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 I B R A H I M I N A M A H A B U B.Sc. (Hons)(Chem. Eng.), The University of Sci . & Tech, Ghana, 1984 M . Sc. (Chem. Eng.), University of 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 Abu , 2007 Abstract In the present study, modified transition metal phosphides (bulk and supported catalysts) were prepared by the temperature programmed reduction (TPR) of 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 (TEM) . The activity of the prepared catalysts were then tested using 4,6-dimethyldibenzothiophene (4 ,6 -DMDBT) for hydrodesulftrrisation (HDS) and carbazole for hydrodenitrogenation (HDN) . A s well , selected catalysts were tested using Light Gas O i l ( L G O ) derived from Athabasca bitumen. The H D S of 4 , 6 - D M D B T showed that small amounts of Co used to modify M2P a n d -M o P produced high selectivity for the direct desulfurization (DDS) product dimethylbiphenyl ( D M B P ) . Coo.4Ni2P/Al203 showed high conversion with little cracked products. Supported Coo.4Ni2P/MCM showed almost complete conversion of 4 , 6 - D M D B T with cracked products. Fluorination marginally increased the conversion of Coo .4Ni 2 P/Al 2 03 with insignificant changes in the product distribution and the addition of Pt enhanced hydrogenation. The N i x M o P (0 < x < 1.1) bulk series and supported Nio.33MoP were tested for the hydrodenitrogenation (HDN) of carbazole at 523-583 K , 3 M P a and a range of space velocity of 8.6-61 x 1 0 2 mol/h gcat. Nio.ovMoP showed the highest bicyclohexyl ( B C H X ) selectivity among the N i x M o P . The improved selectivity was attributed to the enhanced C O uptake and acidity that " resulted in increased hydrogenation of 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 of small amounts of Co to metal phosphides is beneficial as enhanced selectivity to the hydrogenolysis route of 4 , 6 - D M D B T is promoted. Similarly, addition of small amounts of N i is beneficial for the H D N of carbazole. However in this case, the presence of N i increased the hydrogenation of 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 of a fixed bed reactor 6 1.2.1 Mass transfer in a fixed bed reactor 6 1.2.2 Heat transfer in a fixed bed reactor 7 1.3 Methods of formulation to impove the hydoprocessing catalyst activity 8 1.4 Knowledge gap 11 1.5 Motivation 12 1.6 Objectives 13 Chapter 2 Literature Review 14 2.1 Sulfur containing compounds and the hydrodesulfurization (HDS) process 15 2.1.1 Types of common sulfur compounds in liquid fuels 15 2.1.2 Reactivity of sulfur containing compounds 17 i i i 2.1.3 Reaction pathways of hydrodesulfiirization of 4,6-dimethyldibenzothiophene 19 2.2 Nitrogen containing compounds and the H D N process 21 2.2.1 Nitrogen containing compounds in crude oi l 22 2.2.2 Difficulties of H D N 27 2.3 Reaction networks and mechanism of hydrodenitrogenation 29 2.3.1 Quinoline 28 2.3.2 Carbazole 30 2.4 Catalysts used for hydroprocessing 32 2.4.1 Use of transition-metal sulfides 33 2.4.2 Use o f transition metal nitride/carbides 33 2.4.3 Use of transition-metal phosphides 35 2.5 Developing a new phosphide catalyst for enhanced hydroprocessing 44 2.5.1 Promotional effect of a second metal 47 2.5.2 Promotional effect o f a third metal component 48 2.5.3 Effects of support 49 2.5.3.1 Use of alumina 49 2.5.3.2 Use of acidic support 49 2.5.3.3 Effect of addition of fluorine to alumina 50 2.6 Effect o f adding platinum 51 2.7 Effect of process variables 52 iv Chapter 3 Experimental 54 3.1 Preparation of metal phosphides for hydrodesulfurization 54 3.1.1 Preparation of bulk metal phosphides of Co x Ni2P and C o x M o P 55 3.1.2 Preparation o f supported metal phosphide catalysts for hydrodesulfurization of 4,6-dimethyldibenzothiophene 56 3.2 Preparation of N i x M o P as hydrodenitrogenation catalysts 58 3.3 Catalyst Characterization 59 3.3.1 Temperature Programmed Reduction (TPR) 59 3.3.2 Temperature Programmed Reduction using Tapered Element Oscillating Microbalance ( T E O M ) 60 3.3.3 Powder X-ray Diffraction ( X R D ) 61 3.3.4 Brunnauer-Emmett-Teller (BET) Surface Area 62 3.3.5 X-ray Photoelectron Spectroscopy (XPS) 62 3.3.6 Carbon monoxide (CO) uptake 62 3.3.7 n-Propyl amine (n-PA) chemisorption 64 3.3.8 Scanning Electron Microscopy-Energy Dispersive X - R a y Emission ( S E M - E D X ) 65 3.3.9 Transmission Electron Microscopy ( T E M ) 65 3.4 Catalyst activity measurements for hydrodesulfurization of 4,6-dimethyldibenzothiophene and hydrodenitrogenation of carbazole 65 Chapter 4 Hydrodesulfurization over bulk and supported phosphides using 4,6-dimethyldibenzothiophene 68 4.1 Characterization of bulk metal phosphides 68 v 4.1.1 X R D of the prepared catalysts 69 4.1.2 T P R of precursors 73 4.1.3 T E M of prepared catalysts 77 4.1.4 Other catalyst properties 80 4.1.5 X P S of prepared catalysts 80 4.2 Catalyst activity 86 4.3 Discussion on bulk phosphides used for H D S of 4 , 6 - D M D B T 93 4.4 Properties and hydrodesulfurization of 4,6-dimethyldibenzothiophene over supported catalysts 97 4.4.1 X R D of modified AI2O3 supported phosphides 99 4.4.2 X R D of Pt-Coo .4Ni 2 P/Al 203 and P t - C o o . 4 N i 2 P / M 2 03 - W A 101 4.4.3 Properties o f Coo.4Ni2P/AI2O3 prepared on different supports 102 4.4.4 Properties o f Pt-Coo.4Ni2P/Al 203 prepared on different supports 104 4.5 Supported catalysts activity using 4 , 6 - D M D B T 105 4.6 Kinetics of H D S 110 4.7 Discussion of 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 x M o P 123 5.2. Catalyst characterization of prepared metal phosphides 123 5.2.1 T P R of bulk NixMoP 123 vi 5.2.2 X R D of bulk N i x M o P 124 5.2.3 X P S of bulk N i x M o P 127 5.3 Catalyst activity of bulk N i x 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 supported catalysts 138 5.5.1 X R D of modified Nio.3 3 MoP/Al 2 03 140 5.5.2 Properties of prepared supported metal phosphides for H D N 141 5.6 Activi ty of N i o ^ M o P on different supports 142 5.7 Kinetics of the hydrodenitrogenation of carbazole 144 5.8 Discussion on hydrodenitrogenation of supported phosphides 150 Chapter 6 Hydroprocessing using Light Gas Oil over supported catalysts 154 6.1 Results and discussion using Light Gas O i l 155 6.2 Kinetics of hydrodesulfurization and hydrodenitrogenation over selected catalysts 158 Chapter 7 Conclusions and Recommendations 168 7.1 Conclusions 168 7.2 Recommendations 170 References 172 Appendix A . l Establishing validity of plug flow in experimental set-up 190 A . 2 Parameters for testing plug flow 190 v i i A . 3 Mass transfer limitation 192 A . 4 Reactor isothermal operation 194 A.5 Determination of saturation vapor pressure 195 A . 6 Determination of vapor and saturated pressures 196 Appendix B Examples of calculation for characterization data 197 B . 1 Catalysts preparation: Calculation of required chemicals 197 B.2 T P R calibration 200 B.3 Degree of reduction 201 B.4 Temperature programmed reduction of transition metal phosphide precursors in 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 of crystallite sizes 205 B.7 Calculation of lattice parameters 208 B.8 X P S Survey Scan 209 B.9 Determination and repeatability of C O chemisorption 212 B.10 Repeatability o f n-propylamine titration of acid sites on metal phosphides 214 B . l 1 Determination of the specific consumption of 4 , 6 - D M D B T and carbazole 216 Appendix C Hydrodesulfurization experiments 220 C. 1 Summary o f hydrodesulurization experiments 220 C.2 Response Factor and sample calculation of H D S activity , 221 C.3 Example of repeatability o f hydrodesulfurization experiments 224 v i i i C.4 Calculation of standard error (S.E.) 227 C. 5 Data for hydrodesulfurization experiments. 228 Appendix D Hydrodenitrogenation experiments 249 D . 1 Summary of hydrodenitrogenation experiments 249 D.2 Response Factor and sample calculation of H D N activity 250 D.3 Example of repeatability o f hydrodenitrogenation experiments 254 D.4 Data for hydrodenitrogenation experiments 256 Appendix E Program for the Gaussian Newton-Raphson Parameter Estimation 269 ix List of Tables 2.1 Representative sulfur compounds in liquid fuels 16 2.2 Comparison of H D S conversion of 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 of metal phosphides 39 2.6 Summary of activity of metal phosphides promoted with Co or N i 41 4.1 Lattice parameters estimated from P X R D of reduced catalysts 71 4.2 Summary of temperature programmed reduction data „78 4.3 Properties of the prepared metal phosphides 81 4.4 Activities of bulk metal phosphides for the H D S of 4 , 6 - D M D B T measured at 583 K and 3.0 M P a H 2 89 4.5 Comparison of sulfur speciation in liquid product from H D S of 4,6-D M D B T measured at 583 K and 3.0 M P a H 2 over N i 2 P and Coo.o 8Ni 2P 91 4.6 A Properties of supports 103 4.6B Properties of the prepared Coo.4Ni 2P/Al 203 phosphide on different supports 103 4.7 Properties of the prepared Pt-Coo.4Ni 2P/Al 203 phosphide on different support 105 4.8 Activities of supported metal phosphides for the H D S of 4 , 6 - D M D B T measured at 583 K and 3.0 M P a H 2 107 4.9 Estimated 1 s t order rate constant for hydrodesulfurization of 4 , 6 - D M B T 114 5.1 Lattice parameters estimated from X R D of reduced catalysts 126 5.2 Physiochemical properties of prepared Coo.4Ni 2P on different supported metal phosphides 128 x 5.3 Activities of bulk metal phosphides for the H D N of carbazole measured at 583 K and 3.0 M P a H 2 132 5.4 Properties of prepared Nio.33MoP/Al 203 and Nio.33MoP/MCM catalysts 141 5.5 Activities of Nio 33M0P supported metal phosphides for the H D N of carbazole measured at 583 K and 3.0 M P a H 2 144 5.6 Estimated 1 s t order rate constants for the hydrodenitrogenation of carbazole 148 6.1 Characteristics of Light Gas O i l derived from Athabasca bitumen 155 6.2 Apparent kinetic parameters for H D S and H D N of light gas oi l at different temperatures 159 6.3 Comparative apparent activation energy for H D S and H D N of L G O over Nio.33MoP/y-Al203, Coo .4Ni 2 P/Al 2 03, Pt-Coo .4Ni 2 P/Al 203 and commercial sulfided catalysts 162 6.4 Measured catalyst atom ratios before and after reaction in L G O 163 6.5 Atom ratios determined by X P S of supported metal phosphide catalysts before and after reaction in L G O 164 A l Parameters of catalyst and reactor 190 B1 Data for the crystallite sizes of N i x M o P 208 B2 Results of the repeatability of C O uptake on Coo.o8Ni2P 213 B3 Results o f the n - P A repeatability using Coo.o8Ni2P 215 C I Summary of experiments for hydrodesulfurization of 4,6-dimethyl-dibenzothiophene (4,6-DMDBT-3000 ppm) over transition metal phosphides at 3.0 M P a , H 2 pressure for 12 h time on stream 220 xi C2 Results of hydrodesulfurization of 4 , 6 - D M D B T over Co 0 .o8Ni 2P 222 C3 Repeatability of hydrodesulfurization of 4 , 6 - D M D B T using C o o . o s ^ P 224 D l Summary of experiments for hydrodenitrogenation of carbazole (3000 ppm) over transition metal phosphides at 3.0 M P a , H 2 pressure for 12 h time on stream 249 D 2 Results o f hydrodenitrogenation o f carbazole over N10.07M0P 252 D3 Repeatability of hydrodenitrogenation over N10.07M0P using carbazole 254 x i i List of Figures 1.1 Schematic layout of 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 of 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 M o / A l 2 0 3 31 2.4 Crystal structures of some transition metal phosphides 36 2.5 Schematic diagram of the preparation of metal phosphides 38 2.6 Alternative catalytic route for conversion of 4 , 6 - D M D B T 47 3.1 Schematic diagram of the fixed bed reactor 67 4.1 X-ray diffractograms of reduced C02P, CoP and Co x Ni2P catalysts (A - N i i 2 P 5 , * - C o 2 P , T - N iCoP) 70 4.2 X-ray diffractograms of reduced M o P and C00.07M0P catalysts 73 4.3 T P R of Ni2P and C0XM2P catalyst precursors measured in 10 % H2 in A r at a rate of 60 ml(STP)/min 75 4.4 T P R of C02P, CoP, M o P and C00.07M0P catalyst precursors measured in 10 % H 2 in A r at a rate of 60 ml(STP)/min 76 4.5 T E M micrographs of bulk metal phosphides: (i) M 2 P and (ii) Coo .os^P . Estimated d-spacing of 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 of N i 2 P 79 4.6 X P S of (a) Co 2p region (b) P 2p region, and (c) N i 2p region of x i i i Co x Ni2P catalysts after reduction and passivation 83 4.7 X P S of P 2p region and M o 3d region of M o P and C00.07M0P catalysts after reduction and passivation 85 4.8 Correlation of P / M ratio determined by X P S and n - P A : C O uptake ratio determined by adsorption for C o x N i 2 P (•), C00.07M0P ( • ) , CoP (o), C o 2 P (0) with MoP(A) and Ni 2 P(D) as indicated 87 4.9 Conversion of 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 2 , plotted as a function of C O uptake: C o x N i 2 P (•), C00.07M0P (A), C o 2 P (0) and CoP (o) with MoP(A) and Ni 2 P(D) as indicated 92 4.10 X R D of prepared metal phosphides on different acidic supports 100 4.11 X R D of Pt- C o o . 4 N i 2 P / A l 2 0 3 and Pt- C o o . 4 N i 2 P / A l 2 0 3 - W A 101 4.12 Plot of In (1 -X A ) versus space time at 3.0 M P a using P t - C o 0 4 N i 2 P / A l 2 O 3 109 4.13 Simplified reaction scheme for the H D S of 4 , 6 - D M D B T I l l 4.14 Correlation o f the experimental data versus the predicted from the model for conversion of D M D B T (0) and yields of products (DMBP:» ; M C H T : A ; D M B C H : « ) 112 4.15 Plots of In k versus 1000/T for the hydrodesulfurization of 4 , 6 - D M D B T using P t - C o 0 4 N i 2 P / A l 2 O 3 115 4.16 Plots of Bransted acidity as a function of both conversion and D M B P selectivity on all supported catalysts: N i 2 P / A l 2 0 3 (•), C o 0 4 N i 2 P / A l 2 O 3 ( A ) , C o o . 4 N i 2 P / A l 2 0 3 - F ( • ) , C o 0 4 N i 2 P / M C M (•) Pt- C o 0 4 N i 2 P / A l 2 O 3 (•), Pt- C o 0 . 4 N i 2 P / A l 2 O 3 - W A (A ) 117 4.17 Comparison of the X R D profiles of the fresh (reduced) and spent xiv (after activity test) metal phosphides 119 4.18 Comparison of the X P S of the fresh (reduced) and spent (after activity studies) metal phophides 120 5.1 T P R of calcined catalysts precursors ( N i x M o P for 0.0 < x < 1.11) measured in 10 % H 2 in A r at flowrate of 60 ml(STP)/min 124 5.2 X-ray diffractograms of all reduced N i x M o P for 0.0 < x < 1.11 125 5.3 X P S spectra of the N i 2P, M o 3d and P 2p of the prepared phosphides after reduction and passivation 129 5.4 Correlation of P / M ratio determined by X P S and n - P A : C O uptake ratio determined by adsorption for N i x M o P for 0.0 < x .< 1.11(«) and M o P (A) 130 5.5 Conversion of carbazole and selectivity to B C H X over various metal phosphide catalysts at 583 K and 3.0 M P a H 2 , plotted as a function o f C O uptake: N i x M o P (•) 0 .0<x< 1.11 and M o P ( A ) 134 5.6 S E M of Nio.33MoP/Al 20 3 139 5.7 X R D of Nio.33MoP/Al 20 3 140 5.8 Plot of In (1 -X A ) versus space time at 533 K , 3.0 M P a using Ni 0 .33MoP/Al 2O 3 145 5.9 Simplified reaction network of carbazole 146 5.10 Correlation of the experimental data versus the predicted from the model for conversion of carbazole (0) and yields of products ( B C H X : * ; T H C Z : A ) 149 5.11 Plots of In k versus 1000/T for the hydrodenitrogenation of carbazole over Ni 0 .33MoP /Al 2 O 3 151 5.12 Plots of Bronsted acidity as a function of both conversion and D M B P xv selectivity on all supported catalysts:Coo.4Ni2P/Ai203 (A), C00.4N12P/AI2O3-F (•) ,Co 0 . 4 Ni 2 P/MCM (•), N10 .33M0P/ai2o3 (•), Ni 0 3 3MoP/MCM (A) 152 6.1 Total sulfur conversions over selected catalysts using LGO at 613, 623 K, 633 K and 648 K. P = 8.8 MPa, LHSV = 2 h"1, H 2 to oil ratio = 600 ml/ml: • Coo.4Ni2P/Al203 •Pt-Coo.4Ni2P/Ai203 • Sulfided NiMo/Al 2 0 3 a Nio.33MoP/Al203 156 6.2 Total nitrogen conversions over selected catalysts using LGO at 613 K, 623 K 633 K and 648 K. P = 8.8 MPa, LHSV = 2 h"1, H 2 to oil ratio = 600 ml/ml: • C00.4N12P/AI2O3 •Pt-Coo.4Ni2P/Al203 A Sulfided NiMo/Al 2 0 3 a Nio.3 3MoP/Al 20 3 157 6.3 Arrhenius plots for determining the apparent activation energy for the HDS (A) and HDN (B) of LGO over Ni0.33MoP/Y-Al2O3 and commercial sulfided catalysts 161 6.4 Comparison of XRD diffractograms obtained for Nin.33MoP supported on AI2O3 and AI2O3-F, before and after reaction in LGO 165 6.5 XPS of Ni 2p and Mo 3d region for C00.4N12P/AI2O3 and Ni 0 . 3 3MoP/Al 2O3 catalysts before and after reaction with LGO 167 B l 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 Co0.o8Ni2P and C00.07M0P 211 B6 Repeatability of CO uptake using CO0.08N12P Conditions, CO flow = xvi 0.3 ml/min, He(mix) = 30 ml/min, size of sample loop = 1ml Run 1: C O uptake = 0.90 mmols/g, Run 2: C O uptake = 0.99 mmols/g 212 B 7 Repeated n -PA data using Coo.o8Ni2P 214 B8 Arrhenius plots for determining the apparent activation energy for the H D S ( A ) and H D N (B) of L G O over ( • ) C o o . 4 N i 2 P / A l 2 0 3 , and (v) Pt-Coo .4Ni 2 P/Al 2 0 218 B9 Diagram of the M C M - 4 1 crystallite 219 C I Calibration curve for 4 , 6 - D M D B T used to determine the response factor 221 C2 Profile o f product ratio using N i 2 P / A l 2 0 3 (A) and C00.4N12P/AI2O3 (B) for the H D S of 3000 ppm of 4 , 6 - D M D B T at 583 K and 3.0 M P a 248 D l Calibration curve for carbazole used to determine the response factor 250 D2 Calibration curve for B C H X used to determine the response factor 251 xv i i List of Abbreviations 4 , 6 - D M D B T Dimethyl dibenzothiophene refractory sulfur containing compound 4 , 6 - D M T H D B T Dimethyltetrahydro-dibenzothiophene 4 , 6 - D M H H D B T Dimethylhexahydro-dibenzothiophene 4 , 6 - D M P H D B T Dimethylpentahydro-dibenzothiophene A A S C Aliphatic and non-heterocyclic aromatic sulfur compounds A E D Atomic emission detector A G O Atmospheric Distilled Gas O i l B . E Binding Energy B C H X Bicyclohexyl b/d Barrels per day B P Biphenyl B T Benzothiophene C B Z Carbazole C H B Cyclohexylbenzene C H C H E 3-cyclohexyl-cyclohexene c p Heat capacity per unit mass of fluid C P M C H Cyclopentylmethyl-cyclohexane D B T Dibenzothiophene D D S Direct desulfurization D H Q Decahydroquinoline D M C H B Dimethylcyclohexane benzene D M B C H Dimethylbicyclohexane D M B P Dimethyl biphenyl xv i i i E D X Energy Dispersion Spectroscopy E P A Environmental Protection Agency F C C Fluid catalytic cracker FID Flame ionization detector FTIR Fourier Transfom Infrared Reflectance G Mass velocity of fluid G C - M S Gas chromatograph-mass spectroscopy G M molal velocity (mol mixture/sec.cm 2 of total bed cross section) h Heat transfer coefficient, H C H Hexylcyclohexane H C O Hydrocracker Gas O i l H D N Hydrodenitrogenation H D O Hydrodeoxygenation H D S Hydrodesulfurization H H C B Z 1,2,4,4a,9a-hexahydrocarbazole ICP Inductive Coupled Plasma J D Mass transfer group symbol k Thermal conductivity k c Mass transfer coefficient k G Mass transfer coeffiecient related to pressure L C O Light Cycle O i l L H S V Liquid hourly space velocity L P G Liquified Petroleum Gas M C M - 4 1 M o b i l Crystalline Material x ix M P a Mega Pascal n -PA n-propyl amine N p r Prandtl number, O H C B Z 1,2,3,4,4a,9a-octahydrocarbazole O P A o-propylaniline P D F Powdered Diffraction Files P H C B Z Perhydrocarbazole ppm Parts per mil l ion P V Pore volume P X R D Powdered X-ray Diffraction q Heat flux Q Quinoline R S H Thiols R S R Sulfides S A B E T surface area SBET B E T surface area S V Space velocity T C D Thermal conductivity detector T E M Transmission electron microscopy T H C Z 1,2,3,4-tetrahydrocarbazole T 0 Fluid stream temperature T O F Turn over frequency T P D Temperature programmed desorption T s Surface temperature of the pellet, xx T P R Temperature programmed reduction THQ1 1,2,3,4-tetrahydroquinoline V G O Vacuum Gas O i l W A Weakly acidic W H S V Weight Hourly Space Velocity X R D X-ray diffraction X P S X-ray Photoelectron spectroscopy y-AI2O3 Also AI2O3 (alumina) s V o i d fraction p Density of fluid X Heat of adsorption xx i Acknowledgement First I would like to express my sincere thanks and gratitude to my supervisor, Professor Kev in 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 in my future ambition. I would also like to specially thank my thesis committee members Professor C. J im L i m , Professor Ajay K . Dalai, Professor Elod Gyenge and Dr. John Adjaye for the advice and suggestions in completing of this thesis. M y sincere thanks also go to Dr. K e n Wong o f the Advanced Material and Process Engineering Laboratory, Mary Mager from the Department of Metals and Materials Engineering and Lina Mandilao from the Wine Research Centre for the help with catalyst characterization. To the office staff, stores and workshop of Chemical and Biological Engineering, thanks for helping. I want to also thank the Canadian Scholarship Secretariat and the Government of Ghana for their financial support. Thanks to my wife Zaria, my kids, Anas, Adel , 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. xx i i Introduction Chapter 1 Introduction Recently, demands for cleaner burning fuels have led to a reduction in the allowable sulfur and nitrogen content of fuel. For example, in 2010 the U S Environmental Protection Agency (EPA) 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 of having to process heavier crude since the light and sweeter crude containing low contents of S (0.5 wt%) and N (0.1 wt%) are fast diminishing. Furthermore, synthetic crude derived from oi l sands that contain large quantities of 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 of oi l sands covering an area of 46,800 square kilometers containing 137 bil l ion cubic meters (862 bil l ion barrels) of oi 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. Among 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 4,6-dimethyldibenzothiophene (4 ,6 -DMDBT) 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 of 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 of petroleum feedstock has been extensively practiced in the petroleum industry and in the primary upgrading of heavy crudes and synthetic fuels. Hydroprocessing is also an integral part of the production of liquid fuels from coal and biomass. In the refinery, hydroprocessing involves a variety of catalytic reactions such hydrogenation (that lead to saturation of aromatics, olefins, etc) hydrocracking and removal of 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 in the overall molecular structure such that the boiling points of the different oi 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, in 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 of 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 in 116 countries with 202 of the refineries located in As i a Pacific, 160 in North America and 105 in Western Europe. These refineries process about 50% (81.9 mil l ion b/d) of 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. With the high volume of hydropocessing applications, and based on the amount of 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 of oi l and hydrogen that usually results in unfavorable H 2 and H 2 S concentration profiles through the reactor. The removal of the last ppm of S is inhibited in this type of configuration. Therefore a counter current operation is preferable. The process scheme of 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 in Sweden and Lyondell-Citgo refinery in the U S . In this scheme, the oi l feed is introduced to the reactor at the top and hydrogen is introduced at the reactor outlet. The SynSat process merges the fields of catalysis and reactor engineering and uses different catalyst beds within a single reactor shell with intermediate removal of byproduct gas to achieve deep hydrodesulphurization. In this system, catalysts A and B are sulfided catalysts such as sulfided Ni-Mo/Al 2 C»3. Catalyst C is a noble metal on an acidic support. 3 Introduction Gas and Light Gasoline Crude oil E 5 Light Ends Plant Naphtha Kerosene Diesel AGO Vacuum distillation Residuum VGO • LPG Isomerization Plant Hydro-cracker Reformer HCO r «— Catalytic cracker AJkylation Plant Butanes Isomerate Reformate Alkylate 'iqht r.ydP on Coker Hydro-treater Gasoline -•Jet Diesel Fuel oil Asphalt Coke Figure 1.1 Schematic layout of a fully integrated refinery (Nakamura, 2002) Between the two catalyst beds A and B , H 2 S and other gases are removed using a vapor-liquid 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 2 g a s Recycled liquid Catalyst A Catalyst B Catalyst C Diesel Product Gas removal (H2S, etc) A t V / L separator Make-up 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 of the design of 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 in 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 in 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 of equation 1.1 (Froment, 1990; Rase, 1990) N s h = ^ = f(NKe,NSc) 1.1 m where, Nsh, NR e and N s c are the dimensionless Sherwood, Reynold's and Schmidt's numbers respectively, k c is the mass transfer coefficient, d p is the equivalent diameter and D m is the molecular diffusion coefficient for the diffusing species. The mass transfer from gas to liquids in packed beds is given by equations 1.2 and 1.3 (Geankoplis, 1993). Vo = M 5 5 < 7 V R e < 1500 1.2 9 D = - ^ r for 0.0016<iV R e <55 1.3 Re 6 Introduction where 6 is the void space between pellets as a fraction of the total volume of the bed and jo is a grouped symbol = N2'3 = ^ N2'3 1 4 G sc GM sc where kG = kc IR T, p is the fluid density, G is the mass velocity of the fluid and G M is the molal velocity (mol mixture/sec.cm 2 o f total bed cross section). 1.2.2 Heat Transfer The mechanism of 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 : 3 1.5 C P G where h = — - — 1.6 T -T s o C n M k h is the heat transfer coefficient, G is the mass velocity, c p is the heat capacity per unit mass of fluid, N p r is the Prandtl number, T s is the surface temperature of the pellet, T 0 is the fluid stream temperature and k is the thermal conductivity. In a trickle bed reactor, the liquid flows cocurrently down the bed of catalyst with the gas, however, counter current flow is also practiced. Cocurrent flow is desired because much better distribution of the liquid over the catalyst is preferred and higher flow rates of the liquid are possible without flooding. Note that in the refinery, counter current flow is preferred because there is a favorable concentration of the H2 and H2S throughout the bed that allows the last ppm Introduction of S to be removed and that is the main objective of 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 in Chapter 4 and Appendix B . l 1. 1.3 Methods of catalyst formulation to improve the hydroprocessing activity In order to meet some of 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 of the promoter with the main metal is complex. For example, different explanations have been reported for the promoting effect of Co 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 of cobalt suggesting that Co makes S more mobile and thus creating more active sites. Other authors also report that the promotion effect o f Co was attributed to the decrease in the strength of sulfur-molybdenum bond and thus facilitating the desorption of H2S from the active sites (Kabe et al., 1998). Shuit et al. (1973) reported that Co (present as Co ) was assumed to be in the tetrahedral positions in the surface of the AI2O3, replacing A l 3 + ions and that the promotional effect was due to an increase in the stability of the 8 Introduction M o monolayer. The catalytically active sites were M o ions, produced in the presence of hydrogen from M o ions by removal of some S " ions. One approach is to improve the activity of existing metal sulfide catalysts and there are many reports in the literature in this regard (Lee et al. 2005; Mosio-Mosiewski and Morawski , (2005); Kwak et al., 1999; Bataille, 2001). The other approach is the investigation of 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; Coll ing 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 in modifying the metal phosphides to increase their activity and selectivity. The promotion of a second metal to form ternary phosphides is one modification used in anticipation of improved catalyst activity. The idea was drawn from the fact that in commercial sulfided catalysts, when N i or Co 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 Co-M o - S catalysts consist of crystallites of M0S2 decorated at their crystal edges with Co atoms (Kabe et al., 1992). The 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 of 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 in the absence of H2S and N i M o P was the least active when these phosphides were 9 Introduction tested for H D N of o-propylaniline. However, the authors reported that elemental sulphur analysis of the used catalysts showed higher levels of sulfur on the M o P and N i M o P than on the CoP and C o M o P , indicating that the Co containing catalysts were resistant to sulfur. Sun et al. (2004) have reported the addition of N i to M o P catalyst for H D S and in this study, increased N i content of the N i - M o - P catalyst resulted in increased activity for H D S of dibenzothiophene (DBT) . However, no synergistic effect between the N i and M o was observed and N i 2 P had higher activity than all the N i - M o - P catalysts tested. They concluded that N i was not a promoter of H D S activity in the N i - M o - P system either. Similarly, Rodriguez et al. (2003) have shown that a M o N i P / S i 0 2 catalyst was much less active than either M o P / S i 0 2 or N i 2 P / S i 0 2 for H D S of thiophene. Since hydroprocessing involves the removal of 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 of the tightening of environmental legislation regarding the release of N O x . In addition, N compounds have undesirable effects such as poisoning of acidic catalysts. H D N is also becoming important because of increasing interest in converting petroleum residua, coal, shale and tar sands, which contain higher concentrations of N than conventional crude oils. Refineries now have to process more heavy feedstocks such as vacuum gas oi l ( V G O ) , coker gas o i l , and light cycle oi 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. Carbazole is a refractory N containing compound found in petroleum feedstock and also difficult to remove. The effect of alkyl groups is less important because of the higher electron density of 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 of 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 of N i in N i 2 P with one mole of M o to form M o N i P , and the intrinsic activity of this catalyst was shown to be less than the M o P and C02P catalysts. The use of lower Co and N i concentrations has not been investigated. Typically, commercial hydroprocessing catalysts such as sulfided C0M0/AI2O3 and MM0/AI2O3 use 3-5 wt% of Co 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 of 4 , 6 - D M D B T . 