (after activity test) metal phosphides 4.18  119  Comparison o f the X P S o f the fresh (reduced) and spent (after activity studies) metal phophides  5.1  120  T P R o f calcined catalysts precursors ( N i M o P for 0.0 < x < 1.11) x  measured in 10 % H in A r at flowrate o f 60 ml(STP)/min  124  5.2  X-ray diffractograms o f all reduced N i M o P for 0.0 < x < 1.11  125  5.3  X P S spectra o f the N i 2P, M o 3d and P 2p o f the prepared phosphides after  2  x  reduction and passivation 5.4  129  Correlation o f P / M ratio determined by X P S and n - P A : C O uptake ratio determined by adsorption for N i M o P for 0.0 < x .< 1.11(«) and M o P ( A ) x  5.5  130  Conversion o f carbazole and selectivity to B C H X over various metal phosphide catalysts at 583 K and 3.0 M P a H , plotted as a function o f C O uptake: 2  N i M o P (•) 0 . 0 < x < 1.11 and M o P ( A )  134  5.6  S E M o f Nio.33MoP/Al 0  139  5.7  X R D o f Nio.33MoP/Al 0  x  5.8  2  2  A  0  5.10  140  3  Plot o f In ( 1 - X ) versus space time at 533 K , 3.0 M P a using Ni .33MoP/Al O  5.9  3  2  145  3  Simplified reaction network o f carbazole  146  Correlation o f the experimental data versus the predicted from the model for conversion o f carbazole (0) and yields o f products ( B C H X : * ; THCZ:A)  5.11  149  Plots o f In k versus 1000/T for the hydrodenitrogenation o f carbazole over Ni .33MoP/Al O 0  5.12  2  151  3  Plots o f Bronsted acidity as a function o f both conversion and D M B P  xv  selectivity on all supported catalysts:Coo.4Ni2P/Ai20 ( A ) , 3  C00.4N12P/AI2O3-F ( • ) , C o . N i P / M C M (•), 0  4  2  N10.33M0P/ai2o3 (•), N i 3 M o P / M C M (A)  152  03  6.1  Total sulfur conversions over selected catalysts using LGO at 613, 623 K, 633 K and 648 K. P = 8.8 MPa, L H S V = 2 h" , H to oil ratio = 600 ml/ml: 1  2  • Coo. Ni P/Al 03 4  2  •Pt-Coo. Ni P/Ai203 • Sulfided N i M o / A l 0  2  4  2  2  3  a Nio.3 MoP/Al 03 3  6.2  156  2  Total nitrogen conversions over selected catalysts using L G O at 613 K , 623 K 633 K and 648 K. P = 8.8 MPa, LHSV = 2 h" , H to oil ratio = 600 ml/ml: 1  2  • C00.4N12P/AI2O3 a Nio.3 MoP/Al 0 3  6.3  2  •Pt-Coo.4Ni2P/Al 03 A Sulfided N i M o / A l 0 2  2  3  157  3  Arrhenius plots for determining the apparent activation energy for the HDS (A) and HDN (B) of L G O over Ni . 3MoP/Y-Al2O and commercial 3  0 3  sulfided catalysts 6.4  161  Comparison of XRD diffractograms obtained for Nin.33MoP supported on AI2O3 and AI2O3-F, before and after reaction in L G O  6.5  165  XPS of Ni 2p and Mo 3d region for C00.4N12P/AI2O3 and Ni . MoP/Al O3 catalysts 0  33  before and after reaction with LGO Bl  2  167  TPR of transition metal phosphide precursors  202  B2  Mass change during TPR of M2P phosphide precursor using T E O M  203  B3  X R D patterns of calcined precursors of metal phosphides  204  B4  Integration of the most intense peak of the MoP diffractrogram  206  B5  XPS survey scan of Co .o8Ni P and C00.07M0P  211  B6  Repeatability of CO uptake using CO0.08N12P Conditions, CO flow =  0  2  xvi  0.3 ml/min, He(mix) = 30 ml/min, size o f sample loop = 1ml Run 1: C O uptake = 0.90 mmols/g, R u n 2: C O uptake = 0.99 mmols/g  212  B7  Repeated n - P A data using Coo.o8Ni P  214  B8  Arrhenius plots for determining the apparent activation energy for the  2  H D S ( A ) and H D N (B) o f L G O over ( • ) C o o . N i P / A l 0 , and (v) Pt4  2  2  3  Coo.4Ni P/Al 0  218  B9  Diagram o f the M C M - 4 1 crystallite  219  CI  Calibration curve for 4 , 6 - D M D B T used to determine the response  2  2  factor C2  221  Profile o f product ratio using N i P / A l 0 ( A ) and C00.4N12P/AI2O3 (B) 2  2  3  for the H D S o f 3000 ppm o f 4 , 6 - D M D B T at 583 K and 3.0 M P a Dl  Calibration curve for carbazole used to determine the response factor  D2  248  250  Calibration curve for B C H X used to determine the response factor  251  xvii  List of Abbreviations 4,6-DMDBT  Dimethyl dibenzothiophene refractory sulfur containing compound  4,6-DMTHDBT  Dimethyltetrahydro-dibenzothiophene  4,6-DMHHDBT  Dimethylhexahydro-dibenzothiophene  4,6-DMPHDBT  Dimethylpentahydro-dibenzothiophene  AASC  Aliphatic and non-heterocyclic aromatic sulfur compounds  AED  Atomic emission detector  AGO  Atmospheric Distilled Gas O i l  B.E  Binding Energy  BCHX  Bicyclohexyl  b/d  Barrels per day  BP  Biphenyl  BT  Benzothiophene  CBZ  Carbazole  CHB  Cyclohexylbenzene  CHCHE  3-cyclohexyl-cyclohexene  c  Heat capacity per unit mass o f fluid  p  CPMCH  Cyclopentylmethyl-cyclohexane  DBT  Dibenzothiophene  DDS  Direct desulfurization  DHQ  Decahydroquinoline  DMCHB  Dimethylcyclohexane benzene  DMBCH  Dimethylbicyclohexane  DMBP  Dimethyl biphenyl  xviii  EDX  Energy Dispersion Spectroscopy  EPA  Environmental Protection Agency  FCC  Fluid catalytic cracker  FID  Flame ionization detector  FTIR  Fourier Transfom Infrared Reflectance  G  Mass velocity o f fluid  GC-MS  Gas chromatograph-mass spectroscopy  GM  molal velocity (mol mixture/sec.cm o f total bed cross section)  h  Heat transfer coefficient,  HCH  Hexylcyclohexane  HCO  Hydrocracker Gas O i l  HDN  Hydrodenitrogenation  HDO  Hydrodeoxygenation  HDS  Hydrodesulfurization  HHCBZ  1,2,4,4a,9a-hexahydrocarbazole  ICP  Inductive Coupled Plasma  JD  Mass transfer group symbol  k  Thermal conductivity  2  k  c  Mass transfer coefficient  k  G  Mass transfer coeffiecient related to pressure  LCO  Light Cycle O i l  LHSV  Liquid hourly space velocity  LPG  Liquified Petroleum Gas  MCM-41  M o b i l Crystalline Material  xix  MPa  M e g a Pascal  n-PA  n-propyl amine  N  Prandtl number,  p r  OHCBZ  1,2,3,4,4a,9a-octahydrocarbazole  OPA  o-propylaniline  PDF  Powdered Diffraction Files  PHCBZ  Perhydrocarbazole  ppm  Parts per million  PV  Pore volume  PXRD  Powdered X-ray Diffraction  q  Heat flux  Q  Quinoline  RSH  Thiols  RSR  Sulfides  SA  B E T surface area  SBET  B E T surface area  SV  Space velocity  TCD  Thermal conductivity detector  TEM  Transmission electron microscopy  THCZ  1,2,3,4-tetrahydrocarbazole  T  Fluid stream temperature  0  TOF  Turn over frequency  TPD  Temperature programmed desorption  T  Surface temperature o f the pellet,  s  xx  TPR THQ1  Temperature programmed reduction 1,2,3,4-tetrahydroquinoline  VGO  Vacuum Gas O i l  WA  Weakly acidic  WHSV  Weight Hourly Space Velocity  XRD  X-ray diffraction  XPS  X-ray Photoelectron spectroscopy  y-AI2O3  A l s o AI2O3 (alumina)  s  V o i d fraction  p  Density o f fluid  X  Heat o f adsorption  xxi  Acknowledgement First I would like to express my sincere thanks and gratitude to my supervisor, Professor K e v i n J. Smith for the excellent guidance and support throughout the program. I w i l l always remember and respect his wise and consistent mentoring that w i l l guide me i n my future ambition. I would also like to specially thank my thesis committee members Professor C . J i m L i m , Professor Ajay K . Dalai, Professor E l o d Gyenge and Dr. John Adjaye for the advice and suggestions in completing o f this thesis. M y sincere thanks also go to D r . K e n W o n g o f the Advanced Material and Process Engineering Laboratory, Mary Mager from the Department o f Metals and Materials Engineering and L i n a Mandilao from the Wine Research Centre for the help with catalyst characterization. To the office staff, stores and workshop o f Chemical and Biological Engineering, thanks for helping. I want to also thank the Canadian Scholarship Secretariat and the Government o f Ghana for their financial support. Thanks to my wife Zaria, my kids, Anas, A d e l , A m a l and Abubakar for their patience, love and support. Finally I want to express my thanks to all my colleagues in the catalysis group.  xxii  Introduction  Chapter 1 Introduction Recently, demands for cleaner burning fuels have led to a reduction in the allowable sulfur and nitrogen content o f fuel. For example, i n 2010 the U S Environmental Protection Agency ( E P A ) allowable sulfur in gasoline, highway diesel and jet fuels currently at 330 ppm, 500 ppm and 3000 ppm respectively w i l l decrease to 30 ppm, 15 ppm and 5 ppm respectively (Song, 2003). Apart from environmental concerns, petroleum refineries are facing the challenge o f having to process heavier crude since the light and sweeter crude containing low contents o f S (0.5 wt%) and N (0.1 wt%) are fast diminishing. Furthermore, synthetic crude derived from o i l sands that contain large quantities o f S (2-7 wt%) and N (0.2-0.7 wt%) must be hydroprocessed to allowable S and N levels. In Western Canada, there is a large deposit o f oil sands covering an area o f 46,800 square kilometers containing 137 billion cubic meters (862 billion barrels) o f o i l (Montgomery, 1964). The changing environmental regulations and the need to process heavier crude means that there is tremendous pressure on refineries to improve processing technology in order to meet these challenges. The approaches employed to solve these problems include: (1) developing new processes such as sulfur adsorption from fuels; (2) tailoring reaction and process conditions; (3) designing new reactor configurations and (4) improving the catalytic activity by formulating new catalysts or improve existing ones. A m o n g these approaches, improving the catalytic activity by formulating new catalysts or improving the existing ones has received the most attention (Song, 2003). A new catalyst must be able to remove refractory sulfur compounds such as 4,6dimethyldibenzothiophene ( 4 , 6 - D M D B T ) and the non-basic N containing compounds such as  1  Introduction carbazole, as it is well known that the 100-500 ppm residual S and N o f conventionally hydroprocessed crude, contain these refractory compounds. However, before discussing the methods that are used to improve catalyst  activity, it is important  to understand  the  hydroprocessing process since any new catalyst formulated must be tested in the appropriate hydroprocessing environment.  1.1  Terminology: Before proceeding further,  some terminology w i l l be defined. Hydroprocessing o f  petroleum feedstock has been extensively practiced i n the petroleum industry and i n the primary upgrading o f heavy crudes and synthetic fuels. Hydroprocessing is also an integral part o f the production o f liquid fuels from coal and biomass. In the refinery, hydroprocessing involves a variety o f catalytic reactions such hydrogenation (that lead to saturation o f aromatics, olefins, etc) hydrocracking and removal o f heteroatoms such as sulfur (hydrodesulfurization-HDS), nitrogen, (hydrodenitrogenation-HDN),  oxygen,  (hydrodeoxygenation-HDO)  and  metals  (hydrodemetallization). These hydroprocessing reactions are classified according to the severity of the operation. They include hydrotreating, mild hydrocracking or hydroconversion and hydrocracking. In hydrotreating, essentially there is a small change i n the overall molecular structure such that the boiling points o f the different o i l fractions barely change. Typical process conditions for carrying out hydrotreating are 3-20 M P a H 2 and 523-673 K . Hydrocracking on the other hand, i n addition to removing the S, N , O and metals, also cracks the hydrocarbons to give lighter products and is carried out at high pressure (10 M P a H 2 ) and 673 K (Katzer and Sivasubramanian,  1979; Massoth, 1978; Topsoe and Massoth, 1996). M i l d hydrocracking  operates at moderate conditions to obtain low cracking and at the same time achieve some degree of hydrotreating.  2  Introduction Figure 1.1 shows a flowsheet o f a typical petroleum refinery and it can be seen how extensively hydroprocessing is applied (Topsoe and Massoth, 1996). Presently, there are 722 o i l refineries i n 116 countries with 202 o f the refineries located i n A s i a Pacific, 160 in North America and 105 in Western Europe. These refineries process about 50% (81.9 million b/d) o f the world crude distillation capacity (Nakamura, 2003; Prada, 2003). Hydroprocessing is also used to protect catalysts in other downstream processing since nitrogen containing compounds poison the catalysts used in these processes. W i t h the high volume o f hydropocessing applications, and based on the amount o f catalysts sold per year in the world market, hydrotreating catalysts rank third in catalyst production capacity after exhaust gas catalysts and fluid catalytic cracking catalysts (Reddy and Mastikhin, 1998). Conventionally, hydroprocessing is carried-out using fixed bed reactors with co-current supply o f oil and hydrogen that usually results in unfavorable H and H S concentration profiles 2  2  through the reactor. The removal o f the last ppm o f S is inhibited in this type o f configuration. Therefore a counter current operation is preferable. The process scheme o f a commercial hydotreating reactor, based on the SynSat and A B B Lummus reactor, is shown in Figure 1.2 (Maxwell, 1997) and is used by Scanreff and Preem refineries i n Sweden and Lyondell-Citgo refinery i n the U S . In this scheme, the o i l feed is introduced to the reactor at the top and hydrogen is introduced at the reactor outlet. The SynSat process merges the fields o f catalysis and reactor engineering and uses different catalyst beds within a single reactor shell with intermediate removal o f byproduct gas to achieve deep hydrodesulphurization. In this system, catalysts A and B are sulfided catalysts such as sulfided Ni-Mo/Al C»3. Catalyst C is a noble 2  metal on an acidic support.  3  Introduction  Gas and Light Gasoline  • LPG  Light Ends Plant  Butanes  Isomerization Plant Naphtha  Reformer  AJkylation Plant Crude oil  Isomerate Gasoline Reformate  Alkylate  Kerosene Diesel  -•Jet  E  5  Hydrocracker  AGO  Diesel  VGO Vacuum distillation Residuum  Catalytic cracker  'iqht r.ydP on  HCO  r«—  Coker  Hydrotreater Fuel oil  Asphalt Coke  Figure 1.1 Schematic layout o f a fully integrated refinery (Nakamura, 2002)  Between the two catalyst beds A and B , H 2 S and other gases are removed using a vaporliquid separator and nearly all the sulfur is converted and removed as H 2 S before the fuel feed reaches the noble-metal catalyst bed C (Maxwell, 1997) where hydrogenation takes place.  4  Introduction  Fresh feed  H gas 2  Catalyst A  Catalyst B Recycled liquid Diesel Product Catalyst C  Gas removal ( H 2 S , etc) A t V / L separator  M a k e - u p Hydrogen Diesel Product  Figure 1.2: Commercial approach using SynSat process and Criterion/Lummus hydrotreating reactor technology (Maxwel, 1997)  5  Introduction 1.2  Some aspects of a fixed bed reactor Since hydroprocessing uses fixed bed reactors (in the form o f trickle bed), some aspects  o f the design o f fixed bed reactors w i l l be discussed in this section.  1.2.1 Mass Transfer In a fixed bed reactor, flow passing over the catalyst pellet develops a boundary layer i n which the velocity parallel to the surface varies rapidly over a short distance normal to the flow. In the bulk stream, mixing occurs and reactants and products are transported at rates depending on the nature o f the flow. Since the flow passages i n fixed bed reactors are complex, correlations are developed for the mass transfer between the pellet and the flowing gas or liquid. The correlations are generally given in the form o f equation 1.1 (Froment, 1990; Rase, 1990)  N  s  h  = ^  = f(N ,N ) Ke  1.1  Sc  m  where, Nsh, N R and N s are the dimensionless Sherwood, Reynold's and Schmidt's numbers e  c  respectively, k is the mass transfer coefficient, d is the equivalent diameter and D c  p  m  is the  molecular diffusion coefficient for the diffusing species. The mass transfer from gas to liquids i n packed beds is given by equations 1.2 and 1.3 (Geankoplis, 1993). Vo =  9D=  M  -^r  for  5 5 < 7 V < 1500  1.2  0.0016<iV  1.3  R e  Re  <55  Re  6  Introduction where 6 is the void space between pellets as a fraction o f the total volume o f the bed and jo is a  grouped symbol =  where k  N'  2 3  G  sc  = ^  N'  14  2 3  G  sc  M  = k IR T, p is the fluid density, G is the mass velocity o f the fluid and G  G  c  M  is the  molal velocity (mol mixture/sec.cm o f total bed cross section). 2  1.2.2  Heat Transfer The mechanism o f heat transfer between the packed bed and the fluid is similar to the  mass transfer hence a similar correlation is given by equations 1.5 (Geankoplis, 1993). j  H  = - ^ N i : C  P  1.5  3  G  where  h = — - — T -T s  1.6  o  Cn M  k h is the heat transfer coefficient, G is the mass velocity, c is the heat capacity per unit mass o f p  fluid, N  p r  is the Prandtl number, T is the surface temperature o f the pellet, T s  0  is the fluid stream  temperature and k is the thermal conductivity. In a trickle bed reactor, the liquid flows cocurrently down the bed o f catalyst with the gas, however, counter current flow is also practiced. Cocurrent flow is desired because much better distribution o f the liquid over the catalyst is preferred and higher flow rates o f the liquid are possible without flooding. Note that in the refinery, counter current flow is preferred because there is a favorable concentration o f the H2 and H2S throughout the bed that allows the last ppm  Introduction of S to be removed and that is the main objective o f industry. Trickle bed reactors approach plug flow behavior. The mass and heat transfer effects discussed (as intrareactor, interphase or intraparticle gradients) can disguise the results obtained by using fixed-bed reactors and often lead to misinterpretation (Fogler, 1981, Mears, 1971, Dautzenberg, 1988). Therefore before accurate and intrinsic kinetic data can be established, these disguises must be eliminated by adjusting experimental conditions so as to obtain plug flow. Steps for minimizing the effects o f the internal and external mass transport were taken and are discussed in Appendix A . l - A . 4 . The kinetic equations for the fixed bed operations are also developed i n Chapter 4 and Appendix B . l 1.  1.3  Methods of catalyst formulation to improve the hydroprocessing activity In order to meet some o f the challenges outlined in section 1.1, several new catalysts  formulations have been prepared and tested in the last decade. In hydroprocessing, the function of the promoter is to enhance the catalyst activity by increasing conversion or selectivity through the interaction with the main metal. The interaction o f the promoter with the main metal is complex. For example, different explanations have been reported for the promoting effect o f C o on M o in sulfided C0M0/AI2O3 and a few w i l l be reported. Kabe et al. (1999) reported an increase in the labile S with the addition o f cobalt suggesting that C o makes S more mobile and thus creating more active sites. Other authors also report that the promotion effect o f C o was attributed to the decrease in the strength o f sulfur-molybdenum bond and thus facilitating the desorption o f H2S from the active sites (Kabe et al., 1998). Shuit et al. (1973) reported that C o (present as C o ) was assumed to be i n the tetrahedral positions in the surface o f the AI2O3, replacing A l  3 +  ions and that the promotional effect was due to an increase in the stability o f the  8  Introduction M o monolayer. The catalytically active sites were M o hydrogen from M o  ions, produced i n the presence o f  ions by removal o f some S " ions.  One approach is to improve the activity o f existing metal sulfide catalysts and there are many reports i n the literature in this regard (Lee et al. 2005; Mosio-Mosiewski and M o r a w s k i , (2005); K w a k et al., 1999; Bataille, 2001). The other approach is the investigation o f new materials such as carbides and bimetallic carbides (Tiancum et al., 2002; Oyama et al., 1999) nitrides and oxynitrides (Oyama, et al., 1998; Senzi et al., 1994; Colling et al; 1994; Y u , et al., 1998; Stancy et a l , 1998) and most recently, phosphide catalysts (Oyama et al., 2001; Fujikawa et a l , 2006; Mizutani et al., 2005; Oyama, 2003; Jian et al., 1996; Stinner et al., 2001). The metal phosphides have been reported to have higher activity than commercial sulfided catalysts and there is presently interest i n modifying the metal phosphides to increase their activity and selectivity. The promotion o f a second metal to form ternary phosphides is one modification used i n anticipation o f improved catalyst activity. The idea was drawn from the fact that i n commercial sulfided catalysts, when N i or C o are added to supported  M0S2 on alumina, the activity increases.  A s a result, sulfided C o - M o catalysts are principally used for H D S reactions. Conventional C o M o - S catalysts consist o f crystallites o f (Kabe et al., 1992). The  M0S2 decorated at their crystal edges with C o atoms  M0S2 has a layered structure in which the basal planes are largely  inactive. A s noted by others (Stinner et al., 2000; Zuzaniuk and Prins,- 2003) the metal phosphides are not layered structures but rather adopt various crystal structures that can accommodate the large P atom that is usually found at the center o f a triangular prism. M o P has the hexagonal WC-type structure whereas  Ni2P has the hexagonal Fe2P structure. Stinner et al.  (2000) and Zuzaniuk and Prins (2003) also showed that  C0M0P produced comparable activity  with M o P i n the absence o f H2S and N i M o P was the least active when these phosphides were  9  Introduction tested for H D N o f o-propylaniline. However, the authors reported that elemental sulphur analysis of the used catalysts showed higher levels o f sulfur on the M o P and N i M o P than on the C o P and C o M o P , indicating that the C o containing catalysts were resistant to sulfur. Sun et al. (2004) have reported the addition o f N i to M o P catalyst for H D S and in this study, increased N i content o f the N i - M o - P catalyst resulted in increased activity for H D S o f dibenzothiophene ( D B T ) . However, no synergistic effect between the N i and M o was observed and N i P had higher 2  activity than all the N i - M o - P catalysts tested. They concluded that N i was not a promoter o f H D S activity i n the N i - M o - P system either. Similarly, Rodriguez et al. (2003) have shown that a MoNiP/Si0  2  catalyst was much less active than either M o P / S i 0  2  or N i P / S i 0 2  2  for H D S o f  thiophene. Since hydroprocessing involves the removal o f both S and N containing compounds it is important that catalysts developed be tested for also H D N . Recently, much attention has been directed towards H D N because o f the tightening o f environmental legislation regarding the release o f N O x . In addition, N compounds have undesirable effects such as poisoning o f acidic catalysts. H D N is also becoming important because o f increasing interest i n converting petroleum residua, coal, shale and tar sands, which contain higher concentrations o f N than conventional crude oils. Refineries now have to process more heavy feedstocks such as vacuum gas o i l ( V G O ) , coker gas o i l , and light cycle o i l produced from V G O and since these heavy feedstocks contain high N concentrations, it is necessary to improve H D N catalysts i n order to obtain not only better product quality and distribution but also to meet stringent environmental legislation. Carbazole is a refractory N containing compound found i n petroleum feedstock and also difficult to remove. The effect o f alkyl groups is less important because o f the higher electron density o f the five-membered rings and the wide bond angles between the N atoms and  10  Introduction the neighboring alkyl groups. Since the N radius is small, (0.75 A) the C atoms o f the neighboring benzene rings screen the N hereoatom causing steric hindrance.  1.4  Knowledge Gap Some important issues arise from the preceeding studies. The ternary phosphides reported in  the literature were prepared such that the C o : M o and N i : M o ratio was 1:1. For example, in preparing M o N i P , Rodriguez et al. (2003) replaced one mole o f N i in N i P with one mole o f M o 2  to form M o N i P , and the intrinsic activity o f this catalyst was shown to be less than the M o P and  C02P catalysts. The use o f lower C o and N i concentrations has not been investigated. Typically, commercial hydroprocessing catalysts such as sulfided C0M0/AI2O3 and MM0/AI2O3 use 3-5 wt% o f C o or N i (Katzer and Sivasubramanian, 1979; Massoth, 1978; Topsoe et al., 2002). Another important approach to improving metal phosphide catalysts is to modify catalyst properties to increase the conversion and/or selectivity for the H D S o f 4 , 6 - D M D B T . 4 , 6 - D M B T is one o f the refractory  S-containing compounds that are difficult to desulfurize using  commercial sulfided catalysts and therefore serves as a model compound for investigation because once this molecule can be desufurized others that have less steric hindrance can be easily desulfurized. Previous reports (Kilanowski et al., 1978; Landau, 1997) have indicated that the position o f the two methyl groups on the D B T backbone, influences the reactivity. For example, the conversion o f 2,8-dimethyldibenzothiophene is much greater than 4 , 6 - D M D B T over sulfided C0M0/AI2O3 (Kilanowski et al., 1978). The authors attributed the l o w reactivity o f 4 , 6 - D M D B T to the presence o f the two methyl groups at the 4 and 6 position o f the molecule. The authors explained that at these positions, the methyl groups hinder the access o f the sulfur atom to the catalyst surface and consequently C-S cleavage does not occur. Therefore in the present research, Bnansted acidity w i l l be incorporated on metal phosphide catalysts so that the  11  Introduction two methyl groups at the 4- and 6- positions o f 4,6 D M D B T can be isomerized to reduce the steric hindrance associated with the C-S hydrogenolysis, thereby increasing the conversion and selectivity. Another method o f removing S from 4 , 6 - D M D B T is by first hydrogenating the aromatic rings and subsequent hydrogenolysis o f the C - S . There is no report i n the literature exploring the effect o f adding noble metals to metal phosphides to enhance hydrogenation i n 4 , 6 - D M D B T . Therefore Pt w i l l be added to modify metal phosphides to provide excellent hydrogenation capability, which can subsequently lead to increased conversion and selectivity o f 4 , 6 - D M D B T (Landau et al., 1997). The reactivity o f carbazole which is a more resistant N containing compound in petroleum feedstock has not been reported i n the literature using metal phosphides. Finally,  there is little information on the relationship between the activity and  the  physiochemical properties o f metal phosphides on modified metal phosphides.  1.5  Motivation Based on the above issues, the present study was focused on Co Ni2P (0.08 < x < 0.8) and x  N i M o P (0.07 < x < 1.11) catalysts and their catalytic activity for 4 , 6 - D M D B T H D S and x  carbazole H D N , respectively. B y incorporating small amounts o f C o and N i i n the N i P and 2  M o P respectively, the metal phosphides are expected to be metal rich and therefore increase activity o f these catalysts, as reported elsewhere (Stinner et al., 2001). B y supporting the prepared metal phospides on materials with Bronsted acidity such as the M o b i l Catalytic Materials ( M C M - 4 1 ) , it is expected that the catalyst isomerisation activity w i l l  increase,  suggesting that the two methyl groups o f 4 , 6 - D M D B T can undergo migration away from the S  12  Introduction atom, thereby reducing steric hindrance and allowing direct C-S cleavage o f the 4 , 6 - D M D B T molecule. It is also expected that by incorporating small amounts o f N i to modify metal phosphides, the hydrogenation capability w i l l be enhanced leading to faster hydrogynolysis on Bronsted acid sites and hence enhanced conversion and selectivity.  1.6  Objectives of the study Based on the knowledge gap in the literature and the motivation, the objectives o f the  present research are: A.  To investigate the H D S and H D N activity o f modified metal phosphides when 3-24 wt% o f Co and N i are added to N i P and M o P , respectively. 2  B.  To study the role o f acidic components i n the form o f supports (such as M o b i l Crystalline Materials, ( M C M - 4 1 or M C M ) and fluorine added to A 1 0 supports) for H D S and H D N 2  3  activity. C.  To determine the effect o f adding platinum on the catalyst activity i n order to explore the H D S hydrogenation pathway o f 4 , 6 - D M D B T .  E.  To examine the kinetics o f the H D S and H D N reactions on the metal phosphides.  F.  To quantify and relate the catalyst properties with the H D S and H D N activities as measured using model compounds.  G.  To examine the selected catalysts using Light Gas O i l derived from Athabasca bitumen.  13  Literature review  Chapter 2 Literature Review Over the past decades, the petroleum industry has gone through significant changes i n order to meet new challenges. Presently, refineries must remove more N and S from petroleum feedstocks i n order to meet new environmental regulations. In addition, light sweet crude is fast diminishing  and  so refineries  must  process  heavy petroleum  crude  containing higher  concentrations o f N and S compounds. Both developments have put pressure on refiners to develop solutions in order to meet these challenges. Hydroprocessing is most widely used by refineries to remove the S (hydrodesulfurization) and  N  (hydrodenitrogenation)  from  petroleum  feedstocks.  However,  conventional  hydroprocessing catalysts (sulfided C0M0/AI2O3 and N i M o / A l 0 3 ) have difficulty meeting these 2  new challenges and therefore new catalysts must be developed. Metal carbides, nitrides and phosphides have been investigated as alternative hydroprocessing catalysts. Although the metal phosphides have shown significant promise, there is a need to improve the catalytic activity o f these materials and test them using real petroleum feedstock since most o f the activity reported for metal phosphides have used model compounds. The improvement can be made by studying the reaction mechanisms and the properties o f N and S compounds i n petroleum feedstock, in order to tailor a catalyst to effectively hydroprocess and meet the challenges mentioned above. Therefore, in this section, a review o f the type, quantities, chemical behavior, mechanisms and the catalysts that have been investigated in anticipation o f meeting these challenges, is presented.  14  Literature review 2.1  Sulfur containing compounds and the hydrodesulfurization (HDS) process Hydrodesulfurization is a refinery process where S in the form o f organic compounds is  removed from distillate streams. The process involves the use o f H  2  at high temperature and  pressure to remove S in the form o f H S . H D S is a heterogeneously catalyzed reaction. It is 2  exothermic and essentially irreversible under the reaction conditions employed. Traditional catalysts used for H D S are supported metal sulfides. In particular, sulfided CoMo/Al C>3 and 2  C o W / A l 0 3 are commercial catalysts used worldwide for H D S . The promoters are either N i or 2  Co but more recently, sulfided N i - M o / A l C > 3 has been commercially employed for H D S (Song, 2  2000).  2.1.1  Types of common sulfur compounds in liquid fuels  It is important to have a good knowledge o f the types and reactivity o f S compounds since this information can provide the basic understanding o f how these S compounds can be removed from petroleum feedstock. The common types o f S compounds in liquid fuels, shown i n Table 2.1, can be divided into five classes namely thiols, sulfides, thiophenes, benzothiophenes ( B T ) and dibenzothiophenes ( D B T ) . The five classes can be further divided into aliphatic and nonheterocyclic aromatic sulfur compounds ( A A S C ) that include thiols, thiophenes and sulfides forming the aliphatic group. The polyaromatic S compounds include the B T ' s and the D B T ' s . The three major types o f transportation  fuels gasoline, diesel and jet fuels differ in  composition and properties (Gary and Handwerk, 1994; M a y o et al., 2001; H s u et al., 2000; Song, 2000). For example, gasoline derived from naphtha and fluid cracking ( F C C ) naphtha contains thiols ( R S H ) , sulfides (RSR), disulfides ( R S S R ) , thiophenes, alkylated thiophenes, and benzothiophenes.  Jet  fuel  derived  from  heavy  naphtha  and  middle  distillates  contain  benzothiophenes and diesel fuel derived from the middle distillates and light cycle o i l ( L C O ) ,  ••'  15  Literature review Table 2.1 Representative sulfur compounds in liquid fuels  Compound  Formula  Representation  Thiols  RSH  R-S-H  Sulfides  RSR  R  Disulfides  RSSR  R-S-S-R  "  S  "  R  R-  Thiophenes  RC H S 4  3  Benzothiophene  RC H S  Dibenzothiophene  RC  4,6-dimethyldibenzothiophene  8  C  1 4  1 2  H  J  5  H S  1 2  7  S S CH  3  CH  3  contain alkylated benzothiophenes, benzothiophenes and alkylated dibenzothiophenes. R is an alkyl or phenyl group and typically contains C 1 - C 4 alkyl groups (Landau, 1997).  16  Literature review 2.1.2  Reactivity of sulfur containing compounds Under normal hydropocessing conditions, the aliphatic and non-heterocylic aromatic sulfur  compounds are not difficult to hydroprocess and are therefore considered infinitely reactive i n practical high-conversion processes conversion  of  alkyl  (Phillipson, 1971; Gates et al., 1997). However, the  substituted  dibenzothiophenes  and  especially  the  4,6-  dimethyldibenzothiophene ( 4 , 6 - D M D B T ) is difficult. In fact, Kabe et al. (1992) reported that S compounds remaining in diesel fuels at S levels lower than 500 ppm are the benzothiophenes with alkyl substituents at the 4- and/or 6-positions on the aromatic ring. These species are termed refractory sulfur compounds and their low reactivity is attributed to steric hindrance and electronic factors. Only the alkyl groups that are positioned close to the S atom exert steric hindrance.  For  example,  4-methyldibenzothiophene,  4-ethyldibenzothiophene,  6-  methyldibenzothiophene, 6-ethyldibenzothiophene, 4,6-dimethyldibenzothiophene, 4-methyl, 6ethyl dibenzothiophene all exert steric hindrance. The position o f the alkyl substituents on the aromatic ring has a greater impact on the H D S reactivity than the number o f substituents. The alkyl groups positioned at a distance away from the S atom posses only electronic effects and do not sterically hinder the H D S reactivity (Landau, 1997). Table 2.2 shows a summary comparison o f the H D S conversion o f individual polyaromatic S-containing compounds in different solvents over sulfided C0M0/AI2O3. Although direct comparison o f the numbers cannot be made because different conditions (temperature and solvents) were reported for the H D S different, it can be inferred from Table 2.2 that 2,8dimethyldibenzothiophene  has  high  HDS  reactivity  (94-100%)  while  4,6-  dimethyldibenzothiophene has low H D S reactivity (7-9%). A t atmospheric pressure, methylsubstituted  benzothiophenes  showed  lower  HDS  conversion  (39-42%)  compared  nonsubstituted benzothiophene (58%) using n-Ci + n-C\2 solvent at 450 °C. Introduction o f a  17  to  Literature review  Table 2.2 Comparison o f H D S conversion o f individual polyaromatic S-containing compounds using different solvents over sulfided C o M o / A l 0 ( L a n d a u , 1997) 2  Solvent  n-C +n-Ci 7  2  n-Ci  6  Testing conditions: 450  300  1  50  Temperature, °C Pressure, atm Conversion %  Benzothiophene ( B T )  58  2-methylbenzothiophene  42  3 -methy lbenzothiophene  39  7-methylbenzothiophene  42  2,3 -dimethy lbenzothiophene  -  3,7-dimethylbenzothiophene  14  2,7-dimethylbenzothiophene  -  2,3,7-trimethylbenzothiophene  -  -  Dibenzothiophene(DBT)  50  100  4-methyldibenzothiophene  19  9  4,6-dimethyldibenzothiophene  9  7  2,8 -dimethyldibenzothiophene  100  94  3,7-dimethyldibenzothiophene  14  49  18  3  Literature review second methyl group to form 3,7-dimethylbenzothiophene produced a further decrease in H D S to 14%. Introduction o f methyl groups to D B T at the 4-position, especially i n the 4,6-positions, reduced the H D S conversion to 7-32% o f D B T , the 4 , 6 - D M D B T being 30-80% less reactive than the 4-methyl-DBT. However, introduction o f the two methyl groups at other positions in D B T (2,8- or 3,7-) had less impact on the conversion. Consequently, the conversion o f the sterically hindered dibenzothiophenes largely determines the required process conditions.  2.1.3  Reaction pathways of hydrodesulfurization of 4,6-dimethyldibenzothiophene The production o f ultra-low sulfur gasoline starts from June 2006 and the sulfur content o f  diesel w i l l reduce from present levels o f 500 ppm to 15 ppm by 2010. A t 15 ppm, alkyl substituted D B T s are the main S-containing compounds remaining in the o i l (Girgis and Gates, 1991).  Gates et al. (1997) pointed out that the 4-methyldibenzothiophene  dimethyldibenzothiophenes  are the most appropriate  and the 4,6-  compounds that should be used for  investigations for activity and reaction mechanisms for H D S . It is not surprising that recently a number o f reports (Prins et al.,2006; Yang et al., 2005; Rabarihoela-Rakotovao et al., 2004; Breysse, et al., 2003) on H D S are based on the activity o f the 4 , 6 - D M D B T refractory sulfur compound. Figure 2.1 shows the reaction pathway for the H D S o f 4 , 6 - D M D B T (Prins et al., 2006). The conversion o f this refractory 4 , 6 - D M D B T compound occurs through two routes. In the first route, direct elimination o f the S takes place to form dimethyl biphenyl ( D M B P ) and this route is termed the direct desulfurisation route (DDS). The second route is known as the hydrogenation route ( H Y D ) and the desulfurized molecules are 3,3'-dimethylbicyclohexyl ( D M B C H ) and methylcyclohexyl- toluene ( M C H T ) . The intermediates are formed as a result o f a series o f  19  Literature review  Figure 2.1  Reaction networks o f the H D S o f 4 , 6 - D M D B T (Prins et al., 2006)  20  Literature review hydrogenation o f the aromatic ring to form the tetrahydro 4,6 dimethyldibenzothiophene (4,6D M - T H - D B T ) and the hexahydro 4,6 dimethyldibenzothiophene ( 4 , 6 - D M - H H - D B T ) .  These  intermediates are formed relatively fast but the C-S bond breakage is slow (Prins et al., 2006). D M B P could also be hydrogenated to form M C H T . The reactivity o f the 4 , 6 - D M D B T is affected by the position o f the methyl groups. A s reported earlier, at the 4 and 6 positions, the substituted methyl groups possess strong steric hindrance, thereby suppressing the reactivity o f the molecule for D D S (Landau, 1997). When 4 , 6 - D M D B T is adsorbed on the surface o f the catalyst horizontally, the methyl groups and the S lie in the same plane. The methyl groups at the 4,6-positions are adjacent to the sulfur atom and are more spacious than the o orbitals o f the sulfur atom and therefore hinder the molecule from binding on to the catalyst surface. Hence adsorption o f 4 , 6 - D M D B T is weak resulting i n the D D S pathway being strongly suppressed (Landau, 1997). O n the other hand, hydrogenation is promoted by n adsorption. In this mode the methyl groups do not hinder adsorption since they donate electrons to the ring (Landau, 1997; Prins et al., 2006). Therefore, the study o f the reaction pathway o f the 4 , 6 - D M D B T reveals that in order to enhance reactivity o f this molecule, the methyl groups at the 4- and 6- positions considered as the most hindered positions should be moved away by isomerisation which could be achieved by incorporating acidity to the catalyst. The second method o f increasing the reactivity is increasing 7i adsorption so that hydrogenation is enhanced.  2.2  Nitrogen compounds and the hydrodenitrogenation process Hydrodenitrogenation refers to the removal o f N-containing compounds from petroleum  feed stock. Historically, refineries have been more concerned with the removal o f S-containing  21  Literature review compounds than N-containing compounds. However, more recently, much attention has been directed towards H D N because o f the tightening o f environmental legislation regarding the release o f N O x . In addition, N compounds have undesirable effects such as poisoning o f acidic catalysts. H D N is also becoming important because o f increasing interest i n converting petroleum residua, coal, shale and tar sands, which contain higher concentrations o f N than conventional crude oils. Refineries now have to process more heavy feedstocks such as vacuum gas o i l ( V G O ) , coker gas o i l , and light cycle o i l produced from V G O and since these heavy feedstocks contain high N concentrations, it is necessary to improve H D N catalysts in order to obtain not only better product quality and distribution but also to meet stringent environmental legislation.  2.2.1  Nitrogen compounds in crude oils In order that H D N be effectively accomplished, the mechanism, reaction network and  catalyst used to accomplish H D N needs to be properly understood. Therefore i n this section a discussion o f the N containing compounds in petroleum feedstock, the mechanisms and reaction networks that have been proposed for H D N , w i l l be presented. The types, quantities and chemical behavior o f N-containing compounds in o i l derived from sand oils and shale derived o i l is an important knowledge that is required for the development o f an improved catalyst for H D N activity. Detailed review on this aspect has been documented i n the literature (Katzer and Sivasubrumanian, 1997, Jin et al., 1997; Harvey et al., 1 9 8 5 ; M a j o l s k y e t a l . , 1987). N-containing  compounds  in petroleum are normally divided  into heterocyclic and  nonheterocyclic compounds. Katzer and Sivasubramanian, (1979) reported that most o f the nitrogen found in these feedstocks is found in heterocyclic compounds and that they are resistant  22  Literature review to hydrodenitrogenation. Non-heterocyclic organonitrogen such as aliphatic amines and nitriles are found in smaller quantities and are relatively more active towards H D N than the heterocyclic compounds. Consequently, the nonheterocyclic N compounds are o f less industrial importance during H D N . The heterocyclic compounds are present in larger quantities and are the most difficult to remove. For example, Satterfield et al. (1984) using a batch reactor at 640 K and 13.6 M P a , reported that with addition o f phosphorous to  MM0S/AI2O3, the pseudo-first order rate  constants for H D N decreased in the order pyridine (20.9 min" ) > quinoline (6.6 min" ) > acridine 1  1  (2.2 min" ) > benz(a)acridine (1.0 min" ). Generally, the heterocyclic compounds are divided into 1  1  basic and non-basic nitrogen containing compounds as shown on Table 2.3. The basic N compounds consist o f six-membered heterocyles such as pyridine, acridine quinoline and their substituted analogues. The heteronitrogen atom in these compounds has one pair o f electrons that are not contributing to the TX -electron cloud o f the heterocyclic ring and therefore they are available for donation to acid sites on catalyst surfaces. O n the other, the non-basic N-containing compounds consist o f five-membered heterocycles such as pyrrole, indole, carbazole and substituted carbazoles. The heteronitrogen atom i n the five- membered ring contains two lone pair o f electrons that are delocalised and are not available for donation to acidic catalytic surfaces. The non-basic nitrogen containing compounds are converted into basic compounds upon hydrogenation. Majolsky et al. (1987) and Frankman et al. (1987) reported the presence o f neutral pyrrole benzologues and basic pyridine benzologues i n Athabasca bitumen. Jokuty et al. (1991) and Jokuty et al. (1992) have discussed the presence o f neutral N compounds i n synthetic crude o i l derived from Athabasca bitumen. The authors reported that the major N-containing compounds are the alkyl-substituted carbazoles. They also  23  Literature review Table 2.3: Representative nitrogen containing compounds in liquid fuels  Name  Formula  Representation  Nonheterocyclic compounds  Aniline  C H NH 6  5  -NH, 2  Nonbasic heterocyclic compounds  C H N 4  5  Pyrrole  H C H N 8  7  Indole  H  Carbazole  C H N 1 2  9  Basic Heterocycles  Pyridine  C H N 5  5  Quinoline  C H N  Acridine  C ^ N  9  7  concentrated the basic N obtained from the synthetic o i l and found that the ring N-containing compounds were substituted 5,6,7,8 tetrahydroquinolines and octahydrobenzoquinolines (Jokuty et al. 1992). Dinneen et al. (1985) found that in terms o f the number o f aromatic rings in the 24  Literature review molecule present in gasoline fractions o f shale o i l , single ring N-containing compounds comprised 35 wt%, indoles and quinolines 25 wt% and then 40 wt% multi-ring that also contain N and O. The amount o f N-containing compounds present i n gas o i l derived from tar sand is typically 2 to 5 times that obtained from petroleum crude o i l (Katzer and Sivasubramanian, 1979). Frequently, the higher molecular weight N-containing compounds contain sulfur and oxygen. In principle, methods used to hydrodenitrogenate petroleum liquid feedstock should also apply to the heavy feedstocks such as those obtained from Athabasca bitumen since they both contain common N compounds. However, special considerations would have to be taken in order to accomplish high H D N activity i n these synthetic derived oils because o f the presence o f high nitrogen content, high aromaticity and l o w hydrogen.  