MAR 0 8 1984 Liranit3 MASSACHUSETTS INSTITUTE of TECHNOLOGY

REACTIVITY OF NICKEL PORPHYRINS IN

CATALYTIC HYDRODEMETALLATION

ROBERT ADAMS WARE

B.S., Worcester Polytechnic Institute (1977)

M.S., Cornell University (1979)

Submitted to the Department of Chemical Engineering in Partial Fulfillment of the Requirements for the Degree of

DOCTOR OF SCIENCE

at the

MASSACHUSETTS INSTITUTE OF TECHNOLOGY

September 1983 i,. - r c q

Massachusetts Institute of Tecnology 1983

Signature of Author Department of Chemical Engineering September 15, 1983

Certified by James Wei Thesis Supervisor

Accepted by Robert C. Reid Chairman, Departmental Committee on Graduate Students Archives MASSANHETTS 1 'TUE MAR 0 8 1984 LIRAnIt3 MASSACHUSETTS INSTITUTE OF TECHNOLOGY

DEPARTMENT OF CHEMICAL ENGINEERING

Room number: Cambridge, Massachusetts Telephone: 02139

September 15, 1983 Professor Jack P. Ruina Secretary of the Faculty Massachusetts Institute of Technology Cambridge, MA 02139

Dear Professor Ruina:

In accordance with the regulations of the Faculty, I herewith submit a thesis, entitled, "Reactivity of Nickel Porphyrins in Catalytic Hydrodemetallation," in partial fulfillment of the requirements for the degree of Doctor of Science in Chemical Engineering at the Massachusetts Institute of Technology.

Respectfully submitted,

Robert A. Ware REACTIVITY OF NICKEL PORPHYRINS IN

CATALYTIC HYDRODEMETALLATION

ROBERT ADAMS WARE

Submitted to the Department of Chemical Engineering on September 15, 1983 in partial fulfillment of the requirements for the Degree of Doctor of Science in Chemical Engineering

ABSTRACT

The kinetics and reaction network of hydrodemetallation (HDM) reactions and the pattern of metal deposition have been investigated using a model residuum oil, consisting of nickel porphyrins (Ni- etioporphyrin (Ni-EP), Ni-tetraphenylporphyrin (Ni-TPP), and Ni- tetra(3-methylphenyl)porphyrin (Ni-T3MPP)) dissolved in a clean (N and S free) mineral oil. Reactions were carried out over a commercial CoO-MoO./Y-Al 2 03 catalyst at 285 - 345 *C and 4.24 - 10.44 MPa H2 (600 - 1500 psig). The porphyrins react via sequential pathways, first involving hydrogenation of peripheral double bonds to "activate" the porphyrin for the final hydrogenolysis steps which fragment the ring and deposit the metal on the catalyst. The extent of initial hydrogenation, the complexity of the reaction pathway, and the overall rate limiting step for metal removal are dependent on the porphyrin structure. One route to deposition with Ni-T3MPP and Ni-TPP is through a non-porphyrinic nickel intermediate.

The behavior of a commercial HDS catalyst in promoting HDM reactions is consistent with the dual functionality concept of hydrogenation activity associated with Co and Mo sites and hydrogenolysis (metal deposition) activity associated with acid sites on the support and Mo.

The presence of nitrogen (pyridine) and sulfur (CS 2 , sulfided catalyst) did not change the reaction pathway of Ni-T3MPP during HDM but did alter the hydrogenation/hydrogenolysis reaction selectivity. Pyridine preferentially retarded the hydrogenation steps whereas sulfiding the catalyst selectively enhanced the metal deposition steps.

The feasibility of controlling the Ni-T3MPP hydrogenation/ hydrogenolysis (metal deposition) reaction selectivity was demonstrated by doping the CoMo/Al 20 3 catlyst. Changes as dramatic as a shift in the overall rate limiting step were possible. Alkali (Cs, Na) neutralized catalyst acidity and drastically reduced the metal deposition activity with only a marginal reduction in the high hydrogenation activity. Sulfur and halogens (I, Cl) added in a reducing environment resulted in high metal deposition activity attributed to an increase in Bronsted acidity on the surface. Hydrogenation activity became successively lower (S>I>Cl) and rate limiting with the halogens and was interpreted as arising from a strong interaction between Mo vacancies and the halogen which deactivated hydrogenation sites.

Nickel deposition profiles obtained under diffusion limited conditions reflect the intrinsic reactivity of the nickel porphyrins. At the reactor entrance all profiles were characterized by an internal maxima in the metal deposition pattern, termed M-shape. These maxima shifted to the edge of the catalyst pellets at the end of the reactor bed. The enhanced metal deposition activity of the sulfided catalyst compared to the oxide catalyst was manifested in steeper, more U-shape metal profiles.

Theoretical calculations based on sequential HDM reaction schemes coupled with diffusion successfully interpreted the metal profiles. 2 Diffusivities for the nickel species were on the order of 10-6 cm /sec with the hydrogenated intermediates having diffusion coefficients 2 to 3.5 times larger than the starting porphyrin.

The impact of the variations in Ni-T3MPP reaction selectivity obtained by doping the catalyst was apparent in metal deposition profiles generated under diffusion limited conditions.

Thesis Supervisor: Dr. James Wei Department Head and Warren K. Lewis Professor of Chemical Engineering In Loving Memory of Rosemary J. Wojtowicz

Our life together was painfully short yet during this time we lived life to its fullest, enjoying so much in each others company.

She brought a special joy and happiness into my life that I would have otherwise not known. The beloved memories of what we did and shared together will always have a special place in my heart.

I dedicate this work to Rosie. The example of her accomplishments and the thought of her constant encouragement and companionship have been a continuing source of inspiration for me.

As she now lives, so too shall my memories of her.

"Blessed are the pure in heart: for they shall see God."

Matthew 5:8 ACKNOWLEDGEMENTS

I would first like to express my gratitude to my thesis advisor,

Professor James Wei. His guidance, encouragement, and interest in this project and in my professional development over the past years is sincerely appreciated.

Thanks are also extended to Professors Michael P. Manning and

Charles N. Satterfield for their helpful suggestions during the course of this work and for serving on my thesis committee.

The financial support of the Department of Chemical Engineering and the National Science Foundation is gratefully acknowledged.

Much of this work would not have been possible were it not for the technical assistance and cooperation of others. Special thanks are extended to Drs. Dave Green and Chi Wen Hung of Chevron Research

Company for their contributions of time and talent in serving as the liaisons in the collaborative aspects of this project with Chevron

Research Company. Their assistance along with Mr. Jack Gilmore and

Mr. Mark Meiser of Chevron Research Company in providing the electron microprobe measurements in this thesis is especially appreciated.

Thanks are also extended to Dr. Rene LaPierre and Mr. Paul

Brigandi of Mobil Research and Development Corporation for providing the ammonia desorption characterization of the modified catalysts in the latter stages of this work.

The assistance of Mr. William Dark of Waters Associates in providing technical advice on HPLC and his generosity in providing the liquid chromatographic columns is appreciated.

The discussions on porphyrin chemistry with Dr. Peter Hambright of Howard University and the numerous porphyrin samples he provided were

most beneficial.

The analytical assistance of several colleagues at MIT including

Jim Bentsen of the Chemistry Department for the IR results, Art Lafleur

of the Mass Spectrometry Facility for the HPLC UV-vis scans, and John

Martin of the Surface Analysis Central Facility for the XPS results,

are all acknowledged.

My education at MIT would have been incomplete were it not

for the friendships, lively discussions, and reckless times shared

with my fellow graduate students. To my many labmates, Ian A. Webster,

Shan Hsi Yang, George A. Huff, Harvey G. Stenger, Thomas M. Bartos,

Robert Summerhayes, C. Morris Smith, Margaret Ingalls, Robert T. Hanlon,

and David K. Matsumoto, a sincere thanks for the great memories. The

good times we shared over the past years will be hard to match.

I am pleased to have shared four years as a research colleague

and comrade with Ian Webster. His intellect, intriguing personality,

and Scottish brogue left never a dull moment. I wish him all the best.

To Norman Margolus, thanks for contributing to the great memories

at 13D.

The concern and moral support of many, but most especially The

Wojtowicz Family, Suzanne Strempek, Mary Koss, Peggy Murphy, and

Debbie Luper, at a time of great personal loss and grief in my life

is greatly appreciated. I look forward to maintaining these friendships.

Finally, no words of thanks can express my appreciation to my

parents, Mr. and Mrs. Charles L. Ware, Jr., and my sisters Wendy and

Kathy for their love, understanding, and constant encouragement

throughout my entire education and most especially during the past year.

For all in life you have provided me, I will forever be grateful. - 7 -

TABLE OF CONTENTS

Page

CHAPTER I. INTRODUCTION 16

I.A. Background and Motivation 16

I.B. Characterization of Nickel and Vanadium Compounds in Petroleum 21

I.C. Hydrodemetallation of Petroleum Residua 29

I.D. Model Compound Hydrodemetallation Studies 38

I.E. Thesis Objectives 42

CHAPT ER II. APPARATUS AND EXPERIMENTAL PROCEDURES 45

II.A. Materials 45

II.A.l. Metalloporphyrins 45 II.A.2. Oil 48 II.A.3. Preparing Nickel Porphyrins in Mineral Oil 54 II.A.4. Catalyst 55

II.B. Reactor Design 61

II.C. Reactor Operating Procedure 71

II.D. Analytical Procedures 77

II.D.l. Liquid Samples 77 II.D.2. Catalyst Samples 89

II.E. Initial Transient Behavior of the Catalyst 93

CHAPTER III. PORPHYRIN REACTIVITY IN CATALYTIC HYDRODEMETALLATION 98

III.A. Intrinsic Kinetics on the Oxide Catalyst 98

III.A.l. Ni-Etio 98 III.A.2. Ni-T3MPP 108 III.A.3. Ni-TPP 133

III.B. Discussion of Porphyrin Reactivity 133

III.C. Demetallation Reactions Under Diffusion Limited Conditions 137

III.C.l. Metal Deposition Profiles 137 III.C.2. Metal Profile Model Development and Discussion 141 - 8 -

Page

CHAPTER IV. HYDRODEMETALLATION OF NI-T3MPP IN THE PRESENCE OF NITROGEN AND SULFUR COMPOUNDS 153

IV.A. Demetallation in the Presence of Pyridine 153

IV.B. Demetallation on the Sulfided Catalyst 158

IV.B.l. Reaction Pathway and Intrinsic Kinetics 158 IV.B.2. Metal Deposition Results and Discussion 170

IV.C. Comments on the Influence of Nitrogen and Sulfur Compounds 178

CHAPT ER V. VARIATIONS IN NI-T3MPP REACTION SELECTIVITY ON MODIFIED CoMo/Al2 0 3 CATALYSTS 180

V.A. Introduction 180

V.B. Characterization of Prepared Catalysts 183

V.C. Kinetic Results on Modified Catalysts 193

V.C.l. Initial Transient Results 193 V.C.2. Steady State Kinetic Results 198

V.D. Discussion of Active Sites on Modified Catalysts 214

V.E. Reaction Engineering Implications 221

CHAP TER VI. CONCLUSIONS 229

NOMENCLATURE 236

BIBLIOGRAPHY 238

APPENDIX A. Assessment of Transport Limitations in the Reactors 248

APPENDIX B. Solution of Kinetic Model for Ni-T3MPP Reaction Pathway 255

APPENDIX C. Computer Programs for Kinetic and Metal Profile Calculations 258

APPENDIX D. Experimental Data from Intrinsic Kinetic Runs 277 - 9 -

LIST OF FIGURES

Page

I-1 Nickel and vanadium contents of various crude oils. 19 (Barwise and Whitehead, 1980).

1-2 Porphyrin skeletal structure and metallo-porphyrins 23 representative of those in petroleum. (Yen, 1975)

1-3 Representative metal environments comprising non- 26 porphyrinic fraction of metals in petroleum. (Yen, 1975a, 1974)

1-4 Hypothetical structure of petroleum asphaltene from 28 Speight and Moschopedis (1981) and asphaltene cluster from Yen (1978).

I-5 Relationship between demetallation and desulfurization 31 of various resids and between mean pore diameter and sulfur and vanadium removal. (Ohtsuka, 1977)

1-6 Relationship between vanadium and nickel removal dur- 33 ing desulfurization of petroleum oils. (Ohtsuka, 1977)

1-7 Intraparticle vanadium and nickel concentration 35 profiles of aged residuum hydrotreating catalysts from Sato et al., (1971) and Oxenreiter et al., (1972).

1-8 Intraparticle vanadium, nickel, and iron concentration 36 profiles of aged residuum hydrotreating catalysts from inlet and outlet reactor positions. (Tamm et al., 1981)

II-1 Structure of model nickel compounds. 46

11-2 Pore size distribution for HDS-16A CoMo/Al203 catalyst. 57

11-3 Schematic diagram of flow reactor. 63

11-4 Schematic diagram of 0.52 cm I.D. reactor. 68

11-5 Calculated mole fraction of hydrogen dissolved in 74 n-hexadecane and Nujol at 25*C compared to estimate of Agrawal (1980).

11-6 Ni-etioporphyrin HPLC chromatograms for reference 84 sample and effluent oil sample. - 10 -

11-7 Ni-tetra(3-methylphenyl)porphyrin HILC chromatograms 85 for reference sample and effluent oil sample.

11-8 Transient behavior of catalyst during Ni-T3hTP de- 95 metallation at 60 ppm Ni feed, 345*C, 6.99 MPa H2 (1000 psig) and W/Q = 0.35 g cat hr/ml oil.

III-1 Reaction sequence for Ni-etioporphyrin. 100

111-2 Absorption spectrum of effluent oil sample during 101 demetallation of Ni-etioporphyrin at 317 0 C and 6.99 MPa H 2 (1000 psig); background is xylene.

111-3 Absorption spectra of Ni-etioporphyrin and Ni- 102 etiochlorin; background is hexane.

111-4 Concentration versus contact time results for Ni- 104 etioporphyrin at 27 ppm Ni feed, 343*C, and 6.99 MPa H2 (1000 psig) on the oxide catalyst.

111-5 Arrhenius plots for the first order rate parameters 105 for Ni-etioporphyrin at 6.99 MPa H 2 (1000 psig) on the oxide catalyst. (Agrawal, 1980).

111-6 Hydrogen pressure dependence for the first order rate 107 parameters for Ni-etioporphyrin at 27 ppm Ni feed and 343*C on the oxide catalyst. (Agrawal, 1980)

III--7 Absorption spectrum of effluent oil sample during de- 109 metallation of Ni'-tetra(3-methylphenyl)porphyrin at 345*C and 6.99 MPa H 2 (1000 psig); background is xylene.

111-8 Absorption spectra of Ni-tetra(3-methylphenyl)porphyrin, 111 Ni-T3NP chlorin, and Ni-T3MP isobacteriochlorin; background is hexane.

111-9 Concentration versus contact time results for Ni-T3MPP 112 at 26 ppm Ni feed, 345*C, and 6.99 MPa H2 (1000 psig) on the oxide catalyst.

III-10 Concentration versus contact time results for Ni-T3NPP 113 at 63 ppm Ni feed, 345*C, and 6.99 Ma H2 (1000 psig) on the oxide catalyst.

III-11 Concentration versus contact time results for Ni-T3MPP 114 at 66 ppm Ni feed, 285*C, and 6.99 MPa H2 (1000 psig) on the oxide catalyst.

111-12 Absorption spectra of effluent oil sample at high Ni-T3MPP 116 conversion, "Ni-X" species isolated from this oil sample during HPLC analysis, and Ni-corrin type species from Rasetti (1979); background is hexane. - 11 -

111-13 Mass spectrum of Ni-X. 118

111-14 Infrared spectra of Ni-T3MPP and Ni-X. Samples ref- 119 erenced to tetrachloroethylene.

111-15 Proposed sequence for ring contraction and Ni-X forma- 121 tion.

111-16 Reaction sequence for Ni-tetra(3-methylphenyl)porphyrin. 124

111-17 Arrhenius plots for the first order rate parameters for 128 Ni-T3MPP at 63 ppm Ni feed and 6.99 MPa H2 (1000 psig) on the oxide catalyst.

111-18 Arrhenius plots for the first order rate parameters 129 for Ni-T3MPP at 26 ppm Ni feed and 6.99 MPa H 2 (1000 psig) on the oxide catalyst.

111-19 Hydrogen pressure dependence for the first order rate 132 parameters for Ni-tetra(3-methylphenyl)porphyrin at 63 ppm Ni feed and 345*C on the oxide catalyst.

111-20 Nickel deposition profiles in 1/16" diameter catalyst 139 pellets at various reactor axial positions from Ni- etioporphyrin demetallation at 345*C and 6.99 MPa H2 (1000 psig) on the oxide catalyst.

111-21 Nickel deposition profiles in 1/16" diameter catalyst 140 pellets at various reactor axial positions from Ni- tetra(3-methylphenyl)porphyrin demetallation at 345*C and 6.99 MPa H 2 (1000 psig) on the oxide catalyst.

IV-1 Concentration versus contact time results for Ni-T3MPP 155 at 71 ppm Ni feed, 345*C, and 6.99 MPa H2 (1000 psig) on the oxide catalyst with 100 ppm N as pyridine in feed.

IV-2 Concentration versus contact time results for Ni-T3MPP 159 at 60 ppm Ni feed, 345*C, and 6.99 MPa H2 (1000 psig) on the sulfided catalyst.

IV-3 Concentration versus contact time results for Ni-T3NPP 161 at 62 ppm Ni feed, 285*C, and 6.99 MPa H2 (1000 psig) on the sulfided catalyst.

IV-4 Arrhenius plots for the first order rate parameters for 165 Ni-T3MPP at 60 ppm Ni feed and 6.99 MPa H 2 (1000 psig) on the sulfided catalyst.

IV-5 Hydrogen pressure dependence for the first order rate 169 parameters for Ni-T3NPP at 60 ppm Ni feed and 3450 C on the sulfided catalyst. - 12 -

IV-6 Nickel deposition profiles in 1/16" diameter catalyst 171 pellets at various reactor axial positions from Ni-T3MPP 0 demetallation at 345 C and 6.99 ?Ta H 2 (1000 psig) on the sulfided catalyst.

IV-7 Nickel deposition profiles in 1/16" diameter catalyst 173 pellets at various reactor axial positions from Ni- T3MPP demetallation at 345 0 C and 10.4 NPa H2 (1500 psig) on the sulfided catalyst.

V-1 XPS Spectra of Al 2p level in fresh oxide, pre-iodized, 188 and pre-chlorided catalysts.

V-2 XPS Spectra of Mo 3d and 3d levels in fresh oxide, 189 pre-iodized, and pree lorided3 atalysts.

V-3 XPS Spectra of Co 2p3 / and 2p 2 levels in fresh oxide, 191 pre-iodized, and pre-Lclorided catalysts.

V-4 Ammonia temperature desorption profiles for the fresh 194 oxide and pre-sulfided CoMo/Al203 catalysts.

V-5 Initial transient behavior during in-situ chloriding 197 of the catalyst during demetallation of Ni-T3MPP at 61 ppm Ni feed, 345*C, and 6.99 MPa H 2 (1000 psig) at W/Q = 0.35 g cat hr/ml oil.

V-6 Concentration versus contact time results for Ni-T3MPP 199 at 63 ppm Ni feed, 345*C, and 6.99 MPa H2 (1000 psig) on the cesium doped catalyst.

V-7 Concentration versus contact time results for Ni-T3NPP 200 at 56 ppm Ni feed, 345*C, and 6.99 NTa H2 (1000 psig) on the sodium doped catalyst.

V-8 Concentration versus contact time results for Ni-T3MPP 202 at 63 ppm Ni feed, 345*C, and 6.99 MPa H2 (1000 psig) on the oxide catalyst.

V-9 Concentration versus contact time results for Ni-T3MPP 203 at 60 ppm Ni feed, 345*C, and 6.99 MPa H2 (1000 psig) on the sulfided catalyst.

V-10 Concentration versus contact time results for Ni-T3MPP 205 0 at 63 ppm Ni feed, 345 C, and 6.99 4Pa H 2 (1000 psig) on the iodized catalyst.

V-lb Concentration versus contact time results for Ni-T3MPP 206 at 61 ppm Ni feed, 345'C, and 6.99 MPa H 2 (1000 psig) on the chlorided catalyst. - 13 -

V-12 Concentration versus contact time results for Ni-T3MPP 207 at 58 ppm Ni feed, 3451C, and 6.99 MPa H2 (1000 psig) on the alumina support.

V-13 First order total metal removal plots for Ni-T3MPP at 211 345*C and 6.99 MPa H2 (1000 psig) on the modified catalysts.

V-14 Ni-T3MPP kinetic rate parameters at 60 ppm Ni feed, 212 345*C, and 6.99 MPa H2 (1000 psig) versus electron affinity of dopant on the modified catalysts.

V-15 Calculated nickel deposition profiles in 1/16" diameter 222 catalyst pellets at the reactor entrance for Ni-T3MPP at 345*C and 6.99 MPa H2 (1000 psig) on the modified catalysts.

V-16 Reaction selectivity in Ni-T3MPP demetallation at 345*C 227 and 6.99 MPa H2 (1000 psig) versus electron affinity of dopant on the modified catalysts.

A-1 Demonstration of the absence of intraparticle diffusional 250 limitations in the demetallation of Ni-T3MPP at 345C and 6.99 MPa H2 (1000 psig) on the oxide catalyst. - 14 -

LIST OF TABLES

Page

I-1 Properties of Petroleum Residua. 17

II-1 Porphyrin Solubility in Nujol at 250C. 49

11-2 Nujol Specifications. 51

11-3 Porphyrin Solubility in Petroleum Oils at 25*C. 53

11-4 Chemical and Physical Properties of the HDS-16A Catalyst. 56

11-5 Equipment Description for Figure 11-3. 64

11-6 Calibration Factors for Porphyrinic Species. 81

11-7 High Pressure Liquid Chromatography Parameters. 86

III-1 Ni-etioporphyrin Kinetic Parameters at 343*C and 106 6.99 MPa H2 (1000 psig) on the Oxide Catalyst.

111-2 Ni-tetra(3-methylphenyl)porphyrin Kinetic Parameters 126 0 at 345 C and 6.99 MPa H2 (1000 psig) on the Oxide Catalyst.

111-3 Parameters for Ni-etioporphyrin Metal Profile Model- 147 ling at 345*C and 6.99 MPa H (1000 psig) on the Oxide Catalyst (Figure III-26).

111-4 Parameters for Ni-tetra(3-methylphenyl)porphyrin 148 Modelling at 345*C and 6.99 MPa H (1000 psig) on the Oxide Catalyst (Figure III-215.

IV-1 Ni-tetra(3-methylphenyl)porphyrin Kinetic Parameters 156 in the Presence of Pyridine on the Oxide Catalyst.

IV-2 Ni-tetra(3-methylphenyl)porphyrin Kinetic Parameters 164 on the Sulfided Catalyst.

IV-3 Variation in Ni-tetra(3-methylphenyl)porphyrin Kinetic 167 Parameters on Pre-treated Catalysts.

IV-4 Parameters for Ni-tetra(3-methylphenyl)porphyrin 175 Metal Profile Modelling at 345*C and 6.99 MPa H2 (1000 psig) on the sulfided catalyst (Figure IV-6).

IV-5 Parameters for Ni-tetra(3-methylphenyl)porphyrin Metal 177 Profile Modelling at 345*C and 10.4 MPa H (1500 psig) on the sulfided catalyst (Figure IV-7). - 15 -

V-1 Modified Catalyst Properties. 185

V-2 Temperature Programmed Ammonia Desorption Titration 192 Results.

V-3 Kinetic Rate Parameters for Ni-tetra(3-methylphenyl) 210 porphyrin at 345*C and 6.99 MPa H 2 (1000 psig) on the Modified Catalysts.

V-4 Characteristic Thiele Moduli and Metal Distribution 224 Parameters for Ni-tetra(3-methylphenyl)porphyrin Demetallation on the Modified Catalysts. - 16 -

CHAPTER I

INTRODUCTION

I.A. Background and Motivation

The current supply of crude oil available for refiners is dictating an increase in the utilization of low quality, heavy crudes to meet the demand for lighter, more valuable products such as motor gasoline, diesel fuel, jet fuel, and petrochemical feedstocks. These crudes are characterized by high contents of heteroatoms such as sulfur, nitrogen, oxygen, and trace amounts of metals which are concentrated in the residuum or bottom of the barrel fractions. The presence of these species poses serious environmental hazards in the form of SOx and NOx emissions if these components are present during combustion. Similarly, these heteroatoms act as poisons making for difficult and expensive processing in catalytic conversion steps (catalytic cracking, reform- ing) downstream in the refinery or reduce the quality of the finished petroleum product.

Concentrations of heteroatom species encountered in typical petroleum residua are listed in Table I-1. Of all the elements present in crude oils (nearly one half of the elements in the Periodic Table have been identified, Yen (1975)), the metals, chiefly vanadium and nickel, existing only in trace levels, are the most troublesome.

Depending on the origin of the crude, vanadium is present in w w

TABLE I-1

Properties of Petroleum Residua

Kuwait Venezuela Burzurgan Boscan Khafji California

Residuum Type Atmospheric Vacuum Atmospheric Atmospheric Vacuum Vacuum

Fraction of Crude Vol % 42 21 74 52 78 - - 23

Gravity, API 13.9 5.5 9.6 3.1 5.0 14.4 6.5 43

Sulfur, wt% 4.4 5.4 2.6 6.2 5.9 4.1 5.3 2.3 -.4

Nitrogen, wt% 0.26 0.39 0.61 0.45 0.79 - - 0.98

Nickel, ppm 14 32 94 76 133 37 53 120

Vanadium, ppm 50 102 218 233 1264 89 178 180

Asphaltenes, wt%

(heptane insoluble) 2.4 7.1 9 18.4 15.3 - - -

Source: Speight (1981) - 18 -

concentrations ranging from 8 ppm by weight to 1200 ppm, while nickel varies from 4 ppm to 100 ppm (Yen, 1975). As illustrated in Figure

I-1, vanadium is generally more abundant than nickel.

Trace quantities of metals have likewise been identified in alternative liquid fuel sources. Tar sand bitumen contains 200 - 300 ppm V and 70 - 100 ppm Ni (Towson, 1983) whereas oil shale deposits have metal concentrations ranging between 30 - 200 ppm V and 10 - 50 ppm Ni (Tuttle et al., 1983).

Hydrotreating processes are typically used in petroleum refining to remove these heteroatom species. Under conditions of elevated temperature and pressure, the oil is contacted with hydrogen over a catalyst, usually CoMo/Al2 03 , to remove the heteroatom. Quantitative desulfurization of middle distillates can be achieved at temperatures ranging from 650 - 750'F (350 - 400*C) and hydrogen pressures of

500 - 700 psi. Catalyst lifetimes are generally in the range of 2 to

3 years. Heavy oils and residua usually require more severe hydrodesulfurization conditions to produce low sulfur products for further processing. Temperatures as high as 800*F (425*C) and pressures in excess of 2000 psi are not uncommon (Speight, 1981).

In processing these oils with relatively high metals contents numerous problems are encountered. Conditions suitable for hydrodesulfurization (HDS) and hydrodenitrogenation (HDN) promote conversion of these metal species through hydrodemetallation (HDM) reactions.

Unlike sulfur and nitrogen which are removed as gaseous products

(H2 S, NH3 ), elimination of Ni and V from the oil results in deposition of these metals on the hydrotreating catalysts leading to catalyst - 19 -

A CANADA 9 USA 9 VENEZUELA N USSR o MIDDLE EAST + WESIERN AFRICA 1000- 0 OTHERS VENEZUELA -0(TERTIARY) >. -- S00.

200- ANGOLA (TERTIARY) COLUMBIA EQUADOR 100- (ORIENTE) IRAN, IRAK,KUWAIT - CALIFORNIA VOLGA. URAL so. W. CANADA (MISS. TO L.CRET) TUNIS IA (TERT PER MIAN BASIN > 20

100 --- 100

-- + 7 W.AFRICA: GABON TO ANGOLA 5- -w (POST APTIAN) ^LIBYA CRET

2 /-INDONESIA SMPLES +

2'510 20 50 10o 200 sbo (ppm) W.AFRICA NICKEL L.CRETACEOUS

Figure I-1. Nickel and vanadium contents of various crude oils. (Barwise and Whitehead, 1980) - 20 -

poisoning. Deactivation occurs by direct fouling of active sites and by physical obstruction of the entrance to the fine pores making the more active interior sections of the catalyst inaccessible to feedstock and hydrogen. Tolerance levels for metals deposits on new catalysts have been reported to be 50 to 65 wt% (Ohtsuka, 1977) before unacceptably low activity is attained. Typical catalyst lifetime is on the order of one year when treating metals containing residua. Hence consumption of precious Co(Ni)Mo/Al 2 03 hydrotreating catalysts through irreversible poisoning by metals is a problem.

Speight (1981) estimates hydrotreating a 100 ppm (Ni + V) oil would require 0.13 lb CoMo catalyst/barrel of oil. The current price of

$3.85/lb CoMo catalyst can make catalyst replacement a major cost in resid hydrotreating. In some instances this cost is prohibitively expensive dictating alternative processing of the bottom of the barrel fraction to first reduce the metals content.

In addition to fouling HDS catalysts, Ni + V are detrimental to the performance of fluid catalytic cracking (FCC) catalysts. These metals, especially nickel, decrease selectivity in FCC units producing more coke and light gas at the expense of desirable liquid fuels and must therefore be removed from the feedstock (Venuto and Habib, 1979).

Metals concentrations on FCC catalysts as low as 500 ppm Ni and V

(0.05 wt%) can decrease conversion by as much as 5% requiring catalyst replacement. In order to maintain an acceptable metals loading on the catalyst inventory without excessive replacement, the FCC feed should optimally contain less than 5 ppm Ni and V.

Other important reasons for demetallation of fuel oils are that - 21 -

metal oxides, formed during combustion, corrode and erode turbine blades and furnace linings. Similarly, the emission of these metal oxides to the atmosphere poses a health hazard to humans and plants (Yen,

1975).

In light of the continued decline anticipated in the availability of conventional light crude oil and the need to upgrade the bottom of the barrel, the expansion of hydrotreating capacity of refineries and the development of new processes for treating heavy feedstocks will be required. A recent report issued by the National Petroleum Council projected a 20% increase in hydrotreating and residual conversion capacity over 1982 levels will be required in the United States by the year 1990 (Oil and Gas Journal, 1980). At this time, the combined U.S. residual conversion capacity is anticipated to be 1.8 million barrels per day. Removal of 100 ppm metals from this total capacity will be equivalent to removing 28 tons of metal per day.

I.B. Characterization of Nickel and Vanadium Compounds in Petroleum.