4 , 6 - D M B T is one of 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 of the two methyl groups on the D B T backbone, influences the reactivity. For example, the conversion of 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 low reactivity of 4 , 6 - D M D B T to the presence of the two methyl groups at the 4 and 6 position of the molecule. The authors explained that at these positions, the methyl groups hinder the access of 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 of 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 of removing S from 4 , 6 - D M D B T is by first hydrogenating the aromatic rings and subsequent hydrogenolysis of the C-S. There is no report in the literature exploring the effect of adding noble metals to metal phosphides to enhance hydrogenation in 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 of 4 , 6 - D M D B T (Landau et al., 1997). The reactivity of carbazole which is a more resistant N containing compound in petroleum feedstock has not been reported in the literature using metal phosphides. Finally, there is little information on the relationship between the activity and the physiochemical properties of metal phosphides on modified metal phosphides. 1.5 Motivation Based on the above issues, the present study was focused on CoxNi2P (0.08 < x < 0.8) and N i x 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 carbazole H D N , respectively. B y incorporating small amounts of Co and N i in the N i 2 P and M o P respectively, the metal phosphides are expected to be metal rich and therefore increase activity of 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 of 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 of the 4 , 6 - D M D B T molecule. It is also expected that by incorporating small amounts of 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 of the present research are: A . To investigate the H D S and H D N activity of modified metal phosphides when 3-24 wt% of Co and N i are added to N i 2 P and M o P , respectively. B . To study the role of acidic components in the form of 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 2 0 3 supports) for H D S and H D N activity. C. To determine the effect of adding platinum on the catalyst activity in order to explore the H D S hydrogenation pathway of 4 , 6 - D M D B T . E . To examine the kinetics of 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 in order to meet new challenges. Presently, refineries must remove more N and S from petroleum feedstocks in 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 of 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 2 0 3 ) have difficulty meeting these 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 of these materials and test them using real petroleum feedstock since most of the activity reported for metal phosphides have used model compounds. The improvement can be made by studying the reaction mechanisms and the properties of N and S compounds in petroleum feedstock, in order to tailor a catalyst to effectively hydroprocess and meet the challenges mentioned above. Therefore, in this section, a review of the type, quantities, chemical behavior, mechanisms and the catalysts that have been investigated in anticipation of meeting these challenges, is presented. 1 4 Literature review 2.1 Sulfur containing compounds and the hydrodesulfurization (HDS) process Hydrodesulfurization is a refinery process where S in the form of organic compounds is removed from distillate streams. The process involves the use of H 2 at high temperature and pressure to remove S in the form of H 2 S . H D S is a heterogeneously catalyzed reaction. It is 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 2 C>3 and C o W / A l 2 0 3 are commercial catalysts used worldwide for H D S . The promoters are either N i or Co but more recently, sulfided Ni -Mo/Al 2 C>3 has been commercially employed for H D S (Song, 2000). 2.1.1 Types of common sulfur compounds in liquid fuels It is important to have a good knowledge of the types and reactivity of S compounds since this information can provide the basic understanding of how these S compounds can be removed from petroleum feedstock. The common types of S compounds in liquid fuels, shown in Table 2.1, can be divided into five classes namely thiols, sulfides, thiophenes, benzothiophenes (BT) and dibenzothiophenes (DBT) . The five classes can be further divided into aliphatic and non-heterocyclic 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 of transportation fuels gasoline, diesel and jet fuels differ in composition and properties (Gary and Handwerk, 1994; Mayo et al., 2001; Hsu et al., 2000; Song, 2000). For example, gasoline derived from naphtha and fluid cracking (FCC) naphtha contains thiols (RSH), sulfides (RSR), disulfides (RSSR), 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 (LCO) , ••' 15 Literature review Table 2.1 Representative sulfur compounds in liquid fuels Compound Formula Representation Thiols RSH R-S-H Sulfides RSR R " S " R Disulfides RSSR R-S-S-R R-Thiophenes R C 4 H 3 S Benzothiophene R C 8 H 5 S Dibenzothiophene R C 1 2 H 7 S 4,6-dimethyldibenzothiophene C 1 4 H 1 2 S J S C H 3 C H 3 contain alkylated benzothiophenes, benzothiophenes and alkylated dibenzothiophenes. R is an alkyl or phenyl group and typically contains C1-C4 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 in practical high-conversion processes (Phillipson, 1971; Gates et al., 1997). However, the conversion of alkyl substituted dibenzothiophenes and especially the 4,6-dimethyldibenzothiophene (4 ,6 -DMDBT) 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, 6-ethyl dibenzothiophene all exert steric hindrance. The position of the alkyl substituents on the aromatic ring has a greater impact on the H D S reactivity than the number of 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 of the H D S conversion of individual polyaromatic S-containing compounds in different solvents over sulfided C0M0/AI2O3. Although direct comparison of 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 2,8-dimethyldibenzothiophene has high H D S reactivity (94-100%) while 4,6-dimethyldibenzothiophene has low H D S reactivity (7-9%). A t atmospheric pressure, methyl-substituted benzothiophenes showed lower H D S conversion (39-42%) compared to nonsubstituted benzothiophene (58%) using n-Ci + n-C\2 solvent at 450 °C. Introduction of a 17 Literature review Table 2.2 Comparison of H D S conversion of individual polyaromatic S-containing compounds using different solvents over sulfided C o M o / A l 2 0 3 ( L a n d a u , 1997) Solvent n -C 7 +n-Ci 2 n - C i 6 Testing conditions: 450 Temperature, °C 1 Pressure, atm Conversion % 300 50 Benzothiophene (BT) 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 Literature review second methyl group to form 3,7-dimethylbenzothiophene produced a further decrease in H D S to 14%. Introduction of methyl groups to D B T at the 4-position, especially in the 4,6-positions, reduced the H D S conversion to 7-32% of D B T , the 4 , 6 - D M D B T being 30-80% less reactive than the 4-methyl-DBT. However, introduction of 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 of ultra-low sulfur gasoline starts from June 2006 and the sulfur content of diesel w i l l reduce from present levels of 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 oi l (Girgis and Gates, 1991). Gates et al. (1997) pointed out that the 4-methyldibenzothiophene and the 4,6-dimethyldibenzothiophenes are the most appropriate compounds that should be used for investigations for activity and reaction mechanisms for H D S . It is not surprising that recently a number of 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 of the 4 , 6 - D M D B T refractory sulfur compound. Figure 2.1 shows the reaction pathway for the H D S of 4 , 6 - D M D B T (Prins et al., 2006). The conversion of this refractory 4 , 6 - D M D B T compound occurs through two routes. In the first route, direct elimination of 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 of a series of 19 Literature review Figure 2.1 Reaction networks of the H D S of 4 , 6 - D M D B T (Prins et al., 2006) 20 Literature review hydrogenation of the aromatic ring to form the tetrahydro 4,6 dimethyldibenzothiophene (4,6-D 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 of the 4 , 6 - D M D B T is affected by the position of 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 of the molecule for D D S (Landau, 1997). When 4 , 6 - D M D B T is adsorbed on the surface of 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 of the sulfur atom and therefore hinder the molecule from binding on to the catalyst surface. Hence adsorption of 4 , 6 - D M D B T is weak resulting in the D D S pathway being strongly suppressed (Landau, 1997). On 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 of the reaction pathway of the 4 , 6 - D M D B T reveals that in order to enhance reactivity of 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 of 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 of N-containing compounds from petroleum feed stock. Historically, refineries have been more concerned with the removal of S-containing 21 Literature review compounds than N-containing compounds. However, more recently, much attention has been directed towards H D N because of the tightening of environmental legislation regarding the release of 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 of increasing interest in converting petroleum residua, coal, shale and tar sands, which contain higher concentrations of 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 in this section a discussion of 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 oi l is an important knowledge that is required for the development of an improved catalyst for H D N activity. Detailed review on this aspect has been documented in the literature (Katzer and Sivasubrumanian, 1997, Jin et al., 1997; Harvey et al., 1985;Majolskyetal . , 1987). N-containing compounds in petroleum are normally divided into heterocyclic and nonheterocyclic compounds. Katzer and Sivasubramanian, (1979) reported that most of 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 of 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 of phosphorous to MM0S/AI2O3, the pseudo-first order rate constants for H D N decreased in the order pyridine (20.9 min"1) > quinoline (6.6 min"1) > acridine (2.2 min"1) > benz(a)acridine (1.0 min" 1). Generally, the heterocyclic compounds are divided into basic and non-basic nitrogen containing compounds as shown on Table 2.3. The basic N compounds consist of six-membered heterocyles such as pyridine, acridine quinoline and their substituted analogues. The heteronitrogen atom in these compounds has one pair of electrons that are not contributing to the TX -electron cloud of the heterocyclic ring and therefore they are available for donation to acid sites on catalyst surfaces. On the other, the non-basic N-containing compounds consist of five-membered heterocycles such as pyrrole, indole, carbazole and substituted carbazoles. The heteronitrogen atom in the five- membered ring contains two lone pair of 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 of neutral pyrrole benzologues and basic pyridine benzologues in Athabasca bitumen. Jokuty et al. (1991) and Jokuty et al. (1992) have discussed the presence of neutral N compounds in synthetic crude oi 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 Nonheterocycl ic c o m p o u n d s Aniline Nonbas ic heterocycl ic c o m p o u n d s Pyrrole Indole Carbazole B a s i c Heterocycles Pyridine Quinoline Acridine C 6 H 5 N H 2 C 4 H 5 N C 8 H 7 N C 1 2 H 9 N C 5 H 5 N C 9 H 7 N C ^ N -NH, H H concentrated the basic N obtained from the synthetic oi 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 of the number of aromatic rings in the 24 Literature review molecule present in gasoline fractions of 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 of N-containing compounds present in gas oi l derived from tar sand is typically 2 to 5 times that obtained from petroleum crude oi 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 in these synthetic derived oils because of the presence of high nitrogen content, high aromaticity and low hydrogen. 2.2.2 Difficulties of H D N Before discussing the reaction mechanisms of H D N , it is important to discuss the difficulty of 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 in Table 2.4, the bond energy of the C - N bond is 308 kJ/mol and it is higher than the bond energy of the C-S bond (259 kJ/mol). Also the C=N bond energy is 615 kJ/mol which is higher than the bond of 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 2 before the N heteroatom can be removed. The saturation reduces the relatively large energy of the C = N bonds and therefore facilitates easier scission of the C - N bond. The difficulty of H D N is also due to steric hindrance. For example, the C atoms of the neighboring benzene rings of 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 of magnitude while on quinoline introduction of the alkyl groups did not significantly affect the H D N reactivity. The electronic nature of the substituted alkyl groups predominantly determines the Table 2.4 Bond Energies for some hetero-atoms in aromatic and saturated molecules [Fox, M . A . and Whitesell, J. K , 1994] Bond Energy, Bond Energy, Type kJ/mol Type kJ/mol C - H 413 C — N 308 c - c 348 C = N 615 C = C 614 891 C = C 839 c - s 259 N - H 391 C = S 577 26 Literature review reactivity of alkyl pyridines. In pyrroles, alkyl substituents are less important as there is high electron density of 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 of N-alkylated pyridines and pyrroles seems to be the result of complete screening of 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 in structure and size. These heavy feedstocks w i l l also have higher boiling points and the bulky nature of the molecules w i l l reduce their accessibility to the catalyst surface. H D N is also difficult because some of 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 self-inhibitors both of 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 of pyridine (Mcllvried, 1971; Hanlon, 1987; Sonnemans and Van den, 1973) quinoline (Katzer and Sivasubramanian, 1979; Kherbeche et al., 1991; Satterfield and Cocchetto, 1981; Mi l l e r and Hinnemann, 1984; Perot, 1991; Minderhoud and Van Veen, 1993; Jian and Prins, 1998), acridine (Zawadski et al., 1982) indole (Odbunmi and Oll is , 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 of N-containing compounds (basic and non-basic heterocyclic) that have to be removed in H D N . Therefore in this section, the reaction networks of quinoline (basic) and carbazole (non-basic) are presented. Quinoline is representative of basic heterocyclic N-containing compounds and carbazole the non-basic heterocyclic N-containing compounds. The non-heterocyclic N -containing compounds are facile in H D N and therefore they wi l l not be considered. Generally, the pathways for removing the heteronitrogen atom from heterocyclic compounds is as follows: 1) hydrogenation of the N-ring, 2) C - N bond scission to an amine and 3) hydrogenolysis of 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 of the presence of an abundance of alkenes in the H D N products, it was earlier conceived that elimination of 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 Quinol ine Quinoline (Q) has a six-membered heterocyclic ring as well as a phenyl ring and consequently hydrogenation and hydrogenolysis are present in 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 of quinoline where it can be denitrogenated in two ways: via o-propylaniline (OPA) and via 28 Literature review Q THQ-1 OPA THQ-5 DHQ Figure 2.2 H D N network of quinoline over sulfided M - M 0 / A I 2 O 3 catalysts (Jian and Prins, 1998) decahydroquinoline (DHQ). Both pathways require that the strong C - N bond in the aromatic ring be hydrogenated and therefore hydrogenation of the N-r ing or both the N-r ing and the benzenoid ring takes place. In the pathway to O P A , hydrogenation of Q leads to the formation of 1,2,3,4, tetrahydroquinoline (THQ-1) and THQ-1 is mostly in equilibrium with Q (Satterfield and Cocchetto, 1981). After hydrogenation, ring opening of the heterocycle of the THQ-1 to O P A 29 Literature review takes place via the highly active O P A intermediate. The rate of 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 of H 2 S only about 40% of the H D N is converted. After the formation of 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 of THQ-1 . High conversion of 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 MPa) , 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 of 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 of carbazole has been reported (Nagai et al., 1988) using a micro reactor and M0S2/AI2O3 catalyst and also on nitrided M o / A l 2 0 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 of the C - N bond without hydrogenation is difficult, biphenyl (BP) 30 Literature review H Perhydrocarbazole Bicyclohexyl Figure 2.3: General reaction scheme of H D N of carbazole (Nagai et al. , 2000) is not observed. Hydrogenation of carbazole leads to the formation of 1,2,3,4-tetrahydrocarbazole, which is further hydrogenated to form equilibrium products of hexahydrocarbazole, decahydrocarbazole and perhydrocarbazole. The major product bicyclohexyl is formed from the C - N bond cleavage of perhydrocarbazole. Cyclohexylcyclobenzene product was formed from decahydrocarbazole although 31 H Carbazole B i P h e n y ' H H H 1,2,3,4-Tetrahydro- Hexahydro- Decahydro-carbazole carbazole carbazole Cyclohezylbenzene Cyclohexyl-cyclobenzene Literature review decahydrocarbazole was not detected on the nitrided catalyst. Cyclohexylbenzene was formed from hexahydrocarbazole. The authors also reported that upon sulfidation of the reduced catalysts, the hydrogenation was enhanced with the formation of hexylcyclohexane. On the nitrided catalysts, Nagai et al. (2000) reported an H D N rate of 0.34 pmolh"'m~2 after 3 h of time 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 of the commercial sulfided metal catalyst w i l l be presented. Subsequently, the introduction and development of 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 of the sulfides, nitrides and phosphides. Surfaces that are able to transfer electron density into the C-S antibonding orbitals of the reacting molecule should facilitate decomposition of the C-S bond (Rodriguez et al., 2003). The role of 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 of the S 2 H 2 • 4 H + + 4 e (2.1) S 2" + 2 H + • H 2 S +•(sulfur vacancy) (2.2) reacting molecule. In the case of butadiene it is formed from thiophene as follows (Kabe et al., 1999). C 4 H 4 + 2 H + + 4e • C 4 H 6 + S2" (2.3) 32 Literature review 2.4.1 Use of transition-metal sulfides as hydroprocessing catalysts Conventionally, the transition-metals Co or N i is added to M0/AI2O3 or W/AI2O3 catalysts as promoters for simultaneous hydrodesulfurization and hydrodenitrogenation. Traditionally, sulfided forms of 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 2 0 3 is the desired catalyst for H D N (Prins, 2001). The N i W / A ^ C h system is usually employed in cases where hydrocracking is required. Since C0M0/AI2O3, Ni-Mo/Al20"3 and N i W / A ^ C h are commercial hydroprocessing catalysts, many 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 of 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 of S after sulidation. Sulfidation of C0M0/AI2O3 and N i M o / A l 2 0 3 has been observed using X P S (Houalla and Delmon, 1981) and E X A F S have been used to identify N i - M o - S structure in sulfided MM0/AI2O3 (Louwers and Prins, 1992). 2.4.2 Use of transition metal nitride and carbides as hydroprocessing catalysts. The development of 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 of 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 of catalysis by nitriding M0O3 and WO3 with ammonia to obtain high surface area M02N and W2N, respectively. Subsequently, other investigators (Oyama, 1996; Nylon et al., 1999) have reported the use M02N for H D N and H D S and have 3 3 Literature review found the nitride catalyst to have comparable activities to those of commercial alumina-supported metal sulphide catalysts (Volpe et al., 1983). Details of the preparation of nitrides can be found in the literature (Volpe et al., (1985); Oyama, 1996; Nylon et al., 1999; L iaw et al., 1995). Essentially, the preparation of the nitrides involves the use of 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 of mixtures of H2 and N2 instead of 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 of Y-M02N and (3-Mo2Nn.78 metallic M o and unconverted M0O2. The face centered cubic crystal (fee) of Y-M02N was the main species found at lower temperatures. The authors also reported that the final properties of the nitride catalysts are influenced by the space velocity of the synthesis gas, heating rates and composition of the reactant gas mixture. The main problem with Y-M02N is its stability due to its reduction in H2S as given by the equation: M o 2 N + H 2 S • 2 M o S 2 + N H 3 + 2.5 H 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 of some of the metal phosphides. M o P is isostructural with W C with the non-metal containing prisms stacked on top of each other. The metal phosphides o f groups 6-10 adopt the M n P and N i P structures both of which can be regarded as distortions of 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 of metal phosphides are similar to transition metal borides and selicides and also in 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 in H D N and H D S (Oyama et al., 2001). 35 Literature review N i 2 P , Fe 2 P P"62m Hexagonal CoP, C o 2 P , F e P M n P *• nma Orthorhombic N i P Pbca Orthorhombic FeP, £ nm2l Orthorhombic metal Figure 2.4: Crystal structures of some transition metal phosphides 36 Literature review The mechanism by which phosphides (e.g. N i 2 P ) react to remove nitrogen from organic 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 of the supported catalysts (Oyama, 2003). Oyama (2003) also reported that unlike the sulfides that use P-carbon, the reactivity order of 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 of 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 of metal phosphides. The ability of N i 2 P to selectively remove S without excessive hydrogenation of the aromatic ring was suggested to be due to the exposure of many crystallites corners and edges. Although detailed preparations of bulk and supported metal phosphides w i l l be presented in chapter 3, a general description of the preparation of these catalytic materials is shown in 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 in an oven at 110 °C and then calcined at 500 °C for 6 h. The oxide precursor obtained is reduced in H 2 using temperature programmed reduction. The reduced phosphide is then passivated using 2% 0 2 in A r . 37 Literature review Prepare a solution containing both the metal and the phosphorous Evaporate solution on hotplate to form paste Further dry in oven at 110 °C until all water is removed i Calcine dried product at 500 °C for 6 h i Apply temperature programmed reduction using H 2 I Passivate using 0 2 / A r Figure 2.5 Schematic diagram of the preparation of 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 of the hydroprocessing activity of metal phosphides that has been reported. Stinner et al. (2001) reported the activity of bulk C o 2 P , N i 2 P , M o P , W P , C0M0P and N i M o P in 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 of the o-propylaniline, were given as 4.6%, 4.9%, 16.0% and 21.5% for C o 2 P , N i 2 P , M o P and W P , respectively. Intrinsically, M o P was reported as the most active catalyst. 38 Literature review Table 2.5: Summary of hydroprocessing activity of transition metal phosphides Catalyst Testing conditions Conversion Reference C02P Feed: o-propylaniline N i 2 P Micro-reactor at 673 K 3.0 M P a M o P W P N i 2 P / S i 0 2 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 M o P / S i 0 2 Feed: 3.2 mo l% thiophene in H 2 . F low reactor: 643 K , 1 atm. N i 2 P / S i 0 2 Feed: 3.2 mo l% thiophene in H 2 . M o P / S i 0 2 F low reactor: 643 K , 1 atm. M o P / S i 0 2 Feed: o-methylaniline Continuous flow reactor: 643 K , 3.0 M P a N i 2 P / S i 0 2 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 N i 2 P / S i 0 2 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 N i 2 P / M C M Feed : D B T at 613 K , 5.0 M P a 4.6 a Stinner et al. 2001 4.9 a 16.0 a 21.5 a 90.6 a 1 ' 30.4 b l Oyama et al. 2002 32 & i ' ^ 1 bi Wang et al. 2002 4.21 c Phill ips et al. 2002 14.8 d 7.4 d 26 99 a l 9 1 b . 1.3e 2.5 f 100 m Rodriguez et al. 2003 Zuzaniuk et al. 2003 Oyama et al. 2004 Oyama, 2003 Wang et al. 2005 a: HDN conversion of o-propylaniline; al: conversion of DBT; bl: conversion of quinoline; c: conversion of thiophene on MoP/Si02 relative to sulfided Mo/Al203. d: conversion of thiophene relative to sulfided Mo/Si02 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 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 of transition metal phosphides for the H D S of D B T and H D N of quinoline at 643 K and 3.9 M P a H 2 follow the order Fe 2 P < CoP < M o P < N i 2 P . The author further reported that the H D S of D B T and H D N of quinoline over N i 2 P / S i 0 2 were 1.3 times and 2.5 times more than that of a commercial sulfided N i M o / A l 2 0 3 respectively. Since N i 2 P and M o P have been reported to be catalytically active, a number of reports have been presented in the literature on the activity of supported N i 2 P / S i 0 2 phosphide (Rodriguez et 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 of 30 wt% N i 2 P / S i 0 2 using 3.2 mo l% thiophene in H 2 at 643 K and 1 atm in a flow reactor was 14.8 times that of sulfided M o / S i 0 2 and that M o P / S i 0 2 showed 7.5 times the activity of the sulfided M o / S i 0 2 . Thus in all the reports, N i 2 P / S i 0 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 of M o S 2 hydrotreating catalysts with Co or N i is well known (Kabe et al., 1999) but metal phosphides are not promoted by Co 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 of activity studies with the addition of Co 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 of metal phosphides promoted with Co or N i Conversion H D S / H D N Catalyst Testing conditions route Reference Bulk Ratio Co(Ni) :Mo C o M o P (1:1) N i M o P (1:1) C0M0P/AI2O3 -N i M o P / S i 0 2 CoMoP/SiC-2 -M o P / S i 0 2 N i M o P / S i 0 2 N i 2 P / S i 0 2 N i M o P / S i 0 2 -C o M o P / A l 2 0 3 Feed: o-propylaniline Micro-reactor at 673 K 3.0 M P a Feed: 4 , 6 - D M D B T Fixed bed reactor: 603 K , 3.0 M P a Feed: o-methylaniline Continuous flow reactor: 643 K , 3.0 M P a Feed:5000 ppm D B T / decaline Fixed bed reactor, 593 K , 3.0 M P a , 12 h"1 W H S V Feed: 3.2 mol% thiophene in H 2 . F l o w reactor: 643 K , 1 atm. Feed: Thiophene + Pyridine 643 K , 3.0 M P a 15% a 1 8 % a Stinner et al. 2001 93 Hydrogenolysis Mizutani et al. 2005 0.5B 100 e 100 f Zuzaniuk and Prins, 2003 0.96° Hydrogenation 0.96° Hyrogenation 0.1 d Hydrogenolysis 0.8 d Hydrogenolysis Sun et al. 2004 3.0 d Hydrogenolysis Rodriguez et al. 2003 L i et al. 2005. a: HDN conversion of o-propylaniline relative to MoP; b: Thiophene conversion relative to Ni-MoS2/Si02; c: HDN conversion of o-methylaniline relative to MoP/Si02 ; d: TOF (xlO'3 s"1) based on CO chemisorption; e and f: conversion of thiophene and pyridine at conditions in reference. 41 Literature review of 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 of 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 of the nickel atoms in N12P with M o atoms (i.e. a N i / M o ratio of 1/1). The authors then reported the activity of supported NiMoP/SiC»2 using 3.2% thiophene in H 2 in a flow reactor at 1 atm and 643 K , to be half that of the activity of sulfided M o / S i 0 2 . Zuzaniuk and Prins (2003) also prepared N i M o P / S i 0 2 and CoMoP/S i02 with N i / M o and C o / M o ratios of 1/1 and tested the activity of the prepared catalysts using o- methylaniline in a continuous flow reactor at 643 K and 3 M P a . The authors reported the activities of the N i M o P / S i 0 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 2 P / S i 0 2 respectively. The 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 of M o P / S i 0 2 , a series of N i M o P / S i 0 2 and N i 2 P / S i 0 2 using a fixed bed reactor at 3.0 M P a , 593 K and 12 h"1 W H S V . The authors reported TOFs of 0.1 x 10"3 s"1 for M o P , 0.1-1.4 x 10"3 s"1 for the series of N i M o P / S i 0 2 and 3.0 x 10"3 s"1 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 of 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 of H D S of D B T . Mizutani et al. (2005) reported the activity of C o M o P / A l 2 0 3 using 4 , 6 - D M D B T in the presence of quinoline, in a fixed bed reactor at 623 K , 3.0 M P a and 13.5 h"1 L H S V . The authors 42 Literature review reported high D M B P selectivity (42%) and low M C H T (27%) over C o M o P / A l 2 0 3 . Thus, the hydrogenolysis route of the H D S of 4 , 6 - D M D B T was preferred to the hydrogenation route. Recently, L i et al. (2005) reported the activity of C0M0P/AI2O3 to be higher than M0P/AI2O3 using pyridine. The authors also reported that small amounts of Co cause a significant increase on the hydrotreating activity of M0P/AI2O3. Previous reports (Zuzaniuk and Prins, 2003; Rodiguez et al., 2003) on the effect of addition of a second metal such as Co or N i to metal phosphides suggests that when the Co or N i was added to M o in a ratio of 1:1 to form C0M0P or N i M o P , the activity was less than M o P and therefore addition of Co or N i was not beneficial. However there are few reports that examine the addition of small quantities of Co (less than stoichiometric quantities) and N i to M o P or M2P. Small quantities of Co and N i have been used as promoters for commercial sulfided catalysts and it is expected that when added in such small quantities the dispersion on the surface of the catalyst w i l l increase and hence increase activity. Since the S stability of the catalyst is important during hydroprocessing, the stability of transition metal phosphides was reported by Stinner et al. (2001). The authors reported that transition metals are thermodynamically stable in the presence of H2S and S. However, Oyama et al. (2002) reported the presence of a phosphosulfide phase after reaction. The authors tested Ni2P/Al203 with 3000 ppm D B T and 2000 ppm of quinoline in a three-phase packed-bed reactor at 643 K and 3.1 M P a and analysed the surface of 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 (DFT) and confirmed the presence of 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 of the N i 2 P consisted of 50% sulfur replacing phosphorous and some atomic 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 of refractory S compounds such as 4,6-D 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 of transition metal phosphides for hydroprocessing. 