2.2.2  Difficulties of H D N Before discussing the reaction mechanisms o f H D N , it is important to discuss the  difficulty o f carrying out H D N . Typically, H D N is more difficult than H D S and it is partially because it is more difficult to break the C - N bond than the C-S bond. Table 2.4 shows the bond energies for some hetero-atoms in the aromatic and saturated molecule. A s shown i n Table 2.4, the bond energy o f the C - N bond is 308 kJ/mol and it is higher than the bond energy o f the C-S bond (259 kJ/mol). A l s o the C = N bond energy is 615 kJ/mol which is higher than the bond o f C=S (577 kJ/mol) [57]. This higher C = N bond energy explains why it is necessary to saturate by hydrogenation the aromatic N-containing heterocycles with H before the N heteroatom can be 2  removed. The saturation reduces the relatively large energy o f the C = N bonds and therefore facilitates easier scission o f the C - N bond. The difficulty o f H D N is also due to steric hindrance. For example, the C atoms o f the neighboring benzene rings o f acridine, screen the N hereoatom (because o f the small atomic  25  Literature review radius, 0.75 A) causing steric hindrance and hence N hetero cannot get access to the catalyst surface to react. A l k y l groups substituted on pyridine decreased the H D N by one order o f magnitude while on quinoline introduction o f the alkyl groups did not significantly affect the H D N reactivity. The electronic nature o f the substituted alkyl groups predominantly determines the  Table 2.4 Bond Energies for some hetero-atoms i n aromatic and saturated molecules [Fox, M . A . and Whitesell, J. K , 1994]  Bond  Energy,  Bond  Energy,  Type  kJ/mol  Type  kJ/mol  C—N  C - H  413  c-c  348  C=C  614  C=C  839  c-s  N-H  391  C=S  C=N  308  615  891  26  259  577  Literature review reactivity o f alkyl pyridines. In pyrroles, alkyl substituents are less important as there is high electron density o f the five-membered rings and the wider bond angle between the N atoms and the neighboring atoms (Landau, 1997). However, the lowest H D N reactivity o f N-alkylated pyridines and pyrroles seems to be the result o f complete screening o f the N-heteroatom (Landau, 1997). V G O residue, light cycle oil from V G O residue and coal-derived liquids are termed heavy feedstocks because they have high boiling points and usually contain much higher concentrations of N-containing compounds than the straight-run streams and are therefore more difficult to denitrogenate as the molecules are more complex i n structure and size. These heavy feedstocks w i l l also have higher boiling points and the bulky nature o f the molecules w i l l reduce their accessibility to the catalyst surface. H D N is also difficult because some o f the intermediate products react to form secondary N containing compounds that are more resistant to N removal than the starting molecules. These N containing compounds also have strong adsorption capacity on the catalyst surface and are selfinhibitors both o f which result in low H D N rate. Coke formed during H D N is mainly due to the condensation o f the basic-nitrogen containing compounds on Lewis acid sites o f the catalysts and this reduces the H D N activity (Fox and Whitesell, 1994).  2.3  Reaction networks and mechanism of hydrodenitrogenation The reaction networks, mechanisms and kinetics o f pyridine (Mcllvried, 1971; Hanlon,  1987; Sonnemans and V a n den, 1973) quinoline (Katzer and Sivasubramanian, 1979; Kherbeche et al., 1991; Satterfield and Cocchetto, 1981; M i l l e r and Hinnemann, 1984; Perot, 1991; Minderhoud and V a n Veen, 1993; Jian and Prins, 1998), acridine (Zawadski et al., 1982) indole (Odbunmi and Ollis, 1983) and several anilines on sulfided catalysts have been reported in the  27  Literature review literature. However, there is little information in the literature on carbazole. There are different types o f N-containing compounds (basic and non-basic heterocyclic) that have to be removed i n H D N . Therefore in this section, the reaction networks o f quinoline (basic) and carbazole (nonbasic) are presented. Quinoline is representative o f basic heterocyclic N-containing compounds and carbazole the non-basic heterocyclic N-containing compounds. The non-heterocyclic N containing compounds are facile i n H D N and therefore they w i l l not be considered. Generally, the pathways for removing the heteronitrogen atom from heterocyclic compounds is as follows: 1) hydrogenation o f the N-ring, 2) C - N bond scission to an amine and 3) hydrogenolysis o f amine to hydrocarbons and ammonia. In such a scheme, the C - N bond breaking becomes possible once the aromaticity is lost and the aliphatic C - N bonds are formed. However, because o f the presence o f an abundance o f alkenes in the H D N products, it was earlier conceived that elimination o f ammonia to form alkenes was the mechanism responsible for C - N bond breaking (Porefaix et al., 1991; Vivier et al., 1991). However, recently, Prins et al. (2006) reported that the dominant mechanism is nucleophilic substitution by H2S.  2.3.1  Quinoline Quinoline (Q) has a six-membered heterocyclic ring as well as a phenyl ring and  consequently hydrogenation and hydrogenolysis are present i n the H D N network o f quinoline (Katzer and Sivasubramanian, 1979; Massoth, 1978; Topsoe et al., 1996). Because quinoline can be separately examined for hydrogenation and hydrogenolysis it has been studied more than other petroleum N-containing molecules. Figure 2.2 shows the H D N reaction network o f quinoline where it can be denitrogenated in two ways: v i a o-propylaniline ( O P A ) and v i a  28  Literature review  Q  THQ-5  Figure 2.2  THQ-1  OPA  DHQ  H D N network o f quinoline over sulfided M - M 0 / A I 2 O 3 catalysts (Jian and Prins, 1998)  decahydroquinoline ( D H Q ) . Both pathways require that the strong C - N bond in the aromatic ring be hydrogenated and therefore hydrogenation o f the N - r i n g or both the N - r i n g and the benzenoid ring takes place. In the pathway to O P A , hydrogenation o f Q leads to the formation o f 1,2,3,4, tetrahydroquinoline (THQ-1) and T H Q - 1 is mostly in equilibrium with Q (Satterfield and Cocchetto, 1981). After hydrogenation, ring opening o f the heterocycle o f the T H Q - 1 to O P A  29  Literature review takes place via the highly active O P A intermediate. The rate o f H D N is suppressed in this pathway due to competitive adsorption when D H Q is present and, as reported by Jian and Prins (1998) in the presence o f H S only about 40% o f the H D N is converted. After the formation o f 2  O P A it is hydrogenated to form the most reactive propylcyclohexylamine intermediate which yields propylcyclohexenes (PCHEs) and NH3. The  alternative  route  is via D H Q and  Q  is hydrogenated  to  give the  5,6,7,8-  tetrahydroquinoline product that subsequently produces the D H Q . However, D H Q is also formed from the hydrogenation o f T H Q - 1 . H i g h conversion o f quinoline to D H Q is promoted by high hydrogen pressure, a low H2S/H2 ratio and a sufficiently high temperature above 623 K . Satterfield et al. (1984) reported that using a batch reactor and with phosphorous added to N1M0S/AI2O3 (350 - 390 °C and 6.9 M P a ) , the overall H D N reaction order was essentially zero and this was attributed to the high N surface coverage. The authors concluded that quinoline kinetics do not follow a pseudo-first order reaction because it is affected by the conversion and feed concentrations and it is strongly inhibited by the presence o f other nitrogen compounds. Therefore quinoline is modelled using a Langmuir-Hinshelwood ( L - H ) rate equation (Satterfield etal., 1984).  2.3.2  Carbazole The H D N o f carbazole has been reported (Nagai et al., 1988) using a micro reactor and  M0S2/AI2O3 catalyst and also on nitrided M o / A l 0 2  3  (Nagai et al., 2000) catalyst tested at 280-  360 °C and 10.1 M P a . Figure 2.3 shows the reaction pathway for carbazole. A s explained earlier, because direct hydrogenolysis o f the C - N bond without hydrogenation is difficult, biphenyl (BP)  30  Literature review  H Carbazole  B i  H  H  1,2,3,4-Tetrahydro-  Hexahydro-  carbazole  carbazole  P  h e n  y'  H  H  Decahydro-  Perhydrocarbazole  carbazole  Cyclohezylbenzene  Cyclohexyl-  Bicyclohexyl  cyclobenzene  Figure 2.3: General reaction scheme o f H D N o f carbazole (Nagai et al., 2000)  is  not  observed.  tetrahydrocarbazole, hexahydrocarbazole, bicyclohexyl  is  Hydrogenation which  is  of  further  decahydrocarbazole formed  Cyclohexylcyclobenzene  from product  the was  carbazole  leads  hydrogenated and C-N  to  the  form  formation equilibrium  perhydrocarbazole. bond  formed  31  to  cleavage  from  The of  of  1,2,3,4-  products  major  of  product  perhydrocarbazole.  decahydrocarbazole  although  Literature review decahydrocarbazole was not detected on the nitrided catalyst. Cyclohexylbenzene was formed from hexahydrocarbazole. The authors also reported that upon sulfidation o f the reduced catalysts, the hydrogenation was enhanced with the formation o f hexylcyclohexane. O n the nitrided catalysts, Nagai et al. (2000) reported an H D N rate o f 0.34 pmolh"'m~ after 3 h o f time 2  on stream.  2.4 Catalysts used for hydroprocessing After understanding the reaction networks for both H D N and H D S , it is important to discuss the catalysts that are used to effect hydroprocesssing. First a discussion o f the commercial sulfided metal catalyst w i l l be presented. Subsequently, the introduction and development o f metal nitrides, carbides and phosphides which are regarded as promising catalytic materials for hydroprocessing w i l l be discussed. In principle, electronic factors are responsible for the variation in the catalytic activity o f the sulfides, nitrides and phosphides. Surfaces that are able to transfer electron density into the C - S antibonding orbitals o f the reacting molecule should facilitate decomposition o f the C - S bond (Rodriguez et al., 2003). The role o f the N , C , S and P is to transfer electrons to H for reduction +  to occur and create a sulfur vacancy for the sulfur molecule to adsorb and react (Kabe et al., 1999). For example in sulfided M o catalyst, the four electrons move to form a reduction o f the S 2H  2  S"  +2H  2  +  •  4H +4e  (2.1)  •  H S + • ( s u l f u r vacancy)  (2.2)  +  2  reacting molecule. In the case o f butadiene it is formed from thiophene as follows (Kabe et al., 1999). C H 4  4  + 2H  +  + 4e  •  32  C H 4  + S" 2  6  (2.3)  Literature review 2.4.1  Use of transition-metal sulfides as hydroprocessing catalysts Conventionally, the transition-metals C o or N i is added to  catalysts  as  promoters  for  simultaneous  hydrodesulfurization and  M0/AI2O3 or W/AI2O3 hydrodenitrogenation.  Traditionally, sulfided forms o f these catalysts are used (Prins, 2001; Topsee and Clausen, 1984). The  C0M0/AI2O3 is used for H D S while N i M o / A l 0 is the desired catalyst for H D N (Prins, 2  3  2001). The N i W / A ^ C h system is usually employed in cases where hydrocracking is required. Since many  C0M0/AI2O3, Ni-Mo/Al20"3 and N i W / A ^ C h are commercial hydroprocessing catalysts, reports  are  available  in  the  literature  describing  these  catalysts  (Katzer  and  Sivasubramanian, 1979; Massoth, 1978; Topsoe and Massoth, 1996; Landau, 1997; Topsee and Clausen, 1984). A number o f characterization techniques including X P S , X R D , T E M and extended x-ray absorption fine structure ( E X A F S ) have been used to investigate the presence o f S after sulidation. Sulfidation o f  C0M0/AI2O3 and N i M o / A l 0 3 has been observed using X P S (Houalla 2  and Delmon, 1981) and E X A F S have been used to identify N i - M o - S structure i n sulfided  MM0/AI2O3 (Louwers and Prins, 1992). 2.4.2  Use of transition metal nitride and carbides as hydroprocessing catalysts. The development o f nitrides and carbides as research catalysts was made possible by the  search for new catalytic materials to replace sulfided catalysts. These catalytic materials were produced almost at the same time and they both use the temperature programmed reduction o f the oxide precursors for preparation. Only the nitrides w i l l be discussed in this section. Volpe et al. (1985) introduced a new branch o f catalysis by nitriding M0O3 and WO3 with ammonia to obtain high surface area  M02N and W2N, respectively. Subsequently, other investigators  (Oyama, 1996; N y l o n et al., 1999) have reported the use  33  M02N for H D N and H D S and have  Literature review found the nitride catalyst to have comparable activities to those o f commercial aluminasupported metal sulphide catalysts (Volpe et al., 1983). Details o f the preparation o f nitrides can be found in the literature (Volpe et al., (1985); Oyama, 1996; N y l o n et al., 1999; L i a w et al., 1995). Essentially, the preparation o f the nitrides involves the use o f a temperature programmed reaction (TPR) technique where NH3 gas is passed over the metal oxide at specified space velocities, heating rates and final temperature. Laser pyrolysis techniques (LPT) have also been used to prepare metal nitrides with high surface area (Liaw et al., 1995). The use o f mixtures o f H2 and N2 instead o f NH3 as a nitriding source has also been reported for preparing nitride catalysts (Wise and Markel, 1994). Molybdate has been used as the M o source for nitrided catalyst (Nagai et al.,1998; Nagai et al., 1998; Nagai et al., 1999). The authors identified several species in various proportions made up o f Y-M02N and (3-Mo Nn.78 metallic M o and unconverted M0O2. The face centered cubic 2  crystal (fee) o f Y-M02N was the main species found at lower temperatures. The authors also reported that the final properties o f the nitride catalysts are influenced by the space velocity o f the synthesis gas, heating rates and composition o f the reactant gas mixture. The main problem with Y-M02N is its stability due to its reduction i n H2S as given by the equation: Mo N + H S 2  2  •  2 M o S + N H + 2.5 H 2  3  2  (2.4)  In summary, nitrides show sulfidation in the hydroprocessing environment. Once sulfided, the catalysts surface w i l l behave similarly to the sulfided catalyst which is not suitable for hydroprocessing to ultra low sulfur levels. In addition, the nitrides are difficult to prepare as high temperature reaction in ammonia is used. Therefore, at present, the nitrides are not suitable candidates for replacing the sulfides.  34  Literature review 2.4.3  Use of transition-metal phosphides The most recently researched catalytic materials for hydropocessing are the transition metal  phosphides. A s a promoter, phosphorous has been incorporated in commercial hydrotreating catalysts containing sulfided N i (or Co) and M o on alumina support (Oyama et al., 2004., Qian et al., 2002). Therefore when transition metal phosphides were first synthesized and tested as hydroprocessing catalysts, it was argued that there was no difference between the phosphorous added to sulphided catalysts and the new metal phosphide catalysts. However, Clark et al. (2001), reported that phosphorous in bulk W P is in a reduced state, which is different from phosphorous used as a promoter in standard alumina-supported sulfide catalysts. Figure 2.4 shows the crystal structure o f some o f the metal phosphides. M o P is isostructural with W C with the non-metal containing prisms stacked on top o f each other. The metal phosphides o f groups 6-10 adopt the M n P and N i P structures both o f which can be regarded as distortions o f the N i A s structure. The phosphorous atoms form chains in the M n P structure while they form pairs in the N i P structure. Transition metal phosphides have high chemical and thermal stability, they are dense, hard and brittle and have high thermal and electrical conductivities (Stinner et al., 2000). The properties o f metal phosphides are similar to transition metal borides and selicides and also i n most cases, carbides and nitrides (Corbridge, 1980). Based on the stoichiometry, phosphides can be classified as metal-rich ( M / P > 1), monophosphides ( M / P = 1), and phosphorous rich ( M / P < 1). The phosphorous rich phosphides are thermally and chemically unstable and are therefore not suitable for hydroprocessing. The metal-rich and monophosphides are those usually applied technologically in the field o f catalysis and the main focus is hydrogenation catalysis as i n H D N and H D S (Oyama et al., 2001).  35  Literature review  Ni P, Fe P P"62m Hexagonal 2  CoP, C o P , F e P *• nma Orthorhombic M  2  2  NiP Pbca Orthorhombic  FeP, £ nm l Orthorhombic 2  metal  Figure 2.4: Crystal structures o f some transition metal phosphides  36  n  P  Literature review The mechanism by which phosphides (e.g. N i P ) react to remove nitrogen from organic 2  molecules involves activation at carbon positions a and P to nitrogen atoms (Oyama, 2003). Comparatively, sulfides perform activation principally through activation at the  P-carbon  position. In addition, both sulfides and phosphides form trigonal prisms as the fundamental blocks in building up crystal structures. However, the sulfides commonly adopt layered structures with flat two-dimensional crystals, which expose  saturated basal planes.  The  phosphides on the other hand have more isotropic external morphologies that give rise to globular shapes that have a considerable impact on the morphology and reactivity o f the supported catalysts (Oyama, 2003). Oyama (2003) also reported that unlike the sulfides that use P-carbon, the reactivity order o f substituted piperidines using phosphides indicated that ring opening proceeded by C - N bond scission involved an a-carbon and therefore this difference might account for the higher activity of phosphides over sulfides. The effect o f P was much more profound for H D N than H D S , and Wang et al. (2002) suggested that H D S must therefore occur on metal centers o f metal phosphides. The ability o f N i P to selectively remove S without 2  excessive hydrogenation o f the aromatic ring was suggested to be due to the exposure o f many crystallites corners and edges. Although detailed preparations o f bulk and supported metal phosphides w i l l be presented in chapter 3, a general description o f the preparation o f these catalytic materials is shown i n Figure 2.5. First, a solution containing both the metal source and the phosphorous is prepared and evaporated on a hotplate while stirring. The resulting paste is dried i n an oven at 110 °C and then calcined at 500 °C for 6 h. The oxide precursor obtained is reduced i n H using temperature 2  programmed reduction. The reduced phosphide is then passivated using 2 % 0  37  2  in A r .  Literature review  Prepare a solution containing both the metal and the phosphorous  Evaporate solution on hotplate to form paste  Further dry i n oven at 110 °C until all water is removed i Calcine dried product at 500 °C for 6 h i A p p l y temperature programmed reduction using H 2  I  Passivate using 0 / A r 2  Figure 2.5 Schematic diagram o f the preparation o f metal phosphides  Since the metal phosphides are recently synthesized catalytic materials, few reports are available in the literature compared to the metal sulfides. Table 2.5 shows a summary o f the hydroprocessing activity o f metal phosphides that has been reported. Stinner et al. (2001) reported the activity o f bulk C o P , N i P , M o P , W P , C0M0P and N i M o P 2  2  i n the H D N o f o-propylaniline using a continuous-flow microreactor at 673 K and 3.0 M P a . The H D N conversions o f the o-propylaniline, were given as 4.6%, 4.9%, 16.0% and 21.5% for C o P , 2  N i P , M o P and W P , respectively. Intrinsically, M o P was reported as the most active catalyst. 2  38  Literature review  Table 2.5: Summary o f hydroprocessing activity o f transition metal phosphides  Catalyst  C02P 2  2  Conversion  Feed: o-propylaniline Micro-reactor at 673 K 3.0 M P a  Ni P MoP WP Ni P/Si0  Testing conditions  2  Feed: 3000 ppm D B T , 2000 ppm quinoline 500 ppm benzofuran, 20 wt% tetralin  4.6 4.9 16.0 21.5  Stinner et al. 2001  a  a  a  a  90.6 ' 30.4 a1  Ni P/Si0 MoP/Si0 2  2  2  2  MoP/Si0  2  Ni P/Si0  2  2  Ni P/Si0 2  2  Ni P/MCM 2  Wang et al. 2002 4.21  Feed: 3.2 m o l % thiophene i n H . F l o w reactor: 643 K , 1 atm.  14.8 7.4  2  Feed: o-methylaniline Continuous flow reactor: 643 K , 3.0 M P a  Oyama et al. 2002  &  Feed: 3.2 m o l % thiophene i n H . F l o w reactor: 643 K , 1 atm. 2  bl  32 i' ^ 1 bi  Trickle bed reactor: 643 K , 3.1 M P a MoP/Si0  Reference  c  d  d  26  Rodriguez et al. 2003 Zuzaniuk et al. 2003  Feed: 3000 ppm D B T , 2000 ppm quinoline 500 ppm benzofuran, 20 wt% tetralin Trickle bed reactor: 643 K , 3.1 M P a  99  Feed: 3000 ppm D B T , 2000 ppm quinoline 500 ppm benzufuran, 20 wt% tetraline Trickle bed reactor: 593 K , 3.9 M P a  1.3 2.5  Feed : D B T at 613 K , 5.0 M P a  100  9 1  Phillips et al. 2002  al  Oyama et al. 2004  b. e  Oyama, 2003 f  m  Wang et al. 2005  a: HDN conversion of o-propylaniline; al: conversion of DBT; bl: conversion of quinoline; c: conversion of thiophene on MoP/Si0 relative to sulfided Mo/Al 0 . 2  2  3  d: conversion of thiophene relative to sulfided Mo/Si0 at conditions in reference (Rodriguez et al. 2003) e: relative conversion of DBT at conditions in reference Oyama, 2003; f: relative conversion of quinoline at 2  conditions in reference (Oyama, 2003); m: conversion of DBT at conditions in reference (Wang et al. 2005).  39  Literature review Oyama (2003) also reported that the catalytic activity o f transition metal phosphides for the H D S o f D B T and H D N o f quinoline at 643 K and 3.9 M P a H follow the order F e P < C o P < 2  2  M o P < N i P . The author further reported that the H D S o f D B T and H D N o f quinoline over 2  N i P / S i 0 were 1.3 times and 2.5 times more than that o f a commercial sulfided N i M o / A l 0 3 2  2  2  respectively. Since N i P and M o P have been reported to be catalytically active, a number o f reports have 2  been presented in the literature on the activity o f supported N i P / S i 0 phosphide (Rodriguez et 2  2  al., 2003; Oyama et al., (2003); Wang et al., 2002; Gallezot et al., 1997) and M o P / S i 0  2  (Phillips  et al., 2002; Zuzaniuk and Prins, 2003). Rodriguez et al. (2003) reported that the H D S activity o f 30 wt% N i P / S i 0 using 3.2 m o l % thiophene i n H at 643 K and 1 atm i n a flow reactor was 14.8 2  2  2  times that o f sulfided M o / S i 0 and that M o P / S i 0 showed 7.5 times the activity o f the sulfided 2  2  M o / S i 0 . Thus in all the reports, N i P / S i 0 2  2  2  and M o P / S i 0  2  showed higher hydroprocessing  activity than commercial sulfided catalysts. Although previous studies on metal phosphides have shown that these new catalytic materials are more active than the commercial sulfided hydroprocessing catalysts, in order to hydroprocess to the allowable S and N levels, the metal phosphides w i l l have to be modified to enhance their activity and selectivity. Promotion o f M o S hydrotreating catalysts with C o or N i is well known (Kabe et al., 1999) 2  but metal phosphides are not promoted by C o or N i (Stinner et al., 2001; Stinner et al., 2000; Sun et al., 2004; Rodriguez et al., 2003; Kilanowski et al., 1978). Table 2.6 summarizes the results o f activity studies with the addition o f C o or N i to metal phosphides. Stinner et al. (2001) prepared C o M o P and N i M o P , such that the C o / M o and N i / M o ratio was 1/1 and tested the H D N activity  40  Literature review Table 2.6: Summary o f activity o f metal phosphides promoted with C o or N i  Conversion Catalyst  HDS/HDN  Testing conditions  Bulk  route  Reference  Ratio Co(Ni):Mo  CoMoP  (1:1)  NiMoP  (1:1)  Feed: o-propylaniline Micro-reactor at 673 K 3.0 M P a  15%  a  18%  a  Stinner et al. 2001  Feed: 4 , 6 - D M D B T -  C0M0P/AI2O3  Fixed bed reactor: 603 K ,  93  Hydrogenolysis  Mizutani et al. 2005  3.0 M P a NiMoP/Si0  2  -  CoMoP/SiC-2  Feed: o-methylaniline  0.96° Hydrogenation  Continuous  0.96° Hyrogenation  flow  reactor:  2003  643 K , 3.0 M P a MoP/Si0  NiMoP/Si0 Ni P/Si0 2  0.1  d  decaline  0.8  d  Fixed bed reactor, 593 K ,  3.0  d  0.5  B  Feed:5000  2  2  2  ppm  Zuzaniuk and Prins,  DBT/  Hydrogenolysis Hydrogenolysis  Sun et al. 2004  Hydrogenolysis  3.0 M P a , 12 h" W H S V 1  NiMoP/Si0  -  2  Feed: 3.2 m o l % thiophene in H . F l o w reactor: 643 K ,  Rodriguez et al.  1 atm.  2003  2  CoMoP/Al 0 2  3  Feed: Thiophene Pyridine 643 K , 3.0 M P a  +  100 100  e  L i et al. 2005.  f  a: HDN conversion of o-propylaniline relative to MoP; b: Thiophene conversion relative to Ni-MoS /Si0 ; c: HDN conversion of o-methylaniline relative to MoP/Si0 ; d: TOF (xlO' s") based on CO chemisorption; e and f: conversion of thiophene and pyridine at conditions in reference. 2  3  2  41  1  2  Literature review o f the prepared catalysts using o-propylaniline in a micro reactor at 673 K and 0.2 M P a . The authors reported that the activity o f the C o M o P and N i M o P were 15% and 18% respectively relatively less than M o P catalysts. Similarly, Rodriguez et al. (2003) prepared N i M o P by replacing half o f the nickel atoms i n N12P with M o atoms (i.e. a N i / M o ratio o f 1/1). The authors then reported the activity o f supported NiMoP/SiC»2 using 3.2% thiophene in H in a flow reactor at 1 atm and 643 K , to be 2  half that o f the activity o f sulfided M o / S i 0 . 2  Zuzaniuk and Prins (2003) also prepared N i M o P / S i 0 2 and C o M o P / S i 0 2 with N i / M o and C o / M o ratios o f 1/1 and tested the activity o f the prepared catalysts using o- methylaniline i n a continuous flow reactor at 643 K and 3 M P a . The authors reported the activities o f the NiMoP/Si0  2  and C o M o P / S i 0  2  to be less than M o P / S i 0  2  and C o P / S i 0 2  respectively. The  2  authors also reported that the product selectivity revealed that the hydrogenation route was preferred to the hydrogenolysis. Sun et al. (2004) reported the D B T activity o f M o P / S i 0 , a series o f N i M o P / S i 0 2  2  and  N i P / S i 0 using a fixed bed reactor at 3.0 M P a , 593 K and 12 h" W H S V . The authors reported 1  2  2  T O F s o f 0.1 x 10" s" for M o P , 0.1-1.4 x 10" s" for the series o f N i M o P / S i 0 and 3.0 x 10" s" 3  1  3  1  3  1  2  for Ni2P/Al2C»3 and that increased N i content in a N i - M o - P catalyst resulted in increased activity for H D S o f D B T . However, no synergistic effect between the N i and M o was observed and M2P had higher activity than all the N i - M o - P catalysts tested. The authors also reported that the product selectivity was very high (79-84%) for biphenyl (BP), which is obtained through the hydrogenolysis route o f H D S o f D B T . Mizutani et al. (2005) reported the activity o f C o M o P / A l 0 2  3  using 4 , 6 - D M D B T i n the  presence o f quinoline, in a fixed bed reactor at 623 K , 3.0 M P a and 13.5 h" L H S V . The authors 1  42  Literature review reported high D M B P selectivity (42%) and l o w M C H T (27%) over C o M o P / A l 0 . Thus, the 2  3  hydrogenolysis route o f the H D S o f 4 , 6 - D M D B T was preferred to the hydrogenation route. Recently, L i et al. (2005) reported the activity o f  C0M0P/AI2O3 to be higher than  M0P/AI2O3 using pyridine. The authors also reported that small amounts o f C o cause a significant increase on the hydrotreating activity o f M0P/AI2O3. Previous reports (Zuzaniuk and Prins, 2003; Rodiguez et al., 2003) on the effect o f addition of a second metal such as C o or N i to metal phosphides suggests that when the C o or N i was added to M o in a ratio o f 1:1 to form  C0M0P or N i M o P , the activity was less than M o P and  therefore addition o f C o or N i was not beneficial. However there are few reports that examine the addition o f small quantities o f Co (less than stoichiometric quantities) and N i to M o P or M2P. Small quantities o f Co and N i have been used as promoters for commercial sulfided catalysts and it is expected that when added i n such small quantities the dispersion on the surface o f the catalyst w i l l increase and hence increase activity. Since the S stability o f the catalyst is important during hydroprocessing, the stability o f transition metal phosphides was reported by Stinner et al. (2001). The authors reported that transition metals are thermodynamically stable i n the presence o f H2S and S. However, Oyama et al. (2002) reported the presence o f a phosphosulfide phase after reaction. The authors tested Ni2P/Al 03 with 3000 ppm D B T and 2000 ppm o f quinoline i n a three-phase packed-bed reactor 2  at 643 K and 3.1 M P a and analysed the surface o f the spent M2P/AI2O3 using Extended X-ray Absorption Fine Structure Spectroscopy ( E X A F S ) . The E X A F S analysis revealed the presence of a phosphosulfide phase. Nelson et al. (2006) examined  Ni2P/Si02 catalyst during hydrotreating using Density  Function Theory ( D F T ) and confirmed the presence o f a phosphosulfide surface. The authors combined the D F T calculations with experimental observations from literature and indicated that  43  Literature review the active phase o f the N i P consisted o f 50% sulfur replacing phosphorous and some atomic 2  sulfur deposited on the three-fold hollow sites. In summary, metal phosphides are catalytically more active in hydroprocessing than the metal sulfides. However, the conversion and selectivity o f refractory S compounds such as 4,6D M D B T using metal phosphides still needs to be enhanced to meet the new stringent S and N removal criteria. Therefore based on the previous literature reports, the next section w i l l discuss the various methods that can be used to enhance the activity and selectivity o f transition metal phosphides for hydroprocessing.  2.5  Developing a new phosphide catalyst for enhanced hydroprocessing H o (2004) reported that using sulfided catalysts, it is possible to hydroprocess to  acceptable H D S rates by vigorously tuning the process variables such as temperature, H pressure and contact time. However, high H  2  2  pressure results i n increased costs and high  temperature reduces fuel quality by cracking the o i l feedsock (Ho, 2004). enhance hydroprocessing activity o f metal phosphides hydroprocessing variables, certain catalytic approaches  Hence, i n order to  without vigorously changing are adopted to modify the  the  metal  phosphides. Similar catalytic methods that were used for enhancing activity in sulfided catalysts are being adopted i n the present study. Some o f the approaches include incorporating different promoters such as C o and N i , using acidic supports or fluorine to induce isomerisation o f the 4,6D M D B T , and increasing the hydrogenation capability o f the metal phosphide by adding noble metals. Traditionally, C o and N i are added to M o S / A l 0 3 and W S / A 1 0 3 to substantially increase 2  2  their activities in hydrotreating (Prins and De Beer, 1989) and different explanations have been  44  Literature review attributed to this enhancement effect o f C o or N i (Maxwell, 1997; Farrgher and Cossee, 1973). One postulation indicates that upon sulfidation C o forms stable C09S8 crystals and N i forms stable M3S2 i n the presence o f M0S2. The promotion o f C o was attributed to the interaction o f separate C o ions with M0S2 and that the activity decreased with the formation o f more CogSg crystals which are regarded as inactive. CogSg crystals are formed when C o / M o ratio > 0.5 (Farrgher and Cossee, 1973). C o - M o - S phase as shown by in-situ emission Mossbauer spectroscopy is responsible for enhancement o f activity o f C o M o S / A l 2 0 3 (Topsoe et al.1996). 2  Sun et al. (2004) using D F T reported that N i added to WS2/AI2O3 form stable N i S at the 3  2  W edge but not at the S edge while C o forms stable CogSg at the S edge and not at the W edge. The authors also reported that while it is possible for N i to form stable N i W S structure at high temperatures, the C o W S structure is not formed at high temperature because a high S-Co bond strength i n C09S8 makes it difficult to redisperse to the WS2 edges once the well-structured C09S& crystallites are formed. Since isomerisation o f 4 , 6 - D M D B T can enhance H D S activity, acidic supports have been the object o f numerous studies (Landau et al., 1996; Isoda et al., 1996; Yumto et al., 1997). Attempts were made with sulfided catalysts to increase the Bronsted acidity to transform the 4,6D M D B T molecule by isomerisation, so that the hetero sulfur atom can gain access to the catalyst surface for enhanced reaction (Landau, 1997). Landau et al. (1996) reported one o f the first attempts to associate acidity to the enhancement o f 4 , 6 - D M D B T . The authors prepared C o M o / A l 2 0 catalyst by impregnation o f the 3  AI2O3 with H Y zeolite and determined the activity o f the prepared catalysts with commercial C0M0/AI2O3 using 4 , 6 - D M D B T . The activity o f the C0M0/AI2O3 containing H Y zeolite was  45  Literature review reported to be three times higher than the commercial C0M0/AI2O3. The authors attributed the enhanced activity o f the zeolite containing C0M0/AI2O3 to the acidity o f the zeolite. Isoda et al. (1996) prepared C0M0/AI2O3 and incorporated 5 wt% o f Y zeolite. The authors then compared the activity o f the prepared catalyst with commercial C0M0/AI2O3 and NiMo/Al20  3  using gas o i l containing D B T and 4 , 6 - D M D B T . The authors reported the same  activity for all the catalysts for the conversion o f D B T . However, for the 4 , 6 - D M D B T , the catalyst containing Y zeolite showed better activity than the commercial catalysts. The authors reported that the desulfurised products obtained using the catalyst with modified zeolite contained 3 , 6 - D M D B T . They therefore concluded that the presence o f the zeolite produced acidity that isomerised 4 , 6 - D M D B T to the less hindered 3 , 6 - D M D B T and making it easier for the S heteroatom to react. Yumoto et al. (1997) also reported the activity o f a commercial C0M0/AI2O3, a phosphorous promoted C0M0/AI2O3 and a phosphorous promoted C0M0/AI2O3 with modified zeolite support. The authors reported that the phosphorous promoted C0M0/AI2O3 with zeolite modified support produced the best conversion among the three catalysts and they attributed the enhanced activity to isomerisation of the 4 , 6 - D M D B T . In order to better understand the role o f acidic supports i n isomerization, Figure 2.6 shows alternative catalytic routes through hydrogenation and isomerisation to enhance the sulfur exposure on the catalyst surface (Landau, 1997). These alternative routes involve some intermediate transformation o f the 4 , 6 - D M D B T molecule to undergo transalkylation or positional isomerisation o f the alkyl substituents as shown on Figure 2.6. Once isomerised, the methyl groups are far from the S heteroatom (at the 3,6- or 2,8- positions) and the steric hindrance is removed so that there is direct contact o f the S heteroatom with the catalyst active surface thereby increasing activity.  46  Literature review O n the other hand, hydrogenation o f the 4 , 6 - D M D B T leads to the formation o f the corresponding hydrogenated D M D B T and these hydrogenated D M D B T display higher H D S reactivity than the parent 4 , 6 - D M D B T (Landau, 1997).  r^N  ^ 4,6-DMDBT  o o o o o  Catalyst surface  Hydrogenation Isomerisation  o o o o co o o o o o  o o o o o  Figure 2.6 Alternative catalytic routes for conversion, 4 , 6 - D M D B T (Landau, 1997)  2.5.1  Promotional effect of a second metal The earlier discussion on addition o f C o or N i to M0S2/AI2O3 or WS2/AI2O3 revealed that  the promotion results in increased activity (Kabe et al., 1999) but metal phosphides are not  47  Literature review promoted by C o or N i (Stinner et al., 2001; Zuzaniuk and Prins, 2003; Rodriguez et al., 2003). A s observed i n earlier reports (Mizutani et al., 2005; Stinner et al., 2001; Zuzaniuk and Prins, 2003; Rodriguez et al., 2003; L i et al., 2005) C o or N i were added such that the C o / M o was 1/1 but as noted, at C o / M o > 0.5 sulfidation o f C0M0S2/AI2O3 produces the less active CogSg phase (Kabe et al., 1999). Therefore in the present study, the C o w i l l be added to the M o such that C o / M o < 0.3. It is expected that with the small amounts o f C o added to M o P , the dispersion w i l l increase leading to the formation o f stable and active sites on the catalyst.  2.5.2  Promotional effect of a third metal component  A combination o f supported C0M0P and N i M o P on alumina is used to form trimetallic (combination o f three metals) catalyst o f C o N i M o P on alumina but the reported results using them for H D N and H D S are contradictory and in addition, studies on the use o f trimetallic CoNiMo/Al203 is limited and the active phases and sites are not yet known (Kabe et al., 1999; G i l et al., 1983; Laine et al., 1984; Benyamna, et al., 1998; K a n g et al., 1998; Severino et al., 2000). Recently, H o m m a et al. (2005) reported the activity o f sulfided C o N i M o / A b O s tested for thiophene H D S and the rates were lower than the corresponding sulfided MM0/AI2O3 or C0M0/AI2O3. The authors attributed the low activity to the formation o f a mixed C o and N i sulfide on the alumina surface which decreased the number o f active sites. The number o f active sites is related to the sulfur-vacancies and measurements o f the adsorption parameters showed that C0M0 and N i M o were higher than C o N i M o catalysts (Homma et al., 2005).  48  Literature review 2.5.3  Effect of supports  2.5.3.1  Use of alumina  Hydroprocessing catalysts are supported on alumina because alumina has high textural and mechanical properties with average crushing strength o f 100 N/granule. A l u m i n a is also relatively cheap (Euzen et al., 2002). A l u m i n a also has the ability to regenerate catalytic active sites after intense use under hydrotreating. Earlier, it was thought that supports such as alumina were inert, but it is now evidenced that supports do interact with the promoters and the active phase (Topsoe and Clause, 1981; Carrier et al., 1997; Candia et al., 1984). These interactions result in changes in the morphology o f the active phase and creation o f acidic and basic sites. Whereas alumina is a very good carrier for sulfided catalysts, it is an inferior carrier for metal phosphides because it reacts with phosphate to form aluminum phosphates on the surface and therefore the active metal phosphide phase is not available (Mangnus et al., 2003; Stinner et al., 2002; Nagai et al., 2005; Sawmill et al., 2005). Silica was reportedly a superior support because o f its weak interaction with the phosphates (Robinson et al., 1996). However, recently, L i et al. (2005) and Wang et al., 2005) reported that the activity o f phosphides on AI2O3 is increased i f the reduction temperature is carried out at or above 900 °C and the authors attributed the observed increase in activity to the re-appearance o f the M o P active phase at this temperature. Therefore the role o f supports in relation to enhancing the activity o f the 4 , 6 - D M D B T is important and w i l l be discussed.  2.5.3.2 Use of acidic supports Based on previous reports (Landau, 1997; Prins et al., 2006), the present study w i l l incorporate Bronsted acidity on the metal phosphides in anticipation o f causing isomerisation o f 4 , 6 - D M D B T to enhance H D S activity and selectivity.  49  Literature review 2.5.3.3 Effect of addition of fluorine to alumina Fluorine has been used i n some refineries as a promoter o f hydrotreating and hydrocracking (Kasztelan et al., 1999). Benitez et al. (1996) fluorinated W / A 1 0 and N i W / A l 0 2  3  2  3  and reported  increased H D N o f pyridine. The authors attributed the enhanced activity to changes i n morphology o f the W S crystallites (higher stacking and larger pores), high surface acidity, and 2  better sulfidation o f the W oxidic phase. V a n Veen et al. (1993) reported enhanced H D N activity after fluorination o f N i M o / A l 0 . The authors attributed the enhancement to the interaction o f 2  3  the N i M o S phase with the A 1 0 that leads to increased stacking o f the M o S 2  3  2  slabs and better  accessibility o f the active sites for adsorption. Hence with higher reactants adsorbing on the active sites, high H D N activity is obtained. Q u and Prins (2003) reported a small increase i n H D N o f methylcyclohexylamine after fluorinating sulfided N i M o / A l 0 2  3  at 310-350°C and 5.0  MPa. Lewandowski et al. (1997) also reported the presence o f Bronsted acidity on sulfided NiMo/Al 0"3 on addition o f fluorine and proposed that hydrogenation takes place on anion 2  vacancies o f M o - S while the breaking o f the C - N hydrogenolysis bond is related to the Bransted acid centers. It is clear from earlier reports (Fierro et al., 1991; Hughes et al., 1969) that addition o f fluorine enhances acidity o f the catalyst which promotes isomerisation. Recently, D i n g et al. (2006) reported the activity o f sulfided N i M o / A l 0 2  3  using 4 , 6 - D M D B T . The authors reported  that after modifying the support with fluorine, the conversion increased from 20 wt% to 50 wt% and explained that the increased conversion was attributed to the increased acidity after fluorination. However, the authors reported that the H D N activity decreased from 60 wt% to 35 wt% when fluorine was added. Most o f the reports on the addition o f fluorine are reported for sulfided catalysts rather than phosphide catalysts. It is therefore important to study how  50  Literature review fluorination  affects the H D S activity o f phosphides since fluorination enhances the Bronsted  acidity on the metal phosphides that leads to isomerisation and increased activity (conversion and selectivity).  2.6  Effect of platinum Hydrogenation in the industry is usually accomplished using supported N i or Pt catalysts. Pt  and N i are deposited on alumina and silica in commercial applications such as aromatics hydrogenation. However, in hydrodesulfurisation, Pt is normally poisoned by small amounts o f sulfur and it was suggested that on Y zeolite support, Pt transfers electrons to the zeolite acid centers thereby enhancing the catalytic activity towards hydrogenation reactions (Gallezot et al., 1977). A l s o on supported solid acids such as H Y Zeolite and M C M , the higher hydrogenation activity (relative to AI2O3 support) was attributed to the presence o f hydrogen spill over i n the metal-support interface region and aromatics could be adsorbed as carbonium ions, which tend to react with the spillover hydrogen (Wang et al., 1998). Rynkwoski et al. (2004) reported that Pt added to N i on supports enhanced the hydrogenation activity o f toluene. Recently, Pinzon et al. (2006) reported that P t - M o / A h C ^ produced high H D S activity in comparison to non-platinated M0/AI2O3. The authors attributed the enhanced activity to the synergy between the Pt and M o i n a way that Pt maintains the surrounding M o particles in a more active form. The authors further explained that the activity is also due to high dispersion o f Pt as Pt interacts with the alumina support to suitably place the Pt and M o in a location on the alumina. It is expected that the enhanced hydrogenation provided by the addition o f platinum to the phosphides w i l l result i n increased conversion o f 4 , 6 - D M D B T as more hydrogenated desulphurization as illustrated earlier i n Figure 2.6.  51  intermediates  will  enhance  Literature review 2.7  Effect of Process Variables Tailoring process conditions such as temperature, pressure and space velocity for  optimum operations o f the catalyst w i l l enhance activity. These process conditions can be tuned without any new capital investment to the existing reactor. Haldor-Topsoe (EPA-Diesel, 2000) reported that an increase in reaction temperature o f 14 °C reduced the sulfur level o f Straight R u n Light Gas O i l ( S R L G O ) and Light Cycle O i l ( L C O ) from 140-120 ppm while U O P projected that a 20 °C increase o f reactor temperature w i l l reduce the sulfur from 140 -120 ppm. However, increase in reaction temperature results i n decrease o f catalyst life hence, catalysts w i l l need to be changed a little more frequently which consequently w i l l increase cost. Lappas et al. (1999) reported that using sulfided C0-M0/AI2O3, they were able to reduce sulfur content o f diesel and also increase hydrogenation with increasing temperature. The authors also observed that although an increase in reactor temperature increases hydrogenation, the products are limited by thermodynamics. In equation 2.5, thermodynamics requires the equilibrium to move to aromatics at high temperature as the reaction is exothermic.  Aromatics  +  <=> Naphthenes  (2.5)  However, Wang et al. (2005) recently studied the H D S o f dibenziothiophene ( D B T ) over supported Ni2P/MCM at different reduction temperatures and reported that below reduction temperatures o f 673 K , the H D S activity did not increase with increase in the reaction temperature. In order to determine the reaction rate constant, the catalyst w i l l be examined at different space velocities. Space velocity also influences the product distribution.  52  Literature review Summary In summary, reports on the activity o f metal phosphides used for hydroprocessing indicate that metal phosphides have higher activity than metal sulfides. Refractory S compounds such as 4 , 6 - D M D B T and N compounds such as carbazole must be removed from the residue o f conventionally hydroprocessed crude o i l since they form a substantial portion o f the residue. 