Residua and heavy oils contain trace quantities of nickel and vanadium chelated or complexed to ligands that are completely compatible and soluble within the petroleum. These organometallic compounds can be classified into two groups, the metallo-porphyrins and the non- porphyrinic metals. The ratio of metallo-porphyrins to non-porphyrin metal compounds can vary from 0.1 to 1.0 depending on the source of the petroleum (Speight, 1981, Filby, 1975),

The metallo-porphyrins have been extensively studied due to their utilization as geological indicators for characterizing the age and - 22 -

origin of the petroleum source. The origin of porphyrins in petroleum was first proposed by Treibs (1936) to be derived from chlorophylls in organic plant remains. Subsequent work summarized by Baker and Palmer

(1978) indicates that this is indeed the likely source of porphyrins in petroleum oils. The basis skeletal structure characteristic of the porphyrins is a closed aromatic ring comprised of four pyrrole groups bridged at the a-carbon positions by methine carbon atoms. This structure shown in Figure 1-2 and known as porphine (C2 0H1 4 N4 ) does not occur naturally. Compounds referred to as porphyrins have various substituents replacing the hydrogens at the eight beta-pyrrolic carbon positions and the four methine bridge carbon positions. Due to the high aromaticity of the porphine ring, these structures are very stable.

Metallo-porphyrins are formed by the chelation of a metal ion by

the porphyrin. This involves incorporation of the metal into the center

of the tetra-pyrrole ring with the simultaneous displacement of the

two protons from the pyrrole nitrogen atoms. The prevalence of nickel and vanadium porphyrins in petroleum is a consequence of the stronger

chelating strength of these metals in the porphyrin nucleus compared to

iron, for example (Buchler, 1975). Nickel is present in petroleum as

Ni+2 (Yen, 1978). The molecular radius of nickel enables it to sit in

the plane of the four pyrrole rings comprising the porphyrin. The vanadyl group, with the oxygen protruding out perpendicular to the porphyrin plane, has a slightly larger radius than can be accomodated

0 in the nucleus and lies approximately 0,3 - 0.6 A above the plane

(Yen, 1975a).

The origin of nickel and vanadium in the organo-metallic constituents - 23 -

C2H5 CH3

H / / 3 C ~N N / 2 H5 NH N-

NV.=O N N N HN H3C CH3

C2HS PORPHINE DPEP

C2H5 CH3 H C / C H 3 N N/ 2H5

- / / \?\ N N H3 / CH3

2EP52HS C2H25H

ET IOPORPHYR IN RHODOPORPHYR IN

Figure 1-2. Porphyrin skeletal structure and metallo-porphyrins representative of those in petroleum. (Yen, 1975) - 24 -

of crude oil has been attributed, primarily, to the biological material from where the crude was derived (Filby, 1975). Plant material and marine organisms have nickel concentrations similar to the amount detected in petroleum. The higher concentration of vanadium in petroleum than in living organisms has been attributed to a contribution from the source rock.

The extractable, or free, metal porphyrins isolated in petroleum are mixtures of a homologous series generally covering the range of

C27 - C39. Several investigators have concluded that the majority of metallo-porphyrins in crude oil can be classified into two groups, a

Deoxophylloerythroetioporphyrin (DPEP) series and an Etioporphyrin (Etio) series (Baker and Palmer, 1978, Yen, 1975). These structures are shown in Figure 1-2 as the vanadyl-porphyrins and are differentiated by the cycloalkano ring on the DPEP series. The alkyl substiuents on the macrocyclic ring are only representative and vary both on the pyrrole positions and methine bridges producing the homologous series of these materials detected in petroleum. The ratio of these two series varies all the way from predominately DPEP to cases where the Etio series dominates. The extent of these distributions and the relative ratios are found to be a geological parameter for estimating the age of a petroleum sample. Wide distributions are characteristic of more mature oil. Similarly, high DPEP/Etio ratios are characteristic of relatively new formations. As the age of petroleum increases, the petroporphyrins proceed from DPEP to Etio due to thermal conversion releasing the isocyclic ring strain in DPEP (Baker and Palmer, 1978).

Both DPEP and Etio series complexed with Ni and VO are found. - 25 -

A third and relatively minor constituent in the porphyrin fraction is the Rhodoporphyrin type series also shown in Figure 1-2 (Yen,

1975). These porphyrins are usually found in formations of old age and may contain one or more benzo rings fused to the pyrrole rings.

Porphyrins have also been found in energy sources other than crude oil. These include oil shale (Ekstrom et al., 1983), oil sands

(Alturki et al., 1971) and coal-like gilsonite (Baker and Smith, 1974,

Neel, 1983).

Two references that deal extensively with the structure, physical properties, and chemistry of porphyrins are "The Porphyrins," Vol. I -

VII edited by David Dolphin (1978) and "Porphyrins and Metalloporphyrins" edited by Kevin M. Smith (1975).

The remaining 50 - 90% of metals in petroleum oils not contained in the free porphyrin fraction have been classified as non-porphyrinic metals, comprised of material with poorly characterized structures and properties. Examining the nature of the nickel and vanadium in these structures is crucial as the majority of metals in petroleum are generally contained in this group. Yen (1974, 1975) has postulated that the metals may exist in highly aromatic bound porphyrins, complexed to tetradentates of mixed N,S,0 ligands, or present in the asphaltene structures of crude oil occupying sites bound by heteroatoms in large aromatic sheets. Structures representing these metal environments are depicted in Figure 1-3. Most are similar to the environment in the porphyrin molecule. There is also some evidence to suggest that metals in porphyrinic type structures are present in the high molecular weight fraction of crude oils. Sugihara et al. (1970) sep- - 26 -

0 4,

Figure 1-3. Representative metal environments comprising non-porphyrinic fraction of metals in petroleum. (Yen, 1975a, 1974) - 27 -

arated a number of Ni and VO-etioporphyrins from Boscan asphaltenes and concluded that metallo-porphyrins are associated with high molecular weight asphaltenes through interaction of their extensive pi systems.

The most important non-porphyrinic metal compounds are the asphaltenes as these can often contain the majority of metals in crude. The molecular nature of the asphaltenic fraction has been the subject of extensive investigation (Bunger and Li, 1981) but determining the actual structures of the constituents in this fraction has been difficult. This is in part due to asphaltenes being, by definition, an insoluble fraction that is precipitated generally by 40 volumes of n-pentane. Spectroscopic studies of asphaltenes isolated from petroleum have revealed that these structures are condensed poly- nuclear aromatic ring systems bearing alkyl side groups, naphthenes, and heteroatoms scattered throughout (Speight and Moschopedis, 1981).

A hypothetical structure of a petroleum asphaltene is shown in Figure

1-4. Metal is generally coordinated to the heteroatoms within the structure.

These asphaltene sheets tend to associate through interaction of their extensive pi systems forming micelles shown schematically in

Figure 1-4. The attraction is primarily physical interaction and hydrogen bonding. X-ray diffraction analysis has revealed the size of the sheets within the micelle to be on the order of 20 to 25 A in diameter. When stacked 3 or 4 together, micelles of 10 A or higher in height are not uncommon. Asphaltene clusters isolated from a Kuwait residuum by Hall and Herron (1981) have been reported to be as large 0 as 200 A in size and to range over a broad size distribution. In *0 w%0 0

CH3 Sheet X, y - /, L 8-15A

CH3 tz CH -/S

/ -H H3 dm 3.55-37

0 = 4-8 dy 5.5- 6.0 A

LC14-28 A CH33

Cross SectionalView of on AspholteneModel

CH3 Av\ representsthe zig-zog configurationof a saturated CCH3 carbonchain or loosenet nophthenicrings representsthe edge of flat sheetscondensed aromatic rings (C7 H92N 2 S2 0)3 molwt. 3449

Figure 1-4. Hypothetical structure of petroleum asphaltene (left) from Speight and Moschopedis (1981) and asphaltene cluster (right) from Yen (1978). - 29 -

comparison, the metallo-porphyrin diameter is approximately 12 A.

Molecular weight distributions for asphaltenes have likewise been reported to vary over a wide range. Long (1981) determined molecular weights to range from 500 to 70,000 MW based on distributions measured for four crudes.

I. C. Hydrodemetallation of Petroleum Residua.

The process for the hydrodesulfurization (HDS) and hydrodenitrogenation (HDN) of petroleum residua results in the simultaneous hydrodemetallation (HDM) of metal-bearing molecules. Recent reviews by

Ohtsuka (1977) and Speight (1981) deal extensively with the nature of industrial hydrotreating and the commercial processes currently available. The theses of Hung (1979) and Agrawal (1980) provide further discussion on industrial HDM. These reactions generally require elevated temperatures and pressures due to the stability and refractory character of the heavier heteroatom species in residua. Conditions as severe as 400*C (750*F) and 2000 psi hydrogen are not uncommon.

Most of the catalysts used for hydrotreating are formed by impregnating various transition metals on a high surface area solid support, generally y-A1 203 . Cobalt and molybdenum are the two most common metals used for hydrodesulfurization catalysts. Other metals may be used such as a combination of nickel and molybdenum. This catalyst has a more active hydrogenation function than CoMo/Al203 which is undesirable in HDS but is required, and hence more active, for HDN.

Numerous investigators have examined the relative reactivities of sulfur, nitrogen, and metal-b-earing molecules in petroleum during - 30 -

hydrotreating. Comparison of these measured rates from one study to the next is difficult, however, as catalyst activity is intimately related to chemical characteristics such as feed composition and catalytic metals, and physical characteristics such as pore size distribution of the catalyst. The generally observed relationship between demetallation and desulfurization for many resids shown in

Figure 1-5 is that HDS is faster. This is especially the case for catalysts with relatively small pore diameters (less than 120 A) as shown in Figure 1-5. The larger molecules generally associated with metals are more readily hindered or obstructed from diffusing to the active sites within the catalyst as the pores get smaller. HEN reactions have been reported to be slower than HDM reactions on a large pore diameter (average > 150 A) CoMo/Al203 catalyst (Riley,

1978).

Kinetic studies to elucidate the reaction order of hydrodemetallation reactions have yielded conflicting results. Riley (1978) observed apparent first order removal kinetics for both nickel and vanadium when hydrotreating a Safaniya atmospheric resid. Higher order demetallation kinetics were reported by van Dongen et al.

(1980) as 1.5 order for vanadium removal. Similarly, Oleck and

Sherry (1977) reported a better description of HDM was obtained with second order kinetics for both nickel and vanadium removal from

Lagomedio (Venezuelan) atmospheric resid. All studies were conducted on CoMo/Al203 catalysts. These apparent HDM reaction orders observed to be greater than one have been attributed to the presence of more than one class of metal compound reacting with different - 31 -

100 - ' i '

80 $0 on OU

60- 0

40-

201I I 20 40 60 80 100 DESULFURIZATION. WT %

.o00 -- _100 Feed. Ir. Hy. 410*C 2 90- 100 kg/cm - go 1.0 h- 1000 H, I1. Oi 80- 80;

70--7

ui 60- 60

50- - 50

80 900 12 140 160 I80 Mean Pore Diameter (A)

Figure 1-5. Relationship between demetallation and desulfurization of various resids (top) and between mean pore diameter and sulfur and vanadium removal (bottom). (Ohtsuka, 1977) - 32 -

rates (Oleck and Sherry, 1977, Cecil et al., 1968). Considering the complexity of residuum material, this seems appropriate. The apparent second order kinetics can be interpreted as a combination of two first order rates whereas the 1.5 order kinetics may reflect reaction orders of individual vanadium-bearing species of less than

1.0. This latter situation could arise if the reactant obeyed

Langmuir-Hinshelwood kinetics with strong adsorption on the surface.

Independent of the difficulties encountered in determining true kinetic reaction orders of metal compounds in residua oils, all reported demetallation data indicates that vanadium removal is faster than that of nickel. This is depicted graphically in Figure 1-6.

The vanadium content of a Khafji crude and residuum is reduced simultaneously with the increase in % desulfurization but the nickel is not removed until higher conversions. An interpretation for this behavior was first proposed by Beuther and Schmid (1963) and later by Larson and Beuther (1966) and Shah and Paraskos (1975). They reasoned that since vanadium was present exclusively as vanadyl

(VO+2), the oxygen, oriented perpendicular to the plane of the remaining organic structure, provided an enhanced site for adsorption and reaction on the catalyst due to the electron density associated +2 with oxygen. The chemistry of nickel (Ni ) does not provide it with a similar axial ligand as all its valences are satisfied by planar coordination.

Hydrodemetallation reactions are revealed to be diffusion limited by examination of metal deposition profiles in catalyst pellets obtained from commercial hydroprocessing reactors. Intra- - 33 -

V Ni 0 0 crude oil 80JiiS0C+ <> O a310,C+

- E 60 :: 06 C 0

0 A0

E 40

0 30

U 20 40 60 50 100

% D esulfur izatIon

Figure 1-6. Relationship between vanadium and nickel removal during desulfurization of petroleum oils. (Ohtsuka, 1977) - 34 -

pellet radial metal profiles measured by electron microprobe analysis show vanadium to be concentrated in the outer regions whereas nickel is more uniformly distributed (Sato et al., 1971, Oxenreiter et al.,

1972). Examples of metal profiles in catalysts obtained from industrial reactors are shown in Figure 1-7. Hence the higher intrinsic activity observed for vanadium removal compared to nickel is manifested in sharper vanadium profiles in these pellets.

Analysis of the shape of the metal profiles by classical diffusion and reaction theory enables estimates of metal compound effec-

tive diffusivities to be obtained knowing the intrinsic kinetic rate constants. Sato et al. (1971) calculated the effective diffusivity of vanadium compounds would have to be less than 10% of that for nickel compounds to explain their observed profiles. Such a large difference seems unlikely and may suggest much stronger adsorption of the vanadium compounds due to the polarity of the vanadyl group, relative to nickel, is the reason for the apparent difference in effective diffusivities.

More recent and careful examination of intra-pellet metal deposition profiles reveals that the profiles depend on the axial location of the catalyst within the packed bed reactor. Data recently reported by Tamm et al. (1981) and shown in Figure 1-8 show that the nickel and vanadium profiles exhibit an internal maxima at the reactor inlet which shifts to the pellets' edge at the bed outlet.

Hardin et al. (1978) have likewise reported metal profiles with internal maxima. The position of these internal maxima was deep enough within the pellets to rule out any explanation on the basis of experimental artifact during the microprobe analysis. The sharp Ir wNO

SW0 400 300 2D0 100 0 100 200 300 400 500 600

251-

20 4.5 .

C" ul > 3.0 -S 1 0 Ni --

C-

'- Ni 6 1.5 5 ..-.....--..... 4 -A II I ~I I-: 0 0 I I 2 0 Fractional distance to particle edge

U2 NO 500 400 300 =0 100 0 100 200300 400 500 ;0

DISTANCEFROM CENTEROF CATALYST (MICRONS)AFTER 50 HOURS ON STREAM

Figure 1-7. Intraparticle vanadium and nickel concentration profiles of aged residuum hydrotreating catalysts from Sato et al., (1971) (left) and Oxenreiter et al., (1972) (right). - 36 -

0.12 S. S K 0 * 0

S 0. 08 F . Vanadium

In 0. 0.04F S I n 0 D 2ea#r I I + 40 n eI Nickel * 2 nA A

0.024r

-eCL 0.016F- . I* . Vanadium I ron

0 S 0 a. 0 0.008 0 S Reactor Outlet 0 SN Nickel *.

0 11a Ii i 1.0 0.5 0.6 0.4 0.2 0 Fractional Radius

Figure 1-8. Intraparticle vanadium, nickel, and iron concentration profiles of aged residuum hydrotreating catalysts from inlet and outlet reactor positions. (Tamm et al., 1981)

- - 37 -

iron profiles were thought to be due to deposits from rust and scale, not organometallic compounds. Their removal occurred by a sieving process, thus delineating the edge of the pellet.

Tamm et al. (1981) have interpreted these internal maxima as arising from a demetallation mechanism requiring H2 S. At the inlet of the reactor the gas phase contains little H2S, therefore near the edge of the catalyst the demetallation rate would be low. As the desulfurization reaction proceeds within the particle, the H 2S concentration builds up and the demetallation rate increases. Near the outlet of the reactor the H 2S concentration in the bulk is sufficiently high so the demetallation rate is now maximized at the pellets' edge.

Analysis of the metal profiles at the end of the bed by the standard Thiele modulus method revealed an effective diffusivity for nickel species approximately twice that for vanadium species.

The nature of the metals as they accumulate on the catalyst is still open to debate. They appear to exist primarily as sulfides and to a much lesser extent, as a form (i.e. vanadyl) slightly different than that seen in the oil and metallo-porphyrins (Kwan and Sato, 1970,

Silbernagel, 1979). Silbernagel (1979) has recently shown, by employing NMR and EPR techniques, a common pattern of vanadium deposition is observed, independent of the feed and catalyst support. At low vanadium loadings a vanadyl species dominates whereas at high metals loadings (greater than 10 wt%) the dominant form is a sulfide with composition close to V2 S3. Oxenreiter et al. (1972) had earlier speculated that all deposited metals were sulfides with the formula

V2S3 and NiS. Equilibrium constants published for Ni and V sulfides - 38 -

are consistent with the presence of these species at industrial hydrotreating conditions (McKinley, 1957, DeVries and Jellinek, 1974).

The chemical nature of these metal deposits is not only of importance in determining their deposition density but also because of their inherent catalytic activity. Vanadium sulfides can themselves function as HDS and HDM catalysts and help to maintain some degree of activity on highly loaded CoMo catalysts (Tamm et al., 1981,

Kameyama and Amano, 1982). An alternative to the conventional fixed- bed HDS process has been the suggestion of using slurry reactors that utilize as catalysts the metals originally present in the crude oil

(Turnock, 1976, Gleim and Gatsis, 1965, Bearden and Aldrige, 1981).

I.D. Model Compound Hydrodemetallation Studies.

Despite the importance of hydrodemetallation phenomena and its intimate relationship to HDS and HDN, few attempts have been made to explore the intrinsic reactivity of metal-bearing compounds and the nature of the active catalytic sites responsible for these reactions with model compounds. This is in sharp contrast to the wealth of information existing in the literature on hydrodesulfurization (Gates et al., 1979, Mitchell, 1980, Broderick, 1980, Vrinat, 1983) and hydrodenitrogenation (Katzer and Sivasubramanian, 1979, Bhinde, 1979,

Satterfield and Cocchetto, 1981, Yang, 1982, Satterfield and Yang,

1983) using model compounds.

Pioneering model compound work on hydrodemetallation was conducted by Hung (1979). In this study fractional order kinetics for total metal removal were reported for the demetallation of Ni and - 39 -

VO-etioporphyrins and Ni-tetraphenylporphyrin on an oxide Colo/Al2 03 catalyst operated at industrial conditions (Hung and Wei, 1980). The

fractional order was centered at 0.5 but varied slightly with tempera-

ture and hydrogen pressure. Total metal removal rates were less than

the disappearance of feed porphyrin suggesting the presence of other

metal species during demetallation. Limited runs with mixed nickel

and vanadyl-porphyrins indicated that vanadium suppressed the nickel

removal reactions.

In conjunction with this work, Agrawal (1980) interpreted the

apparent fractional kinetic order to be the result of a sequential

mechanism. The hydrodemetallation of both nickel and vanadyletiopor-

phyrins on oxide CoMo/Al203 was shown to proceed through a hydrogena-

tion step followed by a hydrogenolysis step. Hydrogenation steps had

activation energies on the order of 20 kcal/mole and were first order

in hydrogen pressure. The hydrogenolysis (metal deposition) steps

had higher activation energies, 25-30 kcal/mole, and second order

pressure dependence. Hence it was demonstrated that metal deposition

occurred from a hydrogenated intermediate not originally in the oil.

The intermediate was isolated and identified as a di-hydrogenated

metallo-porphyrin commonly named a metallo-chlorin. This reaction

sequence could explain the internal maxima in metal deposition profiles

at the entrance of the bed measured both by Agrawal (1980) hydrotreat-

ing a model heavy oil and by Tamm et al. (1981) with a real feedstock.

Subsequent to these original metallo-porphyrin investigations,

other model compound work has appeared in the literature relating to

demetallation. This has included non-hydrogenative demetallation by - 40 -

acid oxidation (Reynolds and Dittenhafer, 1980, Gould, 1980), the exam-

ination of adsorption and diffusion phenomena of porphyrins on catalysts

and supports (Galiasso and Morales, 1983, Morales and Galiasso, 1982,

Wiser, 1981) and the impact of porphyrin deposited metals on HDS

(Mitchell and Valero, 1982). Of these studies, Galiasso and Morales

(1983) were the only to investigate metallo-porphyrin demetallation

conversion on CoMo/Al203 catalysts but not in any depth.

Complementing the work of Agrawal, Rankel (1981, 1983) has investigated the reaction pathway of nickel-etioporphyrin and nickel and vanadyl-tetra phenylporphyrins in the presence of H2 and H 2S both under

thermal and catalytic (sulfided CoMo) conditions. In the absence of a

catalyst, reacting with H2S and H2 produced cleaved porphyrin rings, or polypyrrolics, which were identified on the basis of UV-visible

spectral data. H 2S was reported to be more reactive for polypyrrolic formation than H . Vanadylporphyrins were more reactive than their nickel analogous.

Reactions catalyzed by CoMo in the presence of H2 and H 2S at

240*C produced hydrogenated metallo-porphyrins including metallo- chlorins (up to hexa-hydrogenated species) and polypyrrolics. None of the reported intermediates were ever isolated in the course of the investigation. Total metal analysis revealed that V and Ni were not removed from the solution although the metallo-porphyrins were destroyed. At elevated temperature, 340 - 400*C, and pressure, 5 MPa. -

7 MPa H2 , (725 - 1000 psig) demetallation proceeded rapidly. Metal- free porphyrin structures were not observed as intermediates in the catalytic reactions in agreement with the results of Hung (1979). - 41 -

These results led Rankel to postulate there were two routes to

metal removal; a direct extraction by H2 S and a hydrogenative route

which was catalytic. The latter was the dominant route with metallo-

porphyrins.

This work demonstrated, as Agrawal had reported, that the

metallo-porphyrins undergo structural changes involving loss of

aromatic character and cleavage of the ring during demetallation.

Attempts were made to identify these changes in petroporphyrins under-

going thermal treatment with H2S and H . Degradation products postu-

lated to be polypyrrolics were detected by Rankel.

Kameyama et al. (1981, 1982) similarly confirmed the sequential

nature of HDM reactions proposed by Agrawal. Vanadyl-tetraphenyl-

porphyrin reacting thermally using tetralin as a hydrogen donor or

catalytically (sulfided CoMo) in the presence of hydrogen was observed

to demetallate through hydrogenated intermediates detected by liquid

chromatography and mass spectroscopy. The sequence was reported to be

VO-P -+ VO-PH2 - VO-PH6 - PH6 + deposit where P represents the starting porphyrin. Surprisingly, no tetrahydro-

genated metal species was seen. Hydrogenation of VO-PH2 was considered

to be the rate controlling step in the thermal reaction with tetralin.

The presence of a sulfur compound, dibenzothiophene, was found to retard the initial porphyrin hydrogenation step but accelerated the overall de-vanadization rate.

In the studies by both Rankel and Kameyama, little emphasis was placed on examining, quantitatively, the reaction sequence, the intermediates, and temperature and pressure dependencies of the reaction - 42 -

steps. Measurement of total metal and feed porphyrin disappearance rates comprised the kinetic analysis. Hence there is still a need

for more detailed examination on the mechanism of hydrodemetallation,

the relative reactivity of different porphyrin compounds, and the nature of the catalytic sites responsible for the reactions. All of

this information is of vital importance in the development of improved

technology and new processes to utilize heavy residuum oils.

I.E. Thesis Objectives

The aim of the hydrodemetallation research effort at M.I.T. has

been to contribute to a more fundamental understanding of the phenomena

associated with the reaction of metal bearing compounds in residuum

oil. This information can be considered part of the cornerstone on which new and efficient catalysts and processes will be developed to

upgrade heavy residua.

In light of these general goals of the program, the objectives of

this thesis were the following:

1) To develop a more thorough understanding of the HDM reactions

employing model compounds on a clean catalyst. A comprehensive

intrinsic kinetic model for HDM requires the synthesis of

reaction networks, determination of the most likely reactive

sites on the metal compounds, identification of important

reaction intermediates, and analysis of reaction temperature and

pressure dependencies. The HDM reaction pathway proposed by

Agrawal (1980) for the etioporphyrin family will be examined

with respect to a second type of porphyrin family to investigate - 43 -

the generality of sequential HDM reactions and to explore the sensitivity of such to porphyrin structure.

2) To examine the intrinsic reactivity of model compounds under conditions more representative of industrial hydroprocessing.

Prior work has been limited to relatively low feed concentrations

(< 30 ppm metal) and reaction on clean, oxide CoMo catalysts whereas higher metal feed concentrations mixed with heteroatoms

co-react on sulfided catalysts in commercial operation. There has previously been no detailed examination of the hydrodemetal-

lation network of model compounds on sulfided CoMo/Al2 03 catalysts. It is important to verify that structure/reactivity

patterns elucidated in clean environments are applicable elsewhere.

3) To identify the nature of the catalytic sites responsible for

promoting reactions within the demetallation sequence based on

the knowledge of the reaction pathway of model compounds. The

feasibility of modifying specific functionalities of the catalyst

and hence controlling the relative reactivity of different steps,

i.e. the reaction selectivity, during HDM will also be explored.

The impact of the selectivity variations can be examined in terms

of the pattern of metal deposition obtained in catalyst pellets

under diffusion limited reaction conditions. These metal deposi-

tion profiles can be used to confirm the reaction sequence and

are the key to determining the effective utilization of the

catalyst for metals loading and, consequently, catalyst life.

Demonstration of the ability to selectively suppress or enhance - 44 -

one reaction step with respect to another would provide the capability

to potentially control the location of metal deposition during demetallation reactions. - 45 -

CHAPTER II

APPARATUS AND EXPERIMENTAL PROCEDURES

II.A. Materials

II.A.l. Metalloporphyrins

Investigation of the reactivity of metal bearing species representative of those in residuum oils involved the use of well characterized metallo-porphyrins supplied by Midcentury Chemicals (Posen, IL). These model compounds were chosen to simplify the identification of reaction intermediates and to avoid the complications associated with the simultaneous reaction of many unspecified compounds present in heavy oils.

The three porphyrins discussed in this study are shown in Figure II-1.

These compounds consist of a characteristic aromatic core comprised of four pyrrole rings bridges by four methine carbons. The nickel (d8 configuration) is situated in the center of the structure in a square planar configuration complexed to the four nitrogen atoms. The filled d orbitals of Ni are of the correct symmetry to overlap with the por- * phyrin r orbitals providing back bonding and greater stability to the molecule (Buchler, 1978). The four pyrrole rings lie in the same plane as the nickel atom with the four bridging carbon atoms slightly puckered out of this plane (Hoard, 1975). The fully unsaturated porphyrin macrocycle contains 11 conjugated double bonds and the high aromaticity W wV

MOD E L N I CK E L C0 MPOUND S

N N N i

Ni-ETIOPORPHYRIN Ni-TETRAPHENYLPORPHYRIN Ni-TETRA (3-IETHYLPHENYL) PORPHYRIN (Ni-EP) (Ni-TPP) (Ni-T3MPP)

Figure II-I. Structure of model nickel compounds. - 47 -

of these structures (resonance delocalization energy estimated to be in excess of 400 kcal/mole, Longo et al., 1970) is reflected in their stability.

The Ni-etioporphyrin (Ni-EP) is substituted at the -pyrrolic positions with methyl and ethyl groups and has open methine (meso) positions. The etio type of porphyrins have been identified in crude oil and comprise up to 50% of the metal in the free porphyrin fraction

(Baker and Palmer, 1978). The Ni-tetraphenylporphyrin (Ni-TPP) on the other hand is substituted at the meso bridge positions and has open

-pyrrolic positions. Although not identified in oil, this type of metal environment may be representative of bound porphyrins of higher aromaticity existing in the petroleum asphaltenes (Vaughan et al., 1970).

The limited solubility of Ni-EP and Ni-TPP reported by Hung (1979) in the mineral oil (Nujol) chosen for his demetallation experiments provided a maximum metal loading of 35 ppm by weight of nickel for

Ni-EP and 30 ppm for Ni-TPP which is far short of the several hundred ppm common to real feed stocks. A derivative of Ni-TPP, namely Ni-tetra

(3-methylphenyl) porphyrin (Ni-T3NPP), was found to exhibit much greater solubility in the mineral oil of up to 120 ppm Ni and was used extensively in this study. This porphyrin provided a more representative oil metal loading as well as a substitution pattern on the porphyrin core different from Ni-EP.

In the continuing search to increase the metal loading in the oil, additional solubility tests were performed on other porphyrins. Ni- octaethylporphyrin (ethyl groups substituted at eight -pyrrolic posi-

tions) was found to have a solubility of 56 ppm Ni in Nujol. A sample of a new tetraphenylporphyrin supplied by Professor Peter Hambright of- - 48 -

Howard University (Washington, D.C.) was also tested. This porphyrin was substituted at the meso positions with 4-isopropylphenyl (4IPP) CH groups, C-H3 and was found to have a solubility of 200 ppm C 3 Ni in Nujol. This can be attributed to the greater aliphatic character of the substituents. Just as the solubility of the 3-methylphenyl porphyrin is greater than the 4-methylphenyl porphyrin, so too would an increase in solubility be expected if the isopropyl group was in the

3 position. This porphyrin was not tested however.

The solubility of a third type of porphyrin, Ni-tetrabenzoporphyrin, with open meso positions and fused benzo rings at the beta positions of the four pyrrole rings was also examined. This porphyrin would have provided an interesting structure/reactivity comparison to complement the Ni-Etio and Ni-TPP types but due to its high aromaticity it was extremely insoluble in Nujol (< 1 ppm Ni) and could not be studied extensively.

Table II-1 contains the solubilities of all the nickel porphyrins tested. Several vanadyl porphyrins are also listed for comparison although they were not used in this study.

II.A.2. Oil

The oil used to dissolve the porphyrins in to make the model residuum oil is as critical as the nature of the porphyrins in determining the metal concentration. Hung (1979) and Agrawal (1980) in their model compound work used a commercial mineral oil, Nujol, supplied by

Plough, Inc., (Memphis, TN). Nujol is a mixture of hydrocarbons consisting primarily of naphthenes with some paraffins and isoparaffins. - 49 -

TABLE II-1

Porphyrin Solubility in Nujol at 25*C

Porphyrin ppm Ni (or V)

Ni-etioporphyrin 35

Ni-OEP 56

Ni-TPP 30

Ni-T (3MP)P 120

Ni-T(4IPP)P 200

Ni-tetrabenzoporphyrin <1

VO-etioporphyrin 35

VO-TPP 15

VO-T3MPP 54

Ni-etioporphyrin solubility in doped Nujol, ppm Ni

Nujol w/ Nujol w/ Nujol w/ Nuj ol 0.5 wt% pyrrole 1.0 wt% pyrrole 2.0 wt% benzene

35 35 55 38 - 50 -

The average molecular formula of the oil is C 30H 57. Other pertinent information on the physical properties of the oil is listed in Table

11-2. This oil was chosen by Agrawal (1980) because it met the following important criteria: a) free of sulfur and nitrogen compounds, b) relatively high viscosity, c) liquid at room temperature, d) high boiling point so that at reaction conditions the vapor pressure of the oil is negligible, and e) relatively inert to thermal cracking at reaction conditions. These factors outweighed the drawback that aromatic porphyrins were only slightly soluble in the naphthenic oil.