2.5 Developing a new phosphide catalyst for enhanced hydroprocessing Ho (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 2 pressure and contact time. However, high H 2 pressure results in increased costs and high temperature reduces fuel quality by cracking the oi l feedsock (Ho, 2004). Hence, in order to enhance hydroprocessing activity of metal phosphides without vigorously changing the hydroprocessing variables, certain catalytic approaches are adopted to modify the metal phosphides. Similar catalytic methods that were used for enhancing activity in sulfided catalysts are being adopted in the present study. Some of the approaches include incorporating different promoters such as Co and N i , using acidic supports or fluorine to induce isomerisation of the 4,6-D M D B T , and increasing the hydrogenation capability of the metal phosphide by adding noble metals. Traditionally, Co and N i are added to M o S / A l 2 0 3 and WS/A1 2 03 to substantially increase their activities in hydrotreating (Prins and De Beer, 1989) and different explanations have been 44 Literature review attributed to this enhancement effect of Co 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 in the presence of M0S2. The promotion of Co was attributed to the interaction of separate Co ions with M0S2 and that the activity decreased with the formation of more CogSg crystals which are regarded as inactive. CogSg crystals are formed when C o / M o ratio > 0.5 (Farrgher and Cossee, 1973). Co-Mo-S phase as shown by in-situ emission Mossbauer spectroscopy is responsible for enhancement of activity of CoMoS 2 /Al203 (Topsoe et al.1996). Sun et al. (2004) using D F T reported that N i added to WS2/AI2O3 form stable N i 3 S 2 at the W edge but not at the S edge while Co 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 in C09S8 makes it difficult to redisperse to the WS2 edges once the well-structured C09S& crystallites are formed. Since isomerisation of 4 , 6 - D M D B T can enhance H D S activity, acidic supports have been the object of 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 4,6-D 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 of the first attempts to associate acidity to the enhancement of 4 , 6 - D M D B T . The authors prepared C o M o /A l20 3 catalyst by impregnation of the 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 of 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 of the zeolite. Isoda et al. (1996) prepared C0M0/AI2O3 and incorporated 5 wt% of Y zeolite. The authors then compared the activity of the prepared catalyst with commercial C0M0/AI2O3 and N i M o / A l 2 0 3 using gas oi 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 of 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 of 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 of 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 of acidic supports in 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 of the 4 , 6 - D M D B T molecule to undergo transalkylation or positional isomerisation of 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 of the S heteroatom with the catalyst active surface thereby increasing activity. 46 Literature review On the other hand, hydrogenation of the 4 , 6 - D M D B T leads to the formation of 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 of Co 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 Co or N i (Stinner et al., 2001; Zuzaniuk and Prins, 2003; Rodriguez et al., 2003). A s observed in earlier reports (Mizutani et al., 2005; Stinner et al., 2001; Zuzaniuk and Prins, 2003; Rodriguez et al., 2003; L i et al., 2005) Co 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 of C0M0S2/AI2O3 produces the less active CogSg phase (Kabe et al., 1999). Therefore in the present study, the Co wi 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 Co added to M o P , the dispersion w i l l increase leading to the formation of stable and active sites on the catalyst. 2.5.2 Promotional effect of a third metal component A combination of supported C0M0P and N i M o P on alumina is used to form trimetallic (combination of three metals) catalyst of 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 of 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; Kang et al., 1998; Severino et al., 2000). Recently, Homma et al. (2005) reported the activity of 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 Co and N i sulfide on the alumina surface which decreased the number of active sites. The number of active sites is related to the sulfur-vacancies and measurements of 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 of 100 N/granule. Alumina is also relatively cheap (Euzen et al., 2002). Alumina 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 of the active phase and creation of 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). Sil ica was reportedly a superior support because of 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 of 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 of the M o P active phase at this temperature. Therefore the role of supports in relation to enhancing the activity of 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 of causing isomerisation of 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 in some refineries as a promoter of hydrotreating and hydrocracking (Kasztelan et al., 1999). Benitez et al. (1996) fluorinated W / A 1 2 0 3 and N i W / A l 2 0 3 and reported increased H D N of pyridine. The authors attributed the enhanced activity to changes in morphology o f the W S 2 crystallites (higher stacking and larger pores), high surface acidity, and better sulfidation of the W oxidic phase. Van Veen et al. (1993) reported enhanced H D N activity after fluorination of N i M o / A l 2 0 3 . The authors attributed the enhancement to the interaction of the N i M o S phase with the A 1 2 0 3 that leads to increased stacking of the M o S 2 slabs and better accessibility of the active sites for adsorption. Hence with higher reactants adsorbing on the active sites, high H D N activity is obtained. Qu and Prins (2003) reported a small increase in H D N of methylcyclohexylamine after fluorinating sulfided N i M o / A l 2 0 3 at 310-350°C and 5.0 M P a . Lewandowski et al. (1997) also reported the presence of Bronsted acidity on sulfided NiMo/Al 2 0"3 on addition of fluorine and proposed that hydrogenation takes place on anion vacancies of M o - S while the breaking of 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 of the catalyst which promotes isomerisation. Recently, Ding et al. (2006) reported the activity of sulfided N i M o / A l 2 0 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 of the reports on the addition of 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 of 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 plat inum 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 of 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). Also 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 of hydrogen spill over in 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 of 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 in 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 of 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 of platinum to the phosphides w i l l result in increased conversion of 4 , 6 - D M D B T as more hydrogenated intermediates w i l l enhance desulphurization as illustrated earlier in Figure 2.6. 51 Literature review 2.7 Effect of Process Variables Tailoring process conditions such as temperature, pressure and space velocity for optimum operations of 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 of 14 °C reduced the sulfur level of Straight Run 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 of reactor temperature w i l l reduce the sulfur from 140 -120 ppm. However, increase in reaction temperature results in decrease of catalyst life hence, catalysts w i l l need to be changed a little more frequently which consequently wi l l increase cost. Lappas et al. (1999) reported that using sulfided C0-M0/AI2O3, they were able to reduce sulfur content of 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 of dibenziothiophene (DBT) over supported Ni2P/MCM at different reduction temperatures and reported that below reduction temperatures of 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 of 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 of conventionally hydroprocessed crude oi l since they form a substantial portion of the residue. 4,6-D M D B T are refractory and resist hydroprocessing because the methyl groups at the 4- and 6-positions possess steric hindrance that prevent the hetero S atom of 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 of 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 of small amounts of Co and N i to M o P and N i 2 P has not been previously examined. Reports on the use of metal phosphides for H D N of 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 of metal sulfides. Therefore it is expected that incorporating Bransted acidity either by use of acidic supports or by F w i l l also enhance activity of the metal phosphides. 53 Chapter 3 Experimental Experimental Since the present research deals with the development of active catalysts for H D S and H D N , experiments were carried out to assess the activity of 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 in the study. In addition to describing the experimental procedures and methods of catalyst preparation, this chapter w i l l describe various catalyst characterization methods used in the present study that included, Temperature Programmed Reduction (TPR), Powder X-ray Diffraction ( X R D ) , Energy Dispersive X - R a y Emission ( E D X ) , X-ray Photoelectron Spectroscopy (XPS) , n-propyl amine chemisorption, C O uptake and Transmission Electron Microscopy ( T E M ) . In the first part of the chapter, the preparation of the bulk Co promoted M2P, M o P and the supported forms of these catalysts are described. The catalyst activity tests, product analysis and characterization for H D S of 4 , 6 - D M D B T is also presented. In the second section, the preparation of bulk and supported N i promoted M o P , the catalyst activity tests for H D N of carbazole and the product analysis, is presented. 3.1 Preparation of metal phosphides for hydrodesulphurization Co and N i are known promoters of sulfided commercial hydroprocessing catalysts and their addition to metal phosphides is investigated here. The bulk metal phosphides M2P, M o P , CoP and C o 2 P were prepared, together with CoxNi2P. The preparation of M2P, M o P and CoP has already been reported in the literature (Oyama, 2003). In preparing C o x N i 2 P , the concentration of Co was varied for the CoxNi2P and once the optimum Co content was obtained, a solution containing the optimum amounts of Co, M o and P was used to impregnate A l 2 0 3 , yielding the 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 CoxNi2P and Co x MoP Bulk N i 2 P , M o P , C o 2 P , C o x N i 2 P (0 < x < 0.8) and C00.07M0P were initially prepared as described elsewhere (Oyama et al., 2001; Stinner et al., 2001; Stinner et al., 2000) to provide reference activity data for the synthesis of supported phosphides. The preparation of C o x N i 2 P w i l l be illustrated here (See Appendix B for detailed calculations). The C o x N i 2 P catalysts were prepared by dissolving stoichiometric amounts (Ni/P = 2/1) of nickel nitrate (Ni (N03) 2 . 6H 2 0 , Aldr ich 99% purity) and diammonium hydrogen phosphate ( D A H P , (NH4) 2 HP04 Sigma 99% purity) in a beaker containing 15 ml of de-ionized water. The resulting solution was stirred at room temperature while adding 10 m l of a cobalt nitrate solution, prepared by dissolving an appropriate amount of cobalt hexahydrate ( ( C o ( N 0 3 ) 2 . 6 H 2 0 , Acros) in 10 m l of de-ionized water to give the desired mol% Co in the final C o x N i 2 P catalyst. After addition of 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 x N i 2 P precursor) was ground to a powder (dp < 0.7 mm) and subjected to temperature programmed reduction (TPR) in H 2 (Praxair - 99.99%) at a flow rate of 150 ml(STP)/min and a temperature ramp rate of 1 K / m i n to a final temperature of 1000 K . The final temperature was held for a period of 2 hours. After the reduction, the catalyst was cooled to room temperature in He at a flowrate of 20 ml/min. Prior to removal from the reactor, the C o x N i 2 P was passivated in 2% 0 2 / H e for 2 h at room temperature. The C00.07M0P was prepared similarly using ammonium heptamolybdate ( (NH 4 ) 6Mo 7 024 . 4H 2 0 -Ana l aR (BDH) 99% purity), D A H P and C o ( N 0 3 ) 2 . 6 H 2 0 55 Experimental to give 3.2 m o l % Co 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 of 1 K / m i n in H 2 flow at 150 ml/min per gram sample. 2. Maintain at final temperature for 2 h. 3. Cool to room temperature in He flow at 20 ml/min. The catalysts were compared to bulk CoP, M o P , N i 2 P and C o 2 P prepared using stoichiometric amounts of C o ( N 0 3 ) 2 . 6 H 2 0 and (NH4) 2 HP04 for CoP, stoichiometric amounts of ( N H 4 ) 6 M o 7 0 2 4 . 4 H 2 0 and ( N H 4 ) 2 H P 0 4 for M o P , and stoichiometric amounts (Ni/P = 2/1 mole ratio) of nickel nitrate N i ( N 0 3 ) 2 . 6 H 2 0 and (NH4)2HPC>4 in the case of N i 2 P . A l l solutions were treated thermally as before, although in the absence of 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 2P was supported on alumina (neutral and weakly acidic) and M o b i l Catalytic Material (MCM-composed of 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 of the A 1 2 0 3 (Aldrich- P V = 0.76 ml/g, pellet size = 4.5 mm), used in the present study is 155 m /g whereas the prepared bulk Coo.4Ni 2P is only 7.9 m /g. It is expected that by supporting the Coo.4Ni 2P on the A 1 2 0 3 , the active Coo.4Ni 2P phase w i l l result in increased dispersion and hence more active sites for reaction. The 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 of the bulk except that in this case the support was impregnated with aqueous solutions of the desired species. A two-step impregnation was performed to avoid the formation of an insoluble precipitate of cobalt phosphate upon mixing aqueous solutions of Co(N03)2.6H20 and ( N H 4 ) 2 H P 0 4 . The modified N12P/AI2O3 and M0P/AI2O3 catalysts were prepared using multiple impregnations. The procedure for preparing the Coo.4Ni 2P/Al203 w i l l be described (see Appendix B for detailed calculations). The catalysts were prepared with theoretical amounts of 3 wt% of 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 3)2 .6H20, and (NH4)2HP04 in de-ionised water to give the required amount of 15 wt% N i in Coo.4Ni 2 P/Ai203. 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 in 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 20 was dissolved in 5 ml of de-ionised water to give a solution containing 3 wt% of Co in 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 of 1 K min" 1 but at a higher reduction temperature of 1200 K (see 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 of 1 wt% of fluorine and the resulting product was dried. The loading of the Co and M o on the support, the heat treatments and the reduction processes were carried out as described earlier. In the case of the platinated Coo.4Ni2P/Al203 catalyst, which was prepared for activity measurements of 4 , 6 - D M D B T , the impregnation of the Pt solution was done after preparing the Coo.4Ni2P/Al203 precursor. Hydrogen hexachloro platinate (IV) hydrate ( H 2 P t C l 6 . 6 H 2 0 , Aldr ich 99.9%) was dissolved in 5 ml of de-ionised water to give an equivalent of 5 wt% of Pt in 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/Al203 precursor was just wet. The resulting product was calcined, reduced and passivated using the same procedure as described in section 3.1.1. 3.2 Preparation of Ni x MoP as hydrodenitrogenation catalyst The procedure for preparing the bulk N i x M o P with different N i concentration was similar to that described for C o x N i 2 P in section 3.1.1. The only exception being that different N i loadings on M o P were obtained by dissolving first stoichiometric amounts (Mo/P = 1) o f (NH4)6Mo7024.4H20 and ( N H 4 ) 2 H P 0 4 to form a solution containing M o P . Solutions containing different amounts of Ni (N03 )2 . 6H20 to give the required loadings of 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 of Coo.4Ni2P/Al203 were followed. The Nio.33MoP was impregnated on the support. The case of supported Nio.33MoP/Al203 wi l l be described. The Nio . 3 3 MoP solution was prepared with 58 Experimental theoretical amounts of 3 wt% of N i and 15 wt% of 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 in Nio. 33MoP/Al203 (see Appendix for detailed calculations and procedure). In each step of 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 of impregnation and drying was carried out several times until all the solution was used up. After this the Ni (N03 ) 2 . 6H 2 0 solution containing the 3 wt% of N i was impregnated onto the support. Once the precursor was obtained, the calcination, reduction and passivation was done in a similar procedure as in 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 of the catalysts. Some used catalysts were also characterized. The different types of 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 of carrying out the T P R was to determine the temperature of reduction and to study the promotional effects of Co loading on the reduction of the main metal M o . T P R can also be used to determine the degree of reduction i f the correct stoichiometry is known and all products are quantified. 59 Experimental T P R of the catalyst precursors was conducted in H 2 using 0.2 - 0.4 g of the calcined sample loaded into a stainless steel reactor (i.d. = 9 mm). A r , flowing at a rate of 60 ml (STP)/min, was passed through the reference side of a thermal conductivity detector (TCD) and a similar volumetric flow of 10% H 2 in A r passed through the reactor before entering the sample side of 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 of 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/min . 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 of 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 of reduction was determined from the integrated area of the T P R of the sample, the assumed stoichimetry of the precursor reduction reaction and the T P R calibration. First, the T P R profile of the sample was integrated to obtain the area. Next, from the stoichiometry of reduction of the sample precursor, the theoretical mol of H 2 was determined assuming complete reduction. From the integrated area of the sample, the mol of H 2 consumed during the T P R of the sample was obtained. The degree of reduction was then calculated as the moles of H 2 consumed divided by the theoretical mol H 2 required for complete reduction (see Appendix B for sample calculation of degree of reduction). 3.3.2 Temperature programmed reduction using tapered element oscillating microbalance (TEOM) The mass change of the catalyst during T P R was also measured for the CoP, N12P and Coo 08Ni2P catalyst using a tapered element oscillating microbalance o f resolution 10 6 g ( T E O M 60 Experimental Series 1500 Pulsed Mass Analyzer, Rupprecht and Patashnick). The catalyst was placed in a fixed-bed configuration and continuous gas flow through the bed during the mass measurement. About 0.1-0.2 g of catalyst was placed in the reactor and heated in He ( U H P , Praxair) at a flow rate of 60 ml(STP)/min and 393 K for 2 h before cooling to room temperature. The reduction in H 2 followed using a purge He (UHP, Praxair) flow o f 120 ml(STP)/min and a pure H 2 ( U H P , Praxair) flow of 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 of the T P R output using T E O M ) . 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 of 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 of 3-70 0 with a step size of 0.04 0 and step time of 2 s. The phase identification was carried out after subtraction of the background using standard software. Crystallite size estimates were made using the Scherrer equation, d c = KA/pcos9 where the constant K was obtained from the integrated areas of the sample and the standard B a F 2 , X, is the wavelength of radiation, P is the peak width in radians and 9 is the angle of diffraction. 61 Experimental 3.3.4 Brunnauer-Emmett-Teller (BET) Surface Area The B E T surface areas of 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 of the catalysts was measured by N 2 adsorption-desorption at 77 K using a Micrometrics FlowSorbl l 2300 analyzer. About l g of 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 of 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 CoxNi2P phosphide where M g K a was used because of the N i ( A ) overlap with the Co 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 eV and for the narrow scan it was 48 eV. 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 in vacuum or under nitrogen (UHP) . A l l X P S spectra were corrected to the C i s peak at 285.0 eV. The catalysts chemical compositions were determined using ICP 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 De 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 in 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 of measuring the C O uptake was to determine the metal dispersion of the catalyst. The C O uptake was measured using pulsed chemisorption. About 0.5 - 1.0 g of 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 of 10% H 2 in A r was passed through the sample loaded in the reactor. 2. The sample was heated from 313 to 723 K at a rate of 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 of H 2 . 4. He at 30 ml/min was used to flush the catalyst for 30 min in order to achieve an adsorbate-free, reduced catalyst surface. After pre-treatment, 1 m l pulses of C O were injected into a flow of He (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 of 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 of 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 -PA. NH3 titrates both the Bronsted acid sites and Lewis acid sites while n -PA titrates only Bronsted acid sites (Farneth and Gorte, 1995). When carrying out chemisorption using n-PA, it is important to note that the temperature programmed desorption (TPD) of n-propylamine on zeolite (ZSM5) 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 3 . However, when the decomposition of n-propylamine is monitored by T C D , 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 -PA 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 of 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 in pure He at a flow rate of 30ml/min for 1 h to ensure that physically adsorbed n -PA was removed. The chemisorbed n -PA was then desorbed by ramping the reactor temperature from 383 to 973 K at a rate of 5 K /min and the T C D was used to quantify the amount of n -PA desorbed. The system was calibrated using 3 zeolite samples of known acidity (see Appendix B) . 6 4 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 in the bulk composition. Before the analysis, the samples were crushed using a pestle and mortar, mounted on a small magnetic circular steel-holder with a carbon felt background and mounted on the instrument. The energy dispersion x-ray emission was done on a Hitachi S300N scanning electron microscope at 20 k V . 3.3.9 Transmission Electron Microscopy (TEM) T E M 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 in a fixed bed reactor (i.d. = 9 mm) operated at 583 K and 3.0 M P a H2. The flow diagram of the reactor set-up is shown in 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 of carbazole in xylene was used in the H D N of carbazole. The 3000 ppm of carbazole was higher than the 700 ppm found in heavy oi l because it is necessary to test the limits of the activity of the prepared modified metal phosphides. The resulting solution was fed to the reactor using a Gilson Model 0154E metering 65 Experimental pump. Prior to entering the reactor, the liquid was evaporated into a stream of flowing H2. Gas and liquid flows and catalyst charged to the reactor were chosen to give a range o f space velocities of 7.9-32 x 10" mol/(h gcat). One gram of the passivated catalysts ( d p < 0.7 mm) was loaded into the reactor and supported with quartz wool while the rest of the reactor volume was filled with glass beads of average diameter 0.1 mm. A thermocouple was placed close to the top of 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 of 160 ml(STP)/min. The temperature was then cooled to the reaction temperature of 583 K and the reaction initiated using the appropriate feed flow conditions. The product was collected every two hours in 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 (CP-Si 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 of the product was also measured using an HP6890 G C equipped with a G2350A 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 Brooks Mass Flow controllers Nitrogen Brooks Mass Flow controllers Hydrogen *—rr Liquid Product Gilson Precision Liquid Pump Figure 3.1 Schematic Diagram of the Experimental Set-Up 6 7 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 of 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, C o 2 P , C0M0P and C o x N i 2 P (0.08 < x < 0.8). The bulk materials were studied so that the measured activity and selectivity could be related to the catalyst properties without interference from the complexity of the catalyst support or promoter. In this chapter, characterization results w i l l be followed by a discussion of the activity of the metal phosphides. Then, the characterization and activity data of supported metal phosphides w i l l be presented. 4.1 Characterization of bulk metal phosphides Characterization of the catalyst precursor was done by TPR, 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 -PA chemisorption were done on the re-reduced catalysts. The characterization was performed to determine the properties of 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 of 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 of the diffraction patterns of the calcined precursors were x-ray amorphous, (see appendix B.5) N i O was identified in the rest of the CoxNi2P precursors in agreement with similar observations made for M o P and N i 2 P precursors (Wang et al., 2002). The results suggest that after calcination, N i was in the +2 oxidation state and M o was likely in the +6 oxidation state (Zuzaniuk and Prins, 2003) although the presence of other amorphous species cannot be excluded (Wang et al., 2002). After preparing the metal phosphides by T P R of the oxidic precursors, X R D was used to determine the bulk phases present. Figure 4.1 shows the X R D profiles of Ni- r ich N i 2 P , C o x N i 2 P , CoP and the Co-rich C o 2 P . In all cases the diffractograms showed that metal phosphides had been successfully prepared. The diffractograms were compared with Powdered Diffraction Files (PDF) ( JCPDS, 2005) and showed that C o 2 P was obtained from samples with Co/P ratios of 2/1 and CoP was obtained with Co/P with a ratio of 1/1 as expected. N o metal oxide or metal phosphate species were detected. The X R D data for the C o x N i 2 P catalysts (Table 4.1), with 0 < x < 0.16 (i.e. up to 5.1 mole % Co) showed the presence of N i 2 P with no significant difference in the N i 2 P lattice parameters (a = 0.5877 ± 0.0003 nm and c = 0.3412 ± 0.0008 nm). For x > 0.34, the C o x N i 2 P showed the development of the metal rich phosphides N i i 2 P s and C o 2 P , in addition to N i 2 P . The N i 2 P lattice parameters decreased (a = 0.5862 ± 0.0002 nm and c = 0.3375 ± 0.0021 nm), although the error associated with the estimate of the lattice parameter c increased 69 Chapter 4 A\CoP A A C o f t 7 0 N L P 0.79 2 C o ^ N L P 0.34 2 Co^JvILP 0.16 2 Co N i P 0.08 2 NLP • • 1 A ~ i — i — i — r — i — i — i — i — r i i I I I I I I T I I I | | | | T™ 30 40 50 60 20 Figure 4.1 X-ray diffractograms of reduced C02P, CoP and CoxNi2P catalysts (A - N i i 2 P 5 , • - C o 2 P , T - N iCoP) 70 Chapter 4 Table 4.1 Lattice parameters estimated from P X R D of reduced catalysts N i 2 P phase Catalyst Co content Phases 20, degree Lattice parameters, nm Crystallite size, nm mole % (111) (201) (210) (300) a c do ( H I ) N i 2 P 0 N i 2 P 40.52 44.44 47.16 54.00 0.5879 0.3417 21 ± 2 Co 0.o8Ni 2P 2.5 N i 2 P 40.52 44.40 47.20 53.96 0.5879 0.3417 36 ± 2 C o 0 . , 6 N i 2 P 5.1 N i 2 P 40.60 44.48 47.14 54.16 0.5874 0.3403 39 ± 2 C o 0 3 4 N i 2 P 10.3 N i 2 P , N i , 2 P 5 N i 2 P , 40.88 44.76 47.32 54.16 0.5860 0.3360 32 ± 2 Co 0.79Ni 2P 21.0 N i , 2 P 5 , C o 2 P 40.66 44.59 — 0.5863 0.3389 27 ± 3 M o P ph ase 20, degree Lattice parameters, nm Crystallite size, nm (100) (101) a c d c (100) M o P 0 M o P 31.93 42.89 0.3240 0.3196 1 6 ± 1 C00.07M0P 3.2 M o P 31.93 42.96 0.3240 0.3179 1 5 ± 1 71 Chapter 4 for these catalysts because the X R D diffractograms were a composite of a number of different phases (Figure 4.1) and the (111) peak of N12P was not well resolved. Co (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 CoxNi2P with x < 0.16, the expected increase in the lattice parameter c would be within the measurement error of the powdered x-ray diffraction ( P X R D ) data. Figure 4.2 shows the X R D diffractogram of prepared M o P and C00.07M0P. The X R D pattern of the C00.07M0P was very similar to that obtained for M o P and did not show characteristic CoP peaks, either because the Co concentration was too low (3.2 mol % Co) and/or because Co was well dispersed in the M o P . The P D F of M o P (PDF, 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 of Table 4.1 show that the presence of C o resulted in a decrease in the lattice parameter c o f M o P , suggesting that in this case, a C o x M o P solid solution was formed, as has been reported for the Ni x Mo ( i . x )P system (Stinner et 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). Addit ion of Co to N i 2 P increased the Ni2P crystallite size (Table 4.1), and the B E T surface area (see Table 4.3) also increased. The bulk CoP, M o P or M2P catalysts had low B E T area (SBET) and large particle size (dc0p = 420 nm, d>ji2P = 206 nm and dM0p = 172 nm, estimated from dp = 6/[SBETP] where p is the bulk density). The particle size was significantly greater than the crystallite dimensions (d c) shown in Table 4.1, suggesting significant agglomeration of the metal phosphide crystallites. The increased M2P and M o P crystallite size with addition of Co to M2P and M o P was likely due to the increased time for complete crystallization compared to the Ni2P, because of the added Co(NC>3)2 solution. The increased B E T surface area of the CoxNi2P 72 Chapter 4 and C00.07M0P suggests a reduction in the degree of agglomeration of the larger crystallites when the Co was added. .JU I J Co 0 0 7 MoP I. A A A JU I J MoP L. A 1A I 1 1 1 1 ' 1 1 1 ' 1 20 30 40 50 60 70 2© * P D F of M o P Figure 4.2: X-ray diffractograms of reduced M o P and C00.07M0P catalysts 4.1.2 TPR of precursors The conventional method used to prepare the metal phosphides involves reducing the precursors under flowing hydrogen using TPR. 