4,6D M D B T are refractory and resist hydroprocessing because the methyl groups at the 4- and 6positions possess steric hindrance that prevent the hetero S atom o f 4 , 6 - D M D B T from reacting. However, by enhancing the Bransted acidity on metal phosphides, it is possible to isomerise the methyl groups and allow the S hetereoatom to react, thereby increasing the conversion and selectivity. C o M o P and M o N i P prepared such that the molar ratio o f C o : M o and M o : N i in C o M o P and M o N i P respectively were 1:1, were reported to have lower activity than M o P and  Ni2P. The addition o f small amounts o f C o and N i to M o P and N i P has not been previously 2  examined. Reports on the use o f metal phosphides for H D N o f carbazole are few. However on metal sulfides many reports indicate that Bransted acidity enhances C - N hydrogenolysis. In addition, F added to metal sulfides also enhanced the hydroprocessing activity o f metal sulfides. Therefore it is expected that incorporating Bransted acidity either by use o f acidic supports or by F w i l l also enhance activity o f the metal phosphides.  53  Experimental  Chapter 3 Experimental Since the present research deals with the development o f active catalysts for H D S and H D N , experiments were carried out to assess the activity o f different prepared metal phosphide catalysts using 4 , 6 - D M D B T for H D S and carbazole for H D N . A fixed-bed reactor was used i n the study. In addition to describing the experimental procedures and methods o f catalyst preparation, this chapter w i l l describe various catalyst characterization methods used i n the present study that included, Temperature Programmed Reduction (TPR), Powder X-ray Diffraction  (XRD),  Energy  Dispersive X - R a y  Emission  (EDX),  X-ray  Photoelectron  Spectroscopy ( X P S ) , n-propyl amine chemisorption, C O uptake and Transmission Electron Microscopy ( T E M ) . In the first part o f the chapter, the preparation o f the bulk C o promoted M2P, M o P and the supported forms o f these catalysts are described. The catalyst activity tests, product analysis and characterization for H D S o f 4 , 6 - D M D B T is also presented. In the second section, the preparation o f bulk and supported N i promoted M o P , the catalyst activity tests for H D N o f carbazole and the product analysis, is presented.  3.1  Preparation of metal phosphides for hydrodesulphurization C o and N i are known promoters o f sulfided commercial hydroprocessing catalysts and  their addition to metal phosphides is investigated here. The bulk metal phosphides M2P, M o P , C o P and C o P were prepared, together with Co Ni2P. The preparation o f M2P, M o P and C o P has 2  x  already been reported i n the literature (Oyama, 2003). In preparing C o N i P , the concentration o f x  2  Co was varied for the Co Ni2P and once the optimum C o content was obtained, a solution x  containing the optimum amounts o f Co, M o and P was used to impregnate A l 0 3 , yielding the 2  54  Experimental supported phosphides. The impregnated bimetallic and supported phosphides were further impregnated with platinum and fluorine.  3.1.1  Preparation of the bulk metal phosphides Co Ni2P and C o M o P x  x  B u l k N i P , M o P , C o P , C o N i P (0 < x < 0.8) and C00.07M0P were initially prepared as 2  2  x  2  described elsewhere (Oyama et al., 2001; Stinner et al., 2001; Stinner et al., 2000) to provide reference activity data for the synthesis o f supported phosphides. The preparation o f C o N i P x  2  w i l l be illustrated here (See Appendix B for detailed calculations). The C o N i P catalysts were prepared by dissolving stoichiometric amounts ( N i / P = 2/1) x  2  o f nickel nitrate ( N i ( N 0 3 ) . 6 H 0 , A l d r i c h 99% purity) and diammonium hydrogen phosphate 2  2  ( D A H P , ( N H 4 ) H P 0 4 Sigma 99% purity) i n a beaker containing 15 m l o f de-ionized water. The 2  resulting solution was stirred at room temperature while adding 10 m l o f a cobalt nitrate solution, prepared by dissolving an appropriate amount o f cobalt hexahydrate ( ( C o ( N 0 3 ) . 6 H 0 , Acros) in 2  2  10 m l o f de-ionized water to give the desired m o l % C o in the final C o N i P catalyst. x  After  2  addition o f the cobalt solution the mixture was stirred while evaporating to dryness. The material was further dried in an oven at 373 K for 2 h and calcined at 773 K for 6 h. The calcined product (or C o N i P precursor) was ground to a powder (dp < 0.7 mm) and subjected to temperature x  2  programmed reduction ( T P R ) in H (Praxair - 99.99%) at a flow rate o f 150 ml(STP)/min and a 2  temperature ramp rate o f 1 K / m i n to a final temperature o f 1000 K . The final temperature was held for a period o f 2 hours. After the reduction, the catalyst was cooled to room temperature i n He at a flowrate o f 20 ml/min. Prior to removal from the reactor, the C o N i P was passivated i n x  2  2% 0 / H e for 2 h at room temperature. The C00.07M0P was prepared similarly using ammonium 2  heptamolybdate ( ( N H ) 6 M o 0 2 4 . 4 H 0 - A n a l a R ( B D H ) 99% purity), D A H P and C o ( N 0 ) . 6 H 0 4  7  2  3  55  2  2  Experimental to give 3.2 m o l % C o in M o P . The mixture was treated thermally as before, except that the reduction temperature was increased to 1200 K . The steps for the temperature programmed reduction used for reducing the metal phosphide precursors is outlined as follows: 1. Heat sample from room temperature to final temperature (1000 K - 1200 K ) at a rate o f 1 K / m i n in H flow at 150 ml/min per gram sample. 2  2.  Maintain at final temperature for 2 h.  3.  C o o l to room temperature in He flow at 20 ml/min.  The catalysts were compared to bulk C o P , M o P , N i P and C o P prepared using 2  2  stoichiometric amounts o f C o ( N 0 3 ) . 6 H 0 and (NH4) HP04 for C o P , stoichiometric amounts o f 2  2  ( N H ) 6 M o 0 4 . 4 H 0 and ( N H ) H P 0 4  7  2  2  4  2  4  2  for M o P , and stoichiometric amounts ( N i / P = 2/1 mole  ratio) o f nickel nitrate N i ( N 0 3 ) . 6 H 0 and (NH4) HPC>4 in the case o f N i P . A l l solutions were 2  2  2  2  treated thermally as before, although in the absence o f the cobalt nitrate solution the initial drying time was shorter for the monophosphides.  3.1.2  Preparation of supported metal phosphide catalysts for hydrodesulfurization of 4,6-dimethyldibenzothiophene. Coo.4Ni P was supported on alumina (neutral and weakly acidic) and M o b i l Catalytic 2  Material ( M C M - c o m p o s e d o f amorphous silica with ordered structure and uniform mesopores). Generally, supported catalysts provide higher surface area than bulk catalysts and hence enhanced activity. For example, the B E T surface area o f the A 1 0 2  3  (Aldrich- P V = 0.76 ml/g,  pellet size = 4.5 mm), used i n the present study is 155 m /g whereas the prepared bulk Coo.4Ni P 2  is only 7.9 m /g. It is expected that by supporting the Coo.4Ni P on the A 1 0 , the active 2  2  3  Coo.4Ni P phase w i l l result in increased dispersion and hence more active sites for reaction. The 2  56  Experimental M C M is an acidic support and it is expected to provide the Bransted acidity necessary to isomerise the methyl groups away from the steric hindered position on 4 , 6 - D M D B T to a less hindered position such as 2 , 8 - D M D B T . The procedure for preparing the supported catalysts is similar to that o f the bulk except that in this case the support was impregnated with aqueous solutions o f the desired species. A two-step impregnation was performed to avoid the formation o f an insoluble precipitate o f cobalt phosphate upon mixing aqueous solutions o f Co(N03)2.6H20 and ( N H ) H P 0 4 . The modified 4  2  N12P/AI2O3 and M0P/AI2O3 catalysts were prepared using multiple impregnations. The procedure for preparing the Coo.4Ni P/Al203 w i l l be described (see Appendix B for detailed 2  calculations). The catalysts were prepared with theoretical amounts o f 3 wt% o f Co. The AI2O3 support ( B E T surface area, 155 m /g, pore volume, 0.76 ml/g, pellet size, 4.5 mm) was dried over night before use. Impregnation solutions were prepared by dissolving Ni(N0 )2.6H20, and 3  (NH4)2HP04 in de-ionised water to give the required amount o f 15 wt% N i in Coo.4Ni P/Ai203. 2  The contents were mixed thoroughly and impregnated on the support. The impregnating solution was added in a stepwise procedure to the alumina until incipient wetness. The resulting product was dried i n an oven at 393 K to remove the moisture. The impregnation and drying was carried out several times until all the solution was used up. After this, the Co(N03)2.6H 0 was dissolved 2  in 5 m l o f de-ionised water to give a solution containing 3 wt% o f C o i n C00.4N12P/AI2O3 and impregnated on the support. The precursors obtained were dried for 3 h in the oven and then calcined for 6 h at 773 K . The reduction process was similar to that used for the bulk phosphides using the same heating rate o f 1 K min" but at a higher reduction temperature o f 1200 K (see 1  section 3.11 for temperature programmed treatment). To prepare the fluorinated samples, the same procedure as described previously for the AI2O3 support was used except that the AI2O3 support was impregnated first with ammonium  57  Experimental fluoride hydrate (NH4F, Acros) solution to give an equivalent o f 1 wt% o f fluorine and the resulting product was dried. The loading o f the C o and M o on the support, the heat treatments and the reduction processes were carried out as described earlier. In the case o f the platinated Coo.4Ni2P/Al203 catalyst, which was prepared for activity measurements o f 4 , 6 - D M D B T , the impregnation o f the Pt solution was done after preparing the Coo.4Ni P/Al203 precursor. Hydrogen hexachloro platinate (IV) hydrate ( H P t C l 6 . 6 H 2 0 , A l d r i c h 2  2  99.9%) was dissolved in 5 m l o f de-ionised water to give an equivalent o f 5 wt% o f Pt i n the Coo^lS^P/AbCb. The resulting solution was then impregnated to the Coo.4Ni2P/Al203 precursor using the impregnation procedure as described in section 3.1.2. During impregnation with the Pt solution, it was important to physically observe when the Coo.4Ni2P/Al 03 precursor was just wet. 2  The resulting product was calcined, reduced and passivated using the same procedure  as  described in section 3.1.1.  3.2  Preparation of Ni MoP as hydrodenitrogenation catalyst x  The procedure for preparing the bulk N i M o P with different N i concentration was similar x  to that described for C o N i P in section 3.1.1. The only exception being that different N i x  2  loadings on M o P were obtained by dissolving first stoichiometric amounts ( M o / P = 1) o f (NH4)6Mo 024.4H 0 and ( N H ) H P 0 to form a solution containing M o P . Solutions containing 7  2  4  2  4  different amounts o f N i ( N 0 3 ) 2 . 6 H 2 0 to give the required loadings o f N i were then added to the M o P solution. After preparing the precursor, the calcination, reduction and passivation followed a similar procedure as described earlier in section 3.1.1. In preparing the supported catalysts, similar procedures as described for the preparation o f Coo.4Ni2P/Al203 were followed. The Nio.33MoP was impregnated on the support. The case o f supported Nio.33MoP/Al203 w i l l be described. The N i o . M o P solution was prepared with 33  58  Experimental theoretical amounts o f 3 wt% o f N i and 15 wt% o f M o in the Nio.33MoP/Al203. The AI2O3 support ( B E T surface area, 155 m /g and pore-volume 0.76 ml/g) was dried overnight before use. First, the M o P impregnation solution was prepared by dissolving ( N H ^ M o y O ^ F k O , and (NH4)2HP04 in de-ionised water to give 15 wt% M o i n Nio. 3MoP/Al203 (see Appendix for 3  detailed calculations and procedure). In each step o f impregnation, the solution was added until the alumina was observed to be just wet. The resulting product was dried in an oven at 397 K to remove the moisture. The step-wise procedure o f impregnation and drying was carried out several times until all the solution was used up. After this the N i ( N 0 3 ) 2 . 6 H 0 solution containing 2  the 3 wt% o f N i was impregnated onto the support. Once the precursor was obtained, the calcination, reduction and passivation was done in a similar procedure as i n section 3.1.1.  3.3  Catalyst Characterization After the catalysts were prepared, different characterization methods were used to  ascertain whether the required catalysts had been prepared and also to determine the properties o f the catalysts. Some used catalysts were also characterized. The different types o f characterization techniques w i l l be discussed below.  3.3.1  Temperature Programmed Reduction (TPR) Temperature programmed reduction was carried out using the calcined precursors. The  purpose o f carrying out the T P R was to determine the temperature o f reduction and to study the promotional effects o f C o loading on the reduction o f the main metal M o . T P R can also be used to determine the degree o f reduction i f the correct stoichiometry is known and all products are quantified.  59  Experimental T P R o f the catalyst precursors was conducted in H 2 using 0.2 - 0.4 g o f the calcined sample loaded into a stainless steel reactor (i.d. = 9 mm). A r , flowing at a rate o f 60 m l (STP)/min, was passed through the reference side o f a thermal conductivity detector (TCD) and a similar volumetric flow o f 10% H 2 i n A r passed through the reactor before entering the sample side o f the detector. A furnace with a temperature programmable controller (Omega CN8260) was used to heat the reactor at 1 K / m i n to the final temperature o f 1000 K or 1200 K . The reactor was held at this temperature for 2 h before cooling to 313 K at a rate o f 5 K / m i n . The T P R was calibrated by reducing C U 2 O to C u in H 2 flow at the same conditions as the samples. The T P R profile o f C U 2 O was integrated and based on the stoichiometry (assuming complete reduction) the H 2 consumption per area was obtained and used as the calibration for the T P R (see Appendix B ) . The  degree o f reduction was determined from the integrated area o f the T P R o f the  sample, the assumed stoichimetry o f the precursor reduction reaction and the T P R calibration. First, the T P R profile o f the sample was integrated to obtain the area. Next, from the stoichiometry o f reduction o f the sample precursor, the theoretical m o l o f H 2 was determined assuming complete reduction. From the integrated area o f the sample, the m o l o f H 2 consumed during the T P R o f the sample was obtained. The degree o f reduction was then calculated as the moles o f H 2 consumed divided by the theoretical mol H 2 required for complete reduction (see Appendix B for sample calculation o f degree o f reduction).  3.3.2  Temperature  programmed  reduction  using  tapered  element  oscillating  microbalance ( T E O M ) The mass change o f the catalyst during T P R was also measured for the C o P , N12P and Coo 08Ni2P catalyst using a tapered element oscillating microbalance o f resolution 10 g ( T E O M 6  60  Experimental Series 1500 Pulsed Mass Analyzer, Rupprecht and Patashnick). The catalyst was placed i n a fixed-bed configuration and continuous gas flow through the bed during the mass measurement. About 0.1-0.2 g o f catalyst was placed i n the reactor and heated in He ( U H P , Praxair) at a flow rate o f 60 ml(STP)/min and 393 K for 2 h before cooling to room temperature. The reduction i n H  2  followed using a purge H e ( U H P , Praxair) flow o f 120 ml(STP)/min and a pure H  2  (UHP,  Praxair) flow o f 120 ml(STP)/min. A l l gases were measured using Brooks 5878 mass flow controllers. The temperature was ramped at 1 KVmin to 979 K , holding at this temperature for 2 h before cooling to room temperature (see Appendix B for a sample o f the T P R output using TEOM).  3.3.3 Powder X-ray Diffraction (XRD) Powder X-ray diffraction ( X R D ) analyses were performed on the passivated catalysts using a Siemens D500 C u K a x-ray source o f wavelength, 1.54 A. The passivated catalysts were initially ground to powder using a pestle and mortar. Next, ethanol was added to the ground sample and mixed thoroughly to form a paste. The paste was carefully loaded on a glass slide, which was subsequently mounted on a platform. The analysis was performed using a 40 k V source, a scan range o f 3-70  0  with a step size o f 0.04  0  and step time o f 2 s. The phase  identification was carried out after subtraction o f the background using standard software. Crystallite size estimates were made using the Scherrer equation, d = KA/pcos9 where the c  constant K was obtained from the integrated areas o f the sample and the standard B a F , X, is the 2  wavelength o f radiation, P is the peak width i n radians and 9 is the angle o f diffraction.  61  Experimental 3.3.4  Brunnauer-Emmett-Teller (BET) Surface Area The B E T surface areas o f the freshly prepared phosphide catalysts were measured to  determine the total surface area that is available on the catalysts for reaction. The single-point B E T surface area o f the catalysts was measured by N adsorption-desorption at 77 K using a 2  Micrometrics F l o w S o r b l l 2300 analyzer. About l g o f passivated catalyst was degassed at 393 K for 2 h and the measurement was made using 30% N2/He mixture fed at 15 ml/min.  3.3.5  X-ray Photoelectron Spectroscopy (XPS) and Inductive Coupled Plasma X P S was performed on all passivated and some spent catalysts to determine the chemical  compositions o f the surface and their relative proportions. A Leybold Max200 X-ray photoelectron spectrometer was used for these analysis. A l Kct was used as the photon source for all the metal phosphides except the Co Ni2P phosphide where M g K a was used because o f the x  N i ( A ) overlap with the C o 2p peak when the former is used as the X-ray source. Both A l K a and M g K a radiation was generated at 15 k V and 20 m A . The pass energy for the survey scan was set at 192 e V and for the narrow scan it was 48 e V . A l l catalyst samples were analysed after passivation at room temperature. Exposure of the samples to ambient atmosphere was minimized by transferring the samples either i n vacuum or under nitrogen ( U H P ) . A l l X P S spectra were corrected to the C i s peak at 285.0 e V . The catalysts chemical compositions were determined using I C P by Cantest Laboratories, Vancouver, B C .  3.3.6  Carbon monoxide (CO) Uptake Chemisorption is carried out using different probe molecules such as C O , O2, H2 and N O .  O2 and C O have been used for chemisorbing on metal phosphides. However, on metal sulfides,  62  Experimental oxygen can reduce the sulfide to oxide surface at high temperature and so O2 chemisorption can only be carried out at -78 °C (Prins and D e Beer, 1989). Similarly on metal phosphides, the surface can be reduced to metal oxide at high temperature. C O chemisorption has been successfully used to titrate metal phosphides but the disadvantage is that depending on the adsorption conditions and the particle size, varying adsorption stoichiometry is obtained. However, despite this disadvantage, repeatability data was obtained i n the present research using C O chemisorption (see Appendix B.9). Since the reduced metal phosphides were passivated, it was necessary to re-reduce the prepared metal phosphides to remove the passivation layer before performing the C O uptake measurements. The purpose o f measuring the C O uptake was to determine the metal dispersion o f the catalyst. The C O uptake was measured using pulsed chemisorption. About 0.5 - 1.0 g o f catalyst was loaded in a stainless steel reactor (i.d. 9 mm). The re-reduction to remove the passivation layer was carried out as follows: 1. 60 ml/min o f 10% H i n A r was passed through the sample loaded i n the reactor. 2  2. The sample was heated from 313 to 723 K at a rate o f 2 K / m i n and maintained at the final temperature for 1 h. 3. The sample was then cooled to 298 K in a flow o f H . 2  4. He at 30 ml/min was used to flush the catalyst for 30 m i n i n order to achieve an adsorbate-free, reduced catalyst surface. After pre-treatment, 1 m l pulses o f C O were injected into a flow o f H e (30 ml/min) and the C O uptake was measured using a T C D . C O pulses were repeatedly injected until the response from the detector showed no further C O uptake after consecutive injections and the total C O uptake, reported here in units o f pmol/m for the bulk catalysts, was obtained by dividing the  63  Experimental uptake in pmol/g by the measured B E T area. Assuming a 1:1 adsorption stoichiometry between C O and metal atoms, this value corresponds to the metal site density on the surface o f the bulk catalyst.  3.3.7  n-Propyl amine (n-PA) chemisorption Titration o f the acid sites on metal phosphides can be achieved using NH3 or n - P A . NH3  titrates both the Bronsted acid sites and Lewis acid sites while n - P A titrates only Bronsted acid sites (Farneth and Gorte, 1995). When carrying out chemisorption using n - P A , it is important to note that the temperature programmed desorption (TPD) o f n-propylamine on zeolite ( Z S M 5 ) monitored by a mass spectrometer shows the n-propylamine desorbing at temperatures less than 573 K and the T P D products, propylene and N H 3 desorbing above 573 K . The propylene desorbs earlier than the N H . However, when the decomposition o f n-propylamine is monitored by T C D , 3  there is one main peak observed (Tittensor et al., 1992) hence the procedure is T P D process. The passivated catalysts were first re-reduced using the procedure as described in section 3.3.7 and used for n - P A chemisorption. The catalyst Bronsted acid sites were titrated by n-propyl amine (n-PA) temperature-programmed desorption using the same reactor and T C D used for the C O chemisorption. A flow o f He (30 ml/min) saturated at room temperature with n-propyl amine (Aldrich, 99.8%) passed through heated gas lines to the reactor containing about 1 g o f catalyst. After a 2 h adsorption at 383 K , the reactor was flushed i n pure He at a flow rate o f 30ml/min for 1 h to ensure that physically adsorbed n - P A was removed. The chemisorbed n - P A was then desorbed by ramping the reactor temperature from 383 to 973 K at a rate o f 5 K /min and the T C D was used to quantify the amount o f n - P A desorbed. The system was calibrated using 3 zeolite samples o f known acidity (see Appendix B ) .  64  Experimental 3.3.8  Scanning Electron Microscopy-Energy Dispersive X-Ray Emission (SEM-EDX) A multi-point S T E M - E D X was used to analyze the bulk  M2P, the C00.08N12P and the  Coo.34Ni2P in order to determine the uniformity i n the bulk composition. Before the analysis, the samples were crushed using a pestle and mortar, mounted on a small magnetic circular steelholder with a carbon felt background and mounted on the instrument. The energy dispersion xray emission was done on a Hitachi S300N scanning electron microscope at 20 k V .  3.3.9  Transmission Electron Microscopy ( T E M ) TEM  was performed  under  high-resolution mode  on a Hitachi H7600 electron  microscope operated at 200 k V . The freshly prepared phosphide catalysts were ground to a fine powder with a pestle and mortar. The powder was dispersed in ethanol and deposited on a carbon-coated copper grid. The specimen was dried in air prior to analysis.  3.4  Catalyst activity measurements for hydrodesulfurization of 4,6-dimethyldibenzothiophene and hydrodenitrogenation of carbazole The activity measurements for hydroprocessing using the prepared phosphides were carried  out i n a fixed bed reactor (i.d. = 9 mm) operated at 583 K and 3.0 M P a H2. The flow diagram o f the reactor set-up is shown i n Figure 3.1. A 3000 ppm solution o f 4 , 6 - D M D B T dissolved i n dodecane was used for the H D S activity study. A 3000 ppm solution o f carbazole in xylene was used i n the H D N o f carbazole. The 3000 ppm o f carbazole was higher than the 700 ppm found in heavy o i l because it is necessary to test the limits o f the activity o f the prepared modified metal phosphides. The resulting solution was fed to the reactor using a Gilson M o d e l 0154E metering  65  Experimental pump. Prior to entering the reactor, the liquid was evaporated into a stream o f flowing H2. Gas and liquid flows and catalyst charged to the reactor were chosen to give a range o f space velocities o f 7.9-32 x 10" mol/(h gcat). One gram o f the passivated catalysts ( d < 0.7 mm) was p  loaded into the reactor and supported with quartz wool while the rest o f the reactor volume was filled with glass beads o f average diameter 0.1 m m . A thermocouple was placed close to the top o f the catalyst bed to control the reaction temperature. A l l gas flows were controlled using calibrated Brooks 5878 mass flow controllers. Prior to activity measurements, the catalysts were pre-treated at 723 K for 2 h in H2 at a flow rate o f 160 ml(STP)/min. The temperature was then cooled to the reaction temperature o f 583 K and the reaction initiated using the appropriate feed flow conditions. The product was collected every two hours i n sealed vials and analyzed off-line using a 3400 G C Varian Star Gas Chromatograph equipped with a flame ionization detector (FID). Component separation was achieved using a capillary column ( C P - S i l 19 C B , 25m length and 0.53 m m i.d.). Component identification was confirmed using the same column and a G C M S (Agilent 6890/5973N). In two cases, sulfur speciation among components o f the product was also measured using an HP6890 G C equipped with a G 2 3 5 0 A A E D detector and a R E S T E K 10526 crossbond 50% methylpolysiloxane column. In this case, component identification was confirmed by a T O F M S and these analyses were conducted at the National Center for Upgrading Technology, Devon, Alberta. The reactor was operated in the gas phase as an isothermal, plug-flow reactor, as considered by the calculations given in Appendix A .  66  Experimental  Gilson Precision Liquid Pump  Brooks Mass Flow controllers Nitrogen  Brooks Mass Flow controllers  Hydrogen r  *— r Liquid Product  Figure 3.1  Schematic Diagram o f the Experimental Set-Up  67  Chapter 4  Chapter 4  Hydrodesulfurization of 4,6-dimethyldibenzothiophene over bulk and supported metal phosphides This chapter w i l l present data on bulk and supported metal phosphides for the H D S o f 4 , 6 - D M D B T , a refractory, sterically hindered S-compound used as a model reactant for H D S . First, the chapter w i l l present results on the bulk metal phosphides such as M2P, M o P , CoP, Co P, 2  C0M0P and C o N i P (0.08 < x < 0.8). The bulk materials were studied so that the x  2  measured activity and selectivity could be related to the catalyst properties without interference from the complexity o f the catalyst support or promoter. In this chapter, characterization results will  be followed  by a discussion o f the activity o f the metal phosphides. Then, the  characterization and activity data o f supported metal phosphides w i l l be presented.  4.1  Characterization of bulk metal phosphides Characterization o f the catalyst precursor was done by T P R , X R D , B E T and X P S were  performed  on the passivated metal phosphides while the C O uptake  and the n - P A  chemisorption were done on the re-reduced catalysts. The characterization was performed to determine the properties o f the catalysts and to ascertain whether or not the required planes and Note: A version of this chapter has appeared in J. Catal. 241 (2006) 356-366  68  Chapter 4 compositions had been synthesized. The other objective o f characterizing the catalysts was to provide the data needed to correlate catalyst properties with the catalyst H D S activity.  4.1.1  X R D of the prepared catalysts Whereas some o f the diffraction patterns o f the calcined precursors were x-ray  amorphous, (see appendix B.5) N i O was identified in the rest o f the Co Ni2P precursors in x  agreement with similar observations made for M o P and N i P precursors (Wang et al., 2002). The 2  results suggest that after calcination, N i was i n the +2 oxidation state and M o was likely i n the +6 oxidation state (Zuzaniuk and Prins, 2003) although the presence o f other amorphous species cannot be excluded (Wang et al., 2002). After preparing the metal phosphides by T P R o f the oxidic precursors, X R D was used to determine the bulk phases present. Figure 4.1 shows the X R D profiles o f N i - r i c h N i P , C o N i P , 2  x  2  C o P and the Co-rich C o P . In all cases the diffractograms showed that metal phosphides had 2  been successfully prepared. The diffractograms were compared with Powdered Diffraction Files (PDF) ( J C P D S , 2005) and showed that C o P was obtained from samples with C o / P ratios o f 2/1 2  and C o P was obtained with C o / P with a ratio o f 1/1 as expected. N o metal oxide or metal phosphate species were detected. The X R D data for the C o N i P catalysts (Table 4.1), with 0 < x  2  x < 0.16 (i.e. up to 5.1 mole % Co) showed the presence o f N i P with no significant difference 2  in the N i P lattice parameters (a = 0.5877 ± 0.0003 nm and c = 0.3412 ± 0.0008 nm). For x > 2  0.34, the C o N i P showed the development o f the metal rich phosphides N i i P s and C o P , in x  2  2  2  addition to N i P . The N i P lattice parameters decreased (a = 0.5862 ± 0.0002 n m and c = 0.3375 2  2  ± 0.0021 nm), although the error associated with the estimate o f the lattice parameter c increased  69  Chapter 4  A\  CoP  Co  f t 7 0  0.79  AA  NLP  •  2  •  Co^NLP 0.34  2  Co^JvILP  1  Co  A  0.16  0.08  2  NiP 2  NLP ~i—i—i—r—i—i—i—i—r  30  i  i  I  40  I  I  I  I  I  T  I  I  I  |  50  20 Figure 4.1  X-ray diffractograms of reduced C02P, C o P and CoxNi2P catalysts (A - N i i P , 2  5  • - Co P, T - NiCoP) 2  70  |  |  |  T™  60  Chapter 4 Table 4.1 Lattice parameters estimated from P X R D o f reduced catalysts  N i P phase 2  Catalyst  Co  Phases  content  20, degree  mole % Ni P  Lattice parameters,  Crystallite  nm  size, nm  (111)  (201)  (210)  (300)  a  c  do ( H I )  0  Ni P  40.52  44.44  47.16  54.00  0.5879  0.3417  21 ± 2  Co .o8Ni P  2.5  Ni P  40.52  44.40  47.20  53.96  0.5879  0.3417  36 ± 2  Co ., Ni P  5.1  Ni P  40.60  44.48  47.14  54.16  0.5874  0.3403  39 ± 2  Co 4Ni P  10.3  Ni P, Ni, P  40.88  44.76  47.32  54.16  0.5860  0.3360  32 ± 2  40.66  44.59  —  0.5863  0.3389  27 ± 3  2  0  0  2  6  0 3  2  2  2  2  2  2  2  5  Ni P, 2  Co .79Ni P 0  2  21.0  Ni, P , 2  5  Co P 2  M o P ph ase Lattice parameters,  20, degree  nm  Crystallite size, nm  (100)  (101)  a  c  d (100) c  MoP  0  MoP  31.93  42.89  0.3240  0.3196  16± 1  C00.07M0P  3.2  MoP  31.93  42.96  0.3240  0.3179  15± 1  71  Chapter 4 for these catalysts because the X R D diffractograms were a composite o f a number o f different phases (Figure 4.1) and the (111) peak o f N12P was not well resolved. C o (0.152 nm) and N i (0.149 nm) have very similar atomic radii so that even i f a solid solution were present for Co Ni2P with x x  < 0.16, the expected increase i n the lattice parameter c would be within the  measurement error o f the powdered x-ray diffraction ( P X R D ) data. Figure 4.2 shows the X R D diffractogram o f prepared M o P and C00.07M0P. The X R D pattern o f the C00.07M0P was very similar to that obtained for M o P and did not show characteristic C o P peaks, either because the C o concentration was too l o w (3.2 m o l % Co) and/or because C o was well dispersed i n the M o P . The P D F o f M o P ( P D F , 2005) is shown with the star and has diffraction peaks at 27.95°, 32.17°, 43.14°, 57.48°, 64.93° and 67.03° corresponding to the (001), (100), (101), (110), (111), and (200) crystal planes respectively. It is clear that both the M o P and C00.07M0P show only M o P X R D pattern. The data o f Table 4.1 show that the presence o f C o resulted i n a decrease i n the lattice parameter c o f M o P , suggesting that i n this case, a C o M o P solid solution was formed, as has been reported for the N i M o ( i . ) P system (Stinner et x  x  x  al., 2001; Zuzaniuk and Prins, 2003; M a et al., 2004). The C o content was estimated and Inductive Coupled Plasma (ICP) was used to confirm the composition (see Chapter 6). Addition o f C o to N i P increased the Ni2P crystallite size (Table 4.1), and the B E T 2  surface area (see Table 4.3) also increased. The bulk C o P , M o P or M2P catalysts had l o w B E T area (SBET) and large particle size (dc p 0  =  420 nm,  d>ji2P  = 206 nm and dM p 0  =  172 nm, estimated  from dp = 6/[SBETP] where p is the bulk density). The particle size was significantly greater than the crystallite dimensions (d ) shown i n Table 4.1, suggesting significant agglomeration o f the c  metal phosphide crystallites. The increased M2P and M o P crystallite size with addition o f C o to M2P and M o P was likely due to the increased time for complete crystallization compared to the  Ni2P, because o f the added Co(NC>3)2 solution. The increased B E T surface area o f the Co Ni2P x  72  Chapter 4 and C00.07M0P suggests a reduction in the degree o f agglomeration o f the larger crystallites when the C o was added.  Co  .JU I  J I.  007  MoP  A  A  A  MoP  JU I I  20  1  1 30  * P D F of MoP  A 1A  J L. 1  1  '  40  1 50  1  1  '  1  60  70  2©  Figure 4.2: X-ray diffractograms o f reduced M o P and C00.07M0P catalysts  4.1.2  T P R of precursors The conventional method used to prepare the metal phosphides involves reducing the  precursors under flowing hydrogen using T P R . Since the reaction conditions for transformation of the oxidic precursor to the phosphide are similar to T P R characterisationof the precursor, T P R profiles may provide some insight into the reactions taking place. Therefore it is necessary to characterize these T P R phenomena in order to determine the temperature reduction peaks as such 73  Chapter 4 information w i l l help explain i f there is for example metal-metal interaction i n the case o f  Co Ni2P and C00.07M0P. T P R data can also be used to estimate the degree o f reduction i f the x  reaction stoichiometry is known. The T P R o f the precursors are conveniently presented in Figures 4.3 and 4.4. M2P, and C o N i P are presented in Figure 4.3. T w o reduction peaks associated with the M2P precursor are x  2  attributed to l o w temperature reduction o f N i O to N i , followed by reduction o f the phosphate to M2P at higher temperature (Wang et al., 2002). The low reduction temperature obtained i n the present study (773 K ) was higher than that reported (500-700 K for the different N / P composition) by Wang et al. (2002). The reason for the difference could be because lower H2 flow rate was used i n the present study (60 ml/min) compared to the 1500 ml/min used by W a n g et al. (2002). A similar T P R profile was observed for C02P.  For CoxNi2P with x <  0.16,  increasing C o content increased the temperature o f the second maximum, most likely because o f the high reduction temperature o f the C o P precursor (1173 K ) (see Figure 4.3).  A s the C o  content increased further such that x > 0.34, both peak temperatures increased significantly, suggesting the formation o f different reduced species, consistent with the X R D data that showed the presence o f new phases (especially M12P5) i n the reduced C o N i P catalysts with x > 0.34. x  2  In Figure 4.4, the T P R o f the M o P precursor shows a peak at 955 K . In previous reports on the preparation o f M o P supported on SiCh, two reduction peaks at about 720 and 1090 K were reported (Zuzaniuk and Prins, 2003). The lower temperature peak was attributed to M o  74  6 +  Chapter 4  I  <  600  Figure 4.3  i  •  700  1  •  800  Temperature, K  1  900  1  1— 1000  T P R o f N i P and Co Ni2P catalyst precursors measured 2  x  in 10 % H in A r at a rate o f 60 ml(STP)/min at 1 K / m i n 2  75  Chapter 4  Figure 4.4 T P R o f C o P , C o P , M o P and C00.07M0P catalyst precursors 2  measured in 10% H in A r at a rate o f 60 ml(STP)/min 2  reduction to M o , with subsequent M o 4 +  4 +  and P  5 +  reduction occurring at higher temperature. In  the present work, the single broad peak at 955 K is assigned to the reduction o f M o  + 6  and P . 5 +  Addition o f 3 wt% C o to M o P resulted in a similar peak maximum temperature, although the T P R profile broadened compared to the M o P profile. T P R data were also used to estimate the degree o f reduction. However, the stoichiometry of the reduction reaction is difficult to determine since, as pointed out by Stinner et al. (2003), the catalyst precursors may include polyphosphate chains.  76  Furthermore, Oyama et al. (2002)  ' Chapter 4 have shown that under certain conditions, some phosphate species are volatile. Assuming the species present after calcination can be written as n M O . P 0 5 (Stinner et al., 2003), the degree o f x  2  reduction calculated i n Table 4.2 (see Appendix B for sample calculation) suggests that the precursors were not completely reduced to the metal phosphide. included species such as H P 0 4 " (x  3)  x  However, i f the precursors  or P , the actual degree o f reduction would be higher than 3 +  that calculated. Similarly, i f some unreduced, volatile phosphorous species leave the catalyst during the reduction or calcination process, the actual degree o f reduction would be higher than that calculated in Table 4.2.  Support for the latter case is provided by the mass loss  measurements made during T P R using the T E O M and reported in Table 4.2.  The measured  mass loss was greater than that calculated using the assumed stoichiometry and the measured H  2  consumption, indicative o f the presence o f other forms o f oxide and/or phosphate species that must be present and not accounted for in the reduction stoichiometry.  4.1.3  TEM Figure 4.5 compares T E M micrographs o f the N i P and Coo.o8Ni P bulk catalysts. 2  2  The  T E M images o f N i P and Coo.osNi P were similar and showed a mosaic crystal structure. Crystal 2  2  planes with d-spacings estimated at 5.0 A , 3.4 A and 2.8 A , corresponding to the (1 0 0), (0 0 1) and (0 11) planes o f N i P were identified. A multi-point S T E M - E D X analysis o f the Coo.osNi P 2  2  catalyst (nominal) yielded the composition Coo.16 ± 0.12 Ni .oo ± 0.12P1.08 + 0.0, 2  uniform catalyst composition.  indicative o f a  In the case o f the Coo.34Ni P, however, a wider variation in 2  composition was obtained (C00.15 ± 0.18 Ni .oo ± 0.24P0.11 ± 0.17)? consistent with the presence o f 2  different phases identified by X R D . The T E M estimates the sizes o f the N i P to be 5.0 A , 3.4 A 2  and 2.8 A , and the X R D indicates larger particles 21 ± 1 nm. The error associated i n the  77  Chapter 4  Table 4.2  Summary o f temperature programmed reduction data  Weight Loss  T P R Peak Catalyst  Temp.  Apparent Assumed Reduction Stoichiometry  Reduction  K 778, 840  Ni P 2  Coo.o Ni P 8  2  785,921  4NiO.P 0 + 9H 2  5  2  5  4NiO.P 0 + 9H  955  C0007M0P  1001  5  2  2  2CoP + 7 H 0  2  2  2Ni P + 9 H 0  2  2  3  2  5  from Reduction Degree  Measured using TEOM  mole %  wt%  wt%  72  24  31  74  25  51  62  25  66  25  46  18  2  2Mo0 .P 0 + H H ^ 2 M o P +11H 0 2  2  2CoO.P 0 + 7 H ^ 2 C o P + 7 H 0 2  1086  5  2  2  2Mo0 .P 0 + 11H -*2MOP + 11H 0 3  CoP  2Ni P + 9 H 0  2  2CoO.P 0 + 7H 2  MoP  Degree of  Calculated  2  5  2  2  2CoO.P 0 + 7H ^2CoP + 7H 0 2  5  2  2  78  24  Chapter 4  Figure 4.5 T E M micrographs o f bulk metal phosphides: (i) N i P and (ii) Co .o8Ni P. 2  0  2  Estimated d-  spacing o f lattice fringes shown are ( A ) 5.0 A (B) 3.4 A and (C) 2.8 A corresponding to (1 0 0), (0 0 1) and (0 1 1) planes o f N i P 2  79  Chapter 4 measurements o f the crystallite sizes was within 10%. Note that the T E M was focused on the small particles within the mosaic structure and since the mosaic structure had other irregular sizes obviously the larger particles are not estimated.  4.1.4  Other catalyst properties Properties o f the prepared catalysts are summarized i n Table 4.3, including the B E T  surface area, n - P A uptake and the C O uptake reported per unit B E T area o f the catalyst.  The  data show that after the initial addition o f the 3 wt% o f C o subsequent addition o f more C o to N i P leads to agglomeration o f crystals that reduces the surface area. Table 4.3 shows that the 2  C o P had the highest n - P A uptake among all the metal phosphides examined. The n - P A uptake on Co Ni2P with x < 0.16 was greater than on N i P . The n - P A uptake increased by about 50% x  2  on C00.07M0P compared to M o P . The C O uptake decreased for the C00.07M0P and the Coo.o8Ni P catalysts compared to 2  M o P and N i P , respectively. 2  However, for the C o N i P catalysts, the C O uptake increased as x  2  C o content o f the C o N i P catalysts increased. Excluding C o P , the catalysts with the highest nx  2  P A uptake, C00.07M0P and Coo.osNi P, also had the lowest C O uptake. 2  4.1.5  XPS The X P S spectra o f the P 2p region, the C o 2d region, the M o 3d region and the N i 2p  region obtained on the various catalysts are shown in Figure 4.6 and Figure 4.7. Note that these catalysts had not been re-reduced prior to the X P S analysis, as was done i n the case o f C O adsorption and n - P A adsorption.  The P 2p3/ binding energy ( B E ) associated with metal 2  phosphides has been reported to be 129.5 e V (Senzi and Jae, 1994; Colling and Thompson, 1994) whereas for P 0 the P 2 p 2  5  3/2  B E is 135.2-135.6 e V (Oyama, 2003; Jian and Prins, 1996),  80  Chapter 4  Table 4.3  Properties o f the prepared metal phosphides  Chemisorption Catalyst  B E T area  P/M*  C O uptake  n - P A uptake  Nominal  XPS  m /g  urnole/m  pmole/m  atom ratio  atom ratio  4.1  0.27  5  0.50  0.5  Coo.o8Ni P  7.9  0.11  18  0.48  4.8  Coo.i Ni P  7.8  0.22  7  0.46  1.8  Co .34Ni P  6.9  0.42  1  0.43  1.2  Co .79Ni P  6.5  0.45  1  0.36  -  MoP  5.3  0.21  8  1.00  1.0  C00.07M0P  11.4  0.19  13  0.94  1.3  CoP  2.3  0.42  37  1.00  2.5  Co P  5.6  0.16  11  0.5  1.6  2  Ni P 2  2  6  2  0  2  0  2  2  2  * M : total metals (Co+Ni) or (Co+Mo) and P: phosphorous  81  Chapter 4  Figure 4.6 (a)  Co 2 p C0 P 2  793.8 eV  —  CoP  - ~ - ~ ~ ~ K  ~~~~~ "  N  778.6 eV  _  v  ' I  Co  0.16  Ni P  Co  I  820  2  Ni 0.08 1  -—  P 2  1 810  J \ r ~ ^ ~ ~ ^ ^ ~ ~ ~ ^  '  1 800  '  1 790  1  1 780  '  1 770  •  1  760  B.E., eV  Figure 4.6 (b)  145  140  135  B E . , eV  82  130  125  Chapter 4  853.7eV  N i 2p  852.7eV  Co ., Ni P0  Co  4  2  Ni P 2  0.16  Co„ N L P NLP  870  860  850  840  B.E., eV  Figure 4.6 (c)  Figure 4.6  X P S o f (a) C o 2p region (b) P 2p region, and (c) N i 2p region o f Co Ni2P catalysts after reduction and passivation x  83  Chapter 4 and 133.3 e V for N i ( P 0 ) (Stinner et al., 2001). Hence we assign the l o w B E peak at 129.53  4  2  129.8 e V , present in the C o P , C o P , N i P and M o P spectra to the metal phosphide, and the higher 2  2  B E peak at 133.4-133.8 e V to surface metal phosphate species. For N i P , this assignment is 2  consistent with the N i 2p B E s that show the presence o f N i P and N i ( P 0 4 ) with corresponding 2  3  2  N i 2p / B E o f 853.7 e V and 857.3 eV, respectively. A similar conclusion can be drawn for the 3  2  C o P and C o P catalyst with C o 2p B E at 780.0 - 778.6 eV and 793.8 e V and the M o P catalyst 2  with M o 3d B E at 228.2 e V and 232.2 e V . The effect o f increasing C o content on the X P S spectra o f the C o N i P catalysts is also x  2  shown in Figures 4.6 and 4.7. N o significant shift in B E ' s was observed in the case o f P 2p / , 3  2  whereas in the case o f N i 2p, a feature at lower B E (852.7 e V ) assigned to N i - r i c h phosphide ( N i i P s ) ( L i et al., 2005), increased as the C o content increased. This observation is consistent 2  with the presence o f N i i P s identified by X R D in the C o N i P catalysts with x > 0.34. 2  x  2  In  addition, as C o content increased, the peak assigned to phosphate species became less significant and the peak at 855.9 e V , assigned to N i , increased. These results are consistent with Sawhill et al. (2005) who reported that N i ^ P s is more readily oxidised during passivation than N i P . The 2  feature at ~850 e V is likely the L 3 V V Auger line o f N i O (Khwaja et al., 1989). Comparison o f the C o 2p peaks present i n the C o P and C o N i P (at B E = 780 e V ) suggest the presence o f metal x  2  phosphide although there is a shift to lower B E o f about 1.5 e V i n the case o f C o P , indicative o f 2  a more metallic C o P . In the case o f 2  C00.07M0P, the M o 3ds/ B E and the P 2p / B E are almost 2  3  2  identical to that observed in the case o f M o P . The X P S analyses were also used to calculate the P / M atom ratios o f the catalysts prior to their use. These data (Table 4.3) suggest that the surface o f the M o P catalyst was slightly enriched in P, and addition o f C o resulted in a small increase in the surface P content. In the case o f N i P , the surface P / N i ratio was close to 1/2, but following C o addition, a surface enrichment 2  84  Chapter 4  Mo 3d  232.2 e V  133.8 e V 129.8 e V  228.2 e V  MoP  ,MoP 245  215  145  —I—  140  135  130  125  B.E., e V  Figure 4.7  X P S o f P 2p region and M o 3d region of M o P and C00.07M0P catalysts after reduction and passivation  85  120  115  Chapter 4 in P occurred that was most significant for the C00.08N12P catalyst (P/(Co+Ni) = 4.8). Figure 4.8 shows a correlation between the X P S and chemisorption data obtained on the metal phosphides o f the present study. Assuming that the C O chemisorption titrates metal sites and the n - P A uptake titrates Bransted acid sites, presumably associated with the surface phosphate species, the P / M atom ratio as determined by X P S should correlate with the n - P A to C O uptake ratio as determined by adsorption. The data plotted in Figure 4.8 confirm a linear correlation ( R = 2  0.88), with some deviation associated with experimental error o f the measured ratios that is indicated on the graph. The standard error o f the correlation parameters is ±0.005 for the slope and ± 0.364 for the intercept.  4.2  Catalyst Activity Prior to performing activity measurements, the passivated catalysts were re-activated in a  H2 flow at 723 K for 1 h. Catalyst conversion data were then measured and the conversions typically reached a steady value after about 8 hours time-on-stream.  