In the course of this study the Nujol was doped with pyrrole and benzene in low amounts (1-2 wt%) in an effort to raise the aromaticity of the oil and improve the porphyrin solubility. The results listed in

Table II-1 for Ni-Etio porphyrin indicate that only a slight improvement was attained. Higher concentrations of the aromatic additives were avoided because of the greater hydrogen demand by their presence during hydrotreating. This approach was not pursued further.

A second attempt involved using cuts of real petroleum of varying aromatic character supplied by Dr. Chi Wen Hung of Chevron Research

Company (Richmond, CA). These oils, as shown in Table 11-3, provided an increase in the solubility of both Ni-Etio and Ni-T3MPP but the improvement in metal concentration was at the expense of higher sulfur and nitrogen concentrations. The Alaska North Slope Straight Run Diesel

(SRD) and the Hydrotreated Vacuum Gas Oil (HVGO) appeared the most promising. The HVGO was very waxy and therefore difficult to work with.

The high sulfur content (0.92 wt%) of the SRD prohibited its use in low

sulfur, oxide catalyst experiments. Hydrotreating the SRD to lower the

sulfur content resulted in a significant reduction in the Ni-Etio - 51 -

TABLE 11-2

Nujol Specifications

(Courtesy: Plough Inc., Memphis, Tenn.)

Specific gravity @ 25 0C 0.875 to 0.885

Viscosity @ 37.80 C 72 cP.

Viscosity @ 100 0C 7.1 cP.

Flash point, Pensky-Martin open cup 216*C (Typical)

Pour point -32*C (Typical)

Refractive Index @ 200 C 1.48 (Typical)

Net Optical Density 0.100 maximum

U.V. absorption at - (Typical):

275 nm 0.075

295/299 nm 0.150

299 nm up 0.100

Distillation at atmospheric Pressure (Typical):

IBP 358 0 C

50% 4290C

90% 4700C

0 95% 484 C

0 FBP 497 C - 52 -

TABLE 11-2 (continued)

Distillation at 10 mm Pressure (Typical):

IBP :208 0 C

5% : 2400C

10% : 252 0 C

50% : 274 0 C

90% : 3050C

95% : 317 0 C

FBP : 3290C

Range of carbon atoms : C18 36 Average Formula : C H 30 57 Average Molecular weight 417

Description, solubility, Acidity or alkalinity, readily carbonizable substances, solid paraffin, sulfur compounds, and cloud point all passes U.S.P.

test. - 53 -

TABLE 11-3

Porphyrin Solubility in Petroleum Oils at 25 0 C

Oil Properties Porphyrin Solubility, ppm

Oil S, wt% N,ppm MW IBP,*F Ni-Etio Ni-T3MPP

Nuj ol <1 417 676 35 120

Alaska North Slope 0.92 345 261 358 80 230 Diesel

Hydrotreated Vacuum 0.037 1045 90 Gas Oil

Hydrotreated Heavy <0.02 <1100 279 257 34 Coker Gas Oil *

Hydrotreated North 0.26 18 175 Slope Diesel

*Metals (Ni, V, Fe) less than 2 ppm. - 54 -

solubility to below that of Nujol. The conclusion from these solubility tests was that Nujol was the best oil to use when considering porphyrin solubility, oil stability, and low sulfur and nitrogen contents.

All results reported in this work were obtained using Nujol as the carrier oil.

II.A.3. Preparing Nickel Porphyrins in Mineral Oil

The nickel porphyrins are crystalline solids at room temperature and it is necessary to heat the oil to dissolve the compounds. It was determined by Agrawal (1980) that high temperature exposure of the porphyrins to oxygen in the atmosphere or dissolved in the oil resulted in anomalous metal concentration measurements and irreproducible reaction behavior. It is therefore necessary to eliminate oxygen prior to dis- solving the porphyrin in the oil. The procedure described was con- ceived and developed by Agrawal (1980). The oil was first filtered through 5 micron filter paper (LSWP 047-00, Millipore Corp., Bedford,

MA) into a 2000 ml flask containing a pre-weighed quantity of porphyrin to give the desired concentration of metal in the final solution. The oil was observed to foam during this filtering procedure as the dissolved air was degassed. The oil/porphyrin mixture was then stirred for 1 hour under vacuum and heated to a temperature of 100*C (3.5 setting on

Sybron/Thermolyne Type 1000 stir plate) to remove any residual air. The vacuum was disconnected and helium (99.995% Matheson Gas Products,

Gloucester, MA) at low pressure ~ 5 psig, was flushed into the flask.

Under helium, the oil/porphyrin mixture was heated to 250*C (6.5 setting on hot plate) and stirred for 4 to 6 hours then cooled to room temperature overnight. The porphyrin solution was finally filtered through 0.5 - 55 -

micron filter paper (FHUP 047-00, Millipore, Corp.) to remove any un- dissolved porphyrin solids and was ready for use.

This same procedure scaled down to 25 ml of oil was used when preparing oil mixtures for the porphyrin solubility studies. Solubilities were measured after the oil had cooled down to room temperature, 25*C, and been filtered.

II.A.4. Catalyst

The catalyst used in this study was a commercial CoO-MoO3 /y-A12 03 hydrotreating catalyst, American Cyanamid AERO HDS-16A (Lot# MTG-S-0573), supplied in the oxide form as 1/16 inch extrudates. The chemical and physical properties of this catalyst are listed in Table 11-4. The pore size distribution of this catalyst shown in Figure 11-2 demonstrates the unimodal character of this catalyst with an average pore size of 80.4 A.

American Cyanamid no longer manufacturers HDS-16A and does not recommend it for resid upgrading applications. The newer Cobalt-Molybdenum hydrotreating catalysts are made with modified geometries (Tri-lobe, HDS-20) and bimodal pore size distributions (HDS-1442) to reduce diffusional limitations and pore plugging in the catalysts. These newer catalysts have similar cobalt and molybdenum loadings as HDS-16A. Since part of the objectives of this study was to examine the reactivity of metal depositing species under diffusion-free conditions and not the aging characteristics of the catalyst, it was not deemed necessary to use a newer type of catalyst. The small amount of metal deposited in the course of an experiment did not cause appreciable pore plugging. The narrow unimodal pore size distribution also simplified the modelling aspects of - 56 -

TABLE 11-4

Chemical and Physical Properties of the HDS-16A

Catalyst

(Courtesy - American Cyanamid Company)

(1) Chemical, wt. % dry basis:

Coo : 5.7

: 12.2 MoO3 Na2 0 : 0.03

Fe : 0.04

Base : ALUMINA

(2) Physical:

Apparent bulk density (poured), g/ml : 0.737

Average diameter, cm : 0.152 (1/16 inch)

Average length, cm : 0.432

Average Crush Strength, g : 6350.4

Fines, wt % : 0.2 (-16 mesh)

Pore volume (H2 0), ml/g : 0.4331 0.0044

Pore volume (Nujol), ml/g : 0.4389 0.0060

Pore volume (Hg), ml/g : 0.43

Surface area, m 2/g : 176

Density of a particle, g/ml : 1.49

Pore diameter corresponding to 50% of the

total pore Volume, Angstrom : 80.4 14W ww

MERCURY PRESSURE (PSIA)

0 0 0 0I o 0 0 0 0 00 0 0 Cl~ 0 00 Cn 0 CD 0 0 0 0l 0 0--

0.4 0.4 Mercury Contact Angle 130*

0.3 -10.3

0 IJ

I _- I__M-- 0 0.2 - 0.2

0.1 0.1

0.0 I ~ 0.0 1767.6 442 176.8 117.8 88.4 70.7 59.0 44.2 35.4

Pore Diameter (Angstroms)

size distribution of AERO® HDS-16A Figure 11-2. Pore CoO-MoO 3 /y Al203 catalyst extrudates. - 58 -

the project as only effective micro-pore diffusivities were required.

Kinetic results were obtained on 170-200 mesh (75-88 micron diameter)

catalyst particles (or smaller) which were shown to be free of diffusional

limitations. Intra-pellet nickel deposition profiles measured under

diffisuion limited reaction conditions were obtained using the full size

1/16 inch extrudates. For this application the catalyst particles were hand picked to insure uniformity of size (5-6 nm long) and no defects or

cracks which would contribute to irregularities in the metal deposition

profiles.

The base case, or reference catalyst, was the HDS-16A catalyst in the

oxide form. Catalyst samples were also prepared by modifying this

commercial catalyst with dopants (Cs, Na, S, I, and Cl) of varying

electronic character. Several methods of preparation were used depend-

ing on the dopant added to the catalyst.

The cesium and sodium doped catalysts were prepared by the "dry"

impregnation technique. Solutions of CsNO3 (#20, 215-0, Aldrich Chemical)

and NaNO3 (#3770, J.T. Baker Chemical) were prepared in distilled water to

give solutions that would upon impregnation result in nominal alkali

cation concentrations of 2.0 mmole/g fresh HDS-16A. This corresponded

to a concentration of 0.91 g CsNO 3/ml solution and 0.40 g NaNO 3/ml solu-

tion for the two salts. The poor solubility of CsNO3 in water at 25*C

required that the solution be heated to 100C to dissolve the salt.

The impregnation step was likewise done at 100*C with a quantity of

crushed catalyst preheated to 1000 C. The high solubility of NaNO3 in water enabled the sodium to be impregnated at room temperature, 25*C.

Impregnation was done using catalyst which had been dryed at 4000 C for 24 hours under flowing helium (99.995% Matheson Gas Products) to - 59 -

remove adsorbed water. In both instances a quantity of salt solution just sufficient to fill the pore volume of the catalyst (0.43 ml/g) was added. The mixtures were stirred for 30 minutes and then oven dried at 100 0 C for 1 hour. The dried samples were then calcined in a tubular furnace at 550*C under dry air (Matheson Gas Products) at a flow of

80 cc/min for 8 hours. The calcined catalysts were re-ground and stored in a desiccator prior to use.

Freshly prepared samples were analyzed by Galbraith Laboratories

(Knoxville, TN) for alkali metal content. The resulting concentrations were 22.64 wt% Cs corresponding to 2.20 mmole Cs/g fresh HDS-16A and

4.30 wt% Na corresponding to 1.95 mmole Na/g fresh HDS-16A.

The sulfided catalyst was prepared using a standard procedure recommended by American Cyanamid and also used by Cocchetto (1980) for pre-sulfiding. The reactor, containing 0.5 or 0.7 g of crushed catalyst, was heated to 175*C at a rate of 60*C/hour under 100 ml/min flowing helium (99.995% Matheson Gas Products). The flow was then switched to a 10 mole % H 2S in H2 mixture (Matheson Gas Products) at GHSV = 700 hr~ 1 and approximately 15 psig and maintained for 6 hours at 175*C. The

temperature was then raised to 315*C at 60*C/hr, maintaining the same

H 2S/H2 flow. After one hour at 315*C, the H 2S/H2 flow was stopped and

the catalyst exposed to helium. The reactor temperature and pressure were then adjusted to the desired operating values for the run. This procedure passed more than five times the stoichiometric amount of

sulfur needed for complete sulfiding (MoS2 Co 9S 8) over the catalyst.

Complete sulfiding corresponds to a concentration of 2.40 mmole S/g

fresh HDS-16A. During a run 0.05 wt% CS 2 (#4352, Mallinckrodt Chemical) was added to the feed. Bhinde (1979) has shown this amount is sufficient - 60 -

to stabilize activity by maintaining the catalyst in the sulfided form.

The iodized and chlorided catalysts were prepared by similar procedures. Approximately 0.7 g of crushed oxide catalyst was charged to the reactor and dried at 375'C for 2-3 hours under flowing helium at

100 ml/min. The temperature was then lowered to 345*C (corresponding to the temperature of the kinetic runs) and hydrogen (99.999%, Matheson

Gas Products) saturated with CH 3I (#5-P086, J.T. Baker Chemical) or

CHC13 (#AC-2280, Anachemia) was passed over the catalyst at GHSV =

1000 hr 1. The hydrogen saturator operated at 40 psig and 25*C. This provided a gas phase mole fraction of CH3 1 equivalent to 0.142 and of

CHC13 equivalent to 0.068 assuming ideal gas behavior. The formation of

HI and CH from CH 3I and similarily, HCl and CH from CHC13 are strongly favored thermodynamically under these conditions. Five times the stoichiometric amount of I or Cl required to give complete conversion of the cobalt and molybdenum oxides to the halides was passed over the catalyst. This corresponded to a total iodizing time of 4 hours and a chloriding time of 3.5 hours. The catalyst was flushed with helium prior to the start of oil flow.

Analysis of the freshly iodized and chlorided samples (Galbraith

Laboratories) revealed 19.0 wt% I and 11.5 wt% Cl. This corresponded to 1.85 mmole I/g fresh HDS-16A and 3.6 mmole Cl/g fresh HDS-16A. No source of iodine or chlorine was added to the oil in kinetic runs on these pre-iodized and pre-chlorided catalysts.

A second procedure was also demonstrated to be effective for chloriding the fresh oxide catalyst. This involved chloriding in-situ with CHC1 3 (#AC-2280, Anachemia) in the feed in a Cl/Ni atomic ratio - 61 -

of 10. At 65 ppm Ni this corresponded to 440 ppm CHC1 . A gradual, but distinct, variation in effluent oil composition was observed and signified the transition from the oxide to the chlorided catalyst (See section V.C.1). Stable chlorided catalyst activity was achieved after

100 hours on stream. At the end of the run (375 hours on stream) the spent catalyst was washed with xylene (#X-19, Fisher Scientific Co.) in a soxhlet extractor for 24 hours to remove residual oil. The cleaned catalyst was then dried overnight and analyzed for chlorine (Galbraith

Laboratories). This in-situ treatment resulted in a chlorine content of 6.3 wt% which corresponded to 1.89 mmole/g fresh HDS-16A.

II.B. Reactor Design

The majority of hydroprocessing steps developed by the petroleum industry for hydrotreating (S,N, and metals removal) and hydrocracking are carried out in trickle-bed reactors. These units are fixed bed catalytic reactors where the liquid is allowed to flow down over the bed of catalyst and the gas (usually hydrogen) flows either up or down through the void spaces between the wetted pellets. Satterfield (1975) has reviewed the applications and performance of trickle-bed reactors.

Laboratory scale trickle-bed reactors have also been successfully used to investigate hydrodenitrogenation (Satterfield and Yang, 1983) and hydrodemetallation (van Dongen et al., 1980) phenomena. These reactors can approach plug flow behavior but their performance is very sensitive to liquid channeling, catalyst wetting, and backmixing making for difficult operation on a small scale (de Bruijn, 1976, Montagna and Shah, 1975). To avoid these complications a conventional packed - 62 -

bed flow reactor was used in this investigation.

The reactor system was designed and constructed by Dr. Rakesh Agrawal of Air Products and Chemicals, Inc. (Allentown, PA). This reactor was built to minimize all heat and mass transfer effects. The elimination of such artifacts in laboratory reactors is difficult yet essential to obtaining true intrinsic kinetic data (Doraiswamy and Tajbl, 1974,

Mears, 1971). Calculations demonstrating the absence of transport limitations are presented in Appendix A. A full description of the apparatus has been given by Agrawal (1980) and will only be briefly reviewed here with emphasis on modifications made to improve the ease of operation and the versatility of the equipment. The apparatus will be discussed in three sections: the high pressure gas and oil feed systems, the high temperature and high pressure reactor, and the oil sampling system. A schematic of the apparatus is shown in Figure 11-3.

Table 11-5 is the key for the numbers on the diagram.

1) Gases were supplied by Matheson Gas Products (Gloucester, MA).

Ultra high purity hydrogen (99.999%) and helium (99.995%) cylinders

were connected to both the reactor and to the oil storage autoclave.

Trace oxygen in the hydrogen was converted to water by passing

through a column packed with 0.5 wt% Pt on alumina, 1/8 inch

pellets type M catalyst (Engelhard Industries, Newark, N.J.).

Oxygen in the helium was removed by passing through a column packed

with an equal mixture of crushed 10% CuO/Al 203 (Cu-0803 T) and

ZnO (Zn-0401 E) catalysts, both products of Harshaw Chemical Co.

(Beachwood, OH). Water in both gases was removed by passing through

a column packed with 4A Molecular Sieves (Union Carbide Corporation,

Linde Division, South Plainfield, N.J.). A mixture of 10 mole% H2 S V vo ww 0

I K 18

v-18 18 V-17 15

2 4 6 7

11/ 14 19 21 4 I

ON 1 3 5 912 1 I V-8 L~2j 16 T V-7

8 13

Figure 11-3. Schematic diagram of flow reactor. (Key to numbers in Table 11-5) - 64 -

TABLE 11-5

Equipment Description For Figure 11-3

1 - H 2 S/H2 Cylinder

2 - 3A Molecular Sieves

3 - Helium Cylinder

4 - Cu and Zn Oxides

5 - Hydrogen Cylinder

6 - Pt/AQ 2 03 Catalyst

7 - 4A Molecular Sieves

8 - 2-Liter Autoclave

9 - Oil Charging Port

10 - 1-Liter Feed Vessel

11 - 2-micron Line Filter

12 - 0.5-micron Line Filter

13 - Pump

14 - Reactor and Furnace

15 - Heat Exchanger

16 - Sample Valve

17 - Rupture Disk

18 - Back Pressure Regulator

19 - Liquid/Gas Separator

20 - Oil Collection Burette

21 - 1-Liter Sample Cylinder

V-7 - Pump/Reactor Purge

V-8 - Reactor Inlet

V-17 - Reactor Exit

V-18 - Gas Inlet for Pre-treatment - 65 - in H2 (Matheson Gas Products) was supplied to the reactor for catalyst sulfiding. Water was selectively removed from this gas stream by passing through a column of 3A Molecular Sieves (Union

Carbide Corporation, Linde Division). Both molecular sieve columns were activated by purging with helium at 200*C for 4 hours.

A 2-liter autoclave (Autoclave Engineers, Erie, PA, Model

AFP-2005) was used as the oil storage and hydrogen saturator vessel. Oil was introduced through a 1/4 inch ball valve on top and the head was only removed for periodic cleaning or to clean when switching to a new feed porphyrin. The oil was saturated in the autoclave at room temperature with hydrogen at the desired operating pressure and supplied all required hydrogen for the demetallation reactions. A one-liter bomb (#304-HDF4-1000, Whitey

Co., Highland Heights, OH) was added to use as a second feed vessel which enabled continuous operation of the reactor as the autoclave was being recharged. Oil was first saturated with hydrogen in the autoclave and then transferred under pressure (not

to de-gas any hydrogen) to the one-liter vessel.

Saturated oil from either the two-liter autoclave or the one- liter bomb passed through a 2 micron in-line filter (SS-4TF-2, Nupro

Co., Willoughby, OH) and a 0.5 micron in-line filter (XX45 025 00,

Millipore Corp., filters #FHLP 025 00) before entering the pump.

This was to insure that precipitated porphyrin or other particu-

lates did not enter and damage the feed pump. The 2 micron filter was found to plug with porphyrin deposits over the course of

several runs, especially at the higher metal concentrations. It was possible, however, to remove, clean, and re-install a new

filtering frit within 5 minutes enabling the run to be continued - 66 -

if such a problem did arise. The 0.5 micron filter rarely plugged and was changed only during routine cleaning every six months.

Oil was supplied from the feed vessels to the reactor using a

6000A Solvent Delivery Pump (Waters Associates, Milford, MA). The

pump could operate up to 6000 psig and was modified by Agrawal with

a frequency synthesizer (Model 171, Wavetek, San Diego, CA) con-

nected to the control unit of the pump to provide greater flexi-

bility in output flow rates. The synthesizer provided a 50 ohm

triangular wave (0.5 V peak height) in frequencies from 1.000 Hz

to 1.999 MHz. Control of the output flow from the reciprocating

pump was possible by choosing the appropriate frequency on the

synthesizer according to the following calibration:

flow rate (ml/hr) = 0.2754 X setting (Hz)

based on the density of oil equal to 0.88 g/ml. This calibration

equation was within 1% of that quoted by Waters for their con-

troller.

Oil was introduced at the bottom of the reactors such that they

operated in an up-flow configuration, ensuring complete wetting of

the solids.

It was necessary to routinely (every six months) clean the

pump with chloroform or methylene chloride followed by xylene to

prevent porphyrin deposits from accumulating on the internal parts

and lead to malfunctions.

2) Two flow reactors were used in this study. Kinetic measure-

ments were obtained with a 0.52 cm I.D. 316 SS reactor tube - 67 -

(#15-093, Autoclave Engineers). Runs conducted under diffusion limited conditions to measure intra-pellet metal deposition profiles were performed in a 1.75 cm I.D. 316 SS flow reactor (#15-099,

Autoclave Engineers). Both tubes were 45.0 cm (17.75 inches) long and were pressure rates to 116 MPa (16,000 psig) and 58 MPa

(8400 psig) for the small and larger tubes, respectively, at 425*C.

A schematic drawing of the micro reactor (0.52 cm I.D.) is shown in Figure 11-4. The ends of the tube were covered with a frit of 325X2300 mesh 304 SS wire cloth (Small Parts, Inc., Miami,

FL) to support the glass and catalyst and prevent solids from being carried from the reactor. A plug of glass wool was also placed at the exit of the bed to insure no fine solids were carried away.

A 22 cm pre-heat zone at the entrance of the reactor was comprised of a 12 cm zone of 3 mm diameter pyrex glass beads (#11-312-10A,

Fisher Scientific Company) stacked one upon the other followed by a 10 cm zone of 0.1 mm diameter glass beads (#34007-088, VWR

Scientific), the two being separated by a plug of glass wool. The zone of 3 mm beads was not used by Agrawal but was necessary here

to allow tiny agglomerates of porphyrin to re-dissolve in the oil as it was heated. These particules accumulated in sufficient amounts to plug the entrance of the reactor if 0.1 mm diameter glass was used throughout the pre-heat zone.

The catalyst was diluted with inert pyrex glass (crushed to

170-200 mesh) in a ratio of 1 volume catalyst/ 2 volumes glass.

This helped -to minimize temperature gradients within the reactor and provided an effective catalyst bed length of approximately

10 cm. - 68 -

Se- - glass wool

inert packing

----- outer metal jacket

catalyst and pyrex 0 diluent 0 0 40

0 0 ------__ inert packing 1/32" thermocouple -glass wool

o reactor tube 0 3 mm glass beads 0

- 325 x 2300 mesh screen t Oil Flow

Figure 11-4. Schematic diagram of 0.52 cm I.D. reactor - 69 -

The remaining 13 cm of reactor tube at the exit was packed with

0.1 mm glass beads to provide a post-heating zone to prevent heat losses through conduction.

The larger diameter (1.75 cm I.D.) reactor was packed in a similar fashion but used 3 mm diameter glass beads throughout the pre-heat and post-heat zones and 1 mm glass beads (#34007-168, VWR

Scientific) as the diluent for the full size catalyst pellets in a volume ratio, catalyst/ glass, of 1/2.5.

All glass was demonstrated to have negligible catalytic activity for the demetallation reactions.

The reactor tubes were fitted into larger diameter 316 SS tubing that had been machined to ensure close contact between this tubing and the reactor. This outer tubing served as a heat sink to help maintain isothermality in the reactor and held in position two thermocouples along the wall of the reactor to monitor the temperature. These thermocouples, 0.79 mm diameter chromel-alumel

(Type K) (#SCASS-032U-6, Omega Engineering, Inc., Stamford, CT) were situated approximately 2 cm into the catalyst bed at the entrance and 1 cm from the end of the bed at the exit based on the glass packing and dilution scheme previously described for the

0.52 cm I.D. reactor. The larger diameter reactor was also equipped with a 1.5 mm diameter chromel-alumel thermocouple located inside the reactor tube in the center of the catalyst bed. This was positioned at the same axial location as one on the outer surface and enabled 'verification that radial temperature gradients were not significant. - 70 -

All reactor temperatures were read off a digital temperature indicator (Model 2170A, Omega Engineering, Inc.) connected to the thermocouples.

The reactors were heated by placing either tube assembly into a two zone Lindberg tube furnace (#F8408-1, Scientific Products,

Columbia, MD). The temperature was controlled with a Thermo

Electric temperature controller (Model 3813011128) based on an input signal from an iron-constantan (Type J) thermocopule located in the furnace.

3) The oil sampling and pressure let-down equipment were unchanged from Agrawal's design. Effluent oil from the reactor passed in

1/16 inch SS tubing through a water cooled heat exchanger and on

through a 2 ml sample loop on a chromatographic valve (Model

CV-6-UHPa, Valco Instruments, Houston, TX). Oil samples were

collected by turning the valve handle clockwise to isolate the loop

from the reactor and simultaneously blowing the oil out into a

sample vial using a 10 ml syringe connected to the luer-lock port

on the valve. The loop was then flushed, using the syringe, with

5 ml of hexane and reconnected to the reactor by turning the valve handle counterclockwise to its original position. The collection

lines were flushed of remaining solvent with acetone and then

blown dry. A minimum of 10 ml of oil was allowed to pass through

the loop before another sample was collected.

Effluent oil from the sample valve flowed through a back

pressure regulator (set with helium to control the reactor pressure)

and was depressurized to atmospheric pressure. The oil then passed - 71 -

to a knock-out pot where the oil and gas were separated. Gas is

vented to the hood and the oil is collected in a calibrated burette

where it can be monitored to verify the oil flow rate. A one-liter

sample bomb was connected in parallel to the back pressure regu-

lator to dampen any pressure fluctuations that may arise in the

system.

The potential danger associated with the operation of this high

temperature, high pressure system mandates that certain safety pre-

cautions be taken. The entire unit is located behind a steel

baracade and is connected to the laboratory hood system which

continuously removes any hydrogen that may be present. The furnace

temperature is monitored by an on/off controller (Model 3234211005,

Thermo Electric) that shuts the power off to the temperature con-

troller if a pre-set limit is exceeded. The pressure at the outlet

of the pump is likewise monitored by an on/off controller (Model

0303-X, LFE Corp., Process Control Division, Waltham, MA) which

turns the pump off in the event of an excessive pressure rise.

Finally, a rupture disk rated to 17.5 MPa (2500 psig) is located

just down stream of the sample loop to prevent damage to the

sampling system in the event of pressure build-up from a plugged

line. Safe operation of this equipment is achieved by attentive

monitoring of the system and by recognizing the potential hazards

involved. No accidents were encountered in the course of this

investigation.

II.C. Reactor Operating Procedure

In a typical kinetic run in the 0.52 cm I.D. reactor, 0.7 g of - 72 -

170-200 mesh catalyst was mixed with the appropriate amount of pyrex diluent (2.26 g glass/ g catalyst for V. /V = 2.0) and loaded into in cat the reactor with the inert packing. The sides of the reactor tube were lightly rapped while packing to insure uniform distribution of solids.

Runs conducted under diffusion limited conditions in the 1.75 cm I.D. reactor were typically charged with 5.0 g of the 1/16 inch extrudates diluted with the 1 mm glass beads (5.9 g glass/g catalyst for V. /V = in cat 2.5). The reactor was then mounted in the furnace and pressure tested with helium a minimum of 500 psi above the operating pressure of the run. Included in the pressure test were the lines from the reactor to the one-liter sample bomb downstream of the sample loop. It was necessary to close the valve leading to the back pressure regulator during this test since the device was only rated to 2000 psig and the tests were often at pressures close to or greater than this. The pressurized reactor was generally left overnight to enable detection of slow leaks. The pressure was monitored using the gauge on the sample bomb.

The two-liter autoclave was similarly pressure tested with helium at a pressure 500 psi in excess of operating conditions. The vessel was then depressurized and charged with 1700 ml of porphyrin solution through the top. Helium at low pressure was passed through the autoclave and out the vent while stirring at 200 rpm for 30 min to remove any oxygen present. The gas flow was then switched to hydrogen for the next 30-45 min to purge the helium. The vent on the autoclave was closed and the vessel pressurized to the desired reaction pressure at which time the oil was saturated with hydrogen at room temperature by stirring for

3 hours. The solution was then allowed to sit (generally overnight) - 73 -

without agitation to allow bubbles present in the oil to rise to the surface.

Data on the solubility of hydrogen in Nujol was unavailable but has been estimated using the correlation of Sebastian et al. (1981) with critical properties and solubility parameter for Nujol calculated by

Wojtowicz (1981). The results of these calculations, shown in Figure 11-5, indicate the solubility of hydrogen in Nujol is less than that in n-hexadecane (Cukor and Prausnitz, 1972) but slightly higher than that assumed by Agrawal (1980). For a feed concentration of 65 ppm nickel, it can be calculated from this figure that at 6.99 NPa H2 (1000 psig) the mole ratio of hydrogen to porphyrin in the oil is approximately 100.

This demonstrates the large excess of hydrogen present for the demetallation reactions.

Oil batches containing co-reactants (pyridine, #P5, 750-6, Aldrich

Chemical Co.) or additives to treat the catalyst (carbon disulfide, chloroform) were prepared by introducing these species into the oil with a syringe through a fitting on the top of the autoclave. This occurred after purging the oil with hydrogen but before pressurizing the autoclave to reaction pressure. Calculating the hydrogen demand of these species indicated that sufficient hydrogen was still present in the oil. The amount of carbon disulfide added (0.05 wt%) to maintain the sulfided catalyst was close to the maximum allowable amount based on hydrogen consumption.

The increased solubility of hydrogen in hydrocarbons as temperature is raised (Cukor and Prausnitz, 1972) insured that as the oil was heated in the reactor no bubbles would form due to hydrogen de-gassing. Iw %w 0

0.14

0.12 n-hexadecane o0.1 4- U L- LL 0.08

Nujol cal. - Nujol est. -4 0

10.04

0.02

500 psig 1000 psig 1500 psig 2000 psig

0 2.0 4.0 6.0 8o 10.0 12.0 14.0 16.0 Hydrogen Pressure, MPa H2

Figure 11-5. Calculated mole fraction of hydrogen dissolved in n-hexadecane and Nujol at 250C compared to estimate of Agrawal (1980). - 75 -

With the oil solution ready and the reactor leak-tight, the final preparative step was to heat up and purge the reactor. The reactor was first depressurized to 100 psig and under a helium purge (500 ml/min)

the reactor was heated to 375*C over a three hour period. Gas was introduced at the top of the reactor through valves V-17 and V-18 (see

Figure 11-3) and passed through V-8 and out the purge, V-7. This direction of flow was opposite to that of oil during demetallation

experiments. The catalyst was purged for three hours to remove air and

adsorbed water from the system. If the catalyst was to be pre-sulfided,

the drying temperature was lower so to be compatible with the sulfiding

procedure. Catalyst batches that were pre-chlorided or pre-iodized were

dried at 345*C.