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 in the case of CoxNi2P and C00.07M0P. T P R data can also be used to estimate the degree of reduction i f the reaction stoichiometry is known. The T P R of the precursors are conveniently presented in Figures 4.3 and 4.4. M2P, and C o x N i 2 P are presented in Figure 4.3. Two reduction peaks associated with the M2P precursor are attributed to low temperature reduction of N i O to N i , followed by reduction of the phosphate to M2P at higher temperature (Wang et al., 2002). The low reduction temperature obtained in 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 in the present study (60 ml/min) compared to the 1500 ml/min used by Wang et al. (2002). A similar T P R profile was observed for C02P. For CoxNi2P with x < 0.16, increasing Co content increased the temperature of the second maximum, most likely because of the high reduction temperature of the CoP precursor (1173 K ) (see Figure 4.3). A s the Co content increased further such that x > 0.34, both peak temperatures increased significantly, suggesting the formation of different reduced species, consistent with the X R D data that showed the presence of new phases (especially M12P5) in the reduced C o x N i 2 P catalysts with x > 0.34. In Figure 4.4, the T P R of the M o P precursor shows a peak at 955 K . In previous reports on the preparation of 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 6 + 74 Chapter 4 I < i • 1 • 1 1 1— 600 700 800 900 1000 Temperature, K Figure 4.3 T P R of N i 2 P and CoxNi2P catalyst precursors measured in 10 % H 2 in A r at a rate of 60 ml(STP)/min at 1 K / m i n 75 Chapter 4 Figure 4.4 T P R of C o 2 P , CoP, M o P and C00.07M0P catalyst precursors measured in 10% H 2 in A r at a rate of 60 ml(STP)/min reduction to M o 4 + , with subsequent M o 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 of M o + 6 and P 5 + . Addit ion of 3 wt% Co 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 of 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. Furthermore, Oyama et al. (2002) 76 ' Chapter 4 have shown that under certain conditions, some phosphate species are volatile. Assuming the species present after calcination can be written as nMO x . P 2 05 (Stinner et al., 2003), the degree of reduction calculated in Table 4.2 (see Appendix B for sample calculation) suggests that the precursors were not completely reduced to the metal phosphide. However, i f the precursors included species such as H x P04 ( x " 3 ) or P 3 + , the actual degree of reduction would be higher than that calculated. Similarly, i f some unreduced, volatile phosphorous species leave the catalyst during the reduction or calcination process, the actual degree of 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 of the presence of other forms of oxide and/or phosphate species that must be present and not accounted for in the reduction stoichiometry. 4.1.3 T E M Figure 4.5 compares T E M micrographs of the N i 2 P and Coo.o8Ni2P bulk catalysts. The T E M images of N i 2 P and Coo.osNi2P were similar and showed a mosaic crystal structure. Crystal 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 of N i 2 P were identified. A multi-point S T E M - E D X analysis of the Coo.osNi2P catalyst (nominal) yielded the composition Coo.16 ± 0.12 Ni2.oo ± 0.12P1.08 + 0.0, indicative of a uniform catalyst composition. In the case of the Coo.34Ni2P, however, a wider variation in composition was obtained (C00.15 ± 0.18 Ni2.oo ± 0.24P0.11 ± 0.17)? consistent with the presence of different phases identified by X R D . The T E M estimates the sizes of the N i 2 P to be 5.0 A , 3.4 A and 2.8 A , and the X R D indicates larger particles 21 ± 1 nm. The error associated in the 7 7 Chapter 4 Table 4.2 Summary of temperature programmed reduction data Weight Loss T P R Peak Catalyst Temp. Assumed Reduction Stoichiometry N i 2 P K 778, 840 4 N i O . P 2 0 5 + 9 H 2 2 N i 2 P + 9 H 2 0 Apparent Degree of Reduction mole % 72 Calculated from Reduction Degree w t % 24 Measured using T E O M w t % 31 Coo.o 8Ni 2P 785,921 2 C o O . P 2 0 5 + 7 H 2 2CoP + 7 H 2 0 4 N i O . P 2 0 5 + 9 H 2 2 N i 2 P + 9 H 2 0 74 25 51 M o P 955 2 M o 0 3 . P 2 0 5 + H H 2 ^ 2 M o P + 1 1 H 2 0 62 25 2 C o O . P 2 0 5 + 7 H 2 ^ 2 C o P + 7 H 2 0 C 0 0 0 7 M 0 P 1001 66 25 2 M o 0 3 . P 2 0 5 + 1 1 H 2 - * 2 M O P + 1 1 H 2 0 CoP 1086 2 C o O . P 2 0 5 + 7 H 2 ^ 2 C o P + 7 H 2 0 46 18 24 78 Chapter 4 Figure 4.5 T E M micrographs of bulk metal phosphides: (i) N i 2 P and (ii) Co 0 .o8Ni 2 P. Estimated d-spacing of 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 of N i 2 P 7 9 Chapter 4 measurements of 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 in Table 4.3, including the B E T surface area, n -PA uptake and the C O uptake reported per unit B E T area of the catalyst. The data show that after the initial addition o f the 3 wt% of Co subsequent addition o f more C o to N i 2 P leads to agglomeration of crystals that reduces the surface area. Table 4.3 shows that the CoP had the highest n -PA uptake among all the metal phosphides examined. The n -PA uptake on CoxNi2P with x < 0.16 was greater than on N i 2 P . The n -PA uptake increased by about 50% on C00.07M0P compared to M o P . The C O uptake decreased for the C00.07M0P and the Coo.o8Ni2P catalysts compared to M o P and N i 2 P , respectively. However, for the C o x N i 2 P catalysts, the C O uptake increased as Co content of the C o x N i 2 P catalysts increased. Excluding CoP, the catalysts with the highest n-P A uptake, C00.07M0P and Coo.osNi2P, also had the lowest C O uptake. 4.1.5 XPS The X P S spectra of the P 2p region, the Co 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 in the case of C O adsorption and n-PA adsorption. The P 2p3/2 binding energy (BE) associated with metal phosphides has been reported to be 129.5 eV (Senzi and Jae, 1994; Coll ing and Thompson, 1994) whereas for P 2 0 5 the P 2p 3 / 2 B E is 135.2-135.6 eV (Oyama, 2003; Jian and Prins, 1996), 80 Chapter 4 Table 4.3 Properties of the prepared metal phosphides Chemisorption P / M * Catalyst B E T area C O uptake n -PA uptake Nominal X P S m 2 /g urnole/m pmole/m 2 atom ratio atom ratio N i 2 P 4.1 0.27 5 0.50 0.5 Coo.o8Ni 2P 7.9 0.11 18 0.48 4.8 Coo . i 6 Ni 2 P 7.8 0.22 7 0.46 1.8 Co 0 .34Ni 2 P 6.9 0.42 1 0.43 1.2 Co 0 .79Ni 2 P 6.5 0.45 1 0.36 -M o P 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 C o 2 P 5.6 0.16 11 0.5 1.6 * M : total metals (Co+Ni) or (Co+Mo) and P: phosphorous 81 Chapter 4 Figure 4.6 (a) Co 2p 793.8 eV 778.6 eV C 0 2P — - ~ - ~ ~ ~ K N v _ CoP ~~~~~ " ' I - — Co Ni P 0.16 2 Co Ni P J \ r ~ ^ ~ ~ ^ ^ ~ ~ ~ ^ 0.08 2 I 1 1 ' 1 ' 1 1 1 ' 1 • 1 820 810 800 790 780 770 760 B.E., eV Figure 4.6 (b) 145 140 135 130 125 B E . , eV 82 Chapter 4 N i 2p Co 0 . , 4 Ni 2 P-Co N i P 0.16 2 Co„ N L P N L P 853.7eV 852.7eV 870 860 850 840 B.E., eV Figure 4.6 (c) Figure 4.6 X P S of (a) Co 2p region (b) P 2p region, and (c) N i 2p region of CoxNi2P catalysts after reduction and passivation 83 Chapter 4 and 133.3 eV for N i 3 ( P 0 4 ) 2 (Stinner et al., 2001). Hence we assign the low B E peak at 129.5-129.8 eV, present in the C o 2 P , CoP, N i 2 P and M o P spectra to the metal phosphide, and the higher B E peak at 133.4-133.8 eV to surface metal phosphate species. For N i 2 P , this assignment is consistent with the N i 2p B E s that show the presence of N i 2 P and N i 3 ( P 0 4 ) 2 with corresponding N i 2p 3 / 2 B E of 853.7 eV and 857.3 eV, respectively. A similar conclusion can be drawn for the C o 2 P and CoP catalyst with Co 2p B E at 780.0 - 778.6 eV and 793.8 eV and the M o P catalyst with M o 3d B E at 228.2 eV and 232.2 eV. The effect of increasing Co content on the X P S spectra of the C o x N i 2 P catalysts is also shown in Figures 4.6 and 4.7. N o significant shift in B E ' s was observed in the case of P 2p 3/ 2 , whereas in the case of N i 2p, a feature at lower B E (852.7 eV) assigned to Ni- r ich phosphide (Nii 2 Ps) ( L i et al., 2005), increased as the Co content increased. This observation is consistent with the presence of N i i 2 P s identified by X R D in the C o x N i 2 P catalysts with x > 0.34. In addition, as Co content increased, the peak assigned to phosphate species became less significant and the peak at 855.9 eV, 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 2 P . The feature at ~850 eV is likely the L 3 V V Auger line of N i O (Khwaja et al., 1989). Comparison of the Co 2p peaks present in the CoP and C o x N i 2 P (at B E = 780 eV) suggest the presence of metal phosphide although there is a shift to lower B E of about 1.5 eV in the case of C o 2 P , indicative of a more metallic C o 2 P . In the case of C00.07M0P, the M o 3ds/2 B E and the P 2p 3/ 2 B E are almost identical to that observed in the case of M o P . The X P S analyses were also used to calculate the P / M atom ratios of the catalysts prior to their use. These data (Table 4.3) suggest that the surface of the M o P catalyst was slightly enriched in P, and addition of Co resulted in a small increase in the surface P content. In the case of N i 2 P , the surface P / N i ratio was close to 1/2, but following Co addition, a surface enrichment 84 Chapter 4 Mo 3d 245 232.2 e V 228.2 e V 215 145 —I— 140 133.8 e V 129.8 e V 135 130 B.E., e V 125 M o P , M o P 120 115 Figure 4.7 X P S of P 2p region and M o 3d region of M o P and C00.07M0P catalysts after reduction and passivation 85 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 of the present study. Assuming that the C O chemisorption titrates metal sites and the n -PA 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 -PA 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 of the measured ratios that is indicated on the graph. The standard error of 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 in 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 mol balance between the reactor feed and reactor effluent was > 95% for all the data reported herein (see Appendix C) . The reproducibility of the experiments is also found in Appendix C. The metal phosphides had conversions in the range 10 - 49 ± 6 mol%, (see Appendix C for the standard error calculations) and by assuming that the conversion of 4 , 6 - D M D B T on these catalysts was first order at the conditions of the present study, the rate of 4 , 6 - D M D B T 86 Chapter 4 n-PA/CO uptake ratio by adsorp t ion Figure 4.8 Correlation of P / M ratio determined by X P S and n - P A : C O uptake ratio determined by adsorption for C o x N i 2 P (•), C 0 0 . 0 7 M 0 P (A) , CoP (o), C o 2 P ( 0 ) with MoP(A) and Ni 2 P(D) as indicated. The solid line represents the correlation equation: 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 of 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 of catalyst) increased on the C o o . o s ^ P catalyst, but declined on the CoxNi2P catalysts with x > 0.16, compared to M2P. However, the data o f Table 4.4 show that per unit area (BET) , the activities o f al l the CoxNi2P catalysts were lower than the M2P, C02P and CoP catalysts, whereas the T O F based on C O uptake, showed a maximum value for the Coo.o8Ni2P catalyst. The T O F is defined as the number of molecules 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 dimethyl-bicyclohexane ( D M B C H ) , dimethylbiphenyl ( D M B P ) and methylcyclo-hexyltoluene ( M C H T ) with less significant quantities of hydrocarbons that were products of 4 , 6 - D M D B T hydrocracking and hydrogenation. The selectivity of the catalysts to M C H T , D M B C H and D M B P is reported in 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 CoP, and this contrasts with the high selectivity to D D S products that occurs for D B T conversion over N i 2 P and M o P (Sun et al., 2004; Oyama et al., 2002). The low selectivity to D D S products in the case of 4 , 6 - D M D B T is known to be due to steric hindrance by the methyl groups of the molecule. O f particular significance in the present work, however, is the increase in selectivity toward D M B P that occurred when Co was added to both the M o P and Ni2P catalysts, especially for low Co contents. A n increase in D M B P selectivity suggests a significant shift toward direct desulfiirization (DDS) of 4 , 6 - D M D B T rather than aromatic ring hydrogenation. Isomerization of the 4 , 6 - D M D B T molecule likely occurs 88 Chapter 4 Table 4.4 Activities of bulk metal phosphides for the H D S of 4 , 6 - D M D B T measured at 583 K and 3.0 M P a H 2 4,6 D M D B T Consumption Rate Selectivity Catalyst Conversion H D S Specific Areal TOF M C H T D M B C H D M B P mol % mol% 109 mol.g-'.s"1 10 1 0mol.m _ 2 .s" 1 1 0 V mol % N i 2 P 29.1 25.6 2.3 5.5 7.3 2.1 25.7 0.9 Coo.ogNi2P 48.7 47.2 4.4 5.5 18.1 0.8 13.3 57.3 Coo . i 6 Ni 2 P 23.4 20.2 1.8 2.2 3.7 14.6 10.5 40.9 Coo .3 4Ni 2P 17.4 16.8 1.3 1.8 1.6 28.6 4.0 22.2 Co 0 .79Ni 2 P 10.1 8.6 0.7 1.1 0.9 25.1 5.7 16.3 M o P 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 C o 2 P 43.9 41.6 3.8 6.7 1.5 15.7 21.6 33.3 89 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 of sulfur containing products (sulfur speciation) obtained with N12P and Coo.o8Ni2P catalysts is compared in Table 4.5. No 4 , 6 - D M D B T isomerization products were detected, presumably because they underwent rapid D D S on the metal sites of the Coo08N12P catalyst. Evidence for increased acid catalyzed methyl cracking on the Coo.o8Ni2P catalyst versus the M2P catalyst is apparent from the data of Table 4.5, whereas the N i 2 P yielded more hydrogenated product than the C00.08N12P catalyst. These trends are consistent with the promotion of isomerization of 4 , 6 - D M D B T followed by rapid D D S on the more acidic CoxNi2P 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 of the C O uptake in Figure 4.9. For the CoxNi2P series of catalysts, conversion of 4 , 6 - D M B T decreased with increasing 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 -PA uptake decreased, the D M B P selectivity decreased. For the CoxNi2P catalysts studied herein, maximum selectivity to D M B P occurred on the catalyst with the highest n -PA 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 of sulphur speciation in liquid product from H D S of 4 , 6 - D M D B T measured at 583 K and 3.0 M P a H 2 over N i 2 P and Coo.osNi2P Catalyst N i 2 P Co 0 .o8Ni 2P Product Comment m o l % Two methyl groups removed 3 32 One methyl group removed 24 10 Partial hydrogenation of one ring 29 18 Complete hydrogenation of one 44 40 ring 91 CrQ Chapter 4 CO uptake, umole/m2 Figure: 4.9 Conversion of 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 Ff2, plotted as a function of C O uptake: C o x N i 2 P (•), C00.07M0P ( A ) , C o 2 P (0) and CoP (o) with MoP(A) and Ni 2 P(p) as indicated 92 Chapter 4 4.3 Discussion on bulk metal phosphides used for HDS of 4,6-DMDBT Previous studies of 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 of N i / M o ratios, the selectivity to biphenyl was > 80% with D B T as reactant. On N i 2 P / S i 0 2 and C o P / S i 0 2 , Wang et al. (2002) reported 100% selectivity to biphenyl with the same reactant, and with N i x P y prepared over a wide range of N i / P ratios, 100% selectivity to biphenyl during the H D S of D B T was also obtained (Oyama et al., 2002). However, with 4 , 6 - D M D B T as reactant, the hydrogenation activity of N i 2 P was 5 to 7 times greater than the hydrogenolysis (DDS) activity (Oyama et al., 2002). Similar results were obtained in the present study, where selectivities to the D D S product D M B P were < 3% on N i 2 P , CoP and M o P . The low selectivity to D D S is due to steric hindrance effects associated with the methyl groups of 4 , 6 - D M D B T (Kwak et al., 1999; Isoda et al., 2000). A n enhanced D D S of 4 , 6 - D M D B T was reported with P 2 Os addition to conventional metal sulfide catalysts (Kwak et al., 1999) and these authors attributed this effect to an increase in Bronsted acidity associated with the added phosphorous that increased migration of the methyl substituents on the aromatic rings of 4 , 6 - D M D B T . The isomerization would not be important in the H D S o f D B T . The products of skeletal isomerization of 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 in the D D S rate of 4 , 6 - D M D B T has also been reported for P addition to M o 2 C hydroprocessing catalysts (Manoli et al., 2004). The correlation between the surface phosphorous/metal (P/M) ratio determined by X P S , and the ratio of n -PA 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 x P 0 4 ( x " 3 ) , and metal sites that chemisorb C O . The source of the Bronsted acid sites is likely a consequence of the 93 Chapter 4 incomplete reduction of phosphate species, which is accentuated by the presence of Co and the catalyst passivation procedure. Following reduction of the catalyst precursor by TPR, the catalysts were passivated in diluted 0 2 , according to procedures used by others that are known to yield a surface oxidised over-layer (Sawhill et al., 2005). The P 2p spectra of the passivated catalysts showed the presence of metal phosphide and metal phosphate species for all the catalysts. H x P 0 4 ( x 3 ) species, with B E 134.3 - 135.2 eV (Fluck et al., 1974) were not resolved in the X P S spectra, but low concentrations of these species cannot be excluded. The N i 2p spectra of the N i 2 P and Coo.o8Ni2P catalysts showed two peaks that were assigned to N i 3 ( P 0 4 ) 2 and N i 2 P . A s the Co 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 of bulk metal phosphides, after passivation, both metal phosphide and metal phosphate species were present on the catalyst surface, and in some cases metal oxide was observed as well . Prior to determining the C O uptake, the n -PA uptake, or the activity, the catalysts were pretreated in H 2 by heating to 723 K . However, the T P R data show that especially in the case of 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 x P 0 4 ( x " 3 ) and the observed Bransted acidity. The chemisorption data show that CoP, which had the highest precursor reduction temperature, also had the highest n -PA uptake and we assume that this is a consequence o f incomplete reduction and more surface phosphate species following passivation. On 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 of Figure 4.9 suggest that for al l the catalysts, the n - P A uptake decreased as the C O uptake increased, except for CoP which had a very high n -PA uptake that corresponded to a 94 Chapter 4 high reduction temperature of 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.osNi2P, also had the lowest C O uptake and highest n -PA uptake among the C0007M0P and CoxNi2P catalysts. These results demonstrate the importance o f the surface acidity and hence isomerization of 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 x N i 2 P and C o x M o P catalysts is obtained with increasing acid site concentration and decreasing metal site concentration both of which are determined by the amount o f C o added. Note, however, that the C o 2 P , CoP, N i 2 P and M o P do not follow the same correlation between D M B P selectivity and C O uptake shown for the C o x N i 2 P and C00.07M0P catalysts (Figure 4.9). The C o x N i 2 P catalysts must therefore provide unique acidic and metallic active sites that benefit the D D S route. The C o x N i 2 P catalysts had larger N i 2 P crystal size and B E T surface area than the N i 2 P catalyst. Although we do not see direct evidence of a C o x N i 2 P solid solution, the possibility of a well-dispersed C o x N i 2 P that is more active for D D S o f 4 , 6 - D M D B T cannot be excluded. Although C o 2 P may also be present, it seems less likely that increased hydrogenolysis is associated with increased metallic character of the catalysts. The data showed a decrease in D M B P selectivity as more C o was added, corresponding to an increase in C O uptake and X R D patterns that showed more metallic phases (Nii 2 P5 and C o 2 P ) present in the bulk. The increase in metallic character of the N i in C o x N i 2 P as Co was added may also be indicated by the increase in the N i 2p peak at low B E of 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 2 P had the highest D M B P selectivity among the single metal phosphides, the T P R data showed this catalyst was more readily reduced than CoP and had a lower acid site density. The X P S data also show that the B E of Co associated with the 95 Chapter 4 CoP and CoxNi2P catalysts (Figure 4.6 (a)) was shifted about 1.5eV to higher B E (less metallic) compared to the Co associated with C02P (778.6 eV). The significant drop in selectivity observed as more Co is added to the CoxNi2P catalysts suggests that the active phase only has.a high dispersion with x < 0.16. Presumably the CoxNi2P catalysts generate a unique phase, probably a solid solution (Ma et al., 2004), dispersed on the large Ni2P crystals, that result in a better distribution of 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 of Co to M2P, the crystallite size also increases but the conversion as shown in Table 4.4 initially increases but deceases as more Co is added. In the case of 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 Co 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 x N i 2 P catalysts compared to M2P was observed. Furthermore, the change in P / M surface composition as determined by X P S , was consistent with the C O chemisorption data and the n -PA T P D data. A s shown in Figure 4.8, the P / M atom ratio was well correlated with the n -PA to C O uptake ratio. If one assumes that the Bronsted acidity resides in H X P 0 4 ( X " 3 ) species, and all the surface P is in this form with x = 3, then the slope of the line shown in Figure 4.8 would be expected to be 1/3, assuming an adsorption stoichiometry of 1:1 for n -PA: Bronsted acid site and C O uptake: metal site. However, the slope of 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 of the surface P is present as metal phosphide, and this 96 Chapter 4 would tend to increase the slope of the line of Figure 4.8). The limitation of 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 of 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 of ca. 3 mol% Co to N i 2 P and M o P resulted in a significant increase in D D S selectivity during 4 , 6 - D M D B T conversion. The change in selectivity corresponds to a number of observed changes in the catalyst properties, especially surface Bransted acidity and metal sites as determined by n -PA adsorption and C O adsorption, respectively. The most selective catalyst for D D S of 4 , 6 - D M D B T was the C00.08N12P catalyst and this catalyst had the lowest C O uptake and highest n -PA uptake among the CoxNi2P and 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 4,6-dimethyldibenzothiophene The Coo.o8Ni2P catalyst had the highest selectivity for direct desulfurization of 4,6-D M D B T among a series of CoxNi2P (0 < x < 0.34) catalysts and the enhanced selectivity was attributed to the presence of 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 of these catalyst using typical refinery feedstocks is rare. Although the effect of 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 of phosphide catalysts in a real feed that contains both S and N compounds. Furthermore, studies of the phosphides with model compounds suggest that they undergo some sulphidation during reaction, yet the extent of conversion of the phosphides to sulphides or other compounds in the presence of a real feedstock needs to be clarified. The activity of Coo.4Ni2P catalysts, supported on supports of 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 in chapter 6, the activity of the selected catalysts w i l l be presented using light gas oi l ( L G O ) derived from Athabasca bitumen. The properties of the catalysts after reaction with the L G O are also reported in an attempt to quantify the stability of 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 of 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 of the crystallinity of alumina support on the H D S activity of sulfided C0M0 catalysts. The authors also reported that based on the T O F , alumina is better than the other supports. Another important role of the support is in the modification of acid properties of the catalysts. Mauge et al. (2001) used FTIR to study the surface properties of M0S2 catalyst and showed that the unsupported M0S2 catalyst had weaker acidic properties as well as a higher metallic character of M o sites compared to the M0S2 supported Y-AI2O3 catalyst. Y-AI2O3 (Aldrich-155 m 2 /g , 0.76 98 Chapter 4 ml/g), used commercially in hydroprocessing, was chosen as a support in the present study. In addition, supports with different acid properties were also investigated. A s explained earlier in Chapter 2, Bransted acidity is required for isomerisation of the methyl groups on the two benzene rings at the 4,6- positions of D M D B T so as to remove the steric hindrance and allow the S atom to gain access to the surface of 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 of acidity on the isomerization of the 4 , 6 - D M D B T . Supported catalysts were prepared using a similar procedure to that used in 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 of 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 2 0 3 , Coo.4Ni2P/Al203, Coo.4Ni 2 P/Ai203-F, and C o o . 4 N i 2 P / M C M . Also included is the P D F of N i 2 P (JCPDS, 2005). Clearly the NI2P phase is present on the AI2O3 support. The diffraction peaks of 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 of 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. Again consistent with that of the bulk Coo.4Ni2P, the Co phase is not present on the diffractogram as Co was either in small quantities or it was well dispersed on the surface of the catalyst. Since M C M is a mesoporous material, the X R D pattern can only 99 Chapter 4 Co n NLP/MCM 0.4 2 t ' 1 ' 1 1 1 ' 1 ' 1 20 25 30 35 40 45 50 55 60 65 70 2 0 , degrees Figure 4.10 X R D of 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 of 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 of the Coo 4M2P on the support. The appearance of 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 Pt-CoNi 2 P/Al 2 0 3 and Pt-CoNi 2 P/Al 2 0 3 -WA Figure 4.11 shows the X R D diffractrograms of P t - C o 0 4 N i 2 P / A l 2 O 3 and Pt-Co 0 .4Ni 2 P/Al2O3-WA. The X R D patterns indicate that both y - A l 2 0 3 and A 1 2 0 3 - W A supports produce similar catalysts. The diffraction peaks corresponding to P tP 2 were obtained at 27.12° 3 ed fi P t - C o 0 4 N i 2 P / A l 2 O 3 + A P t - C o 0 4 N i 2 P / A l 2 O 3 j + ' A 20 30 40 50 2 0 , degrees + Formation of Ni2P * Formation of PtP 60 70 Figure 4.11 X R D diffractrograms of P t - C o 0 4 N i 2 P / A l 2 O 3 and P t - C o 0 . 4 N i 2 P / A l 2 O 3 - W A 101 Chapter 4 and 31.48°. Pt° has XPvD peaks at 39.6 °, 46.1° and 67.3° (JCPDS, 2005). The diffraction peaks at 40.8 °, 44.7 °, 54.2 °, and 54.9 0 were attributed to the C00.4M2P as observed earlier in 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 of the M C M crystal). After preparing these supported catalysts, they were characterized using B E T , C O uptake and n-P 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 of the pore size is 8.3 nm (83 A) to 19.6 nm (196 A). The size of the 4,6-D M D B T molecule is about 8 A which is much larger than the nitrogen molecule but much smaller than the pore sizes of the supports. Therefore i f nitrogen can access the surface area, then the reactant molecules should be able to gain access to the interior of the pores since that is where most of the B E T surface area is located. Hence, it is valid to normalize the reaction rates with the surface areas. Column 2 of Table 4.6B shows that the B E T surface areas of 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 2 0 3 , Coo.4Ni2P/Al 203-F and Coo.4Ni2P/MCM were 25 umol/g, 24 umol/g and 20 umol/g respectively indicating that the metal dispersion on the surface of these catalysts was the same. The n -PA titrated the Bransted acid sites on the surface of the catalysts and the results in column 4 of Table 4.6B show that the catalyst supported on M C M produced the highest Bransted 102 Chapter 4 Table 4.6 A Properties of the supports Support SBET m 2 /g Pore size dp, nm Pore volume ml/g n -PA uptake pmol/g A 1 2 0 3 155 19.6 0.76 158 A 1 2 0 3 - F 146 18.6 0.68 244 A I 2 O 3 - W A 164 18.0 0.74 180 M C M 455 8.3 0.95 557 Table 4.6B Surface areas, C O uptake and n-PA uptakes of prepared Coo.4Ni 2P on different supports Catalyst Chemisorption B E T area C O uptake n -PA uptake C00.4M2P/AI2O3 C00.4N12P/AI2O3-F C o 0 . 