The conversion and  selectivity data reported i n Table 4.4 and Figure 4.9 are the time-averaged values over a period of at least 4 hours that followed the initial 8 hour stabilization period. In each case, the complete product analysis was used to calculate product selectivities. A 4 , 6 - D M B T m o l balance between the reactor feed and reactor effluent was > 95% for all the data reported herein (see Appendix C ) . The reproducibility o f the experiments is also found in Appendix C . The metal phosphides had conversions i n the range 10 - 49 ± 6 m o l % , (see Appendix C for the standard error calculations) and by assuming that the conversion o f 4 , 6 - D M D B T on these catalysts was first order at the conditions o f the present study, the rate o f 4 , 6 - D M D B T  86  Chapter 4  n - P A / C O u p t a k e ratio by a d s o r p t i o n  Figure 4.8 Correlation o f P / M ratio determined by X P S and n - P A : C O uptake ratio determined by adsorption for C o N i P (•), x  2  C00.07M0P  ( A ) , C o P (o), C o P 2  (0)  with M o P ( A ) and  N i P ( D ) as indicated. The solid line represents the correlation equation: 2  P / M ratio = (0.024±0.005)*(n-PA/CO uptake ratio)+(0.411±0.364) and the dashed lines represent the 95% confidence limits o f the correlation  87  Chapter 4 consumption for each catalyst was calculated (Table 4.4) (see Appendix B . l l for detailed calculations). The specific rate (i.e. rate per gram o f catalyst) increased on the C o o . o s ^ P catalyst, but declined on the Co Ni2P catalysts with x > 0.16, compared to M2P. However, the x  data o f Table 4.4 show that per unit area ( B E T ) , the activities o f all the Co Ni2P catalysts were x  lower than the M2P, C02P and C o P catalysts, whereas the T O F based on C O uptake, showed a maximum value for the Coo.o8Ni P catalyst. The T O F is defined as the number o f molecules 2  reacting per active site per unit. Similar behavior was observed on the M o P versus the C00.07M0P catalyst, except that the T O F was greater on the M o P catalyst than the C00.07M0P catalyst. The major products o f 4 , 6 - D M D B T conversion, determined by G C - M S , were dimethylbicyclohexane ( D M B C H ) , dimethylbiphenyl ( D M B P ) and methylcyclo-hexyltoluene ( M C H T ) with  less  significant  quantities  hydrocracking and hydrogenation.  o f hydrocarbons  that  were  products  of  4,6-DMDBT  The selectivity o f the catalysts to M C H T , D M B C H and  D M B P is reported i n Table 4.4, with the remaining products being hydrogenated S containing products or cracked products.  The data show that with 4 , 6 - D M D B T , hydrogenation products  were favored over Ni2P, M o P and C o P , and this contrasts with the high selectivity to D D S products that occurs for D B T conversion over N i P and M o P (Sun et al., 2004; Oyama et al., 2  2002). The l o w selectivity to D D S products in the case o f 4 , 6 - D M D B T is known to be due to steric hindrance by the methyl groups o f the molecule. O f particular significance i n the present work, however, is the increase i n selectivity toward D M B P that occurred when C o was added to both the M o P and Ni2P catalysts, especially for l o w C o contents.  A n increase i n D M B P  selectivity suggests a significant shift toward direct desulfiirization ( D D S ) o f 4 , 6 - D M D B T rather than aromatic ring hydrogenation.  Isomerization o f the 4 , 6 - D M D B T molecule likely occurs  88  Chapter 4 Table 4.4 Activities o f bulk metal phosphides for the H D S o f 4 , 6 - D M D B T measured at 583 K and 3.0 M P a H  4,6 D M D B T Consumption Rate  2  Selectivity  Conversion  HDS  mol %  mol%  29.1  25.6  2.3  5.5  7.3  2.1  25.7  0.9  Coo.ogNi P  48.7  47.2  4.4  5.5  18.1  0.8  13.3  57.3  Coo.i Ni P  23.4  20.2  1.8  2.2  3.7  14.6  10.5  40.9  Coo.3 Ni P  17.4  16.8  1.3  1.8  1.6  28.6  4.0  22.2  Co .79Ni P  10.1  8.6  0.7  1.1  0.9  25.1  5.7  16.3  MoP  36.2  35.2  2.9  5.6  9.5  32.1  1.4  1.2  C00.07M0P  42.5  40.9  3.6  3.2  6.0  10.7  1.3  46.9  CoP  22.3  20.6  1.7  7.2  6.2  17.8  12.0  2.3  Co P  43.9  41.6  3.8  6.7  1.5  15.7  21.6  33.3  Catalyst  Ni P 2  2  6  4  0  2  2  2  2  Specific  10 mol.g-'.s" 9  TOF  Areal  1  10 mol.m .s" 10  89  _2  1  MCHT  10V  DMBCH  DMBP  mol %  Chapter 4 to yield, for example 2 , 8 - D M D B T , that undergoes rapid H D S (Kwak et al., 1999) by D D S . The distribution o f sulfur containing products (sulfur speciation) obtained with N12P and Coo.o8Ni2P catalysts is compared i n Table 4.5. N o 4 , 6 - D M D B T isomerization products were detected, presumably because they underwent rapid D D S on the metal sites o f the Coo08N12P catalyst. Evidence for increased acid catalyzed methyl cracking on the Coo.o8Ni2P catalyst versus the M2P catalyst is apparent from the data o f Table 4.5, whereas the N i P yielded more 2  hydrogenated product than the C00.08N12P catalyst.  These trends are consistent with the  promotion o f isomerization o f 4 , 6 - D M D B T followed by rapid D D S on the more acidic Co Ni2P x  catalysts, versus less isomerization, more hydrogenation and less D D S on the less acidic M2P catalyst. The 4 , 6 - D M D B T conversion and D M B P selectivity data are plotted as a function o f the C O uptake i n Figure 4.9. For the Co Ni2P series o f catalysts, conversion o f 4 , 6 - D M B T decreased with increasing x  C O uptake. O f more significance however, is the trend in D M B P selectivity. Figure 4.9 shows that as the C O uptake increased and n - P A uptake decreased, the D M B P selectivity decreased. For the Co Ni2P catalysts studied herein, maximum selectivity to D M B P occurred on the catalyst x  with the highest n - P A uptake and the lowest C O uptake, i.e. the Coo.08N12P catalyst. The data also clearly show that the metal phosphides (CoP, M o P and N12P) have very different selectivity behavior compared to the CoxNi2P and C00.07M0P catalysts.  90  Chapter 4  Table 4.5 Comparison o f sulphur speciation i n liquid product from H D S o f 4 , 6 - D M D B T measured at 583 K and 3.0 M P a H over N i P and Coo.osNi P 2  2  2  Catalyst Ni P 2  Product  CrQ  Comment  Two methyl groups removed  Co .o8Ni P 0  2  mol%  3  32  One methyl group removed  24  10  Partial hydrogenation o f one ring  29  18  44  40  Complete hydrogenation o f one ring  91  Chapter 4  C O uptake, umole/m  2  Figure: 4.9  Conversion o f 4 , 6 - D M D B T and selectivity to D M B P over various metal phosphide catalysts at 583 K and 3.0 M P a Ff , plotted as a function o f C O 2  uptake: C o N i P (•), C00.07M0P ( A ) , C o P (0) and C o P (o) with M o P ( A ) x  2  2  and N i P ( p ) as indicated 2  92  Chapter 4 4.3  Discussion on bulk metal phosphides used for HDS of 4,6-DMDBT Previous studies o f H D S on metal sulfides and phosphides have shown lower D D S  selectivity with 4 , 6 - D M D B T as reactant, compared to D B T as reactant.  Sun et al. (2004)  reported that on supported N i M o P catalysts over a range o f N i / M o ratios, the selectivity to biphenyl was > 80% with D B T as reactant.  O n N i P / S i 0 and C o P / S i 0 , Wang et al. (2002) 2  2  2  reported 100% selectivity to biphenyl with the same reactant, and with N i P x  y  prepared over a  wide range o f N i / P ratios, 100% selectivity to biphenyl during the H D S o f D B T was also obtained (Oyama et al., 2002).  However, with 4 , 6 - D M D B T as reactant, the hydrogenation  activity o f N i P was 5 to 7 times greater than the hydrogenolysis ( D D S ) activity (Oyama et al., 2  2002). Similar results were obtained i n the present study, where selectivities to the D D S product D M B P were < 3% on N i P , C o P and M o P . The low selectivity to D D S is due to steric hindrance 2  effects associated with the methyl groups o f 4 , 6 - D M D B T (Kwak et al., 1999; Isoda et al., 2000). A n enhanced D D S o f 4 , 6 - D M D B T was reported with P Os addition to conventional metal 2  sulfide catalysts (Kwak et al., 1999) and these authors attributed this effect to an increase i n Bronsted acidity associated with the added phosphorous that increased migration o f the methyl substituents on the aromatic rings o f 4 , 6 - D M D B T . The isomerization would not be important i n the H D S o f D B T . The products o f skeletal isomerization o f 4 , 6 - D M D B T have been shown to be much more reactive to D D S than 4 , 6 - D M D B T , especially 2 , 8 - D M D B T (Isoda et a l , 2000). A n increase i n the D D S rate o f 4 , 6 - D M D B T has also been reported for P addition to M o C 2  hydroprocessing catalysts (Manoli et al., 2004). The correlation between the surface phosphorous/metal ( P / M ) ratio determined by X P S , and the ratio o f n - P A uptake to C O uptake determined by adsorption (Figure 4.8) suggests a catalyst surface with Bronsted acid sites associated with phosphate species such as H P 0 4 " , and metal (x  3)  x  sites that chemisorb C O . The source o f the Bronsted acid sites is likely a consequence o f the  93  Chapter 4 incomplete reduction o f phosphate species, which is accentuated by the presence o f C o and the catalyst passivation procedure.  Following reduction o f the catalyst precursor by T P R , the  catalysts were passivated in diluted 0 , according to procedures used by others that are known to 2  yield a surface oxidised over-layer (Sawhill et al., 2005). The P 2p spectra o f the passivated catalysts showed the presence o f metal phosphide and metal phosphate species for all the catalysts. H P 0 x  ( x 3 ) 4  species, with B E 134.3 - 135.2 e V (Fluck et al., 1974) were not resolved in  the X P S spectra, but low concentrations o f these species cannot be excluded. The N i 2p spectra o f the N i P and Coo.o8Ni P catalysts showed two peaks that were assigned to N i 3 ( P 0 ) and N i P . 2  2  4  2  2  A s the C o content increased a peak assigned to N i ^ P s and oxidized N i became evident. Hence it is clear that although the X R D data only showed the presence o f bulk metal phosphides, after passivation, both metal phosphide and metal phosphate species were present on the catalyst surface, and i n some cases metal oxide was observed as well.  Prior to determining the C O  uptake, the n - P A uptake, or the activity, the catalysts were pretreated in H by heating to 723 K . 2  However, the T P R data show that especially i n the case o f the cobalt phosphide precursor, a higher temperature is required for complete reduction to the metal phosphide and consequently, this pre-treatment likely only removed the most reactive surface oxygen. The product water could react with phosphorous species yielding H P 0 x  ( x 4  "  3 )  and the observed Bransted acidity. The  chemisorption data show that C o P , which had the highest precursor reduction temperature, also had the highest n - P A uptake and we assume that this is a consequence o f incomplete reduction and more surface phosphate species following passivation. O n supported catalysts the acidity due to the P may be less important because o f reaction between P and the metal oxide support during calcination and reduction (Clark and Oyama, 2003). The data o f Figure 4.9 suggest that for all the catalysts, the n - P A uptake decreased as the C O uptake increased, except for C o P which had a very high n - P A uptake that corresponded to a  94  Chapter 4 high reduction temperature o f the precursor. The 4 , 6 - D M D B T conversion also decreased as C O uptake increased. The catalyst that gave the highest 4 , 6 - D M D B T conversion and highest D M B P selectivity, Coo.osNi P, also had the lowest C O uptake and highest n - P A uptake among the 2  C0007M0P and Co Ni2P catalysts. These results demonstrate the importance o f the surface x  acidity and hence isomerization o f 4 , 6 - D M D B T for enhanced S removal by D D S . The data suggest that high selectivity for D D S o f 4 , 6 - D M D B T on C o N i P and C o M o P catalysts is x  2  x  obtained with increasing acid site concentration and decreasing metal site concentration both o f which are determined by the amount o f C o added. Note, however, that the C o P , C o P , N i P and 2  2  M o P do not follow the same correlation between D M B P selectivity and C O uptake shown for the C o N i P and C00.07M0P catalysts (Figure 4.9). The C o N i P catalysts must therefore provide x  2  x  2  unique acidic and metallic active sites that benefit the D D S route. The C o N i P catalysts had x  2  larger N i P crystal size and B E T surface area than the N i P catalyst. Although we do not see 2  2  direct evidence o f a C o N i P solid solution, the possibility o f a well-dispersed C o N i P that is x  2  x  2  more active for D D S o f 4 , 6 - D M D B T cannot be excluded. Although C o P may also be present, it 2  seems less likely that increased hydrogenolysis is associated with increased metallic character o f the catalysts.  The data showed a decrease i n D M B P selectivity as more C o was added,  corresponding to an increase i n C O uptake and X R D patterns that showed more metallic phases (Nii P5 and C o P ) present i n the bulk. The increase i n metallic character o f the N i i n C o N i P as 2  2  x  2  C o was added may also be indicated by the increase in the N i 2p peak at l o w B E o f 852.7eV (Klein and Hercules, 1983). However, as this peak intensity increased with C o content, the D M B P selectivity decreased.  Although C o P had the highest D M B P selectivity among the 2  single metal phosphides, the T P R data showed this catalyst was more readily reduced than C o P and had a lower acid site density. The X P S data also show that the B E o f C o associated with the  95  Chapter 4 C o P and Co Ni2P catalysts (Figure 4.6 (a)) was shifted about 1.5eV to higher B E (less metallic) x  compared to the C o associated with C02P (778.6 e V ) . The significant drop i n selectivity observed as more C o is added to the Co Ni2P catalysts x  suggests that the active phase only has.a high dispersion with x < 0.16. Presumably the Co Ni2P x  catalysts generate a unique phase, probably a solid solution ( M a et al., 2004), dispersed on the large Ni2P crystals, that result in a better distribution o f acidic and metallic sites and/or provides a catalyst morphology that improves selectivity to D M B P (Wang et al., 2002). Table 4.1 shows that with addition o f C o to M2P, the crystallite size also increases but the conversion as shown i n Table 4.4 initially increases but deceases as more C o is added. In the case o f adding C o to M o P , Table 4.1 shows that the crystallite size is almost invariant but the conversion increased. The particle size does not affect the structure-activity because the activity data reported on Table 4.4 is from a complex system. The Bronsted acidity and the presence o f C o both contribute to the conversion and selectivity data. The P / M ratio as determined by X P S , showed significant deviations from the ratio expected based on the catalyst synthesis conditions (Table 4.2). A P enriched surface for all the C o N i P catalysts compared to M2P was observed. x  2  Furthermore, the change i n P / M surface  composition as determined by X P S , was consistent with the C O chemisorption data and the n - P A T P D data. A s shown i n Figure 4.8, the P / M atom ratio was well correlated with the n - P A to C O uptake ratio.  If one assumes that the Bronsted acidity resides in  H P04 X  ( X  "  3 )  species, and all the  surface P is i n this form with x = 3, then the slope o f the line shown i n Figure 4.8 would be expected to be 1/3, assuming an adsorption stoichiometry o f 1:1 for n - P A : Bronsted acid site and C O uptake: metal site. However, the slope o f the line in Figure 4.8 is ten times smaller than this, suggesting a much lower C O uptake than available surface metal sites as determined by X P S . (Note that the X P S data shows that some o f the surface P is present as metal phosphide, and this  96  Chapter 4 would tend to increase the slope o f the line o f Figure 4.8). The limitation o f C O adsorption to titrate the metal sites on metal phosphides has been discussed in the literature previously, (Sun et al., 2004; Fluck et al., 1974) and the present study supports the conclusion that the C O adsorption underestimates the number o f metal sites in these metal phosphide catalysts. However, reproducibility data using C O chemisorption (see Appendix B ) demonstrate reliability of this characterization technique for titrating the metal sites in metal phosphides. In conclusion, the addition o f ca. 3 m o l % C o to N i P and M o P resulted in a significant 2  increase in D D S selectivity during 4 , 6 - D M D B T conversion.  The change in selectivity  corresponds to a number o f observed changes in the catalyst properties, especially surface Bransted acidity and metal sites as determined by n - P A adsorption and C O adsorption, respectively. The most selective catalyst for D D S o f 4 , 6 - D M D B T was the C00.08N12P catalyst and this catalyst had the lowest C O uptake and highest n - P A uptake among the Co Ni2P and x  C00.07M0P catalysts studied. Consequently, the C00.08N12P bulk catalyst was selected from the rest as a high performing H D S catalyst. The selected catalyst was subsequently supported on different supports.  4.4  Properties of supported metal phosphide catalysts and the hydrodesulfurisation of 4,6dimethyldibenzothiophene The Coo.o8Ni2P catalyst had the highest selectivity for direct desulfurization o f 4,6-  D M D B T among a series o f Co Ni2P (0 < x < 0.34) catalysts and the enhanced selectivity was x  attributed to the presence o f Bransted acid sites that isomerised 4 , 6 - D M D B T and reduced the steric hindrance associated with the S atom.  Although metal phosphides show promise as  alternative hydrotreating catalysts, the assessment o f these catalyst using typical refinery feedstocks is rare. Although the effect o f P added to metal sulfides, carbides and nitrides has  97  Chapter 4 been reported (Sundaramurthy et al. 2006) no data are available on the performance o f phosphide catalysts i n a real feed that contains both S and N compounds. Furthermore, studies o f the phosphides with model compounds suggest that they undergo some sulphidation during reaction, yet the extent o f conversion o f the phosphides to sulphides or other compounds in the presence of a real feedstock needs to be clarified. The activity o f Coo.4Ni2P catalysts, supported on supports o f different acidities w i l l be presented.  The catalyst activity using the model compound 4 , 6 - D M D B T is presented in this  section and i n chapter 6, the activity o f the selected catalysts w i l l be presented using light gas o i l ( L G O ) derived from Athabasca bitumen. The properties o f the catalysts after reaction with the L G O are also reported i n an attempt to quantify the stability o f the metal phosphide catalysts when hydrotreating a commercial feedstock. In addition to anchoring the active phase o f the catalyst and providing increased surface area, supports are known to interact with the active phase thereby influencing the activity. AI2O3 is reported to be an inferior support for metal phosphides because it reacts with phosphate to form aluminum phosphate and so Si02 has been used to support metal phosphides because they have weak interaction with Si02 (Robinson et al., 1996, Mangnus et al., 1990). However, different reports lend support for the use o f AI2O3 as a superior support. For example, Ledoux et al. (1995) reported that Y-AI2O3 was the most active support when they studied the role o f the crystallinity o f alumina support on the H D S activity o f sulfided C0M0 catalysts. The authors also reported that based on the T O F , alumina is better than the other supports. Another important role o f the support is in the modification o f acid properties o f the catalysts. Mauge et al. (2001) used F T I R to study the surface properties o f M0S2 catalyst and showed that the unsupported M0S2 catalyst had weaker acidic properties as well as a higher metallic character o f M o sites compared to the M0S2 supported Y-AI2O3 catalyst. Y-AI2O3 (Aldrich-155 m / g , 0.76 2  98  Chapter 4 ml/g), used commercially in hydroprocessing, was chosen as a support i n the present study. In addition, supports with different acid properties were also investigated. A s explained earlier i n Chapter 2, Bransted acidity is required for isomerisation o f the methyl groups on the two benzene rings at the 4,6- positions o f D M D B T so as to remove the steric hindrance and allow the S atom to gain access to the surface o f the catalyst to react. Hence, supports with Bransted acidity such as weakly acidic alumina, fluorinated alumina and M C M were employed to study the effects o f acidity on the isomerization o f the 4 , 6 - D M D B T . Supported catalysts were prepared using a similar procedure to that used i n the synthesis of the bulk catalysts (see Appendix B l for detailed calculations for preparing metal supported catalysts). Recent studies by L i et al. (2005) reported that the phosphide phase is obtainable i f the T P R is completed at high temperature and low heating rate (1 K/min). Therefore in preparing the supported samples, the reduction was carried out to 1200 K at a heating rate o f 1 K min" . 1  4.4.1  X R D of modified AI2O3 supported phosphides Figure 4.10 shows the X R D diffractograms measured for Y - A 1 0 , Coo.4Ni2P/Al203, 2  3  Coo.4Ni P/Ai203-F, and C o o . N i P / M C M . A l s o included is the P D F o f N i P ( J C P D S , 2005). 2  Clearly the  4  2  2  NI2P phase is present on the AI2O3 support. The diffraction peaks o f M2P appeared  on all the prepared catalysts at 40.8°, 44.69, 47.3°, 54.2° and 54.9° corresponding to the (111), (201), (210), (300) and (211) crystal planes, respectively. The formation o f N12P on the M C M is indicated on Figure 4.10 and appears to be weakly crystallized as compared to when supported on the AI2O3. The Coo.4Ni2P is well dispersed on the M C M support and therefore diffraction peaks are not very well resolved. A g a i n consistent with that o f the bulk Coo.4Ni2P, the C o phase is not present on the diffractogram as C o was either in small quantities or it was well dispersed on the surface o f the catalyst. Since M C M is a mesoporous material, the X R D pattern can only  99  Chapter 4  Co  n  NLP/MCM 2  0.4  t  '  20  25  1  30  '  35  1  40  1  45  1  '  1  '  50  55  60  65  1  70  2 0 , degrees  Figure 4.10 X R D o f prepared metal phosphides on different acidic supports  be resolved at the low angle region which cannot be obtained using the available X R D . In the literature, low angle measurements o f M C M show peak intensities at 2.3°, 3.6°, 4° corresponding to crystal planes (100), (110) and (200) respectively ( L i et al., 2003). The X R D diffractograms have provided conclusive evidence for the presence o f the Coo 4M2P on the support. The appearance o f the active phase is attributed to the high reduction temperature employed and the slow heating rate, as observed earlier L i et al., (2005).  100  Chapter 4 4.4.2  X R D of P t - C o N i P / A l 0 and P t - C o N i P / A l 0 - W A 2  Figure  4.11  2  shows  3  2  the  2  3  X R D diffractrograms  of  Pt-Co Ni P/Al O 0 4  Co .4Ni P/Al2O3-WA. The X R D patterns indicate that both y - A l 0 0  2  2  2  2  3  and A 1 0 - W A  3  2  3  and  supports  produce similar catalysts. The diffraction peaks corresponding to P t P were obtained at 27.12° 2  Pt-Co Ni P/Al O 0 4  2  2  +  3  A 3 ed  Pt-Co Ni P/Al O  fi  0 4  2  2  3  j  20  30  A  '  +  40  50  60  70  2 0 , degrees + Formation of Ni P * Formation of PtP 2  Figure 4.11  X R D diffractrograms o f P t - C o N i P / A l O and 0 4  Pt-Co . Ni P/Al O 0  4  2  2  3  -WA  101  2  2  3  Pt-  Chapter 4 and 31.48°. Pt° has XPvD peaks at 39.6 °, 46.1° and 67.3° ( J C P D S , 2005). The diffraction peaks at 40.8 °, 44.7 °, 54.2 °, and 54.9  0  were attributed to the  C00.4M2P as observed earlier i n section  4.4.1 and the diffraction peaks at 45.92° and 66.64° attributed to the Y - A I 2 O 3 support.  4.4.3  Properties of Coo.4Ni2P prepared on different supports The properties of the supports used are shown in Table 4.6A.  C00.4N12P was supported on  neutral AI2O3, fluorinated AI2O3, and M C M (see Appendix B for a diagram o f the M C M crystal). After preparing these supported catalysts, they were characterized using B E T , C O uptake and nP A chemisorption and the properties are presented in Table 4.6B. The pore size dp, was calculated assuming cylindrical pore and using the relationship dp = 4 x pore volume / B E T surface area. The range o f the pore size is 8.3 n m (83 A ) to 19.6 nm (196 A ) . The size o f the 4,6D M D B T molecule is about 8 A which is much larger than the nitrogen molecule but much smaller than the pore sizes o f the supports. Therefore i f nitrogen can access the surface area, then the reactant molecules should be able to gain access to the interior o f the pores since that is where most o f the B E T surface area is located. Hence, it is valid to normalize the reaction rates with the surface areas. Column 2 o f Table 4.6B shows that the B E T surface areas o f all the prepared supported metal phosphides range from 95-125 m /g. A s expected, the high surface area M C M exhibited the highest surface area, 125 m /g upon loading with  C00.4N12P.  The C O uptake for the three investigated catalysts Coo.4Ni2P/Al 0 , Coo.4Ni2P/Al 03-F 2  3  2  and Coo.4Ni2P/MCM were 25 umol/g, 24 umol/g and 20 umol/g respectively indicating that the metal dispersion on the surface o f these catalysts was the same. The n - P A titrated the Bransted acid sites on the surface o f the catalysts and the results in column 4 o f Table 4.6B show that the catalyst supported on M C M produced the highest Bransted 102  Chapter 4  Table 4.6 A Support  Properties o f the supports SBET  Pore size  Pore volume  n - P A uptake  m /g 2  dp, nm  155  19.6  0.76  158  A1 0 -F  146  18.6  0.68  244  AI2O3-WA  164  18.0  0.74  180  MCM  455  8.3  0.95  557  A1 0 2  2  3  3  ml/g  pmol/g  Table 4.6B Surface areas, C O uptake and n - P A uptakes o f prepared Coo.4Ni P on 2  different supports  Chemisorption  Catalyst  B E T area  C O uptake  n - P A uptake  m /g  pmol/g  pmol/g  C00.4M2P/AI2O3  98  25  304 (3.1)  C00.4N12P/AI2O3-F  95  24  390 (4.1)  Co . Ni P/MCM.  125  20  715 (5.7)  2  0  4  2  MCM: Mobil Catalytic Materials support A1 0 -F: Fluorinated alumina support Values in brackets are n-PA uptakes normalized by the BET area (pmol/m ) 2  3  103  Chapter 4 acidity o f 715 umol/g compared to the catalysts supported on A I 2 O 3 (304 umol/g) and AI2O3-F (390 umol/g).  4.4.4  X  Properties of Pt-Coo.4Ni P on different supports 2  Pt was added to Coo.4Ni2P/Al2C»3 to study the effect o f this metal on the hydrogenation route o f the 4 , 6 - D M D B T (see Figure 2.6). A weakly acidic support was used to prepare PtCoo.4Ni2P/Al20 -WA catalysts in order to study the role o f a weakly acidic support compared to 3  the AI2O3 neutral support. The characterization data for these prepared catalysts are presented in Table 4.7. The B E T surface area o f the Pt-Coo.4Ni2P/Al 0 catalyst was 89 m / g and the Pt2  2  Coo.4Ni2P/Al 03-WA was 95 m / g . Compared to the 2  2  3  C00.4N12P/AI2O3 catalyst, B E T surface area  98 m / g (Table 4.6B) the addition o f Pt showed a modest change i n surface area and could be 2  attributed to the addition o f the 5 wt% Pt that led to agglomeration o f particles and blocking pores and therefore decreased the surface area. The platinated supported metals, Pt-Coo.4Ni P/Al 03 and P t - C o o . 4 N i P / A l 2 0 - W A both 2  2  2  3  exhibited high C O uptake (31 and 30 umol/g respectively). The presence o f Pt probably caused the C O uptake i n these catalysts to increase more than the non-platinated samples (see Table 4.7). Pt could also cover the P on the Coo.4Ni2P thereby increasing the C O uptake since the P does not chemisorb C O .  104  Chapter 4  Table 4.7  Surface areas, C O uptake and n - P A uptakes o f prepared Pt-Coo.4Ni2P on different supports  Chemisorption  Catalyst  B E T area  Pt-Coo.4Ni P/Al 0 2  2  3  Pt-Coo.4Ni P/Al 0 -WA 2  2  3  C O uptake  n - P A uptake  m /g  pmol/g  pmol/g  89  31  240(2.7)  95  30  380(4.0)  A1 0 -WA: Weakly acidic alumina support Pt: Platinum Values in brackets are n-PA uptakes normalized by the BET area (pmol/m ) 2  4.5  3  Supported catalyst activity using 4,6-dimethyldibenzothiophene Catalyst activity measurements over the supported metal phosphides were studied using  the same fixed bed reactor operated at the same hydroprocessing temperature and pressure as for the bulk catalyst test. Prior to performing the measurements, catalysts were pretreated in the same way as the bulk catalysts by passing H at 723 K for 1 h in order to remove the passivated 2  layer. In the present study, 4 , 6 - D M D B T was added to dodecane solvent to provide 3000 ppm o f sulfur feed. The H D S products o f 4 , 6 - D M D B T using these supported metal phosphides were similar to those obtained previously on the bulk metal phosphides. The D M B C H and M C H T are products o f the hydrogenation route and D M B P is the product o f the D D S as shown earlier i n the 4 , 6 - D M D B T reaction scheme (Figure 2.6). These products are in agreement with recent report  105  Chapter 4 by Mizutani et al. (2005) who studied the activity o f 4 , 6 - D M D B T over C0M0P/AI2O3 phosphide catalyst. Table 4.8 shows the activity data of the hydrodesulfurization o f 4 , 6 - D M D B T using all the prepared supported metal phosphides. The conversion ranged from 81.3-99.8 ± 0.06 m o l % (see Appendix C for standard error analysis) with the lowest conversion obtained on the N12P/AI2O3 and the highest on the C o o . N i 2 P / M C M . 4  Mizutani et al. (2005) using a fixed bed reactor operated at 603 K , 3 M P a and W H S V o f 13.5 h ' obtained high conversion o f 93 m o l % over C0M0P/AI2O3. The conversion obtained i n 1  the present study for the. C00.4N12P/AI2O3 was 86 m o l % and this is therefore comparable considering the fact that lower temperature (583 K ) was used i n the present study. Kabe et al. (1999) have reported the use o f first order kinetics for the H D S o f heterocyclic sulfur compounds hence first order kinetics w i l l be used. The specific consumption o f the 4,6 D M D B T obtained varied from 1.10 - 4.07 x 10" mol.g' . s" ( 3.95 - 14.6 mol.g" . h" ). The highest consumption 4  1  1  1  1  was obtained using Coo.4Ni2P/MCM supported catalyst and the least on the Ni2P/Al203. Nagai et al. (2005) reported a specific rate o f 0.19 m o l . g" . h" (5.28 x 10 -5 m o l . g" . s" ) o f 1  dibenzothiophene over N i M o P / A l 0 2  1  1  1  at 573 K , 2 M P a , 8 h" W H S V and 12 h time on stream. 1  3  The product distribution was similar for all the catalysts supported on AI2O3 except those supported on M C M and the selectivity of products was obtained as an average after 4 h o f the 8 h stabilization time. The product selectivity data o f Table 4.8 show that the Ni2P/Al C»3 produced 2  more products by the hydrogenation route (89% o f both D M B C H and M C H T ) and fewer products by the D D S route (2.8 m o l % o f the D M B P ) . O n the other hand, the C o o . 4 N i P / A l 0 2  2  3  showed a l o w selectivity for the hydrogenation route (30 m o l % o f D M B C H ) compared to the higher selectivity to D M B P (52%). This observation is consistent with the observation on the bulk Ni2P and Coo.4Ni2P catalysts where it was earlier reported that the Coo.4Ni2P enhanced the  106  Chapter 4  Table 4.8 Activities o f supported metal phosphides for the H D S o f 4 , 6 - D M D B T measured at 583 K and 3.0 M P a H  2  Selectivityt Catalyst  Conversion  HDS  Specific Consumption  MCHT  10 mol.  DMBCH  DMBP  8  mol%  mol%  81.3  80.1  1.10  36.1  52.8  2.8  Coo.4Ni P/Al 03  86.5  85.8  1.31  2.1  28.0  52.4  Coo.4Ni P/Al 0 -F  87.1  86.4  1.36  5.9  16.6  45.2  Coo. Ni P/MCM  99.8  98.8  4.07  4.0  12.7  1.9  97.2  97.1  2.41  0.0  45.1  49.5  90.0  89.5  1.51  6.7  29.9  48.5  M2P/AI2O3 2  2  4  2  2  3  2  Pt-Coo. Ni2P/Al 0 4  2  3  Pt-Co .4Ni2P/Al O3-WA 0  2  mol %  1 order kinetics: Specific consumption o f 4 , 6 - D M D B T based on the inlet = k C A and k is specific rate constant. 0  f Selectivity based on average o f 4 h after 8 h stabilization period.  production o f more products to the D D S route than the N12P. Table 4.8 also shows the product selectivity obtained using Coo.4Ni P supported on 2  fluorinated AI2O3. Fluorine was incorporated on the support to increase the Bronsted acidity i n order to induce isomerisation o f the 4 , 6 - D M D B T molecule and enhance the D D S route. Comparatively, this catalyst (C00.4N12P/AI2O3-F) shows a slight increase i n conversion (87.1%)  107  Chapter 4 and increased cracked products than the catalyst without fluorination (Coo.4Ni2P/Al2C»3). K i m et al. (2003) using sulfided N i M o and C o M o (both on fluorinated AI2O3) reported an enhancement o f the catalytic activity (compared to sulfided N i M o and C o M o on non-fluorinated A1 0 ) i n 2  3  addition to increased D D S products from 4 , 6 - D M D B T . Another interesting selectivity is shown by the Ni2P/MCM i n Table 4.8. Both the hydrogenation and D D S products are small but with significant cracked products (not shown). Although the conversion was very high, the products obtained from the D D S and hydrogenation routes are further cracked due to the high Bronsted acidity o f the M C M support. After selecting the Coo.4Ni2P/Al203 based on the increased D D S product obtained, it was necessary to explore the possibility o f enhancing the hydrogenation route o f Figure 2.6 in order to increase conversion. Pt was incorporated to C00.4N12P/AI2O3 and the product selectivity is shown on Table 4.8. In addition to the increased conversion obtained (97.2 mol%) compared to 86.7 m o l % for the Coo.4Ni2P/Al203 the hydrogenation products also increased (45.2 m o l % compared to 28 m o l % for the Coo.4Ni2P/Al 03) with little cracked products on this catalyst. 2  Since the Pt-Coo. Ni2P/Al203 produced high conversion with less cracked products, the 4  possibility o f further enhancing the conversion by using a weakly acidic support to increase isomerisation was explored. The product selectivity using this catalyst, Pt-Coo.4Ni P/Al203-WA 2  is shown in Table 4.8. The hydrogenation and D D S product selectivity was 29.9 and 48.5 m o l % respectively and the remainder being cracked products. The weakly acidic support also showed a decrease in the conversion to 90 m o l % . To summarize the effects o f supports on the H D S o f 4 , 6 - D M D B T , it is noted that AI2O3 is still the support o f choice as promoted C00.4M2P phosphide can be impregnated on it to give high Bronsted acidity. In the present study, the effect o f fluorine only marginally increases the conversion. Pt added to C00.4N12P/AI2O3 produces very high conversion and high D D S with  108  Chapter 4 significant hydrogenation products. The low Brensted acidity is suitable for isomerisation while the presence o f Pt promotes hydrogenation. The use o f a weakly acidic support showed increased Brensted acidity, 4.0 umol/m compared to the AI2O3 support (2.7 umol/m ). Therefore based on the high conversion and the relatively high D M B P products obtained on the Pt-Coo.4Ni2P/Al20  3  this catalyst was selected and further examined using the 4 , 6 - D M D B T at different temperatures 533, 548 and 583 K to generate the data as plotted in Figure 4.12. The Arrhenius equation was subsequently used to determine the apparent activation energy Ea. The determined value for the E a is 65 kJ/mol. Kabe et al. (1999) have reported activation energy o f H D S o f D B T in the range  -0.4 y = -7.7942x + 12.759 -0.8  c  A  -1.2  -1.6  -2 1.7  1.74  1.78  1.82  1/171000 K  1.86  1.9  -1  Figure 4.12 Plot o f In k versus 1000/T for the hydrodesulfurization o f 4 , 6 - D M D B T using P t - C o o ^ P / A b C ^ .  109  Chapter 4 o f 87.9-94.9 kJ/mol. The apparent activation energy E a reported i n the present study includes the heat o f adsorption, X (Satterfield, 1991) hence, E a = Es-X,  (4.1)  Es accounts for the surface reaction.  4.6 Kinetics of the H D S Based on the high conversion with few cracking products, the Pt-Coo.4Ni P/Al 03 catalyst 2  2  was selected and examined at different space velocities using 4 , 6 - D M D B T . The data was then used to determine the overall lumped kinetic rate constant, k. The equation governing the activity in the fixed bed reactor as used in this study is related through equation 4.2.  (4.2)  where x is the conversion o f reactant 4 , 6 - D M D B T , W is the weight o f the catalyst, F o is the A  A  molar feed rate o f the 4 , 6 - D M D B T and r  A  is the rate o f reaction. The H D S kinetics o f model  compounds and real feed have been reported using the pseudo-first order equation (Manoli et al., 2004; Fluck et al., 1974). Therefore substituting for r i n equation 4.2 yields equation 4.3 A  (4.3)  dx  A  d  { W K  1  \  =  -r =kC =kC (l-x ) A  A  M  A  AoJ  110  Chapter 4 where k is the kinetic rate constant, C A and C A are the initial and the final concentrations o f the 0  feed 4 , 6 - D M D B T . Equation 4.3 is solved to yield equation 4.4.  -\n(l-x )=kC /SV A  (4.4)  Ao  where S V is the space velocity. Hence a plot o f 1/SV (known as the space time) versus l n ( l - X A ) gives a slope from which k can be found. Figure 4.13 shows a typical plot o f the 1/SV versus l n ( l - x ) obtained from data o f Pt-Coo.4Ni P/Al 03 at 533 K and 3.0 M P a . A  2  2  Figure 4.13 gives an excellent fit o f the measured data to equation 4.1 with a correlation coefficient o f R = 0.999 and with the gradient o f 0.2158. Substituting the slope i n equation 4.1 2  and solving gives the value, 2.13 x 10 ml/gcat. h as the pseudo first order rate constant obtained 4  on the Pt-Coo.4Ni P/Al 03 catalyst. Haji et al. (2005) reported a pseudo first order kinetic rate 2  2  constant k = 3.3 x 10 ml/g cat.h for 5.7 wt % P t / A l 0 4  2  3  obtained using D B T (containing 887  4 -r 3.5 -  y = 0.2189x + 0.3676 R  3^  2  = 0.9995  2.5 -  < 1 c  21.5 1 0.5 0 -• 0  2  6  4  8  10  12  1/space velocity, h-gcat/mol Figure 4.13 Plot o f In (1-XA) versus space time at 533 K , 3 M P a using Pt-Coo. Ni P/Al 0 4  2  2  111  3  Chapter 4 wppm S) at 583 K and 1 atm i n a fixed reactor. In order to further clarify the H D S kinetics o f the 4 , 6 - D M D B T , the lumped pseudo-first order rate constant obtained was further examined using the simplified reaction  scheme  presented i n Figure 4.14. Each step was assumed to follow first order kinetics. The dependence on hydrogen concentration is neglected because hydrogen was supplied at high pressure.  DMBCH (D)  Figure 4.14  Simplified reaction scheme for the H D S o f 4 , 6 - D M D B T  112  Chapter 4 The kinetic equations used i n the simulation are shown i n eqns. (4.5) to (4.8).  dX.  =  (4.5)  r =-(k k )C A  l+  2  A  d \  F  A o J  »  - , -kC  dX  (4.6)  -kC  d \  F  A o J  dX  r  W  d \  F  c  (4.7)  =k C +k C-kC 2  2  A  A  3  i  B  4  c  A o J  dX„  (4.8)  W  d \  F  A o J  The Gaussian Newton-Raphson optimization method (Appendix E ) was used to estimate the rate constants assuming gas phase reaction and first order kinetics i n the fixed bed reactor, using the ordinary differential equations 4.5 to 4.8 and the estimated values are summarized i n Table 4.9. The tolerance for calculating the rate constants is 10" . The model is sensitive to the 6  initial values and the step size.  113  Chapter 4 Table 4.9 Estimated I order rate constants for hydrodesulfurisation o f 4 , 6 - D M D B T at 533 K s  Value x 10" Rate constant  J  (ml/gcat.h)  ki  0.2435 ± 0 . 0 0 9 3  k  2  0.0116 ± 0.0005  k  3  0.0058 ± 0 . 0 0 0 7  k4  0.7219 ± 0 . 0 3 2 5  A l l the estimated rate constants (k/s) represent each step o f the hydrodesulfurization o f 4,6-DMDBT  simplified  network  for the  Pt-Co .4Ni P/Al O 0  2  2  3  at  533  K  and  3.0 M P a .  Hydrodesulfurisation proceeds through the D D S route and/or hydrogenation route that involve k\ and k respectively. The magnitude o f the ratio o f k]/k is 21. This high ratio implies that D D S 2  2  on the P t - C o o . 4 N i P / A l 0 is faster than the hydrogenation. It therefore explains the higher yields 2  2  3  o f the D M B P compared to the M C H T . The magnitude o f L i is also high explaining the high yields o f D M B C H . However, from the reaction scheme Figure 4.14, it is noted that the M C H T product could also come from the hydrogenation o f D M B P and since L i is higher than k  3  the  products quickly convert from M C H T to D M B C H . Hydrogenation is also enhanced once hydrogenolysis takes place and the presence o f platinum on the catalyst could explain this phenomenon o f increased hydrogenation. Figure 4.15 shows the correlation from the experimental data and that obtained from the model. The correlation o f R = 0.9813 is very high and therefore indicates agreement between 2  the data and the predicted. 114  Chapter 4  6.00  0.00  1.00  2.00  3.00  4.00  5.00  6.00  Experimental data: Concentration, mol/ml  Figure 4.15: Correlation o f the experimental data versus the predicted from the model for conversion o f D M D B T (0) and yields o f products ( D M B P : * ; M C H T : A ; DMBCH: •  4.7  Discussion on HDS of supported metal phosphides The supports i n this study were chosen in order to enhance Bransted acidity to achieve  high conversion by creating the desired isomerisation o f the methyl substituents.  Figure 4.16  shows the plots o f Bransted acidity as a function o f both conversion and D M B P selectivity. Except for the Pt-Coo.4Ni P/Ai203, all the prepared phosphides show a trend o f increasing 2  115  Chapter 4  Figure 4.16  Plots o f Bronsted acidity as a function o f both conversion and D M B P selectivity on all supported catalysts: Ni2P/Al 0 (•), Coo.4Ni2P/Al2C»3 2  3  ( A ) , C o o . 4 N i P / A l 0 3 - F ( • ) , C o . N i P / M C M (•) 2  2  0  4  2  Pt- C00.4N12P/AI2O3 (•), Pt- C o . 4 N i P / A l O 3 - W A ( A ) 0  116  2  2  Chapter 4 conversion with increased Bronsted acidity. The Pt-Coo.4Ni2P/Al203 showed this exceptional behavior because it combined both the Bronsted acidity function with the hydrogenation as significant hydrogenation and direct desulfurization products are obtained. The n - P A titration o f the Bronsted acidity shows good repeatability (see Appendix BIO) and therefore the fits obtained in Figure 4.16 are reliable. The least square fit is used to fit the curve for Bronsted acidity with D M B T selectivity with r = 0.9993. This is the expected outcome as the presence o f the C o i n 2  N12P catalysts provides the necessary isomerization o f the methyl substituents to less hindered position o f the 4 , 6 - D M D B T molecule. Recent studies (Mizutani et al., 2005) using promoted C o with M0P/AI2O3 obtained high D M B P products relative to the hydrogenation products. The D M B P selectivity in the present study shows a maximum after which the selectivity begins to decline. The least D M B P selectivity was obtained on the M C M support, which also has the highest Bronsted acidity and higher cracked products. This finding seems to suggest that although Bronsted acidity is necessary for isomerisation of the methyl groups to enhance conversion o f the 4 , 6 - D M D B T molecule, excess Bronsted acidity leads to cracking of the products. Therefore it is necessary to use supports that w i l l provide intermediate Bronsted acidity on catalysts i n order to avoid cracking after the products are formed. The addition of F to Coo.4Ni2P/Al203 did not significantly change the conversion nor the selectivity as shown in Table 4.8. Although the Bronsted acidity increased from 304 pmol/g to 390 pmol/g, the C O uptake did not change as F was added to Coo.4Ni2P/Al203 suggesting that the active sites remained the same hence conversion w i l l not be enhanced. It is noted that i n addition to the  high conversion  obtained  on the  Pt-Coo.4Ni P/Al 03 catalyst, 2  2  the  selectivity  to  hydrogenation products (45.1 mol%) were higher than the non-platinated C00.4N12P/AI2O3 (30.1 mol%). The presence o f Pt was expected to increase the hydrogenation capability of the  117  Chapter 4  C00.4N12P/AI2O3 thereby increasing the conversion of 4,6-DMDBT through the hydrogenation route. Recently, Pinzon et al. (2006) reported high conversion of DBT when Pt was added to  M0/AI2O3 compared to non-platinated M0P/AI2O3. The authors attributed the enhanced HDS activity to an improved dispersion of the active species on the support. However, in the present study, the selectivity to the DDS products obtained from the Pt-Coo.4Ni P/Al C>3 (49.5 mol%) 2  2  was marginally smaller than that obtained from the non-platinated Coo.4Ni2P/Al203 (52.4 mol%). The presence of Pt on the Coo. Ni P/Al 03 probably covered part of the P thereby reducing the 4  2  2  Brensted acidity and could result in the small decrease of the DDS product selectivity. A second explanation is that Pt could interact with the Co and N i resulting in agglomeration which could affect the surface area of the catalyst. The BET surface area of this catalyst was the least, 89 m /g. Therefore, addition of Pt is beneficial to the catalyst as it results in increased conversion of 2  4,6-DMDBT.  4.8  Used catalyst properties of the bulk and supported metal phosphides It is important that the catalysts be examined after use to determine if the surfaces have  been deteriorated since surface deterioration will lead to low activity. Therefore after reaction, the catalysts were cooled in flowing He and then passivated in low pressure oxygen at room temperature. Selected samples of the used catalysts were characterized by X R D , TPR and XPS. The X R D data, shown in Figure 4.17, suggest that the bulk phases present in the catalyst did not change after use, although the sharper peaks are indicative of some crystal growth. The TPR profile of the used catalyst (not shown) was also featureless suggesting that following the passivation and air exposure the catalyst samples had very little reducible surface oxygen.  