Once the catalyst had been purged, further in-situ treatment

(S, I, Cl addition) proceeded. Catalyst sulfiding was achieved by intro-

ducing the H 2S/H2 mixture directly into the reactor through valves

V-17 and V-18. The gas stream for the pre-chloriding and pre-iodizing

treatment was prepared by bubbling hydrogen up through the bottom of

the one-liter feed bomb (which had been cleaned of oil solution and

filled with 100-200 ml of the appropriate halogenated solvent). This

saturated gas stream was likewise introduced at the top of the reactor.

When the drying or pre-treatment was complete, the reactor temp-

erature was set to the desired value. The reactor was also pressurized

with helium 300-500 psi above the desired operating pressure by setting

the back pressure regulator. This excess pressure was chosen to ensure

that no gas bubbles formed in the reactor. This had no effect on the

hydrogen pressure of the reaction which was determined by the saturating

pressure in the autoclave. - 76 -

Oil flow from the pump was started and 20 ml were pumped out the reference valve on the pump to insure no gas bubbles were in the line.

An oil sample was taken at this time. The reference valve was closed and oil was then pumped out through the purge valve situated before the reactor for 15 minutes at a flow rate of 20 ml/hr. Another sample was taken. These two were used to determine the feed concentration of nickel porphyrin. After closing the purge valve, oil was quickly introduced (20 ml/hr for 15 min for 0.52 cm I.D. reactor; 60 ml/hr for 25 min for 1.75 cm I.D. reactor) to cover the catalyst and fill the reactor. The oil flow was then set to the desired value and the reactor was on stream. The standard start-up conditions for the 0.52 cm

I.D. reactor were 345*C, 6.99 MPa H2 (1000 psig) and 2.0 ml/hr (W/Q =

0.35 g cat hr/ ml for 0.7 g catalyst charge). This flow was usually maintained for 80-100 hours until steady state activity was attained.

Similar steady state activity was also observed when starting at a higher flow of 20 ml/hr but this consumed much more oil.

Typical operating conditions were in the range 285-345*C,

4.24-10.44 MPa H2 (600-1500 psig), and contact times, W/Q, of 0.007 to

0.7 g cat hr/ ml oil. Run duraction for a 1700 ml charge of oil to the autoclave was typically 2 weeks. During this period oil samples for a complete range of contact times up to 90% metal removal were obtained at several temperatures. After a sample at one contact time was taken, the pump was adjusted to a new flow rate. The system was allowed to reach a new steady state, taking 2 to 12 hours depending on the flow rate, before the 'next sample was taken.

The operating procedure was slightly different when using the 1.75 cm

I.D. reactor. These runs were conducted at constant flow and temperature - 77 -

for the entire batch of oil charged to the autoclave. In this way the

metal deposition profiles would maintain a unique shape characteristic

of one contact time and not be masked by the composite deposition

patterns from several flow rates.

At the termination of a run the pump controller and temperature

controller were shut off. Pressure in the reactor was reduced to 500

psig by venting gas from the one-liter sample bomb. Oil was drained

from the reactor while hot by opening the purge valve (V-7) and using

the pressure head in the sample cylinder to force oil out of the reactor

in reverse direction to which it was fed. The reactor was then slowly

purged with helium until it had cooled down to room temperature and it

could be disconnected and cleaned for the next run.

II.D. Analytical Procedures

II.D.l. Liquid Samples

Liquid samples were routinely analyzed by several methods. The

total nickel concentration was determined by atomic absorption spectrophotometry (Perkin Elmer Model 360) whereby the metal atoms were ionized

in an air/acetylene flame and subsequently detected (Sychra et al.,

1981). Oil samples were dissolved in xylene (#X-16, Fisher Scientific), generally 0.2 to 1.0 ml oil in 10 ml solution, to obtain samples in a workable concentration range, 1 to 3 ppm Ni. This solvent was chosen since it both dissolved the oil and burned in a controlled fashion.

Weighing the amount of oil and xylene added provided quantitative analysis.

Calibration standards were prepared using a Conostan 5000 ppm Ni - 78 -

standard (Conostan Division, Conoco, Inc.). To a weighed 50 ml graduated flask, 2 ml of the 5000 ppm Ni standard were added and weighed, and then

2 ml of Conostan Stabilizer were added. The flask was filled with

Nujol and weighed again. This made approximately a 175 ppm Ni stock which was stable for six months. From this stock, 1, 3, and 5 ppm Ni calibration standards were prepared in xylene each time an atomic absorption analysis was performed. The instrument was calibrated to read in ppm Ni and had a linear response in this region. The nickel lamp on the spectrophotometer (Perkin Elmer #303-6047) was operated at

25 m amps. The detector was set at a wavelength of 232.4 nm with a normal slit of 0.2 nm. The air (Airco) was supplied at 40 psig, rotameter set at 37 and the acetylene (Middlesex Welding Supply) was supplied at 12 psig, rotameter set at 13. The instrument was zeroed with respect to xylene. The lower limit of detection for nickel was determined to be 0.05 ppm and sample reproducibility was 2.0%.

The extensive conjugation of the porphyrin macrocyclic ring results in strong absorption bands and characteristic spectra in the visible region (Gouterman, 1978). These absorption peaks, resulting * from ff to ff electron transitions, are distinct for both the starting material and hydrogenated intermediates and may be used in conjunction with Beer's Law for quantitative analysis. UV-visible spectrophotometry

(Bausch and Lomb Spectronic 2000) was performed in the range 290 to

800 nm using the identical samples as had been prepared for atomic

absorption analysis. The use of xylene as the background solvent prevented spectra below 290 nm from being obtained. Samples were loaded

in semi-micro, silica cells (#58016-458, VWR Scientific) with a 10 mm

path length. - 79 -

Characteristic peaks above the Soret band (%400 nm) were chosen to quantify the porphyrin species. For Ni-etioporphyrin (Ni-EP) and

Ni-etiochlorin (Ni-EPH2) the corresponding peaks were 552 nm and 616 nm.

A calibration factor for Ni-EP at this wavelength was determined from several pure porphyrin in xylene samples by measuring the nickel concentration with atomic absorption and correlating with the absorbance at this wavelength. The absorbance constant was determined to be

0.478 absorbance/ppm Ni for Ni-EP at 552 nm. From this an extinction -l -l coefficient of 31,900 M cm was calculated which agrees with the -l -l value of 32,000 M1 cm given by Fuhrhop (1970) for the structurally similar Ni-octaethylporphyrin. For Ni-EPH2, Agrawal (1980) isolated the intermediate by column chromatography and calculated a calibration constant of 0.72 absorbance/ppm Ni at 616 nm. This corresponds to an -l -l extinction coefficient of 48,000 M1 cm , identical to the value reported by Fuhrhop (1970) for Ni-octaethylchlorin.

In a similar manner calibration constants for Ni-tetra (3-methyl phenyl) porphyrin (Ni-P) and is hydrogenated derivatives were determined at their wavelengths of characteristic absorbance. The value for Ni-P, obtained from pure material, was calculated to be 0.272 absorbance/ppm -l -l Ni at 526 nm yielding an extinction coefficient of 18,000 M cm . The -l -l reported value for the similar Ni-TPP at 528 nm is 17,100 M cm

(Dorough et al., 1951). The calibration factor for Ni-tetra (3-methyl-

phenyl) chlorin (Ni-PH2 ) was determined from a purchased sample contain-

ing both Ni-P and Ni-PH . Knowing the calibration factor for Ni-P

enabled through a simple mass balance a value of 0.498 absorbance/ppm

Ni at 616 nm to be calculated for Ni-PH . From this value of 33,200 -1 -1 N cm was determined for the extinction coefficient. No spectra for - 80 -

a Ni-tetraphenylchlorin species was found in the literature; however, the extinction coefficient for Co-tetraphenylchlorin at 618 nm is

27,300 M~lcm~1 (Dorough et al., 1952). The similarity in spectral behavior of Co and Ni porphyrins reported by Thomas and Martell (1958) is also seen here for the hydrogenated derivatives.

The procedure for determining the calibration constant for Ni- tetra (3-methylphenyl) isobacteriochlorin (Ni-PH4 ) was more complex.

The dehydrogenation of this species to Ni-PH2 on alumina (Sidorov, 1965) prevented isolation of a large quantity of sample by a dry chromatography procedure outlined by Agrawal for Ni-EPH . The approach used instead was to take a reacted oil sample containing only Ni-P, Ni-PH 2, and

Ni-PH and selectively oxidize Ni-PH4 to Ni-PH2 using 2,3 dichloro 5,6

dicyano 1,4 benzoquinone (#D6,040-0, Aldrich Chemical Co.) according to

Rousseau and Dolphin (1974). With the known calibration values for

yielded Ni-P and Ni-PH2, a mass balance before and after the oxidation

the calibration factor for Ni-PH This procedure was tried several

times yielding an average value of 0.77 absorbance/ppm Ni at 593 nm for

Ni-PH Once again, the corresponding extinction coefficient of 51,400

M~ cm~l could not be directly compared with a literature value for the -l -l identical compound but was similar to the value of 56,200 M cm re-

ported for Ni-octaethylisobacteriochlorin at 593 nm by Johansen et al.

(1980).

Table 11-6 summarizes the calibration factors for the porphyrinc

species. The similarity in absorbance peak wavelengths, expecially for

the hydrogenated derivatives, reflects the minor affect external sub-

stituents have in altering the optical properties (Gouterman, 1978). - 81 -

TABLE 11-6

Calibration Factors For Porphyrinic Species

Porphyrin Wavelength, nm Calibration Factor, Abs/ppm Ni

Ni-EP 552 0.478

Ni-EPH 616 0.72 2

Ni-P 526 0.272

Ni-PH 616 0.498 2 Ni-PH 593 0.77 4 - 82 -

The ability to use visible spectroscopy on the composite oil to identify reaction products was limited to those not having overlapping absorbance peaks in the visible range. Reaction products, especially porphyrin ring fragments, which showed strong absorbance peaks in the

UV region could not be selectively identified by examination of the composite oil. Separation and identification of the reactor effluent by gas chromatography was not possible due to the complexity of the starting oil (Nujol) and the low volatility of the porphyrin species.

A more appropriate and successful separation method involved the use of high pressure liquid chromatography (HPLC). A discussion of the theory of HPLC can be found in Horvath (1980) and a recent bibliography on applications is in the Journal of Chromatography (1981).

The instrumentation consisted of a Du Pont Model 848 Liquid

Chromatograph equipped with an air-operated hydraulic pump and a UV detector at 254 nm. The detector cell volume was 6.3 pl with an 8 mm path length. Used in series with this detector was a Waters Associates

(Milford, MA) Model 440 Absorbance Detector operating at 405 nm. The cell volume on this detector was 15.5 pl and the path length was 10 mm.

Dual wavelength detection provided differentiation between the porphyrinic and non-porphyrinic peaks. A six-port Rheodyne (Berkeley, CA) 70-IOA valve with pneumatic actuator was used for sample introduction. A

Rheodyne 70-11 Loop Filler Port was used to manually load the sample into

the 50 pl sample loop. Tubing from the column to the 254 nm detector was

stainless steel 1/16 inch 0.D., 0.007 inch I.D. tubing (Supelco, Inc.,

Bellefonte, PA). -Likewise, 1/16 inch 0.D., 0.010 inch I.D. tubing was used from the 254 nm to 405 nm detector.

Isocratic analysis of the Ni-EP and Ni-P oil samples was best - 83 -

achieved using two different column/mobile phase pairs. The Ni-EP system developed was a modification of Johansen et al. (1980) consisting of a p-Porasil (Waters Associates) 3.9 mm I.D. X 30 cm column with 10 pm packing and a mobile phase of hexane (#H-302, Fisher Scientific)/1,2 dimethoxyethane (#0-2430, Fisher Scientific) in a ratio of 100/1 (v/v).

The column was operated at a pressure of 1100 psig corresponding to a mobile phase flow of 3.2 ml/min. Samples for injection were prepared by diluting the oil in the mobile phase in a 1/1 ratio. A typical chromatogram for Ni-EP effluent oil is shown in Figure 11-6 along with the calibration chromatogram for Ni-EP. A sandwich complex of Ni-EP and Ni-EPH2 , determined by the UV-visible spectra, eluted as a single peak after Ni-EP. The formation of this structure by the interaction of two rings has been reported for etio but not tetraphenyl porphyrins

(Tsutsui and Taylor, 1975).

Reaction products from Ni-P (Ni-T3MPP) runs were found to adsorb irreversibly on the p-Porasil column with the hexane/1,2 dimethoxyethane mobile phase and with even stronger mobile phases indicating a less polar column was required. A p-Bondapak-CN (Waters Associates, #84210)

3.9 mm I.D. X 30 cm column with 10 pm packing with a mobile phase of hexane (#H-302, Fisher Scientifid)/absolute ethanol (U.S. Industrial

Chemicals, Co., Tuscola, IL) in a ratio of 400/1 (v/v) was developed for

analysis of these samples. The column was operated at 800 psig (2.5

ml/min flow) and samples were again prepared by diluting the oil with

the mobile phase. A sample chromatogram from analysis of Ni-P oil is

shown in Figure 11-7 along with the calibration chromatogram for Ni-P,

Ni-PH and Ni-PH . Elution times for both Ni-EP and Ni-P reaction

products are listed in Table 11-7. w w 0

uj I

..L z 00~

CL z

- A.. J -% -j

I I I I I I I I I I I |I A _ II III I I I -0 . 8 6 4 2 0 4 2 0 time, min time, min

Figure 11-6. Ni-etioporphyrin HPLC chromatograms using conditions in Table 11-7 for reference sample (right) and effluent oil sample (left). WV ~wV fw

0 .. 2

qd- z X

Z z a IJ

j Ii i i iV I I I I I I Il I I I I I I 6 4 2 0 6 4 2 0 time, min time, min

Figure 11-7. Ni-tetra(3-methylphenyl)porphyrin HPLC chromatograms using conditions in Table 11-7 for reference sample (right) and effluent oil sample (left). - 86 -

TABLE 11-7

High Pressure Liquid Chromatography Parameters

Ni-Etioporphyrin

Column: p-Porasil

Mobil phase: hexane/1,2 dimethoxyethane 100/1 (v/v)

Flow plate: 3.2 ml/min (1100 psig)

Species Elution Time

Nujol peak 1.0 min

Ni-EPH 2.3 2 Ni-EP 2.8

Ni-EP/Ni-EPH 2 complex 3.2

pyrrole 6.4

Ni-tetra (3-methylphenyl) porphyrin

Column: p-Bondapak - CN

Mobil phase: hexane/ethanol 400/1 (v/v)

Flow rate: 2.5 ml/min (800 psig)

Species Elution Time

Nujol peak 1.3 min

xylene 1.4

Ni-P 2.0

Ni-PH 2.2 2 Ni-PH 2,8 4 pyrrole 3.7

Ni-X 4.1 - 87 -

The HPLC systems were developed for qualitative purposes to aid in identification of reaction intermediates rather than for quantitative analysis. In this regard, a rapid scanning UV-visible spectrophotometer

(Hewlett Packard 8450A) was also used in conjunction with the LC analysis. This instrument was located in the Mass Spectrometry Facility.

This provided complete spectra in the 200-700 nm range of individual peaks as they eluted from the column.

Successive oil injections and peak collections enabled sufficient quantities of the key nickel intermediates in the demetallation of

Ni-tetra (3-methylphenyl) porphyrin to be isolated for mass-spec and

infrared analysis. Samples collected off the LC were found to contain

traces of hydrocarbon from the Nujol which did not appear in the UV

spectra but interfered with the other analyses. It was speculated that

hydrocarbon oil bled continuously off the column due to a saturation

effect caused by an initially high loading after each injection and also

due to the complexity of the oil which no doubt contained species with

a wide range of elution times. This problem was overcome by a liquid-

liquid extraction process. A large quantity of oil (100 ml) was mixed

with an equal volume of immiscible liquid, acetonitrile (#AX0142-1, MCB,

Cincinnati, OH) and agitated in an ultrasonic bath for one hour and then

let sit. The top layer (CH3 CN), rich in porphyrinic species, was de-

canted into a beaker. This solution was left for several hours to allow

the remaining oil to settle to the bottom. The top layer was again

decanted. This procedure was repeated twice resulting in a final solu-

tion containing very little oil. This extract was then injected into

the LC and the appropriate peaks collected. Approximately 100 injections

were required to obtain a sufficient amount of material for accurate - 88 -

analysis. The large volume of mobile phase solvent accumulated in collecting these peaks was evaporated in a dry glove box with argon

(Middlesex Welding Supply) purge.

Mass Spectroscopy was performed by Skinner and Sherman Laboratories,

Inc. (Waltham, MA) by the chemical ionization technique using ammonia as the ionizing gas. The detection limit for the parent Ni-tetra (3- methylphenyl) porphyrin was determined to be 100 nanograms. For some unexplained reason, methane was found to be ineffective as the chemical ionizing source on experiments performed with the parent porphyrin.

Solution Infrared Spectroscopy was performed on the isolated nickel intermediates using a Nicolet 7000 Series FTIR located in the Spectro- scopy Laboratory in the Chemistry Department. Tetrachloroethylene (Gold

Lable #15,499-7, Aldrich Chemical Co.) was chosen as the solvent due to its transparency in the IR region above 1200 cm 1. For this analysis the LC collected samples were evaporated to dryness and then re-dissolved in the minimum amount of tetrachloroethylene required for the analysis, approximately 0.2 ml. The samples were loaded into a NaCl cell with

0.2 mm path length and scanned over 2000 times with a resolution of -l 4 cm . The pure component spectra were obtained by referencing the sample spectrum to the solvent spectrum.

HPLC analysis determined that just one nickel intermediate was the dominant species in the oil in addition to the Ni-P, Ni-PH2 , and Ni-PH . A calibration factor for this compound at a characteristic wavelength in the UV-visible region was obtained by correlating the metal content

(determined by atomic absorption) with the peak wavelength absorbance for a sample isolated by LC. It was difficult, however, to use this factor for quantitative analysis of the species in the composite oil due - 89 -

to the overlap of absorption bands of other species at the same wavelength. This species was quantified, instead, by subtracting the Ni-P,

Ni-PH2, and Ni-PH concentrations (determined by UV-visible spectrophotometry) from the total metal concentration (determined by atomic absorption spectrophotometry).

II.D.2. Catalyst Samples

Catalyst samples were analyzed and evaluated by both bulk characterization techniques and surface sensitive methods. With one exception, all analyses were performed to evaluate chemical characteristics or compositions of the catalyst. The one exception being surface area measurements of the self-prepared doped catalysts. This analysis was provided by Dr. Chi Wen Hung of Chevron Research Company (Richmond, CA) using the single point nitrogen B.E.T. method.

Bulk nickel determination on spent, crushed catalyst was measured to check the consistency of the metal mass balance. Verification that inert glass packing used as a diluent in the reactor was not a site for nickel deposition was also obtained. The catalyst or glass sample was dissolved in a hydrofluoric acid (#1-9560, J.T. Baker Chemical)/sulfuric acid (#2468, Mallinckrodt) mixture according to the method of Labrecque

(1976) and the nickel concentration in the resulting solution determined by atomic absorption spectroscopy.

Self-prepared catalysts made by doping HDS-16A with various additives were analyzed by Galbraith Laboratories, Inc. (Knoxville, TN) for determining the bulk concentration of the additive. These catalysts were also analyzed to verify the existence of uniform dopant concentra- - 90 -

tions by use of a scanning electron microscope (SEM) (AMR Model 1000A) equipped with an energy dispersive x-ray analyzer (Tracor Northern Model

TN 2000) located in the Center for Materials Science and Engineering.

This instrument when operated in a line scan mode provided qualitative

information on concentration profiles across the particles. The special mounts required for the catalyst samples were prepared in the Earth and

Planetary Science Department. A one inch diameter phenolic resin disk

(Miller Stephenson Chemical Co. Inc., Danbury, CN) with a 1/4 inch hole drilled through was pressed firmly onto two sided tape on the bottom of

an aluminum weighing dish. Freshly prepared catalyst particles were

then sprinkled lightly into the hole onto the tape. The hold was filled with an epoxy resin comprised of one part Epon Z hardener (brown) and

five parts Epon 828 resin (clear) (Buehler LTD., Evanston, IL). The

epoxy on the mount was cured for 10 minutes at 100*C and then let to

sit overnight to harden. The catalyst mount was then lightly ground

using 400 and 600 grit paper (Buehler LTD.) to reveal catalyst particles

approximately at their mid-section (75 microns in diameter). Only minor

sanding was required. The final polishing was done on an 8 inch diameter

polishing wheel with alumina powder of 0.3 micron size (Mark V Labora-

tory, Inc., East Granby, CN). The finished, polished catalyst samples were washed with distilled water and dried at room temperature. Prior

to SEM analysis, the samples were gold coated by evaporative decorating

and then grounded to the instrument by silver paint and metallic tape

to avoid charging. The instrument was run at a beam energy of 20 kV.

The electron beam; several microns in diameter, sampled to a depth of

1.5 microns. - 91 -

X-ray Photoelectron Spectroscopy (XPS) on fresh and modified catalysts was performed using a Physical Electronics Model 548 Spectrophotometer located in the Center for Materials Science and Engineering. Magnesium

Ka radiation was employed as the excitation source. The digital data was processed through Multiple Analysis Computer System (MACS) supplied by Physical Electronics. The XPS instrument was operated at 10 kV and

40 or 50 m amps and at a pressure of 10-8 torr. The instrument sampled an area several millimeters in diameter to a depth of 30 to 40 atomic layers. The Au 4f7/ 2 peak was set at 84.0 eV and used as a reference for determining binding energies (Stephenson and Binkowski, 1976).

Samples for analysis were prepared by compressing catalyst powder into 1.5 mm thick disks at 10,000 psig in a 13 mm diameter die. The disks were dotted with gold through aluminum foil in an evaporative decorator and dried a minimum of 24 hours in a heated (60*C) vacuum desiccator prior to analysis. Despite all attempts to keep samples dry, outgassing in the spectrophotometer was always a problem.

Surface acidity measurement of the fresh self-prepared doped catalysts were provided through Dr. Rene LaPierre and Mr. Paul Brigandi of Mobil Research and Development Corporation (Paulsboro, N.J.). Temp- erature programmed desorption of NH3 chemisorbed on acid sites was examined using a Du Pont Thermogravimetric Analyzer in conjunction with a Metrohm titration assembly. About 25 mg of catalyst was first heated

0 to 350 C at 10*C/min and held at 350*C for 1 hour under a helium purge of 150 cc/min to dry the sample. After cooling to room temperature,

NH3 adsorption was carried out in-situ by exposing the sample to a gas stream of 0.4 wt% NH3 in helium. The adsorption was continued until the sample reached constant weight. The sample was then purged overnight at - 92 -

room temperature with helium (150 cc/min) to remove physically adsorbed

NH Temperature programmed desorption of the chemisorbed NH3 was carried out in a helium stream (150 cc/min) at a heating rate of 10*C/min up to 600*C. The rate of ammonia evolution was monitored by titrating the effluent with sulfamic acid. The advantage of this method over others available (Forni, 1973, Benesi and Winquist, 1978) is that both

the total acidity and nature of the acid strength distribution can be

quantified. The disadvantage is that Bronsted and Lewis acid sites can not be distinguished.

Radial nickel profiles in the 1/16 inch pellets were measured at

Chevron Research Company (Richmond, CA) using an electron microprobe

(Applied Research Laboratories Model EMX/SM). This analysis was made

possible through the efforts of Dr. Dave Green, Dr. Chi Wen Hung, Mr.

Jack Gilmore, and Mr. Mark Meiser. The catalyst particles examind were

from specified positions along the reactor length. These were obtained

by carefully sliding the entire catalyst bed, intact, from the reactor

tube into a plexiglas tube of identical internal diameter. This tube

had been machined to fit snugly over the reactor and was sliced into

two parts along its axis. The plexiglas tube was held together with

hose clamps as the catalyst bed was pushed from the exit end of the

reactor into the tube. Once into the plastic holder, the clamps were

removed and the tube opened revealing catalyst extrudates at all axial

positions.

The catalyst pellets were washed in xylene (#X-16, Fisher Scientific)

in a soxhlet extractor for 24 hours and dried overnight in an oven at

1000 C. The pellets were mounted in Buehler transoptic resin and ground

down so that the cross-section at midlength was exposed. The final - 93 -

surface polishing was done with 3 micron diamond abrasive paper. All catalyst mounting and preparation was done by Chevron.

The electron beam was focused to a diameter of 10 microns and sampled to a depth of 2 to 3 microns. The standard conditions for the beam were 20 kilovolts and 50 nanoamperes. Nickel profiles were measured by moving the sample continuously beneath the electron beam at a speed of 300 microns/min and recording the x-ray counts in about 20 consecu-

tive 10-second intervals. Counting began at the periphery of the pellet and ended at the center. The instrument was calibrated using nickel on alumina standards and had a lower limit of detection less than 0.1 wt%.

The reported profiles were an average of two separate catalyst pellets. The scatter for individual data points was within 10% of

the average. Analysis of several catalyst pellets across the entire

diameter revealed symmetric profiles indicating that analysis of one

radius was sufficient. One catalyst pellet was mounted longitudinally

(axis parallel to plane of mounting disk) and ground down revealing the

cross-section in the axial direction. Radial profiles measured at two

axial positions demonstrated the reproducibility of the analysis.

II.E. Initial Transient Behavior of the Catalyst

Transients in catalytic activity similar to the behavior reported

by Agrawal (1980) with Ni and VO-etioporphyrins was also observed in

this study with Ni-tetra (3-methylphenyl) porphyrin. This transient was

more pronounced on the former. A fresh charge of catalyst would

initially exhibit a high activity (as measured by total metal conver-

sion), then rapidly deactivate during a relatively short period (30 hours), - 94 -

and finally increase in activity at a more gradual rate. This behavior is presented graphically in Figure 11-8 for the oxide, pre-reduced, and pre-sulfided crushed catalysts. Total metal conversion is plotted versus wt% nickel deposited on the catalyst which is related to time on stream.

The amount of deposited nickel was determined by mass balance, calculating the difference between the inlet and effluent metal concentrations during a run. No catalyst samples were analyzed in this time period. At the end of the run, the catalyst samples were dissolved in acid and the resulting solution analyzed for nickel. The concentration of deposited nickel was within 10% of the value calculated by the difference in total nickel fed to the reactor and that leaving it. This demonstrated the closure of the metal balance and confirmed the relia- bility of the method used for determining the amount of nickel deposited as plotted in Figure 11-8.

The complete metal conversion observed at low metal loadings is a consequence of both the high activity of the fresh virgin catalyst and the removal of feed porphyrin by physical adsorption.

The rapid deactivation period which follows is also typical of the behavior of industrial hydroprocessing catalysts treating real oils.

Beuther et al. (1963, 1980) and Galiasso et al. (1983) have attributed this deactivation to poisoning of the most active sites by coke laydown.

One half of the total carbon deposited in a 16 day run was reported by

Beuther et al. (1963) to accumulate in the first 12 hours on stream.

Tamm et al. (1981) on the other hand interpret this initial deactiva-

tion period to result from monolayer coverage of the fresh catalyst with metal (Ni and V) deposits which possess a level of activity below that w0 w

61 PPM Ni-T3MPP 345 0C 1000 Psim W/Q= 0.35 G CAT. HR ML OIL

1.0-

REDUCED z 0.8 SULFIDED 1

0.6 -- z 00

0 0.4 -

-- 0.2 - OXIDE

I I I I I I 0 0.4 0.8 1.2 1.6 2.0 24 WT% Ni DEPOSITED

Figure 11-8. Transient behavior of catalyst during Ni-T3MPP demetallation at 61 ppm Ni feed, 345 *C, 6.99 MPa H2 (1000 psig) and W/Q = 0.350 g cat hr/ml oil. - 96 -

of the original metals (Co and Mo). The calculated metal loading, 0.4-

0.8 wt% Ni, at the onset of deactivation in this study is far below that for monolayer coverage of the catalyst (14 wt%). This suggests the deactivation is more a consequence of active site poisoning by coke as

Agrawal had speculated rather than by metals deposits. The rate of deactivation, as measured by the slope of the conversion versus weight loading plot, was different for each of the three differently prepared charges of catalyst demonstrating that the observed behavior was not just an adsorption breakthrough effect. The oxide catalyst apparently possess sites most susceptible to coking and poisoning as evidenced by the fastest rate of deactivation.

The third phase of transient behavior observed was a longer, more gradual period of increasing activity following the initial deactivation.

It appears unlikely that this is attributable to an activation of the catalyst by in-situ reduction, for example, as the pre-reduced sample showed the same activity increase. Comparable rates of increase were observed for all catalyst charges. This suggests, as Agrawal (1980) reported, a contribution to the activity of the catalyst from the deposited nickel. The catalytic activity of deposited metals (in the form of their sulfides) has been reported by Tamm et al. (1981) and Rankel and

Rollmann (1983) although these investigators have not observed the auto- catalytic behavior seen here. Sie (1980) has demonstrated the auto- catalytic nature of metal removal reactions but by using bare alumina carriers which gradually acquire metals removal activity when exposed to a residuum feedstock under hydroprocessing conditions, Presumable the complexity of real oils and the simultaneous reaction of heteroatom species impairs the observation of this activity increase on CoMo/Al2 03 - 97 -

catalysts seen with the model system.

The deposition of metal continued throughout the entire run but as shown in Figure 11-8, the increase in catalytic activity became insig- nificant after 1.2 - 1.8 wt% nickel deposited on the catalyst. This corresponded to approximately 120 hours on stream. All kinetic data reported were obtained on catalyst batches operating in this stabilized regime. At the low metal loadings encountered in these runs, no deactivation due to pore mouth plugging was observed. - 98 -

CHAPTER III

PORPHYRIN REACTIVITY IN CATALYTIC

HYDRODEMETALLATION

This chapter examines in detail the hydrodemetallation (HDM) mechanism of a series of nickel porphyrins reacting over an oxide

CoMo/Al203 catalyst. The porphyrins investigated have similar central aromatic cores but different substitution patterns around the periphery.

The influence that slight structural differences have in altering the chemistry of the molecule and in modifying the reaction pathway and overall rate limiting step for metal removal are examined with the aim of establishing structure/reactivity relationships. Reaction condi-

tions, approaching those of industrial hydroprocessing, were varied

(285-3450 C, 4.24-10.09 MPaH 2) to enable the temperature and pressure

dependencies of the demetallation reactions to be assessed.

This information is essential when attempting to interpret, and to

predict the pattern of metal deposition in industrial hydroprocessing

catalysts.