4 N i 2 P / M C M . m 2 /g 98 95 125 pmol/g 25 24 20 pmol/g 304 (3.1) 390 (4.1) 715 (5.7) MCM: Mobil Catalytic Materials support A1 20 3-F: Fluorinated alumina support Values in brackets are n-PA uptakes normalized by the BET area (pmol/m ) 103 Chapter 4 acidity of 715 umol/g compared to the catalysts supported on A I 2 O 3 (304 umol/g) and AI2O3-F (390 umol/g). X 4.4.4 Properties of Pt-Coo.4Ni2P on different supports Pt was added to Coo.4Ni2P/Al2C»3 to study the effect of this metal on the hydrogenation route of the 4 , 6 - D M D B T (see Figure 2.6). A weakly acidic support was used to prepare Pt-Coo.4Ni2P/Al20 3 -WA catalysts in order to study the role of a weakly acidic support compared to the AI2O3 neutral support. The characterization data for these prepared catalysts are presented in Table 4.7. The B E T surface area of the Pt-Coo.4Ni2P/Al 2 0 3 catalyst was 89 m 2 /g and the Pt-Coo.4Ni2P/Al 2 03-WA was 95 m 2 /g. Compared to the C00.4N12P/AI2O3 catalyst, B E T surface area 98 m 2 /g (Table 4.6B) the addition of Pt showed a modest change in surface area and could be attributed to the addition of the 5 wt% Pt that led to agglomeration of particles and blocking pores and therefore decreased the surface area. The platinated supported metals, Pt-Coo.4Ni 2 P/Al 2 03 and Pt -Coo.4Ni 2 P/Al20 3 -WA both exhibited high C O uptake (31 and 30 umol/g respectively). The presence of Pt probably caused the C O uptake in 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-PA uptakes of prepared Pt-Coo.4Ni2P on different supports Chemisorption Catalyst B E T area C O uptake n - P A uptake m /g pmol/g pmol/g P t -Coo .4Ni 2 P/Al 2 0 3 89 31 240(2.7) P t - C o o . 4 N i 2 P / A l 2 0 3 - W A 95 30 380(4.0) A1 20 3-WA: Weakly acidic alumina support Pt: Platinum Values in brackets are n-PA uptakes normalized by the BET area (pmol/m ) 4.5 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 2 at 723 K for 1 h in order to remove the passivated layer. In the present study, 4 , 6 - D M D B T was added to dodecane solvent to provide 3000 ppm of sulfur feed. The H D S products of 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 of the hydrogenation route and D M B P is the product of the D D S as shown earlier in 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 of 4 , 6 - D M D B T over C0M0P/AI2O3 phosphide catalyst. Table 4.8 shows the activity data of the hydrodesulfurization of 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 Coo . 4 Ni2P/MCM. Mizutani et al. (2005) using a fixed bed reactor operated at 603 K , 3 M P a and W H S V of 13.5 h ' 1 obtained high conversion of 93 mol% over C0M0P/AI2O3. The conversion obtained in the present study for the. C00.4N12P/AI2O3 was 86 mo l% and this is therefore comparable considering the fact that lower temperature (583 K ) was used in the present study. Kabe et al. (1999) have reported the use of first order kinetics for the H D S of heterocyclic sulfur compounds hence first order kinetics w i l l be used. The specific consumption of the 4,6 D M D B T obtained varied from 1.10 - 4.07 x 10"4 mol.g ' 1 . s"1 ( 3.95 - 14.6 mol.g" 1. h"1). The highest consumption was obtained using Coo.4Ni2P/MCM supported catalyst and the least on the Ni2P/Al203. Nagai et al. (2005) reported a specific rate of 0.19 mol. g"1. h"1 (5.28 x 10 -5 mol . g"1. s"1) o f dibenzothiophene over N i M o P / A l 2 0 3 at 573 K , 2 M P a , 8 h"1 W H S V and 12 h time on stream. 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 of the 8 h stabilization time. The product selectivity data of Table 4.8 show that the Ni2P/Al2C»3 produced more products by the hydrogenation route (89% of both D M B C H and M C H T ) and fewer products by the D D S route (2.8 mo l% of the D M B P ) . On the other hand, the Coo . 4Ni 2 P/Al 2 0 3 showed a low selectivity for the hydrogenation route (30 mol% of 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 of supported metal phosphides for the H D S of 4 , 6 - D M D B T measured at 583 K and 3.0 M P a H 2 Selectivityt Catalyst Conversion H D S Specific Consumption M C H T D M B C H D M B P mo l% mol% 10 8 mol. mol % M2P/AI2O3 Coo.4Ni 2 P/Al 2 03 81.3 86.5 80.1 85.8 1.10 1.31 36.1 2.1 52.8 28.0 2.8 52.4 Coo .4Ni 2 P/Al 2 0 3 -F 87.1 86.4 1.36 5.9 16.6 45.2 C o o . 4 N i 2 P / M C M 99.8 98.8 4.07 4.0 12.7 1.9 Pt-Coo.4Ni2P/Al203 97.2 97.1 2.41 0.0 45.1 49.5 Pt-Co 0 .4Ni2P/Al 2 O3-WA 90.0 89.5 1.51 6.7 29.9 48.5 1 order kinetics: Specific consumption of 4 , 6 - D M D B T based on the inlet = k C A 0 and k is specific rate constant. f Selectivity based on average of 4 h after 8 h stabilization period. production of more products to the D D S route than the N12P. Table 4.8 also shows the product selectivity obtained using Coo.4Ni 2P supported on fluorinated AI2O3. Fluorine was incorporated on the support to increase the Bronsted acidity in order to induce isomerisation of 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 in 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 of the catalytic activity (compared to sulfided N i M o and C o M o on non-fluorinated A1203) i n addition to increased D D S products from 4 , 6 - D M D B T . Another interesting selectivity is shown by the Ni2P/MCM in 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 of 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 of enhancing the hydrogenation route of 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/Al203) with little cracked products on this catalyst. Since the Pt-Coo.4Ni2P/Al203 produced high conversion with less cracked products, the possibility of further enhancing the conversion by using a weakly acidic support to increase isomerisation was explored. The product selectivity using this catalyst, Pt-Coo.4Ni2P/Al203-WA 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 mol%. To summarize the effects of supports on the H D S of 4 , 6 - D M D B T , it is noted that AI2O3 is still the support of choice as promoted C00.4M2P phosphide can be impregnated on it to give high Bronsted acidity. In the present study, the effect of 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 of 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 of H D S of D B T in the range -0.4 y = -7.7942x + 12.759 -0.8 A c -1.2 -1.6 -2 1.7 1.74 1.78 1.82 1.86 1.9 1/171000 K -1 Figure 4.12 Plot of In k versus 1000/T for the hydrodesulfurization of 4 , 6 - D M D B T using P t - C o o ^ P / A b C ^ . 109 Chapter 4 of 87.9-94.9 kJ/mol. The apparent activation energy Ea reported in the present study includes the heat of adsorption, X (Satterfield, 1991) hence, Es accounts for the surface reaction. 4.6 Kinetics of the HDS Based on the high conversion with few cracking products, the Pt-Coo.4Ni 2P/Al203 catalyst 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. where x A is the conversion of reactant 4 , 6 - D M D B T , W is the weight of the catalyst, F A o is the molar feed rate of the 4 , 6 - D M D B T and r A is the rate of reaction. The H D S kinetics of 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 A in equation 4.2 yields equation 4.3 E a = Es-X, (4.1) (4.2) dxA = -rA=kCA=kCM(l-xA) (4.3) d { W \ K1 AoJ 110 Chapter 4 where k is the kinetic rate constant, C A 0 and C A are the initial and the final concentrations of the feed 4 , 6 - D M D B T . Equation 4.3 is solved to yield equation 4.4. -\n(l-xA)=kCAo/SV (4.4) where S V is the space velocity. Hence a plot of 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 A ) obtained from data of Pt-Coo.4Ni 2 P/Al 2 03 at 533 K and 3.0 M P a . Figure 4.13 gives an excellent fit of the measured data to equation 4.1 with a correlation coefficient of R 2 = 0.999 and with the gradient of 0.2158. Substituting the slope in equation 4.1 and solving gives the value, 2.13 x 10 4 ml/gcat. h as the pseudo first order rate constant obtained on the Pt-Coo.4Ni 2 P/Al 2 03 catalyst. Haji et al. (2005) reported a pseudo first order kinetic rate constant k = 3.3 x 10 4 ml/g cat.h for 5.7 wt % P t / A l 2 0 3 obtained using D B T (containing 887 4 -r 3.5 -3 -^ 2.5 -< 1 2-c 1.5 -1 -0.5 -0 -• 0 1/space velocity, h-gcat/mol Figure 4.13 Plot of In (1-XA) versus space time at 533 K , 3 M P a using P t - C o o . 4 N i 2 P / A l 2 0 3 y = 0.2189x + 0.3676 R 2 = 0.9995 2 4 6 8 10 12 111 Chapter 4 wppm S) at 583 K and 1 atm in 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 in 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. D M B C H (D) Figure 4.14 Simplified reaction scheme for the H D S of 4 , 6 - D M D B T 112 Chapter 4 The kinetic equations used in the simulation are shown in eqns. (4.5) to (4.8). dX. d = rA=-(kl+k2)CA \ F A o J (4.5) d dX» - , -kC -kC \ F A o J (4.6) dX d W \ F A o J r =k2CA+kiC-kC c 2 A 3 B 4 c (4.7) dX„ d W \ F A o J (4.8) The Gaussian Newton-Raphson optimization method (Appendix E) was used to estimate the rate constants assuming gas phase reaction and first order kinetics in the fixed bed reactor, using the ordinary differential equations 4.5 to 4.8 and the estimated values are summarized in Table 4.9. The tolerance for calculating the rate constants is 10"6. The model is sensitive to the initial values and the step size. 113 Chapter 4 Table 4.9 Estimated I s order rate constants for hydrodesulfurisation of 4 , 6 - D M D B T at 533 K Value x 10"J Rate constant (ml/gcat.h) k i 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 of the hydrodesulfurization of 4 , 6 - D M D B T simplified network for the P t - C o 0 . 4 N i 2 P / A l 2 O 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 2 respectively. The magnitude of the ratio of k]/k 2 is 21. This high ratio implies that D D S on the P t -Coo .4Ni 2 P/Al 2 0 3 is faster than the hydrogenation. It therefore explains the higher yields of the D M B P compared to the M C H T . The magnitude of L i is also high explaining the high yields of 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 of 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 of platinum on the catalyst could explain this phenomenon of increased hydrogenation. Figure 4.15 shows the correlation from the experimental data and that obtained from the model. The correlation of R 2 = 0.9813 is very high and therefore indicates agreement between 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 of the experimental data versus the predicted from the model for conversion of D M D B T (0) and yields of products ( D M B P : * ; M C H T : A ; D M B C H : • 4.7 Discussion on HDS of supported metal phosphides The supports in this study were chosen in order to enhance Bransted acidity to achieve high conversion by creating the desired isomerisation of the methyl substituents. Figure 4.16 shows the plots of Bransted acidity as a function of both conversion and D M B P selectivity. Except for the Pt-Coo.4Ni 2P/Ai203, all the prepared phosphides show a trend of increasing 115 Chapter 4 Figure 4.16 Plots of Bronsted acidity as a function of both conversion and D M B P selectivity on all supported catalysts: Ni2P/Al203 (•), Coo.4Ni2P/Al2C»3 (A) ,Coo.4Ni 2 P/Al 2 03-F ( • ) , C o 0 . 4 N i 2 P / M C M (•) Pt- C00.4N12P/AI2O3 (•), Pt- C o 0 . 4 N i 2 P / A l 2 O 3 - W A ( A ) 116 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 -PA titration of 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 2 = 0.9993. This is the expected outcome as the presence of the Co in N12P catalysts provides the necessary isomerization of 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 of 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 in 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 in addition to the high conversion obtained on the Pt-Coo.4Ni 2P/Al 203 catalyst, the selectivity to hydrogenation products (45.1 mol%) were higher than the non-platinated C00.4N12P/AI2O3 (30.1 mol%). The presence of 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.4Ni2P/Al2C>3 (49.5 mol%) was marginally smaller than that obtained from the non-platinated Coo.4Ni2P/Al203 (52.4 mol%). The presence of Pt on the Coo.4Ni2P/Al203 probably covered part of the P thereby reducing the 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 Ni resulting in agglomeration which could affect the surface area of the catalyst. The BET surface area of this catalyst was the least, 89 m2/g. Therefore, addition of Pt is beneficial to the catalyst as it results in increased conversion of 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 XRD, TPR and XPS. The XRD 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 4Ni 2P/Al 20 3 * 1 : • 1 . 1 . , . , , , ro 20 30 40 50 60 70 29 Figure 4.17 Comparison of the X R D profiles of 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 of the used M o P and Coo .os^P catalysts, shown in Figure 4.18 A , provided evidence for the presence of surface phosphide with a peak at low B E (129-130 eV). On the used N i 2 P and Cooo 7 MoP, however, the P was present as a surface phosphate (BE = 133.5-134.2 eV), suggesting that.the metal phosphide was more stable on the M o P and Coo .os^P than the 119 Chapter 4 Ni 2 P C o M o P C o N L P Figure 4.18 Comparison of the X P S of 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 of spent metal phophides showed that the prepared metal phosphides used in the present study are stable in the presence of small H2S produced during the H D S of 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 of 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 of the catalyst could be burnt off in air at temperatures 400-600°C. Another method of catalyst regeneration is by treating the organic solvents at temperatures 140 °C to extract organic deposits that plug the pores of 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 of the organo nitrogen compounds present in petroleum feedstock. The chapter w i l l first present the preparation and characterization of bulk M o P and N i x M o P (0.07 < x < 1.11) catalysts. The first section w i l l discuss the different methods used to characterize the bulk and surface properties of the prepared catalysts. The results of activity studies using the bulk metal phosphides w i l l be presented. Finally, the characterization and activity data of supported metal phosphides w i l l be discussed. Recent results on a series of CoxNi2P and C o x M o P catalysts, in which the P content, catalyst acidity and hence selectivity during H D S of 4,6-dimethyldibenzothiophene, were shown to be dependent upon the Co content (see Chapter 4), the present study was aimed at determining i f a similar effect was observed with N i x M o P catalysts for the H D N of carbazole. Carbazole is a non-basic refractory heterocycle used here for the first time to examine the activity of metal phosphide catalysts. Note: A version of this chapter is in the press, Catal. Today, 2007 122 Chapter 5 5.1 Bulk Ni x MoP catalysts The preparation of a series of N i x M o P (0.0 < x < 1.11) catalysts for the H D N of carbazole was done following a similar procedure to that reported earlier (Chapter 4) for the CoxNi2P bulk 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 in Chapter 4. The characterization was performed to determine the properties of 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 in a 60 ml/min flow of 10 % H2 in A r using 0.2-0.4 g of the calcined sample loaded into a stainless steel reactor (i.d. = 9 mm). The T P R profiles of the bulk N i x M o P ( 0 . 0 < x < l . l l ) precursors are shown in Figure 5.1. The data show a clear trend of 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 2 P precursor (Abu and Smith, 2006) but with the peak temperatures occurring at about 50 K higher than for the M2P precursor. The decreased reduction temperature with increased N i content is indicative of a significant interaction between the N i and the M o P . 123 Chapter 5 Figure 5.1 T P R of calcined catalyst precursors ( N i x M o P for 0.0 < x < 0.11) measured in 10% H2 in A r at flowrate of 60 ml(STP)/min and 1 K/min 5.2.2 X R D The X R D diffractograms of the reduced catalysts, presented in 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 x M o P , the formation of a N i M o P phase was clearly evident. For the N i i . n M o P catalyst, the N i M o P phase was dominant. The X R D data were used to estimate the crystallite size (d c) and the lattice parameters of the M o P phase, and the results are shown in 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 x M o P for 0.0 < x < 1.11 size ranged from 1 6 - 2 2 nm for 0 < x < 1.11 and as the N i content of the N i x M o P catalyst 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 nm and c = 0.3191 (JCPDS, 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 of reduced catalysts. M o P phase Catalyst N i content Phases 20, degree Lattice parameters, nm Crystallite size, nm mole % (100) (101) (110) (111) a c d c (101) M o P 0 M o P 31.93 42.89 57.16 64.93 0.3240 0.3196 16 Nio.o7MoP 3 M o P 32.94 42.90 57.02 64.72 0.3255 0.3185 18 N i 0 . i 6 M o P 7 M o P , N i M o P 31.98 42.96 57.08 64.68 0.3254 0.3182 21 NiojgMoP 15 M o P , N i M o P 31.90 42.92 57.06 64.68 0.3256 0.3184 22 N i , . i , M o P 34 M o P , N i M o P 32.04 43.16 57.12 64.72 0.3245 0.3170 22 126 Chapter 5 Ni(2-X)Mo xP catalysts and showed the formation of solid solutions for a range of N i / M o ratios > 1. In the present study, however, the (Ni + Mo) content of the N i x M o P catalysts increased as the N i content increased and the X R D data show that with the preparation method and catalyst composition used herein, a catalyst with a mixture of 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 -PA uptake of the bulk N i x M o P catalysts. The B E T surface area varied marginally from 5.3 - 6.0 m 2 /g for the N i x M o P ( 0 < x < l . l l ) catalysts and there is a modest increase in the surface area upon addition of N i to M o P . The C O uptake and the n -PA uptake reported per unit B E T area of the catalysts, increased significantly with addition of N i to the M o P . Furthermore, as the N i content increased, the C O uptake increased, whereas the n -PA uptake showed a maximum for the N i x M o P catalysts. Also note that the Nio.o7MoP catalyst had the lowest C O uptake and the highest n -PA uptake. 5.2.3 XPS X P S was used to characterize the surfaces of the prepared catalysts. Figure 5.3 shows the N i 2p, M o 3d and the P 2p X P S spectra of 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 2p3/2 region showed peaks at 133.8 eV and another peak at 129.8 eV for al l 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 eV is attributed to surface PO4 " species (Harris and Chianelli , 1986; Okamoto et al., 1980). The lower B E peak at 129.8 eV is attributed to metal phosphides. The B E of the N i 2p is consistent with the presence o f metal phosphides and phosphates, corresponding to B E s o f 853.7 eV and 857.3 eV, respectively. Similarly, M o 3d B E at 228.2 eV and 232.2 eV indicate the formation 127 Chapter 5 Table 5.2 Physiochemical properties of prepared metal phosphide Catalyst B E T area Chemisorption P / M * C O uptake n-PA uptake Nominal X P S m 2 /g umol/m 2 umol/m 2 atom ratio atom ratio M o P 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 N i i . n M o P 5.9 2.05 32 0.60 0.84 * M : total metals (Ni+Mo) and P: phosphorous of phosphide and phosphate species, respectively. The atom percentages of the prepared phosphides were also determined by X P S and the P / M atom ratios (where M is (N i + Mo)) are reported in Table 5.2. The P / M atom ratios show that the Nio.o7MoP enriched in phosphorous (P/(Ni+Mo) ratio = 1.03) and increasing N i content resulted in a decreased P / M ratio at the surface of the catalysts. Figure 5.4 shows a correlation between the X P S and chemisorption data obtained on the metal phosphides of the present study. Assuming that C O chemisorption titrates metal sites and 128 Chapter 5 Mo 3d 245 235 225 Binding Energy, eV 215 Ni 2p 857.3 eV 853.7 eV 870 860 850 Binding Energy, eV Ni, nMoP Nio.38MoP "Ni 0 1 6MoP Mi007MoP 840 830 P 2P 133.8 eV 129.8 eV Jj V \ ^ NimMoP V Ni 0 3 8MoP Ni 0 1 6MoP 1 1 1 1 1 Ni 0 0 7MoP 145 140 135 130 125 120 115 Binding Energy, eV Figure 5.3 X P S spectra of the N i 2P, M o 3d and P 2p of the prepared phosphides 129 Chapter 5 1.05 £ 1.00 >< •B 0.95 as 0.90 0.85 0.80 y = 0.0056x + 0.786 R 2 = 0 . 9 0 6 7 / m • 10 20 30 40 50 n-PA/CO uptake ratio by adsorption Figure 5.4 Correlation of P / M ratio determined by X P S and n - P A : C O uptake ratio determined by adsorption for N i x M o P for 0.0 < x . < 1.11(B) and M o P ( A ) . the n -PA 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 in Figure 5.4 show a good correlation (R 2 = 0.906) and are consistent with a similar correlation determined for the series of CoxNi2P catalysts reported in Chapter 4. 130 Chapter 5 5.3 Catalyst Activity of bulk Ni x MoP 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 of 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 of 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 x M o P catalysts (0.07 < x < 1.11). The carbazole consumption rates in terms of 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 of carbazole consumption ranged from 1.21 - 3.54 x 10"8 mol.g"1.s"1 with the NI0.07M0P showing the highest specific rate of carbazole consumption and the M o P showing the lowest. The areal consumption of 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 9 1 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 of bulk metal phosphides for the H D N of carbazole measured at 583 K and 3.0 M P a H 2 Carbazole Consumption Rate Selectivity Catalyst Total Conversion Specific Areal T O F B C H X C P C H X T H C Z Other products B T A T mol % 10 s mol.g"1.s"1 10* mol.m^.s"1 10-V mol % -M o P 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 0 . i 6MoP 89.6 3.48 5.89 0.5 77.8 0.7 17.0 0.5 0.4,, Nio.3 8MoP 88.8 3.36 5.67 0.4 51.2 2.1 10.0 27.8 6.7 N i i . n M o P 85.8 3.00 5.07 0.2 40.4 3.3 17.5 34.9 2.4 B T : Other products formed with carbon number < 12 A T : Other products formed with carbon number > 12 132 Chapter 5 consumption of the Nio.o7MoP catalyst and this result is in 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 of reaction of carbazole for all the catalysts were bicyclohexane ( B C H X ) and tetrahydrocarbazole (THCZ) . 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 of Table 5.3 show that the selectivity to B C H X of the N i x M o P catalyst was significantly greater than that obtained over M o P . Furthermore, for the N i x M o P catalysts, * the B C H X 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-PA uptake are plotted as a function of the C O uptake in Figure 5.5. A s noted above, the data show a significant increase in 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 in N i x M o P ) the n -PA uptake and B C H X selectivity decreased, with a small reduction in carbazole conversion. 133 Chapter 5 5.4 Discussion on conversion of carbazole using bulk phosphides The data of Table 5.3 show that the activity of the M o P catalyst (2.3 x 10"9 mol 1 0 0 - , o 8 0 -a> CO 6 0 -X O 4 0 -CQ 2 0 - • 1 0 0 -^ 9 0 - " " — Conversion D O O O • C M £ 4 0 -o 1. 3 0 -to ro 2 0 -Q_ < 1 0 -Q. i o - J • ^ ^ ^ ^ • i 0 . 0 1 1 0 . 4 , . , i | i 0 . 8 1.2 1.6 i 2 . 0 1 2 . C O u p t a k e , u m o l / m 2 Figure 5.5 Conversion of carbazole and selectivity to B C H X over various metal phosphide catalysts at 583 K and 3.0 M P a H 2 plotted as a function of C O uptake rati of N i x M o P (•) 0.0< x < 1.11 and M o P ( A ) catalysts m" 2 s"1) was slightly lower than the N i x M o P catalysts and the highest areal activity was obtained on the N 1 0 . 0 7 M 0 P catalyst with a value of 5.9 x 10"9 mol m" 2 s"1 at 583 K and 3.0 M P a H 2 . Areal 134 Chapter 5 rates of 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 of 312 x 10" mol m" s" and 55 x 10" mol m" s" were reported on bulk M o P and N i M o P , respectively. These activities are at least an order of 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 CoP (1.2 x 10"3 s"1) and supported M2P (0.60 x 10"3 s"1) at 643 K and 3.1 M P a in the presence of dibenzothiophene. The relatively low activity of quinoline at this temperature was attributed in part to the inhibiting effects of the sulfur compound. Carbazole is a non-basic, five-membered N-heterocycle in which the extra pair of electrons of 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 low reactivity that makes it difficult to remove the N heteroatom and the data reported herein are consistent with this observation. The data of Table 5.3 also show that the bulk M o P had the highest T O F among the catalysts tested. On bulk M o P , a T O F of 11.1 x 10"3 s"1 was obtained after an initial 8 hours of 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 of 29 x 10"3 s"1 over nitrided 12.5 wt% M0/AI2O3, measured in a fixed bed microreactor at 573 K and 10.1 M P a . However, the catalyst T O F was determined after 30 minutes time-on-stream and the activity decreased by a factor of 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 of 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 CoxNi2P catalysts, in Chapter 4 and (Abu and Smith, 2006). The source o f the Bransted acidity on the bulk phosphide is likely a consequence of the 135 Chapter 5 pretreatment of the catalysts before reaction with H2 at 723 K , that leads to formation of water with the most reactive passivated oxygen. The water may react with surface phosphate to produce the H x P 0 4 ( x 3 ) species responsible for the Bronsted acidity. The N i x M o P catalysts all had higher metal and acid site densities than the M o P catalysts, possibly because o f a higher dispersion of 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 x M o P catalysts. The data of Figure 5.5 show that the acid site density decrease associated with the N i content o f N i x M o P is relatively small when compared to 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 of the N i x M o P catalysts increased, the selectivity to B C H X decreased in favor of lighter products. In all cases, however, the M o P had lower B C H X selectivity than the N i x M o P catalysts. A significant increase in conversion and B C H X selectivity is observed when comparing M o P to the N i x M o P catalysts. However, for the series of N i x M o P catalysts, the conversion is relatively constant, as C O uptake increases, whereas selectivity to B C H X decreases. Indeed the trend in B C H X selectivity follows the trend in n -PA site density as C O uptake increases. Figure 2.3 (Chapter 2) shows the proposed reaction scheme for the H D N of carbazole on nitrided M o / A l 2 0 3 catalyst (Satterfield and Cocchetto, 1981). The authors attributed the B C H X formation to the C - N hydrogenolysis of the perhydrocarbazole and the cyclohexylbenzene and cyclohexylhexene to C - N hydrogenolysis of 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 of 136 Chapter 5 catalysts towards unsaturated hydrocarbons is limited by the hydrogenation ability of the catalyst. The H D N of heterocyclic compounds proceeds by hydrogenation of the aromatic ring and then subsequently by hydrogenolysis of 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 . Addit ion of 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 of N i , the hydrogenation activity increased significantly. Wada et al . (1996) reported that on introduction of the noble metals to Ni/Y-zeoli te, D B T was completely converted as compared to the 88% conversion obtained using Ni/Y-zeoli te without noble metals. The authors attributed the increased conversion to the hydrogenation capability of the noble metal present in Ni/Y-zeoli te . A n increased hydrogenation capability of 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 of the formation of more C(sp ) -N bonds that are readily cleaved. The data of Figure 5.5 also suggest that the bond cleavage is related to the acid sites of the catalysts since for the N i x M o P series, as the metal site density increases, the acid site density decreases as does the selectivity to B C H X . The products from carbazole H D N included small amounts of C P C H X , likely produced as a result of ring opening or isomerization due to Bransted acidity. After hydrogenation of 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 of 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 x M o P phosphides had a lower T O F for the H D N of carbazole compared with M o P . However, the selectivity to bicyclohexane was greater on the N i x M o P 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 x M o P series. The improved selectivity is attributed to enhanced hydrogenation in the presence of the N i x M o P catalyst that was a mixture of M o P and N i M o P phases, and had higher C O uptake than the M o P . The C - N bond cleavage is attributed to the acidity of 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 2 /g, PV=0.76 ml/g, pellet size 4.5mm), fluorinated AI2O3 and M C M . A s already mentioned in chapter 4 of 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 of the active phase (Anderson et al., 1982). Fluorine in the form of NH4F impregnated on AI2O3 provides increased acidity, better dispersion and higher hydrogen chemisorption that enhance H D N compared with non-fluorinated AI2O3 (van Veen et al., 1993; Qui 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 of 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 of M0.33M0P catalysts, supported on supports of 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 of the reduced and passivated Nio.33MoP/Al 203 catalyst taken by S E M . Note that interpretation of this S E M data is difficult because the detailed structure of the metals (Ni 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 2 0 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 203, and AI2O3. 700 -1 600 -500 -cd —^» 400 -C ii 300 -200 -100 -* Formation of M o P Figure 5.7 Comparison of X R D of Nio .33MoP/Al 2 0 3 and A 1 2 0 3 The formation of M o P is indicated on the Nio.33MoP/Ai203 profile. The peak at angle 42.9 0 and plane (101) corresponds to the highest intensity peak of M o P . The other peak occurs at 32.6 0 and corresponds to the (100) plane of M o P . N i was absent in the X R D profile of the Nio.3 3 MoP/Al203 and this observation is consistent with earlier observations with Con.o8Ni2P 140 Chapter 5 (Chapter 4) where small quantities of Co 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 0 peak could not be identified, it could be the result of 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 of Nio.33MoP/Al 203 and Nin .33MoP/MCM were determined using B E T , C O uptake and n-PA chemisorption and the data are presented in Table 5.4 . Table 5.4 Properties of prepared Nin.33MoP/Ai203 and Nio.33MoP/MCM catalysts Chemisorption Catalyst B E T area C O uptake n -PA uptake 2 2 m /g umol/g umol/g umol/m N10.33M0P/AI2O3 65 9.10 319 4.9 M 0 . 