118  Chapter 4  Pt-Coo Ni P/Al 0 2  4  2  3  *  1 ro  :  20  •  1 30  .  1 40  .  , 50  .  , 60  ,  , 70  29  Figure 4.17 Comparison o f the X R D profiles o f the fresh (reduced) and spent (after activity test) metal phosphides  It was noted that S was not detected by X P S on the used catalysts. Furthermore, the P 2p  3/2  X P S spectra o f the used M o P and C o o . o s ^ P catalysts, shown in Figure 4.18 A , provided  evidence for the presence o f surface phosphide with a peak at low B E (129-130 eV). O n the used N i P and Cooo MoP, however, the P was present as a surface phosphate ( B E = 133.5-134.2 2  7  eV), suggesting that.the metal phosphide was more stable on the M o P and C o o . o s ^ P than the  119  Chapter 4  Ni P 2  CoNLP  CoMoP  Figure 4.18 Comparison o f the X P S o f the fresh (reduced) and spent (after activity studies) metal phophides  120  Chapter 4  C00.07M0P and M2P catalysts. The change in P B E noted here was also consistent with the increase in M o 3d and N i 2p B E , as shown in Figure 4.18 B . In summary, characterisation o f spent metal phophides showed that the prepared metal phosphides used i n the present study are stable in the presence o f small H2S produced during the H D S o f 4 , 6 - D M D B T . Although the modified metal phosphides were tested for 14 hours it is expected that during hydroprocessing with industrial feedstock these metal phosphides w i l l maintain the phosphide phase without being sulfided hence the activity o f these catalysts w i l l be better than the metal sulfided catalysts.  4.9  Catalyst regeneration Although the used catalysts were not regenerated, certain methods can be used to  regenerate them. The organic residue on the surface o f the catalyst could be burnt off in air at temperatures 400-600°C. Another method o f catalyst regeneration is by treating the organic solvents at temperatures 140 °C to extract organic deposits that plug the pores o f the catalyst (Satterfield, 1991).  121  Chapter 5  Chapter 5 Hydrodenitrogenation of carbazole over bulk and supported metal phosphides  This chapter w i l l present data on bulk and supported metal phosphides used for the H D N of carbazole, a model compound o f the organo nitrogen compounds present i n petroleum feedstock. The chapter w i l l first present the preparation and characterization o f bulk M o P and N i M o P (0.07 < x < 1.11) catalysts. The first section w i l l discuss the different methods used to x  characterize the bulk and surface properties o f the prepared catalysts. The results o f activity studies using the bulk metal phosphides w i l l be presented. Finally, the characterization and activity data o f supported metal phosphides w i l l be discussed. Recent results on a series o f CoxNi2P and C o M o P catalysts, in which the P content, x  catalyst acidity and hence selectivity during H D S o f 4,6-dimethyldibenzothiophene, were shown to be dependent upon the C o content (see Chapter 4), the present study was aimed at determining i f a similar effect was observed with N i M o P catalysts for the H D N o f carbazole. Carbazole is a x  non-basic refractory heterocycle used here for the first time to examine the activity o f metal phosphide catalysts.  Note: A version of this chapter is in the press, Catal. Today, 2007  122  Chapter 5 5.1  Bulk N i M o P catalysts x  The preparation o f a series o f N i M o P (0.0 < x < 1.11) catalysts for the H D N o f carbazole x  was done following a similar procedure to that reported earlier (Chapter 4) for the Co Ni2P bulk x  catalyst.  5.2  Catalyst characterization T P R was used to characterize the precursors and X R D , X P S , C O uptake and n - P A uptake  were  used to characterize the reduced catalysts as stated  earlier i n Chapter 4. The  characterization was performed to determine the properties o f the catalysts and to ascertain whether or not the required phases and compositions were synthesized.  5.2.1  TPR T P R on the precursors was conducted i n a 60 ml/min flow o f 10 % H2 in A r using 0.2-0.4  g o f the calcined sample loaded into a stainless steel reactor (i.d. = 9 mm). The T P R profiles o f the bulk N i M o P ( 0 . 0 < x < l . l l ) precursors are shown i n Figure 5.1. The data show a clear trend x  o f decreasing reduction temperature as the N i content increased. For the N i 1.1 M o P precursor, two reduction peaks were observed at about 840 K and 995 K . The profile is similar to that reported for the bulk N i P precursor ( A b u and Smith, 2006) but with the peak temperatures 2  occurring at about 50 K higher than for the M2P precursor. The decreased reduction temperature with increased N i content is indicative o f a significant interaction between the N i and the M o P .  123  Chapter 5  Figure 5.1  T P R o f calcined catalyst precursors ( N i M o P for 0.0 < x < 0.11) measured i n 10% x  H2 in A r at flowrate o f 60 ml(STP)/min and 1 K / m i n  5.2.2  XRD The X R D diffractograms o f the reduced catalysts, presented i n Figure 5.2, confirmed that  the metal phosphides had been prepared. M o P phase was present, but as more N i was added to the N i M o P , the formation o f a N i M o P phase was clearly evident. For the N i i . n M o P catalyst, the x  N i M o P phase was dominant. The X R D data were used to estimate the crystallite size (d ) and c  the lattice parameters o f the M o P phase, and the results are shown i n Table 5.1. The crystallite  124  Chapter 5  20  30  40  50  60  70  26 * N i M o P Phase  Figure 5.2  X-ray diffractograms o f all reduced N i M o P for 0.0 < x < 1.11 x  size ranged from 1 6 - 2 2 nm for 0 < x < 1.11 and as the N i content o f the N i M o P catalyst x  increased, the M o P crystallite size also increased. The M o P lattice parameters were a = 0.3241 ± 0.009 nm and c = 0.3180 ± 0.007, in good agreement with the M o P lattice parameters o f a = 0.3223 n m and c = 0.3191 ( J C P D S , 2005). Furthermore there was no consistent reduction in the lattice parameter c as N i content increased. In previous work, M a et al. (2004) prepared  125  Chapter 5  Table 5.1 Lattice parameters estimated from X R D o f reduced catalysts.  M o P phase  Catalyst  Ni content  Phases  mole %  20, degree  Lattice parameters,  Crystallite size,  nm  nm  (100)  (101)  (110)  (111)  a  c  d (101) c  MoP  0  MoP  31.93  42.89  57.16  64.93  0.3240  0.3196  16  Nio.o7MoP  3  MoP  32.94  42.90  57.02  64.72  0.3255  0.3185  18  Ni .i6MoP  7  MoP, NiMoP  31.98  42.96  57.08  64.68  0.3254  0.3182  21  NiojgMoP  15  MoP, NiMoP  31.90  42.92  57.06  64.68  0.3256  0.3184  22  Ni,.i,MoP  34  MoP, NiMoP  32.04  43.16  57.12  64.72  0.3245  0.3170  22  0  126  Chapter 5 Ni(2- )Mo P catalysts and showed the formation o f solid solutions for a range o f N i / M o ratios > 1. X  x  In the present study, however, the ( N i + M o ) content o f the N i M o P catalysts increased as the N i x  content increased and the X R D data show that with the preparation method and catalyst composition used herein, a catalyst with a mixture o f N i M o P and M o P phases was produced. Table 5.2 reports the B E T surface area, the C O chemisorption uptake and the n - P A uptake o f the bulk N i M o P catalysts. The B E T surface area varied marginally from 5.3 - 6.0 x  m / g for the N i M o P ( 0 < x < l . l l ) catalysts and there is a modest increase i n the surface area 2  x  upon addition o f N i to M o P . The C O uptake and the n - P A uptake reported per unit B E T area o f the catalysts, increased significantly with addition o f N i to the M o P . Furthermore, as the N i content increased, the C O uptake increased, whereas the n - P A uptake showed a maximum for the N i M o P catalysts. x  A l s o note that the Nio.o7MoP catalyst had the lowest C O uptake and the  highest n - P A uptake.  5.2.3  XPS X P S was used to characterize the surfaces o f the prepared catalysts. Figure 5.3 shows the  N i 2p, M o 3d and the P 2p X P S spectra o f the prepared catalysts. A l l the spectra showed oxidized N i , M o and P species because after synthesis, the phosphides were passivated using 2 % O2 in He to protect them from deep oxidation. The P 2p /2 region showed peaks at 133.8 e V and 3  another peak at 129.8 e V for all the prepared catalysts. Similar results have been reported by others (Phillips et a l , 2002; Harris and Chianelli, 1986; Okamoto et al., 1980). The peak at 133.8 e V is attributed to surface PO4 " species (Harris and Chianelli, 1986; Okamoto et al., 1980). The lower B E peak at 129.8 e V is attributed to metal phosphides. The B E o f the N i 2p is consistent with the presence o f metal phosphides and phosphates, corresponding to B E s o f 853.7 e V and 857.3 e V , respectively. Similarly, M o 3d B E at 228.2 e V and 232.2 e V indicate the formation  127  Chapter 5  Table 5.2  Physiochemical properties o f prepared metal phosphide  Chemisorption Catalyst  B E T area  P/M*  C O uptake  n - P A uptake  Nominal  XPS  m /g 2  umol/m  umol/m  atom ratio  atom ratio  5.3  0.21  8  1.00  1.00  N10.07M0P  6.0  0.83  40  0.93  1.03  M0.16M0P  5.9  1.15  35  0.88  0.99  Nio.3sMoP  5.9  1.26  29  0.78  0.93  Nii.nMoP  5.9  2.05  32  0.60  0.84  MoP  2  2  * M : total metals (Ni+Mo) and P: phosphorous  o f phosphide and phosphate species, respectively. The atom percentages  o f the prepared  phosphides were also determined by X P S and the P / M atom ratios (where M is ( N i + M o ) ) are reported i n Table 5.2. The P / M atom ratios show that the Nio.o7MoP enriched i n phosphorous (P/(Ni+Mo) ratio = 1.03) and increasing N i content resulted i n a decreased P / M ratio at the surface o f the catalysts. Figure 5.4 shows a correlation between the X P S and chemisorption data obtained on the metal phosphides o f the present study. Assuming that C O chemisorption titrates metal sites and  128  Chapter 5  Mo 3d  Ni 2p 857.3 eV 853.7 eV  Ni, nMoP  Nio. MoP 38  "Ni  016  MoP  Mi MoP 007  245  235  225  215  870  860  Binding Energy, eV  850  840  830  Binding Energy, eV  P 2P  133.8 eV 129.8 eV  Jj V  \ ^  NimMoP  V  Ni  038  Ni Ni 1  145  1  140  1  1  135  007  MoP  016  MoP  MoP  1  130  125  120  115  Binding Energy, eV  Figure 5.3  X P S spectra o f the N i 2P, M o 3d and P 2p o f the prepared phosphides  129  Chapter 5  y = 0.0056x + 0.786  1.05  £ ><  •B as  R  2  = 0.9067/m  1.00  0.95 0.90 0.85  •  0.80 10  20  30  40  50  n-PA/CO uptake ratio by adsorption  Figure 5.4  Correlation o f P / M ratio determined by X P S and n - P A : C O uptake ratio determined by adsorption for N i M o P for x  0.0 < x . < 1.11(B) and M o P ( A )  .  the n - P A uptake titrates Bronsted acid sites, presumably associated with the surface phosphate species, the P / M atom ratio as determined by X P S should correlate with the n - P A to C O uptake ratio as determined by adsorption. The data plotted i n Figure 5.4 show a good correlation ( R = 2  0.906) and are consistent with a similar correlation determined for the series o f Co Ni2P catalysts x  reported i n Chapter 4.  130  Chapter 5 5.3  Catalyst Activity of bulk Ni MoP x  Catalyst activity measurements were carried out using carbazole as the model reactant. The catalysts were passivated following synthesis and, therefore, prior to performing the activity measurements, the passivated catalysts were re-reduced in pure H 2 flow at 723 K for 1 h. Catalyst conversion data were then measured and the conversions typically reached steady values after about 8 hours time-on-stream. The carbazole consumption rates reported in Table 5.3 are the time-averaged values over a period o f at least 4 hours that followed the initial 8 hour stabilization period. A thermal run was carried out using xylene at the testing conditions and the products were always subtracted when the feed containing carbazole and xylene was used. In addition, for all the data reported herein, the carbazole mole balance between the reactor feed and reactor effluent was > 95% (see Appendix D3 and for reproducibility). The data o f Table 5.3 show that the carbazole conversion over the M o P catalyst was 54.5 mol % , whereas a significant increase occurred (86 - 90 mol%) for the N i M o P catalysts (0.07 < x  x < 1.11). The carbazole consumption rates in terms o f the specific rate (assuming first order reaction kinetics), areal rate and T O F are also reported in Table 5.3. The areal rates are based on the measured B E T surface area whereas the specific rates are referred to the mass o f catalyst used. The specific rate o f carbazole consumption ranged from 1.21 - 3.54 x 10" mol.g" .s" with 8  1  1  the NI0.07M0P showing the highest specific rate o f carbazole consumption and the M o P showing the lowest. The areal consumption o f carbazole obtained from the prepared metal phosphides followed a similar trend as the specific consumption rate and ranged from 2.28 - 5.91 x 10" 9  1  9  mol.m" .s" . However based on the T O F , the M o P catalyst showed about twice the carbazol  131  Chapter 5  Table 5.3 Activities o f bulk metal phosphides for the H D N o f carbazole measured at 583 K and 3.0 M P a H  2  Carbazole Consumption Rate Catalyst  Selectivity  Total Conversion mol %  10 mol.g" .s"  MoP  54.5  1.21  2.28  1.1  20.5  4.3  26.0  30.9  16.0  M0.07M0P  90.0  3.54  5.91  0.7  85.6  2.2  7.9  2.9  0.0  Ni .i6MoP  89.6  3.48  5.89  0.5  77.8  0.7  17.0  0.5  0.4,,  Nio.3 MoP  88.8  3.36  5.67  0.4  51.2  2.1  10.0  27.8  6.7  Nii.nMoP  85.8  3.00  5.07  0.2  40.4  3.3  17.5  34.9  2.4  0  8  Specific s  Areal 1  1  10* mol.m^.s"  TOF  BCHX  10-V  1  B T : Other products formed with carbon number < 12 A T : Other products formed with carbon number > 12  132  CPCHX  THCZ  mol %  Other products BT AT -  Chapter 5 consumption o f the Nio.o7MoP catalyst and this result is i n agreement with previous studies that report the M o P catalyst activity is not promoted by addition o f N i (Stinner and Prins, 2001). The main products o f reaction o f carbazole for all the catalysts were bicyclohexane (BCHX)  and tetrahydrocarbazole ( T H C Z ) . Smaller quantities o f cyclopentane cyclohexyl  methane ( C P C H X )  were also identified by G C - M S  (Agilent 6890/5973N). The direct  hydrogenolysis product biphenyl was not observed. In addition, both the aromatic rings o f the products were saturated and no products with partially hydrogenated aromatic rings were observed. There were also some cracked products but these were not identified by G C - M S . The data o f Table 5.3 show that the selectivity to B C H X o f the N i M o P catalyst was significantly x  greater than that obtained over M o P . Furthermore, for the N i M o P catalysts, * the x  BCHX  selectivity decreased as the N i content increased and the highest B C H X selectivity occurred for the Nio.o7MoP catalyst. The carbazole conversion, B C H X selectivity and n - P A uptake are plotted as a function o f the C O uptake i n Figure 5.5. A s noted above, the data show a significant increase i n n - P A uptake, B C H X selectivity and carbazole conversion with N i addition to the M o P .  However, as  the N i content increased (x > 0.07 i n N i M o P ) the n - P A uptake and B C H X selectivity decreased, x  with a small reduction i n carbazole conversion.  133  Chapter 5 5.4  Discussion on conversion of carbazole using bulk phosphides The data o f Table 5.3 show that the activity o f the M o P catalyst (2.3 x 10" m o l 9  1 0 0 -, o  8  0  -  a> CO  60  -  X  •  20  -  1 00  -  90  -  "  D O O  Conversion  ^  4 0 -  "  —  O  O CQ  •  C M  £  o  4 0 -  1. 3 0  -  ^  ^  ^  ^  •  to ro 2 0 Q_  -  < Q.  -  10  •  i  o  -  J  i  0.0  1  1 0.4  ,  .  0.8  CO  ,  i  1.2  |  i  i  1.6  uptake,umol/m  2.0  1 2.  2  Figure 5.5 Conversion o f carbazole and selectivity to B C H X over various metal phosphide catalysts at 583 K and 3.0 M P a H plotted as a function o f C O 2  uptake rati o f N i M o P (•) 0.0< x < 1.11 and M o P ( A ) catalysts x  m" s" ) was slightly lower than the N i M o P catalysts and the highest areal activity was obtained 2  1  x  on the  N10.07M0P  catalyst with a value o f 5.9 x 10" m o l m" s" at 583 K and 3.0 M P a H . Areal 9  2  1  2  134  Chapter 5 rates o f o-propylaniline H D N at 643 K and 3.1 M P a H2 over metal phosphides have been reported by Stinner et al. (2001). Values o f 312 x 10" m o l m" s" and 55 x 10" m o l m" s" were reported on bulk M o P and N i M o P , respectively.  These activities are at least an order o f  magnitude higher than the values reported herein for carbazole H D N . Wang et al. (2002) reported quinoline H D N T O F data for supported C o P (1.2 x 10" s" ) and supported M2P (0.60 x 3  1  10" s" ) at 643 K and 3.1 M P a in the presence o f dibenzothiophene. The relatively l o w activity o f 3  1  quinoline at this temperature was attributed in part to the inhibiting effects o f the sulfur compound. Carbazole is a non-basic, five-membered N-heterocycle in which the extra pair o f electrons o f N are involved in n electron-cloud bonding and therefore are not available for interaction with the catalyst surface (Kilanowski et al. 1978). Hence, carbazole has a l o w reactivity that makes it difficult to remove the N heteroatom and the data reported herein are consistent with this observation. The data o f Table 5.3 also show that the bulk M o P had the highest T O F among the catalysts tested. O n bulk M o P , a T O F o f 11.1 x 10" s" was obtained 3  1  after an initial 8 hours o f reaction at 583 K and 3.0 M P a . The activity compares favorably to that reported for M02N with the same reactant.  Nagai et al. (2000) reported a T O F for carbazole  H D N o f 29 x 10" s" over nitrided 12.5 wt% M0/AI2O3, measured i n a fixed bed microreactor at 3  1  573 K and 10.1 M P a . However, the catalyst T O F was determined after 30 minutes time-onstream and the activity decreased by a factor o f about two within 2 hours time-on-stream Nagai et al. (2000). The catalyst activity and selectivity during carbazole H D N can be related to the catalyst properties. The X P S and adsorption data o f Figure 5.4 suggest that metal sites are associated with C O adsorption and acid sites are associated with P, and these observations are consistent with similar data reported for the Co Ni2P catalysts, i n Chapter 4 and ( A b u and Smith, 2006). x  The source o f the Bransted acidity on the bulk phosphide is likely a consequence o f the  135  Chapter 5 pretreatment o f the catalysts before reaction with H2 at 723 K , that leads to formation o f water with the most reactive passivated oxygen. The water may react with surface phosphate to produce the H P 0 4 x  ( x 3 )  species responsible for the Bronsted acidity. The N i M o P catalysts all x  had higher metal and acid site densities than the M o P catalysts, possibly because o f a higher dispersion o f the mixed N i M o P and M o P phases present on the catalyst. Although a slight P enrichment o f the surface occurred for Nio.o7MoP compared to M o P , the P content decreased as more N i was added to the N i M o P catalysts. x  The data o f Figure 5.5 show that the acid site  density decrease associated with the N i content o f N i M o P is relatively small when compared to x  the M o P data.  The selectivity for B C H X among the metal phosphides was highest on the  Nio.cnMoP catalyst. However as the N i content o f the N i M o P catalysts increased, the selectivity x  to B C H X decreased in favor o f lighter products. In all cases, however, the M o P had lower B C H X selectivity than the N i M o P catalysts. A significant increase in conversion and B C H X x  selectivity is observed when comparing M o P to the N i M o P catalysts. However, for the series o f x  N i M o P catalysts, the conversion is relatively constant, as C O uptake increases, whereas x  selectivity to B C H X decreases. Indeed the trend in B C H X selectivity follows the trend in n - P A site density as C O uptake increases. Figure 2.3 (Chapter 2) shows the proposed reaction scheme for the H D N o f carbazole on nitrided M o / A l 0 catalyst (Satterfield and Cocchetto, 1981). The authors attributed the B C H X 2  3  formation to the C - N hydrogenolysis o f the perhydrocarbazole and the cyclohexylbenzene and cyclohexylhexene to C - N hydrogenolysis o f the hexahydrocarbazole and decahydrocarbazole respectively. Clark et al. (2002) have noted that acid sites alone do not lead to active hydroprocessing catalysts. According to these authors, C - N bond cleavage on M o P catalysts is dominated by a Hoffman type elimination reaction, in which both the a - and P-carbon atoms need to be saturated to allow the C - N hydrogenolysis reaction to take place. Hence the activity o f  136  Chapter 5 catalysts towards unsaturated hydrocarbons is limited by the hydrogenation ability o f the catalyst. The H D N o f heterocyclic compounds proceeds by hydrogenation o f the aromatic ring and then subsequently  by hydrogenolysis o f the C - N bond (Clark and Oyama, 2003).  Hydrogenation and hydrogenolysis occur on two types o f catalytic sites that are responsible for the direct C(sp ) - N bond hydrogenolysis and hydrogenation and C(sp ) - N bond cleavage during H D N reaction (Jian and Prins, 1996). O n the M o P catalyst, moderate activity is obtained with low selectivity to the B C H X and higher selectivity to T H C Z .  Addition o f N i to M o P yields  increased metal sites (based on C O uptake) that also increased catalyst hydrogenation activity. Jian et al. (1996) reported that M o P was a better catalyst for the direct C - N bond hydrogenolysis but on addition o f N i , the hydrogenation activity increased significantly. Wada et al. (1996) reported that on introduction o f the noble metals to Ni/Y-zeolite, D B T was completely converted as compared to the 88% conversion obtained using Ni/Y-zeolite without noble metals. The authors attributed the increased conversion to the hydrogenation capability o f the noble metal present in Ni/Y-zeolite. A n increased hydrogenation capability o f the Nio.o7MoP yielded higher selectivity to B C H X and lower selectivity to T H C Z , compared to M o P . The higher B C H X selectivity is a consequence o f the formation o f more C(sp ) - N bonds that are readily cleaved. The data o f Figure 5.5 also suggest that the bond cleavage is related to the acid sites o f the catalysts since for the N i M o P series, as the metal site density increases, the acid site density decreases as does the x  selectivity to B C H X .  The products from carbazole H D N included small amounts o f C P C H X ,  likely produced as a result o f ring opening or isomerization due to Bransted acidity.  After  hydrogenation o f the aromatic rings and the C - N hydrogenolysis, the resulting molecule may undergo a C - C bond cleavage and isomerization to form a five-member carbon ring while leaving one o f the six-member saturated aromatic ring intact to produce the C P C H X product.  137  Chapter 5 Conclusions on bulk phosphides for conversion of Carbazole In the present study, the N i M o P phosphides had a lower T O F for the H D N o f carbazole x  compared with M o P . However, the selectivity to bicyclohexane was greater on the N i M o P x  catalysts compared to M o P , and the Nio.o7MoP had the highest B C H X selectivity and highest T O F among the N i M o P series. The improved selectivity is attributed to enhanced hydrogenation x  in the presence o f the N i M o P catalyst that was a mixture o f M o P and N i M o P phases, and had x  higher C O uptake than the M o P . The C - N bond cleavage is attributed to the acidity o f the catalyst.  Properties and hydrodenitrogenation study of M0.33M0P supported catalysts This section w i l l present the preparation, characterization and activity measurements o f Nio.33MoP supported on different  supports. The different  supports used include, Y-AI2O3  (Aldrich, SA=155 m /g, PV=0.76 ml/g, pellet size 4.5mm), fluorinated AI2O3 and M C M . A s 2  already mentioned i n chapter 4 o f this study, AI2O3 presents unique interaction with sulfided C o M o for H D S to provide overwhelming H D S efficiency and also provides a high dispersion o f the active phase (Anderson et al., 1982). Fluorine in the form o f NH4F impregnated on AI2O3 provides increased acidity, better dispersion and higher hydrogen chemisorption that enhance HDN  compared with non-fluorinated AI2O3 (van Veen et al., 1993; Q u i and Prins, 2003;  Lewandowski and Sarbak, 1997; Fierro et al., 1991). A previous study (Fierro et al., 1991) has been carried out using sulfided catalysts hence it is important to find the effects o f fluorination on metal phosphides. M C M contains Bronsted acid sites that enhance the C - N conversion during H D N therefore the prepared metal phosphides w i l l be supported on M C M to study the effects on conversion and selectivity.  138  Chapter 5 The supported catalysts were prepared in a similar manner as discussed earlier in Chapter 4 o f the present thesis. Therefore the activity o f M0.33M0P catalysts, supported on supports o f different acidities w i l l be presented for both model compounds and a light gas o i l ( L G O ) derived from Athabasca. Figure 5.6 shows the morphology o f the reduced and passivated Nio.33MoP/Al 03 catalyst 2  taken by S E M . Note that interpretation o f this S E M data is difficult because the detailed structure o f the metals ( N i and M o ) and the AI2O3 support is not conspicuously clear.  139  Chapter 5 5.5.1  X R D of modified Nio^MoP on A I 0 2  3  Since it was earlier reported (Clark and Oyama, 2003) that AI2O3 interacts strongly with phosphorous and only forms the active phosphide phase at high temperature, it is necessary to identify the active Nio.33MoP phase on the prepared AI2O3 support. Figure 5.7 shows the X R D profiles of Nio.33MoP/Al 03, and AI2O3. 2  700  -1  600 500 -  cd ^—»  400 -  C ii  300 200 100 -  * Formation o f M o P Figure 5.7 Comparison o f X R D of N i o . 3 3 M o P / A l 0 and A 1 0 2  3  2  3  The formation o f M o P is indicated on the Nio.33MoP/Ai203 profile. The peak at angle 42.9  and plane (101) corresponds to the highest intensity peak o f M o P . The other peak occurs  0  at 32.6  0  and corresponds to the (100) plane o f M o P . N i was absent i n the X R D profile o f the  Nio.3 MoP/Al203 and this observation is consistent with earlier observations with Con.o8Ni P 3  2  140  Chapter 5 (Chapter 4) where small quantities o f C o added to N12P, were absent from the X R D profile. The N i was not revealed as it was highly dispersed on the catalyst. Although the 39.2 peak could not 0  be identified, it could be the result o f the P interacting with the AI2O3 to produce a aluminophosphate as was reported by Stinner et al. (2000).  5.5.2  Properties of prepared supported phosphides The properties o f Nio.33MoP/Al 03 and Nin.33MoP/MCM were determined using B E T , C O 2  uptake and n - P A chemisorption and the data are presented in Table 5.4 .  Table 5.4 Properties o f prepared Nin.33MoP/Ai203 and Nio.33MoP/MCM catalysts  Chemisorption  Catalyst  B E T area  C O uptake  n - P A uptake  2  2  m /g  umol/g  umol/g umol/m  N10.33M0P/AI2O3  65  9.10  319  4.9  M0.33M0P/MCM  132  7.29  779  5.9  The other supported catalysts used i n the carbazole H D N study are Con.4Ni2P/Ai203, Coo.4Ni2P/Al 0 -F and Coo.4Ni P/MCM. These supported catalysts were already tested for H D S 2  3  2  o f 4 , 6 - D M D B T and their properties were reported earlier. The intention o f using these H D S catalysts is to test their potential for H D N as well. In Table 5.4, the M C M support has a larger  141  Chapter 5 B E T surface than the AI2O3 using the same active  M0.33M0P phase. However the C O uptake is  smaller (7.29 pmol/g) on the M C M support compared to the AI2O3 support (9.10 pmol/g). A s presented on Table 5.4, the n - P A acid titration shows that the M C M support has a higher Bronsted acidity (779 pmol/g) than the AI2O3 supported catalyst (319 pmol/g). In fact the magnitude o f the Bronsted acidity per m  o f M C M is still greater than the AI2O3 supported  catalysts. However it is still valid to compare the total Bronsted acidity o f the two different supported catalysts using the B E T surface area because the supports and active phases contribute to the Bronsted acidity. A s expected, previous studies reported high acidity for sulfided C 0 M 0 / M C M (Turage et al., 2003) and N i / A I M C M catalysts (Fang et al., 2005)  5.6  Activity of Ni<u3MoP on different supports A s i n the case o f the bulk activity study, catalyst activity measurements were carried out  over the Nio.3 MoP on different supports, using the same fixed bed reactor operated at 523 - 583 3  K  temperature  and 3.1  MPa H  2  pressure.  The Coo.4Ni P/Al203 2  >  Coo.4Ni2P/Al 0 -F 2  3  and  Coo.4Ni2P/MCM catalysts previously prepared in Chapter 4 for testing the H D S o f 4 , 6 - D M D B T , were also examined using carbazole. Again, prior to performing the measurements, catalysts were pretreated in the same way as the bulk catalysts by passing H at 723 K for 1 h, in order to 2  remove the passivated layer. In this study, carbazole was added to xylene solvent to provide 3000 ppm o f N in the feed. The H D N products o f carbazole using these supported metal phosphides were similar to those obtained previously on the bulk metal phosphides. Table 5.5 shows the activity data o f the H D N o f carbazole using the supported metal phosphide catalysts. The conversion ranged from 82-99.9 m o l % with the lowest conversion obtained on the Coo.4Ni P/Al20 and the highest on the M 0 . 3 3 M 0 P / M C M . 2  3  142  Chapter 5 Turaga et al. (2003) also reported a high conversion o f carbazole (98 wt%) using C o M o / M C M and about 78 wt% for C0M0/AI2O3. The authors worked with 500 ppmw o f nitrogen in carbazole at 573 K , 45 atm and space velocity o f 4 h" . The higher conversion on the !  M C M supported catalyst indicates that the M C M supported metal phosphides have better activity than the AI2O3 supported metal phosphides. However, again the M C M supported catalysts show high amounts o f cracked products. The Ni .33MoP/Al2O3 showed a conversion o f 96 m o l % which 0  is also much higher than the C0M0/AI2O3 catalyst reported by Turaga et al. (2003). Table 5.5 also reports the carbazole consumption rate over all the prepared supported catalysts used in the present study. Assuming first order kinetics, the specific consumption reported in Table 5.5 varied from 2.63 - 21.21 x 10" mol.g" .s" . The specific consumption o f the 8  1  1  Nio.33MoP/Ai203 was 4.94 x 10' mol.g" .s" and Table 5.5 revealed that this catalyst also 8  1  1  produced the highest B C H X selectivity with minimum C > 12 products. The product distribution is similar for all the catalysts supported on the AI2O3, whereas those supported on M C M produced more cracked products. It is important to state that although the hydrogenated T H C Z was detected, successive hydrogenated carbazole compounds such as hexahydrocarbazole, octahydrocarbazole, decahydrocarbazole and perhydrocarbazole (see Figure 2.3) were not detected i n the present study, probably due to their high reactivity, i n agreement with earlier reports (Turaga et al., 2003; Fang et al. 2005). Table 5.5 also shows the product selectivity obtained from all the metal phosphides used for the H D N o f carbazole. Both the N i . 3 3 M o P / M C M and the Coo.4Ni2P/MCM produced small-hydrogenated B C H X because the 0  B C H X is either almost or completely cracked to products with products o f carbon number less than 12. The high amounts o f cracked products produced by the M C M supports were reported earlier i n the present study with H D S o f 4 , 6 - D M D B T and it is linked to the high acidity  143  Chapter 5  Table 5.5  Activities o f supported metal phosphides for the H D N o f carbazole measured at 583 K and 3.0 M P a H  Catalyst  Total  2  Consumption  Conversion  Product selectivity  Rate  BCHX  CPCHX  THCZ  Other BT  AT  mol %  10 mol.g-'.s-'  Coo.4Ni P/Al 03  82.0  2.63  72.5  5.1  9.6  8.1  2.0  Coo. Ni2P/Al20 -F 3  94.0  4.32  80.5  7.7  7.8  0.1  1.4  Coo. Ni P/MCM  98.0  6.01  9.4  9.8  2.0  41.3  35.0  Ni .33MoP/Al O3  96.0  4.94  82.5  4.7  3.0  6.3  1.7  Nio.3 MoP/MCM  99.9  21.21  13.2  1.1  12.4  43.4  20  2  2  4  4  2  0  2  3  mol %  8  BT: Other products formed with carbon number < 12 AT: Other products formed with carbon number > 12  associated with M C M . Except for the M C M supported catalysts, the Nio.33MoP/Al203 catalyst produced a very high conversion (96 mol%), high B C H X with little cracked products; therefore this catalyst was selected for further examination.  5.7  Kinetics of the hydrodenitrogenation of carbazole As explained previously, the Nio. 3MoP/Al20 was selected based on its performance and 3  3  was further examined at different space velocities. The data were then used to determine the overall or lumped kinetic rate constant, k. 144  Chapter 5 Equation 4.1 is still valid assuming first order kinetics (first order kinetics for carbazole on sulfided catalysts has been reported by Kabe et al. (1991)). Figure 5.8 shows a plot o f the space  0  4  2  6  8  10  12  1/space velocity.h-gcat/mol  Figure 5.8 Plot o f In ( 1 - X A ) versus space time at 533 K , 3.0MPa using N I 0 . 3 3 M 0 P / A I 2 O 3  time (inverse o f the space velocity) versus l n ( l - X ) obtained from data over N I 0 . 3 3 M 0 P / A I 2 O 3 A  and at 533 K and 3.0 M P a , X  A  is the conversion o f carbazole. The correlation gives a good fit o f  R = 0.9988 and the slope is 0.1227. Hence from equation 4.3, the kinetic rate constant is 4.194 x 2  10  ml/hgcat. Based on the reaction network and the products identified as shown in Figure 5.9 (which  is a simplified reaction network o f Figure 2.3) the kinetic rate constants were estimated. The  145  Chapter 5  Figure 5.9  Simplified reaction network o f carbazole  146  Chapter 5 Gaussian Newton-Raphson optimization method was used (as stated earlier i n Chapter 4) to estimate the individual rate constants. Table 5.6 summarizes the results o f the different kinetic rate constants. The kinetic equations used in the simulation are shown in equations 5.1 to 5.3.  dX,  dX  jL-  (5.1)  = r=-k.C.  =  (5.2)  =kC.-kC  r  d K AoJ F  dX fw_\ \  F  =r  =kC. (5.3)  A o J  147  Chapter 5  Table 5.6  Estimated 1 order rate constants for the hydrodenitrogenation o f carbazole at 583 K st  Value x 10 Rate constant  J  (ml/gcat.h)  k,  4.4000 ± 0 . 0 1 2 9  k  0.2570 ± 0.0075  2  The rate constant k i represents hydrogenation o f carbazole to T H C Z and k  2  the C - N  hydrogenolysis o f T H C Z and further hydrogenation o f the product to form B C H X . The value o f k i / k is about 18 and indicates that hydrogenation o f the carbazole is much faster than the C - N 2  hydrogenolysis. Jian et al. (1996) studied the H D N o f ortho-propylaniline over phosphorous promoted N i M o / A l 0 3 at 623 K and 3.0 M P a and reported that there are two kinds o f catalytic 2  sites: the N i - M o - S site (associated with N i ) and the M o site (associated with M o ) . The authors explained that the former was responsible for hydrogenation o f the phenyl group i n the ortho•y  propylaniline. The latter was suited for the C(sp ) - N bond cleavage and it is enhanced by phosphorous. The high hydrogenation rate constant, k i suggests that introduction o f N i into M o P enhanced hydrogenation and subsequent C - N bond cleavage. Direct denitrogenation through hydrogenolysis o f carbazole was not reported contrary to what was observed earlier i n Chapter 4 with the H D S o f 4 , 6 - D M D B T . A s explained earlier in Chapter 2, carbazole has a high C - N bond strength relative to C-S and therefore it is difficult for direct bond cleavage. However, 148  Chapter 5 hydrogenation o f the carbazole ring reduces the C - N bond strength and makes cleavage easier to accomplish. Infact,  this explains why in the  H D N mechanism hydrogenation precedes  hydrogenolysis, which is different from the H D S mechanism. Figure 5.10 shows the plot o f the concentration o f the products and carbazole with the predicted values. The R = 0.9839 indicates good correlation and hence model fits well. The selected N i n j a M o P / A b O s was examined using the carbazole at different temperatures 423, 523 and 583 K to generate the data as plotted in Figure 5.11. The apparent activation energy Ea, was determined using the Arrhenius equation and the value is 59.49 kJ/mol. Szymanska et al.  0.80  0.00  0.10  0.20  0.30  0.40  0.50  0.60  0.70  0.80  Experimental data: Concentration, mol/ml  Figure 5.10 Correlation o f the experimental data versus the predicted from the model for conversion o f carbazole (0) and yields o f products ( B C H X : « ; T H C Z : A )  149  Chapter 5 (2003) studied the H D N o f carbazole using bulk  M02C at different temperatures and reported an  activation energy o f 86.6 kJ/mol. Comparatively, the value obtained in this study was smaller and indicates that it is much easier for the reactants to overcome the small activation energy and react more to increase the conversion.  5.8 Discussion on hydrodenitrogenation of supported phosphides Reports on hydrodenitrogenation o f carbazole over metal phosphides are few. However, previous reports (Satterfield and Cocchetto, 1981) using nitrided  M0/AI2O3 catalysts at 573 K  and 10.1 M P a showed a product distribution that is similar to the present study with no observation o f direct hydrogenolysis products. Figure 5.12 shows plots o f Bransted acidity as a function o f both conversion and B C H X selectivity on all supported catalysts. The conversion increases with Brensted acidity as was observed earlier in the present study in the case o f the H D S o f 4 , 6 - D M D B T . Complete conversion is obtained on the M C M supported catalysts and this is also the catalyst with high Bransted acidity. The B C H X selectivity shows an increase with increased Brensted acidity. However, the B C H X  selectivity decreased  after a maximum  Bransted acidity. It was observed earlier i n the present study that M C M supported catalysts produced cracked products due to the high Bransted acidity.  150  Chapter 5  -1 •  R = 0.9857 2  -2 1  -3 G  -4 H  -5 1.5  1.7  1.9  2.1  1/T, M O O O K "  Figure 5.11  2.3  2.5  1  Plots o f In k versus 1000/T for the hydrodenitrogenation o f carbazole over Nio.33MoP/Al 0 2  3  Summary In summary, the effect o f adding M C M to the selected bulk N i o . M o P leads to enhanced 33  Bronsted acidity and subsequent increased conversion o f carbazole. However, i n terms o f selectivity, the B C H X product was obtained in small quantities when M C M support is used as the high Bronsted acidity leads to cracking o f the products. Fluorination on the AI2O3 support results in moderate acidity on the catalysts and increased conversion, however, the conversion was still less than that obtained on Nio.33MoP/Al203. Although Coo.4Ni2P/Al203 produced high  151  Chapter 5  Bronsted acidity, pmol/m  Figure 5.12  2  Plots o f Bronsted acidity as a function of both conversion and D M B P selectivity on all supported catalysts:Coo.4Ni2P/Al203 ( A ) , Coo.4Ni P/Ai203-F ( • ) , C o . 4 N i P / M C M (•), N i . 3 3 M o P / A l O 2  0  2  0  N i . M o P / M C M (A) 0  3 3  152  2  3  (•)  Chapter 5 selectivity to D M B P that is the direct desulfurisation product o f 4 , 6 - D M D B T , there was no biphenyl product (direct dehydrogenation product from carbazole) observed on this catalyst. In addition, the conversion on this catalyst was low compared to the rest o f the catalysts. Nio.33MoP/Al203 has very high conversion o f carbazole (96 mol%) and also produces less cracked products, hence based on these observations, it is selected to be further tested using real feedstock i n a hydrotreating pilot plant.  153  Chapter 6  Chapter 6 Hydroprocessing using Light Gas Oil (LGO) over supported metal phosphide catalysts Previously, metal phosphides have been tested using model compounds such as 4,6D M D B T , quinoline etc. but it also is important to examine the modified metal phosphides using industrial feed stock. During industrial hydroprocessing, there is interaction among molecules (both S and N containing compounds) resulting in competitive adsorption for active sites. In addition, N-containing compounds inhibit the adsorption o f S compounds on active sites, affecting the H D S activity (Kabe et al., 1999). Therefore, the motivation for testing the selected catalysts, P t - C o o . 4 N i P / A l 0 3 , C o o . 4 N i P / A l 0 3 , and N i . 3 3 M o P / A l O using L G O derived from 2  2  2  2  0  2  3  Athabasca bitumen was to compare the activity with a commercial sulfided catalyst and with the activity data obtained using model compounds. The A l 0 3 ( S A = 240 m / g , P V = 0.65 ml/g) used 2  2  was in the form o f extrudates obtained from Sud Chemical India Limited. The characteristics of the L G O derived from Athabasca bitumen are shown in Table 6.1. The catalyst test procedure with the L G O feedstock has been described elsewhere (Botchwey et al. 2003; Sundaramurthy et al., 2006). The selected catalysts were tested using a trickle bed micro reactor o f inner diameter 10 m m and length 285 m m . Five grams o f the catalyst was diluted with 75 v o l % o f 90 mesh silicon carbide particles to form a bed o f length 12 cm. In order to enhance the radial mixing and prevent axial mixing, silicon carbide particles o f various sizes and glass beads were loaded above and below the catalyst bed. W i t h i n 13 cm o f the reactor length, the temperature profile showed a uniform temperature with ± 2 °C variation, thus ensuring that the 12 cm length o f the catalyst bed was isothermally stable.  154  Chapter 6  Table 6.1  Characteristics o f Light Gas O i l derived from Athabasca bitumen  Boiling range °C  260-315  Sulfur content (ppm)  12664  Nitrogen content (ppm)  166  The catalysts were presulfided for 48 h at 193 °C and 343 °C with 3 v o l % butanethiol i n straight run gas o i l . The flow rate o f the sulfiding solution was 5 ml/h and the H /sulfiding 2  solution ratio was 600 (v/v). Prior to commencing the experiments, catalysts were pretreated with the L G O for a period o f five days at 375 °C, 8.8 M P a , 1 h" L H S V and H / o i l ratio o f 600 1  2  ml/ml. The conditions for testing were 375 °C, 8.8 M P H , and 2 h" L H S V . After three days o f a 1  2  steady decline in catalyst activity, the H D S and H D N activity stabilized by the fourth and fifth days. The experiments were conducted at steady state (after the initial stabilization period) and samples were withdrawn every 24 h. The data presented in this chapter represent the analysis o f the 24 h single point sample. In addition to the selected metal phosphide catalysts, a commercial sulfided N i M o / A l 0 3 2  catalyst was also examined under the same hydroprocessing conditions as the selected catalysts.  6.1  Results and discussion using Light Gas Oil Figures 6.1 and 6.2 show the total S and N conversions o f the catalysts tested using L G O  as feed. Both figures show a similar trend o f increasing conversion with increasing reaction temperature. The sulfur conversion obtained over the temperature range studied (613 - 647 K ) for the C o . N i P / A l O , P t - C o . N i P / A l O 0  4  2  2  3  0  4  2  2  3  and N i  155  0 3 3  M o P / A l O catalysts was 73.3 wt%, 76.1 2  3  Chapter 6 wt%, and 98.6 wt% respectively and the corresponding  N conversions were 58.1 wt%, 67.8  wt%, and 97.1 wt%, respectively. The commercial sulfided N i M o / A l 0 catalyst also showed a 2  3  c o  > C o U  610  620  630  640  650  Temperature, K  Figure 6.1 Total sulfur conversions over selected catalysts using L G O at 613 K , 623 K , 633 K and 648 K . P = 8.8 M P a , L H S V = 2 h" , H to o i l ratio 1  2  = 600 ml/ml: • C o . N i P / A l O 0  4  A sulfidedNiMo/Al 0 2  2  3  2  aNi  0 3 3  156  3  • Pt-Co . Ni P/Al O 0  MoP/Al O 2  4  3  2  2  3  Chapter 6  o >  o U  610  620  630  640  650  Temperature, K Figure 6.2 Total nitrogen conversions over selected catalysts using L G O at 613 K 623 K , 633 K and 648 K . P = 8.8 M P a , L H S V = 2 h" , H to o i l ratio 1  2  = 600 ml/ml: •  • Coo.4Ni P/Al 03 • P t - C o o . N i P / A l 0 2  SulfidedNiMo/Al 0 2  3  2  aNi  4  0 3 3  MoP/Al O 2  2  2  3  3  maximum S conversion o f 94.9 wt% and a maximum N conversion o f 97.1 wt%. Therefore the N i o . 3 M o P / A l 0 catalyst prepared in the present study had higher S but lower N conversion than 3  2  3  the sulfided commercial catalyst, in agreement with previous results on metal phosphides using model compounds (Oyama et al., 2003). B u l k N i  0 3 3  M o P was reduced at a lower temperature in  the presence o f M o . Therefore the enhanced activity is attributed to the improved dispersion o f the P. The C o o . 4 N i P / A l 0 2  2  3  and the P t - C o o . 4 N i P / A l 0 did not show very good conversion 2  2  possibly because the tested catalysts were pretreated by sulfiding, resulting in the loss o f the active phosphide. In the model compound studies reported in Chapters 4 and 5, the metal phosphide catalysts were pretreated in H so that the passivation layer would be re-reduced and 2  157  Chapter 6 provide the needed active sites for hydroprocessing. It is known that commercial C0M0/AI2O3 upon sulfidation can form stable CogSz crystals in the presence o f M0S2. The CooSg crystals are known to be inactive (Farragher and Cossee, 1973) so that i f the C00.4M2P/AI2O3, and the PtCoo.4Ni2P/Al203 catalysts form large amounts o f these CogSg crystals, the conversion w i l l be reduced. O n the other hand commercial NHVI0/AI2O3 forms stable N i S 2 crystals upon sulfidation 3  but these newly formed crystals are known to be active (Farragher and Cossee, 1973).  