III.A. Intrinsic Kinetics on Oxide Catalyst

III.A.l. Ni-Etio

Hydrodemetallation kinetics for Ni-Etio on the oxide form of

CoMo/Al2 0 3 were first reported by Agrawal (1980). The Ni-porphyrin was - 99 -

found to demetallize via a sequential mechanism through a hydrogenated intermediate. The first reaction step involved reversible hydrogenation of a peripheral double bond in one of the four pyrrole rings comprising the macrocycle to form the Ni-Etio chlorin (Ni-EPH2) as shown in

Figure III-1. The Ni-EPH2 then reacted via a hydrogenolysis step which fragmented the ring and deposited the metal on the catalyst.

With Ni-Etio, the visible spectra of effluent oil shown in Figure

111-2 only revealed absorption peaks which could be assigned to Ni-EP and Ni-EPH2 on the basis of Figure 111-3. Recent HPLC analysis of effluent oil samples also indicated relatively few intermediates. One other major peak in addition to the Ni-EP and Ni-EPH 2 species was present with a retention time similar to pyrrole and no absorption bands above 290 nm in its spectra. This third component was presumably the fragmented remains of the ring once the metal had been removed. No metal-free porphyrinic rings are detected in the effluent oil. Experi- ments by Hung (1979) using free-base Etio porphyrin as the feed have demonstrated that the metal-free structures are rapidly destroyed at these reaction conditions. The central metal is thus essential for the stability of these species at demetallation conditions and once removed, the ring rapidly disintegrates.

These results indicate that Ni-EPH2 is the only stable metal- bearing intermediate in the demetallation of Ni-Etio. As a consequence, the total metal in the oil (determined by atomic absorption) can be accounted for by summing the contributions of the Ni-EP and Ni-EPH2 species (determined by visible spectra: Ni-EP, 557 nm; Ni-EPH 2, 616 nm). A typical plot of the variation of reactor outlet concentration as a function of contact time taken from Agrawal (1980) is shown 4w ww W w w

\.-- N N N N- 1 N\ / 3 >DEPOSIT 0 N N 2 N N H I

Ni-EP Ni-EPH2

Figure III-1. Reaction sequence for Ni-etioporphyrin. w 'pw

1.0 I' I II I

0.8-

2~ I ~Li

LU 0.6k C) M 0-

0 N Y) T 0.4k- 0~

0.21-

-L I 250 350 450 550 650 750 850

WAVELENGTH,NM

Figure 111-2. Absorption spectra of effluent oil sample during demetallation of Ni-etioporphyrin at 317 *C and 6.99 MPa H2 (1000 psig); background is xylene. - 102 -

390 557

2.01

Ni \N N L5 z Ni-EP 0 V)~ 1.0

200 300 400 500 600 700 800

WAVELENGTH, NM

0.3

395

.N, , N

N I

H LLJ U z 616 NI-EPH 2 Cm 0 MV) 0.11-

0.0

200 300 400 500 600 700 800 WAVELENGTH, NM

Figure 111-3. Absorption spectra of Ni-etioporphyrin (top) and Ni-etiochlorin (bottom); background is hexane. - 103 -

in Figure 111-4. Initially there is a rapid build-up of Ni-EPH 2 during

which time there is a slight induction period in the rate of total

metal removal. Once the intermediate concentration has built up, the

total metal removal rate increases.

Intrinsic kinetic parameters are evaluated from the experimental

results using the reaction sequence described above assuming first

order reaction steps and constant hydrogen concentration in the oil.

Rate constants used in the calculated model results of Figure 111-4 and

their associated activation energies obtained from the Arrhenius plots

in Figure 111-5 are listed in Table III-1. It is observed for Ni-Etio

that hydrogenation of the porphyrin is the rate limiting step in the

overall demetallation scheme. The hydrogenolysis step following

Ni-EPH2 formation is relatively rapid and the concentration of this

intermediate never amounts to more than 20% of the total metal in

solution at any time. The activation energies for the hydrogenation

(k1 ), dehydrogenation (k2 ), and hydrogenolysis (k3 ) steps are consistent with literature reported values for similar reactions in hydrodenitro-

genation (HDN) (Satterfield and Gultekin, 1981).

The large excess of hydrogen dissolved in the oil ensures that the hydrogen concentration remains constant and can therefore be lumped into

the first order rate coefficient. Variation of the hydrogen pressure during saturation of the oil permits the evaluation of the rate coefficients dependence on hydrogen pressure. Shown in Figure 111-6 is such a determination for the three first order rate coefficients ki, k2, and k3 in the Ni-EP reaction scheme reported by Agrawal (1980).

Based on the slope of the line for k , a first order dependence is observed. This is consistent with the observation that the first - 104 -

a TOTAL Ni

25 o Ni-EP a Ni-EPH 2

20 - o

1 15- 0 z 10-

03 0

0.0 0.05 0.10 0.15 0.20

A/fQ G CAT. HR WV/Q, ML OIL

Figure 111-4. Concentration versus contact time results for Ni-etioporphyrin at 27 ppm Ni feed, 343 *C, and 6.99 MPa H 2 (1000 psig) on oxide catalyst. Solid lines represent model calculations using parameter values in Table III-1. - 105 -

1000 .. I I I I

100

- kk3 -

10 - k,

1.6 1.7 I.j I/T x1O', oK~

Figure 111-5. Arrhenius plots for the first order rate parameters for Ni-etioporphyrin at 6.99 MPa H 2 (1000 psig) on the oxide catalyst. (Agrawal, 1980) - 106 -

TABLE III-1

Ni-etioporphyrin Kinetic Parameters

1 3

Ni-EP Z Ni-EPH - deposit 2 2

Reaction conditions: 0 27 ppm Ni feed, 343 C, 6.99 MPa H2 (1000 psig), oxide catalyst

k = 25.0 ml oil/g cat. hr E, = 16.3 k cal/mole

k2 = 100.0 E2 = 20.5

k3 = 80.0 E3 = 28.7 - 107 -

1000 I i I I

2 u 100

-j k3

600 psig 1000 psig 1400 psig iL I II I I I I I 1.0 1.5 2.0 2.5 LN P (MPa H2 )

Figure 111-6. Hydrogen pressure dependence for the first order rate parameters for Ni-etioporphyrin at 27 ppm Ni feed and 343*C on the oxide catalyst. (Agrawal, 1980) - 108 - intermediate formed is the Ni-etiochlorin, requiring the addition of one hydrogen molecule to the starting porphyrin. For a kinetically controlled dehydrogenation reaction, k should be independent of hydrogen 2 pressure and Figure 111-6 confirms a zero order dependence for this step. The final step, k3, the metal deposition reaction exhibits second order hydrogen dependence. This suggests that more than one hydrogen molecule is required for the final metal removal step and that a sequence of rapid reactions may be lumped into the hydrogenolysis step.

III.A.2. Ni-T3MPP

Investigation of the kinetics of Ni-T3MTP demetallation affords comparison with Ni-Etio of two porphyrins that have similar four pyrrole ring macro structures and conjugation (187T electrons in delocalization pathway) but have structural differences around the periphery of the macrocycle. Ni-Etio has methyl, ethyl substituents on the -pyrrolic positions and open methine bridges, whereas Ni-T3MPP has open a-pyrrolic positions and tolyl substituents oriented perpendicular to the porphyrin plane at the methine bridges. These structural differences result in significant reactivity differences although the global HDM scheme of porphyrin hydrogenation followed by hydrogenolysis and metal deposition still occurs. The higher solubility of Ni-T3MPP in Nujol (up to 120 ppm

Ni) as compared to Ni-Etio (35 ppm Ni) also provides a more realistic metal loading in the "model" oil.

The UV-visible spectra of effluent oil shown in Figure 111-7 reveals many absorption peaks indicating that Ni-T3MPP reacts through more stable intermediates than Ni-Etio. The feed Ni-porphyrin is rapidly hydrogenated at one peripheral double bond resulting in Ni-chlorin formation (Ni-PH2) followed by a second rapid hydrogenation selectivity at an adjacent pyrrole ring forming the Ni-isobacteriochlorin Iw MOw 0

0.75 I I I

I I

0.601

Liu 0.451 z C-

0 0 Y) mCD 0.301- C\j

0.15 -

I 250 350 450 550 650 750 850 WAVELENGTH, NM

Figure 111-7. Absorption spectra of effluent oil sample during demetallation of Ni-tetra(3-methylphenyl)porphyrin at 345 *C and 6.99 MPa H2 (1000 psig); background is xylene. - 110 -

(Ni-PH4 ). See Figure 111-8 for assignment of peaks. Central metals are observed to favor formation of adjacent tetrahydroporphyrins, the isobacteriochlorins, as opposed to opposite tetrahydroporphyrins known as the bacteriochlorins (Scheer, 1978).

Typical concentration versus contact time plots for Ni-T3MPP demetallation at 345*C and feed concentrations at 26 ppm Ni and 63 ppm

Ni are shown in Figures 111-9 and III-10 respectively. Experimental points are represented by the symbols. Both figures show the rapid disappearance of feed porphyrin (Ni-P) concurrently with the rapid production of Ni-PH2 and Ni-PH followed by their more gradual removal.

Beyond a critical W/Q value, 0.025 gcat/ml/hr in Figure 111-9 and 0.04 gcat/ml/hr in Figure III-10, the NiP/NiPH2 and Ni-PH 2/NiPH ratios remain constant indicating that a dynamic equilibrium has been estab-

lished between the porphyrinic species. The most striking feature of

these figures is that the sum of the porphyrinic species in the effluent

oil (concentration determined by visible absorption peaks: Ni-P, 526 nm;

Ni-PH 616 nm; Ni-PH4 , 593 nm) does not account for the total metal

concentration in the oil determined by atomic absorption spectrophotometry.

The discrepancy is represented by the symbol 0. At 95% Ni-T3MPP conver-

sion only 20-30% of the total metal in solution can be accounted for by

porphyrinic species (Ni-P, Ni-PH 2, Ni-PH4 ) whereas for Ni-Etio, the

metal in solution could always be accounted for by the sum of the Ni-EP

and Ni-EPH2 contributions. A similar reaction pattern is observed for

Ni-T3MPP at a lower temperature, 2850 C, as shown in Figure III-11. At

high feed conversions there is again a large discrepancy between

porphyrinic metal and total metal, accentuated in this figure by the

slow rate of metal removal at these conditions. - 111 - 0.20

0.18 414 0- N 0.16

0.14 N N IJ2 z) 0.12 4 526 M 0.10 0 to Ni-P E) 0.08

0.06

0.04

0.02

250 350 450 550 650 750 5 WAVELENGTH, NM

Q25

0.20

o N H 0.15 H z

0 Ni-PH U) 0.10 2 00

616

0.0

200 300 400 500 600 70 8 0 WAVELENGTH, NM

0.16 400

H -

*'2 o N

C3 593 0

0 0.08 Ni-PH 0 4 UO

0.04 F

0.0t 20 0 300 400 500 600 700 800 WAVELENGTH, NM

Figure 111-8. Absorption spectra of Ni-tetra(3-methylphenyl) porphyrin (top), Ni-T3MPchlorin (middle), and Ni-T3MPisobacteriochlorin (bottom); background is hexane. - 112 -

A TOTAL Ni 25 o Ni-P

o Ni-PH 2 9 Ni-PH4 20 A 4 Ni-X

_J 15 LiJ z 10

5 0 e

0.0 0.05 0.10 0.15 0.20 GCAT. HR WA/Q/ ) ML OIL

Figure 111-9. Concentration versus contact time results for Ni-T3MPP at 26 ppm Ni feed, 345 *C, and 6.99 MPa H2 (1000 psig) on the oxide catalyst. Solid lines represent model calculations. - 113 -

70 1 1

60 . a TOTAL Ni o Ni-P

o Ni-PH 2 50. e Ni-PH 4 4 Ni-X

40 -

20- 0

10 - -

0.0 0.1 0.2 0.3 0.4 0.5 0.6

G CAT. HR W/Q ) ML OIL

Figure III-10. Concentration versus contact time results for Ni-T3MPP at 63 ppm Ni feed, 345 'C, and 6.99 MPa H2 (1000 psig) on the oxide catalyst. Solid lines represent model calculations using parameter values in Table 111-2. - 114 -

70 TOTAL Ni o Ni-P 60 - Ni-PH 2 6 Ni-PH4 SNi-X 50

-40

LLJ

-30-

20-

10.

OO 0.16 0.32 0.48 0.64 GCAT. HR W/Q ' ML OIL

Figure Ill-il. Concentration versus contact time results for Ni-T3MPP 0 at 66 ppm Ni feed, 285 C, and 6..99 MPa H 2 (1000 psig) on the oxide catalyst. Solid lines represent model calculations. - 115 -

Examination of the effluent oil UV-visible spectra at high porphyrin conversions presented in Figure 111-12 reveals no dominat- ing absorption peaks at 400 nm. From this it can be concluded that the unaccounted for Ni is not present in a porphyrinic form, since macrocyclic conjugation has a characteristic strong absorption band at this wavelength (Smith, 1975).

HPLC analysis of effluent oil samples has revealed the presence of several compounds in addition to the Ni-P, Ni-PH 2, and Ni-PH .

Combined UV-visible spectra and LC analysis indicates most of these peaks are aromatic fragments with little absorption in the visible region. Effluent samples at high porphyrin conversions with especially poor metal balances contained an additional peak in substantial quantities, more polar than the porphyrinic components. The UV-visible spectra of this compound taken during LC analysis is shown in Figure

111-12 along with the composite spectra of effluent oil from which the injected sample was taken. The absorption peaks of this one species correspond exactly to the maxima in the effluent oil scan indicating that this is the dominant species in the oil.

Successive oil injections and peak collections enabled sufficient quantities of this nickel intermediate to be isolated for analysis and identification. Atomic absorption analysis has confirmed that this one compound can essentially account for the remaining nickel in the metal balance, hereinafter termed Ni-X. In effluent oil samples taken at short contact times the metal balance nearly closes with Ni-P,

Ni-PH and Ni-PH . Correspondingly, this Ni-X species is detected only in trace amounts during HPLC analysis of these samples. - 116 -

1.0

358 0.8- 299 334

LJ zU 0.6-

0 0.4

478

0.2

250 350 450 550 650 750 850 WAVELENGTH, NM

0.05 3s8

0.04 334

299 U 0.03 Cr 0.n (J1 CO 0.02.

450478 0.C I

0.0 - 20 0 300 400 500 600 700 800 WAVELENGTH, NM

.. 1 to

I I ! I , I 300 A00 60 700 nm

Figure 111-12. Absorption spectra of effluent oil sample at high Ni-T3MPP conversion (top), "Ni-X" species isolated from this oil during HPLC analysis (middle), and Ni-corrin type species from Rasetti (1979) (bottom); background is hexane. - 117 -

The mass spectrum of this nickel intermediate presented in Figure

111-13 reveals a parent molecular ion (M+1) at m/e = 633. The presence of nickel is confirmed by the paired peaks at 633 (Ni58) and

635 (Ni60) corresponding to the nickel isotopes in the ratio of 2.5/1.

This molecular weight is similar to that obtained by removing a

toluene group from the starting porphyrin but several factors rule out

this possible interpretation for Ni-X. The actual molecular weight of

such a species corresponds to m/e = 637 (M+1, = 726-90+1). Hence the

removal of 4 additional hydrogens from an already unsaturated species would be required. This is an unlikely occurrence considering both

the nature of the molecule and the strong reducing environment of the

reactions. The loss of a toluene group would also not be expected to

reduce the macrocyclic conjugation of the porphyrin ring. This is not

consistent with the absence of a strong absorption band at 400 nm in

the Ni-X spectra of Figure 111-12.

Further information is obtained by comparing the infrared absorb- -l -l ance bands in the 1800 cm to 900 cm region for the nickel inter-

mediate and the starting porphyrin, Ni-T3MPP, shown in Figure 111-14.

The increased number of bands in this region for Ni-X is an indication

of lower symmetry which would result from hydrogenation or structural

alteration of the starting porphyrin. The appearance of the 1470 cm~

band and 1550 cm~ shoulder have been assigned to -C = N- and -C = C-

stretching vibrations which are forbidden or very weak in the highly

symmetric D4 h porphyrin system (Thomas and Martell, 1959). Similarly,

strong bands in the Ni-T3MPP spectra at "1010 cm 1 and 1352 cm~1

assigned to porphyrin macrocylic ring vibration (Kincaid and Nakamoto,

1975) and -C -N = vibrations of high double bond character (Thomas and w w w' W 0

9 87 I" 0 A . .. 174 -.

cxI

m7 373 . 4.47 963 4. .4R 99 55 599 6 9 35WA 460 450 1-*, 5-8

667 689 77 735 0 750m Sim.. 8 W/e

Figure 111-13. Mass spectrum of Ni-X. - 119 -

1800 1600 1400 1200 1000

WAVENUMBERS CM'

1800 1600 1400 1200 1000

WAVENUMBERS CM 1

Figure 111-14. Infrared spectra of Ni-T3MPP (top) and Ni-X (bottom). Samples referenced to tetrachloroethylene. - 120 -

Martell, 1959), respectively, are absent in the Ni-X spectra. Bands at n,1010 cm~1 have been reported for dihydrogenated and hexahydrogenated nickel porphyrins indicating that reduction alone does not eliminate

this characteristic ring vibration (Angst, 1981). Furthermore, the presence of phenyl groups on the Ni-X species is indicated by the band at 1603 cm- which is likewise strong in Ni-T3NPP (Thomas and Martell,

1959).

An interpretation of the Ni-X intermediate consistent with the analytical results and the reaction environment is required. Such a

species appears to be the result of a concerted reaction involving con-

traction of the ring at a bridge position adjacent to one of the hydro-

genated pyrrole rings of Ni-PH by elimination of the methine carbon as a xylyl group. The central nickel metal is known to exert a strong

templating affect in such cyclizations by ensuring a favorable posi-

tioning between the adjacent c-pyrrolic carbon atoms (Grigg, 1978).

Similar reactions are common in the synthesis of analogous contracted ring structures referred to as metallo-corrins. A possible sequence

to interpret this ring contraction route is depicted in Figure 111-15.

The initial step involves hydrogenation of a meso bridge adjacent to a hydrogenated pyrrole ring in Ni-PH followed by opening of the ring by a concerted electrocyclic reaction. As will be discussed later, the meso positions on hydrogenated porphyrin rings are highly reactive.

The key step in the sequence which follows involves an acid catalyzed ring contraction step (Grigg, 1978) with subsequent elimination of the former bridge position carbon atom as a xylyl group,

For the case at hand, the molecular weight of this newly generated species is 628 indicating that the actual Ni-X intermediate has under- "p 0

2HH H H HHHH

H / meso bridge

hydrogenation H/ H

0 \ ring opening and

acid attack

Ni-PHLI

H 0 H H HH H HH H H H H H Hl H

H N HH Ni N f inal Hn H H hydrogenation H contraction HH and xylyl 0H H elimination

Ni-X m/e = 628 m/e = 632

Figure 1II-15. Proposed sequence for ring contraction and Ni-X formation. - 122 -

gone further hydrogenation after elimination of the xylyl group. The maxima in the visible spectra of the Ni-X intermediate presented in

Figure 111-12 shows resemblance to that of a nickel-corrin hydrogenated at additional pyrrole positions (Rasetti, 1979). These may likewise be the hydrogenated sites in the Ni-X molecule. The lack of any evidence for a Ni-PH 6 species in the oil (visible absorption peak at 670 nm (Johansen et al., 1981)) is an indication that this hydrogenation occurs after ring contraction.

Several features of the proposed Ni-X structure shown in the

reaction scheme of Figure 1I-15 are now consistent with the observed analytical and experimental results. Foremost is the agreement be-

tween.lack of macrocyclic conjugation detected in the visible spectra

of this species and the corresponding interrupted porphyrin ring

structure with extensive hydrogenation. Additionally, the IR spec-

trum is characterized by the absence of porphyrin ring vibrational modes and a decrease in molecular symmetry. Absorbance bands reported

for contracted ring systems at %1566 cm-1 (Ofner, 1981) and 1500 cm~1

(Burger, 1975) also appear identically at 1566 cm~1 and as a shoulder

at 1500 cm~ in the Ni-X sample. The nickel, verified by both

atomic absorption and mass spectra, is retained in a square planar

configuration with the four nitrogen atoms analogous to its coordina-

tion in the starting porphyrin. This indicates that no dramatic

structural rearrangements were required during the formation of this

species. Finally, the relatively slow reactivity of this nickel

intermediate to-demetallation is consistent with the reported obser-

vation that ring contraction stabilizes the metal-nitrogen bonds when

the central metal is small enough to allow such a contraction - 123 -

(Buchler, 1975). The coordination cavity of the corrin-type ligand system is apparently closer to the spatial coordination optimum of the

Ni transition metal ion than the hydrogenated porphyrin ligand

(Angst et al., 1981).

Analysis of the total metal concentration as a function of W/Q for Ni-T3MPP demetallation in Figures 111-9 and III-10 revealed a discrepancy when attempting to interpret the behavior in terms of a single deposition step through Ni-X. In both figures the steepest slope on the total nickel curves (corresponding to the fastest rate of metal removal) occurs at short contact times when the oil metal balance nearly closes with Ni-P, Ni-PH2, and Ni-PH and the Ni-X concentration is low. At longer contact times when the concentration of Ni-X is the largest and comprises essentially all metal in the oil, the rate of total metal removal is slower. Hence a route for metal deposition other than from the Ni-X species alone is required. A semi-logarithmic plot of the total nickel concentration data contained in either Figure

111-9 or III-10 does in fact reveal two straight line regimes indicative of two first order removal rates.

The kinetics of Ni-T3MPP demetallation have been modelled as a sequence of first order reactions in the network shown in Figure 111-16. Feed porphyrin is successively and reversibly hydrogenated twice to form the tetra- hydrospecies. This Ni-PH can then react via two paths, one leading to direct deposition of nickel on the catalyst surface and the other leading to formation of the stable Ni-X intermediate. The relative ratio of these two reactions may be dependent on the proximity of the molecule to both a hydrogenation and an acid site which are speculated to be necessary for ring contraction. The direct deposition step k 7 is analogous to the direct vo ww Now w 0

o 0 H H 7 N -~.- DEPOSIT N N \N

H H H HH 0

Ni-P N Ni-PH IN) I-PH2 H

H H HH H H H -N 6 -,j- DEPOSIT

H H

HH HH H

Ni-X

Figure 111-16. Reaction sequence for Ni-tetra(3-methylphenyl)porphyrin. - 125 -

deposition of metal from Ni-EPH2 in the Ni-Etio sequence. This reaction (k7 ) may proceed through other intermediates but presumably they are very reactive and exist only in trace amounts at reaction conditions. The formation of the Ni-X intermediate via reaction k results in production of a pool of relatively unreactive nickel in the oil. The hydrogenolysis step k 6 leading to metal deposition is slow and determines the rate of total metal removal at the higher contact times as evidenced by the disappearance of Ni-X paralleling the total metal removal rate. Presumably a second macrocyclic ring cleavage step is important in the final metal deposition sequence but no metal free fragments were identified in the oil indicating the preferred site for attack on the ring.

The overall metal removal rate from Ni-T3MPP can therefore be characterized by two rates; a fast decay at short contact times governed by k and a slow removal rate at longer times governed by k 6 * Attempts to interpret the results in terms of a single deposition route through the Ni-X species did not result in an adequate repre-

sentation of the total amount of metal and its removal rate at both short and long contact times.

The coupled set of rate expressions for this model are solved by use of the Wei-Prater technique (Wei and Prater, 1962) in Appendix B.

Kinetic parameters were obtained from non-linear least squares fitting of the experimental data to this solution or by the Himmelblau-Jones-

Bischoff technique (Himmelblau et al., 1967) using the computer pro-

grams in Appendix C. Both methods gave similar results.

Kinetic parameters used in the solid line model calculations of

Figures 111-9 and III-10 are listed in Table 111-2. In contrast to - 126 -

TABLE 111-2

Ni-tetra(3-methyl phenyl)porphyrin Kinetic Parameters

1 3 7 Ni-P . Ni-PH ' Ni-PH4 -- P deposit 2 4

Ni-X -- deposit 6

0 345 C, 6.99 MPa H 2 (1000 psig), oxide catalyst

Feed concentration, ppm Ni rate coefficient 16 26 63 90

k ml oil/gcat hr 86.0 85.0 105.0 94.0

100.0 90.0 94.0 84.0

120.0 130.0 120.0 130.0

k4 145.0 150.0 150.0 155.0

55.0 66.0 49.0 60.0

4.20 4.60 3.80 4.25

24.0 25.0 18.0 19.5

Activation Energy

E kcal/mole 17.8 16.4

21.8 20.0

18.9 17.3

23.1 21.2

21.2 19.0

20.0 20.0

24.0 23.3 - 127 -

Ni-Etio, hydrogenation of Ni-T3MPP is rapid and the hydrogenolysis

steps k6 and k7 are rate determining in the demetallation sequence.

Activation energies for the rate parameters at 63 ppm Ni feed are

obtained from the Arrhenius plots of Figure III-17. Values likewise

determined at 26 ppm Ni feed (Arrhenius plots shown in Figure 111-18) were similar to the activation energies for the seven steps at the

higher concentration. Negative activation energy differences for the

hydrogenation/dehydrogenation steps (E1 - E2 , E3 - E4) were qualita-

tively consistent with the exothermicity of hydrogenation reactions.

Thermodynamic data on the enthalpies of reaction between these porphyrinic

compounds were unavailable for quantitative comparison. Due to the

lumping of the hydrogen concentration into the first order kinetic

parameters, a true estimate of the enthalpy of reaction from the acti-

vation energies would have to take into account the heat of solution

of hydrogen in the oil.

The kinetic rate parameters listed in Table 111-2 for Ni-T3MPP at

345C span a 5.6 fold variation in feed concentration ranging from 16

to 90 ppm Ni. At each of the four feed concentrations the same reac-

tion sequence, depicted in Figure 111-16, was observed. The similarity

in the kinetic parameters suggests a concentration independence over

the range examined. The slight variation observed between some of the

rate parameters determined at different feed concentrations is far less

than the magnitude of the concentration changes between these runs.

These parameter variations may arise from different levels of activity

from one batch of catalyst to the next due to catalyst ageing and

metals loading. Different catalyst charges were used for each of the

four runs and the amount of total nickel deposited (determined by mass - 128 -

I | I 1 I -k4 k3 100

10 -

: -H k6

' 0.1 i iI iI 1.6 1.7 1.8 1/T x 10', K1

Figure 111-17. Arrhenius plots for the first order rate parameters .for Ni-T3MPP at 63 ppm Ni feed and 6.99 MPa H2 (1000 psig) on the oxide catalyst. - 129 -

500

- k4~

100 ki k2

k5 k

10 -0

-ko 00

- -

0.5 1.6 1.7 1.8 1/T x 10 K

Figure 111-18. Arrhenius plots for the first order rate parameters for Ni-T3MPP at 26 ppm Ni feed and 6.99 MPa H2 (1000 psig) on the oxide catalyst. - 130 - balance) at the end of each run varied from 0.67 wt% for the 16 ppm Ni run to 3.7 wt% Ni for the 90 ppm Ni experiment.

The first order rate parameters in the demetallation scheme for

Ni-T3MPP therefore appear to be independent of porphyrin concentration.

Agrawal (1980) reported similar behavior in the HDM of VO-porphyrins although over a much narrower concentration range.

An interpretation of this concentration independence is possible by considering a surface kinetic model. Reactions in heterogeneously catalyzed systems depend on surface species concentration and therefore are typically influenced by adsorption/desorption phenomena on the active catalytic sites. The reactions when interpreted in terms of simple first order rate expressions can yield kinetic rate parameters exhibiting a concentration dependence. This behavior may be interpreted in terms of adsorptivity on catalytic sites through the use of a Langmuir-Hinshelwood kinetic model shown below: k.. C. rate = I + 1 + K.i where the summation includes all species, both reactants and products.

A common assumption employed is that each species has an equal adsorptivity, K. This reduces the denominator to 1 + K I C . Since the summation includes all species, it remains constant and equal to the feed concentration, C . Hence for strongly adsorbing species, KC >> 1 and a bulk pseudo first order rate coefficient for such a system exhibits an inverse feed concentration dependence. For catalytic sites pith relatively weak adsorption interactions, KC << 1 and concentration independent first order rate parameters are expected. The concentration independence of the seven rate paramters in the Ni-T3MPP reaction scheme indicate at these conditions the term KC is not important relative 0 - 131 - to 1. The total number of active sites therefore remains essentially constant with changes in feed concentration.

Adsorption studies for Boscan petroporphyrins on alumina supports and on CoMo/Al203 catalysts have been reported by Morales and co-workers

(Morales and Galiasso, 1982, Andreu et al., 1981). Both support and catalyst were determined to have similar adsorption properties at electron acceptor sites on the surfaces but low heats of adsorption indicative of weakly bound porphyrins. At temperatures in excess of

350*C or in strong solvents more than 90% of the room temperature adsorbed porphyrin could be removed. The adsorption of nickel porphyrins on these materials was also observed to be lower than of vanadyl porphyrins. This information lends further insight and support to the concentration independent rate parameters determined for Ni-T3MPP at 345*C. Based on these results, no detailed modelling of the intrinsic kinetic behavior in terms of the more fundamental Langmuir-

Hinshelwood model was attempted or deemed necessary.

The hydrogen pressure dependence of the rate coefficients presented in Figure 111-19 was determined in an analogous manner to that for

Ni-Etio porphyrin. Similar results for the two porphyrins were obtained. Over the pressure range of 4.58 - 10.09 MPa H2 (650 - 1450 psig H2) the hydrogenation steps k and k3 exhibited first order hydrogen dependence whereas the dehydrogenation steps k 2 and k were independent of hydrogen pressure. Step k5 exhibited second order hydrogen dependence. The proposed structure for the Ni-X intermediate is consistent with the requirement for the addition of more than one hydrogen molecule in its formation from Ni-PH Likewise, the final metal deposition steps k6 and k exhibited second order dependence suggesting again that a sequence of hydrogen consuming - 132 -

1000 , I I

k 4 0 '-I - -6- 100 K

k3 -II 0

k5 10

1 k6

650 psig 1000 psig 1450 psig Ii I I iI I si 1.0 1.5 2.0 2.5 LN P (MPa H2 )

Figure 111-19. Hydrogen pressure dependence for the first order rate parameters for Ni-tetra(3-methylphenyl)porphyrin at 63 ppm Ni feed and 345 *C on the oxide catalyst. - 133 -

reactions may be lumped into these final steps.

III.A.3. Ni-TPP

The behavior of Ni-TPP under HDM conditions is similar to that of

Ni-T3MPP. The meta-substituted methyl groups on the phenyl rings are observed to have little effect on the porphyrin reactivity.

Total metal removal kinetics for Ni-TPP on CoMo/Al203 were first reported by Hung (1979) in his model compound hydrodemetallation work.