3 3 M 0 P / M C M 132 7.29 779 5.9 The other supported catalysts used in the carbazole H D N study are Con.4Ni2P/Ai203, Coo.4Ni2P/Al 20 3 -F and Coo.4Ni 2 P/MCM. These supported catalysts were already tested for H D S of 4 , 6 - D M D B T and their properties were reported earlier. The intention of 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 -PA 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 of the Bronsted acidity per m of M C M is still greater than the AI2O3 supported catalysts. However it is still valid to compare the total Bronsted acidity of 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 C0M0/MCM (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 in the case of the bulk activity study, catalyst activity measurements were carried out over the Nio .3 3 MoP on different supports, using the same fixed bed reactor operated at 523 - 583 K temperature and 3.1 M P a H 2 pressure. The Coo.4Ni 2P/Al203 > Coo.4Ni2P/Al 2 0 3 -F and Coo.4Ni2P/MCM catalysts previously prepared in Chapter 4 for testing the H D S of 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 2 at 723 K for 1 h, in order to remove the passivated layer. In this study, carbazole was added to xylene solvent to provide 3000 ppm of N in the feed. The H D N products of carbazole using these supported metal phosphides were similar to those obtained previously on the bulk metal phosphides. Table 5.5 shows the activity data of the H D N of carbazole using the supported metal phosphide catalysts. The conversion ranged from 82-99.9 mo l% with the lowest conversion obtained on the Coo.4Ni 2 P/Al20 3 and the highest on the M0.33M0P/MCM. 142 Chapter 5 Turaga et al. (2003) also reported a high conversion of carbazole (98 wt%) using C o M o / M C M and about 78 wt% for C0M0/AI2O3. The authors worked with 500 ppmw of nitrogen in carbazole at 573 K , 45 atm and space velocity of 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 of cracked products. The Ni 0.33MoP/Al2O3 showed a conversion of 96 mo l% which 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"8 mol.g"1.s"1. The specific consumption of the Nio.33MoP/Ai203 was 4.94 x 10' 8 mol.g"1.s"1 and Table 5.5 revealed that this catalyst also 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 in the present study, probably due to their high reactivity, in 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 of carbazole. Both the N i 0 . 3 3 M o P / M C M and the Coo.4Ni2P/MCM produced small-hydrogenated B C H X because the B C H X is either almost or completely cracked to products with products of carbon number less than 12. The high amounts of cracked products produced by the M C M supports were reported earlier in 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 of supported metal phosphides for the H D N of carbazole measured at 583 K and 3.0 M P a H 2 Catalyst Total Conversion Consumption Rate Product selectivity B C H X C P C H X T H C Z Other B T A T mol % 10 8mol.g-'.s-' mol % Coo.4Ni2P/Al203 82.0 2.63 72.5 5.1 9.6 8.1 2.0 Coo.4Ni2P/Al203-F 94.0 4.32 80.5 7.7 7.8 0.1 1.4 Coo. 4 Ni 2 P/MCM 98.0 6.01 9.4 9.8 2.0 41.3 35.0 Ni0.33MoP/Al2O3 96.0 4.94 82.5 4.7 3.0 6.3 1.7 Nio .3 3 MoP/MCM 99.9 21.21 13.2 1.1 12.4 43.4 20 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.33MoP/Al203 was selected based on its performance and 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 of the space 0 2 4 6 8 10 12 1/space velocity.h-gcat/mol Figure 5.8 Plot of 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 of the space velocity) versus l n ( l - X A ) obtained from data over N I 0 . 3 3 M 0 P / A I 2 O 3 and at 533 K and 3.0 M P a , X A is the conversion of carbazole. The correlation gives a good fit o f R 2 = 0.9988 and the slope is 0.1227. Hence from equation 4.3, the kinetic rate constant is 4.194 x 10 ml/hgcat. Based on the reaction network and the products identified as shown in Figure 5.9 (which is a simplified reaction network of Figure 2.3) the kinetic rate constants were estimated. The 145 Chapter 5 Figure 5.9 Simplified reaction network of carbazole 1 4 6 Chapter 5 Gaussian Newton-Raphson optimization method was used (as stated earlier in Chapter 4) to estimate the individual rate constants. Table 5.6 summarizes the results of the different kinetic rate constants. The kinetic equations used in the simulation are shown in equations 5.1 to 5.3. dX, = r=-k.C. (5.1) dX d jL- = r=kC.-kC K FAoJ (5.2) dX fw_\ \ F A o J = r =kC. (5.3) 147 Chapter 5 Table 5.6 Estimated 1 s t order rate constants for the hydrodenitrogenation of carbazole at 583 K Value x 10 J Rate constant (ml/gcat.h) k, 4.4000 ± 0 . 0 1 2 9 k 2 0.2570 ± 0.0075 The rate constant k i represents hydrogenation of carbazole to T H C Z and k 2 the C - N hydrogenolysis of T H C Z and further hydrogenation of the product to form B C H X . The value of k i / k 2 is about 18 and indicates that hydrogenation of the carbazole is much faster than the C - N hydrogenolysis. Jian et al. (1996) studied the H D N of ortho-propylaniline over phosphorous promoted N i M o / A l 2 0 3 at 623 K and 3.0 M P a and reported that there are two kinds of catalytic sites: the N i - M o - S site (associated with Ni ) and the M o site (associated with M o ) . The authors explained that the former was responsible for hydrogenation of the phenyl group in 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 of N i into M o P enhanced hydrogenation and subsequent C - N bond cleavage. Direct denitrogenation through hydrogenolysis of carbazole was not reported contrary to what was observed earlier in Chapter 4 with the H D S of 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 of 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 of the concentration of the products and carbazole with the predicted values. The R = 0.9839 indicates good correlation and hence model fits well . The selected NinjaMoP/AbOs 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 of the experimental data versus the predicted from the model for conversion of carbazole (0) and yields of products (BCHX:« ; T H C Z : A ) 149 Chapter 5 (2003) studied the H D N of 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 of direct hydrogenolysis products. Figure 5.12 shows plots of Bransted acidity as a function of 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 of the H D S of 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 in the present study that M C M supported catalysts produced cracked products due to the high Bransted acidity. 150 Chapter 5 -1 • G -4 H -2 1 -3 R 2 = 0.9857 -5 1.5 1.7 1.9 2.1 2.3 2.5 1/T, M O O O K " 1 Figure 5.11 Plots of In k versus 1000/T for the hydrodenitrogenation of carbazole over Nio .33MoP/Al 2 0 3 In summary, the effect of adding M C M to the selected bulk N i o . 3 3 M o P leads to enhanced Bronsted acidity and subsequent increased conversion of carbazole. However, in terms of 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 of 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 Summary 151 Chapter 5 Bronsted acidity, pmol/m 2 Figure 5.12 Plots of Bronsted acidity as a function of both conversion and D M B P selectivity on all supported catalysts:Coo.4Ni2P/Al203 ( A ) , Coo.4Ni2P/Ai203-F ( • ) , C o 0 . 4 N i 2 P / M C M (•), N i 0 . 3 3 M o P / A l 2 O 3 (•) N i 0 . 3 3 M o P / M C M (A) 152 Chapter 5 selectivity to D M B P that is the direct desulfurisation product of 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 of the catalysts. Nio.33MoP/Al203 has very high conversion of carbazole (96 mol%) and also produces less cracked products, hence based on these observations, it is selected to be further tested using real feedstock in 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 4,6-D 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 of 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 -Coo .4Ni 2 P/Al 2 03, Coo .4Ni 2 P/Al 2 03, and N i 0 . 3 3 M o P / A l 2 O 3 using L G O derived from Athabasca bitumen was to compare the activity with a commercial sulfided catalyst and with the activity data obtained using model compounds. The A l 2 0 3 (SA = 240 m 2 /g , P V = 0.65 ml/g) used was in the form of 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 of inner diameter 10 mm and length 285 mm. Five grams of the catalyst was diluted with 75 v o l % o f 90 mesh silicon carbide particles to form a bed of 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. Within 13 cm of the reactor length, the temperature profile showed a uniform temperature with ± 2 °C variation, thus ensuring that the 12 cm length of the catalyst bed was isothermally stable. 154 Chapter 6 Table 6.1 Characteristics of Light Gas O i l derived from Athabasca bitumen Boi l ing 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 in straight run gas o i l . The flow rate of the sulfiding solution was 5 ml/h and the H 2 /sulfiding solution ratio was 600 (v/v). Prior to commencing the experiments, catalysts were pretreated with the L G O for a period of five days at 375 °C, 8.8 M P a , 1 h"1 L H S V and H 2 / o i l ratio of 600 ml/ml . The conditions for testing were 375 °C, 8.8 M P H 2 , and 2 h"1 L H S V . After three days of a 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 of the 24 h single point sample. In addition to the selected metal phosphide catalysts, a commercial sulfided NiMo/Al 2 03 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 of the catalysts tested using L G O as feed. Both figures show a similar trend of increasing conversion with increasing reaction temperature. The sulfur conversion obtained over the temperature range studied (613 - 647 K ) for the C o 0 . 4 N i 2 P / A l 2 O 3 , P t - C o 0 . 4 N i 2 P / A l 2 O 3 and N i 0 3 3 M o P / A l 2 O 3 catalysts was 73.3 wt%, 76.1 155 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 2 0 3 catalyst also showed a c o > C o U 610 620 630 640 Temperature, K 650 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" 1, H 2 to o i l ratio = 600 ml/ml: • C o 0 . 4 N i 2 P / A l 2 O 3 • P t - C o 0 . 4 N i 2 P / A l 2 O 3 A s u l f i d e d N i M o / A l 2 0 3 a N i 0 3 3 M o P / A l 2 O 3 156 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" 1, H 2 to o i l ratio = 600 ml/ml: • Coo.4Ni 2 P/Al 2 03 • P t - C o o . 4 N i 2 P / A l 2 0 3 • S u l f i d e d N i M o / A l 2 0 3 a N i 0 3 3 M o P / A l 2 O 3 maximum S conversion of 94.9 wt% and a maximum N conversion of 97.1 wt%. Therefore the N i o . 3 3 M o P / A l 2 0 3 catalyst prepared in the present study had higher S but lower N conversion than the sulfided commercial catalyst, in agreement with previous results on metal phosphides using model compounds (Oyama et al., 2003). Bulk N i 0 3 3 M o P was reduced at a lower temperature in the presence of M o . Therefore the enhanced activity is attributed to the improved dispersion of the P. The C o o . 4 N i 2 P / A l 2 0 3 and the Pt -Coo .4Ni 2 P/Al 2 0 did not show very good conversion possibly because the tested catalysts were pretreated by sulfiding, resulting in the loss of the active phosphide. In the model compound studies reported in Chapters 4 and 5, the metal phosphide catalysts were pretreated in H 2 so that the passivation layer would be re-reduced and 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 Pt-Coo.4Ni2P/Al203 catalysts form large amounts of these CogSg crystals, the conversion w i l l be reduced. On the other hand commercial NHVI0/AI2O3 forms stable Ni 3 S2 crystals upon sulfidation 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 of the H D S and H D N of L G O (Sundaramurthy et al., 2006) and the apparent rate constants, k s and kN were calculated using equations 6.1 and 6.2 assuming that H D S follows 1.5 th order kinetics and H D N first order kinetics respectively. f k = 1 rr0.5 Q.0.5 LHSV (n-\) (6.1) kN = I n f e l W (6.2) N P S F and N F are the S and N concentrations (wt%) in the feed respectively and S p and N p are the S and N concentrations in 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 s and 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%" 0 5) was higher than the carbide (8.3 h _ 1 wt%" 0 5 ) , nitride (7.1 h"1wt%"0'5) and the commercial sulfided catalyst 158 Chapter 6 (9.1 KlvA%~°5). Similarly, at 633 K , the H D N kinetic rate constant over N io . 3 3 MoP catalyst (6.2 h"1) was higher than the carbide (3.7 h"1), nitride (3.4 h"1) and the commercial sulfided catalyst (3.3 h"1). However at 648 K , the H D N kinetic rate constant was comparable to the commercial 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 3, and Pt-Coo.4Ni2P/Al20 3 are shown in Table 6.2 and the values range from 0.8 - 3.3 h"'wt%" 0 5 and 0.6 - 3.7 Klvrt%~05 respectively while the H D N kinetic rate constants range from 0.2 - 1.7 h"1 and 0.2 - 2.3 h" 1, respectively, at the reaction temperature range o f 613 - 648 K . Figure 6.3 (A) and (B) show plots of the Arrhenius equation for the data obtained on the Nin . 3 3 MoP/Al20 3 catalyst prepared in the present study and the commercial sulfided catalyst. Table 6.2 A Apparent kinetic parameters for H D S of Light Gas O i l at different temperatures Temp k s (h_1wt%-K Nio.33MoP/Al 20 3 Co 0 .4Ni 2 P/Al 2 O 3 Pt-Coo . 4 N i 2 P / A l 2 0 3 Carbide* Nitride* Sulfide 613 9.3 0.1 0.6 4.7 3.9 5.9 623 10.4 1.2 0.9 6.0 5.4 8.8 633 14.5 1.6 1.3 8.3 7.1 9.1 648 27.0 3.3 3.7 - - -* Phosphorous doped NiMo/y-Al 20 3: Carbide and Nitride; Sulfide : Commercial catalyst 159 Chapter 6 Table 6.2 B Apparent kinetic parameters for H D N of Light Gas O i l at different temperatures Temp k N (h"1) K N i 0 3 3 M o P / A l 2 O 3 Coo. 4 Ni 2 P/Al 2 0 3 Pt-Co 0.4Ni 2P/Al 2O 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 3 catalysts (see Appendix B) and the apparent activation energies are reported in Table 6.3. Also included in Table 6.3 are the apparent activation energies obtained from testing Pt-Con.4Ni2P/Ai203 and Nin. 33MoP/Al203 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 of 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 of 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) of L G O over ( • ) Ni 0 .33MoP/Al 2 O3 and (•)commercial sulfided catalysts 161 Chapter 6 Table 6.3 Comparison of apparent activation energies for H D S and H D N o f L G O overNi 0 .33MoP/Al 2O3, Coo.4Ni 2P/Al203, Pt-Co 0.4Ni 2 P/Al 2O 3 and commercial sulfided catalyst Activation Energy,(kJ/mol) N i o . 3 3 M o P / A l 2 0 3 Coo .4Ni 2 P/Al 2 0 3 , P t - C o o . 4 N i 2 P / A l 2 0 3 Commercial Sulfide H D S 103 138 193 62 H D N 78 177 112 130 4 , 6 - D M D B T - - 65 -Carbazole 60 - - -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 203 and the Coo.4oNi2P/Al203 catalysts underwent the exact same pre-sulphiding and stabilization treatments as the conventional N i M o S / A l 203 catalyst. This approach was taken to ensure that a comparison under the same process conditions could be made. In addition, the pre-sulphiding of 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 of the used catalysts, therefore, the catalysts were recovered after reaction and were analysed following extraction in tetrahydrofuran to remove soluble coke material. The bulk chemical composition of 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 of S after reaction in 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 of Figure 6.4, showing that the C00.40NI2P/AI2O3 catalyst probably formed a metal phospho-sulphide, whereas no change in the bulk properties of the Nio.33MoP/Ai203 catalyst were apparent from the X R D data. Note that the conventional N i M o S / A l 2 0 3 catalyst also contained a significant amount of P and had the highest S content following reaction among the catalysts tested. These results clearly demonstrate that the oxide precursor of the conventional catalyst was readily sulphided, despite the presence of P, whereas the metal phosphides were resistant to sulphidation, the Nio.33MoP/Al 203 catalyst more so than the Con.4oNi2P/Al203 catalyst. The state o f the P in the conventional catalyst is clearly different to that of the metal phosphide. Table 6.4 Measured catalyst atom ratios before and after reaction in L G O Co N i M o atom ratio p * S Coo.4Ni2P/Ai203 - after T P R 0.20 1.00 - 0.55 -Coo.4Ni2P/Al203 - 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 203 - 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 * atom ratio relative to M o or N i as determined by ICP of the supported catalysts 163 Chapter 6 Table 6.5 Atom ratios determined by X P S of supported metal phosphide catalysts before and after reaction in L G O P / M M / A l S / M Before rxn After rxn Before rxn After rxn After rxn atom ratio* Coo.4Ni 2P/Ai203 "047 21)2 61)4 (X02 5.00 N i o . 3 M o P / A l 2 0 3 1.11 0.61 0.10 0.16 0.30 The data of Table 6.5 report the results of X P S analysis of the C00.40NL2P/AI2O3 and the Nio 3 3 M o P / A l 2 0 3 catalysts before and after reaction with L G O . Before reaction, the P / M ratio ( M = Co + N i or N i + Mo) indicated someP enrichment of the catalyst surface and similar observations were made on bulk N i x M o P and C o x N i 2 P catalysts (Abu and Smith, 2006, A b u and Smith, 2006). The S / M ratio measured after reaction was high for the Coo.4oNi 2P/Al 203 catalyst and for both catalysts the S / M ratio determined by X P S was much greater than that determined from the bulk chemical analysis of 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/Al203 catalysts, whereas the opposite was true for the Coo.4oNi 2P/Al203 catalyst, indicative o f the former catalyst having a higher resistance to sulphidation than the Coo .4oNi 2 P/Al 2 03 catalyst. The partial 164 Chapter 6 3 ro w CD Before rxn MoP B I After rxn V i Before rxn L 30 40 50 2 Theta 60 30 40 50 2 Theta 60 Figure 6.4 Comparison of X R D diffractograms obtained for N i 0 3 3 M o P supported on A 1 2 0 3 and AI2O3 -F, before and after reaction in L G O . 165 Chapter 6 sulfidation of the metal phosphides is also evident from the X P S N i 2p and M o 3d spectra of Figure 6.5. The B E for N i 2p and M o 3d of the catalysts before reaction (i.e. following reduction and passivation) are in agreement with those measured on the corresponding bulk catalysts (Abu 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 eV and the presence of N i 2 P by the N i 2p peak at B E of 853.7 eV. Following reaction, there is an increase in the B E of the M o 3d peak at 229.5 eV, consistent with the formation of M o S 2 , whereas the intensity of the N i 2p peak at 853.7 eV is significantly reduced. Finally we note that the M / A l ratio determined by X P S suggests a higher dispersion of the metal phosphide on the Ni0.33MoP/Al2O3 catalyst than the Coo.4oNi2P/Al2C>3 catalysts and this may be part of the reason for the lower activity observed for the latter catalyst. In summary, the Nio .3 3MoP/Al 203 catalyst prepared in the present study produced higher 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 of competitive adsorption with other molecules. 166 Chapter 6 Figure 6.5 X P S of 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 of modified transition metal phosphides has been investigated for the H D S of 4 , 6 - D M D B T . The addition of 3 - 24 wt% of Co to bulk N i 2 P indicated that the catalyst containing 3 wt% of Co (C00.08N12P), produced the highest selectivity to the direct desulfurization (DDS) product dimethylbiphenyl ( D M B P ) when tested with the refractory model-sulphur compound 4,6-dimethyldibenzothiophene (4 ,6 -DMDBT) . However, the activity of 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 of the surface phosphorous/metal (P/M) ratio determined by X P S and the ratio of n-propylamine uptake to C O uptake determined by adsorption indicated that the surfaces of the CoxNi2P have Bronsted acidity that isomerizes the methyl groups to less sterically hindered positions and enhances direct desulfurization of 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 2 0 3 , the Coo.4Ni 2P on fluorinated AI2O3 produced a modestly higher conversion than the neutral AI2O3 support. Pt was also added to the Coo.4Ni 2P to examine the possible enhancement of the hydrogenation route of 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 2P/Ai203 was due to the increased hydrogenation due to the presence o f the Pt. The Pt-Coo.4Ni2P/Al203-WA showed decreased conversion with more cracked products than the neutral AI2O3 support. H D N of carbazole over bulk N i x M o P and Nio. 33MoP on different supports were also studied and in agreement with previous reports no direct biphenyl product was obtained. Nio.33MoP/Ai203 showed a conversion of 96 mol% and high B C H X product (83 mol%) while Nio.33MoP/MCM showed 100 mol% conversion but with high cracked products and little B C H X (13.2 mol%) products as a result of high Bronsted acidity. Correlation of conversion and B C H X selectivity with the Bronsted acidity suggests that increased Bronsted acidity promotes conversion of 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 (HDS = 78.9 wt%, H D N = 56.7 wt%) at 613 K and comparatively also produced higher activity than the carbides (HDS = 82 wt%, H D N = 63 wt%) and nitrides (HDS = 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 of the kinetic rate constant of the direct desulfurization to the hydrogenation obtained using 4 , 6 - D M D B T was found to be high (k\/k2 =21) 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 of small quantities of Co to M2P enhances the D D S route of the H D S of 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 of 4 , 6 - D M D B T is enhanced leading to observed D D S selectivity. Another major contribution of the present study is that the kinetics of the H D N of 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 of industrial significance on the activity of metal phosphides using industrial feed (LGO) . The catalyst life has not been evaluated because such evaluation needs extended length of 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 of modified metal phosphides. • Increasing the surface area of bulk metal phosphides The B E T surface area of bulk Coo .os^P and Nin.07MoP were small ( < 10 m 2/g). It w i l l be desirable to increase the activity of the bulk by preparing bulk metal phosphides with high B E T surface area ( > 100 m 2/g). Recently, Yang et al. (2006) reported the preparation of bulk N i 2 P with high surface area (130 m /g) using a surfactant-assisted method-addition of polyethylene glycol tert-octylphenyl ether and ethylene glycol (Yang et al., 2006). • Nanocrystal metal phosphides can also be used to study structure-activity relationships. 170 Chapter 7 • Incorporate other sources of 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 in order to increase the conversion and D D S selectivity of the 4 , 6 - D M D B T . • Zeolites could also be used as supports for the prepared modified phosphides since they have large pores that can enhance reactivity of the S and N containing molecules. • Effect o f H 2 S The effect of H 2 S and S on the modified metal phosphides should be studied at longer times using a pilot plant as this study was limited to about 14 h time-on-stream. • Effects of mixed reactants The model compounds study did not take into consideration the possibility o f interaction of other reactants in the feed. Therefore the effects of other sulfur or nitrogen containing model compounds in the feed should be investigated. 171 References Abu , I. I., Smith, K . J. The effect of cobalt addition to bulk M o P and N i 2 P catalysts for the hydrodesulfurization of 4,6-dimethyldibenzothiophene. Journal of Catalysis, 241, 356-366 (2006). A b u , 1.1., Smith, K . J . Hydrodenitrogenation of carbazole over a series of 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 of 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 of fluorine on the structure and activity of N i W / A l 2 0 3 catalysts for H D S of thiophene and H D N of pyridine. 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 of the association of Co and N i on the catalytic properties of sulfided C o N i M o / A l 2 03 . American. Chemical Society, Divison of Petroleum Chemistry 43, 39-42 (1998). Botchwey, C , Dalai, A . , K . , Adjaye, J. Product selectivity during hydrotreating and mild hydrocracking of bitumen-derived gas oi l . 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 in 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 of 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 of a new catalyst for hydroprocessing: Tungsten phosphide. Journal of Catalysis, 200, 140-147 (2001). Col l ing, C. W. , Thompson, L . T. The structure and function of supported molybdenum nitride hydrodenitrogenation catalysts. Journal of Catalysis, 146, 193-203 (1994). Corbridge, D . E . C , Studies in inorganic Chmistry 2: Phosphorous an Outline of its Chemistry, Biochemistry and Technology, Elsevier, Scientific Publishing Company, Amsterdam, 1980. Corma, A . , Martinez, A . , Martinez-Soria, V . Hydrogenation of 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 t h A C S Meeting, Los Angelos California, 99-119 (1998). Dineen, G . U . , Cook, G . L . , Jensen, H . B . Estimation of types of nitrogen compounds in shale-oil gas oi l . Analytical Chemistry, 30, 2026-2030 (1985). Ding, L . , Zhang, Z . , Zhang, Y . , Ring, Z . , Chen, J. Effect of fluorine and boron modifications on the H D S , H D N and H A D activity of hydrotreating catalysts. Applied Catalysis A: General, 301, 241-250 (2006). EPA-Diese 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 . , Le Loarer, J. L . , Jolivet, J. P., Fridefond, C , in: Schuth, F., Sing, K . S., Weitkamp, J. (Eds.), Handbook of Porous Solids, vol . 3, (2002). Fang, K . , Ren, J., Sun, Y . Effect of nickel precursors on the performance of 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 of characterizing Zeolite Acidi ty. Chemical Review 95, 615-631 (1995). Farragher, A . L . , and Cossee, P., in: Proceedings of the 5 t h International Congress on Catalysis Ed . 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 of 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 of Chemical Reaction Engineering", 3 r d 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, New York (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 in phosphorous(III) compounds, phosphonium salts and oxyacids of phosphorous. Z . Naturfomsch, B29, 603-603 (1974). Fujikawa, T., Kimura, H . , Kiriyama, K . , Hagiwara, K . Development of ultr-deep H D S catalysts for production of clean diesel fuels. Catalysis Today, 111, 188-193 (2006). Gallezot, P., Dakta, J., Marssardier, J., Primat, M . , Imelik, B . , in: Proceedings of the Sixth International Congress on Catalysis, London, 2, 696 (1977). Gary, J. H . , Handwerk, G.E . , Petroleum Refinery Technology and Economics, Marcel Dekker, New york, 465 (1994). Geankoplis, C. J., "Transport Processes and Unit Operations." PrnticeHall, New Jersey. Louwers, (1993). Haji, S., Zhang, Y . , Kang, D . , Aindow, M . , Erkey, C. Hydrodesulfurization of model diesel using P t /A^Os catalysts prepared by supercritical deposition. Catalysis Today, 99, 365-371 (2005). Hanlon, R. T. Effects of PH2S, PH2 and PH2S/PH2 on the hydrodenitrogenation of pyridine. Energy & Fuels, 1,424-430 (1987). Harris, S. and Chianelli, R. R. Catalysis by transition metal sulfides: A theoretical and experimental study of 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 of the batch hydrotreatment of Rundle shale oi l . Fuel, 64, 925-930 (1985). 175 Ho, C. T. Deep H D S of diesel fuel: Chemistry and catalysis. Catalysis Today, 98, 3-18 (2004). Houlla, M . and Delmon, B . Use of X P S to detect variations in dispersion of 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 of methyl-substituted dibenzothiophenes catalyzed by sulfided Co-M o / y - A l 2 0 3 . Journal of Catalysis, 61, 523-527 (1980). Homma, T., Echard, M . , Leglise, J. Investigation of CoNiMo/Al 2 03 catalysts: Relationships between H 2 S adsorption and H D S activity. Catalysis Today, 106, 238-242 (2005). Hsu, C. S., Genowitz, M . W. , Dechert, G . J., Abbott, D.J . , Barbour, R., Molecular characterization of diesel fuels by modern analytical techniques, in: Song, C , Hsu, S., Mochida, I., (Eds), Chemistry of Diesel Fuels, Taylor & Francis, New York, (2000). Hughes, T. R., White, H . M . , White, R. J. Bronsted and Lewis acid site concentrations in fluorinated alumina from the infrared spectra of adsorbed pyridine species. Journal of Catalysis, 13, 58 -64 (1969). Isoda, T., Takase, Y . , Kusakabe, K . , Morooka, S. Changes in desulfurization reactivity of 4,6-Dimethyldibnzothiophene 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 Chemical Society, Preparation Divivison of Petroleum Chemistry, 41, 559-564 (1996). Jian, M . , and Prins, R. Mechanism of the hydodenitrogenation of quinoline over NiMo(P)/Al 2 03 catalyst. Journal of Catalysis, 179, 18-27 (2002). 176 Jian, M . , Kapteijin, A . , Prins, R. Kinetics of hydrodenitogenation of ortho-propylanilne on N i M o ( P ) / A l 2 0 3 catalysts. Journal of Catalysis, 168,491-500 (1997). Jian, M . and Pr ins /R. Existence of 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 of 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 of catalysts comprised of 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 of Pyridine hydrodenitogenation on MoSx-based catalysts: Experimental evidence. Journal of Catalysis, 131,204-214(1991). 177 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 of thiophene, benzothiophene, dibenzothiophene, and related compounds catalyzed by sulfided CoO—M 0 O 3 / Y-AI2O3: Low-pressure reactivity studies. Journal of Catalysis, 55, 129-137(1978). K i m , H . , Lee, J. J., Moon, S. H . Hydrodesulfurization of dibenzothiophene compounds using fluorinated NiMo/Al203 catalysts. Applied Catalysis B: Environmental, 44, 287-299 (2003). Kle in , J. C , Hercules, D . M . Surface characterization of model Urushibara catalysts. Journal of Catalysis, 82, 424-441 (1983). Kwak, C , K i m , M . Y . , Choi , K . , Sang, H . P. Effect of phosphorus addition on the behavior of 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 of the activity o f cobalt and nickel as promoters of 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 of routes to deep desulfurization. Journal of Catalysis, 158, 236-245 (1996). Landau, M . V . Deep hydrotreating of middle distillate from crude and shale oils. Catalysis Today, 36,393-429(1997). 178 Lappas, A . A . , Budistenau, R., Drakak, K . , Vasalos, I. A . Production of low aromatics and low sulfur diesel in a hydrodesulfurization (HDS) pilot plant unit. Global Nest: The International Journal, 1, 15-22 (1999). Ledoux, M . J. , Andre., J., Blekkan, E . A . , Luck, F. The role of the nature and the purity of the alumina support on the hydrodesulfurization activity of C o M o sulfides. Applied Catalysis A: General, 133, 321-333 (1995). Lee, J. J., K i m , H . , Koh , J. H . , Jo, A . , Moon, S. H . Performance of fluorine-added C0M0S/AI2O3 prepared by sonochemical and chemical vapor deposition methods in the hydrodesulfurization of dibenzothiophene and 4,6-dimethyldibenzothiophene. Applied Catalysis Environmental, 61, 274-280 (2005). Leon, R., Margolese, D . , Stucky, G . , Petroff, P. M . Nanocrystalline Ge filaments in the pores o f a mesosilicate. Physical Review B, 52, 4 2285-2288 (1995). Lewandowski, M . , Sarbak, Z . Acid-base properties and hydrofining activity of 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 of reduction temperature on hydrofining performance of supported molybdenum phosphide catalyst. Journal Natural Gas Chemistry, 14, 233-237 (2005). L i , Q., Brown, S., E . , Broadbelt, L . , J. , Zeng, J., Wu . , N . , Q. Synthesis and characterization o f MCM-41-supportd Ba2Si04 base catalyst. Microporous and Mesoporous Materials, 59, 105-111 (2003). Liaw, S. J., Raje, A . , B i , X . X . , Eklund, P. C , Geaham, D . M . , Davies, B . H . Hydrotreatment of naphtha with molybdenum nitride catalysts. Energy & Fuels, 9, 921-927 (1995). 