6.2  Kinetics of hydrodesulfurization and hydrodenitrogenation over selected catalysts. The power law model has been applied to determine the kinetics o f the H D S and  H D N o f L G O (Sundaramurthy et al., 2006) and the apparent rate constants, k and kN were s  calculated using equations 6.1 and 6.2 assuming that H D S follows 1.5 order kinetics and H D N th  first order kinetics respectively. f  k =  k  N  1  LHSV  rr0.5  Q.0.5  (6.1)  (n-\)  = I n f e l W N  (6.2)  P  SF and N F are the S and N concentrations (wt%) i n the feed respectively and S and N are the S p  p  and N concentrations i n the product respectively. L H S V is the liquid hourly space velocity (h" ). 1  Tables 6.2 A and B show the rate constants k and s  calculated for the C00.4M2P/AI2O3,  Pt-Coo.4Ni2P/Ai203 and Nio.33MoPAA.l2O3 catalysts. The data from the present study are compared to data reported by Sundaramurthy et al. (2006) using L G O (S = 1.5 wt% and N = 0.02 wt%) from Athabasca bitumen and tested at similar conditions over a phosphorous doped NiMo/Al203 that was carbided and phosphorous doped NiMo/Al203 that was nitrided. A t 633 K , Table 6.2 shows that the H D S kinetic rate constant over the Nio.33MoP catalyst (14.5 h"'wt%" ) was higher 05  than the carbide (8.3 h w t % " ) , nitride (7.1 h" wt%" ' ) and the commercial sulfided catalyst 158 _1  05  1  0 5  Chapter 6 (9.1 K vA%~° ). l  Similarly, at 633 K , the H D N kinetic rate constant over N i o . M o P catalyst (6.2  5  33  h" ) was higher than the carbide (3.7 h" ), nitride (3.4 h" ) and the commercial sulfided catalyst 1  1  1  (3.3 h" ). However at 648 K , the H D N kinetic rate constant was comparable to the commercial 1  catalyst and these results are in agreement with previously reported data (Stinner et al., 2001). The H D S kinetic rate constants for the Coo.4Ni2P/Al20 , and Pt-Coo.4Ni2P/Al20 are shown i n 3  Table 6.2 and the values range from 0.8 - 3.3 h"'wt%"  3  and 0.6 - 3.7 K vrt%~  05  l  05  respectively  while the H D N kinetic rate constants range from 0.2 - 1.7 h" and 0.2 - 2.3 h" , respectively, at 1  1  the reaction temperature range o f 613 - 648 K . Figure 6.3 (A) and (B) show plots o f the Arrhenius equation for the data obtained on the N i n . M o P / A l 2 0 catalyst prepared in the present study and the commercial sulfided catalyst. 33  3  Table 6.2 A  Apparent kinetic parameters for H D S o f Light Gas O i l at different temperatures  k (h wt%_1  s  Temp K  Nio.33MoP/Al 0  613  9.3  0.1  623  10.4  633 648  Co .4Ni P/Al O  Pt-Coo . N i P / A l 0  Carbide*  Nitride*  Sulfide  0.6  4.7  3.9  5.9  1.2  0.9  6.0  5.4  8.8  14.5  1.6  1.3  8.3  7.1  9.1  27.0  3.3  3.7  -  -  -  2  3  0  2  2  3  4  2  2  3  * Phosphorous doped NiMo/y-Al 0 : Carbide and Nitride; Sulfide : Commercial catalyst 2  3  159  Chapter 6 Table 6.2 B  Apparent kinetic parameters for H D N o f Light Gas O i l at different temperatures  Temp K  k Ni  0 3 3  MoP/Al O 2  3  Coo. Ni P/Al 0 4  2  2  3  (h" ) 1  N  Pt-Co .4Ni P/Al O 0  2  2  3  Carbide*  Nitride*  Sulfide  613  3.1  0.2  0.2  2.2  2.0  1.7  623  4.6  0.4  0.3  2.9  2.5  2.7  633  6.2  0.6  0.4  3.7  3.4  3.3  648  7.1  1.7  2.3  Table  6.3  shows  the  activation  energy  _  determined  from  Figures  6.3  for  the  Nio.33MoP/Al203 and the sulfided catalyst. Arrhenius plots similar to Figures 6.3 were also constructed for the Coo.4Ni2P/Al203, and the Pt-Coo.4Ni2P/Al20 catalysts (see Appendix B ) and 3  the apparent activation energies are reported in Table 6.3. A l s o included i n Table 6.3 are the apparent activation energies obtained from testing Pt-Con.4Ni2P/Ai203 and Nin. 3MoP/Al203 3  using 4 , 6 - D M D B T and carbazole respectively (see Chapters 4 and 5). Comparatively, the H D S apparent activation energy over Pt-Coo.4Ni2P/Ai203 using the L G O derived from Athabasca bitumen was 193 kJ/mol while using 4 , 6 - D M D B T it was only 65 kJ/mol. The apparent activation energy includes both the surface reaction activation energy and the heat o f adsorption  (see  equation 4.1). The low apparent activation energy obtained using the model compound signifies that there are no molecular interactions and so the molecules adsorb on active sites. Note that the activation energy comprises the surface activation energy and the heat o f adsorption. A similar  160  Chapter 6  Figure 6.3  Arrhenius plots for determining the activation energy for the H D S (A) and H D N (B) o f L G O over ( • ) Ni .33MoP/Al O3 and 0  (•)commercial sulfided catalysts  161  2  Chapter 6 Table 6.3  Comparison o f apparent activation energies for H D S and H D N o f L G O overNi .33MoP/Al O3, Coo.4Ni P/Al 03, Pt-Co .4Ni P/Al O and 0  2  2  2  0  2  2  3  commercial sulfided catalyst Commercial Activation  Nio. MoP/Al 03 33  2  Coo.4Ni P/Al 0 , 2  2  3  Pt-Coo. Ni P/Al 0 4  2  2  3  Sulfide  Energy,(kJ/mol) HDS  103  138  193  62  HDN  78  177  112  130  -  -  65  -  60  -  -  -  4,6-DMDBT Carbazole  trend is shown by the H D N activation energy over Nio.33MoPAA.bO3. Using the L G O , the activation energy was 78 kJ/mol whereas using carbazole the activation energy was 60 kJ/mol. The Nio.33MoP/Al 03 and the Coo.4oNi P/Al 03 catalysts underwent the exact same pre2  2  2  sulphiding and stabilization treatments as the conventional N i M o S / A l 0 3 catalyst. 2  This  approach was taken to ensure that a comparison under the same process conditions could be made. In addition, the pre-sulphiding o f the metal phosphides represents a severe test o f their ability to remain in the phosphide form, rather than be converted to a metal sulphide, during reaction.  In order to determine the state o f the used catalysts, therefore, the catalysts were  recovered after reaction and were analysed following extraction i n tetrahydrofuran to remove soluble coke material. The bulk chemical composition o f the catalysts, determined before and after reaction, is given in Table 6.4, whereas Table 6.5 reports the surface composition as  162  Chapter 6 amounts o f S after reaction i n L G O . The S content was lower on the Nio 33M0P/AI2O3 catalyst than the Coo.4oNi2P/Al203 catalyst, which had a lower activity than the Nio.33MoP/Al203 catalyst. These observations are consistent with the X R D data o f Figure 6.4, showing that the  C00.40NI2P/AI2O3 catalyst probably formed a metal phospho-sulphide, whereas no change in the bulk properties o f the Nio.33MoP/Ai203 catalyst were apparent from the X R D data. Note that the conventional  NiMoS/Al20  3  catalyst also contained a significant amount o f P and had the  highest S content following reaction among the catalysts tested.  These results clearly  demonstrate that the oxide precursor o f the conventional catalyst was readily sulphided, despite the presence  o f P, whereas the metal phosphides were resistant  to sulphidation, the  Nio.33MoP/Al 03 catalyst more so than the Con.4oNi2P/Al203 catalyst. The state o f the P i n the 2  conventional catalyst is clearly different to that o f the metal phosphide.  Table 6.4 Measured catalyst atom ratios before and after reaction in L G O Co  Ni  Mo  p  S  atom ratio * Coo.4Ni P/Ai203 - after T P R  0.20  1.00  -  0.55  -  Coo.4Ni2P/Al 0 - after reaction in L G O  0.18  1.00  -  0.59  0.67  M0.3M0P/AI2O3 - after T P R  -  0.33  1.00  0.51  -  Nio.3MoP/Al 03 - after reaction in L G O  -  0.30  1.00  0.21  0.03  MM0S/AI2O3 - calcined oxide precursor  -  0.37  1.00  0.35  -  MM0S/AI2O3 - after reaction in L G O  -  0.40  1.00  0.77  2.23  2  2  3  2  * atom ratio relative to M o or N i as determined by I C P of the supported catalysts  163  Chapter 6 Table 6.5 A t o m ratios determined by X P S o f supported metal phosphide catalysts before and after reaction i n L G O  P/M Before rxn  M/Al After rxn  Before rxn  S/M  After rxn  After r x n  atom ratio* Coo.4Ni P/Ai203 2  Nio. MoP/Al 0 3  2  3  "047  21)2  61)4  (X02  5.00  1.11  0.61  0.10  0.16  0.30  The data o f Table 6.5 report the results o f X P S analysis o f the C00.40NL2P/AI2O3 and the Nio 3 M o P / A l 0 3 catalysts before and after reaction with L G O . 3  Before reaction, the P / M ratio  2  ( M = C o + N i or N i + M o ) indicated someP enrichment o f the catalyst surface and similar observations were made on bulk N i M o P and C o N i P catalysts ( A b u and Smith, 2006, A b u and x  x  2  Smith, 2006). The S / M ratio measured after reaction was high for the Coo.4oNi P/Al 03 catalyst 2  2  and for both catalysts the S / M ratio determined by X P S was much greater than that determined from the bulk chemical analysis o f the same catalysts (Table 6.4). Furthermore, Table 6.5 shows that after reaction, the S / M ratio was less than the P / M ratio on the Nio.33MoP/Al 03 catalysts, 2  whereas the opposite was true for the Coo.4oNi P/Al203 catalyst, indicative o f the former catalyst 2  having a higher resistance to sulphidation than the Coo.4oNi P/Al 03 catalyst. The partial 2  164  2  Chapter 6  B  After rxn  I  3  ro w CD  Before rxn  i Before rxn  V  L MoP  30  40  50  60  30  40  2 Theta  Figure 6.4  50  60  2 Theta  Comparison o f X R D diffractograms obtained for N i 3 M o P supported on A 1 0 and 0 3  AI2O3-F, before and after reaction in L G O .  165  2  3  Chapter 6 sulfidation o f the metal phosphides is also evident from the X P S N i 2p and M o 3d spectra o f Figure 6.5. The B E for N i 2p and M o 3d o f the catalysts before reaction (i.e. following reduction and passivation) are in agreement with those measured on the corresponding bulk catalysts ( A b u and Smith, 2006, A b u and Smith, 2007). The presence o f M o P is evidenced by the M o 3d B E o f 228.2 e V and the presence o f N i P by the N i 2p peak at B E o f 853.7 e V . Following reaction, 2  there is an increase in the B E o f the M o 3d peak at 229.5 e V , consistent with the formation o f M o S , whereas the intensity o f the N i 2p peak at 853.7 e V is significantly reduced. Finally we 2  note that the M / A l ratio determined by X P S suggests a higher dispersion o f the metal phosphide on the Ni .33MoP/Al2O3 catalyst than the Coo.4oNi2P/Al2C>3 catalysts and this may be part o f the 0  reason for the lower activity observed for the latter catalyst. In summary, the Nio.3 MoP/Al 03 catalyst prepared in the present study produced higher 3  2  H D S and H D N conversions than commercial sulfided N i M o / A b C ^ catalysts and comparatively also produced higher activity than carbides and nitrides reported elsewhere. The H D S and H D N activation energies obtained on the selected metal phosphides were lower using model compounds than using the L G O derived from Athabasca bitumen, implying that the model compounds undergo reaction more easily probably because there is an absence o f competitive adsorption with other molecules.  166  Chapter 6  Figure 6.5 X P S o f N i 2p and M o 3d region for C00.4N12P/AI2O3 and Ni0.33MoP/Al2O3 catalysts before and after reaction with L G O .  167  Chapter 7  Chapter 7 Conclusions and Recommendations 7.1  Conclusions The preparation, characterization and activity o f modified transition metal phosphides has  been investigated for the H D S o f 4 , 6 - D M D B T . The addition o f 3 - 24 wt% o f C o to bulk N i P 2  indicated that the catalyst containing 3 wt% o f C o (C00.08N12P), produced the highest selectivity to the direct desulfurization ( D D S ) product dimethylbiphenyl ( D M B P ) when tested with the refractory model-sulphur compound 4,6-dimethyldibenzothiophene ( 4 , 6 - D M D B T ) . However, the activity o f NI2P using the same 4 , 6 - D M D B T model compound showed high selectivity to the hydrogenation product bicyclohexane ( B C H X ) . Therefore it was concluded that the presence o f Co enhanced the selectivity to the direct desulfurization product. The correlation o f the surface phosphorous/metal ( P / M ) ratio determined by X P S and the ratio o f n-propylamine uptake to C O uptake determined by adsorption indicated that the surfaces o f the Co Ni2P have Bronsted acidity x  that isomerizes the methyl groups to less sterically hindered positions and enhances direct desulfurization o f 4 , 6 - D M D B T . Coo.4Ni2P was supported on AI2O3, weakly acidic AI2O3, fluorinated AI2O3 and M C M and tested again using 4 , 6 - D M D B T . It was found that conversion was almost complete over the M C M supported catalyst but this M C M supported catalyst also produced a very high proportion of cracked products. The high cracking on the M C M support was ascribed to the high Bransted acidity. Consistent with earlier reports using sulfided catalyst with fluorinated A l 0 3 , the 2  Coo.4Ni P on fluorinated AI2O3 produced a modestly higher conversion than the neutral AI2O3 2  support. Pt was also added to the Coo.4Ni P to examine the possible enhancement 2  o f the  hydrogenation route o f the 4 , 6 - D M D B T . The increase in conversion and hydrogenation products 168  Chapter 7 compared to the non-platinated Coo.4Ni P/Ai203 was due to the increased hydrogenation due to 2  the presence o f the Pt. The Pt-Coo. Ni2P/Al203-WA showed decreased conversion with more 4  cracked products than the neutral AI2O3 support. H D N o f carbazole over bulk N i M o P and Nio. 3MoP on different supports were also x  3  studied and i n agreement with previous reports no direct biphenyl product was obtained. Nio.33MoP/Ai203 showed a conversion o f 96 m o l % and high B C H X product (83 mol%) while Nio.33MoP/MCM showed 100 m o l % conversion but with high cracked products and little B C H X (13.2 mol%) products as a result o f high Bronsted acidity. Correlation o f conversion and B C H X selectivity with the Bronsted acidity suggests that increased Bronsted acidity promotes conversion o f carbazole but that excess Bronsted acidity leads to decreased B C H X selectivity as products are cracked. The Nio.33MoP/Al2C>3 catalyst prepared in the present study produced higher H D S (94.9 wt%) and H D N (78.9 wt%) conversions than the commercial sulfided MM0/AI2O3 catalysts ( H D S = 78.9 wt%, H D N = 56.7 wt%) at 613 K and comparatively also produced higher activity than the carbides ( H D S = 82 wt%, H D N = 63 wt%) and nitrides ( H D S = 81 wt%, H D N = 61 wt%) reported elsewhere when tested in L G O derived from Athabasca bitumen. The activation energy obtained using 4 , 6 - D M D B T over the selected Pt-Coo.4Ni2P/Al203, was smaller (65 kJ/mol) than that obtained (192.9 kJ/mol) using L G O derived from Athabasca butimen. A ratio o f the kinetic rate constant o f the direct desulfurization to the hydrogenation obtained using 4 , 6 - D M D B T was found to be high (k\/k2 = 2 1 ) suggesting that the C-S bond cleavage occurred much faster than hydrogenation on the acidic surface. One major contribution o f the present study is that addition o f small quantities o f C o to M2P enhances the D D S route o f the H D S o f 4 , 6 - D M B T subsequently increasing the D M B P product selectivity. Consequently, modifying metal phosphides by incorporating intermediate 169  Chapter 7 Bransted acidity was desirable as isomerisation o f 4 , 6 - D M D B T is enhanced leading to observed D D S selectivity. Another major contribution o f the present study is that the kinetics o f the H D N o f carbazole on metal phosphides is now reported for the first time. Prior to the present study, there was no report o f H D N o f carbazole over metal phosphides. Previously all reported activity data including conversion, H D S and H D N kinetic rate constants on metal phosphides were obtained using model compounds. The activity data using L G O derived from Athabasca bitumen has never been reported. This is a major contribution since the present thesis provides literature with data o f industrial significance on the activity o f metal phosphides using industrial feed ( L G O ) . The catalyst life has not been evaluated because such evaluation needs extended length o f time and the present research cannot accomplish that within the limited time period.  7.2  Recommendations Some future research directions and recommendations can be studied to improve H D S  and H D S o f modified metal phosphides. •  Increasing the surface area o f bulk metal phosphides  The B E T surface area o f bulk C o o . o s ^ P and Nin.07MoP were small ( < 10 m /g). It w i l l 2  be desirable to increase the activity o f the bulk by preparing bulk metal phosphides with high B E T surface area ( > 100 m /g). Recently, Y a n g et al. (2006) reported the preparation o f bulk 2  Ni P 2  with high surface  area (130 m /g) using a surfactant-assisted  method-addition o f  polyethylene glycol tert-octylphenyl ether and ethylene glycol (Yang et al., 2006). •  Nanocrystal metal  phosphides  can  relationships.  170  also  be  used  to  study  structure-activity  Chapter 7 •  Incorporate other sources o f Bronsted acidity.  In this study, weakly acidic AI2O3 and M C M were used to enhance Bronsted acidity. Other supports such as silica-alumina with moderate Bronsted acidity should be exploited to incorporate on metal phosphides i n order to increase the conversion and D D S selectivity o f the 4,6-DMDBT. •  Zeolites could also be used as supports for the prepared modified phosphides since they have large pores that can enhance reactivity o f the S and N containing molecules.  •  Effect o f H S 2  The effect o f H S and S on the modified metal phosphides should be studied at longer 2  times using a pilot plant as this study was limited to about 14 h time-on-stream. •  Effects o f mixed reactants  The model compounds study did not take into consideration the possibility o f interaction o f other reactants in the feed. Therefore the effects o f other sulfur or nitrogen containing model compounds in the feed should be investigated.  171  References A b u , I. I., Smith, K . J. The effect o f cobalt addition to bulk M o P and N i P catalysts for the 2  hydrodesulfurization o f 4,6-dimethyldibenzothiophene. Journal of Catalysis, 241, 356366 (2006). A b u , 1.1., Smith, K . J . Hydrodenitrogenation o f carbazole over a series o f bulk N i x M o P catalysts. Catalysis Today, In press. Bataille, F., Lemberton, J. L . , Perot, G . , Leyrit, P., Cseri, T., Marchal, N . , Kasztelan, S. Sulfided M o and C o M o supportd on zeolite as hydrodesulfurization catalysts: transformation o f dibenzothiophene and 4,6-dimethyldibnzothiophene. Applied Catalysis A: General,  220,  191-205 (2001). Baum, E . , J. (1998) Chemical Property Estimation:Theory and Application. C R C Press. Benitez, A . , Ramirez, J., Fierro, J. L . G . , Agudo, L . A . Effect o f fluorine on the structure and activity o f N i W / A l 0 3 catalysts for H D S o f thiophene and H D N o f pyridine. 2  Applied Catalysis A: General, 144, 343-364 (1996). Benyamna, A . , Bennouna, C , Leglise, J., van Gestel, J., Duyme, M . , Duchet, J- C , Dutartre, R., Moreau, C . Effect o f the association o f C o and N i on the catalytic properties o f sulfided C o N i M o / A l 0 3 . American. 2  Chemical Society, Divison of Petroleum  Chemistry  43, 39-42 (1998). Botchwey, C , Dalai, A . , K . , Adjaye, J. Product selectivity during hydrotreating and m i l d hydrocracking o f bitumen-derived gas oil. Energy & Fuels, 17, 1372-1381 (2003). Breysse, M . , Djega-Mariadassou, G , Pessayre, S., Geantet, C , Vrinat, M . , Perot, G . , Lemaire, M . Deep desulfurization: reactions, catalysts and technological challenges. Catalysis Today, 84, 129-138 (2003). Candia, R., S0rensen, V . , Villadsen, J., Topsoe, N . Y . , Clausen, B . S., Tops0e, H . The structure  of cobalt sulfide phase i n carbon supported and C o - M o sulfide catalysts as studied studied by exafs and xanes. Bulletin Society of Chimistry,  Belgrade,  93, 763-772 (1984). Carrier, X . , Che, M . Comment on: N i and M o interaction with Al-containing M C M - 4 1 support and its effect on the catalytic behavior in D B T hydodesulfurization. Applied Catalysis A: General, 253, 317-320 (2003). Clark, P. A . , Oyama, S. T. Alumina-supported molybdenum phosphide hydroprocessing catalysts. Journal of Catalysis, 218 78-87 (2003). Clark, P . A . , Wang, X . , Deck, P., Oyama, S. T. Push-pull mechanism o f hydrodenitrogenation over silica-supported M o P , W P and  M0S2 hydroprocessing catalysts. Journal of Catalysis,  210, 116-126 (2002). Clark, P., L i , W . , Oyama, S. T. Synthesis and activity o f a new catalyst for hydroprocessing: Tungsten phosphide. Journal of Catalysis, 200, 140-147 (2001). Colling, C . W . , Thompson, L . T. The structure and function o f supported molybdenum nitride hydrodenitrogenation catalysts. Journal of Catalysis, 146, 193-203 (1994). Corbridge, D . E . C , Studies in inorganic Chmistry 2: Phosphorous an Outline o f its Chemistry, Biochemistry and Technology, Elsevier, Scientific Publishing Company, Amsterdam, 1980. Corma,  A . , Martinez, A . , Martinez-Soria, V . Hydrogenation o f aromatics in diesel fuels on P t / M C M - 4 1 catalysts. Journal of Catalysis, 169,480-489 (1997).  Dautzenberg, F. M . Characterization and Catalyst Development. ACS Symposium Series, Chapter 11, 196 A C S Meeting, Los Angelos California, 99-119 (1998). th  Dineen, G . U . , Cook, G . L . , Jensen, H . B . Estimation o f types o f nitrogen compounds i n shale-oil gas o i l . Analytical Chemistry, 30, 2026-2030 (1985).  Ding, L . , Zhang, Z . , Zhang, Y . , Ring, Z . , Chen, J. Effect o f fluorine and boron modifications on the H D S , H D N and H A D activity o f hydrotreating catalysts. Applied  Catalysis  A:  General, 301, 241-250 (2006). E P A - D i e s e l R I A , Regulatory Impact Analysis: Heavy Duty Engines and Vehicles Standards and Highway Diesel Fuel Sulfur Control Requirements,  United States Environmental  Potection Agency, A i r and Radiation, EPA420_R-00-026, December (2000). Euzen, P., Raybaud, P., Krokidis, X . , Toulhaot, H . , L e Loarer, J. L . , Jolivet, J. P., Fridefond, C , in: Schuth, F., Sing, K . S., Weitkamp, J. (Eds.), Handbook o f Porous Solids, v o l . 3, (2002). Fang, K . , Ren, J., Sun, Y . Effect o f nickel precursors on the performance o f N i / A 1 M C M - 4 1 catalysts for n-dodecane hydroconversion. Journal of Molecular  Catalysis. A 229, 51-58  (2005). Farneth, W . E . , Gorte, R. J., Methods o f characterizing Zeolite Acidity. Chemical Review 95, 615-631 (1995). Farragher, A . L . , and Cossee, P., in: Proceedings o f the 5 International Congress on Catalysis E d . th  Hightower, J . W . , V o l . 2, North-Holland, Amsterdam, 1301 (1973). Fierro, J. L . G . , Cuevas, R., Ramirez, J., Agudo, L . A . , Activity and selectivity trends o f F modified N i M o catalysts in various hydrotreating ( H D S / H D N , H Y D ) reactions.  Bulletin  Society Chimi Belgrade, 100, (1991) 945. Fogler, S.H., "Elements o f Chemical Reaction Engineering", 3  rd  ed., M c G r a w - H i l l , N e w  York, (1981). Fox, M . A . , Whitesell, J. K , Organic Chemistry. Jones and Barlett Publishers, Boston, (1994). Froment, G . F., Bischoff, K . B . , Chemical Reactor Analysis and Design. Wiley, N e w Y o r k (1990).  Frankman, Z . , Ignasiak, T. M , Montgomery, D . S., Strauz, O.P., Nitrogen compounds in Athabasca asphaltenes: The carbazoles. AOSTRA Journal of Research 3 131-138 (1987). Fluck, E . , Weber, D . , P 2p binding energies i n phosphorous(III) compounds, phosphonium salts and oxyacids o f phosphorous. Z . Naturfomsch,  B29, 603-603 (1974).  Fujikawa, T., Kimura, H . , Kiriyama, K . , Hagiwara, K . Development o f ultr-deep H D S catalysts for production o f clean diesel fuels. Catalysis Today, 111, 188-193 (2006). Gallezot, P., Dakta, J., Marssardier, J., Primat, M . , Imelik, B . , in: Proceedings o f the Sixth International Congress on Catalysis, London, 2, 696 (1977). Gary, J. H . , Handwerk, G . E . , Petroleum Refinery Technology and Economics, Marcel Dekker, N e w york, 465 (1994). Geankoplis, C . J., "Transport Processes and Unit Operations." PrnticeHall, N e w Jersey. Louwers, (1993). Haji, S., Zhang, Y . , Kang, D . , Aindow, M . , Erkey, C . Hydrodesulfurization o f model diesel using P t / A ^ O s catalysts prepared by supercritical deposition. Catalysis  Today, 99, 365-371  (2005). Hanlon, R. T. Effects o f  P H 2 S , PH2  and  PH2S/PH2  on the hydrodenitrogenation o f pyridine. Energy  & Fuels, 1,424-430 (1987). Harris, S. and Chianelli, R. R. Catalysis by transition metal sulfides: A theoretical and experimental study o f the relation between the synergic systems and the binary transition metal sulfides. Journal of Catalysis, 98, 17-31 (1986). Harvey, T. G . , Matheson, T. W . , Pratt, K . C , Stanborough, M . S. Studies o f the batch hydrotreatment o f Rundle shale oil. Fuel, 64, 925-930 (1985).  175  H o , C . T. Deep H D S o f diesel fuel: Chemistry and catalysis. Catalysis Today, 98, 3-18 (2004). Houlla, M . and Delmon, B . Use o f X P S to detect variations i n dispersion o f impregnated and ion-exchanged N i O / S i 0 2 systems. Surface nd Interface Analysis, 3,103-105 (1991). Houalla, M . , Broderic, D . H . , Sapre, A . V . , Nag, N . K . , de Beer, V . H . V . , Gates, B . , Kwart, H . Hydrodesulfurization o f methyl-substituted dibenzothiophenes catalyzed by sulfided C o M o / y - A l 0 . Journal of Catalysis, 61, 523-527 (1980). 2  3  Homma, T., Echard, M . , Leglise, J. Investigation o f C o N i M o / A l 0 3 catalysts: Relationships 2  between H S adsorption and H D S activity. Catalysis Today, 106, 238-242 (2005). 2  Hsu,  C . S., Genowitz, M . W . , Dechert, G . J., Abbott, D . J . , Barbour, R., Molecular characterization o f diesel fuels by modern analytical techniques, in: Song, C , Hsu, S., Mochida, I., (Eds), Chemistry o f Diesel Fuels, Taylor & Francis, N e w Y o r k , (2000).  Hughes, T. R., White, H . M . , White, R. J. Bronsted and Lewis acid site concentrations i n fluorinated alumina from the infrared spectra o f adsorbed pyridine species. Journal  of  Catalysis, 13, 58 -64 (1969). Isoda, T., Takase, Y . , Kusakabe, K . , Morooka, S. Changes i n desulfurization reactivity o f 4,6Dimethyldibnzothiophene by skeletal isomerization using Ni-supported Y-type zeolite. Energy & Fuels, 14, 585-590 (2000) Isoda, T., Nagao, S., Korai, Y . , Mochida, I., 4,6 dimethyl-dibenzothiophene hydrodesulfurization reaction mechanism 1. N i W based catalyst behavior. American Preparation  Chemical  Society,  Divivison of Petroleum Chemistry, 41, 559-564 (1996).  Jian, M . , and Prins, R. Mechanism o f the hydodenitrogenation o f quinoline over N i M o ( P ) / A l 0 3 2  catalyst. Journal of Catalysis, 179, 18-27 (2002).  176  Jian, M . , Kapteijin, A . , Prins, R. Kinetics o f hydrodenitogenation o f ortho-propylanilne on N i M o ( P ) / A l 0 catalysts. Journal of Catalysis, 168,491-500 (1997). 2  3  Jian, M . and P r i n s / R . Existence o f different catalytic sites in H D N catalysts. Catalysis Taday, 30, 127-134(1996). Jokuty, P. L . and Gray, M . R. Nitrogen bases resistant to hydrodenitrogenation: evidence against using quinoline as a model compound. Industrial Engineering  Chemistry Research,  31,  1445-1449 (1992). Jokuty, P. L . and Gray, M . R. Resistant nitrogen compounds in hydrotreated gas o i l from Athabasca. Energy & Fuels, 5, 791 -795 (1991). Kabe, T., Ishihara, A . , Qian, W . "Hydrodesulfurization and Hydrode-nitrogenation", W i l e y - V C H , Tokyo, Japan, (1999). Kabe, T., Ishiharam, A . , Tajima, H . Hydrodesulfurization o f sulfur-containing polyaromatic compounds in Light O i l . Industrial & Engineering  Chemistry Research, 31, 1577-1580  (1992). Katzer, J. R., Sivasubramanian, R. Process and catalyst needs for hydrodenitrogenation. Catalysis Review Science & Engineering, 20, 2 155-208. (1979). Kasztelan, S., Benazzi, E . , Geaoge-Marchal, N . Hydrocracking o f catalysts comprised o f beta zeolite and a group V B element. US patent 6 171 474 (2001). Kherbeche, A . , Hubaut, R., Bonnelle, J. P., Grimblot, J. Dual mechanism o f Pyridine hydrodenitogenation on MoSx-based catalysts: Experimental evidence. Journal Catalysis, 131,204-214(1991).  177  of  Khwaja, E . E . , Salim, M . A . , Khan, M . A . , A l - A d e l , F. F., Khattak, G . D . , Hussain, Z . Journal Non-crystalline  Solids, 33,110 (1989).  Kilanowski, D . , Teeuwen, H . , de Beer, V . H . J., Gate, B . C , Shuit, G . C . A . , Kwart, H . Hydrodesulfurization  o f thiophene,  benzothiophene,  dibenzothiophene,  and  related  compounds catalyzed by sulfided C o O — M 0 O 3 / Y-AI2O3: Low-pressure reactivity studies. Journal of Catalysis, 55, 129-137(1978). K i m , H . , Lee, J. J., M o o n , S. H . Hydrodesulfurization o f dibenzothiophene compounds using fluorinated NiMo/Al203 catalysts. Applied  Catalysis  B: Environmental,  44, 287-299  (2003). K l e i n , J. C , Hercules, D . M . Surface characterization o f model Urushibara catalysts. Journal of Catalysis, 82, 424-441 (1983). K w a k , C , K i m , M . Y . , Choi, K . , Sang, H . P. Effect o f phosphorus addition on the behavior o f  C0M0S/AI2O3  catalyst  in  hydrodesulfurization  of  dibenzothiophene  and  4,6-  dimethldibenzothiophene. Applied Catalysis A: General, 185, 19-27 (1999). Laine, J., Severino, F., Golding, R. Comparative study o f the activity o f cobalt and nickel as promoters o f H D S catalysts. Journal of Chemical Technology  &. Biotechnology,  34A,  387-397(1984). Landau, M . V . , Berger, M . , Herskowitz, M . Hydrodesulfurization o f methyl-substiuted dibenzothiophenes: Fundamental study o f routes to deep desulfurization. Journal Catalysis, 158, 236-245 (1996). Landau, M . V . Deep hydrotreating o f middle distillate from crude and shale oils. Catalysis Today, 36,393-429(1997).  178  of  Lappas, A . A . , Budistenau, R . , Drakak, K . , Vasalos, I. A . Production o f low aromatics and low sulfur diesel i n a hydrodesulfurization ( H D S ) pilot plant unit. Global Nest: The International Journal, 1, 15-22 (1999). Ledoux, M . J., Andre., J., Blekkan, E . A . , Luck, F . The role o f the nature and the purity o f the alumina support on the hydrodesulfurization activity o f C o M o sulfides. Applied Catalysis A: General, 133, 321-333 (1995). Lee, J. J., K i m , H . , K o h , J. H . , Jo, A . , M o o n , S. H . Performance o f fluorine-added C0M0S/AI2O3 prepared by sonochemical and chemical vapor deposition methods in the hydrodesulfurization o f dibenzothiophene and 4,6-dimethyldibenzothiophene. Applied Catalysis Environmental, 61, 274-280 (2005). Leon, R., Margolese, D . , Stucky, G . , Petroff, P. M . Nanocrystalline Ge filaments i n the pores o f a mesosilicate. Physical Review B, 52, 4 2285-2288 (1995). Lewandowski, M . , Sarbak, Z . Acid-base properties and hydrofining activity o f N i M o catalysts incorporated on alumina modified with F" and CI". Applied Catalyis A: General, 156, 181-192 (1997). L i , F., Zhao, Z . , L i , Q., Zhao, T., L i , C , Sun, G . Effect o f reduction temperature on hydrofining performance  o f supported  molybdenum phosphide  catalyst. Journal Natural Gas  Chemistry, 14, 233-237 (2005). L i , Q., B r o w n , S., E . , Broadbelt, L . , J., Zeng, J., W u . , N . , Q . Synthesis and characterization o f MCM-41-supportd Ba2Si04 base catalyst. Microporous and Mesoporous Materials, 59, 105-111 (2003). L i a w , S. J., Raje, A . , B i , X . X . , Eklund, P. C , Geaham, D . M . , Davies, B . H . Hydrotreatment o f naphtha with molybdenum nitride catalysts. Energy & Fuels, 9, 921-927 (1995).  179  M a , D . , X i a o , T., X i e , S., Zhou, W . , Gonzalez-Cortes, S. L . , Green, M . L . H . Synthesis and structure o f bimetallic nickel molybdenum phosphide solid solution. Chemical Material, 16, 2697-2699 (2004). M a , X . , Sakanishi, K . , Mochida, I. Hydrodesulfurization reactivities o f various sulfur compounds in diesel fuel. Industrial Engineering Chemistry Research, 33, 218-222 (1994). Majolsky, T. W . , Montgomery, M . S., and Strauz, O.P. The basic nitrogen compounds i n Athabasca maltene. AOSTRA Journal of Research 3, 25-35 (1986). M a n o l i , J . - N . , D a Costa, P., Brun, M . , Vrinat, M . , Mauge, F . , Potvin, C . Hydrodesulfurization o f 4,6-dimethyldibenzothiophene over promoted (Ni,P) alumina-supported molybdenum carbide catalysts: activity and characterization o f active sites. Journal of Catalysis, 221, 365-377 (2004). Massoth, F . E . Characterization o f molybdena catalysts. Advances in Catalysis, 27, 266310(1978). Mague, F . , Lamotte, J., Nesterenko, O., Tsyganenko, A . A . F T - I R study o f surface properties o f unsupported M o S . Catalysis Today, 70, 271-284 (2001). 2  M a y o , S., Brevoord, E . , Gerritsen, L . , Plantenga, F. Process ultra-low sulfur diesel. Hydrocarbon Process 80 2, 84A (2001). M a x w e l l , I. E . Innovation i n applied catalysis. Cattech 1 5-13 (1997). Mcllvried, H . G . Kinetics o f hydrodenitrification o f pyridine. Industrial Engineering Chemistry Process Development, 10 ,125-130 (1971). Mears, D . E . Tests for transport limitations i n experimental catalytic reactions. Industrial Engineering Chemistry Process Development, 10 541-547 (1971). M i l l e r , J . T., Hinemann, M . F . Non-first-order hydrodenitrogenation kinetics o f quinoline Journal of Catalysis, 85, 117-126 (1984).  Minderhoud, J . K . , van Veen J. A . R . First Stage Hydrocracking: Process and Catalytic Aspects. Fuel Processing Technology. 35 87 (1993). Mizutani, H . , Godo, H . , Ohsaki, T., Kato, Y . , Fujikawa, T., Saih, Y . , Funamoto, T., Segawa, K . Inhibition effects o f nitrogen compounds on C0M0P/AI2O3 cataysts with alkali or zeolite added i n hydrodesulfurization o f dibenzothiophene and 4,6-dimethyldibenzothiophene. Applied Catalysis A: General, 295, 193-200 (2005). Montgomery, D . S . Department o f Mines and Technical Surveys Information Circular, IC 169 Sept. (1964). Mosio-Mosiewski, J., Morawski, I. Study on single-stage hydrocracking o f vacuum residue i n the suspension o f N i - M o catalyst. Applied Catalysis A: General, 283, 147-155 (2005). Nagai, M . , Fukiage, T., Kurata, S. Hydrodesulfurization o f dibenzothiophene over aluminasupported nickel molybdenum phosphide catalysts. Catalysis Today, 106, 201-205 (2005). Nagai, M . , Goto, Y . , Irisawa, A . , O m i , S. Catalytic activity and surface properties o f nitrided molybdena-alumina for carbazole hydrodentrogenation. Journal of Catalysis. 191, 128137(2000). Nagai, M . , Goto, Y . , Kiyoshi, M . , Hada, K . , Oshikawa, K . , O m i , S. Temperature -programmed reduction and X R D studies o f ammonia-treated molybdenum oxide and its activity for carbazole hydrodenitrogenation. Journal of Catalysis, 182, 292-301 (1999). Nagai, M . , Koyama, H . , Sokomoto, S., O m i , S., in: Hydrotreatment and Hydrocracking o f O i l fractions. Eds. Delmon, B . , Froment, G . , Grauge, E . , Elsevier, Amsterdam 195 (1999). Nagai, M . , Masunaga, T., Hana-oka, N . hydrodentrogenation o f carbazole o n a M0/AI2O3 catalysts. Effects o f sulfiding and sulfur compounds. Energy & Fuels, 2, 645-651 (1988).  181  Nagai, M . , Goto, Y . , Uchino, O., O m i , S. T P D and X R D studies o f molybdenum nitride and its activity for hydrodenitogenation o f carbazole. Catalysis Today, 43, 249-259 (1998). Nakamura, D . , Stell, J. World wide refinery Capacity climbs to highest level ever. Oil Gas Journal. 52 68 (23 December, 2002). th  Nelson, A . E . , Sun, M . , Junaid, A . S. M . O n the structure and composition o f the phosphosulfide overlayer on N i P at hydrotreating conditions. Journal of Catalysis, 241, 2  180-188 (2006). N y l o n , M . K . , Choi, S., K w o n , H . , Curry, K . E . , Thompson, L . T. Catalytic properties o f early transition metal nitrides and carbides: tt-butane hydrogenolysis, dehydrogenation and isomerization. Applied Catalysis, 183, 253-263(1999). Odebunmi, E . O . , Ollis, D . F . Catalytic hydrodexygenation III. Interactions between catalytic hydrodeoxygenation o f m-cresol and hydrodenitrogenation  o f indole. Journal of  Catalysis, 80, 76-89 (1983). Okamoto, Y . , Niha, Y . , Imanaka, T., Teranishi, S. Surface state and catalytic activity and selectivity o f nickel catalysts i n hydrogenation reactions III. Electronic and catalytic properties o f nickel catalysts. Journal of Catalysis, 64, 397-404 (1980). Oyama, S. T., Wang, X . , Lee, Y . K'., Chun, W . - J . Active phase o f Ni P/Si02 i n hydroprocessing 2  reactions. Journal of Catalysis, 221 263 (2004). Oyama, S. T. N o v e l catalysts for advanced hydroprocessing: transition metal phosphides. Journal of Catalysis, 216, 343-352 (2003). Oyama, S. T., Wang, X . , Lee, Y . K . , Bando, K . , Requejo, F. G . Effect o f phosphorus content i n nickel phosphide catalysts studied by X A F S and other techniques. Journal of Catalysis, 210, 207-217(2002).  182  Oyama, S. T.,  Clark, T., Teixeira da Silva, L . S., Lede, E . J., Reuejo, F. G . , J. X A F S  characterization o f highly active alumina-supported molybdenum phosphide catalysts (M0P/AI2O3) for hydrotreating. Journal of Physical Chemistry B, 105, 4961-4966 (2001). Oyama, S. T., Y u C . C , Ramanathan, S. Transition metal bimetallic oxycarbides: Synthesis, characterization and activity studies. Journal of Catalysis 184, 535 (1999). Oyama, S. T., in: The Chemistry o f Transition Metal Carbides and Nitrides, Blackie Academic & Professional, London, (1996). Oyama, S. T, Schlatter, J. C . Metcalfe, J. E . , Lambert Jr. J. M . Preparation and characterization o f early transition metal carbides and nitrides. Industrial  Engineering  Chemistry  Research, 27, 1639-1648 (1988). Perot,-G. Hydrotreating catalysts containing zeolites and related materials-mecahnistic aspects related to deep desulfurization. Catalysis Today, 10, 111-128 (2003). Phillips, D . C , Sawhill, S. J., Self, R., Bussel, M . E . Synthesis, Characterization, and hydrodesulfurization properties o f silica-supported molybdenum phosphide  catalysts.  Journal of Catalysis, 20,266-273 (2002). Phillipson, J. J. Kinetics and hydrodesulfurization o f light and middle distillates in: Proceedings of the Paper Presented at the American Institute o f Chemical Engineers Meeting, Houston, (1971). Portefaix, J. L . , Cattenot, M . , Guerriche, M . , Thivolle-Cazat, J., Breysse, M . Conversion o f saturated cyclic and noncyclic amines over a sulphided MM0/AI2O3 catalyst : mechanisms o f carbon - nitrogen bond cleavage. Catalysis Today, 10 473-487 (1991). Powdered  Diffraction  Powder Files,  JCPDS  Swathmore, P A , (2005).  183  International  Center  for  Diffraction  Data,  Pinzon, M . H . , Centeno, A . , Giraldo, S. A . Role o f Pt i n high performance P t - M o catalysts for hydrotreatment reactions. Applied Catalysis A: General, 302 118-126 (2006). Prada, R . S., Reyes, E . , Galiasso, R., Romero, Y . , Munoz, R . Catalyst for m i l d hydrocracking feedstocks and method for its preparation. European Patent application. (1992). Prins, R . Egorova, M . , Rothlisberger, A . , Zhao, Y . , Sivasankar, N . , K u k u l a , P . Mechanisms o f hydrodesulfurization and hydrodenitrogenation.  Catalysis Today, 111, 84-93 (2006).  Prins, R., Catalytic hydrodenitrogenation. Catalytic hydrodenitogenation. Advances in Catalysis, 46, 399-464 (2001). Prins, R., D e Beer, V . H . J . Structure and function o f catalyst and promoter i n C o - M o hydrodesulfurization catalysts. Catalysts Review Science Engineering. 31 (1&2), 1-41 (1989) . Prins, R . , N i E X A F S  studies o f the N i - M o - S structure in carbon-supported and alumina-  supported N i - M o catalyst. Journal of Catalysis 133 94-111 (1992). Qian, W . , Hachiya, Y . , Wang, D . , Hirabayashi, K . , Ishihara, A . , Kabe, T., Okazaki, H . , Adachi, M . Elucidation o f promotion effect o f nickel on M0/AI2O3 and C0-M0/AI2O3 catalysts i n hydrodesulfurization using a  S radioisotope tracer method. Applied Catalysis. A, 227,  19-28 (2002). Qu, L . , Prins, R. Different Active sites i n hydrodenitrogenation as determined by the influence o f the support and fluorination. Applied Catalysis A. General, 250 105-115 (2003). Rabarihoela-Rakotovao, V . , Brunet, S., Berhault, G . , Perot, G . , D i e h l , F . Effect o f acridine and of actahydroacridine on the H D S o f 4,6-dimethyldibenzothiphene catalyzed by sulfided N i M o P / A I 2 O 3 . Applied Catalysis A: General, 267, 17-25 (2004). Rase, H . F . Fixed-Bed Reactor Design and Diagonostics: Gas-Phase Reaction, Butterworth, (1990) .  Reddy, B . M . , Mastikhin, V . M . , i n "Proc. 9 Int. Congr. Catal. Calgary" (M.J.Phillips t h  and M . Terman, Eds) p.82 Chemical Institute o f Canada, Ottawa, (1988). Robinson, W . R. A . M . , V a n Gestel, J. N . M . , Koranyi, T. I., Eijsbouts, S., V a n der Kraan, A . M . , V a n Veen, J. A . R., De Beer, V . H . J. Phosphorus promotion o f Ni(Co)-containing M o free catalysts i n quinoline hydrodenitrogenation. Journal of Catalysis, 161, 539-550 (1996). Rodriguez, J . A . , K i m , J. Y . , Hanson, J . C , Sawhill, S. J . Bussell, M . E . Physical and chemical properties o f M o P , N i P and M o N i P hydrodesulfurization catalysts: Time-resolved x-ray 2  diffraction, density functional, and hydrodesulfurization  activity studies. Journal of  Physical Chemistry B, 107, 6276-6285 (2003). Rynkwoski, J., Rajski, D . , Szyska, I., Grzechowwiak, J. R . Effect o f platinum on the hydrogenation activity o f nickel catalysts. Catalysis Today, 90 159-166 (2004). Satterfield, C . N . and Yang, S. H . Catalytic hydrodenitrogenation o f quinoline i n trickle bed reactors. Comparison with vapor phase. Industrial Engineering Chemistry Process Design and Development, 23, 11-19 (1984). Satterfield, C . N . and Cocchetto, J. F . Reaction network and kinetics o f vapor phase catalytic hydrodenitrogenation o f quinolie. Industrial Engineering Chemistry Process Design and Development, 20, 53-62 (1981). Satterfield, C . N . "Heterogeneous Catalysis i n industrial practice." M c G r a w - H i l l N e w Y o r k (1991) Sawhill, S. J., Layman, K . A . , V a n W y k , D . R., Engelhard, M . H . , Wang, C , Bussel, M . K . Thiophene hydrodesulfurization over nickel phosphide catalysts:effect composition and support. Journal of Catalysis, 231, 300-313 (2005).  185  o f the precursor  Senzi, L . , Jae, L . S . , Taeghwan, H . , Suslick, K . S . Catalytic hydrodenitogenation o f indole over molybdenum nitride and carbides with different structure. Applied Catalysis A: General, 184, 1-9 (1999). Severino, F., Laine, J., Agudo, A . L . Compensation effect and dual promotions effect i n activated carbon supported C o N i M o hydrodesulfurization catalysts. Journal of Catalysis 189, 244246 (2000). Shuit, G . C . A . and Gates, B . C . Chemistry and Engineering o f catalytic hydrodesulfurization. Journal of. American Institute of Chemical Engineering, 19, 33 417-438 (1973). Song, C , Introduction to chemistry o f diesel fuel, in: Song, C , Hsu, S., Mochida, I., (Eds), Chemistry o f Diesel Fuels, Taylor & Francis, N e w York, (2000). Song, C . A n overview o f new approaches to deep desulfurization for ultra-clean gasoline, diesel fuel and jet fuel. Catalysis Today, 86, 211-263 (2003). Sonnemans, J., van den Berg, G . H . , Mars, P. The mechanism o f pyridine hydogenolysis on molybdenum-containing catalysts. Journal of Catalysis, 31, 220-230 (1973). Stancy, K . , K i m , S. H . , Sayag, C , Brodzki, D . , Djega-Mariadassou, G . H D N o f 1-4 terahydroquinoline over M o N x O y and N b N x O y : Effect o f transition metal, solvent and ammonia. Catalysis Letter.53 59 (1998). Stinner, C , Tang, Z . , Haouas, M . , Weber, T h . , Prins, R . Preparation and 3 I P N M R characterization o f nickel phosphide on silica. Journal of Catalysis, 208, 456-466 (2002). Stinner, C , Prins, R., Weber, T h . Binary and ternary transition-metal phosphides as H D N catalysts. Journal Catalysis, 202, 187-194 (2001). Stinner, C , Prins, R., Weber, T h . Formation, structure and H D N activity o f unsupported molybdenum phosphide. Journal of Catalysis, 191, 434-444 (2000).  186  Sun, F.,Wu, W., Guo, J., Wei, Z., Yang, Y., Jiang, Z., Tian, F., L i , C. Dibenzothiophene hydrodesulfurization activity and surface sites of silica-supported MoP, an Ni-MoP catalysts. Journal of Catalysis, 228, 298-310 (2004). Sun, M . , Nelson, A . E., Adjaye, J. First principles study of heavy oil organonitrogen adsorption on NiMoS hydrotreating catalysts. Journal of Catalysis, 226, 41-53 (2004). Sundaramurthy, V . , Dalai, A., K., Adjaye, J. Comparison of P-containing Y-AI2O3 supported Ni-Mo bimetallic carbide, nitride and sulfide catalysts for H D N and HDS of gas oils derived from Athabasca bitumen. Applied Catalysis A: General, 311, 155-169 (2006). Szymanska, A., Lewandowsky, M . , Sayag, C , Djega-Mariadassou, G. Kinetic study of the hydrodenitrogenation of carbazole over bulk molybdenum carbide. Journal of Catalysis, 218, 24-31 (2003). Tiancum, X . , Haritao, W., Jianwen D., Coleman, K.S., Green, M . L . H . Study of the preparation and catalytic performance of molybdenum carbide catalysts prepared with C2H2/H2 Carburizing Mixture. Journal Catalysis, 211, 183-191 (2002). Tittensor, J. G., Gorte, R.J., Chapman, D. M . Isopropylamine adsorption for the characterization of acid sites in silica-alumina catalysts. Journal of Catalysis, 138, 714-720 (1992). Topsoe, H., Clausen, B. S., Massoth, F. E., in "Hydrotreating Catalysis Science and Technology" (J.R. Anderson and M . Boudart, Eds.), Springer, Berlin, (1996). Topsoe, H . , Clausen, B. S. Importance of Co-Mo-S type structures in hydrodesulfurization Catalysis Review Science & Engineering, 26, 396-417 (1984). Topsoe, H . , Clausen, B. S, Active sites and support effects in hydrodesulfurization catalysts. Journal of Applied Calysis 25 273-293 (1984). Turaga, U . T., Ma, X., Song, C. MCM-41 supported Co-Mo catalysts for deep  hydrodesulfurization o f light cycle oil. Catalysis Today, 86 265-275 (2003). V a n Veen, J . A . R., Colijn, H . A . , Hendriks, P . A . J . M . , van Welsenes, A . J . O n the formation o f type I and type II N i M o S phases i n N i M o / A l 2 0 3 hydrotreating catalysts and its catalytic implications. Fuel Processing Technology, 35, 137-157 (1993). V i v i e r , L . , Dominguez, V . , Perot, G . , Kasztelan, S. Mechanism o f carbon-nitrogen bond scission. Hydrodenitrogenation o f 1,2,3,4-tetrahydroquinoline and o f 1,2,3,4-tetrahydoisoquinoline. Journal of Molecular Catalysis, 67, 267-277 (1991). Volpe, L . , Boudart, M . , J. Compounds o f molybdenum and tungsten with high specific surface area : I. Nitrides. Journal of Solid State Chemistry, 59, 332-347 (1985). Volpe, L . , Oyama, S. T., Boudart, M . Preparation o f Catalysts, Elsevier, Amsterdam, 147 (1983). Wada, T., Kareda, K . , Murata, S., Nomura, M . , Effect o f modified P d metal on hydrocracking o f polyaromatic compounds over Ni-loaded Y-type zeolite. Catalysis Today, 31, 113-120 (1996). Wang, A . , Ruan, L . , Teng, Y . , L i , X . , L u , M . , Ren, J., Wang, Y . , H u , Y . Hydrodesulfurization o f dibenzothiophene over siliceous M C M - 4 1 - supported nickel phosphide catalysts. Journal of Catalysis 229 314-321 (2005). Wang, J., Huang, L . , L i , Q . Influence o f different diluents in PI/AI2O3 catalyst on the hydrogenation o f benzene, toluene and o-xylene. Applied Catalysis A: General, 175,191-199 ( 1998). Wang, X . , Clark, P. A . , Oyama, S. T. Synthesis, characterization and hydrotreating activity o f several iron group transition metal phosphides. Journal of Catalysis, 208, 321-331 (2002). Wise, R . S., Markel, E . J . Synthesis o f high surface area molybdenum nitride i n mixtures o f nitrogen and hydrogen. Journal of Catalysis, 145 344-355 (1994).  