At that stage, the importance of intermediates in the overall demetallation scheme were not fully appreciated. UV-visible spectra of effluent oil samples revealed, however, the presence of absorption peaks at 616 rum and 593 nm characteristic of the Ni-PH2 and Ni-PH4 species. Re-examination of Hung's batch data indicates that like

Ni-T3NPP, the porphyrinic fraction fails to account for all metal in effluent oil. At 90% Ni-TPP conversion at reaction conditions of

344*C and 6.99 MPa H2 (1000 psig), only 20% of the Ni in the oil can be accounted for by the Ni-P, Ni-PH2, and Ni-PH4 components.

III.B. Discussion of Porphyrin Reactivity

The model compound results presented in the previous section demonstrated the sequential nature of hydrodemetallation on the oxide form of CoMo/Al203 involving both hydrogenation and hydrogenolysis reactions. The initial ring hydrogenation of the pyrrole group appears to be the pathway resulting in the least disruption of the porphyrin molecule under reaction conditions. Even though some strain is induced into the macrocycle to accomodate the sp3 carbon atom of - 134 -

the pyrroline ring (Scheer, 1978), the porphyrin aromaticity is not significantly interrupted since only 18 of the original 22 7r electrons in the porphyrin are included in any one delocalization pathway (Smith,

1975a). The different degrees of hydrogenation observed between

Ni-Etio and Ni-T3MPP (also Ni-TPP) can be rationalized in terms of porphyrin basicity and steric factors. The electron donating effect of the methyl and ethyl groups in Ni-Etio renders the -pyrrolic positions more basic and hence less susceptible to electron addition ocur- ring upon reduction than Ni-T3MPP where only hydrogen atoms exist at these positions. The larger -pyrrolic substituents in Ni-Etio would also appear to hinder access to the catalytic hydrogenation sties.

Reduction potentials, directly related to basicity, are a commonly used parameter in correlating porphyrin reactivity (Worthington et al.,

1980). Reduction potential values determined by Hambright (1982) are lower for Ni-T3MPP as compared to Ni-Etio, implying that Ni-T3IPP should be more readily reduced. The first order hydrogenation rate constants for Ni-T3MPP (ki = 85.0 ml/gcat hr) and Ni-Etio (k1 = 25.0 ml/gcat hr) reflect this as does the greater extent to which Ni-T3MPP is reduced (Ni-PH formation).

A more significant result that the slight disruption of aromaticity resulting from porphyrin hydrogenation is the enhancement in reactivity of the methine bridges adjacent to the newly formed pyrroline ring toward electrophilic attack (Fuhrhop and Subramanian, 1976, Scheer and

Inhoffen, 1978, Fuhrhop, 1978). In the porphyrin macrocycle, the four pyrrolic rings are maintained as aromatic subunits by borrowing electron density from neighboring methine bridges. The reduced pyrrole ring in the chlorin is no longer such a subunit and thus the - 135 -

electron density at the neighboring methine positions is increased.

Kwart et al. (1980) have recently proposed a mechanism for HDS on sulfided CoMo/Al 203 catalysts. The catalyst is postulated to act as an electrophile seeking to fill the anion vacancies of its Mo atoms through interactions with bonds of high electron density. A similar interpretation is useful to explain the HDM behavior observed in this study on oxide CoMo/Al 2 03 . The hydrogenated porphyrins presumably interact at the highly reactive methine bridges adjacent to the pyrroline ring with Mo anion vacancies. If the molecule adsorbs in a planar configuration, the R orbitals of the methine bridge carbons can be coordinated by the electron-deficient Mo centers. Interaction of

Tr electrons with the metal would lead to transformation of the C-C

Tr bond to yield a a bond which at the same time would activate the methine bridge for hydrogenation. Once the aromaticity of the bridge has been destroyed by hydrogenation, stabilization of the molecule by macrocylic conjugation is lost. The resulting strain in the now sp a-pyrrolic carbon atom may favor cleavage of the ring and formation of an open ring metallo-tetrapyrrole structure (Fuhrhop, 1978). One possible route for this reaction would be a concerted 18 r electron electrocylic reaction (Grigg, 1978). The presence of an acidic catalyst site may likewise facilitate this cleavage via a carbonium ion cracking mechanism. Similar open chain tetra-pyrrolic structures

(M-biliverdinates) have been found to be unstable under catalytic hydrogenating conditions (Subramanian and Fuhrhop, 1978) so presumably the metal is quickly deposited on the catalyst and the remaining aromatic structure destroyed. - 136 -

With Ni-Etio, the lack of stable intermediates (other than the chlorin) detected indicates that this sequence proceeds rapidly when the methine bridges are open and exposed to attack.

In contrast to Ni-Etio, the reaction of Ni-T3MPP involves relatively slow rates of hydrogenolyis and the generation of a stable non- porphyrinic intermediate. This may be interpreted as resulting from the influence exerted by the tolyl groups at the methine bridges.

Their presence may be envisioned as sterically hindering interaction of the methine bridge sites with the catalyst resulting in inhibition of the hydrogenation of the double bond and the subsequent cleavage step. A second influence is the ability of the tolyl group to stabilize the methine carbon as a leaving group during formation of the nickel-corrin type intermediate, the Ni-X. Once hydrogenation of the methine bridge and subsequent opening of the ring has occurred, a rapid series of reactions follows. The rate of contraction and reclosure of the ring would appear to determine the relative amount of

Ni-PH4 reacting to deposit metal directly (k7 ) or to produce Ni-X (k5 ).

Unlike the strain associated with the sp3 a-carbon in the porphyrin system, the same sp3 orbital configuration in the a-carbon of the newly formed contracted ring molecule is known to be stable (Fuhhrop,

1978). The success of the central nickel in exerting a templating effect to contract the ring is apparent by the k5 /k7 ratios consistently greater than one. Accompanying this bridging of the two a-carbon atoms in the adjacent pyrrole rings is the simultaneous elimination of the former methine carbon as a xylyl group. This reaction occurring to relieve the steric crowding at the contracted bridge bond. - 137 -

The cleavage of the carbon-carbon bond in this sequence, beta to the tolyl ring is a reaction commonly encountered in carbonium ion and free radial reactions. This same ring contraction/carbon elimination reaction were it to occur with nickel-etioporphyrin would involve loss of a methyl group. Formation of a methyl carbonium ion, for example, is energetically 50 kcal/mole higher than the corresponding xylyl carbonium ion and hence highly unfavorable (Alder et al., 1971). This may explain in part the absence of the analogous non-porphyrinic nickel intermediate in the demetallation of Ni-etioporphyrin on CoMo/Al2 03

III.C. Demetallation Reactions Under Diffusion Limited Conditions

III.C.l. Metal Deposition Profiles

Metal deposition profiles measured in catalysts from commercial hydroprocessing units reveal that HDM reactions are diffusion limited.

Vanadium tends to be deposited in sharp, U-shape, profiles (Todo et al.,

1971, Sato et al., 1971) whereas nickel has been observed in both

U-shape (Todo et al., 1971) and uniform (Sato et al., 1971) profiles.

Agrawal (1980) and Tamm et al. (1981) have shown that metal deposition profiles depend on the axial location of the catalyst within the reactor. Nickel and vanadium both exhibited profiles with internal maxima, termed M-shape, at the reactor entrance. These maxima were found to shift to the pellets' edge at the reactor outlet, generating the classical U-shape profile. Wei and Wei (1982) have recently shown how all these types of profiles can be explained by a consecu- tive HDM mechanism involving a metal depositing intermediate which is - 138 -

not originally in the oil.

Nickel deposition profiles obtained with Ni-Etio and with Ni-T3MPP as a function of axial position are presented in Figures 111-20 and

111-21. Experimental points are represented by the open circles. At the reactor entrance, no metal depositing species are present in the oil. The nickel porphyrins must first react to form the metal depositing intermediates before metal accumulates on the pellets. As feed material diffuses into the first catalyst particles and reacts, the concentration of intermediates rises and the M-shape profiles appear.

The position of the internal maxima within the first pellets is determined by the reaction network, the temperature, and the hydrogen pressure.

The internal maxima, present close to the edge of the pellet at the reactor entrance with Ni-Etio in Figure 111-20, is observed to shift to the pellet's edge due to the build-up of the metal depositing Ni-EPH2 in the bulk. Metal profiles from pellets at the middle and exit of the reactor exhibit the classical shape characteristic of a first order reaction with diffusion.

The internal maxima at the reactor entrance for Ni-T3MPP is significantly more pronounced and moved from the edge as shown in

Figure 111-21. The maxima occurs at r/R = 0.75 whereas for Ni-Etio it was 0.90. This is easily explained in terms of the longer HDM sequence and the slow metal removal steps previously discussed for

Ni-T3MPP. Metal profiles taken from pellets a quarter of the distance into the bed and -even from the middle of the bed reveal indications of an internal maximum which further confirms this point. Another

interesting feature of the Ni-T3MPP metal profiles is the relatively - 139 -

1.4 1.4 ENTRANCE 1.2 MIDDLE 1.2 0 I 1.O -J 1.01. 0 0 0.8 0 a(- Lo 0.8 000oo 0 F- 0.6 .0 0~ 0000 0000 I- 0.6 0

0.4. 0.4 0g 0000000 o oo 0.2- 000-0 0.2

1.0 0.8 0.6 0.4 0.2 0.0 1.0 0.8 0.6 04 0.2 0.0 f/R r/R

1.4 EXIT 1.2

1.0

LU 0.8- z I- 0.6

0 0.4 OA 0 00 0 0.2 - 000

1.0 0.8 0.6 0.4 0.2 0.0 r/R

Figure 111-20. Nickel deposition profiles in 1/16" diameter catalyst pellets at various reactor axial positions from 'Ni-etioporphyrin demetallation at 345 *C and 6.99 MPa H2 (1000 psig) on the oxide catalyst. Solid lines represent model calculations using parameter values in Table 111-3. - 140 -

6 1. r 1.6 ENTRANCE FIRST QUARTER

0 0 0 0 1.2 1.2

0 00 0 -J 0 0 LiJ -J LiJ 00 0 0 0 U0.8 0 0.8- 0 00 o 000 rx 00o 0H1 00 00

0.4 - 0.4 H

I - 1.0 0.8 0.6 0.4 0.2 [ 0.0 1.0 0.8 0.6 0.4 0.2 0.0 r/ R r/R

1.6 1.6, MIDDLE EXIT

1.2 1.2 00 00 -J 0 LiJ 0 0 LiJ 00 z 0.8 0.8. 0 0 00 0- O 00 0 00 0 H- 0 00 0 00 0 0 0 0 4 0 0 . - OA

1.0 0.8 0.6 0.4 0.2 0.0 1.0 0.8 0.6 0.4 0.2 0.0 r/R r/R

Figure 111-21. Nickel deposition profiles in 1/16" diameter catalyst pellets at various reactor axial positions from Ni-tetra(3-methylphenyl)porphyrin demetallation at 345 *C and 6.99 MPa H2 (1000 psig) on the oxide catalyst. Solid lines represent model calculations using parameter values in Table 111-4. - 141 -

uniform metal loadings in the inner half of the pellets. This is a

direct result of the two metal deposition routes characteristic of

Ni-T3MPP. The fast reaction (k7 ) operates under strong diffusional

limitations giving rise to the high metal loadings at the edge whereas

the slow deposition step (k6 ) dominates in the center since it is less

influenced by diffusional effects and results in relatively uniform metal profiles.

III.C.2. Metal Profile Model Development and Discussion

The pattern of metal deposition is useful as a fingerprint of the

HDM kinetic scheme and reflects the radial position of metal-bearing

intermediates just prior to deposition. The profiles have been modelled using the theory of coupled multi-component first order reac-

tion and diffusion given by Wei (1962). The approach used for the

Ni-T3MPP kinetic scheme will be outlined, that for the Ni-Etio scheme

has been given elsewhere (Agrawal, 1980).

There are several assumptions involved in the model development; no axial and radial dispersion effects, no interphase resistance be-

tween the liquid and catalyst particles, and end effects on the catalyst pellets are neglected. The first two assumptions are justified in

Appendix A.

The steady-state equations for diffusion with the hydrodemetal-

lation scheme can be written as: - 142 -

D 0 0 0 C k -k2 0 0

0 D2 0 0 C2 -k1 (k2+k) -k 0 C2 1 - (1)

PC 0 0 D 0 C 0 -k (k4 +k5 +k 7) 0 C3

0 0 0 D4 C4 0 0 -k5 k6 C4

where C1 , C 2, C3, and C4 are the concentrations in ppm Ni of Ni-P,

Ni-PH2, Ni-PH4 , and Ni-X, respectively, inside the catalyst pellet and

D1, D2, D and D4 the respective effective diffusivities in cm2/sec. These diffusion coefficients are assumed to be constant on the basis that the low level of deposited nickel is not sufficient to restrict diffusion in the catalyst pores. The rate coefficients, k, in units of ml oil/ g cat hr, are from the kinetic scheme previously discussed.

The catalyst density, pC, is necessary to balance the units. The boundary conditions accompanying eq. (1) are:

at r = 0, V 0 (2) C3

LC 4

C 1 C ls

at r = R, C [C] (3) C3 C3s C 4] C 4sj where the subscript s refers to the concentration in the oil at the outer surface of the catalyst. - 143 -

The solution to the above set of equations for cylindrical geometry

is given by:

I r/R) Cl 0 0 0 C ls

I (2 r/R) C02 2 0 0 C02 20 Y I ( 3 r/R) y C (4) C 3 0 0 0 03 3s 0 3 I (4 r/R) 0 04. 0 0 C 4 s I (04 0 4 )

where Y and Y are the eigenvector matrix and the inverse matrix of the p D K matrix and I the modified Bessel function of the first kind of zero order. Defining AX,2X 3, and X4 as the eigenvalues of pC K, then the Thiele modulus for each characteristic species is = R for i = 1, 2, 3, 4 where R is the pellet radius.

Using the solution given by eq (4), the overall species reaction rates for each catalyst pellet can be obtained from the fluxes of reactants across the boundary. The gradient at R is required and is given by: - 144 -

0 0 0 Cl $'

C 0 0 0 C2s 2 42 10(42 d C 2 (5) dr R ~o () C 0 0 0 C3s 3 0310(03

C C 4 0 0 0 4 s

which can be simplified to

d - (6) TC R Rs s I0W

where $ is the diagonal matrix in eq (5) and C the concentration 0 vector. The rate of reaction per unit mass of catalyst is then given by:

27TpRLD C R2pW =-I (7) 22 =s WR LPc RP 0

where L is the pellet length and p the oil density. This flux expres-

sion may be reduced to:

p K C (8) 5

where i is the diffusion-disguised overall rate coefficient matrix

based on the bulk oil concentrations given by: - 145 -

+ 2 D -1 (9) R P I()

The problem then is simply one of using this local rate expression in a differential material balance in the reactor. For a plug flow reactor assuming uniform catalyst distribution and no interphase transport resistances, the set of mass balances is given by:

dC Q = -K C (10) where Q is the volumetric oil flow rate and dW the differential weight of catalyst along the reactor length. The initial condition for eq (10) is the feed composition to the reactor:

Cls CW

at W = 0 C 2s 0 (11) C3s 0 LC 4J L0

Integrating eq (10) over the total weight of catalyst, W, yields:

Cls exp (-24 W/Q) 0 0 0 C 0 C =X 0 exp W/Q) 0 0 2s (-2 2 0 X (12) C 3 s 0 0 exp (-X3 W/Q) 0 0

0 0 0 exp (-A W/Q) 0 C4 s

=+ =+1 where X and X are the eigenvector matrix and the inverse matrix of the diffusion disguised rate constant matrix K and A , A' 3, and 1 2' 3 A its characteristic eigenvalues. With eq (12) it is possible 4 to - 146 -

calculate the concentration profiles in the bulk oil through the reactor

as a function of catalyst weight (or equivalently, reactor length).

Concentration distributions within individual catalyst pellets at

specific axial positions are determined from eq (4) using the bulk oil

concentration given by eq (12). The final step, the calculation of the

amount of deposited metal as a function of pellet radius at any posi-

tion along the reactor is accomplished using:

M = [k6 C4 + k7 C3] Pt x 104 (13) where M is the weight percent deposited nickel and t the reactor time

on stream in hours. The factor of 10- is necessary when the units of

concentration are in parts per million. The two routes to metal

deposition with Ni-T3MPP are clearly identified in eq (13).

The kinetic rate parameters used in solving eq (4) and eq (12) have been independently determined in the intrinsic studies leaving the

four diffusion coefficients as the only unknowns in these equations.

Values for these parameters were then adjusted to obtain a best fit between the measured and predicted values of both the intra-pellet metal profiles along the reactor length and the species concentration in the effluent oil. The computer program used in these calculations

is located in Appendix C. Listed in Table 111-4 are the parameter values used in calculating the solid line model profiles in Figure

111-21 and the resulting effluent oil composition. Results from modelling the Ni-Etio profiles of Figure II-20 are presented in

Table 111-3.

The model curves in Figure 111-21 predict the M-shape behavior for Ni-T3MPP at the reactor inlet with the maxima shifting to the edge - 147 -

TABLE 111-3

Parameters for Ni-etioporphyrin Metal Profile Modelling

(Figure 111-20)

= 25.0 ml oil/g cat hr t = 75 hrs

= 100.0 p = 0.88 g/ml

= 80.0 PC = 1.49 g/ml

-6 2 D = 2.5 X 10 cm /sec Q = 15.0 ml/hr

6 = 4.8 X 10- W = 5.0 g

R = 0.076 cm

C = 32.0 ppm Ni 0

Effluent Oil Concentration

Experimental Calculated

Ni-EP 4.8 ppm Ni 4.7

Ni-EPH2 0.70 0.63 5.5 5.33 - 148 -

TABLE 111-4

Parameters for Ni-tetra(3-methylphenyl)porphyrin Metal

Profile Modelling (Figure 111-21)

105.0 ml oil/g cat hr t = 109 hrs

94.0 p = 0.88 g/ml

120.0 Pc = 1.49 g/ml

150.0 Q = 15.0 ml/hr

49.0 w = 5.0 g

3.80 R = 0.076 cm

18.0 C = 66 ppm Ni 0

1.0 x 10- 6 cm 2/sec Effluent Oil Concentration

2.0 X 10- 6 Experimental Calculated

2.0 X 10- 6 Ni-P 9.9 ppm Ni 6.3

2.0 X 10 6 Ni-PH2 4.6 5.1

Ni-PH4 2.6 3.0

Ni-X 17.5 18.7 34.6 33.1 - 149 -

of the pellet toward the end of the reactor. Similarly, the concentration of metal at the point of maximum loading within a pellet is predicted to increase, initially, with axial position before declining further down the length of the bed. This behavior has been observed experimentally. Both of these features are a direct consequence of the sequential metal deposition pathway.

Due to the rapidly changing concentration of intermediates at the entrance of the bed, the deposited metal concentration calculated by the model at the surface of a catalyst pellet at this position was very sensitive to axial location. Mathematically, the metal concentration is zero at r = R on a pellet at the entrance of the bed (z = 0). Experi- mentally this was never observed due to the finite size of the catalyst particles (6 mm long) and the possible effects of dispersion and adsorption. The experimental profiles at the entrance were instead modelled by relaxing the constraint on the entrance position location and using metal profiles calculated generally with a 0.5 cm distance of the mathematically defined entrance position.

Quantitative agreement between the predicted and experimental metal profiles is satisfactory, especially in the front half of the reactor when using unequal diffusivities for the nickel species. The best fit values for the effective diffusivities are in the range 1 - 2 x 10-6 cm 2/sec. An estimate for the values of the diffusion coefficients for the porphyrinic species can be obtained from the Stokes-Einstein equa- * tion. Assuming a molecular diameter for Ni-T34PP of 15 A, the bulk -5 liquid phase diffusivity in Nujol at 350*C is calculated to be 4.8 x 10- cm /sec. An effective diffusivity in the catalyst is obtained by correcting for configurational effects with the Spry and Sawyer (1975) - 150 -

0 factor and by assuming a catalyst pore diameter of 80 A, a void fraction of 0.5, and a tortuosity factor of 6 (Satterfield, 1970). The result- -6 2 ing effective diffusivity is calculated to be 1.7 x 10 cm /sec which

suggests the values determined from modelling the metal profiles are

reasonable.

With the chosen values of the four diffusion coefficients, the

model accurately matches the total metal concentration of the effluent

oil but predicts a slightly lower Ni-P concentration and a higher Ni-X

concentration than what was observed. Since the metal profiles are

dependent on the bulk concentration of metal depositing species, any

errors in the calculated concentration distribution will be reflected

in the calculated metal profiles. This discrepancy in effluent oil

composition therefore explains the slight disagreement between the pre-

dicted and experimental metal profiles at the exit of the bed. Adjust-

ment of the four diffusivities was possible to match exactly the exit

concentration from the reactor. The resulting metal profile at the end

of the bed then accurately modelled the observed profile but the

agreement between profiles at the front of the bed was poor. It is

thus thought that the diffusion coefficient values listed in Table 111-4

are the optimized set.

The three porphyrinic species would be expected to have similar

diffusivities from a size consideration. The difference of a factor of

two between the values for Ni-P and Ni-PH 2, Ni-PH while not large, was

required to obtain a better fit to the metal profiles. It is conceiv-

able that the strength of the interaction with the catalyst surface is

different for the porphyrin and hydrogenated derivatives. This would

alter the adsorption and surface diffusion processes for the species, - 151 -

both of which contribute to the observed effective diffusivity. The

similarity of the Ni-X diffusion coefficient to those of the porphyrinic

species suggests the size and surface interactions of this species,

despite being a contracted ring, are not all that different from the

porphyrin.

The Ni-Etio metal profiles of Figure 111-20 were modelled in a

similar fashion. It was again determined that unequal diffusivities for

Ni-EP and Ni-EPH2 were needed to give an accurate representation of the

metal profiles throughout the bed as well as the effluent oil concentra-

tion (See Table 111-3). The values for the diffusion coefficients were

well within the expected range and of similar magnitude, yet consistently

larger than the Ni-T3NPP values which was expected based on the smaller

size of Ni-etioporphyrin. Once again, the hydrogenated derivative

(Ni-EPH2 ) was determined to have an effective diffusion coefficient twice

that of the porphyrin (Ni-EP).

An earlier interpretation suggested that the internal maxima in metal profiles at the entrance of the reactor was related to an increasing concentration of H 2S, speculated to be necessary for demetallation

(Tamm et al., 1981). The results presented here in the absence of sulfur demonstrate that the M-shape metal deposition pattern is solely a consequence of the reaction network of the metal species. This reaction pathway for nickel porphyrins under hydrodemetallation conditions on an oxide CoMo/Al203 catalyst has been shown to involve a sequential mechanism. The porphyrins are initially hydrogenated, forming precursor species which subsequently react via hydrogenolysis steps to deposit the metal on the catalyst surface. Structural differences on the periphery of the metallo-porphyrin molecules are shown to significantly influence - 152 -

the complexity of the reaction pathways and the relative rates of hydrogenation and hydrogenolysis. The substituent groups at the -pyrrolic and methine bridge carbon atoms in the porphyrin appear to exert a dominate role, both steric and chemical, in determining these reactivity differences. Variations in the overall rate limiting step for metal removal are possible from one porphyrin to the next but the global sequence of hydrogenation followed by hydrogenolysis is still preserved.

Radial nickel deposition profiles in catalyst pellets obtained under diffusion limited conditions are dependent on the axial location of the catalyst within the reactor. These profiles have been interpreted in terms of the sequential HDM reaction schemes developed for the porphyrins using diffusion coefficients for all the nickel species on -6 2 the order of 10 cm /sec. Structural differences between nickel porphyrins which influence the intrinsic kinetics are likewise reflected in the metal deposition patterns. - 153 -

CHAPTER IV

HYDRODEMETALLATION OF NI-T3NPP IN THE PRESENCE OF

NITROGEN AND SULFUR COMPOUNDS

IV.A. Demetallation in the Presence of Pyridine

The investigation of the reactivity of Ni-porphyrins on oxide

CoMo/Al 0 has enabled an understanding of the porphyrin demetallation 2 3 sequence to be attained in the absence of external artifacts or competing reactions. In contrast, actual demetallation processing occurs in the presence of nitrogen and sulfur compounds on sulfided catalysts where competitive adsorption of heteroatom species and inherently different catalytic sites determine HDM reactivity. The aim of this chapter is therefore to examine the intrinsic demetallation phenomena of Ni-T3MPP under conditions more representative of industrial hydroprocessing and draw comparisons to the clean model system (oxide catalyst, S and N free).

Pyridine and quinoline are typical of the heterocyclic nitrogen compounds found in petroleum feedstocks. The lone pair of electrons associated with the nitrogen atom contributes to the basicity of the molecules enabling them to adsorb on sites on CoMo/Al 2 03 with both

Bronsted and Lewis acid character (Ratnasamy and Knozinger, 1978). When present with other heteroatom species they competitively adsorb on active sites and inhibit both hydrodesulfurization (HDS) and hydrodenitrogenation

(HDN) reactions (Bhinde, 1979). - 154 -

The intrinsic kinetics of Ni-T34PP demetallation on oxide

CoMo/Al203 in the presence of 100 ppm N as pyridine have been investigated. Pyridine was added to the porphyrin/oil mixture prior to satura-

tion with hydrogen. The concentration versus contact time profile for reaction at 345*C and 6.99 MPa H 2 (1000 psig) is presented in Figure

IV-l. Experimental points are represented by the different symbols.

Feed porphyrin was rapidly removed in comparison to the total metal, producing the hydrogenated nickel chlorin (Ni-PH2) and nickel isobacterio-

chlorin (Ni-PH4 ) species. At short contact times (W/Q < 0.06 g cat hr/ml)

the metal balance closed with these species, but at higher conversions

the non-porphyrinic Ni-X intermediate was again present. In contrast to

the results obtained in the absence of pyridine, the non-porphyrinic

intermediate constituted no more than 40% of the total metal in the oil

at any time. With the clean oxide catalyst, as much as 80% of the metal

in the oil was present as Ni-X at high feed conversions.

Analysis of effluent oil samples by UV-visible spectroscopy and

HPLC revealed no additional nickel bearing species other than in trace

quantities. This suggests the demetallation mechanism has not been

changed by the presence of this basic nitrogen compound.

Kinetic rate parameters were determined from the experimental

data in Figure IV-1 using the sequential hydrogenation/hydrogenolysis model for Ni-T3NPP demetallation developed in Section III.A.2. Pre-

sented in Table IV-1 are the seven rate coefficients used in the calcu-

lated solid curves in Figure IV-1 and the corresponding rate coefficients

previously reported in the absence of pyridine. An inhibition is ob-

served for all reaction steps in the presence of this nitrogen compound. - 155 -

A TOTAL Ni

70 o Ni-P o Ni-PH e Ni-PH 60. Ni-X

0- 40-

-J Li.J LU0

20 - 200

0.0 0.1 0.2 0.3 0.4 0.5 0.6 G CAT. HR W/Q, LOL

Figure IV-l. Concentration versus contact time results for Ni-T3MPP demetallation at 71 ppm Ni feed, 345 'C, and 6.99 MPa H 2 (1000 psig) on the oxide catalyst with 100 ppm N as pyridine in feed. Solid lines represent model calculations using parameter values in Table IV-l. - 156 -

TABLE IV-1

Ni-tetra(3-methylphenyl)porphyrin Kinetic Parameters

in the Presence of Pyridine

1 3 7 Ni-P , Ni-PH ' Ni-PH -+- deposit 2 4

5 Ni-X -+ deposit 6

Reaction conditions: 345 0 C, 6.99 MPa E 2 (1000 psig)

oxide CoMo/Al2 03

66 ppm Ni 71 ppm Ni k pyridine 0 ppm N 100 ppm N k oxide k, ml/g cat hr 105.0 44.0 0.42

94.0 53.0 0.56 120.0 42.0 0.35 150.0 78.0 0.52 49.0 7.20 0.15 3.80 2.30 0.61 18.0 14.0 0.78 - 157 -

Satterfield et al. (1980) have reported a similar decrease in both C-S hydrogenolysis and double bond hydrogenation rates for thiophene

HDS in the presence of pyridine. Likewise, Bhinde (1979) reported a decrease in all steps in the hydrodesulfurization of dibenzothiophene in the presence of quinoline. Both investigators reported stronger

inhibition of the hydrogenation sites compared to the C-S bond scission

sites, suggesting that two types of sites poisoned to different extents are present on sulfided CoMo/Al203 and sulfided NiMo/Al2 03' The degree of inhibition observed for Ni-T3MPP demetallation de-

pends in a similar fashion on the nature of the reaction type suggesting

multiple sites exist on the oxide form of CoMo/Al2 0 The hydrogenation/

dehydrogenation steps (k1 to k4 ) were all reduced approximately 50% at

100 ppm N whereas the metal deposition steps (k6 + k7 ) were inhibited to

a lesser extent.

Ratnasamy and Knozinger (1978) report the majority of pyridine

adsorbed on the oxide form of CoMo catalysts is associated with Lewis

acid sites such as molybdenum and cobalt vacancies which are thought to

be sites for hydrogenation. Bronsted sites (speculated to possess

hydrogenolysis activity) associated with hydroxyl groups on the alumina

and Mo showed less affinity for pyridine. Hence the greater reduction

observed in hydrogenation activity compared to metal deposition activity

is consistent with the preferred adsorption of pyridine on the hydro-

genation sites.

The observed small contribution of Ni-X to the total pool of

nickel previously discussed reflects a decline in the reaction select-

ivity of Ni-PH4 to Ni-X (k5 /k7). The larger decrease in the rate

coefficient for Ni-X production compared to direct deposition is the - 158 -

reason for this selectivity drop. Recalling the mechanism for Ni-X formation proposed in Section III.A.2, both hydrogenation and ring contraction by an acid catalyzed reaction are required. Pyridine has been demonstrated to deactivate both of these catalytic functions so the strong inhibition in Ni-X formation is not surprising.

IV.B. Demetallation on the Sulfided Catalyst

IV.B.l. Reaction Pathway and Intrinsic Kinetics

The incorporation of sulfur into the CoMo/Al203 catalyst was achieved by pre-sulfiding with a 10% H2S/H gas mixture prior to intro- 2 H2 gamitrprotoit- duction of the oil. Carbon disulfide was also added intentionally to the oil (0.05 wt%) to provide a source of sulfur to maintain constant activity of the catalyst. Under reaction conditions the CS2 was rapidly converted to H 2S. A complete discussion of this sulfiding procedure may be found in Section II.A.4.

Kinetic results on the sulfided catalyst have been obtained for

Ni-tetra(3-methylphenyl)porphyrin at 60 ppm Ni over a range of temperatures (285C - 345*C) and pressures (4.24 - 10.44 MPa H 2 ) similar to that reported for the oxide catalyst. Analysis of effluent oil samples has again revealed the presence of only those intermediates that were produced with the oxide catalyst. This is evident in the concentration versus contact time plot presented in Figure IV-2 for reaction at 345*C.