179 M a , D . , Xiao , T., X ie , S., Zhou, W. , Gonzalez-Cortes, S. L . , Green, M . L . H . Synthesis and structure of bimetallic nickel molybdenum phosphide solid solution. Chemical Material, 16, 2697-2699 (2004). M a , X . , Sakanishi, K . , Mochida, I. Hydrodesulfurization reactivities of 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 in Athabasca maltene. AOSTRA Journal of Research 3, 25-35 (1986). Manol i , J . -N. , Da Costa, P., Brun, M . , Vrinat, M . , Mauge, F. , Potvin, C. Hydrodesulfurization of 4,6-dimethyldibenzothiophene over promoted (Ni,P) alumina-supported molybdenum carbide catalysts: activity and characterization of active sites. Journal of Catalysis, 221, 365-377 (2004). Massoth, F. E . Characterization of molybdena catalysts. Advances in Catalysis, 27, 266-310(1978). Mague, F. , Lamotte, J., Nesterenko, O., Tsyganenko, A . A . FT- IR study of surface properties o f unsupported M o S 2 . Catalysis Today, 70, 271-284 (2001). Mayo, S., Brevoord, E . , Gerritsen, L . , Plantenga, F. Process ultra-low sulfur diesel. Hydrocarbon Process 80 2, 84A (2001). Maxwel l , I. E . Innovation in applied catalysis. Cattech 1 5-13 (1997). Mcl lvr ied, H . G . Kinetics of hydrodenitrification of pyridine. Industrial Engineering Chemistry Process Development, 10 ,125-130 (1971). Mears, D . E . Tests for transport limitations in experimental catalytic reactions. Industrial Engineering Chemistry Process Development, 10 541-547 (1971). Mi l le r , J. T., Hinemann, M . F. Non-first-order hydrodenitrogenation kinetics of 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 of nitrogen compounds on C0M0P/AI2O3 cataysts with alkali or zeolite added in hydrodesulfurization of dibenzothiophene and 4,6-dimethyldibenzothiophene. Applied Catalysis A: General, 295, 193-200 (2005). Montgomery, D.S . Department of Mines and Technical Surveys Information Circular, IC 169 Sept. (1964). Mosio-Mosiewski , J., Morawski, I. Study on single-stage hydrocracking of vacuum residue in the suspension of N i - M o catalyst. Applied Catalysis A: General, 283, 147-155 (2005). Nagai, M . , Fukiage, T., Kurata, S. Hydrodesulfurization of dibenzothiophene over alumina-supported nickel molybdenum phosphide catalysts. Catalysis Today, 106, 201-205 (2005). Nagai, M . , Goto, Y . , Irisawa, A . , Omi , S. Catalytic activity and surface properties of nitrided molybdena-alumina for carbazole hydrodentrogenation. Journal of Catalysis. 191, 128-137(2000). Nagai, M . , Goto, Y . , Kiyoshi , M . , Hada, K . , Oshikawa, K . , Omi , S. Temperature -programmed reduction and X R D studies of ammonia-treated molybdenum oxide and its activity for carbazole hydrodenitrogenation. Journal of Catalysis, 182, 292-301 (1999). Nagai, M . , Koyama, H . , Sokomoto, S., Omi , S., in: Hydrotreatment and Hydrocracking of 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 on a M0/AI2O3 catalysts. Effects of sulfiding and sulfur compounds. Energy & Fuels, 2, 645-651 (1988). 181 Nagai, M . , Goto, Y . , Uchino, O., Omi , S. T P D and X R D studies of molybdenum nitride and its activity for hydrodenitogenation of 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 t h December, 2002). Nelson, A . E . , Sun, M . , Junaid, A . S. M . On the structure and composition of the phosphosulfide overlayer on N i 2 P at hydrotreating conditions. Journal of Catalysis, 241, 180-188 (2006). Nylon , M . K . , Choi , S., Kwon , H . , Curry, K . E . , Thompson, L . T. Catalytic properties of early transition metal nitrides and carbides: tt-butane hydrogenolysis, dehydrogenation and isomerization. Applied Catalysis, 183, 253-263(1999). Odebunmi, E .O. , Oll is , D . F. Catalytic hydrodexygenation III. Interactions between catalytic hydrodeoxygenation of m-cresol and hydrodenitrogenation of indole. Journal of Catalysis, 80, 76-89 (1983). Okamoto, Y . , Niha, Y . , Imanaka, T., Teranishi, S. Surface state and catalytic activity and selectivity of nickel catalysts in hydrogenation reactions III. Electronic and catalytic properties of nickel catalysts. Journal of Catalysis, 64, 397-404 (1980). Oyama, S. T., Wang, X . , Lee, Y . K'., Chun, W.-J . Active phase of Ni 2P/Si02 in hydroprocessing reactions. Journal of Catalysis, 221 263 (2004). Oyama, S. T. Novel 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 of phosphorus content in 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 of 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 of 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 of 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 of silica-supported molybdenum phosphide catalysts. Journal of Catalysis, 20,266-273 (2002). Phillipson, J. J. Kinetics and hydrodesulfurization of light and middle distillates in: Proceedings of the Paper Presented at the American Institute of Chemical Engineers Meeting, Houston, (1971). Portefaix, J. L . , Cattenot, M . , Guerriche, M . , Thivolle-Cazat, J., Breysse, M . Conversion of saturated cyclic and noncyclic amines over a sulphided MM0/AI2O3 catalyst : mechanisms of carbon - nitrogen bond cleavage. Catalysis Today, 10 473-487 (1991). Powdered Diffraction Powder Files, J C P D S International Center for Diffraction Data, Swathmore, P A , (2005). 183 Pinzon, M . H . , Centeno, A . , Giraldo, S. A . Role of Pt in high performance Pt-Mo 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 mild hydrocracking feedstocks and method for its preparation. European Patent application. (1992). Prins, R. Egorova, M . , Rothlisberger, A . , Zhao, Y . , Sivasankar, N . , Kukula , 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., De Beer, V . H . J. Structure and function of catalyst and promoter in 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 of 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 of promotion effect of nickel on M0/AI2O3 and C0-M0/AI2O3 catalysts in hydrodesulfurization using a S radioisotope tracer method. Applied Catalysis. A, 227, 19-28 (2002). Qu, L . , Prins, R. Different Active sites in hydrodenitrogenation as determined by the influence of the support and fluorination. Applied Catalysis A. General, 250 105-115 (2003). Rabarihoela-Rakotovao, V . , Brunet, S., Berhault, G . , Perot, G . , Diehl , F . Effect o f acridine and of actahydroacridine on the H D S of 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 . , in "Proc. 9 t h Int. Congr. Catal. Calgary" (M.J.Phill ips and M . Terman, Eds) p.82 Chemical Institute of Canada, Ottawa, (1988). Robinson, W . R. A . M . , Van 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 of Ni(Co)-containing M o -free catalysts in 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 of M o P , N i 2 P and M o N i P hydrodesulfurization catalysts: Time-resolved x-ray 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 of nickel catalysts. Catalysis Today, 90 159-166 (2004). Satterfield, C . N . and Yang, S. H . Catalytic hydrodenitrogenation of quinoline in 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 of vapor phase catalytic hydrodenitrogenation of quinolie. Industrial Engineering Chemistry Process Design and Development, 20, 53-62 (1981). Satterfield, C . N . "Heterogeneous Catalysis in industrial practice." M c G r a w - H i l l N e w York (1991) Sawhill , S. J., Layman, K . A . , V a n Wyk, D . R., Engelhard, M . H . , Wang, C , Bussel, M . K . Thiophene hydrodesulfurization over nickel phosphide catalysts:effect o f the precursor composition and support. Journal of Catalysis, 231, 300-313 (2005). 185 Senzi, L . , Jae, L .S . , Taeghwan, H . , Suslick, K . S . Catalytic hydrodenitogenation of 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 in activated carbon supported C o N i M o hydrodesulfurization catalysts. Journal of Catalysis 189, 244-246 (2000). Shuit, G . C. A . and Gates, B . C. Chemistry and Engineering of catalytic hydrodesulfurization. Journal of. American Institute of Chemical Engineering, 19, 33 417-438 (1973). Song, C , Introduction to chemistry of diesel fuel, in: Song, C , Hsu, S., Mochida, I., (Eds), Chemistry of Diesel Fuels, Taylor & Francis, New York, (2000). Song, C . A n overview of 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 of 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 of 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, Th., Prins, R. Preparation and 3IP N M R characterization of nickel phosphide on silica. Journal of Catalysis, 208, 456-466 (2002). Stinner, C , Prins, R., Weber, Th. Binary and ternary transition-metal phosphides as H D N catalysts. Journal Catalysis, 202, 187-194 (2001). Stinner, C , Prins, R., Weber, Th. Formation, structure and H D N activity of 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., Li , 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 HDN 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 of light cycle oi l . 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. On the formation of type I and type II N i M o S phases in N i M o/Al203 hydrotreating catalysts and its catalytic implications. Fuel Processing Technology, 35, 137-157 (1993). Vivier , L . , Dominguez, V . , Perot, G . , Kasztelan, S. Mechanism of carbon-nitrogen bond scission. Hydrodenitrogenation of 1,2,3,4-tetrahydroquinoline and of 1,2,3,4-tetrahydoisoquinoline. Journal of Molecular Catalysis, 67, 267-277 (1991). Volpe, L . , Boudart, M . , J. Compounds of 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 of Catalysts, Elsevier, Amsterdam, 147 (1983). Wada, T., Kareda, K . , Murata, S., Nomura, M . , Effect o f modified Pd metal on hydrocracking of 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 . , Hu , Y . Hydrodesulfurization of 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 of different diluents in PI/AI2O3 catalyst on the hydrogenation of 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 of several iron group transition metal phosphides. Journal of Catalysis, 208, 321-331 (2002). Wise, R. S., Markel , E . J. Synthesis of high surface area molybdenum nitride in mixtures of nitrogen and hydrogen. Journal of Catalysis, 145 344-355 (1994). Yang, H . , Chen, J., Briker, Y . , Szykarczuk, K . , Pving, Z . Effect of nitrogen removal from light cycle oil on the hydrodesulfurization of dibenzothiophene, 4-methyldibenzothiophene and 4,6-dimethyldibenzothiophene. Catalysis Today, 109, 16-23 (2005). Yang, 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. New catalysts for hydroprocessing: Bimetallic oxynitrides M i - M n - 0 - N ( M I , M H = M o , V , Nb, Cr, M n and Co) Part I. Synthesis and characterization. Journal of Catalysis. 173, 1-9 (1998). Yumoto, M . , Usui , K . , Watanabe, K . , Idei, K . , Yamazaki, H . Development of 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 of Fossil Energy, Department of Energy, Washington, D C (1982). Zuzaniuk, V . , Prins, R. Synthesis and characterization of silica-supported transition-metal phosphide as H D N catalysts. Journal of Catalysis, 219, 85-96 (2003). 189 Appendix A Appendix A A l : 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 of the internal and external mass transport were taken. A2: Parameters for testing plug flow The dimensions of the reactor and the catalyst are listed in Table A . 1. Table A l Parameters of catalyst and reactor Parameter Dimension Diameter of reactor D = 9.53 x 10" m Diameter of catalyst particle dp = 7 x 10"4 m Length o f reactor L = 6.1 x 10"1 m Length of catalyst bed L B = 2 x l 0"2 m Great care must be exercised to ensure that the flow pattern is ideal even when plug flow prevails. To ensure that the influence of the reactor walls on the flow is eliminated, the diameter of the PFR, must be at least 10 times the catalyst particle diameter, i.e. D/dp > 10. In addition, by virtue of convection, axial gradients exist. These effects are minimized by selecting the correct ratio of the length of the reactor to the particle diameter. Thus L/dp > 50. In this set up, D/dp =9.53 x 10"3 m / 7 x l O ^ m A . l 190 Appendix A = 13.6 L/dp = 6.1 x 10"' m / 7 x 10"*m A . 2 = 871 Since D/dp = 13.6 > 10 and L / d = 871 > 50, both axial and wall effects w i l l be minimized in 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 p e > N p e min, where N p e is the Peclet number and N p e m j n is the minimum Peclet number. For the gas phase operation, N p e =(0.087) < 2 3 A A.3 Npemin = 8n I n A . 4 1 - X n = 1, NRe is Reynold's number and x is conversion. The superficial velocity is given by u = 4 V / TCD 2 A .5 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 / 6 0 x(583/298) , M , , u = r - r - = 1.07x10 Tn/s 3.14JC (9.53JC10~ 3) 2 Assume the viscosity of the fluid is the same as the solvent dodecane, u. = 1.006 x 10"3 Ns /m 2 , NRe can be calculated as, N R e = udpp/p. _ 1.07 x\Q-*ms x 7x\Q-4 x 741.1 kg/m3 1.006 xl0~3Ns/m2 From equation A . 3 , N Pe = 0.087 x ( 0 .055) 0 2 3 (871) = 38.92 and from equation A . 4 and for x = 0.291 N i 2 P conversion, 191 .3=0.0552 A . 6 Appendix A N Pemin= 8 x In 1/(1-0.291) = 2.75 N p e (38.92) > Npemin (2.75) therefore deviation from plug flow is acceptable. A.3 Diagnostic Test for internal and external mass transport effects Some considerations wi l l be given to the effects of 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 wi 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>2 = r A P p R 2 / D A B C A A . 7 where r A is the observed reaction rate( for bulk N i 2 P = 2.06 x 10"9 m o l g ' V 1 ) , p p ( N i 2 P = 7.09 g cm"3) is the catalyst packing density, R = 0.035 cm, is the radius of the catalyst particle C A = 9.82xl0" 6 mol cm" 3 is the concentration of the reactant and D,VB = 1.18 x 10"4 c m 2 s"1 is the effective difftisivity of 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-* = Q ^ 1 .18JC10 _ 4 X9.82X10- 6 therefore, (j) = 0.124. From a graph of 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 of the supported NI2P/AI2O3, the effective diffusivity in the pores w i l l taken into account. The effective diffusivity, De, in a porous catalyst is given by: T „ ^ A B "p De = p -— A . 9 where D A B is the relative diffusivity of the 4 , 6 - D M D B T in dodecane, e P ; is the porosity, a is the constriction factor and x is the tortuosity. Typical values of ep, o, and 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" 3cw/s A.IO 3.0 For the N i 2 P / A l 2 0 3 , R = 0.035 cm, p = 0.37 g/cm 3 , (values obtained from Wang et al., 2002), r A = I 3 . l x 10"9 and C A = 9.82 x 10"6 mol/cm 3 hence putting in equation A.7 , , (3.5xl0"2)2x0.37x n . lx lO" 9 4 $ 2 = - - — : = 4 . 7 9 x l 0 - 4 A . l l 1.26JC10"3X 9.82X10 6 Hence § = 0.0219. This value also corresponds to also an effectiveness factor of 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 of external mass transfer is tested using the following condition (Foggier, 1981): - r A p R n / k A C 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 of ICA is obtained from the Frosling correlation: Sh = k A dp/Deff - 2 + 0.6 R e 0 5 Sc 0 3 3 A . 13 where Sh is the Sherwood number, Re is the Reynold's number (0.0552), and Sc is the_Schmidt's number (p/pD AB =115) and the rest o f the symbols take their definitions as given earlier. Hence from equation A . 9 , k A = (2+0.6Re 0 - 5 Sc 0 3 3)xDAB /dp . A.14 = (2+0.6*0.0552° 5 x 1150.33) x 1.18 x 10"4/ 0.07 = 4.51 x 10"3 cm 2/s Substituting all values in equation A .8 and for n = 1, 2 . 0 6 x l 0 ' 9 molls.g xO.lAlglcm3 x 0.035cm 4.51 x 10"3 cm2 / j x 9.82 x l O ' 6 mol I cm3 = 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 of reaction for organo-sulfur and nitrogen containing compounds are in the ranges of -68 to -13 and -65 to - 7 kcal/mol respectively. Also , the heats of adsorption of 4,6-D M D B T is 20-30 kcal/mol (Kabe et al., 1999). Since these heats of 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 of 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 of 4 , 6 - D M D B T was estimated to be 649.61 K . The vapor pressure, P s o f 4 , 6 - D M D B T is estimated from the modified Clausis-Clapeyron equation (Baum, 1995). f A C/T' \ \ In (P s) = AS(TB) -6.%x(TM/T-l) A.15 R[\.S x(TBIT-\)- 0.8 x \n{TBIT)\ where A S ( T B ) = Entropy, J/mol-K at the boiling temperature, = 81.119+ 13.083 log T B - 2 5 . 7 6 9 x T B / M +0.146528 x T B 2 / M -2.1362 x 1 0 " 4 x T B 3 / M A.16 P s = saturation vapor pressure at the specified temperature, atm, T M = melting point temperature of the compound, K , T = specified temperature and M = molecular weight of compound, g/mol. For Dodecane, the P s is estimated from the Antoine equation, A . 17 log(P s ) = A l - ( B 1 / ( T + C 1 ) ) A.17 where A l = 4.09978, B l = 1625.928, C I = -92.839. A t the experimental conditions, T = 583 K , and molecular weight of 4 , 6 - D M D B T = 212.31 g/mol, equations A.15 and A.16 are solved to give P s of 4 , 6 - D M D B T = 4.54 kPa and equation A . 13 gives P s of dodecane = 28.98 kPa. 195 Appendix A A . 6 Determination of vapor pressure of feed Assumptions: 1. Concentration of 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 v = y i P = X j P s A . 18 Where P v is the partial pressure of component, and y i and Xj are the vapor and liquid mol fractions respectively, P s is the saturation vapor pressure But for our system, y A + y B + y c = 1 (gas phase) A . 19 x A + XB = 1 (liquid phase) A.20 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 of 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 of 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 of M o P catalyst = 3 . 0 g * * * * • ™ r. 95.94 g/mol Mo • Amount of M o in M o P • x 3 ? 126.94 gl mol MoP = 2.27 g Amount of ammonium heptamolybdate (NH 4 )6Mo7024 .4H20 ( A H M ) required = 221 gx 1235.86 gl mol AHM 95.94 glmolMoxl = 4.17g Amount of diammonium hydrogen phosphate (NH4)2HP04 required = (3.00-2.27) x 132.06 30.97 = 3.12g A d d 4.1725 g (3.38 mmols) and 3.1241 g (23.66 mmols) of (NH 4)6Mo7024.4H 20 and (NH4)2HP04 respectively in a 60 ml beaker and add 15 ml of 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 Coo .os^P Required amount of Coo.o8Ni2P = 3.0 g , Amount of Co = 0.03 x 3 g = 0.09 g Amount of cobalt hexahydrate ( (Co(N03)2 .6H20 required 290.59 x 0.09 = = 0.44 g 58.93 Amount of N i 2 P = 3.0-0.09 g = 2.91 g A • f x r - M - D 58.69 x 2 x 2.91 Amount of N i in N12P = 148.35 = 2.30g Amount o f nickel nitrate Ni(N03)2.6H 20, required = 290.81 x 2.30 58.69 = 11.41 g Amount of diammonium hydrogen phosphate ( N T - L ^ H P C M required = (3 .00 -2 .30-0 .09) 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 3 ) 2 . 6 H 2 0 and (NH 4 )2HP04 respectively in a 60 ml beaker and dissolve with 15 ml of de-ionised water. Stir to form M 2 P precursor. Next dissolve 0.44 g (1.53 mmols) of cobalt hexahydrate ((Co(NO"3)2.6H20 in 10 ml of de-ionised water and add to the beaker containing the M 2 P precursor. Stir continuously while evaporate in hot plate. Then follow procedure as described in Chapter 3. 198 Appendix B 3. Calculations for preparing Coo.4Ni 2 P/Al 203 ( 3 wt% Co and 15 wt% N i ) Required amount of Coo .4Ni 2P/Al 203 = 8.0 g Amount of Co • = 0.03 x 8 g = 0.24 g Amount of cobalt hexahydrate ((Co(N03) 2 .6H 2 0 required 290.59 x 0.24 = = 1.18 g 58.93 Amount of N i = 0 . 1 5 x 8 = 1.2 g 148.35* 1.2 Amount of N i 2 P = = 1.52 g 2 x 58.69 = 2.30g Amount of nickel nitrate N i ( N 0 3 ) 2 . 6 H 2 0 , required = 290.81xl .20 58.69 = 5.95g Amount of diammonium hydrogen phosphate (NH4) 2 HP04 required = (1.52-1.2) x 132.06 30.97 = 1-35 g Amount of A 1 2 0 3 = 8.0-1.52-0.24 g = 6.24 g Required volume of solution = 0.76 ml/g x 6.24 g = 4.75 ml . Since the volume of solution is small (4.75 ml), 15 ml of 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) of N i ( N 0 3 ) 2 . 6 H 2 0 and (NH4) 2 HP04 respectively with 15 ml of de-ionised water and stirred. 15 ml was chosen to 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) of cobalt hexahydrate ( ( C o ( N 0 3 ) 2 . 6 H 2 0 in 5 ml of de-ionised water and impregnated on the AI2O3 and dried again at 393 K . After step 2, the precursor was calcined, reduced in H 2 and passivated following the same procedure as described in Chapter 3. B.2 TPR 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 of the T P R profile was obtained by integration using Origin software. Area o f obtained from T P R profile of CU2O = 3.469 a.u 2. Reduction stoichiometry of CU2O. C u 2 0 + H 2 • 2Cu + H 2 0 B . l From equation B . l , 1 mol of CU2O requires 1 mol of H 2 (assume complete reduction). Initial amount of CU2O sample = 0.05 g M o l of C u 2 0 = 0.05 g/ 95.545 g/mol = 5.2331 x 10" 4 mol Therefore mol of H2 consumed = 5.2331 x 10"4 mol The H2 consumption per area = ^-2331 x 10—rnol_ _ j ^ Qgg x i Q-4 J^QI/^QQ 3.469 a.u Hence T P R calibration = 1.5086 x 10 " 4 mol/a.u 200 Appendix B B.3 Degree of reduction: Example of Ni2P The degree of reduction of N i 2 P was determined as follows: 1. Integrated area of T P R profile of N i 2 P precursor = 21.1467 a.u 2. N i O . P 2 0 5 phase was identified as the precursor using X R D hence the reduction stoichiometry was given as: 4 N i O . P 2 0 5 + 9 H 2 • 2 N i 2 P + 9 H 2 0 B.2 Initial mass of N i O . P 2 0 5 precursor = 0.4269 g mol N i O . P 2 0 5 = 0.4269 g / 216. 63 g/mol = 1.9706 x 1 0 ' 3 mol From stoichiometry mol H 2 (assumed complete reduction) = 4.4612 x 10 " mol 3. T P R integrated area of sample =21.1467 a.u M o l H 2 consumed = 1.5086 x lO" 4 (mol/a.u) x 21.1467 = 3.19 x 10' 3 mol v i - 3.19 x l O - 3 i n n Hence degree of reduction = 100 4.4612 x 10 ~3 = 71.94% 201 Appendix B B.4 Temperature programmed reduction of transition metal phosphide precursors in H 2 using tapered element oscillating microbalance (TEOM). The T P R of N i 2 P , Coo.osNi2P and M o P precursors in H 2 were obtained using T E O M and Figure B l shows the profile of the mass change of 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 of the single high temperature peak o f Figure B l T P R of 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 B2 shows the mass change of the N i 2 P precursor with time of reduction using the T E O M . The percent mass loss was estimated as follows: 0 -0.07 -I , 1 1 , 1 0 10000 20000 30000 40000 50000 Time, s Figure B2 Mass change during T P R of Coo.o8Ni2P phosphide precursor using T E O M Initial amount of Coo.o8Ni2P precursor loaded = 0.1264 g Mass change after reduction = 0.064 g Therefore % mass loss = 0 0 6 4 x 100 0.1264 = 51 % (as shown in 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 5 5 6 0 6 6 7 0 ] Ni 2-phosphide precursor 10 15 2 0 2 5 3 0 3 5 4 0 4 5 5 0 5 5 6 0 6 5 10 15 2 0 2 5 X 3 5 4 0 4 5 5 0 5 5 6 0 6 5 7 0 20 Figure B3 X R D patterns of calcined precursors. Phases identified: Mo-phosphide precursor: M 0 O 3 . P 2 O 5 ; Co-phosphide precursor: CoO.P20 5 ;Ni 2-phosphide precursor: N i . O P 2 O s ; Coo.o8Ni2-phosphide precursor: CoO.P 2C>5, N i .OP 2 Os Coo.osMo-phosphide precursor: CoO.P 2 Os ; Mo03.P2Os 204 Appendix B B.6: Determination of crystallite sizes. The crystallite sizes of the bulk metal phosphides were determined from the X R D diffractograms. The N i x M o P series prepared in Chapter 4 w i l l be used as examples for determining the crystallite sizes. The X R D diffractogram for M o P is shown in Figure B4 . 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. Apply integral breadth: B (specimen) = — ^ x -J— B.3 Hp ISO where A p is the total area of peak and Hp = Peak height (both obtained from peak integration). 4. Separate instrumental broadening: where X = wavelength (1.54056), 0 is the angle at the peak of integration and B(standard) is the B a F 2 X R D pattern which is the instrument. P*(total) P (specimen) * cos (O)A, B.4 P*(specimen) = p* (total)- (P*(standard))2/ p*(total) 5. Crystal size, dp, = 1/ P*(specimen) 205 Appendix B Chi A 2 = 282.1 193485 C O D = 0.9974 Date:2/24/2006 Corr Coef=0.9987 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 Fitting Results 4 0 A r e a F i t 1 1 51 . 9 7 5 1 4 2 4 4 4 6 2 the t a 4 8 5 0 M a x H e i a h t 1 7 9 6 . 3 1 2 4 5 F W H M 0 . 4 0 8 2 5 BaseLine: C O N S T A N T Figure B4 Integration of the most intense peak of the M o P diffractrogram For example to calculate the M o P crystallite size from Figure B3 , 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 = 0.011187 p (standard) = (3.14/180) * 197.8/1708 = 0.00202 B*(Total)-specimen = 0.011187 *Cos (42.86/2) / 1.54056 =, 0.00676 P*(Total)-standard = 0.00202 *Cos (24.85/2) / 1.54056 = 0.00128 Therefore from equation B.4, p*(specimen) = 0.0067 - 0.00128 2 / 0.0067 = 0.00652 Hence, dp = 1/p* (specimen) = 158.433 A = 16 nm 207 Appendix B The results for the rest of 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 3 (Total) [3* (specimen) d. Ni0.07MoP 550.53 982.4 42.91 0.0098 0.0059 0.0056 18 N i 0 1 6 MoP 268.81 558.15 42.98 0.0084 0.0051 0.0048 21 Nio.3 8MoP 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 BaF2(standard) 197.8 1708 24.85 0.0020 0.0013 B.7 Calculation of Lattice Parameters. The lattice parameters of the prepared bulk M o P , C00.07M0P and CoxNi2P (0< x <0.34) were calculated using equation B.5 for hexagonal phases. -Xr = —x (h2 + hk +k2)/a2 + l— B.5 d 3 . c2 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 of the X R D source and 0 is the angle of the diffraction peak. 208 Appendix B The example of 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 - -x (h2 + hk + k2)la2 -Xr=0 B.7 d2 3 c2 Hence when B.7 is satisfied to some acceptable error then the parameters are accepted. 1/d2 = 1 / 2 sin (21.34/180* TC) = 0.2246 jx (h2 + hk +k2)/a2 =4/3 * ( 1 + 0 + 0 )/a2 = 1.3333/a2 From equation B.7, 0.2246 - - l a 2 - \ = 0 B.8 3 c2 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 of 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 of 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 of the survey scan is shown in Figure B5 . Both the bimetallic C00.07M0P and Co 0.o8Ni 2P reveal the presence of N i 2p, M o 3d, P 2p and P 2s. In addition, the X P S profiles also show O (A), O l s and C Is. 210 Appendix B 2.50E+05 2.00E+05 </) Q. O >M.50E+05 CO c cu 1.00E+05 5.00E+04 0.00E+00 CO0.07M0P 0(A) 01s Mo3d 1100 1000 900 800 700 600 500 400 300 200 100 0 Binding Energy, eV Figure B5 XPS survey scan of Coo.osNiiP and C00.07M0P 211 Appendix B B.9 Determination and repeatability of C O chemisorption The C O uptake data from Co 0 .o8Ni 2 P wi l l be used to show the repeatability of the 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 of sample loop = 1ml Run 1: C O uptake = 0.90 mmols/g, Run 2: C O uptake = 0.99 mmols/g The determination of the C O uptake is carried out as follows: In'ected C O mols = ^'^m^1m^nCO) (sample loop size (\ml)) njec e mo s (30m// min He) (22414m/1 mol) = 4 . 4 6 x 10"7 mol 212 Appendix B Table B2 shows the determined areas and sample sizes used. Assume the final peak area corresponds to the mol C O injected (since at saturation the sample does not take in anymore C O and the area is constant). Table B2 Results of the repeatability of C O uptake on C o o . o s ^ P Sample wt. Integrated Area (a.u) C O Uptake (g) Final Peak Difference Total umols R u n l 1.0010 0.01601 3.2 x 10"2 0.90 Run2 1.0801 0.01350 3.14 x 10"2 0.96 The uptake is calculated as follows: Run 1: C O uptake = 0.0320 x 4.46 x 10 ~7 mols / 0.01601 / 1.0010 g = 0.90 umols/g Run 2: C O uptake = 0.0314 x 4.46 x 10 ~7 mols / 0.01350 / 1.0801 g = 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 -PA chemisorption data for Con.osNiiP w i l l be used to show the repeatability of the experiments. Figure B7 shows two runs using C00.08N12P. 0 2000 4000 6000 8000 10000 12000 Time.s Figure B7 Repeated n -PA data using Con.os^P / ' Determination of Brensted acidity The system was calibrated using 3 zeolite samples of known acidity. C00.08N12P n-PA chemisorption w i l l be used to show how the Brensted acidity was calculated. Average integrated area for zeolites =0.16208 V.s Flow rate of He =30 ml/min Flowrate of n -PA =12.5 ml/min 214 Appendix B Volume of sample loop = 1 ml Therefore mols of n -PA = 1 m l x 1 2 5 m l / m i n 30 ml/min x 22400 ml/mol = 2.03x 10"5 mol Table B3 shows the sample sizes and integrated areas. The Bronsted acidity is calculated as follows: Table B3 Results of the n -PA repeatability using Co 0.o8Ni 2P Sample wt. Integrated Area Bransted acidity (g) (V . s ) (umols) R u n l 0.9666 1.12102 145.40 Run2 0.7986 0.84701 133.13 R u n l : Bransted acidity Similarly for Run 2: Error 2.035x10" 5 molx \A2 \V .s x 1000 0.1621 V.s x 0.9666g =145.40 umols/g 2.035x10"5 mol x 0.8470 V.s x 1000 0.1621 V.s x 0.7986g ; 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 of 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 B8 is re-written as: Fdx = -rdW (B8) Ao A Substituting for - r A = kCAo, solution of equation B8 gives (l-xA) = e-k,m (B9) v 0 is the volumetric flow rate. Different values of 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 of the T O F , the specific consumption is obtained assuming the reactant at the entrance of 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 B9 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"6 mol/gcat.s. C O uptake = 1.11 pmol/g 216 Appendix B Therefore, _ 8.11 x lO"° mollgcatls 1 O r — 1.11 x 10"° mollg = 7.31 s" 217 Appendix B 1.5 -1 -I , , , , , — | 1.52 1.54 1.56 1.58 1.6 1.62 1.64 0.5 -2 -I , 1 , , , 1 1.52 1.54 1.56 1.58 1.6 1.62 1.64 Figure B8 Arrhenius plots for determining the apparent activation energy for the H D S ( A ) and H D N (B) of L G O over ( • ) C00.4N12P/AI2O3, and (•) P t - C o o A P / A ^ O s 218 Appendix B Figure B9 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 of experiments for hydrodesulfurization of 4,6-dimethylmethyl dibenzothiophene (4 ,6-DMDBT-3000 ppm) over transition metal phosphides at 3.0 M P a , H 2 pressure for 12 h time on stream Exp. Reactant Catalyst Temp. S V Conversion # K xlO" 2 mol/h-g 1 4 , 6 - D M D B T N i 2 P 583 7.9 29.1 2 4 , 6 - D M D B T M o P 583 7.9 36.2 3 4 , 6 - D M D B T CoP 583 7.9 22.3 4 4 , 6 - D M D B T Co 0.o8Ni 2P 583 7.9 48.7 5 4 , 6 - D M D B T Coo . i 6 Ni 2 P 583 7.9 23.4 6 4 , 6 - D M D B T Coo.3 6Ni 2P 583 7.9 17.4 7 4 , 6 - D M D B T Coo.79Ni2P 583 7.9 10.1 8 4 , 6 - D M D B T C00.07M0P 583 7.9 42.5 9 4 , 6 - D M D B T Co 0.o8Ni 2P (Repeat) 583 7.9 47.2 10 4 , 6 - D M D B T C o 2 P 583 7.9 43.9 11 4 , 6 - D M D B T N i 2 P / A l 2 0 3 583 7.9 81.3 12 4 , 6 - D M D B T C o o , 4 N i 2 P / A l 2 0 3 583 7.9 86.5 13 4 , 6 - D M D B T Co 0 .4Ni 2 P/Al 2 O 3 (repeat) 583 7.9 84.4 14 4 , 6 - D M D B T C o 0 . 