Yang, H . , Chen, J., Briker, Y . , Szykarczuk, K . , Pving, Z . Effect o f nitrogen removal from light cycle oil on the hydrodesulfurization o f dibenzothiophene, 4methyldibenzothiophene and 4,6-dimethyldibenzothiophene. Catalysis Today, 109, 16-23 (2005). Y a n g , S., Liang, C , Prins, R . Preparation and hydrotreating activity o f unsupported nickel phosphid with high surface area. Journal of Catalysis, 241, 465-469 (2006). Y u , C . C , Ramanathan, S., Oyama, S. T. N e w catalysts for hydroprocessing: Bimetallic oxynitrides M i - M - 0 - N ( M I , M H = M o , V , N b , Cr, M n and Co) Part I. n  Synthesis and characterization. Journal of Catalysis. 173, 1-9 (1998). Yumoto, M . , Usui, K . , Watanabe, K . , Idei, K . , Yamazaki, H . Development o f a Cosmo deep H D S catalyst for diesel fuel. Catalysis Today, 35, 45-50 (1997). Zawadski, R., Shih, S. S., Reiff, E . , Katzer, J. R., Kwart, H . Tenth and eleventh Quartely Reports for the Period august 16 1981 to Feb 14 1982; Prepared for the Office o f Fossil Energy, Department o f Energy, Washington, D C (1982). Zuzaniuk, V . , Prins, R. Synthesis and characterization o f silica-supported transition-metal phosphide as H D N catalysts. Journal of Catalysis, 219, 85-96 (2003).  189  Appendix A  Appendix A  Al:  Some considerations for the fixed-bed reactor  Heat or mass transfer effects caused by intrareactor, interphase or intraparticle gradients can disguise the results obtained by using fixed-bed reactors and often lead to misinterpretation (Fogler, 1981; Mears, 1971; Dautzenberg, 1988; Rase, 1990; Froment, 1990). Therefore before accurate and intrinsic kinetic data can be established, these disguises must be eliminated by adjusting experimental conditions so as to obtain plug flow. Steps for minimizing the effects o f the internal and external mass transport were taken. A2:  Parameters for testing plug flow  The dimensions o f the reactor and the catalyst are listed in Table A . 1. Table A l  Parameters o f catalyst and reactor  Parameter  Dimension  Diameter o f reactor  D = 9.53 x 10" m  Diameter o f catalyst particle  dp = 7 x 10" m  Length o f reactor  L = 6.1 x 10" m  Length o f catalyst bed  L B = 2 x l 0" m  4  1  2  Great care must be exercised to ensure that the flow pattern is ideal even when plug flow prevails. To ensure that the influence o f the reactor walls on the flow is eliminated, the diameter o f the P F R , must be at least 10 times the catalyst particle diameter, i.e. D/dp > 10. In addition, by virtue o f convection, axial gradients exist. These effects are minimized by selecting the correct ratio o f the length o f the reactor to the particle diameter. Thus L/dp > 50. In this set up, D/dp  =9.53 x 10" m / 7 x l O ^ m  A.l  3  190  Appendix A = 13.6 L/dp  = 6.1 x 10"' m / 7 x 10"*m  A.2  = 871 Since D / d p = 13.6 > 10 and L / d = 871 > 50, both axial and wall effects w i l l be minimized i n the present study as the laboratory conditions for plug flow are satisfied for the fixed bed. Alternatively, acceptable plug flow conditions can be assumed i f the N N  p e  is the Peclet number and N N  p e  =(0.087) <  Npemin = 8n  2 3  1-  p e  > N min, where pe  j is the minimum Peclet number. For the gas phase operation,  p e m  n  A  A.3  I  n  A  .  4  X  n = 1, NRe is Reynold's number and x is conversion. The superficial velocity is given by u = 4 V / TCD  A.5  2  where V = 0.04 ml/min, is the volumetric flow rate and D is the reactor diameter. Substituting values, 4 x 0 . 0 4 w / / m i n x l 0 - / 6 0 x(583/298) , , , r-r - = 1.07x10 Tn/s 3.14JC (9.53JC10~ ) 6  u=  M  3  2  Assume the viscosity o f the fluid is the same as the solvent dodecane, u. = 1.006 x 10" N s / m , 3  NRe can be calculated as, N  R e  = udpp/p. _ 1.07 x\Q-*ms x 7x\Q-  x 741.1  4  1.006  kg/m.3  3  xl0~ Ns/m 3  2  From equation A . 3 , N e = 0.087 x ( 0 . 0 5 5 ) P  023  (871) = 38.92  and from equation A . 4 and for x = 0.291 N i P conversion, 2  191  =0.0552  A.6  2  Appendix A N emin= 8 x In 1/(1-0.291) = 2.75 P  N  p e  A.3  (38.92) > Npemin (2.75) therefore deviation from plug flow is acceptable.  Diagnostic Test for internal and external mass transport effects  Some considerations w i l l be given to the effects o f diffusion i f they are known to limit the reaction rates.  Both internal and external mass transfer limitations w i l l be approached as  follows: A.3  Mass transfer limitation  Internal mass transfer limitation The Thiele modulus w i l l be used to determine i f internal mass transfers limitations are present. The equation for Thiele modulus is given as (Fogler, 1981): (t> = r A P p R / D 2  where r  2  A B  CA  A.7  is the observed reaction rate( for bulk N i P = 2.06 x 10" m o l g ' V ) , p ( N i P = 7.09 g 9  A  2  1  p  2  cm" ) is the catalyst packing density, R = 0.035 cm, is the radius o f the catalyst particle C 3  A  =  9.82xl0" m o l cm" is the concentration o f the reactant and D,VB = 1.18 x 10" c m s" is the 6  3  4  2  1  effective difftisivity o f the 4 , 6 - D M D B T in dodecane . In this study we establish that the Thiele Modulus is small and hence the effectiveness factor is almost 1. Substituting for Thilele Modulus <> |,  2  =  ( 3 . 5 x 1 0 ^ * 7.09 x 2.06*10-* 1.18JC10 X9.82X10_4  =  Q  ^  6  therefore, (j) = 0.124. From a graph o f effectiveness factor versus Thiele modulus (Foggier, 1981), the effectiveness factor is 0.99. This implies that there is no internal mass transfer limitation. 192  Appendix A In the case o f the supported NI2P/AI2O3, the effective diffusivity i n the pores w i l l taken into account. The effective diffusivity, De, in a porous catalyst is given by:  „  "p T -—  ^ A B  De =  A.9  p  where D A B is the relative diffusivity o f the 4 , 6 - D M D B T i n dodecane, e  P;  is the porosity, a is the  constriction factor and x is the tortuosity. Typical values o f e , o, and p  x are 0.4, 0.8 and 3.0  respectively (Fogler, 1981). Therefore De is calculated as:  1.18 xWx  0.4 JC 0.8  , ^  De =  =  l.26xl0" cw/s  A.IO  3  3.0 For the N i P / A l 0 , 2  2  3  R = 0.035 cm, p = 0.37 g/cm , (values obtained from Wang et al., 2002), r 3  = I 3 . l x 10" and C 9  A  A  = 9.82 x 10" m o l / c m hence putting i n equation A . 7 , 6  ,  3  (3.5xl0" ) x0.37x n.lxlO" 2  $ =-  2  -—:  2  9 4  =4.79xl0-  1.26JC10" X 9.82X10 3  4  A.ll  6  Hence § = 0.0219. This value also corresponds to also an effectiveness factor o f 0.9 indicating that internal diffusion does not affect the fixed bed reactor used i n the present study.  External mass transfer limitation The influence o f external mass transfer is tested using the following condition (Foggier, 1981): -r  A  pRn/k C A  A  < 0.15  A.12  If the condition is satisfied, then there are no external mass transfer limitations.  193  Appendix A ICA is the mass transfer rate coefficient cm/s and n is the reaction order. The value o f ICA is obtained from the Frosling correlation: S h = k dp/Deff - 2 + 0.6 R e  Sc  05  A  A . 13  0 3 3  where Sh is the Sherwood number, Re is the Reynold's number (0.0552), and Sc is the_Schmidt's number ( p / p D B =115) and the rest o f the symbols take their definitions as given earlier. Hence A  from equation A . 9 , k = (2+0.6Re - Sc 0  5  033  A  )xDAB/dp  .  A.14  = ( 2 + 0 . 6 * 0 . 0 5 5 2 ° x 1150.33) x 1.18 x 10" / 0.07 5  4  = 4.51 x 10" cm /s 3  2  Substituting all values i n equation A . 8 and for n = 1, 2 . 0 6 x l 0 ' molls.g xO.lAlglcm 9  4.51 x 10" cm / j x 9.82 x l O ' 3  =  x 0.035cm  3  2  6  mol I cm  3  0.0012  Since the L H S o f equation A . 8 = 0.0012 < 0.15, there is no external mass transfer limitation.  A . 4 Reactor Isothermal operation The heats o f reaction for organo-sulfur and nitrogen containing compounds are i n the ranges o f - 6 8 to - 1 3 and - 6 5 to - 7 kcal/mol respectively. A l s o , the heats o f adsorption o f 4,6D M D B T is 20-30 kcal/mol (Kabe et al., 1999). Since these heats o f reactions and adsorption are very small, negligible heat w i l l be dissipated hence, heat effects can be neglected. In addition, the catalyst bed is small therefore; negligible temperature gradients w i l l be set up. Therefore we can assume isothermal reactor operation.  194  Appendix A A.5 Determination of saturation vapor pressure Since the physical properties o f the 4 , 6 - D M D B T were not known, they were estimated (Baum, 1995). Using the Group Conribution Method (Baum, 1995), the boiling point, T B o f 4 , 6 - D M D B T was estimated to be 649.61 K . The vapor pressure, P o f 4 , 6 - D M D B T is estimated from the modified Clausis-Clapeyron s  equation (Baum, 1995).  AS(T ) A C/T'  f  In (P ) =  \  \  B  s  R[\.S x(T IT-\)-  0.8 x  B  \n{T IT)\  -6.%x(T /T-l)  A.15  M  B  where A S ( T B ) = Entropy, J/mol-K at the boiling temperature, = 81.119+ 13.083 log T - 2 5 . 7 6 9 x T / M +0.146528 x T B  2.1362 x 1 0 " x T 4  P  s  B  2 / M B  A.16  3 / M B  = saturation vapor pressure at the specified temperature,  atm, T M = melting point  temperature o f the compound, K , T = specified temperature and M = molecular weight o f compound, g/mol. For Dodecane, the P is estimated from the Antoine equation, A . 17 s  log(P ) = A l - ( B 1 / ( T + C 1 ) )  A.17  s  where A l = 4.09978, B l = 1625.928, C I = -92.839. A t the experimental conditions, T = 583 K , and molecular weight o f 4 , 6 - D M D B T = 212.31 g/mol, equations A . 1 5 and A . 1 6 are solved to give P o f 4 , 6 - D M D B T = 4.54 k P a and equation s  A . 13 gives P o f dodecane = 28.98 kPa. s  195  Appendix A A.6  Determination of vapor pressure of feed  Assumptions: 1. Concentration o f 4 , 6 - D M D B T is small (3000 ppm) and solution is dilute therefore Dalton's and Raoult's laws apply. 2. For the reactor operating conditions assume: Total Pressure = 3.1 M P a , and 5 g solution prepared. Combining Dalton's and Raoult's Laws,  P = yiP = XjP  s  v  A . 18  Where P is the partial pressure o f component, and y i and Xj are the vapor and liquid mol v  fractions respectively, P is the saturation vapor pressure s  But for our system, y A + y B + y c = 1 (gas phase)  A . 19  x + XB = 1 (liquid phase)  A.20  A  where A is 4 , 6 - D M D B T , B is dodecane and C is hydrogen. Combining equations A . 13 A . 15 the partial pressures o f the 4 , 6 - D M D B T = 0.0254 kPa, and dodecane = 9.65 kPa.  196  Appendix B  Appendix B B.l  Catalysts preparation-Calculation of required chemicals  The preparation o f catalysts was done following procedures described elsewhere (Oyama et al., 1998; Stinner et al., 2001). For example the required chemicals for preparing M o P are calculated as follows: 1. Calculations for preparing M o P Required amount o f M o P catalyst * * * * • ™ r. Amount o f M o in M o P  =3.0g 95.94 g/mol Mo •  •  x 3?  126.94 gl mol MoP = 2.27 g Amount o f ammonium heptamolybdate (NH )6Mo7024.4H20 ( A H M ) required 4  =  221 gx 1235.86 gl mol AHM 95.94  glmolMoxl  = 4.17g Amount o f diammonium hydrogen phosphate (NH4)2HP04 required =  ( 3 . 0 0 - 2 . 2 7 ) x 132.06 30.97  = 3.12g A d d 4.1725 g (3.38 mmols) and 3.1241 g (23.66 mmols) o f (NH )6Mo7024.4H 0 4  2  and (NH4)2HP04 respectively i n a 60 m l beaker and add 15 m l o f de-ioinsed water. Stir until completely dissolved and for the rest of the procedure refer to Chapter 3.  197  Appendix B 2. Calculations for preparing C o o . o s ^ P Required amount o f Coo.o8Ni P  = 3.0 g ,  Amount o f C o  = 0.03 x 3 g = 0.09 g  2  Amount o f cobalt hexahydrate ((Co(N03)2.6H20 required  =  290.59 x 0.09  = 0.44 g  58.93  Amount of N i P  = 3.0-0.09 g = 2.91 g  A • fxr- M - D Amount of N i i n N12P  =  2  58.69 x 2 x 2.91 148.35  = 2.30g Amount o f nickel nitrate Ni(N03)2.6H 0, required 2  =  290.81 x 2.30 58.69  = 11.41 g Amount o f diammonium hydrogen phosphate ( N T - L ^ H P C M required =  ( 3 . 0 0 - 2 . 3 0 - 0 . 0 9 ) x 132.06 30.97  = 2.59g A d d 11.41 g (39.23 mmols) and 2.59 g (19.6 mmols) of N i ( N 0 ) . 6 H 0 and 3  2  2  (NH )2HP04 respectively in a 60 m l beaker and dissolve with 15 m l o f de-ionised water. 4  Stir to form M 2 P precursor. Next dissolve 0.44 g (1.53 mmols) o f cobalt hexahydrate ((Co(NO"3)2.6H20 i n 10 m l o f de-ionised water and add to the beaker containing the M 2 P precursor. Stir continuously while evaporate i n hot plate. Then follow procedure as described i n Chapter 3.  198  Appendix B 3. Calculations for preparing Coo.4Ni P/Al 03 ( 3 wt% C o and 15 wt% N i ) 2  2  Required amount o f Coo.4Ni P/Al 03  = 8.0 g  Amount o f C o •  = 0.03 x 8 g = 0.24 g  2  2  Amount o f cobalt hexahydrate ( ( C o ( N 0 3 ) . 6 H 0 required 2  2  290.59 x 0.24 = Amount o f N i  =0.15x8  Amount o f N i P  =  2  = 1.18 g  58.93  = 1.2 g  148.35* 1.2 2 x 58.69  = 1.52 g  = 2.30g Amount o f nickel nitrate N i ( N 0 3 ) . 6 H 0 , required 2  2  =  290.81xl.20 58.69  = 5.95g Amount o f diammonium hydrogen phosphate (NH4) HP04 required 2  =  (1.52-1.2) x 132.06 30.97  = 1-35 g Amount o f A 1 0 2  = 8.0-1.52-0.24 g = 6.24 g  3  Required volume o f solution  = 0.76 ml/g x 6.24 g = 4.75 m l .  Since the volume o f solution is small (4.75 ml), 15 m l o f de-ionised water was used to dissolve the required chemicals and a two-step multiple impregnation was performed. Step 1: A d d 5.95 g (20.01 mmols) and 2.59 g (10.02 mmols) o f N i ( N 0 ) . 6 H 0 and 3  2  2  (NH4) HP04 respectively with 15 m l o f de-ionised water and stirred. 15 m l was chosen to 2  199  Appendix B make sure that all the contents were well dissolved. The M 2 P precursor was impregnated on the AI2O3 and dried at 393 K . Step 2: Dissolve 1.18 g (4.07 mmols) o f cobalt hexahydrate ( ( C o ( N 0 ) . 6 H 0 i n 5 3  2  2  m l o f de-ionised water and impregnated on the AI2O3 and dried again at 393 K . After step 2, the precursor was calcined, reduced i n H and passivated following 2  the same procedure as described in Chapter 3.  B.2  T P R calibration. T P R was calibrated as follows: 1. CU2O was reduced in H2 under the same reduction conditions as the sample and  the area o f the T P R profile was obtained by integration using Origin software. Area o f obtained from T P R profile o f CU2O  = 3.469 a.u  2. Reduction stoichiometry o f CU2O. Cu 0 + H 2  •  2  2Cu + H 0 2  From equation B . l , 1 m o l o f CU2O requires 1 m o l o f H (assume complete 2  reduction). Initial amount o f CU2O sample  = 0.05 g  M o l of C u 0  = 0.05 g/ 95.545 g/mol  2  = 5.2331 x 10" mol 4  Therefore m o l o f H2 consumed  = 5.2331 x 10" m o l 4  The H2 consumption per area = ^-2331 x 10—rnol_ _ j ^Qgg i Q-4 3.469 a.u x  Hence T P R calibration = 1.5086 x 10 " mol/a.u 4  200  J^QI/^QQ  B.l  Appendix B Degree of reduction: Example of Ni2P  B.3  The degree o f reduction o f N i P was determined as follows: 2  1. Integrated area o f T P R profile o f N i P precursor  = 21.1467 a.u  2  2. N i O . P 0 5 phase was identified as the precursor using X R D hence the reduction 2  stoichiometry was given as: 4NiO.P 0 + 9H 2  5  •  2  2Ni P + 9 H 0 2  2  B.2  Initial mass o f N i O . P 0 5 precursor  = 0.4269 g  mol N i O . P 0  = 0.4269 g / 216. 63 g/mol  2  2  5  = 1.9706 x 1 0 ' m o l 3  F r o m stoichiometry mol H (assumed complete reduction) = 4.4612 x 10 " m o l 2  3. T P R integrated area o f sample  =21.1467 a.u  M o l H consumed  = 1.5086 x l O " (mol/a.u) x 21.1467 4  2  = 3.19 x 10' m o l 3  v i Hence degree of reduction  =  3.19 x l O  - 3  4.4612 x 10 ~ = 71.94%  201  100 i  3  n  n  Appendix B B.4  Temperature programmed reduction of transition metal phosphide precursors in H using tapered element oscillating microbalance ( T E O M ) . 2  The T P R o f N i P , Coo.osNi P and M o P precursors in H were obtained using 2  2  2  T E O M and Figure B l shows the profile o f the mass change o f the precursors with the reduction temperature. The reduction temperatures obtained using the T E O M agreed very well with the reduction temperatures obtained using the T C D for measuring the same samples. For example the reduction temperature o f the single high temperature peak o f  Figure B l T P R o f transition metal phosphide precursors  202  Appendix B M o P using T C D is 955 K (Table 4.3) whereas using the T E O M it is 675 °C (947 K ) from Figure B l . Figure B 2 shows the mass change o f the N i P precursor with time o f reduction 2  using the T E O M . The percent mass loss was estimated as follows:  0  -0.07 -I 0  ,  1  10000  20000  ,  1  30000  40000  1 50000  Time, s  Figure B 2 Mass change during T P R o f Coo.o8Ni P phosphide precursor using T E O M 2  Initial amount o f Coo.o8Ni P precursor loaded  = 0.1264 g  Mass change after reduction  = 0.064 g  2  =  Therefore % mass loss  0  0 6 4  0.1264  x 100  = 51 % (as shown i n Table 4.3)  203  B.5  X R D of calcined precursors of metal phosphide  Coo.osMo-phosphide precursor  10  15  2  0  2  5  3  0  3  5  4  0  4  5  5  0  ]  10  10  5  5  6  0  6  6  7  0  Ni -phosphide precursor 2  15  15  2 0 2 5 3 0 3 5 4 0 4 5 5 0 5 5 6 0 6 5  2  0  2  5  X  3  5  4  0  4  5  5  0  5  5  6  0  6  5  7  0  20 Figure B 3 X R D patterns o f calcined precursors. Phases identified: Mo-phosphide precursor: M0O3.P2O5; Co-phosphide precursor: CoO.P20 Ni -phosphide 5;  2  precursor: N i . O P O s Coo.o8Ni -phosphide precursor: CoO.P C>5, N i . O P O s 2  ;  2  2  Coo.osMo-phosphide precursor: C o O . P O s Mo03.P Os 2  204  ;  2  2  Appendix B B.6:  Determination of crystallite sizes.  The crystallite sizes o f the bulk metal phosphides were determined from the X R D diffractograms. The N i M o P series prepared in Chapter 4 w i l l be used as examples for x  determining the crystallite sizes. The X R D diffractogram for M o P is shown in Figure B 4 . The procedure for calculating the crystallite sizes is outlined as follows: 1. Separate C u K A 2 from peak- X R D software used to do this after pattern taken and it is known as the Rachinger correction. 2.  Subtract background noise.  3. A p p l y integral breadth: B (specimen)  = —^ x - — Hp ISO  B.3  J  where A p is the total area o f peak and H p = Peak height (both obtained from peak integration). 4.  Separate instrumental broadening:  P*(total)  P (specimen) * cos (O)A,  P*(specimen)  B.4  = p* (total)- (P*(standard)) / p*(total) 2  where X = wavelength (1.54056), 0 is the angle at the peak o f integration and B(standard) is the B a F X R D pattern which is the instrument. 2  5. Crystal size, dp,  = 1/ P*(specimen)  205  Appendix B  C h i 2 = 282.1 193485 A  Corr Coef=0.9987  Date:2/24/2006  C O D = 0.9974 D eg ree of F reedom = 24 5  # of Data Points=248  2.0x10' 1 .5x10" Q. O  f 1 .0x10  3  c  5.0x10  2  0.0 40  42  Fitting Results  AreaFit 1 1 51 . 9 7 5 1  44 46 2 theta  MaxHeiaht 1 796.31 245  48  50  FWHM 0.40825  BaseLine: C O N S T A N T  Figure B 4 Integration o f the most intense peak o f the M o P diffractrogram  For example to calculate the M o P crystallite size from Figure B 3 , Total Area (Ap)  = 1151.96 cps. degree  Peak height (Hp)  = 1796.31  Angle 20  = 42.86  and for the standard sample B a F  2  206  Appendix B Total Area (Ap)  = 197.8 cps. degree  Peak height (Hp)  = 1708 cps  Angle 20  = 42.86 degree  From equation B . 3 , 6 (specimen)  = (3.14/180) * 1151.96/1796.31 =  p (standard)  = (3.14/180) * 197.8/1708 =  B*(Total)-specimen  0.011187  0.00202  = 0.011187 *Cos (42.86/2) / 1.54056 = , 0.00676  P*(Total)-standard  =  0.00202 *Cos (24.85/2)  =  0.00128  / 1.54056  Therefore from equation B . 4 ,  p*(specimen) = 0.0067 - 0.00128 / 0.0067 2  Hence, dp = 1/p* (specimen)  =  0.00652  =  158.433 A  =  16 nm  207  Appendix B  The results for the rest o f the calculation is shown on Table B l .  Table B1 Data for the crystallite sizes of N i x M o P  P* Total area (Ap) Peak height(HP) 29  (Total)  3  [3* (specimen) d.  Ni . 7MoP  550.53  982.4  42.91  0.0098  0.0059  0.0056  18  Ni  MoP  268.81  558.15  42.98  0.0084  0.0051  0.0048  21  Nio. MoP  347.29  744.9  42.93  0.0081  0.0049  0.0046  22  Nh.nMoP  276.64  602.67  43.15  0.0080  0.0048  0.0045  22  MoP  1151.96  1796.31  42.86  0.0112  0.0068  0.0065  15  197.8  1708  24.85  0.0020  0.0013  0  0  0 1 6  38  BaF (standard) 2  B.7  Calculation of Lattice Parameters. The lattice parameters of the prepared bulk M o P , C00.07M0P and Co Ni2P (0< x x  <0.34) were calculated using equation B.5 for hexagonal phases.  -Xr d  = —x (h 3  2  + hk +k )/a . 2  2  +  —  l  c  2  B.5  Where d is the plane spacing, hkl are miller indices and a and c are the lattice parameters. The Bragg's angle is used to determine d and is given by equation B . 6 . d = X12  sin (9)  B.6  where X is the wavelength o f the X R D source and 0 is the angle o f the diffraction peak.  208  Appendix B The example o f M o P w i l l be calculated. 1.  The angles for the two intense peaks from the M o P diffractogram are 21.43° and 31.92°.  2.  h = l , k = 0 a n d l = 1 at 21.43°  A 'trial and error' method is used to solve equation B.5 by transforming equation B.5 as:  -V d  - -x 3  2  (h  2  + hk + k )la 2  2  -Xr=0 c  B.7  2  Hence when B . 7 is satisfied to some acceptable error then the parameters are accepted. 1/d  jx  2  = 1 / 2 sin (21.34/180* TC) = 0.2246  (h  2  + hk +k )/a 2  =4/3 * ( 1 + 0 + 0 )/a = 1.3333/a  2  2  2  From equation B . 7 , 0.2246 - - l a 3  2  - \  c  2  = 0  B.8  choose a= 3.243 and c = 3.196 and substitute in equation B . 7 and answer is 2.047E-10. Similarly use angle 31.92° and for h = 1, k = 0 and 1 = 0, equation B . 7 satisfies a = 3.240 and c = 3.196 where the answer for equation B . 7 is 0.0001759 and does not vary. The values for M o P lattices w i l l therefore be a = 3.240 and c = 3.196 as shown on Table 4.1. The lattice parameters for the rest o f the prepared metal phosphides were calculated following the same procedure and it is found on Table 4.1.  B.8  XPS Survey Scan In order to identify the components on the surface o f the metal phosphides, X P S  survey scan was obtained. Once the elements were identified using survey scan, the  209  Appendix B narrow scan was obtained for the individual elements and from which the atom ratios can be calculated. A n example o f the survey scan is shown in Figure B 5 . Both the bimetallic C00.07M0P and Co .o8Ni P reveal the presence o f N i 2p, M o 3d, P 2p and P 2s. In 0  2  addition, the X P S profiles also show O ( A ) , O l s and C Is.  210  Appendix B  2.50E+05  01s  2.00E+05  </)  CO0.07M0P  0(A)  Q. O  Mo3d  >M.50E+05  CO  c cu  1.00E+05  5.00E+04  0.00E+00 1100  1000  900  800  700  600  500  400  300  Binding Energy, eV  Figure B 5  XPS survey scan of Coo.osNiiP and C00.07M0P  211  200  100  0  Appendix B B.9 Determination and repeatability of C O chemisorption The C O uptake data from Co .o8Ni P w i l l be used to show the repeatability o f the 0  2  experiments. Figure B 6 shows two runs using COQ.OS^P.  c  CO Q O  1200  Figure B 6 Repeatability C O uptake using CO0.08N12P Conditions: C O flow = 0.3 ml/min, He(mix) = 30 ml/min size o f sample loop = 1ml Run 1: C O uptake = 0.90 mmols/g, R u n 2: C O uptake = 0.99 mmols/g  The determination o f the C O uptake is carried out as follows: In'ected C O mols njec e mo s  = ^'^ ^ ^ CO) (sample loop size (\ml)) (30m// m i n He) (22414m/1 mol) m  =  4.46  1m  n  x 10" m o l 7  212  Appendix B Table B 2 shows the determined areas and sample sizes used. Assume the final peak area corresponds to the m o l C O injected (since at saturation the sample does not take i n anymore C O and the area is constant).  Table B 2 Results o f the repeatability o f C O uptake on C o o . o s ^ P Sample wt.  Integrated Area (a.u)  C O Uptake  (g)  Final Peak  Runl  1.0010  0.01601  3.2 x 10"  Run2  1.0801  0.01350  3.14 x 10"  Difference Total  umols 0.90  2  2  0.96  The uptake is calculated as follows: R u n 1: C O uptake  = 0.0320 x 4.46 x 10 ~ mols / 0.01601 / 1.0010 g 7  = 0.90 umols/g R u n 2: C O uptake  = 0.0314 x 4.46 x 10 ~ mols / 0.01350 / 1.0801 g 7  = 0.96 mols Error  = (0.96-0.9)/0.96 = 6.25 %  213  Appendix B B.10 Repeatability of n-propylamine titration of acid sites on metal phosphides The n - P A chemisorption data for Con.osNiiP w i l l be used to show the repeatability o f the experiments. Figure B 7 shows two runs using C00.08N12P.  0  2000  4000  6000  8000  10000  12000  Time.s  Figure B 7 Repeated n - P A data using Con.os^P /'  Determination of Brensted acidity The system was calibrated using 3 zeolite samples of known acidity. C00.08N12P n - P A chemisorption w i l l be used to show how the Brensted acidity was calculated. Average integrated area for zeolites  =0.16208 V . s  F l o w rate o f He  = 3 0 ml/min  Flowrate o f n - P A  =12.5 ml/min  214  Appendix B Volume o f sample loop  = 1 ml  Therefore mols o f n - P A  =  1 m l  x  1  2  5  m  l  /  m  i  n  30 ml/min x 22400 ml/mol  = 2 . 0 3 x 10" m o l 5  Table B 3 shows the sample sizes and integrated areas. The Bronsted acidity is calculated as follows:  Table B 3 Results o f the n - P A repeatability using Co .o8Ni P 0  Sample wt.  2  Integrated Area  (g)  Bransted acidity  (V.s)  (umols)  Runl  0.9666  1.12102  145.40  Run2  0.7986  0.84701  133.13  Runl: 2.035x10" molx\A2\V.s 5  Bransted acidity  x 1000  0.1621 V.s x 0.9666g =145.40 umols/g 2.035x10" mol x 0.8470 V.s x 1000 5  Similarly for R u n 2:  0.1621 V.s x 0.7986g ;  Error  133.13 umols/g  = (145.40-133.13)/145.40 = 8.44 % 215  Appendix B B . l l Determination of the specific consumption of 4,6-DMDBT and carbazole In order to determine the consumption o f either the 4 , 6 - D M D B T or carbazole, the specific rate constant, k was obtained then the specific consumption obtained assuming first order kinetics. Equation B 8 is re-written as:  Fdx  = -rdW  Ao  (B8)  A  Substituting for - r = kCAo, solution o f equation B 8 gives A  (l-x )  = e-  v  0  (B9)  k,m  A  is the volumetric flow rate. Different values o f k are obtained from equation B l l and  substituted in equation B.9 using the first order equation to obtain the consumption rate.  Example of calculating T O F for  Ni2P:  In the case o f the T O F , the specific consumption is obtained assuming the reactant at the entrance o f the reactor. The rate constants ks' are then multiplied by the initial concentration and divided by the C O uptake. The rate constant k, is obtained from equation B 9 and for XA = 29.1 % and vo = 2.4 ml/h k = 0.83 ml/gcat.h and k C A o = 8.11 x 10" mol/gcat.s. 6  C O uptake  =  1.11 pmol/g  216  Appendix B Therefore,  1 Or  _ 8.11 x l O " ° mollgcatls — 1.11 x 10"° mollg = 7.31 s"  217  Appendix B  1.5  -I  ,  ,  ,  ,  ,  1.52  1.54  1.56  1.58  1.6  1.62  ,  1  ,  ,  ,  1  1.54  1.56  1.58  1.6  1.62  1.64  -1  —| 1.64  0.5  -2 -I 1.52  Figure B 8 Arrhenius plots for determining the apparent activation energy for the H D S ( A ) and H D N (B) o f L G O over ( • ) C00.4N12P/AI2O3, and (•) P t - C o o A P / A ^ O s  218  Appendix B  Figure B 9  Diagram of an M C M - 4 1 crystal (Leon et al. 1995)  219  Appendix C  Appendix C Hydrodesulfurization Experiments C.l  Summary of hydrodesulurization experiments  Table C I  Summary o f experiments for hydrodesulfurization o f 4,6-dimethylmethyl dibenzothiophene ( 4 , 6 - D M D B T - 3 0 0 0 ppm) over transition metal phosphides at 3.0 M P a , H pressure for 12 h time on stream 2  Exp.  Reactant  Catalyst  #  Temp. S V K xlO" mol/h-g 583 7.9 583 7.9 583 7.9 583 7.9 583 7.9 583 7.9 583 7.9 583 7.9  29.1 36.2 22.3 48.7 23.4 17.4 10.1 42.5  583 583 583 583 583 583 583 583 583 583 583 583 548 533  47.2 43.9 81.3 86.5 84.4 87.1 99.8 97.2 90.0 93.5 63.5 51.7 64.3 76.6  Conversion  2  1 2 3 4 5 6 7 8  4,6-DMDBT 4,6-DMDBT 4,6-DMDBT 4,6-DMDBT 4,6-DMDBT 4,6-DMDBT 4,6-DMDBT 4,6-DMDBT  Ni P MoP CoP Co .o8Ni P Coo.i Ni P Coo.3 Ni P Coo.79Ni P  9 10 11 12 13 14 15 16 17 18 19 20 21 22  4,6-DMDBT 4,6-DMDBT 4,6-DMDBT 4,6-DMDBT 4,6-DMDBT 4,6-DMDBT 4,6-DMDBT 4,6-DMDBT 4,6-DMDBT 4,6-DMDBT 4,6-DMDBT 4,6-DMDBT 4,6-DMDBT 4,6-DMDBT  Co .o8Ni P (Repeat) Co P  2  0  2  6  2  6  2  2  C00.07M0P 0  2  2  Ni P/Al 0 Coo, Ni P/Al 0 Co .4Ni P/Al O (repeat) Co . Ni P/Al O -F Co . Ni P/MCM. Pt-Co Ni P/Al O Pt-Co Ni P/Al O -WA Pt-Co Ni P/Al O Pt-Co Ni P/Al O Pt-Coo. Ni P/Al 0 Pt-Co Ni P/Al O Pt-Co Ni P/Al O 2  2  4  3  2  0  2  2  0  4  2  0  4  2  0 4  2  3  2  3  2  3  2  3  0 4  2  2  3  0 4  2  2  3  0 4  2  4  2  2  3  2  3  0 4  2  2  3  0 4  2  2  3  220  7.9 7.9 7.9 7.9 7.9 7.9 7.9 7.9: 7.9 9.2 32.0 62.0 7.9 7.9  Appendix C C.2  Response Factor and sample calculation of HDS activity  Response factor o f the 4 , 6 - D M D B T in dodecane was obtained by injecting known composition o f 0.1 umol o f the liquid sample into gas chromatography ( G C ) . Four samples o f each fixed composition were injected and the average reading was used to plot the graph o f concentration versus the G C area. A s shown in Figure C I , the slope = 3.09 x 10" mol/area was 9  used as the response factor.  8.E+06 x = G C area  Figure C . l  Calibration curve for 4 , 6 - D M D B T used to determine the response factor  221  Appendix C Sample calculation for the HDS of 4,6-DMDBT. Data for the experiment 4 is used for illustrating the calculations as shown on Table C 2 . Table C 2 Results o f hydrodesulfurization o f 4 , 6 - D M D B T over Coo.osNiiP 4,6-DMDBT  4,6-DMDBT  DMBP  Area  4,6-DMDBT mol x10"  Conversion m o l %  2  2061910  6.38  4  1785485  6  Time, h  DMBCH  Area  DMBP mol x10"  Area  DMBCH mol x 10"  35.07  103518  0.32  267750  8.30  5.52  43.77  1487214  4.60  171987  5.32  1214631  3.76  61.75  1419508  4.39  163596  5.06  8  1289328  3.99  59.40  1275195  3.94  203621  6.30  10  1857220  5.74  41.52  657055  2.03  206880  6.40  12  1742719  5.39  45.12  723095  2.24  208007  6.43  1 0  9  11  initial mols  Total mols  Area  Others* mol x10"  Mol Balance error  x10"  x10"  %  476220  14.73  9.82  9.75  0.74  304398  9.41  9.82  9.81  0.09  1.63  357103  11.04  9.82  9.80  0.17  16505  0.51  317330  9.81  9.82  9.66  1.66  0.28  6004  0.19  366937  11.35  9.82  9.60  2.28  0.18  0  0.00  315983  9.77  9.82  9.26  5.67  Time  MCHT  MCHT  h  Area  2  134304  4.15  108360  4  32614  1.01  26010  6  62747  1.94  52718  8  20961  0.65  10  9214  12  5780  rnolxlO"  11  DMPU Area  DMPU rnolxlO"  11  3.35 • 0.80  Others  11  * others include unidentified products with small products cracked  222  9  '  9  Appendix C The initial concentration o f 4 , 6 - D M D B T = average area o f 4 , 6 - D M D B T x response factor o f 4 , 6 - D M D B T = 3175554 area x 3.09 x 10" mol/area 9  = 9.82x l O mol - 9  Conversion o f 4 , 6 - D M D B T after 12 h =  [4,6 - DMDBT]  - [4,6 -  I=0  [4,6 _ 9.82-5.39 9.82  DMDBT]  I=I  -DMDBT]  I=0  x 100  = 45.1 %  mol balance, % error  =  [ 4,6 - DMDBT]  -[4,6 - DMDBT + all products]. '~° J^L [4,6-DMDBT] 0  n  l=0  9.82-9.26 9.82  h  1  x 100  5.67 %  223  x  100  Appendix C C.3 Example of repeatability of hydrodesulfurization experiments Table C3 shows the data for 2 runs o f hydrodesulfurization  o f 4 , 6 - D M D B T over  Coo.o8Ni P. The standard error, S.E. calculations were obtained as shown i n equations C 1 - C 3 . 2  Table C3 Repeatability of hydrodesulfurization of 4 , 6 - D M D B T using Coo.o8Ni P 2  Experiment # 4 & 9  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H : 25 ml/min  Catalyst: Co .o8Ni P  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :7.9 x 10" mol/h-gcat  0  2  2  2  (4,6-DMDBT+dodecane): 0.04 ml/min  Time  mols 4,6-DMDBT  mols 4,6-DMDBT  Average  S.E.  h  run 1  run 2  mols  mol  2  6.37584E-09  6.37713E-09  6.37649E-09  0.00  0.01  4  5.52107E-09  5.99757E-09  5.75932E-09  0.05  5.28  6  3.75588E-09  3.64814E-09  3.70201 E-09  0.01  1.19  8  3.98686E-09  5.01723E-09  4.50205E-09  0.11  11.43  10  5.74289E-09  5.21204E-09  5.47747E-09  0.06  5.89  12  5.38883E-09  5.31022E-09  5.34953E-09  0.01  0.87  224  %S.E.  Appendix C  Time  mols D M B P  mols D M B P  Average  S.E.  h  run 1  run 2  mols  mol  2  0.32  0.33  0.33  0.02  2.18  4  4.60  4.57  4.59  0.00  0.46  6  4.39  4.07  4.23  0.05  5.35  8  3.94  3.57  3.76  0.07  6.97  10  2.03  2.06  2.05  0.01  1.04  12  2.24  2.07  2.16  0.06  5.58  Time  mols D M B C H x I O " mols D M B C H x I O " 11  11  %S.E.  Average  S.E.  %S.E.  h  run 1  run 2  mols  mol  2  8.30  8.92  8.61  0.05  5.09  4  5.32  5.73  5.53  0.05  5.25  6  5.06  5.60  5.33  0.07  7.16  8  6.30  5.74  6.02  0.07  6.58  10  6.40  6.26  6.33  0.02  1.56  12  6.43  5.92  6.18  0.06  5.84  225  Appendix C  Time  mols M C H T x I O "  11  mols M C H T x I O "  11  Average  S.E.  %S.E.  h  run 1  run 2  mols  mol  2  4.15  4.42  4.29  0.04  4.46  4  1.01  1.04  1.03  0.02  9.00  6  1.94  1.84  1.89  0.04  3.74  8  0.65  0.67  0.66  0.02  2.14  10  0.28  0.28  0.28  0.00  0.00  12  0.18  0.19  0.19  0.04  3.82  run 1  Run2  48.7  Average  S.E.  47.2  %S.E.  47.95  0.02  2.21  Conversion  The average conversion is determined here as the average after the initial stabilization period, i.e. 8-12 h of time on stream.  226  Appendix C C.4  Calculation of standard error (S.E.) The average o f the sample,  a j = ^ -  C.l  where, aj is the sample size and n is the number o f sample. The standard error is calculated follows:  S.E.  C.2  n-\  and the percent standard error is: ^ % S.E. = n /  0  S.E.xlOO  a,  227  C.3  Appendix C C.5  Data for hydrodesulfurization experiments  Experiment # 1  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H2 : 25 ml/min  Catalyst: N i P  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :7.9 x 10" mol/h-gcat  2  2  (4,6-DMDBT+dodecane) : 0.04 ml/min  Time, h  4,6-DMDBT  4,6-DMDBT  Time, h  mol  2  Conversion m o l %  3.95E-09  4  59.78  6.03E-09  6  6.44E-09  38.57  8  6.94E-09  10  29.34  6.96E-09  29.07  12  6.97E-09  29.03  Avg  29.14  34.39  DMBCH  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  mols  mols  mols  mols  mols  mols  mols  mols  % error  2  2.25E-09 3.12E-10 8.00E-10 1.01E-10 3.30E-09  2.83E-09  9.60E-09  9.82E-09  2.25  4  1.05E-09 7.12E-10 2.89E-10 6.01E-10 2.79E-09  3.89E-09  9.33E-09  9.82E-09  4.96  6  4.67E-10 3.41E-10 2.34E-10 5.68E-10 4.78E-09  2.51 E-09  8.90E-09  9.82E-09  9.31  8  4.30E-11 0.00E+00 8.29E-10 3.14E-11 6.65E-09  1.47E-09  9.02E-09  9.82E-09  8.13  10  0.00E+00 9.79E-11 7.83E-10 4.78E-11 7.13E-09  1.62E-09  9.68E-09  9.82E-09  1.47  12  1.41E-10 0.00E+00 5.95E-10 0.00E+00 7.09E-09  1.77E-09  9.60E-09  9.82E-09  2.23  228  Appendix C Experiment # 2  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H2 : 25 ml/min  Catalyst: M o P  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :7.9 x 10  mol/h-gcat  2  (4,6-DMDBT+dodecane): 0.04 ml/min  4,6-DMDBT Time, h  4,6-DMDBT  mol  2  Conversion m o l %  4.69E-09  4  4.57E-09  52.20  6  3.97E-09  8  59.57  6.18E-09  10  37.02  6.29E-09  12  35.96  6.32E-09  35.61  53.41  Avg  Time, h D M B C H mols  36.21  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  mols  mols  mols  mols  mols  mols  mols  % error  2  2.10E-10  2.77E-10 3.47E-10 0.00E+00 6.79E-09 1.61 E-09  9.23E-09  9.82E-09  6.00  4  7.30E-11 0.00E+00 1.13E-09 6.20E-11 6.67E-09 1.56E-09  9.49E-09  9.82E-09  3.32  6  0.00E+00 4.18E-10 2.08E-09 0.00E+00 2.85E-09 4.43E-09  9.78E-09  9.82E-09  0.43  8  0.00E+00 1.54E-10  1.69E-09 0.00E+00 5.63E-09 1.98E-09  9.45E-09  9.82E-09  3.77  10  8.71 E-11 5.70E-11 1.11E-09 4.36E-11 6.31 E-09 1.78E-09  9.38E-09  9.82E-09  4.49  12  6.39E-11 0.00E+00 9.55E-10 3.66E-11 6.84E-09 1.50E-09  9.40E-09  9.82E-09  4.31  229  Appendix C Experiment # 3  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H2: 25 ml/min  Catalyst: C o P  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :7.9 x 10" mol/h-gcat 2  (4,6-DMDBT+dodecane): 0.04 ml/min  4,6-DMDBT  Time, h  4,6-DMDBT  Time, h  mol  Conversion m o l %  2  7.84E-09  20.12  4  6.83E-09  30.40  6  7.50E-09  23.65  8  7.56E-09  23.03  10  7.71 E-09  21.46  12  7.63E-09  22.35  Avg  22.28  DMBCH  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  mols  mols  mols  mols  mols  mols  mols  mols  % error  2  2.24E-10  1.94E-10 3.33E-10  4.99E-11 7.84E-09 1.08E-09 9.73E-09 9.82E-09  0.95  4  2.25E-10 .1.01 E-10 6.42E-10 4.99E-11 6.83E-09 1.54 E-09 9.40E-09 9.82E-09  4.31  6  1.78E-10 6.02E-11 4.94E-10 4.99E-11 7.50E-09 1.48E-09 9.76E-09 9.82E-09  0.57  8  3.24E-10 0.00E+00 3.47E-10  4.99E-11 7.56E-09 1.17E-09 9.45E-09 9.82E-09  3.71  10  2.69E-10 0.00E+00 3.90E-10 4.99E-11 7.71 E-09 1.22E-09 9.64E-09 9.82E-09  1.83  12  1.82E-10 8.92E-11 4.12E-10 4.99E-11 7.63E-09 1.16E-09 9.52E-09 9.82E-09  3.05  230  Appendix C Experiment # 5  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H2 : 25 ml/min  Catalyst: C o . i N i P  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :7.9 x 10" mol/h-gcat  0  6  2  2  (4,6-DMDBT+dodecane): 0.04 ml/min  4,6-DMDBT Time, h  4,6-DMDBT  mol  2  Conversion m o l %  6.90E-09  4  29.70  6.55E-09  6  33.49  7.10E-09  8  27.54  7.35E-09  10  25.17  7.55E-09  12  22.90  7.65E-09  22.16  Avg  23.41273271  DMBP  DMBCH  MCHT  DMPU  Others  Initial  Total  error  Time, h  mols  mols  mols  mols  mols  mols  mols  %  2  9.39E-10  2.79E-10  2.31 E-10  1.14E-10  5.74E-10  9.82E-09  9.04E-09 7.93  4  9.86E-10  1.90E-10  2.24E-10  9.98E-11  9.79E-10  9.82E-09  9.03E-09 8.02  6  8.69E-10  1.55E-10  2.87E-10  8.89E-11  1.19E-09  9.82E-09  9.70E-09 1.22  8  9.64E-10  1.50E-10  2.24E-10  9.30E-11  9.00E-10  9.82E-09  9.69E-09 1.36  10  1.18E-09  2.29E-10  2.44E-10  9.09E-11  3.97E-10  9.82E-09  9.69E-09 1.29  12  6.78E-10  3.46E-10  5.36E-10  1.08E-10  4.50E-10  9.82E-09  9.77E-09 0.48  231  Appendix C Experiment # 6  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H : 25 ml/min  Catalyst: Co .34Ni P  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :7.9 x 10" mol/h-gcat  0  2  2  2  (4,6-DMDBT+dodecane): 0.04 ml/min  4,6-DMDBT Time, h  mol  2  1.356E-08  4  1.451E-08  6  1.478E-08  8  1.564E-08  10  1.625E-08  12  1.672E-08  4,6-DMDBT Conversion m o l % 30.91 26.04 24.71 20.31 17.18 14.78  Avg  Time, h  17.42  DMBCH  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  mols  mols  mols  mols  mols  mols  mols  mols  % error  2  3.48E-10 8.02E-11 3.03E-10 8.93E-10 6.78E-09 1.41 E-09  9.81 E-09  9.82E-09  0.06  4  2.14E-10 8.28E-11 2.83E-10 5.48E-10 7.26E-09 1.40E-09  9.79E-09  9.82E-09  0.29  6  5.83E-11 1.27E-10 3.05E-10 4.68E-10 7.39E-09 1.21 E-09  9.56 E-09  9.82E-09  2.64  8  1.74E-10 1.06E-10 5.85E-10 4.42E-10 7.82E-09 4.30E-10  9.56E-09  9.82 E-09  2.69  10  3.04E-11 1.08E-10 6.34E-10 4.01 E-10 8.13E-09 4.83E-10  9.79E-09  9.82E-09  0.30  12  2.54E-11 0.00E+00 2.48E-10 2.96E-10 8.37E-09 4.67E-10  9.40E-09  9.82E-09  4.23  232  Appendix C Experiment # 7  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H : 25 ml/min  Catalyst: Co .8oNi P  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :7.9 x 10" mol/h-gcat  0  2  2  2  (4,6-DMDBT+dodecane) : 0.04 ml/min  4,6-DMDBT Time, h  mol  2  1.75E-08  4  1.62E-08  6  1.66E-08  8  1.72E-08  10  1.72E-08  12  1.85E-08  4,6-DMDBT Conversion m o l % 11.08 17.25 15.67 12.20 12.42 5.79  Avg  Time, h  10.14  DMBCH  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  mols  mols  mols  mols  mols  mols  mols  mols  % error  2  6.94E-11 0.00E+00 2.20E-10 3.24E-10 8.73E-09  3.56E-10  9.70E-09  9.82E-09  1.21  4  0.00E+00 0.00E+00 3.06E-10 2.99E-10 8.13E-09  8.87E-10  9.62E-09  9.82E-09  2.06  6  7.12E-11 0.00E+00 2.52E-10 2.88E-10 8.28E-09  8.84E-10  9.78E-09  9.82E-09  0.44  8  9.72E-11 4.42E-11 3.18E-10 2.03E-10 8.62E-09  3.95E-10  9.68E-09  9.82E-09  1.44  10  7.17E-11 9.39E-11 2.54E-10 2.09E-10 8.60E-09  3.78E-10  9.61 E-09  9.82E-09  2.16  12  8.72E-12 0.00E+00 1.78E-10 7.45E-11 9.25E-09  1.89E-10  9.70E-09  9.82E-09  1.21  233  Appendix C Experiment # 8  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H2 : 25 ml/min  Catalyst: C00.07M0P  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :7.9 x 10" mol/h-gcat 2  (4,6-DMDBT+dodecane): 0.04 ml/min  4,6-DMDBT Time, h  4,6-DMDBT  mol  2  Conversion m o l %  8.164E-09  4  16.86  7.148E-09  6  27.21  5.406E-09  8  5.124E-09  44.95  10  5.742E-09  12  6.081 E-09  47.82 41.52 38.07  Avq  Time, h  42.47  DMBCH  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  mols  mols  mols  mols  mols  mols  mols  mols  % error  2  3.99E-10 3.24E-10 8.03E-10 3.02E-10  6.04E-09  1.68E-09  9.54E-09  9.82E-09  2.81  4  3.57E-11 8.69E-11 3.29E-10 2.35E-09  4.19E-09 2.08E-09  9.08E-09  9.82E-09  7.56  6  1.61E-11 7.01 E-11 4.14E-10 2.87E-09  3.16E-09 2.79E-09  9.32E-09  9.82E-09  5.12  8  5.96E-11 1.37E-10 4.98E-10 1.87E-09  5.39E-09  1.54E-09  9.50E-09  9.82E-09  3.29  10  6.05E-11 5.98E-11 4.20E-10 1.98E-09  5.61 E-09 1.00 E-09  9.13E-09  9.82E-09  7.01  12  4.60E-11 6.93E-11 4.25E-10 1.99E-09  5.95E-09  9.50E-09  9.82E-09  3.26  234  1.02E-09  Appendix C Experiment #10  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H : 25 ml/min  Catalyst: C o P  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :7.9 x 10  2  2  mol/h-gcat  (4,6-DMDBT+dodecane) : 0.04 ml/min  4,6-DMDBT  4,6-DMDBT  Time, h  mol  2  Conversion m o l %  3.556E-09  4  63.78  3.962E-09  6  4.696E-09  59.65  8  5.435E-09  10  5.588E-09  12  5.954E-09  52.17 44.65 43.09 39.37  Avq  Time, h  43.87  DMBCH  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  mols  mols  mols  mols  mols  mols  mols  mols  % error  2  2.56E-09 0.00E+00 1.76E-10 2.45E-09 3.56E-09 8.76E-10 9.62E-09  9.82E-09  1.99  4  2.62E-09 0.00E+00 1.97E-10 2.67E-09 3.96E-09 2.89E-10 9.74E-09  9.82E-09  0.82  6  1.75E-09 0.00E+00 1.66E-10 2.06E-09 4.70E-09 9.28E-10 9.60E-09  9.82E-09  2.28  8  1.32E-09 0.00E+00 6.17E-10 1.40E-09 5.43E-09 1.17E-10 8.88E-09  9.82 E-09  9.53  10  1.38E-09 0.00E+00 6.69E-10 1.38E-09 5.59E-09 3.35E-10 9.35E-09  9.82E-09  4.78  12  1.30E-09 0.00E+00 6.69E-10 1.38E-09 5.95E-09 4.52E-10 9.75E-09  9.82E-09  0.72  235  Appendix C Experiment #11  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H : 25 ml/min  Catalyst: N i P / A l 0  Reaction period: 12 h  2  2  2  3  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :7.9 x 10" mol/h-gcat 2  (4,6-DMDBT+dodecane) : 0.04 ml/min  4,6-DMDBT Time, h  4,6-DMDBT  mol  Conversion m o l %  2  4.83E-09  50.81  4  2.01 E-09  79.51  6  1.94E-09  80.25  8  1.86E-09  81.03  10  1.83E-09  81.40  12  1.82 E-09  81.45  Avq  Time, h D M B C H mols  81.29  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  mols  mols  mols  mols  mols  mols  mols  % error  2  1.38 E-09 0.00E+00 9.44E-10 2.18E-10 6.33E-09  2.77E-10  9.14E-09  9.82E-09  6.87  4  1.52 E-09 0.00E+00 1.15E-09 2.20E-10 2.01 E-09  4.36E-09  9.26E-09  9.82E-09  5.65  6  3.70E-09 0.00E+00 2.69E-09 2.18E-10 1.94E-09  1.01 E-09  9.56E-09  9.82E-09  2.67  8  3.88E-09 0.00E+00 2.77E-09 2.22E-10 1.86E-09  1.02 E-09  9.76E-09  9.82E-09  0.62  10  4.31 E-09 0.00E+00 2.92E-09 2.23E-10 1.83E-09  4.21 E-10  9.70E-09  9.82E-09  1.24  12  4.48E-09 0.00E+00 2.98E-09 2.23E-10 1.82 E-09  1.02E-10  9.61 E-09  9.82 E-09  2.16  236  Appendix C Experiment #12  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H : 25 ml/min  Catalyst: Coo.4Ni P/Al 03  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :7.9 x 10" mol/h-gcat  2  2  2  2  (4,6-DMDBT+dodecane): 0.04 ml/min  4,6-DMDBT Time, h  4,6-DMDBT  mol  2  Conversion m o l %  2.86E-09  4  70.92  2.45E-09  6  75.08  2.06E-09  8  1.40E-09  79.01  10  1.30E-09  12  1.28E-09  86.95  AVG  86.49  Time, h D M B C H mols  85.78 86.73  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  • mols  mols  mols  mols  mols  mols  mols  % error  2  6.61 E-10 1.451 E-10 2E-10 2.313E-09 2.8551 E-09 3.2538E-09  9.38E-09  9.82E-09  4.43  4  1.47 E-09 1.392E-10 2E-10 3.04E-09 2.4474E-09 2.1303E-09  9.40E-09  9.82E-09  4.23  6  2.17E-09 1.629E-10 2E-10 3.167E-09 2.0614E-09 1.6293E-09  9.39E-09  9.82E-09  4.37  8  2.21 E-09 5.868E-11 2E-10 4.244E-09 1.3965E-09 1.3223E-09  9.48E-09  9.82E-09  3.47  10  2.26E-09 5.753E-11 2E-10 4.321 E-09 1.3026E-09 1.4175E-09  9.52E-09  9.82E-09  3.04  12  2.65E-09 5.649E-11 1E-10 4.79E-09 1.2814E-09 6.8365E-10  9.59E-09  9.82E-09  2.30  237  Appendix C Experiment #13 (Repeat run)  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H2 : 25 ml/min  Catalyst: C00.4N12P/AI2O3  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :7.9 x 10" mol/h-gcat 2  (4,6-DMDBT+dodecane): 0.04 ml/min  4,6-DMDBT Time, h  4,6-DMDBT  mol  2  Conversion m o l %  2.96E-09  4  69.88  2.48E-09  6  74.74  1.66E-09  10  83.07  1.51 E-09  12  1.42E-09  84.63 85.52  Avq  Time, h  84.41  DMBCH  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  mols  mols  mols  mols  mols  mols  mols  mols  % error  2  2.24E-09 2.06E-10 4.16E-10 1.98E-09 2.96E-09 1.75E-09 9.56E-09 9.82E-09  2.66  4  2.19E-09  1.79E-10 4.03E-10 2.88E-09 2.48E-09 1.39E-09 9.53E-09 9.82 E-09  2.98  6  2.13E-09  1.60E-10 5.17E-10 3.41 E-09 1.66E-09 1.41 E-09 9.29E-09 9.82E-09  5.40  10  2.12E-09  1.67E-10 4.02E-10 4.44E-09 1.51 E-09 1.06E-09 9.70E-09 9.82E-09  1.23  12  2.15E-09  1.63E-10 3.65E-10 5.08E-09 1.42 E-09 6.03E-10 9.78E-09 9.82E-09  0.37  238  Appendix C Experiment #14  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H : 25 ml/min  Catalyst: Coo.4Ni P/Al 03-F  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :7.9 x 10" mol/h-gcat  2  2  2  2  (4,6-DMDBT+dodecane): 0.04 ml/min  4,6-DMDBT Time, h  Time, h  4,6-DMDBT  mol  