Similar arguments used to describe the concentration behavior of nickel species on the oxide catalyst are reiterated here: rapid feed porphyrin removal occurring concurrently with the production of hydrogenated - 159 -

70 i

&TOTA L Ni 601 o Ni-P o Ni-P-H2 e Ni-P H4 50 e Ni-X

0~ 40 0~

-J LU

0 30 - 2

10 e 0

r 0.0 0.05 0.10 0.15 GCAT. HR W/Q ML OIL

Figure IV-2. Concentration versus contact time results for Ni-T3MPP demetallation at 63 ppm Ni feed, 345 *C, and 6.99 MPa H2 (1000 psig) on the sulfided catalyst. Solid lines represent model calculations using parameter values in Table IV-2. - 160 -

intermediates; establishment of a dynamic equilibrium between the feed and hydrogenated porphyrins; and initially (after a slight induction period) a rapid rate of total metal removal followed by, at larger contact times, a slower rate of total metal removal from the non- porphyrinic nickel intermediate. The similarity in reaction mechanism on the sulfided and oxide CoMo/Al203 catalyst is apparent. This demonstrates the validity of investigating such reactions on the oxide catalyst and eliminates any concern that may exist in extrapolating to the more complex sulfided system.

Further comparison between the oxide and sulfided catalyst data at 345*C reveals an enhancement in reaction rates on the sulfided catalyst.

Approximately 80% total nickel removal is achieved at W/Q = 0.075 g cat hr/ml on the sulfided catalyst whereas the equivalent conversion on the oxide catalyst required a contact time of 0.42 g cat hr/ml (See Figure

III-10). Results obtained at 285*C on the sulfided catalyst presented in Figure IV-3 exhibit a similar enhancement in total metal removal rate compared to the oxide catalyst at 285*C (oxide data in Figure III-11).

Increases in hydrodesulfurization reactions are common on sulfided catalysts and are attributed to molybdenum sulfides being the active catalytic species. Gissy et al. (1980) and Daly (1978) reported substantially greater benzothiophene conversion on the pre-sulfided catalyst compared to the fresh oxide CoMo/Al 2 03 catalyst. Similarly, the presence of H 2S has been reported by Rankel and Rollmann (1983) to increase the demetallation of vanadyl-tetraphenylporphyrin on an aged

CoMo/Al2 03 catalyst.

At all temperatures the selectivity of Ni-PH4 to Ni-X relative to direct metal deposition (k5/k ) was smaller on the sulfided catalyst - 161 -

60 TOTAL Ni o Ni-P o Ni-PH _ A e Ni-PH 4 Ni-X

- 0. L40- -J Lu -30 - z

20- 0

2 10

0.0 0.1 0.2 0.3 O. 0.5

GCAT. HR W/Q ML OIL

Figure IV-3. Concentration versus contact time results for Ni-T3MPP at 62 ppm Ni feed, 285 *C, and 6.99 MPa H 2 (1000 psig) on the sulfided catalyst. Solid lines represent model calculations using parameter values in Table IV-2. - 162 -

compared to the oxide catalyst. This is reflected most noticeably in the concentration profiles on the sulfided catalyst at 2850 C in Figure

IV-3. At 90% Ni-T3MPP conversion, only 20% of the metal in the oil was present as Ni-X whereas on the oxide catalyst (Figure III-11) at an equivalent feed conversion this non-porphyrinic intermediate accounted for 65% of the metal. This lower selectivity (k5 /k7 ) could be attributed to an enhanced metal deposition activity on the sulfided catalyst.

The importance of acid sites reported for hydrogenolysis reactions (Maternova, 1983, Katzer et al., 1980, Nelson and Levy, 1979) and proposed for Ni-X formation suggests that changes in surface acidity are a contributing factor to these observed reactivity differences on the oxide and sulfided catalysts. Topsoe (1980) and Sivasanker et al. (1980) have reported the disappearance of strong Bronsted sites associated with surface hydroxyl groups resulting from sulfiding CoMo/Al203 while new Bronsted sites associated with the generation of sulfhydryl groups have been reported by Maternova (1983, 1982). Similarly, Lewis acid sites present as anion vacancies on both catalysts are more prevalent on the sulfided form (Sivasanker et al., 1980) in the absence of H2 S.

Stronger acidity by itself does not necessarily correspond to greater activity. Competing reactions such as coke formation by carbonium ion pathways are also promoted by strong acid sites but lead to site deactivation.

The enhanced metal deposition activity observed with the sulfided catalyst may be the result of new Bronsted sites arising from the dissociation of H S adsorbed on molybdenum vacancies (Lewis sites) on the 2 catalyst. Yang and Satterfield (1983) have recently proposed that molybdenum surface anion vacancies (sites for hydrogenation and - 163 -

hydrogenolysis) on the sulfided catalyst are transformed into Bronsted acid sites (responsible for hydrogenolysis and isomerization) by the dissociation of H2 S into protons and sulfhydryl groups. The presence of H 2 S provides a source for continually generating new sites. No such transformation of sites occurs in the oxide system at reaction conditions. It is apparent that the number of active sites and their stability toward deactivation are of crucial importance in determining the observed reactivity on the oxide and sulfided catalysts.

Further evidence for the parallel metal deposition mechanism proposed for Ni-T3MPP in the previous chapter is obtained by considering both the low and high temperature data on the sulfided catalyst. At

2854C the metal balance is clearly closed with Ni-P, Ni-PH and Ni-PH at short contact times (W/Q < 0.05 g cat hr/ml) yet measurable quantities of metal are removed in this period without reacting through the

Ni-X intermediate. The maximum in metal removal rate is observed to correspond to the maximum in the Ni-PH concentration. The data at

345*C in Figure IV-2 shows a similar rapid metal removal rate at short contact times but at W/Q > 0.05 g cat hr/ml a transformation occurs.

The majority of metal is present as Ni-X and the total metal removal rate parallels the disappearance of this species in a first order fashion.

Metal removal from more than one intermediate is required to rationalize these observed results.

Kinetic modelling of the experimental concentration profiles in terms of the sequential reaction scheme has been attempted. Parameters used in the model solid curves of Figures IV-2 and IV-3 are presented in Table IV-2 with the corresponding activation energies obtained from the Arrhenius plots of Figure IV-4. Comparison of the relative magnitude - 164 -

TABLE IV-2

Ni-tetra(3-methylphenyl)porphyrin Kinetic Parameters on

Sulfided Catalyst.

1 3 7 z Ni-PH -- deposit Ni-P Ni-PH 2,"' 4 2 4

5 Ni-X -+ deposit 6

Reaction Conditions: 6.99 MPa H (1000 psig) 2 sulfided Co~o/Al 20 360 ppm Ni

2850 C 3450C E, kcal/mole k, ml oil/g cat hr 22.0 183.0 23.0 k 24.0 164.0 24.9 2 46.0 250.0 19.7 59.0 400.0 23.1 4.80 171.0 40.3 2.00 27.0 29.4 26.0 120.0 20.0 - 165 -

1000

100 -

I 1.6 1.7 1.8 1/T x le), K~

Figure IV-4. Arrhenius plots for the first order rate parameters for Ni-T3MPP at 60 ppm Ni feed and 6.99 MPa H2 (1000 psig) on the sulfided catalyst. - 166 -

of the individual rate parameters reveals that the metal deposition steps

(k6 , k7 ) are rate limiting in the overall demetallation scheme. Metal deposition was likewise rate limiting on the oxide catalyst. Hydro- genation/dehydrogenation reactions are facile on both catalysts compared to the hydrogenolysis steps although the magnitude of the difference between the two types of reactions is diminished on the sulfided catalyst.

Activation energies for the hydrogenation/dehydrogenation steps

(k1 to k ) and metal deposition steps (k6 , k7 ) are similar on both the oxide and sulfided catalysts. A substantially larger value, twice that on the oxide catalyst, was obtained for the activation energy of the step producing Ni-X (k5). A consequence of this large activation energy is that the relative rate of reaction of Ni-PH4 by direct deposition

(k7 ) compared to production of Ni-X (k5 ) is strongly temperature dependent on the sulfided catalyst. The activation energy difference E - E = 20.3 kcal/mole favors direct deposition from Ni-PH at low temperatures on the sulfided catalyst. Only a minor variation in relative rates (E7 - E5 = 4.3 kcal/mole) was observed with the oxide system with

Ni-X production, rather than direct deposition, being favored at low temperatures.

The selective enhancement observed in metal deposition reactions upon sulfiding the oxide catalyst is quantified in Table IV-3 where relative rates for each step at 345*C are presented. Also included is data obtained onapre-reduced catalyst prepared by in-situ treatment with hydrogen at 450*C for 4 hours prior to contact with the Ni-T3MPP/ oil mixture. Pre-reduction is shown to increase by approximately two- fold the rates of all steps irrespective of the nature of the reaction.

A similar doubling in the hydrogenation/dehydrogenation rates is - 167 -

TABLE IV-3

Variation in Ni-tetra(3-methylphenyl)porphyrin Kinetic Parameters

on Pre-treated Catalysts

1 3 7 Ni-P Ni-PH2 Ni-PH4 -3p- deposit 2 4

5 Ni-X -+- deposit 6

Reaction conditions: 345*C,6.99 MPa H2 (1000 psig)

60 ppm Ni

kreduced/ ksulfided k oxide k Reaction Step oxide

2.5 1.7 2.1 1.7 1.9 2.0 1.4 2.6 2.5 3.5 2.2 7.1 2.2 6.7 - 168 -

Qbserved on the sulfided catalyst suggesting this may likewise be due to the simultaneous reduction of the catalyst which occurs upon sulfiding (Chin and Hercules, 1982). Such treatment results in creation of anion vacancies on the reduced molybdenum atoms which have been attributed to possess hydrogenation activity (Grange, 1980).

The more pronounced impact of sulfiding the catalyst which is not seen by pre-reduction, is the dramatic increase in activity for bond cleavage reactions when compared to the oxide system. The metal deposition steps (k6 , k7 ) are shown to increase by seven-fold in the sulfided environment. This selective enhancement in hydrogenolysis rates is similar to the effect seen with H 2S in the hydrodenitrogenation, of quinoline reported by Satterfield and Gultekin (1981). If it is assumed these bond cleavage reactions are associated with Bronsted sites on the catalyst, this selective enhancement is consistent with an increase in

Bronsted acidity on the surface due to H 2S dissociation recently proposed by Yang and Satterfield (1983).

The hydrogen pressure dependence of the kinetic rate parameters was determined over the pressure range 4.24 MPa H2 to 10.44 MPa H 2 (600 - 1500 psig) at 345*C from the plots in Figure IV-5. (The behavior at 285*C and 320*C was identical to that at 345*C). Most results were similar to those reported for Ni-T3NPP on the oxide catalyst. In this pressure regime on the sulfided catalyst, the hydrogenation steps k1 and k3 were first order in hydrogen pressure, consistent with the stoichiometry of the reactions, whereas the dehydrogenation steps k2 and k showed no hydrogen pressure dependence. Reaction steps k and k in 4 5 6 Figure IV-5 exhibited a pressure dependence of 1.8 and 2.2 at 345 0 C.

It may be concluded that both of these steps approximate second order - 169 -

1000 I I I I

A k 100 3

-j I

10-

600 psig 1000 PsIg 1500 psig

i I I I II I I 10.- 1.5 2.0 2.5 LN P (MPa H2 )

Figure IV-5. Hydrogen pressure dependence for the first order rate parameters for Ni-tetra(3-methylphenyl)porphyrin at 60 ppm Ni feed and 345 *C on the sulfided catalyst. - 170 -

hydrogen dependence and follow the same behavior observed on the oxide

catalyst. The direct metal deposition step from Ni-PH4 (k ) was de-

termined to have a third order dependence on hydrogen pressure at all

temperatures. The analogous step on the oxide catalyst was determined

to have a second order dependence. This discrepancy may reflect a subtle

change in the extent of hydrogenation occurring within the rapid se-

quence of reactions comprising the apparent direct deposition step.

The stronger hydrogenating ability of the sulfided catalyst has been

demonstrated by the relatively faster rate of feed porphyrin removal on

the sulfided catalyst compared to the oxide catalyst.

IV.B.2 Metal Deposition Results and Discussion

Metal deposition profiles obtained under diffusion conditions on

the sulfided catalyst are presented in this section. The similarities

in the reaction mechanism of Ni-T3MPP observed on the oxide and sulfided

catalysts are reflected in the metal profiles. Likewise, the selective

enhancement in the rate of metal deposition on the sulfided catalyst is

apparent in the metal profiles.

Shown in Figure IV-6 are radial nickel deposition profiles in

catalyst pellets as a function of reactor axial position obtained with

Ni-T3MPP at 345*C and 6.99 MPa H2 (1000 psig). The experimental data is

represented by the open circles. The internal maximum in metal concen-

tration at the entrance of the bed (M-shape profile) resulting from the

sequential reaction scheme shifts rapidly to the edge of the pellet

(U-shape profile) a short distance into the reactor as the concentration

of metal depositing species in the bulk increases. Throughout the - 171 -

4 4 ENTRANCE FIRST QUARTER

3 3 0

-j IJ

z 2 0 0 o0 z 2 0 H- 0 H- - __ 0 0 0 - I 0 I 0 00 0 0 0 0 0000

10 0.8 0.6 0.4 0.2 0 1.0 0.8 0.6 OA 0.2 0 rIR /R

4 4 MIDDLE EXIT

3 3

-J .. -LJ - C) 2 00 2 F- 0 H: 0

I 0 1 0 0 0 0 0 0 0 00 - 0 0 00 0 000

1.0 0.8 0.6 0.4 0.2 0 1.0 0.8 0.6 OA 0.2 0 T/R r/R

Figure IV-6. Nickel deposition profiles in 1/16" diameter catalyst pellets at various reactor axial positions from Ni-tetra(3-methylphenyl)porphyrin demetallation at 345 *C and 6.99 MPa H2 (1000 psig) on the sulfided catalyst. Solid lines represent model calculations using parameter values in Table IV-4. - 172 -

remainder of the reactor the conventional U-shape profile remains although becoming increasingly smaller as metal is depleted from the oil.

The profiles once again are dominated by a strong diffusion limited component resulting in the steep metal gradient at the edge of the pellets and a relatively diffusion free component resulting in the more uniform metal concentrations at the center of the pellets. Both characteristics were present throughout the reactor and provide further evidence for the parallel deposition routes in the demetallation scheme for Ni-T3MPP.

The selective enhancement in the metal deposition steps observed upon

sulfiding the catalyst is apparent in the sharper profiles and in the proximity of the internal maximum to the edge of the pellet (r/R = 0.93)

compared with the oxide results (Figure 111-21, r/R = 0.75) at identical

conditions.

Metal profiles have similarly been obtained with Ni-T3MPP at 345*C and a higher hydrogen pressure of 10.44 MPa H2 (1500 psig). These re-

sults are presented as the open circles in Figure IV-7. The higher

order hydrogen dependence of the metal deposition steps discussed in

the section on intrinsic reactivity (Section IV.B.1) results in a selec-

tive enhancement of these reactions as the pressure is raised. This

is clearly reflected in the very steep metal profiles throughout the

bed and in the inability to experimentally resolve the internal maximum

in catalyst pellets at the entrance of the reactor. Activity of the

catalyst under these conditions is sufficient to achieve over 90% metal

removal by the end of the reactor resulting in the low metal loadings

observed in pellets taken from the exit.

All metal profiles have been modelled using the coupled diffusion

and kinetic reaction scheme developed in Section III.C.2. The unknowns - 173 -

9 - 9 0 ENTRANCE FIRST 8 QUARTER

6 - 6 -j Lii 5- 5 C)

4- 0 H1-0 4 0I 3- 3

0 2- 2 0

I 0o

0 1 000 1.0 0.8 0.6 04 0.2 0 Z- 1.0 0.8 0.6 04 0.2 0 1/R r/R

8 8 MIDDLE EXIT 7 - 7-

6~ 6-

5 -J 5-

z 4 2 4-

3-0 3-

2- 2 -

1000 0

00 1. O.6 0.6 0.4 0.2 0 1.0 0.8 0.6 OA 0.2 o fI/R I/R

Figure IV-7. Nickel deposition profiles in 1/16" catalyst pellets at various reactor axial positions from Ni-tetra(3-methylphenyl)porphyrin demetallation at 345 *C and 10.4 MPa H2 (1500 psig) on the sulfided catalyst. Solid lines represent model calculations using parameter values in Table IV-5. - 174 -

in this steady state model are the seven kinetic rate parameters and the four species diffusion coefficients. These parameters are all assumed to remain constant on the basis that the low level of metal deposited is not sufficient to poison a substantial number of catalytic sites or constrict diffusion within the pores of the pellets. Having independently determined the kinetic parameters in the intrinsic reaction studies, the four diffusion coefficients are left as the only unknowns.

Estimates for these parameters have previously been obtained in the metal profile modelling on the oxide catalyst. For consistency of the model, these should be applicable on the sulfided catalyst as well.

First attempts to model the metal profiles obtained at 345*C and

6.99 MPa H2 (1000 psig) were with the corresponding intrinsic kinetic parameters and the diffusion coefficient values used in modelling the oxide catalyst profiles at identical conditions. This set of parameters resulted in a calculated concentration of Ni-X 2.5 times higher than experimentally measured in the effluent and poor agreement between the measured and calculated total metal effluent. Likewise, the model consistently predicted a lower metal loading at the center of pellets than had been measured.

These discrepancies were both corrected by adjustment of the -6 2 diffusivity of Ni-X. A value of 3.5 x 10 cm /sec resulted in a more satisfactory agreement between experiment and theory than the value of

2.0 x 10-6 used on the oxide catalyst. The remaining three diffusivities for Ni-P, Ni-PH and Ni-PH4 were unchanged. The complete set of parameters used in generating the solid model curves in Figure IV-6 are listed in Table IV-4. The features of the experimentally determined profiles are both qualitatively and quantitatively reproduced with the - 175 -

TABLE IV-4

Parameters for Ni-tetra(3-methylphenyl)porphyrin Metal Profile

Modelling (Figure IV-6) ki = 183.0 ml oil/g cat hr t = 84 hrs k2 = 164.0 p = 0.88 g/ml k3 = 250.0 PC = 1.49 g/ml k = 400.0 Q = 20 ml/hr k5 = 171.0 W = 5.0 g k6 = 27.0 R = 0.076 cm k = 120.0 Co = 60 ppm Ni

D = 1 x 10-6 cm2/sec Effluent Oil Concentration 1 - 2 x 10-6 D2 Experimental Calculated D = 2 x 106 Ni-P 5.6 ppm Ni 3.6 D = 3.5 x 106 Ni-PH 2.2 2.3 2 Ni-PH 4 1.0 0.9 N i-X 2.8 4.7 11.6 11.5 - 176 -

model serving as evidence in support of the reaction scheme developed for Ni-T3MPP and of the accuracy of the kinetic parameters.

The calculated metal profiles of Figure IV-7 at the higher hydrogen pressure were determined by a similar procedure using the parameter set in Table IV-5. In this instance the model was used in a truly predic-

tive fashion. No adjustment of parameters was required to obtain the

excellent agreement between the experimental and theoretical profiles

throughout the reactor as well as the match in effluent oil composition.

The intrinsic kinetic parameters determined on crushed catalyst at the higher hydrogen pressure (10.44 MPa H2 ) were used in conjunction with

the values for the four diffusion coefficients used at 6.99 MPa H2 on

the sulfided catalyst.

The consistency of the diffusion coefficients for the four nickel

species present in the oil with changes in the nature of the catalyst

and operating conditions is encouraging in terms of verifying the re-

liability of the model. Liquid phase diffusivities are relatively

insensitive to pressure changes of this magnitude. Similarly, a varia-

tion of these parameters with a change from the oxide to the sulfided

catalyst would only be expected if solid support interactions such as

adsorption or surface diffusion contributed in a substantially different manner to the apparent diffusion coefficient. The difference in value

for the diffusion coefficient of Ni-X on the oxide and sulfided catalysts

while not large, was necessary to achieve the better fit. Similarly,

using this larger value to interpret the metal profiles on the oxide

catalyst in Section III.C.2 resulted in poorer agreement. No

phenomenological explanation as to why this one diffusion coefficient

should vary on the two catalysts was evident. A larger value for the - 177 -

TABLE IV-5

Parameters for Ni-tetra(3-methylphenyl)porphyrin Metal Profile

Modelling (Figure IV-7)

= 240.0 ml oil/g cat hr t = 69.5 hrs k 1 = 150.0 p = 0.88 g/ml = 400.0 PC = 1.49 g/ml = 450.0 Q = 25 ml/hr = 246.0 W = 5.0 g k5 = 44.0 R = 0.076 cm = 560.0 Co =61 ppm Ni

= 1 x 10--6 6 cm 2/S2 Effluent Oil Concentration x 10-6 = 2 Experimental Calculated = 2 x 10 Ni-P 2.3 ppm Ni 2.3 = 3.5 x 10-6 Ni-PH2 1.0 1.3

Ni-PH 4 0.5 0.4 Ni-X 1.2 2.2 5.0 6.2 - 178 -

structurally smaller Ni-X molecule is consistent from a hydrodynamic

standpoint but the magnitude of the change is not large and can not be realistically interpreted for a lumped parameter such as this.

IV.C. Comments on the Influence of Nitrogen and Sulfur Compounds

The results of this chapter have demonstrated the demetallation

pathway of Ni-T3MPP on a CoMo/Al203 catalyst is similar in the presence

or absence of nitrogen and sulfur compounds. Reaction occurs through

the same stable intermediates with multiple hydrogenation preceding the

generally slower, rate limiting, metal deposition steps in the overall

sequence. The in-depth investigation of the reactivity of Ni-T3MPP on

the sulfided catalyst has revealed temperature and hydrogen pressure

dependencies for the individual steps comprising the demetallation se-

quence similar to those previously discussed for the oxide system. The

consistency of the proposed demetallation pathway has been further

demonstrated by the success achieved in modelling the pattern of metal

deposition in catalyst pellets throughout the reactor in both the oxide

and sulfided systems. Agreement such as this justifies the extrapolation

of information obtained on demetallation pathways with the clean (N and

S free) oxide system to more complex environments.

While no mechanistic differences have been observed by the addi-

tion of N and S compounds, a potentially important difference in

reaction selectivity has been noted. The relative hydrogenation to

hydrogenolysis rates within the demetallation sequence in the presence

of nitrogen and sulfur were found to vary substantially compared to the

clean oxide system. Basic nitrogen in the form of pyridine was observed - 179 -

to strongly deactivate the hydrogenation sites (k to k ) but showed 1 4 less of an inhibition on the metal deposition steps (k6 and k7). Bhinde

(1979) and Satterfield et al. (1980) have previously discussed the implications of a similar situation where pyridine could be used to improve the selectivity of a catalyst for hydrodesulfurization (i.e. hydrogenolysis) relative to hydrogenation. Similarly, sulfiding the

CoMo/Al203 catalyst was shown to enhance the metal deposition steps

(k6 and k 7 ) by a factor of seven compared to only a two-fold increase in the hydrogenation type reactions (k1 to k . This leads to the conclusion that at least two separate types of sites are responsible for the overall demetallation reaction. The presence of N and S was also shown to decrease the reaction selectivity of Ni-PH4 to Ni-X (k5 /k7 ).

The relatively slow demetallation of this non-porphyrinic species (k6 ) suggests a high selectivity is desirable from the standpoint of obtaining uniform metal deposits in catalyst pellets.

An important consequence of this catalyst dual functionality and the ability to alter the relative hydrogenation/hydrogenolysis rates within the demetallation sequence is the potential improvement in catalyst performance and lifetime attainable through control of metal deposition location. The influence this hydrogenation/hydrogenolysis selectivity has in determining the intrapellet metal distribution can be substantial as has been demonstrated by the differences between the oxide and sulfided catalyst metal profiles. The apparent relationship of this reaction selectivity to the acidic nature of the catalytic sites which has been intimated by the nitrogen and sulfur results will be explored in the next chapter. - 180 -

CHAPTER V

VARIATIONS IN NI-T3MPP REACTION SELECTIVITY ON MODIFIED

CoMo/Al203 CATALYSTS.

V.A. Introduction

Demetallation reactions occurring under strongly diffusion limited conditions are characterized by rapid catalyst deactivation due to a concentration of deposited metals at the catalyst periphery, excluding access

to the inner unpoisoned regions of the catalyst. Maximization of a

catalyst'scapacity for metals deposits and, consequently, maximization of

its useful lifetime would clearly be achieved by a uniform deposition of metal throughout the catalyst. Attempts to achieve this, not at the

expense of lower conversions at lower temperatures, have often involved physical alterations of the catalyst structure. Shape and size variations

and incorporation of larger pores into the catalyst through multi-modal

pore size distributions have demonstrated the potential for increasing

the metals capacity of a catalyst (Tamm et al., 1981, Riley, 1978,

Plumail et al., 1983).

An alternative approach which by comparison has received far less

attention is to control the selectivity of the hydrodemetallation reac-

tions by modifying, the chemical composition of the catalyst in order to

potentially control the location of metal deposition. Two mechanisti- - 181 -

cally different reactions (hydrogenation and hydrogenolysis) occurring on different sites have been demonstrated to comprise the reaction network of the model system. It is reasonable to speculate that one type of reaction can be enhanced or inhibited with respect to the other by suitable modification of the catalyst sites through alteration of their chemical environment. That is, the selectivity of the reaction mechanism may be controlled.

Numerous attempts and methods for altering the catalytic properties of Co(Ni)Mo/Al203 hydrotreating catalysts have been reported. These generally have been motivated by the desire to control one function of the catalyst to improve activity and selectivity for hydrodesulfurization (HDS) and hydrodenitrogenation (HDN). These reactions, while similar to hydrodemetallation (HDM), are different in one important aspect. They are not inherently deactivating. The heteroatom is removed as a gaseous product (H2S, NH3 ) rather than deposited as a perm- anent poison on the catalyst surface. Hence modification of catalyst selectivity is often directed toward minimization of cracking activity to reduce coking rates and improvements in hydrogenation selectivity to lower hydrogen consumption. Understanding the control and modification of these catalytic functions may however prove to be useful in controlling the selectivity of hydrodemetallation reactions.

In this investigation, a series of modified catalysts were prepared to examine the influence of surface treatment with dopants of varying electronic character on the HDM reaction network of Ni-T3MPP. The additives included Cs, Na, S, I, and Cl with the oxide CoMo/A12 03 catalyst

(HDS-16A) as the base case or reference catalyst. The catalysts were - 182 -

prepared by modifying the existing commercial catalyst rather than the alumina support prior to Co and Mo impregnation.

The method of addition may alter the influence the additive exerts in controlling surface properties of the catalyst but there is contro- versy in the literature as to whether this difference is always detect- able in catalytic activity measurements. Massoth and co-workers

(MuraliDhar et al., 1981, Wiser et al., 1980) in a systematic study on the effect of dopants of varying acidity on the desulfurization, hydrogenation, and cracking activity of modified CoMo/Al203 catalysts, observed no difference based on the order of addition. Basic additives such as Na and Ca at low loadings (0.5 wt%) resulted in a lowering of both HDS and cracking activity. This was interpreted on the basis that these reactions are related to catalyst acidity. Similarly, the addition of F and Cl increased the HDS and cracking rates. Hydrogenation activity was the least sensitive function to be affected by the additives.

Boorman et al. (1982) observed similar patterns of behavior but unlike

Massoth, the magnitude of the change was dependent on the order of addition of the Na and F dopants.

The presence of additives such as Na (Ratnasamy et al., 1974,

Ramaswamy et al., 1976), P (Stanulonis and Pederson, 1980, Fitz and Rase,

1983), and F (Boorman et al., 1982) on the support are known to affect the surface acidity, the support/metal interaction, and the coordination of Co and Mo after impregnation. An extensive investigation probing the surface structure and properties of Co-Mo catalysts supported on y-Al203 doped with alkali metals (LiNa,K,Rb, and Cs) has been conducted by Lycourghiotis (1980, 1980a) and co-workers (Kordulis et al., 1982).

Activity measurements in the HDS of thiophene on these catalysts have - 183 -

been correlated with modified surface characteristics (Lycourghiotis et al., 1982, 1982a). All modifiers were observed to neutralize alumina surface acidity, with complete supression at the higher loadings (1.1 mmole/g). Similarly, the transformation of Mo(VI) from octahedral to tetrahedral coordination and the dispersion of Co as determined by the relative amounts of CoAl204 and Co304 were influenced by these dopants.

Generation of such Co-aluminate species draws Co from the oxide Co-Mo bi-layer which is thought to be the oxide precursor to the active catalyst (Gajardo et al., 1980). Likewise, reduction of Mo(VI) to Mo

(IV) occurs with octahedral Mo whereas tetrahedral Mo is reportedly only reducible to Mo(V) (Chin and Hercules, 1982). Generation of MoS2 and anion vacancies on Mo, speculated to be the active sites for HDS and hydrogenation reactions on the sulfided catalyst, are therefore facili-

tated by octahedral Mo.

In contrast, the addition of dopants to the calcined catalyst is not expected to significantly influence the structural characteristics of

the Co and Mo phases already present on the catalyst. Coordination of

dopant with sites on these metals is however likely, in addition to

dopant interaction with the support, potentially modifying the activity

of these metals. Similarly, modification of the total surface acidity

can be achieved by adding dopants of various acid character. A decrease

in surface acidity has been reported by Martinez and Mitchell (1980) upon

addition of the basic component Mg to a CoMo/Al203 catalyst.

V.B Characterization of Prepared Catalysts

The doping procedures discussed in Section II.A.4 to prepare the - 184 -

catalyst samples were determined to be reliable for obtaining catalysts with no internal gradients. Scanning Electron Microscope (SEM) analysis of catalyst particles revealed a uniform dopant concentration (to within the 1 micron resolution of the instrument) in all samples.

The concentration of dopant, listed in Table V-1, varied substantially from sample to sample when considered on a weight basis. When these concentrations were calculated on a comparable basis, i.e., mmole dopant/g CoMo catalyst, the additive levels were essentially equivalent.

Similarly, the surface areas of the doped samples were approximately the same despite the high additive loading in some instances, when calculated on the basis of weight of Co-Mo oxide catalyst.

Variations in additive concentration were possible but not thoroughly investigated to determine minimum requirements for an effect. The cesium and sodium levels were controlled by using the appropriate concentration in the impregnating solution. The sulfur, iodine, and chlorine levels could be controlled by varying the pre-treatment time although no attempt was made to correlate the additive level with pre-treatment time.

Despite the ability to vary the amount of dopant added to the catalyst, there was no attempt to investigate the dependence of catalytic activity on the level of dopant concentration. The molar concentration of each additive on the catalyst was comparable to the molar concentration of each of the active metals (Co, Mo). This insured that their influence would not be overshadowed by the activity of the base case oxide catalyst, which would be the case if the dopant were present in trace quantities.- The sodium level was chosen to exceed the concentration required to neutralize acid sites on alumina (1.1 mmole/g Y-Al2 03 determined by Lycourghiotis et al. (1980a). The 4.30 wt% loading a44,ww 4w w 10 wo 0 0

TABLE V-1

Modified Catalyst Properties

dopant concentration surface area

wt % mmole/g CoMo cat. m /g m 2/g CoMo cat.