4 N i 2 P / A l 2 O 3 - F 583 7.9 87.1 15 4 , 6 - D M D B T C o 0 . 4 N i 2 P / M C M . 583 7.9 99.8 16 4 , 6 - D M D B T P t - C o 0 4 N i 2 P / A l 2 O 3 583 7.9: 97.2 17 4 , 6 - D M D B T P t - C o 0 4 N i 2 P / A l 2 O 3 - W A 583 7.9 90.0 18 4 , 6 - D M D B T P t - C o 0 4 N i 2 P / A l 2 O 3 583 9.2 93.5 19 4 , 6 - D M D B T P t - C o 0 4 N i 2 P / A l 2 O 3 583 32.0 63.5 20 4 , 6 - D M D B T P t - C o o . 4 N i 2 P / A l 2 0 3 583 62.0 51.7 21 4 , 6 - D M D B T P t - C o 0 4 N i 2 P / A l 2 O 3 548 7.9 64.3 22 4 , 6 - D M D B T P t - C o 0 4 N i 2 P / A l 2 O 3 533 7.9 76.6 220 Appendix C C.2 Response Factor and sample calculation of HDS activity Response factor of the 4 , 6 - D M D B T in dodecane was obtained by injecting known composition of 0.1 umol of the liquid sample into gas chromatography (GC). Four samples of each fixed composition were injected and the average reading was used to plot the graph of concentration versus the G C area. A s shown in Figure C I , the slope = 3.09 x 10"9 mol/area was 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 C2 . Table C2 Results of hydrodesulfurization of 4 , 6 - D M D B T over Coo.osNiiP Time, h 4,6-DMDBT 4,6-DMDBT 4,6-DMDBT DMBP DMBP DMBCH DMBCH Area mol x10" 1 0 Conversion mol% Area mol x10" 9 Area mol x 10" 1 1 2 2061910 6.38 35.07 103518 0.32 267750 8.30 4 1785485 5.52 43.77 1487214 4.60 171987 5.32 6 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 Time MCHT MCHT DMPU DMPU Others Others* initial mols Total mols Mol Balance error h Area rnolxlO"11 Area rnolxlO"11 Area mol x10" 1 1 x10" 9 ' x10" 9 % 2 134304 4.15 108360 3.35 476220 14.73 9.82 9.75 0.74 4 32614 1.01 26010 • 0.80 304398 9.41 9.82 9.81 0.09 6 62747 1.94 52718 1.63 357103 11.04 9.82 9.80 0.17 8 20961 0.65 16505 0.51 317330 9.81 9.82 9.66 1.66 10 9214 0.28 6004 0.19 366937 11.35 9.82 9.60 2.28 12 5780 0.18 0 0.00 315983 9.77 9.82 9.26 5.67 * others include unidentified products with small products cracked 222 Appendix C The initial concentration of 4 , 6 - D M D B T = average area of 4 , 6 - D M D B T x response factor of 4 , 6 - D M D B T = 3175554 area x 3.09 x 10"9 mol/area = 9.82x l O - 9 mol Conversion of 4 , 6 - D M D B T after 12 h = [4,6 - DMDBT]I=0 - [4,6 - DMDBT]I=I [4,6 -DMDBT]I=0 _ 9.82-5.39 x 100 9.82 = 45.1 % [ 4,6 - DMDBT] 0 -[4,6 - DMDBT + all products]. n h mol balance, % error = '~° 1J^L x 100 [4,6-DMDBT]l=0 9.82-9.26 9.82 5.67 % x 100 223 Appendix C C.3 Example of repeatability of hydrodesulfurization experiments Table C3 shows the data for 2 runs of hydrodesulfurization of 4 , 6 - D M D B T over Coo.o8Ni2P. The standard error, S.E. calculations were obtained as shown in equations C1-C3 . Table C3 Repeatability of hydrodesulfurization of 4 , 6 - D M D B T using Coo.o8Ni2P Experiment # 4 & 9 Solvent: Dodecane Reactant: 4 , 6 - D M D B T (3000 ppm) Flowrate H 2 : 25 ml/min Catalyst: Co 0 .o8Ni 2P Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K Flow rate (4,6-DMDBT+dodecane): 0.04 ml/min Space Velocity :7.9 x 10"2 mol/h-gcat Time mols 4,6-DMDBT mols 4,6-DMDBT Average S.E. %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 Appendix C Time mols DMBP mols DMBP Average S.E. %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 DMBCHxIO" 1 1 mols DMBCHxIO" 1 1 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 MCHTxIO" 1 1 mols MCHTxIO" 1 1 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 Average S.E. %S .E . Conversion 48.7 47.2 47.95 0.02 2.21 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 of the sample, a j = ^ - C . l where, aj is the sample size and n is the number of sample. The standard error is calculated follows: S.E. n-\ and the percent standard error is: C.2 n / 0 ^ S.E.xlOO % S.E. = C.3 a, 227 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 2 P Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K Flow rate (4,6-DMDBT+dodecane) : 0.04 ml/min Space Velocity :7.9 x 10"2 mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mol% 2 3.95E-09 59.78 4 6.03E-09 38.57 6 6.44E-09 34.39 8 6.94E-09 29.34 10 6.96E-09 29.07 12 6.97E-09 29.03 Avg 29.14 Time, h 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 Flow rate (4,6-DMDBT+dodecane): 0.04 ml/min Space Velocity :7.9 x 102 mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mol% 2 4.69E-09 52.20 4 4.57E-09 53.41 6 3.97E-09 59.57 8 6.18E-09 37.02 10 6.29E-09 35.96 12 6.32E-09 35.61 Avg 36.21 Time, h DMBCH DMBU MCHT DMBP DMDBT Others total Initial mol balance mols 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: CoP Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K Flow rate (4,6-DMDBT+dodecane): 0.04 ml/min Space Velocity :7.9 x 10"2 mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mo 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 Time, h 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 0 . i 6 N i 2 P Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K Flow rate (4,6-DMDBT+dodecane): 0.04 ml/min Space Velocity :7.9 x 10"2 mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mol% 2 6.90E-09 29.70 4 6.55E-09 33.49 6 7.10E-09 27.54 8 7.35E-09 25.17 10 7.55E-09 22.90 12 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 2 : 25 ml/min Catalyst: Co 0 .34Ni 2P Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K Flow rate (4,6-DMDBT+dodecane): 0.04 ml/min Space Velocity :7.9 x 10"2 mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mo l% 2 1.356E-08 30.91 4 1.451E-08 26.04 6 1.478E-08 24.71 8 1.564E-08 20.31 10 1.625E-08 17.18 12 1.672E-08 14.78 Avg 17.42 Time, h 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 2 : 25 ml/min Catalyst: Co 0 .8oNi 2P Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K Flow rate (4,6-DMDBT+dodecane) : 0.04 ml/min Space Velocity :7.9 x 10"2 mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mol% 2 1.75E-08 11.08 4 1.62E-08 17.25 6 1.66E-08 15.67 8 1.72E-08 12.20 10 1.72E-08 12.42 12 1.85E-08 5.79 Avg 10.14 Time, h 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 Flow rate (4,6-DMDBT+dodecane): 0.04 ml/min Space Velocity :7.9 x 10"2 mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mo l% 2 8.164E-09 16.86 4 7.148E-09 27.21 6 5.406E-09 44.95 8 5.124E-09 47.82 10 5.742E-09 41.52 12 6.081 E-09 38.07 Avq 42.47 Time, h 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 1.02E-09 9.50E-09 9.82E-09 3.26 234 Appendix C Experiment #10 Solvent: Dodecane Reactant: 4 , 6 - D M D B T (3000 ppm) Flowrate H 2 : 25 ml/min Catalyst: C o 2 P Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K Flow rate (4,6-DMDBT+dodecane) : 0.04 ml/min Space Velocity :7.9 x 10 mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mol% 2 3.556E-09 63.78 4 3.962E-09 59.65 6 4.696E-09 52.17 8 5.435E-09 44.65 10 5.588E-09 43.09 12 5.954E-09 39.37 Avq 43.87 Time, h 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 2 : 25 ml/min Catalyst: N i 2 P / A l 2 0 3 Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K Flow rate (4,6-DMDBT+dodecane) : 0.04 ml/min Space Velocity :7.9 x 10"2 mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mol% 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 81.29 Time, h DMBCH DMBU MCHT DMBP DMDBT Others total Initial mol balance mols 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 2 : 25 ml/min Catalyst: Coo .4Ni 2P/Al 203 Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K Flow rate (4,6-DMDBT+dodecane): 0.04 ml/min Space Velocity :7.9 x 10"2 mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mo l% 2 2.86E-09 70.92 4 2.45E-09 75.08 6 2.06E-09 79.01 8 1.40E-09 85.78 10 1.30E-09 86.73 12 1.28E-09 86.95 A V G 86.49 Time, h DMBCH DMBU MCHT DMBP DMDBT Others total Initial mol balance mols • 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 Flow rate (4,6-DMDBT+dodecane): 0.04 ml/min Space Velocity :7.9 x 10"2 mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mol% 2 2.96E-09 69.88 4 2.48E-09 74.74 6 1.66E-09 83.07 10 1.51 E-09 84.63 12 1.42E-09 85.52 Avq 84.41 Time, h 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 2 : 25 ml/min Catalyst: Coo .4Ni 2P/Al 203-F Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K Flow rate (4,6-DMDBT+dodecane): 0.04 ml/min Space Velocity :7.9 x 10"2 mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mo l% 2 3.59E-09 63.47 4 1.81 E-09 81.57 6 1.35E-09 86.30 10 1.23E-09 87.46 12 1.22E-09 87.62 A V G 87.13 Time, h 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 2 : 25 ml/min Catalyst: C o o . 4 N i 2 P / M C M Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K Flow rate (4,6-DMDBT+dodecane): 0.04 ml/min ~ Space Velocity :7.9 x 10" mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mo 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 A V G 99.84 Time, h 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 2 : 25 ml/min Catalyst: Pt-Coo .4Ni 2 P/Al 2 0 3 Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K Flow rate (4,6-DMDBT+dodecane): 0.04 ml/min _ Space Velocity :7.9 x 10" mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mo l% 2 6.79E-09 30.85 4 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 97.21 Time, h DMBCH 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 1.12E-10 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 0.13 241 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 Flow rate (4,6-DMDBT+dodecane): 0.04 ml/min >y ' Space Velocity :7.9 x 10 mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mo l% 2 5.30E-09 '46.07 4 4.92E-09 49.92 6 3.73E-09 62.03 8 1.85E-09 81.12 10 6.30E-10 93.59 12 4.62E-10 95.29 A V G 90.00 Time, h DMBCH DMBU MCHT DMBP DMDBT Others total Initial mol balance mols 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 5.59E-11 1.28E-11 5.05E-09 4.62E-10 1.08E-09 9.62E-09 9.82E-09 2.07 242 Appendix C Experiment # 18 Solvent: Dodecane Reactant: 4 , 6 - D M D B T (3000 ppm) Flowrate H 2 : 15 ml/min Catalyst: Pt -Coo .4Ni 2 P/Al 203 Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K Flow rate (4,6-DMDBT+dodecane): 0.156 ml/min Space Velocity : 9.2 x 10"2 mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mol% 2 4.76E-09 51.51 4 3.32E-09 66.24 6 1.86E-09 81.08 8 9.51 E-10 90.32 10 5.28E-10 94.62 12 4.30E-10 95.62 AVG 93.52 Time, h DMBCH DMBU MCHT DMBP DMDBT Others total Initial mol balance mols mols mols mols mols mols mols mols % error 2 3.07E-11 4.91 E-11 1.18E-10 1.42E-10 4.76E-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 2 : 60 ml/min Catalyst: Pt-Coo .4Ni 2P/Al 203 Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K F low rate (4,6-DMDBT+dodecane) : 0.624 ml /min ' Space Velocity : 3.2 x 10"' mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mol% 2 4.96E-09 49.46 4 4.33E-09 55.93 6 4.12E-09 58.00 8 3.74E-09 61.95 10 3.62E-09 63.17 12 3.41 E-09 65.31 A V G 63.48 Time, h DMBCH DMBU MCHT DMBP DMDBT Others total Initial mol balance mols 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 2 : 100 ml/min Catalyst: Pt-Coo .4Ni 2P/Al 203 Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K F low rate (4,6-DMDBT+dodecane) : 1.04 ml/min Space Velocity : 6.2 x 10"' mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mol% 2 8.32E-09 15.26 4 5.53E-09 43.64 6 5.37E-09 45.29 8 5.01 E-09 48.94 10 4.63E-09 52.83 12 4.59E-09 53.28 A V G 51.69 Time, h 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 Flow rate (4,6-DMDBT+dodecane) : 0.04 ml/min Space Velocity :7.9 x 10"2 mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mo l% 2 6.45E-09 34.31 4 5.00E-09 49.08 6 3.83E-09 60.95 8 3.48E-09 64.59 10 1.81 E-09 81.60 12 1.61 E-09 83.56 A V G 76.58 Time, h 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 2 : 25 ml/min Catalyst: Pt-Co 0 .4Ni 2 P/Al 2 O 3 Reaction period: 12 h Pressure : 3.0 M P a Temperature : 533 K F low rate (4,6-DMDBT+dodecane): 0.04 ml/min Space Velocity :7.9 x 10"2 mol/h-gcat 4,6-DMDBT 4,6-DMDBT Time, h mol Conversion mo l% 2 5.49E-09 44.05 4 4.42E-09 54.97 6 •4.07E-09 58.57 8 3.58E-09 63.51 10 3.49E-09 64.48 12 3.41 E-09 65.30 Avg 64.43 Time, h 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 • D M B C H A MCHT • DMBP ODMDBT _— • x * • ii * ft * t - O 2 4 6 8 10 12 Time-on-stream,h, Figure C2 Profile of product ratio using N i 2 P / A l 2 0 3 (A) and C o 0 . 4 N i 2 P / A l 2 O 3 (B) for the H D S of 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 Temp. K SV xlO"2 mol/h-g Conversion 23 Carbazole MoP 583 8.6 54.5 24 Carbazole Nio.oyMoP 583 8.6 90.0 25 Carbazole Ni 0 . i6MoP 583 8.6 89.6 26 Carbazole Ni0.38MoP 583 8.6 88.8 27 Carbazole Nii.nMoP 583 8.6 85.8 28 Carbazole Co0.4Ni2P/Al2O3 583 8.6 82.0 29 Carbazole C00.4N12P/AI2O3-F 583 8.6 94.0 30 Carbazole Co 0.4Ni 2P/MCM 583 8.6 98.0 31 Carbazole Nio.33MoP/Al 20 3 583 8.6 96.0 32 Carbazole M 0 . 3 3 M 0 P / M C M 583 8.6 100 33 Carbazole N10.33M0P (Repeat) 583 8.6 87.5 34 Carbazole Nio .33MoP /Al 2 0 3 523 8.6 69.1 35 Carbazole Nio.33MoP/Al 20 3 523 15.0 44.9 36 Carbazole M0 .33M0P/AI2O3 523 61.0 3.57 37 Carbazole Nio .33MoP /Al 2 0 3 523 8.6 21.1 249 Appendix D D.2 Response Factor and sample calculation of H D N activity Response factor of the carbazole in xylene was obtained by injecting known composition of 0.1 umol of the liquid sample into gas chromatography (GC). Four samples of each fixed composition were injected and the average reading was used to plot the graph of concentration versus the G C area. A s shown in Figure D l , the slope = 6.283 x 10"21 mol/area was used as the response factor. 5.E-14 -| 5.E-14 - y = 6.283E-21x S 4.E-14 - R* = 0.9981 4.E-14 -3.E-14 -3.E-14 -2.E-14 -2.E-14 -1.E-14 -5.E-15 -0.E+00 - 1 1 1 0.E+00 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 of the bicyclohexane ( B C H X ) product in xylene was obtained by injecting known composition of 0.1 umol of the liquid sample into the G C . Four samples of each fixed composition were also injected and the average reading was used to plot the graph of 250 Appendix D concentration versus the G C area. A s shown in Figure D2, the slope = 6.283 x 10 mol/area was used as the response factor. Figure D2 Calibration curve for B C H X used to determine the response factor In calculating the mols, the R F factor of 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 D2 251 Appendix D Table D2 Results of hydrodenitrogenation of carbazole over N10.07M0P Time Carbazole Carbazole BCHX BCHX CPHX CPHX THCZ THCZ h Area mol Area mol Area mol Area 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 mol balance h Area . mol Area mol mol mol %error 2 205218 3.30E-15 0 0 2.27E-14 2.30E-14 1.28 4 109833 1.77E-15 0 2.30E-14 2.30E-14 0.22 6 59644 9.60E-16 0 0 2.17E-14 2.30E-14 5.65 8 61866 9.96E-16 0 0 2.25E-14 2.30E-14 2.40 10 40617 6.54E-16 0 0 2.22E-14 2.30E-14 3.44 12 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 of carbazole = average area of carbazole x response factor of carbazole = 3564514 area x 6.29 x 10"2 1 mol/area = 2.304 x 10"1 4 mol Conversion of 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 mol balance, % error [Carbazole] t = 0 -[Carbazole + all products] t=12h x100 [Carbazole] t = 0 2.304-2.19 2.19 x 100 = 5.00% 253 Appendix D D.3 Repeatability of hydrodenitrogenation experiments Table D3 Repeatability of 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 Flow rate (4,6-DMDBT+dodecane): 0.04 ml/min Space Velocity :8.6 x 10"2 mol/h-gcat Time mols Carbazole mols Carbazole Average S.E. %S .E . h run 1 run 2 mols mol 2 1.64E-14 1.76E-14 1.70E-14 0.05 4.99 4 8.28E-15 7.97E-15 8.13E-15 0.01 1.28 6 7.07E-15 6.27E-15 6.67E-15 0.03 3.33 8 2.32E-15 1.82E-15 2.07E-15 0.02 2.08 10 2.31E-15 2.01E-15 2.16E-15 0.01 1.25 12 2.28E-15 1.10E-15 1.69E-15 0.05 4.91 Time mols BCHX mols BCHX Average S.E. %S .E . h run 1 run 2 mols mol 2 9.14E-16 9.93E-16 9.54E-16 5.91 E-02 5.91 4 1.05E-14 1.14E-14 1.09E-14 6.04E-02 6.04 6 1.27E-14 1.17E-14 1.22E-14 5.55E-02 5.55 8 1.77E-14 1.66E-14 1.71E-14 4.47E-02 4.47 10 1.77E-14 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 CPHX mols CPHX Average S.E. %S .E . h run 1 run 2 mols mol 2 1.47E-15 1.31E-15 1.39E-15 7.83E-02 7.83 4 1.77E-15 1.72E-15 1.74E-15 1.89E-02 1.89 6 3.61 E-16 3.85E-16 3.73E-16 4.51 E-02 4.51 8 3.95E-16 3.60E-16 3.78E-16 6.55E-02 6.55 10 4.62E-16 4.13E-16 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. %S .E . h run 1 run 2 mols mol 2 6.53E-16 6.12E-16 6.33E-16 4.57E-02 4.57 4 6.91E-16 6.15E-16 6.53E-16 8.30E-02 8.30 6 6.77E-16 6.65E-16 6.71 E-16 1.26E-02 1.26 8 6.52E-16 5.91 E-16 6.22E-16 6.89E-02 6.89 10 6.48E-16 6.88E-16 6.68E-16 4.23E-02 4.23 12 6.30E-16 6.22E-16 6.26E-16 8.04E-03 0.80 run 1 Run2 Average S.E. %S .E . Conversion 90.01 87.53 88.77 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 Flow rate (4,6-DMDBT+dodecane) : 0.04 ml/min Space Velocity :8.6 x 10"2 mol/h-gcat Carbazole Carbazole Time, h mol Conversion mo 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 T H C Z <C12 >C12 Total Initial mol h 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 Flow rate (4,6-DMDBT+dodecane) : 0.04 ml/min Space Velocity :8.6 x 10"2 mol/h-gcat Carbazole Carbazole Time, h mol Conversion mol% 2 4.84E-15 78.99 4 3.94E-15 82.88 6 2.55E-15 88.93 10 • • 2.52E-15 89.06 12 2.14E-15 90.70 Avg 89.56 Time BCHX CPHX T H C Z <C12 > C12 Total Initial mol h 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 0 . 3 8 M o P Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K Flow rate (4,6-DMDBT+dodecane): 0.04 ml/min Space Velocity :8.6 x 10"2 mol/h-gcat Carbazole Carbazole Time, h mol Conversion mol% 2 3.89E-15 83.10 4 3.05E-15 86.75 6 2.97E-15 87.11 10 2.53E-15 89.00 12 2.24E-15 90.30 Avg 88.80 Time BCHX CPHX T H C Z <C12 > C12 Total Initial mol h 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 Flow rate (4,6-DMDBT+dodecane): 0.04 ml/min Space Velocity :8.6 x 10"2 mol/h-gcat Carbazole Carbazole Time, h mol Conversion mo 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 T H C Z <C12 >C12 Total Initial mol h 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 2 : 25 ml/min Catalyst: Coo .4Ni 2P/Al 203 Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K F low rate (4,6-DMDBT+dodecane): 0.04 ml/min Space Velocity :8.6 x 10 mol/h-gcat Carbazole Carbazole Time, h mol Conversion mo l% 2 7.47E-15 67.56 4 6.98E-15 69.68 6 6.12E-15 73.44 8 4.84E-15 78.98 10 4.29E-15 81.36 12 3.33E-15 85.55 Avg 81.96 Time BCHX CPHX T H C Z <C12 >C12 . Total , Initial mol h mol mol mol mol mol 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 2P/Al20 3_F Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K Flow rate (4,6-DMDBT+dodecane) : 0.04 ml/min Space Velocity :8.6 x 10" mol/h-gcat Carbazole Carbazole Time, h mol Conversion mol% 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 T H C Z < C12 > C12 Total Initial mol h 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 2 : 25 ml/min Catalyst: C o o . 4 N i 2 P / M C M Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K Flow rate (4,6-DMDBT+dodecane): 0.04 ml/min Space Velocity :8.6 x 10"2 mol/h-gcat Carbazole Carbazole Time, h mol Conversion mo l% 2 3.44E-15 85.05 4 4.13E-15 82.07 6 2.84E-15 87.66 8 9.23E-16 95.99 10 5.75E-16 97.50 12 7.00E-17 99.70 Avq 97.73 Time BCHX CPHX T H C Z <C12 > C12 Total Initial mol h 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: M0 . 3 3 M 0 P / A I 2 O 3 Reaction period: 12 h Pressure : 3.0 M P a Temperature : 583 K Flow rate (4,6-DMDBT+dodecane): 0.04 ml/min Space Velocity :8.6 x 10"2 mol/h-gcat Carbazole Carbazole Time, h mol Conversion mol% 2 9.39E-15 59.25 4 7.25E-15 68.55 6 5.65E-15 75.47 8 1.33E-15 94.22 10 8.34E-16 96.38 12 8.36E-16 96.37 Avg 95.66 Time BCHX CPHX T H C Z <C12 >C12 Total Initial mol h 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 low rate (4,6-DMDBT+dodecane): 0.04 ml/min Space Velocity :8.6 x 10"2 mol/h-gcat Carbazole Carbazole Time, h mol Conversion mol% 2 1.24E-15 94.61 4 0.00E+00 100.00 6 0.00E+00 100.00 10 0.00E+00 100.00 12 0.00E+00 100.00 Avg 100.00 Time BCHX CPHX T H C Z <C12 > C12 Total Initial mol h 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 2 : 25 ml/min Catalyst: N i 0 . 3 3 M o P / M C M Reaction period: 12 h Pressure : 3.0 M P a Temperature : 523 K Flow rate (4,6-DMDBT+dodecane) : 0.04 ml/min Space Velocity : 8.6 x 10" mol/h-gcat Carbazole Carbazole Time, h mol Conversion mol% 2 1.37E-14 40.55 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 T H C Z <C12 > C12 Total Initial mol h 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 2 : 50 ml/min Catalyst: Ni 0 .33MoP/Al 2O3 Reaction period: 12 h Pressure : 3.0 M P a Temperature : 523 K F low rate (4,6-DMDBT+dodecane): 1.77 ml/min Space Velocity :1.5 x 10"' mol/h-gcat Carbazole Carbazole Time, h mol Conversion mo l% 2 1.90E-14 17.41 4 1.80E-14 22.06 6 1.36E-14 41.13 10 1.27E-14 44.94 12 1.18E-14 48.67 Avg 44.91 Time BCHX CPHX T H C Z <C12 > C12 Total Initial mol h 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 low rate (4,6-DMDBT+dodecane): 7.08 ml/min Space Velocity :6.1 x 10"' mol/h-gcat Carbazole Carbazole Time, h mol Conversion mo l% 2 2.22E-14 3.50 4 2.18E-14 5.37 6 2.22E-14 3.49 10 2.22E-14 3.46 12 2.22E-14 3.77 Avg 3.57 Time BCHX CPHX T H C Z <C12 > C12 Total Initial mol h 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 2 : 25 ml/min Catalyst: N i o . 3 3 M o P / A l 2 0 3 Reaction period: 12 h Pressure : 3.0 M P a Temperature : 423 K Flow rate (4,6-DMDBT+dodecane) : 0.04 ml/min Space Velocity :8.62 x 10"' mol/h-gcat Carbazole Carbazole Time, h mol Conversion mol% 2 8.86E-15 61.54 4 1.11E-14 51.84 6 1.80E-14 21.78 10 1.82E-14 21.14 12 1.83E-14 20.39 Avg 21.11 Time BCHX CPHX T H C Z < C12 >C12 Total Initial mol h 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 H D S _ 4 , 6 - D M D B T Parameter Estimation adapted from Applied c Parameter Estimation c Program Adopted by: Ibrahim Inamah A b u c c c P A R A M E T E R E S T I M A T I O N R O U T I N E F O R O D E M O D E L S c Based on Gauss-Newton method with Pseudolnverse and Marquardt's c Modification. Hard B O U N D A R I E S on parameters can be imposed. c Bayesian parameter priors can be used as an option. c c 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) c 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 c or c c W E I G H T E D L E A S T S Q U A R E S Formulation (when IGLS=0) c 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 The User must provide Subroutine M O D E L [that describes the c mathematical model dx/dt=f(x,p)] c and subroutine J X [computes the Jacobeans (df/dx) & (df/dp)] c where x(i), i = l , N X is the State vector. c c The Output vector y(i), i = l , N Y is assumed to correspond to c the first N Y elements of the State vector c c c The Following Variables M U S T be Specified in P A R A M E T E R statement: c c N X Z (=NX) = Number of state variables in the model c N Y Z (=NY) = Number of measured variables in the model c N P A R Z (=NPAR) = Number of unknown parameters c N R U N Z = Greater or equal to the maximum number of runs (experiments) c to be regressed simultaneously c N P Z = Greater or equal to the maximum number of measurements per run c 269 Appendix E c c The Following Variables M U S T be Specified in M a i n Program: c c I G L S = Flag to use Generalized Least Squares (IGLS=1) or Weighted LS(IGLS=0) c F ILEFN = Name of Input file c F I L O U T = Name o f Output file cQ(i), i = l , N Y = Constant weights for each measured variable (Use 1.0 for Least Squares). c I B O U N D = 1 to enable hard constraints on the parameters (0 otherwise) c N S T E P = Maximum number of reduction allowed for Bisection rule (default is 10) c K I F , K O S , K O F = Input file unit, Screen output unit, Output file unit. c N S I G = Approx. Number of Significant digits (Tolerence EPS = 10* *(-NSIG)) c EPSPSI = Min imum Ratio of eigenvalues before the Pseudo-inverse approximation c is used. c EPSMPW = A very small number (used to avoid division by zero). c T O L = Tolerence for local error required by O D E solver (default is 1 .Oe-6) c c — + U S E M S I M S L 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 — Set convergence tolerence & other parameters c 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— Set Filenames and Open the files for I/O c filein = 'DataINl_HDS.texf 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 format ( /20x , 'PARAMETER 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 format( / /20x, 'GENERALIZED 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 of State Variables ( N X ) = ',i3, & /20x,'Number of Output Variables ( N Y ) = ' ,i3, & /20x,'Number of 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 Read PRIOR, M I N & M A X parameter values c read(kif,*) dummyline read(kif,*) (pprior(j),j=l,npar) read(kif,*) dummyline read(kif,*) (vprior(j),j=l,npar) read(kif,*) dummyline read(kif,*) (pmin(j)j=l,npar) read(kif,*) dummyline read(kif,*) (pmax(j),j=l,npar) c c Read N I T E R , IPRINT & EPS_Marquardt c read(kif,*) dummyline read(kif,*) niter, iprint, epsmrq c c Read Number of Runs to be Regressed c 271 Appendix E read(kif,*) dummyline read(kif,*) nrun i f (nrun .gt. nrunz) then write(kos,74) nrun 74 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 75 format(//20x,'Input Data for',i3,' Runs',//) dfr= - npar c c Read Measurements for each Run 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 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) 85 format(//5x/Run:',i2,4x,'NP=',i3,4x, & T0=\g l2 .4 ,4x , 'X(0)= ' ,10gl l .4 / ) ; c 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) 91 format(5x,'Time=',gl2.4,10x,'Y-data(i)=',10gl2.4) 90 continue c do 92 j=l,npar p(j)=ppriorG) 92 continue write(kof,93) (p(j),j=l,npar) 93 format(///2x,'Pprior(j)=',8gl2.5) write(kos,94) (p(j),j=l,npar) 94 format(//2x,'Pprior(j)=',8gl2.5) write(kos,96) (vprior(j),j=l,npar) write(kof,96) (vprior(j),j=l,npar) 96 format(lx,T/VARprior=',8gl2.5//) write(kof,97) (pmin(j),j=l,npar) vvTite(kos,97) (pmin(j),j=l,npar) 97 format(lx,' Pmin(j) - ,8gl2 .5 / / ) write(kof,98) (pmax(j),j=l,npar) vvTite(kos,98) (pmax(j),j=l,npar) 98 format(lx,' PmaxG) =',8gl2.5//) 272 Appendix E close(kif) c c Ma in iteration loop c do 500 nloop=l,niter c 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)=0.0 104 continue do 106 i=l,npar a(i,i)=vprior(i) 106 continue SSEtotal=SSEprior c c Go through each Run 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 Run do 112 j=l ,nx 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 c Integrate ODEs for this Run 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 273 Appendix E c Update Objective function and matrix A , b 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 c Cal l D I V P A G again with INDEX=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 of p(i) do 211 i=l,npar p0(i)=p(i) 211 continue c — & * A d d Bayes influence of 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 j2 ) , j2=l,npar),b(jl) 216 format(l lgl2.4) 218 continue iprint=l end i f c c Decompose matrix A (using D E V C S F from I M S L ) c call devcsf(npar,a,nparz,s,v,nparz) 274 Appendix E c 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) 220 continue 225 continue c c Use Pseudo-inverse (if Cond(A) > 1/epspsi) ipsi=0 do 230 k=l,npar i f (s(k)/s(l) .It. epspsi) then bvs(k)=0.0 ipsi=ipsi + 1 else c Include M A R Q U A R D T ' S modification bvs(k)=bv(k)/(s(k) + epsmrq) end i f 230 continue c write(kos,235) nloop,epsmrq,epspsi,ipsi,conda write(kof,235) nloop,epsmrq,epspsi,ipsi,conda 235 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) 237 format(lx, 'Eigenvalues-,10gl 1.4) c c 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) 240 continue c c Compute new vector p and ||dp|| dpnorm=0.0 do 250 i=l,npar dpnorm=dpnorm + abs(dp(i)) p( i )=p0(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 format(lx, 'PG)byG-N=',8gl2.5) 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 EPS is nonzero i f (epsmrq .gt. s(npar)*0.01) then write(kof,347) epsmrq write(kos,347) epsmrq 347 f o r m a t ( / / l x , ' » » » 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 f o r m a t ( / / l x , ' » » » From now on EPS_Marquardt = 0.0') epsmrq=0.0 goto 500 end i f end i f 276 Appendix E sigma=sqrt((SSEtotal-SSEprior)/dfr) write(kos,285) sigma,dfr write(kof,285) sigma,dfr 285 format(///5x,'++++++ C O N V E R G E D ++++++',/5x,'LS-Sigma=',gl 1.5, & /5x,'Degrees of Freedom=',f4.0,//) write(kos,286) (p(j),j=l,npar) write(kof,286) (p(j),j=T,npar) 286 format(5x,'BestP0)-,8gl2.5) c c — Go and get Standard Deviations & Sigma goto 600 else c 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 NSTEP=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, ODEs 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 Compute N E W value of the Objective function do 320 j=l ,ny c c— Select appropriate weighting factor (GLS 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 of ffeedom=',f4.0) write(kof,586) (pG),j=l,npar) 586 format(/5x,'LastP(j)=',8gl2.5) c c Compute Sigma & Stand. Deviations of the Parameter 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 )*100 j= l , npa 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, ' -— P R O G R A M E N D ',/) write(kos,891) filout 891 format(//lx,60('-'),/5x,'Program O U T P U T stored in f i le : ' , & a24,/lx,60('-'),/) close(kof,status='keep') pause stop end c c M O D E L c c Ordinary Differential Equation (ODE) Model of the form dX7dt=f(X,P) c c where X( i ) , i = l , N X is the vector of state variables c P(i), i=l , N P A R is the vector of unknown parameters c c —+ c c The User must specify the ODEs to be solved c dX(l) /dt = f l ( X ( l ) , . . . X ( N X ) ; P(1),. . .P(NPAR)) c dX(l) /dt = f2 (X( l ) , . . .X(NX) ; P(1),. . .P(NPAR)) c dX(l) /dt = 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 , N P A R c c M O D E L 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 Model 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 c —Jacobian (df/dx) . 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 c Jacobian (df/dp) 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 continue return end c c J X c J A C O B E A N of the c Ordinary Differential Equation (ODE) Model c required by the O D E solver (version for stiff systems) c c + c c The User must specify the Jacobean matrix: c dfdx(i, j)=[df(i)/dXG)] i = l , N X & j = l , N X c c J X 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~ Initialize Jacobian matrix Fx do 20 i=l ,n do 20 j= l ,n fx(i,j)=0.0 20 continue c c Jacobian matrix Fx 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 ,nx jj=ll+j do 50 i=l,nx ii=ll+i fx(ii,jj)=fx(i,j) 50 continue ll=ll+nx 60 continue return end S E T P A R subroutine SETPAR(nn,param) Initialize vector P A R A M for D I V P A G double precision param(nn) do 10 i=l,nn - param(i)=0 10 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