2  Conversion m o l %  3.59E-09  4  63.47  1.81 E-09  6  81.57  1.35E-09  10  1.23E-09  86.30  12  1.22E-09  87.62  AVG  87.13  87.46  DMBCH  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  mols  mols  mols  mols  mols  mols  mols  mols  % error  2  4.81E-10 3.05E-10 7.49E-10 2.28E-09 3.59E-09 1.98E-09 9.39E-09  9.82E-09  4.40  4  6.93E-10 3.90E-10 1.14E-09 2.55E-09 1.81 E-09 2.8E-09  9.39E-09  9.82E-09  4.42  6  8.33E-10 1.80E-10 1.33E-09 3.06E-09 1.35E-09 2.94E-09 9.69E-09  9.82E-09  1.32  8  1.45E-09 1.49E-10 4.91E-10 4.08E-09 1.08E-09 2.37E-09 9.62E-09  9.82E-09  2.00  10  1.60E-09 9.27E-11 4.83E-10 4.25E-09 8.45E-10 2.18E-09 9.44E-09  9.82E-09  3.81  239  Appendix C Experiment #15  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H : 25 ml/min  Catalyst: C o o . N i P / M C M  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  4  2  2  ~  F l o w rate  Space Velocity :7.9 x 10" mol/h-gcat  (4,6-DMDBT+dodecane): 0.04 ml/min  Time, h  4,6-DMDBT  4,6-DMDBT  Time, h  mol  Conversion m o l %  2  4.09E-10  95.83  4  1.71E-10  98.25  6  1.68E-10  98.29  8  1.84E-11  99.81  10  1.79E-11  99.82  12  1.18E-11  99.88  AVG  99.84  DMBCH  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  mols  mols  mols  mols  mols  mols  mols  mols  % error  2  6.00E-10  1.45E-10 9.59E-10 1.67E-10 4.09E-10 7.10E-09 9.38E-09 9.82E-09  4.48  4  5.93E-10  1.39E-10 1.04 E-09 1.72E-10 3.57E-10 7.20E-09 9.50E-09 9.82E-09  3.24  6  6.99E-10  1.63E-10 1.15E-09 1.69E-10 2.90E-10 7.20E-09 9.67E-09 9.82E-09  1.57  8  1.34E-09  1.81E-10 7.14E-10 1.85E-10 1.63E-10 7.19E-09 9.77E-09 9.82E-09  0.49  10  1.18E-09  2.80E-10 1.94E-10 1.88E-10 2.16E-11 7.59E-09 9.46E-09 9.82E-09  3.69  12  1.31 E-09  3.16E-10 2.72E-10 2.11E-10 1.18E-11 7.69E-09 9.82E-09 9.82E-09  0.04  240  Appendix C Experiment #16  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H : 25 ml/min  Catalyst: Pt-Coo.4Ni P/Al 0  Reaction period: 12 h  2  2  2  3  Pressure : 3.0 M P a  Temperature : 583 K  _  F l o w rate  Space Velocity :7.9 x 10" mol/h-gcat  (4,6-DMDBT+dodecane): 0.04 ml/min  4,6-DMDBT  4,6-DMDBT  mol  Conversion m o l %  2  6.79E-09  4  30.85  3.99E-09  59.41  6  3.18E-09  67.64  8  3.77E-10  96.16  10  2.35E-10  97.61  12  2.11E-10  97.85  Time, h  97.21  Time, h D M B C H  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  mols  mols  mols  mols  mols  mols  mols  mols  % error  2  2.18E-10  9.97E-11  2.00E-10  2.32E-09 6.79E-09 1.40E-10  9.77E-09  9.82E-09  0.49  4  1.32 E-09 9.57E-11  1.83E-10 2.93E-09 3.99E-09 8.56E-10  9.37E-09  9.82E-09  4.54  6  2.14E-09  2.09E-10  3.70E-09 3.18E-09 2.54E-10  9.60E-09  9.82E-09  2.27  8  4.26E-09 0.00E+00  0.00E+00 4.78E-09 3.77E-10 2.47E-11  9.44E-09  9.82E-09  3.88  10  4.33E-09 0.00E+00  0.00E+00 4.98E-09 2.35E-10 1.99E-11  9.57E-09  9.82E-09  2.58  12  4.34E-09 0.00E+00  0.00E+00 5.26E-09 2.11E-10 4.22E-12  9.81 E-09  9.82E-09  0.13  1.12E-10  241  Appendix C Experiment #17  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H2 : 25 ml/min  Catalyst: P t - C o o ^ P / A b O s - W A  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :7.9 x 10  >y '  mol/h-gcat  (4,6-DMDBT+dodecane): 0.04 ml/min  4,6-DMDBT Time, h  4,6-DMDBT  mol  Conversion m o l %  2  5.30E-09  4  '46.07  4.92E-09  6  3.73E-09  49.92  8  1.85E-09  10  6.30E-10  12  93.59  4.62E-10  95.29  62.03 81.12  AVG  Time, h D M B C H mols  90.00  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  mols  mols  mols  mols  mols  mols  mols  % error  2  1.86E-10 4.87E-11 3.33E-10  5.18E-10  5.30E-09 2.96E-09 9.34E-09  9.82E-09  4.91  4  2.72E-10  6.70E-11 3.06E-10  1.13E-09  4.92E-09 3.09E-09 9.79E-09  9.82E-09  0.33  6  2.97E-10  6.55E-11 3.05E-10  1.82 E-09  3.73E-09 3.11 E-09 9.33E-09  9.82E-09  4.99  8  2.19E-09  6.63E-11 2.72E-10  3.80E-09  1.85E-09 1.41 E-09 9.60E-09  9.82E-09  2.20  10  2.79 E-09 5.88E-11 1.59E-11 4.01 E-09  6.30E-10 1.84 E-09 9.35E-09  9.82E-09  4.77  12  2.96E-09  4.62E-10 1.08E-09 9.62E-09  9.82E-09  2.07  5.59E-11 1.28E-11 5.05E-09  242  Appendix C Solvent: Dodecane  Experiment # 18 Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H : 15 ml/min  Catalyst: Pt-Coo.4Ni P/Al 03  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity : 9.2 x 10" mol/h-gcat  2  2  2  2  (4,6-DMDBT+dodecane): 0.156 ml/min  4,6-DMDBT mol 4.76E-09 3.32E-09 1.86E-09 9.51 E-10 5.28E-10 4.30E-10 AVG  Time, h 2 4 6 8 10 12  Time, h DMBCH DMBU  MCHT  mols  mols  mols  DMBP DMDBT mols  mols  4,6-DMDBT Conversion mol% 51.51 66.24 81.08 90.32 94.62 95.62 93.52  Others  total  Initial  mol balance  mols  mols  mols  % error  2  3.07E-11 4.91 E-11 1.18E-10 1.42E-104.76E-09 4.23E-09  9.33E-09  9.82E-09  4.99  4  4.21 E-10 4.30E-11 2.14E-10 4.33E-10 3.32E-09 5.31 E-09  9.74E-09  9.82E-09  0.84  6  5.44E-10 4.21 E-11 1.97E-10 2.35E-09 1.86E-09 4.47E-09  9.46E-09  9.82E-09  3.61  8  8.17E-10 1.58E-11 1.39E-10 2.50E-09 9.51E-10 5.11 E-09  9.53E-09  9.82E-09  2.93  10  2.81 E-09 1.03E-11 1.12E-10 5.03E-09 5.28E-10 1.12E-09  9.60E-09  9.82E-09  2.19  12  2.92E-09 9.32E-12 4.34E-11 5.34E-09 4.30E-10 8.28E-10  9.57E-09  9.82E-09  2.51  243  Appendix C Experiment #19  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H : 60 ml/min  Catalyst: Pt-Coo.4Ni P/Al 03  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity : 3.2 x 10"' mol/h-gcat  2  2  2  (4,6-DMDBT+dodecane) : 0.624 m l / m i n '  4,6-DMDBT Time, h  4,6-DMDBT  mol  2  Conversion m o l %  4.96E-09  4  49.46  4.33E-09  6  4.12E-09  55.93  8  3.74E-09  10  61.95  3.62E-09  12  63.17  3.41 E-09  65.31  AVG  63.48  Time, h D M B C H mols  58.00  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  mols  mols  mols  mols  mols  mols  mols  % error  2  3.86E-10 1.31E-10 2.07E-10  5.23E-10  4.96E-09 3.28E-09  9.49E-09  9.82E-09  3.32  4  5.91 E-10 1.23E-10 1.67E-10 6.61 E-10  4.33E-09 3.48E-09  9.35E-09  9.82E-09  4.82  6  6.02E-10 1.07E-10 1.52E-10 8.61 E-10  4.12E-09 3.62E-09  9.47E-09  9.82E-09  3.61  8  8.29E-10 9.39E-11 4.78E-11 2.80E-09  3.74E-09 1.83E-09  9.33E-09  9.82E-09  4.94  10  1.07 E-09 8.72E-11 4.34E-11 3.30E-09  3.62E-09 1.24E-09  9.37E-09  9.82E-09  4.61  12  1.21 E-09 8.29E-11 4.20E-11 3.60E-09  3.41 E-09 9.87E-10  9.33E-09  9.82E-09  4.97  244  Appendix C Experiment # 20  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H : 100 ml/min  Catalyst: Pt-Coo.4Ni P/Al 03  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity : 6.2 x 10"' mol/h-gcat  2  2  2  (4,6-DMDBT+dodecane) : 1.04 ml/min  Time, h  4,6-DMDBT Time, h  4,6-DMDBT  mol  2  Conversion m o l %  8.32E-09  4  15.26  5.53E-09  6  43.64  5.37E-09  8  5.01 E-09  45.29  10  4.63E-09  12  52.83  4.59E-09  53.28  AVG  51.69  48.94  DMBCH  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  mols  mols  mols  mols  mols  mols  mols  mols  % error  2  4.13E-10  4.91 E-11 3.33E-10  4.78E-10  7.90E-09 2.33E-10 9.40E-09 9.82E-09  4.25  4  5.06E-10  4.30E-11 3.06E-10  6.34E-10  5.53E-09 2.28E-09 9.30E-09 9.82E-09  5.30  6  5.44E-10  4.21 E-11 3.48E-10  8.89E-10  5.37E-09 2.12E-09 9.31 E-09 9.82E-09  5.14  8  5.75E-10  3.70E-11 7.30E-11 2.50E-09  5.01 E-09 1.57E-09 9.77E-09 9.82E-09  0.49  10  6.92E-10  3.58E-11 4.98E-11 2.84E-09  4.63E-09 1.52 E-09 9.77E-09 9.82E-09  0.50  12  8.00E-10  3.48E-11 4.17E-11 2.96E-09  4.59E-09 1.37E-09 9.79E-09 9.82E-09  0.33  245  Appendix C Experiment #21  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H2 : 25 ml/min  Catalyst: P t - C o o ^ P / A b C b  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 548 K  F l o w rate  Space Velocity :7.9 x 10" mol/h-gcat 2  (4,6-DMDBT+dodecane) : 0.04 ml/min  4,6-DMDBT  Time, h  4,6-DMDBT  Time, h  mol  2  6.45E-09  4  5.00E-09  6  3.83E-09  8  3.48E-09  10  1.81 E-09  12  1.61 E-09  83.56  AVG  76.58  Conversion m o l % 34.31 49.08 60.95 64.59 81.60  DMBCH  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  mols  mols  mols  mols  mols  mols  mols  mols  % error  2  1.01 E-09  1.66E-10 1.80E-10 1.38E-09 6.45E-09  1.50E-10 9.34 E-09  9.82E-09  4.87  4  1.36E-09  1.60E-10 2.07E-10 1.87E-09 5.00E-09  1.12E-09 9.71 E-09  9.82E-09  1.07  6  1.36E-09  1.87E-10 2.32E-10 2.05E-09 3.83E-09  1.68E-09 9.34E-09  9.82E-09  4.87  8  1.41 E-09  6.73E-11 1.09E-10 2.63E-09 4.75E-09 3.67E-10  9.33E-09  9.82E-09  4.97  10  1.74E-09  6.59E-11 9.55E-11 3.01 E-09 1.81 E-09 2.62E-09  9.34E-09  9.82E-09  4.89  12  1.76E-09  6.48E-11 6.47E-11 5.93E-09 1.61 E-09 3.42E-10  9.78E-09  9.82E-09  0.43  246  Appendix C Experiment # 22  Solvent: Dodecane  Reactant: 4 , 6 - D M D B T (3000 ppm)  Flowrate H : 25 ml/min  Catalyst: Pt-Co .4Ni P/Al O  Reaction period: 12 h  0  2  2  2  3  Pressure : 3.0 M P a  Temperature : 533 K  F l o w rate  Space Velocity :7.9 x 10" mol/h-gcat 2  (4,6-DMDBT+dodecane): 0.04 ml/min  4,6-DMDBT  Time, h  4,6-DMDBT  Time, h  mol  2  Conversion m o l %  5.49E-09  4  44.05  4.42E-09  54.97  6  •4.07E-09  58.57  8  3.58E-09  10  63.51  3.49E-09  12  64.48  3.41 E-09  65.30  Avg  64.43  DMBCH  DMBU  MCHT  DMBP  DMDBT  Others  total  Initial  mol balance  mols  mols  mols  mols  mols  mols  mols  mols  % error  2  1.66 E-09  2.55E-10 6.25E-11 1.71 E-09 5.49E-09 3.12E-10  9.49E-09  9.81944E-09  3.32  4  1.69E-09  7.18E-10 5.26E-10 1.80E-09 4.42E-09 3.97E-10  9.55E-09  9.81944E-09  2.79  6  1.40E-09  1.91 E-10 1.64E-10 2.13E-09 4.07E-09 1.68E-09  9.64E-09  9.81944E-09  1.84  8  1.21 E-09  1.91E-10 1.37E-10 2.51 E-09 3.58E-09 1.64E-09  9.27E-09  9.81944E-09  5.59  10  1.24E-09  1.91 E-10 9.55E-11 2.63E-09 3.49E-09 1.77E-09  9.41 E-09  9.81944E-09  4.14  12  1.27E-09  1.91 E-10 6.47E-11 3.00E-09 3.41 E-09 1.66E-09  9.59E-09  9.81944E-09  2.33  247  Appendix C Sample profiles of time on stream versus product ratio  Time-on- stream, h  1 •DMBCH A MCHT • DMBP ODMDBT  •  _—  x  ii  *  ft  2  4  6  *  • -  *  t  8  10  O  12  Time-on-stream,h,  Figure C 2 Profile o f product ratio using N i P / A l 0 (A) and C o . N i P / A l O (B) 2  2  3  0  4  2  2  3  for the H D S o f 3000 ppm of 4 , 6 - D M D B T at 583 K and 3.0 M P a  248  Appendix D  Appendix D Hydrodenitrogenation Experiments D.l  Summary of hydrodenitrogenation experiments  Table D l  Summary of experiments for hydrodenitrogenation of carbazole (3000 ppm) over transition metal phosphides at 3.0 MPa, H 2 pressure for 12 h time on stream  Exp.  Reactant  Catalyst  23  Carbazole  MoP  Temp. SV K xlO" mol/h-g 583 8.6  24  Carbazole  Nio.oyMoP  583  8.6  90.0  25  Carbazole  Ni .i6MoP  583  8.6  89.6  26  Carbazole  Ni .38MoP  583  8.6  88.8  27  Carbazole  Nii.nMoP  583  8.6  85.8  28  Carbazole  Co .4Ni2P/Al O3  583  8.6  82.0  29  Carbazole  C00.4N12P/AI2O3-F  583  8.6  94.0  30  Carbazole  Co .4Ni P/MCM  583  8.6  98.0  31  Carbazole  Nio.33MoP/Al 0  583  8.6  96.0  32  Carbazole  M0.33M0P/MCM  583  8.6  100  33  Carbazole  N10.33M0P (Repeat)  583  8.6  87.5  34  Carbazole  Nio.33MoP/Al 0  3  523  8.6  69.1  35  Carbazole  Nio.33MoP/Al 0  3  523  15.0  44.9  36  Carbazole  M0.33M0P/AI2O3  523  61.0  3.57  37  Carbazole  Nio.33MoP/Al 0  523  8.6  21.1  #  Conversion  2  0  0  0  0  2  2  2  3  2  2  2  3  249  54.5  Appendix D D.2  Response Factor and sample calculation of H D N activity Response factor o f the carbazole i n xylene was obtained by injecting known composition  of 0.1 umol o f the liquid sample into gas chromatography ( G C ) . Four samples o f each fixed composition were injected and the average reading was used to plot the graph o f concentration versus the G C area. A s shown in Figure D l , the slope = 6.283 x 10" mol/area was used as the 21  response factor.  5.E-14 -| 5.E-14 -  y = 6.283E-21x  4.E-14 -  R* = 0.9981  S  4.E-14 3.E-14 3.E-14 2.E-14 2.E-14 1.E-14 5.E-15 0.E+00 0.E+00  1  1  1  2.E+06  4.E+06  6.E+06  8.E+06  x = G C Area  Figure D . 1  Calibration curve for carbazole used to determine the response factor  Similarly response factor o f the bicyclohexane ( B C H X ) product i n xylene was obtained by injecting known composition o f 0.1 umol o f the liquid sample into the G C . Four samples o f each fixed composition were also injected and the average reading was used to plot the graph o f  250  Appendix D concentration versus the G C area. A s shown in Figure D 2 , the slope = 6.283 x 10  mol/area  was used as the response factor.  Figure D 2  Calibration curve for B C H X used to determine the response factor  In calculating the mols, the R F factor o f carbazole was also used for tetrahydrocarbazole ( T H C Z ) and for the products the response factor for bicyclohexane ( B C H X ) was used to calculate the mols.  Sample calculation for the H D N of Carbazole. Data for the experiment 24 is used for illustrating the calculations as shown on Table D 2  251  Appendix D Table D 2  Time h  Results o f hydrodenitrogenation o f carbazole over N10.07M0P  Carbazole Carbazole  BCHX  BCHX  CPHX  CPHX  THCZ  THCZ  Area  Area  mol  Area  mol  Area  mol  mol  2  2609377  1.64E-14  56756  9.14E-16  91099  1.47E-15  103913  6.53E-16  4  1317044  8.28E-15  650823  1.05E-14  109832  1.77E-15  109949  6.91 E-16  6  1123984  7.07E-15  786951  1.27E-14  22453  3.61E-16  107733  6.77E-16  8  438865  2.32E-15 1098161  1.77E-14  24539  3.95E-16  103725  6.52E-16  10  437328  2.31E-15 1101313  1.77E-14  28673  4.62E-16  103034  6.48E-16  12  432245  2.28E-15 1108004  1.78E-14  30637  4.93E-16  100149  6.30E-16  Time  <C12  < C12  >C12  > C12  Total  Initial  h  Area .  mol  Area  mol  mol  mol  2  205218  3.30E-15  4  109833  1.77E-15  6  59644  9.60E-16  8  61866  10 12  0  mol balance %error  0  2.27E-14  2.30E-14  1.28  0  2.30E-14  2.30E-14  0.22  0  0  2.17E-14  2.30E-14  5.65  9.96E-16  0  0  2.25E-14  2.30E-14  2.40  40617  6.54E-16  0  0  2.22E-14  2.30E-14  3.44  40078  6.45E-16  0  0  2.23E-14  2.30E-14  3.09  C < 12 : Products formed with carbon number less than 12 C >12 : Products formed with carbon number less than 12  252  Appendix D The initial concentration o f carbazole = average area o f carbazole x response factor o f carbazole = 3564514 area x 6.29 x 10" mol/area 21  = 2.304 x 10" m o l 14  Conversion o f 4 , 6 - D M D B T after 12 h _ [Carbazole]  t=0  -[Carbazole]  1=t  [Carbazole]lt=0 2.304-0.228 2.304  x 100  = 90.0 % The m o l balance, % error [Carbazole]  t=0  -[Carbazole + all products] [Carbazole]  2.304-2.19 2.19  t=12h x100  t=0  x 100  = 5.00%  253  Appendix D D.3 Repeatability of hydrodenitrogenation experiments Table D 3 Repeatability o f hydrodenitrogenation over N10.07M0P using carbazole Experiment # 24 & 33  Solvent: Xylene  Reactant: Carbazole (3000 ppm)  Flowrate H2 : 25 ml/min  Catalyst: M0.07M0P  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :8.6 x 10" mol/h-gcat 2  (4,6-DMDBT+dodecane): 0.04 ml/min  Time  mols Carbazole  mols Carbazole  Average  S.E.  h  run 1  run 2  mols  mol  %S.E.  2  1.64E-14  1.76E-14  1.70E-14  4  8.28E-15  0.05  7.97E-15  8.13E-15  0.01  6  7.07E-15  6.27E-15  1.28  6.67E-15  0.03  3.33  4.99  8  2.32E-15  1.82E-15  10  2.31E-15  2.07E-15  0.02  2.01E-15  2.08  2.16E-15  0.01  1.25  12  2.28E-15  1.10E-15  1.69E-15  0.05  4.91  Time  mols B C H X  mols B C H X  Average  S.E.  %S.E.  h  run 1  run 2  mols  2  mol  9.14E-16  9.93E-16  9.54E-16  4  5.91 E-02  1.05E-14  1.14E-14  1.09E-14  5.91  6.04E-02  6.04  6  1.27E-14  1.17E-14  1.22E-14  1.77E-14  5.55E-02  8  1.66E-14  1.71E-14  5.55  4.47E-02  10  1.77E-14  4.47  1.66E-14  1.72E-14  4.60E-02  4.60  12  1.78E-14  1.79E-14  1.79E-14  2.08E-03  0.21  254  Appendix D Time  mols C P H X  mols C P H X  Average  h  run 1  run 2  mols  mol  2  1.47E-15  1.31E-15  1.39E-15  7.83E-02  4  1.77E-15  1.72E-15  1.74E-15  7.83  1.89E-02  6  3.61 E-16  3.85E-16  1.89  3.73E-16  4.51 E-02  8  3.95E-16  3.60E-16  4.51  3.78E-16  6.55E-02  10  4.62E-16  4.13E-16  6.55  4.37E-16  7.91 E-02  7.91  12  4.93E-16  4.33E-16  4.63E-16  9.27E-02  9.27  Time  mols T H C Z  mols T H C Z  Average  S.E.  h  run 1  run 2  %S.E.  mols  mol  2  6.53E-16  6.12E-16  4  6.91E-16  6.33E-16  4.57E-02  6.15E-16  6.53E-16  4.57  8.30E-02  6  6.77E-16  6.65E-16  8.30  6.71 E-16  8  6.52E-16  1.26E-02  5.91 E-16  6.22E-16  1.26  6.89E-02  10  6.48E-16  6.88E-16  6.89  6.68E-16  12  4.23E-02  6.30E-16  6.22E-16  4.23  6.26E-16  8.04E-03  0.80  Conversion  S.E.  %S.E.  run 1  Run2  Average  S.E.  90.01  87.53  88.77  %S.E.  0.02  1.97  255  Appendix D D.4  Data for hydrodenitrogenation experiments  Experiment #23  Solvent: Xylene  Reactant: Carbazole (3000 ppm)  Flowrate H2: 25 ml/min  Catalyst: M o P  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :8.6 x 10" mol/h-gcat 2  (4,6-DMDBT+dodecane) : 0.04 ml/min  Carbazole  Carbazole  Time, h  mol  Conversion m o l %  2  7.95E-15  65.50  4  8.09E-15  64.88  6  8.80E-15  61.79  8  1.04E-14  54.87  10  1.05E-14  54.38  12  1.05E-14  54.26  Avg  54.50  Time  BCHX  CPHX  THCZ  <C12  >C12  Total  Initial  h  mol  mol  mol  mol  mol  mol  mol  mol balance %error  2  9.08E-15  4.64E-16  6.53E-16  9.90E-16  3.13E-15  2.23E-14  2.30E-14  3.37  4  7.82E-15  4.76E-16  6.91 E-16  1.16E-15  3.54E-15  2.18E-14  2.30E-14  5.47  6  6.86E-15  5.02E-16  6.77E-16  1.44E-15  3.93E-15  2.22E-14  2.30E-14  3.59  8  2.75E-15  5.33E-16  6.52E-16  6.63E-15  1.62E-15  2.26E-14  2.30E-14  2.02  10  2.56E-15  5.34E-16  6.48E-16  6.55E-15  1.70E-15  2.25E-14  2.30E-14  2.27  12  2.42E-15  5.56E-16  6.30E-16  5.78E-15  1.82E-15  2.17E-14  2.30E-14  5.66  256  Appendix D Experiment # 25  Solvent: Xylene  Reactant: Carbazole (3000 ppm)  Flowrate H2 : 25 ml/min  Catalyst: Nio.ieMoP  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :8.6 x 10" mol/h-gcat 2  (4,6-DMDBT+dodecane) : 0.04 ml/min  Time, h  Carbazole  Carbazole  mol  Conversion m o l %  2  4.84E-15  4  78.99  3.94E-15  82.88  6  2.55E-15  88.93  10  • •  12  2.52E-15  89.06  2.14E-15  90.70  Avg  89.56  Time  BCHX  CPHX  THCZ  <C12  > C12  Total  Initial  h  mol  mol  mol  mol  mol  mol  mol  mol balance %error  2  1.28E-14  2.13E-16  6.53E-16  1.21 E-16  3.51 E-15  2.22E-14  2.30E-14  3.81  4  1.43E-14  8.81 E-17  6.91 E-16  2.08E-17  3.57E-15  2.26E-14  2.30E-14  1.92  6  1.60E-14  1.32E-16  6.77E-16  3.31E-17  3.41 E-15  2.28E-14  2.30E-14  1.19  10  1.61E-14  1.56E-16  6.52E-16  3.40E-17  3.41E-15  2.29E-14  2.30E-14  0.76  12  1.61E-14  1.26E-16  6.48E-16  3.84E-17  3.73E-15  2.28E-14  2.30E-14  1.09  257  Appendix D Experiment # 26  Solvent: Xylene  Reactant: Carbazole (3000 ppm)  Flowrate H2 : 25 ml/min  Catalyst: N i . M o P  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :8.6 x 10" mol/h-gcat  0  3 8  2  (4,6-DMDBT+dodecane): 0.04 ml/min  Carbazole  Carbazole  Time, h  mol  2  Conversion m o l %  3.89E-15  4  83.10  3.05E-15  6  2.97E-15  86.75  10  2.53E-15  12  2.24E-15  87.11 89.00 90.30  Avg  88.80  Time  BCHX  CPHX  THCZ  <C12  > C12  Total  Initial  h  mol  mol  mol  mol  mol  mol  mol  mol balance %error  2  7.58E-15  4.83E-16  6.53E-16  9.37E-15  0.00E+00  2.20E-14  2.30E-14  4.59  4  8.97E-15  4.21 E-16  6.91 E-16  8.39E-15  3.79E-16  2.19E-14  2.30E-14  4.88  6  1.02E-14  3.59E-16  6.77E-16  7.40E-15  1.09E-15  2.27E-14  2.30E-14  1.37  10  1.04E-14  4.37E-16  6.52E-16  7.40E-15  3.95E-16  2.19E-14  2.30E-14  5.08  12  1.08E-14  5.22E-16  6.48E-16  6.70E-15  1.04E-15  2.19E-14  2.30E-14  4.93  258  Appendix D Experiment # 27  Solvent: Xylene  Reactant: Carbazole (3000 ppm)  Flowrate H2 : 25 ml/min  Catalyst: Nii.nMoP  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :8.6 x 10" mol/h-gcat 2  (4,6-DMDBT+dodecane): 0.04 ml/min  Carbazole  Carbazole  Time, h  mol  Conversion m o l %  2  8.01 E-15  65.22  4  5.44E-15  76.40  6  4.52E-15  80.38  8  3.29E-15  85.73  10  3.28E-15  85.77  12  3.22E-15  86.02  Avg  85.84  Time  BCHX  CPHX  THCZ  <C12  >C12  Total  Initial  h  mol  mol  mol  mol  mol  mol  mol  mol balance %error  2  3.22E-20  4.64E-16  6.53E-16  3.05E-15  4.66E-15  2.20E-14  2.30E-14  4.70  4  6.44E-20  4.76E-16  6.91 E-16  3.85E-15  5.20E-15  2.23E-14  2.30E-14  3.18  6  9.66E-20  5.02E-16  6.77E-16  4.90E-15  4.89E-15  2.27E-14  2.30E-14  1.64  8  1.29E-19  5.33E-16  6.52E-16  5.06E-15  5.21E-15  2.27E-14  2.30E-14  1.29  10  1.61E-19  5.34E-16  6.48E-16  5.29E-15  5.18E-15  2.29E-14  2.30E-14  0.53  12  1.93E-19  5.56E-16  6.30E-16  5.39E-15  4.97E-15  2.28E-14  2.30E-14  1.17  259  Appendix D Experiment #28  Solvent: Xylene  Reactant: Carbazole (3000 ppm)  Flowrate H : 25 ml/min  Catalyst: Coo.4Ni P/Al 03  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :8.6 x 10  2  2  2  mol/h-gcat  (4,6-DMDBT+dodecane): 0.04 ml/min  Carbazole Time, h 2  Carbazole  mol  Conversion m o l %  7.47E-15  67.56  4  6.98E-15  6  6.12E-15  8  73.44  4.84E-15  10  78.98  4.29E-15  12  81.36  3.33E-15  85.55  69.68  Avg  81.96  Time  BCHX  CPHX  THCZ  <C12  >C12  h  mol  mol  mol  mol  mol  .  Total  ,  mol  Initial mol  mol balance %error  2  6.10E-15  2.96E-15  5.31E-15  1.61E-16  8.75E-16  2.29E-14  2.30E-14  0.67  4  7.67E-15  2.52E-15  3.37E-15  2.22E-16  1.82E-15  2.26E-14  2.30E-14  1.92  6  1.08E-14  9.43E-16  2.89E-15  1.62E-15  1.22E-16  2.25E-14  2.30E-14  2.26  8  1.25E-14  1.21E-15  1.86E-15  1.58E-15  1.81E-16  2.22E-14  2.30E-14  3.70  10  1.40E-14  7.44E-16  1.73E-15  1.55E-15  1.95E-16  2.25E-14  2.30E-14  2.37  12  1.46E-14  9.33E-16  1.84E-15  1.44E-15  1.81E-16  2.24E-14  2.30E-14  2.96  260  Appendix D Experiment #29  Solvent: Xylene  Reactant: Carbazole (3000 ppm)  Flowrate H2 : 25 ml/min  Catalyst: Coo.4Ni P/Al20 _F  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :8.6 x 10" mol/h-gcat  2  3  (4,6-DMDBT+dodecane) : 0.04 ml/min  Carbazole  Carbazole  Time, h  mol  Conversion m o l %  2  2.81E-15  87.82  4  1.23E-15  94.68  6  1.07E-15  95.34  8  1.15E-15  95.02  10  1.40E-15  93.93  12  1.60E-15  93.07  Avg  94.01  Time  BCHX  CPHX  THCZ  < C12  > C12  Total  Initial  h  mol  mol  mol  mol  mol  mol  mol  mol balance %error  2  7.93E-15  3.93E-15  1.77E-15  3.90E-15  3.79E-16  2.07E-14  2.30E-14  10.06  4  1.36E-14  3.95E-15  2.29E-15  3.78E-17  4.00E-16  2.15E-14  2.30E-14  6.65  6  1.64E-14  2.69E-15  1.83E-15  2.16E-17  2.17E-16  2.22E-14  2.30E-14  3.67  8  1.74E-14  1.70E-15  1.74E-15  0.00E+00  2.34E-16  2.22E-14  2.30E-14  3.57  10  1.74E-14  1.70E-15  1.56E-15  1.99E-17  0.00E+00  2.21 E-14  2.30E-14  4.08  12  1.75E-14  1.61E-15  1.45E-15  0.00E+00  0.00E+00  2.22E-14  2.30E-14  3.71  261  Appendix D Experiment #30  Solvent: Xylene  Reactant: Carbazole (3000 ppm)  Flowrate H : 25 ml/min  Catalyst: C o o . N i P / M C M  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :8.6 x 10" mol/h-gcat  4  2  2  2  (4,6-DMDBT+dodecane): 0.04 ml/min  Carbazole Time, h  Carbazole  mol  Conversion m o l %  2  3.44E-15  4  85.05  4.13E-15  82.07  6  2.84E-15  87.66  8  9.23E-16  10  95.99  5.75E-16  97.50  12  7.00E-17  99.70  Avq  97.73  Time  BCHX  CPHX  THCZ  <C12  > C12  Total  Initial  h  mol  mol  mol  mol  mol  mol  mol  mol balance %error  2  1.68E-14  7.74E-16  8.57E-16  1.34E-16  7.56E-17  2.21 E-14  2.30E-14  3.99  4  7.26E-15  6.61 E-16  1.49E-15  9.00E-15  2.49E-16  2.28E-14  2.30E-14  1.02  6  3.37E-15  1.28E-15  1.41E-15  8.27E-15  4.69E-15  2.19E-14  2.30E-14  5.09  8  2.35E-15  8.27E-16  1.27E-15  1.10E-14  5.36E-15  2.17E-14  2.30E-14  5.73  10  1.99E-15  9.19E-16  2.87E-17  1.60E-14  1.47E-15  2.10E-14  2.30E-14  8.79  12  1.99E-15  2.79E-15  6.20E-18  1.51E-14  1.28E-15  2.12E-14  2.30E-14  8.00  262  Appendix D Experiment # 31  Solvent: Xylene  Reactant: Carbazole (3000 ppm)  Flowrate H 2 : 25 ml/min  Catalyst: M 0 . 3 3 M 0 P / A I 2 O 3  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :8.6 x 10" mol/h-gcat 2  (4,6-DMDBT+dodecane): 0.04 ml/min  Carbazole Time, h  Carbazole  mol  2  Conversion m o l %  9.39E-15  4  59.25  7.25E-15  6  68.55  5.65E-15  8  75.47  1.33E-15  10  94.22  8.34E-16  12  96.38  8.36E-16  96.37  Avg  95.66  Time  BCHX  CPHX  THCZ  <C12  >C12  Total  Initial  h  mol  mol  mol  mol  mol  mol  mol  mol balance %error  2  2.35E-15  2.52E-16  2.47E-15  3.98E-15  3.51E-15  2.20E-14  2.30E-14  4.70  4  8.25E-15  1.70E-16  4.45E-15  1.20E-15  9.08E-18  2.13E-14  2.30E-14  7.43  6  1.31E-14  6.80E-17  3.06E-15  1.09E-16  1.30E-17  2.20E-14  2.30E-14  4.65  8  1.78E-14  1.07E-15  7.52E-16  3.42E-16  1.21 E-16  2.14E-14  2.30E-14  7.13  10  1.83E-14  1.13E-15  6.05E-16  1.97E-16  1.05E-16  2.11E-14  2.30E-14  8.19  12  1.85E-14  8.85E-16  6.21E-16  3.77E-16  2.17E-17  2.12E-14  2.30E-14  7.80  263  Appendix D Experiment #32  Solvent: Xylene  Reactant: Carbazole (3000 ppm)  Flowrate H2 : 25 ml/min  Catalyst: NI0.33M0P/MCM  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 583 K  F l o w rate  Space Velocity :8.6 x 10" mol/h-gcat 2  (4,6-DMDBT+dodecane): 0.04 ml/min  Carbazole  Carbazole  mol  Conversion m o l %  Time, h 2  1.24E-15  4  94.61  0.00E+00  100.00  6  0.00E+00  10  0.00E+00  100.00  12  0.00E+00  100.00  Avg  100.00  100.00  Time  BCHX  CPHX  THCZ  <C12  > C12  Total  Initial  h  mol  mol  mol  mol  mol  mol  mol  mol balance %error  2  1.44E-14  4.79E-15  6.53E-16  0.00E+00  0.00E+00  2.11E-14  2.30E-14  8.35  4  1.03E-14  2.71 E-15  6.91E-16  6.91 E-16  6.33E-15  2.08E-14  2.30E-14  9.85  6  3.53E-15  1.74E-15  6.77E-16  1.42E-14  1.84E-15  2.20E-14 •  2.30E-14  4.44  10  2.91E-15  1.95E-15  6.52E-16  1.26E-14  3.09E-15  2.12E-14  2.30E-14  8.00  12  2.68E-15  0.00E+00  6.48E-16  1.40E-14  4.67E-15  2.20E-14  2.30E-14  4.57  264  Appendix D Experiment #34  Solvent: Xylene  Reactant: Carbazole (3000 ppm)  Flowrate H : 25 ml/min  Catalyst: N i . 3 M o P / M C M  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 523 K  F l o w rate  Space Velocity : 8.6 x 10" mol/h-gcat  0  2  3  (4,6-DMDBT+dodecane) : 0.04 ml/min  Carbazole  Carbazole  Time, h  mol  Conversion m o l % 40.55  2  1.37E-14  4  1.09E-14  52.80  6  8.12E-15  64.73  10  6.68E-15  71.02  12  6.56E-15  • 71.52  Avg  69.09  Time  BCHX  CPHX  THCZ  <C12  > C12  Total  Initial  h  mol  mol  mol  mol  mol  mol  mol  mol balance %error  2  5.97E-16  5.11E-16  3.09E-.16  3.55E-15  3.66E-15  2.23E-14  2.30E-14  3.11  4  1.14E-15  5.72E-16  1.34E-15  4.39E-15  4.43E-15  2.28E-14  2.30E-14  1.23  6  4.46E-15  6.34E-16  4.23E-15  1.67E-15  3.39E-15  2.25E-14  2.30E-14  2.28  10  6.20E-15  6.51E-16  4.19E-15  1.82E-15  2.91E-15  2.24E-14  2.30E-14  2.54  12  7.87E-15  6.59E-16  4.04E-15  1.76E-15  1.37E-15  2.23E-14  2.30E-14  3.32  265  Appendix D Experiment #35  Solvent: Xylene  Reactant: Carbazole (3000 ppm)  Flowrate H : 50 ml/min  Catalyst: Ni .33MoP/Al O3  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 523 K  F l o w rate  Space Velocity :1.5 x 10"' mol/h-gcat  0  2  2  (4,6-DMDBT+dodecane): 1.77 ml/min  Carbazole Time, h  Carbazole  mol  2  Conversion m o l %  1.90E-14  4  17.41  1.80E-14  6  22.06  1.36E-14  10  1.27E-14  41.13  12  1.18E-14  48.67  Avg  44.91  44.94  Time  BCHX  CPHX  THCZ  <C12  > C12  Total  Initial  h  mol  mol  mol  mol  mol  mol  mol  mol balance %error  2  6.92E-16  3.10E-16  4.39E-15  1.43E-15  1.38E-15  2.28E-14  2.30E-14  0.89  4  2.00E-15  5.24E-16  7.96E-15  1.47E-15  6.58E-16  2.26E-14  2.30E-14  1.85  6  2.79E-15  2.80E-16  6.05E-15  1.75E-15  2.54E-15  2.09E-14  2.30E-14  9.20  10  2.97E-15  5.41 E-16  5.52E-15  1.68E-15  4.23E-15  2.21 E-14  2.30E-14  4.04  12  3.20E-15  1.10E-15  4.69E-15  1.47E-15  4.82E-15  2.24E-14  2.30E-14  2.71  266  Appendix D Experiment #36  Solvent: Xylene  Reactant: Carbazole (3000 ppm)  Flowrate H2: 100 ml/min  Catalyst: M 0 . 3 3 M 0 P / A I 2 O 3  Reaction period: 12 h  Pressure : 3.0 M P a  Temperature : 523 K  F l o w rate  Space Velocity :6.1 x 10"' mol/h-gcat  (4,6-DMDBT+dodecane): 7.08 ml/min  Carbazole Time, h  Carbazole  mol  2  Conversion m o l %  2.22E-14  4  3.50  2.18E-14  6  2.22E-14  10  2.22E-14  12  2.22E-14  5.37 3.49 3.46 3.77  Avg  3.57  Time  BCHX  CPHX  THCZ  <C12  > C12  Total  Initial  h  mol  mol  mol  mol  mol  mol  mol  mol balance %error  2  1.04E-16  0.00E+00  3.37E-15  4.16E-17  1.57E-16  2.25E-14  2.30E-14  2.18  4  2.30E-16  3.01E-17  6.21E-15  4.78E-17  3.23E-16  2.24E-14  2.30E-14  2.63  6  5.50E-17  3.03E-17  7.07E-15  7.95E-17  4.09E-16  2.28E-14  2.30E-14  1.01  10  5.30E-17  3.39E-17  7.30E-15  8.95E-17  3.10E-16  2.27E-14  2.30E-14  1.35  12  5.23E-17  3.37E-17  7.35E-15  9.00E-17  1.94E-16  2.25E-14  2.30E-14  2.16  267  Appendix D Experiment #37  Solvent: Xylene  Reactant: Carbazole (3000 ppm)  Flowrate H : 25 ml/min  Catalyst: N i o . 3 M o P / A l 0  Reaction period: 12 h  3  2  2  3  Pressure : 3.0 M P a  Temperature : 423 K  F l o w rate  Space Velocity :8.62 x 10"' mol/h-gcat  (4,6-DMDBT+dodecane) : 0.04 ml/min  Carbazole  Carbazole  Time, h  mol  2  Conversion m o l %  8.86E-15  4  61.54  1.11E-14  51.84  6  1.80E-14  10  21.78  1.82E-14  21.14  12  1.83E-14  20.39  Avg  21.11  Time  BCHX  CPHX  THCZ  < C12  >C12  Total  Initial  h  mol  mol  mol  mol  mol  mol  mol  mol balance %error  2  6.92E-16  3.10E-16  4.39E-15  1.43E-15  1.38E-15  2.21 E-14  2.30E-14  4.26  4  2.00E-15  5.24E-16  7.96E-15  1.47E-15  6.58E-16  2.26E-14  2.30E-14  2.10  6  2.79E-15  2.80E-16  6.05E-15  1.75E-15  2.54E-15  2.15E-14  2.30E-14  6.52  10  2.97E-15  5.41 E-16  5.52E-15  1.68E-15  4.23E-15  2.19E-14  2.30E-14  4.76  12  3.20E-15  1.10E-15  4.69E-15  1.47E-15  4.82E-15  2.26E-14  2.30E-14  1.67  268  Appendix E  Appendix E  c c c c  c c c c c c c c c c c c c c c  H D S _ 4 , 6 - D M D B T Parameter Estimation adapted from Applied Parameter Estimation Program Adopted by: Ibrahim Inamah A b u  P A R A M E T E R ESTIMATION ROUTINE FOR ODE MODELS Based on Gauss-Newton method with Pseudolnverse and Marquardt's Modification. Hard B O U N D A R I E S on parameters can be imposed. Bayesian parameter priors can be used as an option. G E N E R A L I Z E D L E A S T S Q U A R E S Formulation (when IGLS=1) with the D I A G O N A L W E I G H T I N G M A T R I X , Q(i)/Ydata(i)**2, i = l , N Y or W E I G H T E D L E A S T S Q U A R E S Formulation (when IGLS=0) with the D I A G O N A L W E I G H T I N G M A T R I X , Q(i), i= 1 , N Y  +  c  c c c c c c c c c c c c c c c c c c  The User must provide Subroutine M O D E L [that describes the mathematical model dx/dt=f(x,p)] and subroutine J X [computes the Jacobeans (df/dx) & (df/dp)] where x(i), i = l , N X is the State vector. The Output vector y(i), i = l , N Y is assumed to correspond to the first N Y elements o f the State vector  The Following Variables M U S T be Specified in P A R A M E T E R statement: N X Z (=NX) = Number o f state variables in the model N Y Z (=NY) = Number o f measured variables in the model N P A R Z ( = N P A R ) = Number o f unknown parameters N R U N Z = Greater or equal to the maximum number o f runs (experiments) to be regressed simultaneously N P Z = Greater or equal to the maximum number o f measurements per run  269  Appendix E c c c c c c cQ(i), c c c c c c c c c  The Following Variables M U S T be Specified in M a i n Program: I G L S = Flag to use Generalized Least Squares (IGLS=1) or Weighted LS(IGLS=0) F I L E F N = Name o f Input file F I L O U T = Name o f Output file i = l , N Y = Constant weights for each measured variable (Use 1.0 for Least Squares). I B O U N D = 1 to enable hard constraints on the parameters (0 otherwise) N S T E P = M a x i m u m number o f reduction allowed for Bisection rule (default is 10) K I F , K O S , K O F = Input file unit, Screen output unit, Output file unit. N S I G = Approx. Number o f Significant digits (Tolerence E P S = 10* *(-NSIG)) E P S P S I = M i n i m u m Ratio o f eigenvalues before the Pseudo-inverse approximation is used. EPSMPW = A very small number (used to avoid division by zero). T O L = Tolerence for local error required by O D E solver (default is 1 .Oe-6) —  c  USE MSIMSL E X T E R N A L jx,model P A R A M E T E R (nxz=4,nyz=4,nparz=4,ntot=nxz*(nparz+1 ),npz= 100,nrunz= 1) D O U B L E P R E C I S I O N ydata(nrunz,nyz,npz),tp(nrunz,npz),p(nparz) D O U B L E P R E C I S I O N xO(rirunz,nxz),q(nyz),tO(nrunz),nobs(nrunz) D O U B L E P R E C I S I O N pO(nparz),dp(nparz),x(ntot),v(nparz,nparz) D O U B L E P R E C I S I O N a(nparz,nparz),b(nparz),s(nparz),ag(l,l) D O U B L E P R E C I S I O N stdev(nparz),bv(nparz),bvs(nparz),pivpag(50) D O U B L E P R E C I S I O N pmin(nparz),pmax(nparz) D O U B L E P R E C I S I O N pprior(nparz),vprior(nparz) D O U B L E P R E C I S I O N t, tend, tol COMMON/gn/imode,nx,npar,irun,p C H A R A C T E R * 4 dummyline C H A R A C T E R * 2 4 filein, filout c c c  —  Set convergence tolerence & other parameters  D A T A kos/6/,kif/2/,kof/7/,epsmin/l .e-80/,tol/l .d-6/,igls/0/ D A T A nsig/4/,epspsi/l .e-30/,nstep/l 0/,q/nyz* 1.0/,ibound/l / c eps = 10.**(-nsig) nx=nxz ny=nyz npar=nparz nxx=nx +1 n=nx*(npar + 1 ) c c— c  Set Filenames and Open the files for I/O filein = ' D a t a I N l _ H D S . t e x f filout = 'DataOUT HDS.txt' 270  +  Appendix E open(kif,file=filein) open(kof,file=filout) c  Print header information write(kos,60) write(kof,60) 60 f o r m a t ( / 2 0 x , ' P A R A M E T E R E S T I M A T I O N P R O G R A M F O R O D E M O D E L S using', & /20x,the G A U S S - N E W T O N Method with Marquardf s Modification,', & /20x,'Pseudo-Inverse Approximation, Bisection Rule,', & /20x,'Bounded Parameter Search, Bayesian Parameter Priors') i f (igls .eq. 1) then write(kos,65) write(kof,65) 65 f o r m a t ( / / 2 0 x , ' G E N E R A L I Z E D L E A S T S Q U A R E S Formulation...') else write(kos,67) write(kof,67) 67 format(//20x,*WEIGHTED L E A S T S Q U A R E S Formulation...*) end i f write(kos,70) nx, ny, npar write(kof,70) nx, ny, npar 70 format(//20x,'Number o f State Variables ( N X ) = ',i3, & /20x,'Number o f Output Variables ( N Y ) = ',i3, & /20x,'Number o f Parameters ( N P A R ) = ',i3) write(kos,71) filein write(kof,71) filein 71 format(//40x,'Data Input Filename: ',a24) write(kos,72) filout write(kof,72) filout 72 format(40x,'Program Output File: ',a24)  c c c  Read P R I O R , M I N & M A X parameter values read(kif,*) read(kif,*) read(kif,*) read(kif,*) read(kif,*) read(kif,*) read(kif,*) read(kif,*)  c c c  dummyline (pprior(j),j=l,npar) dummyline (vprior(j),j=l,npar) dummyline (pmin(j)j=l,npar) dummyline (pmax(j),j=l,npar) Read N I T E R , I P R I N T & EPS_Marquardt  read(kif,*) dummyline read(kif,*) niter, iprint, epsmrq c c c  Read Number o f Runs to be Regressed  271  Appendix E  74  75  read(kif,*) dummyline read(kif,*) nrun i f (nrun .gt. nrunz) then write(kos,74) nrun format(///20x,'NRUNZ in P A R A M E T E R must be at leasf,i3//) stop 111 end i f write(kos,75) nrun write(kof,75) nrun format(//20x,'Input Data for',i3,' Runs',//) dfr= - npar  c Read Measurements for each R u n  c  c do 90 irun=l,nrun read(kif,*) dummyline read(kif,*) nobs(irun),tO(irun),(xO(irun,j),j=l ,nx) np=nobs(irun) dfr=dfr + ny*np c  85 &  write(kos,85) irun,np,tO(irun),(xO(irun,j),j=l ,nx) write(kof,85) irun,np,tO(irun),(xO(irun,j),j=l,nx) format(//5x/Run:',i2,4x,'NP=',i3,4x, T0=\gl2.4,4x,'X(0)=',10gll.4/);  91 90  read(kif,*) dummyline do 90 j=l,np read(kif,*) tp(irun,j),(ydata(irun,i,j),i=l,ny) write(kof,91) tp(irun,j),(ydata(irun,i,j),i=l,ny) format(5x,'Time=',gl2.4,10x,'Y-data(i)=',10gl2.4) continue  c  c  92 93 94  96  97  98  do 92 j=l,npar p(j)=ppriorG) continue write(kof,93) (p(j),j=l,npar) format(///2x,'Pprior(j)=',8gl2.5) write(kos,94) (p(j),j=l,npar) format(//2x,'Pprior(j)=',8gl2.5) write(kos,96) (vprior(j),j=l,npar) write(kof,96) (vprior(j),j=l,npar) format(lx,T/VARprior=',8gl2.5//) write(kof,97) (pmin(j),j=l,npar) vvTite(kos,97) (pmin(j),j=l,npar) format(lx,' Pmin(j) - , 8 g l 2 . 5 / / ) write(kof,98) (pmax(j),j=l,npar) vvTite(kos,98) (pmax(j),j=l,npar) format(lx,' PmaxG) =',8gl2.5//) 272  Appendix E close(kif) c c c c  104  106  M a i n iteration loop do 500 nloop=l,niter Initialize matrix A , b and Objective function SSEprior=0.0 do 104 i=l,npar SSEprior=SSEprior + vprior(i)*(pprior(i)-p(i))**2 b(i)=vprior(i)*(pprior(i)-p(i)) do 104 1=1,npar a(i,l)=0.0 continue do 106 i=l,npar a(i,i)=vprior(i) continue SSEtotal=SSEprior  c c  G o through each R u n do 210 irun=l,nrun write(kof,108) irun 108 format(/4x,'Current M O D E L output vs. D A T A for RUN#',i2, & /5x,'Time Y-data(i) & Y-model(i), i = l , N Y ' ) c c Initialize O D E s for current R u n do 112 j = l , n x x(j)=x0(irun,j) 112 continue do 114 j=nxx,n xG)=0.0 114 continue c c Initialize O D E solver: D I V P A G np=nobs(irun) call SETPAR(50,pivpag) index=l t=t0(irun) imode=l c Integrate O D E s for this R u n do 200 i=l,np tend=tp(irun,i) call DIVPAG(index,n,model,jx,ag,t,tend,tol,pivpag,x) i f (iprint .eq. 1) then write(kof, 135) t,(ydata(irun,k,i),x(k),k= 1 ,ny) 135 format(lx,gll.4,3x,4gl2.5) end i f c  c 273  Appendix E Update Objective function and matrix A , b  c  do 140 j=l,ny c  Select appropriate weighting factor ( G L S vs W L S ) i f (igls .eq. 1) then qyj= q(j)/(ydata(irun,j,i)**2 + epsmin) else  qyj= qO) end i f SSEtotal=SSEtotal + qyj*(ydata(irun,j,i) - x(j))**2 do 140 l=l,npar U=l*nx b(l)=b(l) + qyj*xO+ll)*(ydata(irun,j,i) - xO)) do 140 k=l,npar kk=k*nx a(l,k)=a(l,k) + qyj*x(j+ll)*x(j+kk) 140 continue Call D I V P A G again with I N D E X = 3 to release M E M O R Y from I M S L i f (i.eq.np) then index=3 call DIVPAG(index,n,model,jx,ag,t,tend,tol,pivpag,x) end i f 200 continue 210 continue c c Keep current value o f p(i) do 211 i=l,npar p0(i)=p(i) 211 continue c—&* A d d Bayes influence o f A & b do 212 i=l,npar b(i) = b(i) + vprior(i)*(pprior(i)-p0(i)) a(i,i) = a(i,i) + vprior(i) 212 continue c c Print matrix A & b i f (iprint .eq. 1) then write (kof,214) 214 format(/lx,' Matrix A and vector b....') do 218 jl=T,npar write(kof,216) (afjl j 2 ) , j2=l,npar),b(jl) 216 format(llgl2.4) 218 continue iprint=l end i f  c  c c c  Decompose matrix A (using D E V C S F from I M S L ) call  devcsf(npar,a,nparz,s,v,nparz) 274  Appendix E c c  220 225 c  Compute condition number & (V**T)*b conda=s( 1)/s(npar) do 220 i=l,npar bv(i)=0.0 do 220 j=l,npar . bv(i)=bv(i) + vG,i)*bG) continue continue Use Pseudo-inverse (if Cond(A) > 1/epspsi)  c  c  230 c  235 & & 237 c c  240 c c  ipsi=0 do 230 k=l,npar i f (s(k)/s(l) .It. epspsi) then bvs(k)=0.0 ipsi=ipsi + 1 else Include M A R Q U A R D T ' S modification bvs(k)=bv(k)/(s(k) + epsmrq) end i f continue write(kos,235) nloop,epsmrq,epspsi,ipsi,conda write(kof,235) nloop,epsmrq,epspsi,ipsi,conda format(//lx;iTERATION=',i3,/lx,'EPS_Marq.=',gl0.4,4x, 'EPS_PseudoInv =',gl0.4,4x,'No. Pseudolnv. Apprx.=',i2, /lx,'Cond(A)=',el2.4) write(kof,237) (sG)j=l,npar) format(lx,'Eigenvalues-,10gl 1.4) Compute dp = (V**T)*b do 240 i=l,npar dp(i)=0.0 do 24.0 j=l,npar dp(i)=dp(i) + v(i j)*bvsG) continue  Compute new vector p and ||dp|| dpnorm=0.0 do 250 i=l,npar dpnorm=dpnorm + abs(dp(i)) p ( i ) = p 0 ( i ) * ( l . + dp(i)) 250 continue dpnorm = dpnorm/npar i f (dpnorm .le. 10*eps*npar) then istep=0 else istep=l 275  Appendix E end i f write(kos,263) SSEtotal,SSEprior/SSEtotal 263 format(lx,'SSE_total =',gl 1.5,5x,'SSEprior/SSEtot =',fl4.4,/) write(kof,264) SSEtotal,SSEprior 264 format(lx,'SSE_total =',gl 1.5,3x,'Prior_SSE =',gl 1.5,/) write(kos,265) (p(j)j=l,npar) write(kof,265) (p(j),j=l,npar) 265 f o r m a t ( l x , ' P G ) b y G - N = ' , 8 g l 2 . 5 ) c c Enforce H A R D M P N & M A X parameter boundaries i f (ibound .eq. 1) then c Use bisection to enforce: Pmin(i) < P(i) < Pmax(i). do 460 j=T,npar 440 continue i f (p(j).le.pmin(j) .or. p(j).ge.pmax(j)) then do 450 i=T,npar dp(i)=0.50*dp(i) p(i)=pO(i)*(l. + dp(i)) 450 continue end i f i f (p(j).le.pmin(j) .or. p(j).ge.pmax(j)) goto 440 460 continue write(kos,465) (p(j),j=l,npar) write(kof,465) (p(j),j=l,npar) 465 format(lx,'P(j) Bounded=',8gl2.5) end i f c— Test for convergence i f (dpnorm .It. eps ) then i f (ipsi .eq. 0) then c Check i f Marquardt's E P S is nonzero i f (epsmrq .gt. s(npar)*0.01) then write(kof,347) epsmrq write(kos,347) epsmrq 347 format(//lx,'»»» Converged with EPS_Marquardt =',gl4.5) epsmrq=epsmrq*0.01 goto 500 else i f (epsmrq .gt. 0.0) then write(kos,347) epsmrq write(kos,349) write(kof,347) epsmrq write(kof,349) 349 format(//lx,'»»» From now on EPS_Marquardt = 0.0') epsmrq=0.0 goto 500 end i f end i f  276  Appendix E  285 &  286  sigma=sqrt((SSEtotal-SSEprior)/dfr) write(kos,285) sigma,dfr write(kof,285) sigma,dfr format(///5x,'++++++ C O N V E R G E D ++++++',/5x,'LS-Sigma=',gl 1.5, /5x,'Degrees o f Freedom=',f4.0,//) write(kos,286) (p(j),j=l,npar) write(kof,286) (p(j),j=T,npar) format(5x,'BestP0)-,8gl2.5)  c c—  Go and get Standard Deviations & Sigma goto 600 else  c Include one more singular values epspsi=epspsi* l.e-1 goto 225 end i f end i f c STEP-SIZE C O M P U T A T I O N S : c Use full step i f it is not needed or N S T E P = 0 i f (istep .eq. 0 .or. nstep .eq. 0) then goto 500 end i f c Compute step-size by the bisection rule do 400 kks=l,nstep c— Initialize Objective function, O D E s and D I V P A G SSEprior=0.0 do 301 i=l,npar SSEprior=SSEprior + vprior(i)*(pprior(i)-p(i))**2 301 continue SSEnew=SSEprior do 390 irun=l,nrun do 300 i=l,nx x(i)=x0(irun,i) 300 continue np=nobs(irun) call SETPAR(50,pivpag) index=l t=t0(irun) imode=0 do 340 ii=l,np tend=tp(irun,ii) call DIVPAG(index,nx,model,jx,ag,t,tend,tol,pivpag,x)  c  c c  Compute N E W value o f the Objective function do 320 j = l , n y  c c—  Select appropriate weighting factor ( G L S vs W L S ) 277  Appendix E i f (igls .eq. 1) then qyj= q(j)/(ydata(irun,j,ii)**2 + epsmin) else qyj= qO) end i f SSEnew=SSEnew + qyj*(ydata(irunj,ii) - x(j))**2 320 continue c c~ If L S Obj. function is not improved, half step-size i f (SSEnew .ge. SSEtotal) then do 330 i=l,npar dp(i)=0.50*dp(i) p(i)=pO(i)*(l. + dp(i)) 330 continue c~ Release M E M O R Y from I S M L (call with index=3) index=3 call DIVPAG(index,nx,model,jx,ag,t,tend,tol,pivpag,x) goto 400 end i f i f (ii.eq.np) then index=3 call DIVPAG(index,nx,model,jx,ag,t,tend,tol,pivpag,x) end i f 340 continue 390 continue write(kof,394) (pG),j=l,npar) write(kos,394) (p(j),j=l,npar) 394 format(lx,'PG) stepped=',8gl2.5/) kkk=kks-1 write(kof,395) kkk,SSEtotal,SSEnew write(kos,395) kkk,SSEtotal,SSEnew 395 format(/2x,'Stepsize= (0.5)**',i2,4x;SSE_total_old=',gl 1.5,3x, & 'SSE_total_new=',gl 1.5/) goto 500 400 continue write(kof,496) (jj(jj),jj=l,npar) 496 format(/lx,'PG)laststep=',8gl2.5) 500 continue c c Alert user that It did not converge... sigma=sqrt((SSEtotal-SSEprior)/dfr) write(kof,585) sigma,dfr 585 format(///5x,'***** D i d N O T converged yet *****',/5x, & 'LS-Sigma=',gl 1.5,/5x,'Degrees o f ffeedom=',f4.0) write(kof,586) (pG),j=l,npar) 586 format(/5x,'LastP(j)=',8gl2.5) c Compute Sigma & Stand. Deviations of the Parameter c  278  Appendix E c— C A L C U L A T I O N S A R E B A S E D O N T H E D A T A O N L Y (Prior info ingnored) 600 continue do 670 i=l,npar stdev(i)=0.0 do 670 j=T,npar stdev(i)=stdev(i) + v(i,j)**2/s(j) 670 continue do 680 i=l,npar stdev(i)=sqrt(stdev(i))* sigma 680 continue write(kos,686) ( s t d e v G ) * 1 0 0 j = l , n p a r ) write(kof,686) ( s t d e v G ) * 100 j=l,npar) 686 format(/5x,'St.Dev.(%)=',8gl2.5) c c Alert user whether G L S or W L S was used... i f (igls .eq. 1) then write(kof,688) 688 format(//lx,50('-'),/5x, & ' G E N E R A L I Z E D L E A S T S Q U A R E S Formulation Used',/lx,50('-'),/) else write(kof,689) 689 fomiat(//lx;50('-'),/5x/WEIGHTED L E A S T S Q U A R E S Formulation Used' & ,/lx,50('-'),/) end i f write(kof,890) 890 format(/lx,'-— PROGRAM END ',/) write(kos,891) filout 891 format(//lx,60('-'),/5x,'Program O U T P U T stored in file:', & a24,/lx,60('-'),/) close(kof,status='keep') pause stop end c c MODEL c c Ordinary Differential Equation ( O D E ) M o d e l o f the form dX7dt=f(X,P) c c where X ( i ) , i = l , N X is the vector o f state variables c P(i), i = l , N P A R is the vector o f unknown parameters c c —+ c c The User must specify the O D E s to be solved c d X ( l ) / d t = f l ( X ( l ) , . . . X ( N X ) ; P(1),...P(NPAR)) c d X ( l ) / d t = f 2 ( X ( l ) , . . . X ( N X ) ; P(1),...P(NPAR)) c d X ( l ) / d t = f J ( X ( l ) , . . . X ( N X ) ; P(1),...P(NPAR)),... etc 279  Appendix E c c and the Jacobean matrices: dfdx(i,j)=[df(i)/dX(j)] i = l , N X & j = l , N X c dfdp(i,j)=[df(i)/dPG)] i = l , N X & j=l,NPAR c c MODEL subroutine model(n,t,x,dx) double precision x(n),dx(n),fx(4,4),fp(4,4),p(4) double precision t common/gn/imode,nx,npar,irun,p c c  M o d e l Equations (dx/dt) . dx(l)=-(p(l)+p(2))*x(l) dx(2)=p(l)*x(l)-p(3)*x(2) dx(3)=(p(2)*x(l))+(p(3)*x(2))-p(4)*x(3) dx(4)=p(4)*x(3)  c —Jacobian (df/dx)  c  . i f (imode.eq.O) return fx(l,l)=-p(l)-p(2) fx(l,2)=0.0 fx(l,3)=0.0 fx(l,4)=0.0 fx(2,l)=p(l) fx(2,2)=-p(3) fx(2,3)=0.0 fx(2,4)=0.0 fx(3,l)=p(2) fx(3,2)=p(3) fx(3,3)=-p(4) fx(3,4)=0.0 fx(4,l)=0.0 fx(4,2)=0.0 fx(4,3)=p(4) fx(4,4)=0.0 c Jacobian (df/dp)  c  fp(l,l)=-x(l) fp(l,2)=-x(l) fp(l,3)=0.0 fp(l,4)=0.0 fp(2,l)=x(l) fp(2,2)=0.0 fp(2,3)=-x(2) fp(2,4)=0.0 fp(3,l)=0.0 280  Appendix E fp(3,2)=x(l) fp(3,3)=x(2) fp(3,4)=-x(3) c c-  Set up the Normalized Sensitivity Equations do 10 k=T,npar kk=k*nx do 10j=l,nx dx(j+kk)=fpG,k)*p(k) do 101=l,nx dx(j+kk)=dxG+kk)+fxG,l)*x(l+kk)  10  c c c c c c c c c c c  continue return end JX J A C O B E A N of the Ordinary Differential Equation ( O D E ) M o d e l required by the O D E solver (version for stiff systems) + The User must specify the Jacobean matrix: dfdx(i,j)=[df(i)/dXG)] i = l , N X & j = l , N X JX  c  S U B R O U T I N E jx(n,t,x,fx) double precision x(n),fx(n,n),p(4) double precision t common/gn/imode,nx,npar,irun,p c c~  20  Initialize Jacobian matrix F x do 20 i = l , n do 20 j = l , n fx(i,j)=0.0 continue  c c  Jacobian matrix F x fx(l,l)=-p(l)-p(2) fx(l,2)=0.0 fx(l,3)=0.0 fx(l,4)=0.0 fx(2,l)=p(l) fx(2,2)=-p(3) fx(2,3)=0.0 fx(2,4)=0.0 fx(3,l)=p(2) fx(3,2)=p(3) 281  fx(3,3)=-p(4) fx(3,4)=0.0 fx(4,l)=0.0 fx(4,2)=0.0 fx(4,3)=p(4) fx(4,4)=0.0  Set up the expanded Jacobian i f (imode .eq. 0) return ll=nx do 60 kk=l,npar do 50 j = l , n x  50 60  jj=ll+j do 50 i=l,nx ii=ll+i fx(ii,jj)=fx(i,j) continue ll=ll+nx continue return end SETPAR subroutine SETPAR(nn,param) Initialize vector P A R A M for D I V P A G  10  double precision param(nn) do 10 i=l,nn - param(i)=0 continue param(l)=1.0d-8 param(3)=1000. param(4)=500 param(5)=50000 param(6)=5 param(9)=l param(l 1)=1 param(12)=2 param(13)=l param(18)=1.0d-15 param(20)=l return end  282  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0058993/manifest

Comment

Related Items