Co-Mo/Y-Al203 (HDS-16A) (a) 176 (b) 176

Cs-CoMo/Al203 22.6 2.20 130 168

Na-CoMo/Al203 4.30 1.95 175 182

U1 pre-sulfided CoMo/A1203 4.83 1.59 170 178 pre-iodized CoMo/Al203 19.0 1.85 145 179 pre-chlorided CoMo/Al203 11.5 3.6 155 175

(4.3 mmole/g A1 2 03)

in-situ chlorided CoMo/Al203 6.3 1.89 - -

Y-Al203 - 219 -

pre-chlorided y-Al 2 03 6.36 1.92 mmole/g Al 203 a) Base catalyst 5.7 wt% Co (0.79 mmole Co/g CoMo catalyst) 12.2 wt% MoO 3 (0.85 mmole Mo/g CoMo catalyst) 2 b) Surface area based on alumina is 214 m /g A1 203 - 186 -

corresponded to43% coverage of the surface area of the fresh catalyst assuming sodium was present as a monolayer after calcining as sodium oxide, Na 20, with an area of 13 A2. An equivalent atomic loading was

chosen for cesium. The surface coverage of cesium on this catalyst was

sufficient for complete coverage of the fresh catalyst assuming a mono- 02 layer and a molecular area for Cs 20 of 32 A2. At this high surface

coverage, it is unlikely that higher concentrations of cesium would

further influence the catalytic activity.

The sulfur concentration measured may represent a lower limit

attainable from the sulfiding treatment used. This sample was exposed

to air at room temperature prior to analysis, which may have led to

removal of sulfur through oxidation of the top sulfide layers (de Beer

et al., 1976).

The incorporation of the halogens into the catalyst in a reducing

atmosphere as HI and HCl at 345 0 C was similar to the sulfiding treatment

of the catalyst. This method facilitated interaction of I and Cl with

the catalytic metals through vacancies generated by reduction, and not

just interaction with the alumina support. This method was somewhat

different than the more common approach of adding halogens in the form

of their ammonium salts via an impregnation step (Wiser et al., 1980,

Boorman et al., 1982) which results mainly in addition of the halogen

to the support and not the metals. Samples of fresh CoMo/A1 2 03 catalyst

and y-Al 203 support pre-chlorided (i.e., before contact with oil) under

identical conditions resulted in a higher chlorine loading on the catalyst.

(See TAble V-1) This indicates a greater number of sites for chlorine on

the catalyst and serves as indirect evidence for chlorine addition to the

Co and Mo metals. - 187 -

Identification of the chemical nature of the halogens on the catalyst

surface was attempted using X-ray Photoelectron Spectroscopy (XPS). The

samples were not prepared in the instrument environment and came in contact with air prior to analysis. All spectra were referenced to Au 4f7/2 at

84.0 eV which was equivalent to C ls at 284.4 eV. The binding energies

for Al 2 p 74.2 = eV, Mo 3d5/2 = 232.7 eV, Mo 3d3/2 = 236.1 eV, Co 2 p3/2 2 781.7 eV, and Co p1 /2 = 797.3 eV core levels measured on the oxide

catalyst were similar to values reported by Chin and Hercules (1982).

The aluminum 2p level on the iodized catalyst shown in Figure V-1

was unchanged from the oxide catalyst indicating the Al 2 03 support was

not chemically modified by iodine. In contrast, the aluminum signal on

the chlorided catalyst displayed a shoulder at binding energies higher

than the support suggesting some chloriding of the alumina occurred.

The results for molybdenum presented in Figure V-2 show a substan-

tial increase in the complexity of the chemical state of Mo resulting

from treatment with Cl or I. The reduction of Mo(VI) present as MoO3 on

the fresh catalyst (Mo 3d5/2 = 232.7 eV) to Mo(V) (Mo 3d5/2 = 232.0 eV)

and Mo(IV) (Mo 3d = 229.0 eV) oxide species was determined by reference 5/2 values reported by Zingg et al. (1980). This was observed for both the

chlorine and iodine treated samples. A shoulder at 230.8 eV above the

Mo(IV) peak on the chlorided catalyst was consistent with the presence

of a chlorided Mo(IV) species (unsupported MoCl Mo 3d5/2 = 230.5 eV)

reported by Walton (1977). There may likewise have been chlorine inter-

action with Mo(V) but the reported binding energy for chlorided Mo(V)

(unsupported Mo 2 C 10, Mo 3d5/2 = 231.4 eV) was too close to distinguish

from the oxide Mo(V) species (Walton, 1977).

Data for Mo binding energies in Mo-I complexes was not located in - 188 -

6 ...... ------

5 ...... 74. 2

3 ......

2 ......

...... ------......

------77 76 -75 -74 -73 -72 -71 -70 -69 -68 DINDING ENERGY, EV

......

...... ------......

Li 4 ------...... 74.4 ...... 3 ------

------I......

......

-79 -77 -76 -75 -74 -73 -72 -71 -79 -69 -6 IINDINC ENERGY, EV

74 .2 Z

...... - ......

/ ......

-79 -78 -76 -7S 7 3 -72 -71 IINDIN; EWEPGY. Ey

Figure V-1. XPS Spectra of Al 2p level in fresh oxide (top), pre-iodized (middle), and pre-chlorided (bottom) catalysts. - 189 -

232.7 . b ...... - ...... [

r ...... - ...... 236.1

......

...... - ... -- - - -242 -240 -238 -236 -234 -232 -23e -229 -226 -224 BINDING ENERGY. EV

......

232.u -_ ---- _ --_ ------......

......

...... 229.0 229,0

......

------242 -240 -238 -236 -234 -232 -230 -228 -226 -224 BINDING ENERGY, EY

21.2,9

? ......

...... 1_220 6

-242 -240 -238 -236 -234 -232 -230 -229 -226 -224 BINDING EHEPCY. EV

Figure V-2. XPS Spectra of Mo 3d 5/2 and 3d 3/2 levels in fresh oxide (top), pre-iodized (middle), and pre-chlorided (bottom) catalysts. - 190 -

the literature. The similarities in the electron configuration of chlorine and iodine and in the Mo signal for the two catalyst systems suggests the presence of Mo-I species on the catalyst is possible.

Information regarding the chemical nature of cobalt species on the surface is presented in Figure V-3. The low intensity of the binding energy signal reflects a low surface concentration of cobalt. This can be interpreted in terms of the bi-layer structure of these catalysts proposed by Gajardo et al. (1980) in which Co exists as a layer beneath the Mo or diffuses into the alumina 2 matrix. The Co p3/ 2 and 2p1/2 levels were essentially unchanged from the fresh oxide catalyst on the chlorided and iodized samples. Binding energies for chlorided cobalt

2 (unsupported CoCl2 , Co p 3/2 = 783.5 eV) and iodized cobalt (unsupported

2 CoI 2, Co p 3/ 2 = 782.3 eV) reported by Frost et al. (1974) were absent.

This may in part result from the inaccessibility of cobalt to the Cl or

I due to its existence under the molybdenum layer and strong interaction with the alumina. There was likewise no indication of metallic cobalt

2 (Co p3/ 2 = 778.1 eV) present on the surface (Chin and Hercules, 1982).

This can not be considered surprising as metallic cobalt is oxidized upon exposure to air (Brinen and Armstrong, 1978).

From the XPS results it was concluded that the majority of the Cl and I was associated with Mo on the catalyst. Chlorine also appeared to be associated with Al whereas the I was not. No evidence for cobalt- halogen bonding was found.

In conjunction with the chemical characterization of the catalysts, temperature programmed ammonia desorption (TPAD) was performed on the samples and provided qualitative comparison of the total acidity and acid strength distribution on the surfaces. Listed in Table V-2 are the acid - 191 -

781.7 ......

797.3 4 ......

...... :......

2 ...... -

------......

-91S -sie eer, -;L9 7195 -79e -785 -780 IINDINQ ENERGY. EV 775 -770

.781.5

...... ------

796.7 ......

...... \ ......

...... - ......

......

I -910 _885 -M -77S -779 BINDING ENERGY. EV

7 .0

79b.0

...... - ...... -

......

...... -9 -896 _9@0 _1?5 -7?S -78S -780 -775 -770 -71S IINDING ENERGY. EV

Figure V-3. XPS Spectra of Co 2p 3/2 and 2p 1/2 levels in fresh oxide (top), pre-iodized (middle), and pre-chlorided (bottom) catalysts. - 192 -

TABLE V-2

TPAD Titration Results

2 6 * Acidic m.eq. sites/g m..eq. sites/cm x 10 T max ,*C Effluent fresh CoMo/Al203 0.549 0.31 180 no

Cs doped 0.00 0.0 - no

Na doped 0.029 0.016 - no pre-sulfided 0.546 0.32 189 yes pre-iodized 0.658 0.45 180 yes pre-chlorided 0.547 0.35 173 yes y-alumina 0.08 0.037 318** no

* Precision 5C ** Precision 10*C - 193 -

site equivalents and peak desorption temperatures for the catalysts and the y-Al 2 03 support. The peak desorption temperature was essentially the same for the fresh and modified catalysts suggesting the surfaces are primarily composed of sites of similar strength and that differences in acidity reflect a difference in number of sites. Representative ammonia desorption profiles for the oxide and pre-sulfided samples are shown in

Figure V-4. When compared to the fresh catalyst, the pre-sulfided (and pre-halogenated) catalyst shows a steeper descent on the high temperature side of the desorption peak indicative of a slightly lower concentration of strong sites. This, however, may be misleading as above 400*C acidic effluent was detected with the S, I, and Cl-treated catalysts. The on-set of desorption of acidic species from these catalysts would mask any NH 3 desorbing simultaneously. Thus the acid site concentrations for

these samples likely represent minimum values.

The use of a mass basis for acid site measurement comparisons is

somewhat inaccurate due to the wide range in dopant loadings. A more appropriate quantity, the acid site surface density, measuring the acid

site equivalents per unit surface area has been proposed by Hou and Wise

(1982). The values so determined using the catalyst surface areas in

Table V-1 are presented in Table V-2.

V.C. Kinetic Results on Modified Catalysts

V.C.l. Initial Transient Results

All catalyst charges were observed to go through an initial transient

period similar to that described in Section II.E for the oxide and - 194 -

TEMPERATURE PROGRAMMED AMMONIA DESORPTION .025

.02

I I

015

0 0 0 0 w t c0 0

.05- 00 0% 0

0 '' ~~ 0 100 200 300 400 500 6oo 700 TEMP (DEG C) TEMPERATURE PROGRAMMED AMMONIA DESORPTION 25

0 . 02-

15-

01- 0 0

C 105-

10 100 200 300 400 500 Goo 700 TEMP (DEG C)

Figure V-4. Ammonia temperature desorption profiles for the fresh oxide (top) and pre-sulfided (bottom) CoMo/Al2 03 catalysts. - 195 -

sulfided systems. The initial deactivation on the Na and Cs doped samples was not as rapid as on the oxide catalyst. This was expected considering the neutralizing influence of these components on the acid sites of the catalyst responsible for cracking and coke deposition.

The initial activity of the pre-iodized and pre-chlorided samples was not only high for demetallation but also for cracking. The viscosity of the effluent oil during the first 48 hours on stream was noticeably lower than what was common with the oxide catalyst. Eventually this thinning of the oil ceased as the strong cracking sites were presumably poisoned. There was also evidence to suggest that the iodine and chlorine content on the freshly prepared samples was not the level at steady state activity. Removal of the halogen from the surface in the high pressure hydrogen environment was apparent from the slight odor of HCl and HI in effluent samples. Runs conducted on the pre-treated catalysts did not have an I or Cl source in the feed so this stripping phenomena was not surprising. Analysis of the spent iodized catalyst after 200 hours on stream revealed an iodine content of less than 0.1 mmole I/g catalyst confirming the loss of iodine from the surface. This continuous removal of halogen from the surface prevented the maintenance of stable activity for long periods of time (>50 hours) although regions of relatively stable activity were observed.

Kinetic runs conducted with chlorine (CHCl3 ) in the feed to induce chloriding of the catalyst (i.e., during reaction with oil) did not experience this long term deactivation attributed to halogen removal.

Analysis of this in-situ chlQrided catalyst after 375 hours on stream revealed a chlorine loading equivalent to 1.89 mmoles Cl/g oxide CoMo catalyst (see Table V-1). This presumably represents a steady state - 196 -

chlorine loading on the catalyst. Kinetic data quantitatively similar to that obtained on this catalyst was obtained with the pre-chlorided catalyst after 75 hours on stream. This suggests that the fresh pre- chlorided sample was stripped to a chlorine loading similar to that of

the in-situ treated sample or that the catalytic activity was insensitive

to the dopant loading at these concentrations. The kinetic data dis-

cussed in the next Section is that obtained with CHCl3 in the feed.

In addition to the usual initial variations in activity as measured by total nickel removal, a distinct transition in effluent oil composi-

tion with time on stream was observed during the in-situ chloriding of the

catalyst. See Figure V-5. At early times (less than 80 hours) the

catalyst had not been in contact with a sufficient amount of CHC13 in

the feed to achieve stable chlorided activity. The effluent oil composi-

tion was representative of that seen with the oxide catalyst with the

porphyrinic species (Ni-P, Ni-PH and Ni-PH ) comprising only 50% of the

total metal in solution. With continued exposure to chlorine, the

activity of the catalyst changed in that Ni-P and Ni-PH2 became the only

species present in appreciable quantities at all contact times. Figure

V-5 demonstrates this activity pattern was maintained for periods as long

as 200 hours provided CHC3 was in the feed. When operating the catalyst

in this condition, it was possible to close the metal balance with these

two porphyrinic compounds. (See kinetic results in Figure V-ll and

associated discussion.) This unique ability to close the metal balance

with Ni-P and Ni-PH2 became a characteristic of the chlorided catalyst. w w

25 I I I I A

201- A A TOTALNi A m- F Ni-P, Ni-PH2,Ni-PH a- 4 A z 151-- A wz Z) LL 10 w A

5 - 8

a-LI I 0 50 100 150 200 250 300 TIME ON STREAM, HRS

Figure V-5. Initial transient behavior during in-situ chloriding of catalyst during demetallation of Ni-T3MPP at 61 ppm Ni feed, 345 *C, and 6.99 MPa H 2 (1000 psig) at W/Q = 0.35 g cat. hr/ml oil. - 198 -

V.C.2. Steady State Kinetic Results

Concentration profiles (ppm of nickel versus contact time) for

Ni-T3MPP reaction at 345*C and 6.99 MPa H2 are shown in Figures V-6 through V-ll for each of the five modified catalysts and the base case oxide catalyst. The contact times, W/Q, were calculated on the basis of

0.7 g of oxide catalyst charged to the reactor. This was necessary as

the weight of catalyst after in-situ S, I, and C1 treatment was not known. Similarly, when the weight of the modified catalyst was known

(Cs and Na doped) an amount of catalyst equivalent to 0.7 g oxide Co-Mo

catalyst was used.

There are several key points to consider in proceeding through the

six figures. The first being the kinetic characteristics revealed by

the rate of total metal removal and the relative rate of feed porphyrin

(Ni-P) removal to total metal removal. These give an indication as to

the influence of the additive on the catalytic activity and, more im- portantly, the impact on the selectivity within the demetallation sequence.

Secondly, the presence of similar nickel intermediates on all catalysts

demonstrates the mechanism of Ni-T3MPP demetallation is unchanged in

the presence of these additives. Differences in the number of intermediates can be interpreted in terms of selectivity variations within the

reaction network.

For the alkali metal doped samples, Figures V-6 and V-7, feed

porphyrin was rapidly hydrogenated forming the Ni-PH2 and Ni-PH4 species.

These three then, as evidenced by the constant relative concentration

ratios, maintained a dynamic equilibrium indicative of relatively rapid

hydrogenation/dehydrogenation reactions. The lack of Ni-X present with - 199 -

Cs doped 70 1

A TOTAL Ni a o Ni-P 60 -o Ni -PH 2 G Ni-PH4

0 0 -LJ 300

20-

0.0 0.1 0.2 0.3 0.4 0.5 0.6 IAQ) f G CAT. HR YV/'~) ML OIL

Figure V-6. Concentration versus contact time results for Ni-T3MPP at 63 ppm Ni feed, 345 *C, and 6.99 MPa H 2 (1000 psig) on the cesium doped catalyst. Solid lines represent model calculations using parameter values in Table V-3. - 200 -

Na doped 70 - --

a TOTAL Ni

0Ni-P 60 0 o Ni-PH 2 e Ni-PH4 50 - Ni-X

-J LAL LLJ

- 0030- 20

20 - 0 0

0.0 0.1 0.2 0.3 OA 0.5 0.6

GCAT. HR W/Q P ML OIL Figure V-7. Concentration versus contact time results for Ni-T3XPP at 56 ppm Ni feed, 345 *C, and 6.99 MPa H 2 (1000 psig) on the sodium doped catalyst. Solid lines represent model calculations using parameter values in Table V-3. - 201 -

the cesium doped catalyst and the small concentration seen with the sodium

sample indicates a decline in the relative rate of Ni-PH reacting to

Ni-X (k5 ) compared to direct metal deposition (k7 ) in the presence of

these additives. The stronger effect seen with cesium suggests the ex-

tent of the influence is dependent on the characteristics of the addi-

tive and not just the absolute amount. Slow metal removal rates were

also observed, being more pronounced on the cesium sample. The resulting

large discrepancy between the initial rate of Ni-P and total metal re-

moval was an indication that metal deposition (hydrogenolysis) was the

rate limiting step in the reaction sequence. On the cesium doped

catalyst the rate of metal removal through Ni-PH was so slow that the

conversion of the feed porphyrin was limited by the dynamic equilibrium

established among the three porphyrinic species. The increase in Ni-P

conversion measured over the entire range of contact times after estab-

lishment of this equilibrium was from 50% to only 60%. As a comparison,

on the oxide catalyst Ni-P conversions in excess of 90% were attained at

similar conditions. It was evident that the addition of the basic com-

ponents Cs and Na resulted in selective inhibition of the bond cleavage

or hydrogenolysis reactions.

The results on the oxide and sulfided catalysts, Figures V-8 and V-9,

have previously been discussed (Section III.A.2 and IV.B.1) and will

only be briefly reiterated. On both catalysts feed porphyrin was rapidly

hydrogenated forming Ni-PH2 and Ni-PH . The concentration of these

species then declined as metal was removed from the oil and as Ni-X was

generated. At the higher contact times essentially all metal in the oil was present as this non-porphyrinic intermediate and its removal determined

the total removal rate. The base case oxide catalyst exhibited a total - 202 -

oxide 70

60 .A TOTAL Ni o Ni-P o Ni-PH 2 50 .e Ni-PH4 0 Ni-X

3 40 R- 0.

Ld U30-

10 -A .

0.0 0.1 0.2 0.3 0.4 0.5 0.6 GCAT. HR W/Q / ML OIL

Figure V-8. Concentration versus contact time results for Ni-T3MPP at 63 ppm Ni feed, 345 *C, and 6.99 MPa H2 (1000 psig) on the oxide catalyst. Solid lines represent model calculations using parameter values in Table V-3. - 203 -

sulfided 70 i I

A TOT/ L Ni 601 o Ni-P o Ni-P e Ni-P 14 50 + Ni-X

40 0.

.-J LUI

0_ 30 2

10 e

0.0 0.05 0.10 0.15 GCAT. HR W/Q, ML OIL

Figure V-9. Concentration versus contact time results for Ni-T3MPP at 60 ppm Ni feed, 345 *C, and 6.99 NPa H 2 (1000 psig) on the sulfided catalyst. Solid lines represent model calculations using parameter values in Table V-3. - 204 -

metal removal rate greater than that obtained on the alkali doped samples but the rapid disappearance of Ni-P compared to total metal indicated that hydrogenolysis was still rate limiting. Sulfiding the catalyst enhanced all reaction steps relative to the oxide catalyst but not to the same extent. The metal deposition reactions were selectively accelerated as demonstrated by the comparatively short contact time scale in Figure V-9 and were only marginally rate limiting.

Treatment of the CoMo/Al203 catalyst with additives iodine and chlorine of relatively higher acid character complemented the results obtained from treating with the alkali metals. The iodized catalyst

(data in Figure V-10) produced lower concentrations of hydrogenated intermediates (Ni-PH2 and Ni-PH ) and a relatively slow rate of feed porphyrin removal compared to the oxide catalyst. There was likewise only a small contribution of Ni-X to the total pool of metal. The approach of the Ni-P disappearance rate to the total metal removal rate suggests that hydrogenation may now be the rate limiting step on this catalyst. A similar but more pronounced effect was seen on the chlorided catalyst in Figure V-lb. Here the feed porphyrin accounted for 80% of the metal in the oil at all times compared to generally less than 20% on the oxide catalyst. The parallel feed porphyrin and total metal removal curves indicated the initial hydrogenation step was rate limiting.

The low concentration of Ni-PH2 and the absence of Ni-PH and Ni-X can be interpreted on the basis that hydrogenated intermediates, once generated, react rapidly in relation to their production rate to deposit metal.

The activity of the alumina support was also examined for comparison.

The reactivity of Ni-T3MPP on this material, shown in Figure V-12, was - 205 -

Iodized 70

A TOTAL Ni 60 o Ni-P o Ni-PH2 9 Ni-PH4 50 - 4 Ni-X

M 40 CL

30- 00

20-

10-

0.0 0.1 0.2 0.3 wAIh G CAT. HR W/Q ) ML OIL

Figure V-10. Concentration versus contact time results for Ni-T3MPP at 63 ppm Ni feed, 345 *C, and 6.99 MPa H 2 (1000 psig) on the iodized catalyst. Solid lines represent model calculations using parameter values in Table V-3. - 206 -

chlorided 7 0

A TOTAL Ni 6 0' o Ni-P o Ni-PH 2

-4

LLJ z_z 3

k.- 2(

0 -

0.0 0.10 0.20 0.30 0.40 W/Q) GCAT. HR ML OIL

Figure V-li. Concentration versus contact time results for Ni-T3MPP at 61 ppm Ni feed, 345 C, and 6.99 MPa H2 (1000 psig) on the chlorided catalyst. Solid lines represent model calculations using parameter values in Table V-3. - 207 -

alumina 70

A TOTAL Ni 60 o Ni-P o Ni-PH 2 9 Ni-PH4 50 $ Ni-X

0-a- -

30 U z 0A

10 -0

0.0 0.1 0.2 0.3 0.4 0.5 0.6 A/Ci G CAT. HR WV/Q ) ML OIL

Figure V-12. Concentration versus contact time results for Ni-T3MPP at 58 ppm Ni feed, 345 *C, and 6.99 MPa H 2 (1000 psig) on the alumina support. Solid lines represent model calculations using parameter values in Table V-3. - 208 -

mechanistically unchanged from the CoMo/Al203 catalyst but all reaction steps were slower. Considering the absence of active metals (Co, Mo) on the surface, the activity of the alumina may be considered high. It is difficult, however, to assess the contribution of just the alumina to the observed activity. As demetallation proceeds, a layer of nickel is deposited generating a surface possibly resembling a supported nickel catalyst. The hydrogenation and hydrogenolysis activity of nickel would

therefore contribute to the observed behavior.

Quantitative interpretation of the kinetic results on all catalysts was obtained both in terms of pseudo first order kinetics for total metal removal and in terms of the sequential reaction network proposed for

Ni-T3MPP in Section III.A.2:

1 3 7 Ni-P W Ni-PH 2 Ni-PH -- deposit 2 4

Ni-X deposit 6

It was also possible to use this network to explain the chlorided

catalyst data even though Ni-PH2 was the only stable intermediate detected.

Based on the reasoning that metal removal from Ni-PH in particular .42

step k , is much faster than the rate of Ni-PH4 production, k 3, the network simplifies to:

1 3'

Ni-P ; Ni-PH - deposit 2 2

Hence any Ni-PH2 reacting to Ni-PH4 results in further rapid reaction

and destruction of the Ni-PH with the appearance of a direct deposition - 209 -

step (k3 ') from Ni-PH 2 . The similarity of this sequence to that used to describe the behavior of nickel-etioporphyrin on oxide CoMo/Al203 is apparent. Subsequent investigation has demonstrated that the Ni-Etio behavior is not a result of chlorine contamination on the catalyst.

The kinetic parameters obtained from least-squares fitting of the experimental data to the solution (discussed in Section III.A.2) of the coupled set of rate equations are listed in Table V-3. These parameters have been used to generate the respective model curves (solid lines) in

Figures V-6 through V-12. The pseudo first order metal removal rate constants, k., in Table V-3 were obtained from the plots of Figure V-13.

Selected rate parameters from the set in Table V-3 are presented in

Figure V-14 to assist in characterizing the catalytic behavior of the doped catalysts. Those chosen include the total metal removal rate

constant, kM, which although being a pseudo kinetic parameter is a measure of the activity of the catalyst, and the porphyrin hydrogenation rate (k1 ) and the hydrogenolysis, or metal deposition rates (k6 + k7 ).

The latter two represent distinctly different functions of the catalyst and enable evaluation of modifications in reaction selectivity provided by the dopants. Trends observed in these rate coefficients could be

correlated with the electron affinity of the added element. This parameter is a measure of the electron attraction power of the atom

(MacKay and MacKay, 1973). The larger (more negative) the electron

affinity, the greater the attraction of an electron to an available

orbital in the atom. Considering the complexity of the system under

study, the apparent success of this correlation may be a fortuitous

coincidence and should be viewed with caution. It is tempting however,

to interpret this trend as possible evidence for a systematic modifica- w we 0

TABLE V-3

Kinetic Rate Parameters for Ni-T3MPP Demetallation on Modified Catalysts

345*C, 1000 psig H2, 63 ppm Ni

Cs-doped Na-doped oxide HDS-16A sulfided iodized chlorided y-Al2 03

(a) 102.0 105.0 183.0 12.0 7.8 29.0 k 1 89.0 94.0 164.0 17.0 7.0 35.0 k 2 126.0 134.0 120.0 250.0 45.0 22.0 k 3 130.0 130.0 150.0 400.0 82.0 33.0 k 4 300.0 300.0 6.4 0 k 0 3.3 49.0 170.0 20.0 0 2.2 3.8 27.0 13.0 k = 114.0 0.21 k 6 2.7 5.0 18.0 120.0 66.0 7.9 k 7 (b) 0.33 0.75 3.4 19.3 3.8 6.4 1.0 km

a) rate coefficients are in units of ml oil/g cat hr b) first order total metal removal - 211 -

1.0

0 Cs doped

0 Na doped 0.5-

oxide

0.1 0.2 0.3 0.4 0.5 1.0 -A

M iodized M

chlorided

s ulf ided

0.0 0.1 0.2 0.3 GCAT. HR W/Q)W/) ML OIL

Figure V-13. First order total metal removal plots for Ni-T3MPP at 345 *C and 6.99 MPa H2 (1000 psig) on the modified catalysts. - 212 -

100

1000 10

0 ki

-- -- I 100 0 N -J

k6 k7 10 0.1 0

Cs Na 0 S I CI I I I I I I It C -20 -40 -60 -80 -1 ELECTRON AFFINITY, kCAL / MOLE

Figure V-14. Ni-T3MPP kinetic rate parameters at 60 ppm Ni feed, 345 *C, and 6.99 MPa H (1000 psig) versus electron affinity of dopant on ihe modified catalysts. - 213 -

tion of the surface chemistry or acidity of the catalyst in the presence of these dopants which influences the observed catalytic behavior. Spec- ulation as to the mechanism of modification by the dopants will be discussed in the next section.

Variations in catalytic activity for total metal removal aslarge as

50-fold were observed on the doped catalysts. When plotted in the fashion of Figure V-14, the overall demetallation rate exhibited a

"volcano" shape curve with maximum metal removal activity occurring on the sulfided catalyst. By resolving this lumped parameter into the individual reaction steps within the network, it was possible to explain why the maximum was observed. The sulfided catalyst exhibited high activity for both hydrogenation (two times greater than oxide catalyst) and metal deposition (seven times greater than oxide catalyst) resulting in the high overall demetallation rate.

On the base case oxide catalyst and alkali-doped catalysts, the hydrogenation activity was similar. The metal deposition rate parameters were substantially lower, hence rate limiting, and became progressively smaller moving from the oxide, to sodium, to cesium doped catalyst. The sharp rise in overall demetallation activity observed in progressing from the cesium-doped through to the sulfided catalyst (increasing electron affinity of additive) was attributable to this sharp increase in the rate limiting step of metal deposition.

A quite different effect was seen when pretreating the catalyst with additives of even higher electron affinity (S, I, Cl) in a reducing environment. The lower overall demetallation activity on the iodized and chlorided catalyst was due to a significant decline in hydrogenation

(k1 ) activity of the catalyst which became rate limiting when progressing - 214 -

from sulfur to chlorine on the surface. As shown in Figure V-14, the metal deposition activity (k6 + k7 , k3 ' for chlorided catalyst) displays only a marginal drop with increasing electron affinity of dopant whereas the hydrogenation activity is twenty times lower on the chlorided catalyst compared to the sulfided catalyst. This sharp decline in hydrogenation rate results in a shift in the rate limiting step from hydrogenolysis on the sulfided catalyst (also on the oxide and alkali- doped catalysts) to hydrogenation on the halogen treated catalysts.

A measure of the consistency of the kinetic parameters can be ob-

tained by examining the relative k /k2 and k 3/k ratios for all catalysts.

These ratios reflect the thermodynamic equilibrium between Ni-PH 2/Ni-P, and Ni-PH 4 /Ni-PH2 and should therefore be invariant from one catalyst to

the next. The k 1/k2 ratio was equal to 0.91 0.20 whereas k 3/k was

equal to 0.58 0.19. Variations in these ratios were on the order of

20 to 30%. These ratios were sensitive to the concentration of Ni-P,

Ni-PH 2, and Ni-PH4 at high Ni-P conversion (>50%) where accurate measurement by UV-visible spectroscopy of the composite oil was most difficult.

These variations did not seem excessive in view of this limitation.

V.D. Discussion of Active Sites on Modified Catalysts

More significant than the wide variation in total demetallation

activity for Ni-T3NPP observed by modifying the CoMo/Al203 catalyst is

the selective alteration of catalytic sites with different functionality

by these dopants. It has been demonstrated that the changes in global

activity are a result of selective enhancement or inhibition of hydro-

genation or hydrogenolysis steps within the demetallation network. - 215 -

Changes as dramatic as a shift in the nature of the rate limiting step from a hydrogenolysis reaction to a hydrogenation reaction are possible by suitable modification of the active catalytic sites.

The reactivity patterns observed for Ni-T3NPP on the doped catalysts result from the complex interactions of the support, the active metals, and the dopant. Yet by assimilating and carefully interpreting data representative of a wide range of catalyst behavior, it is possible to speculate on the nature of the active sites in the demetallation network and to assess the impact of dopants on the reactivity of these sites.

Consistent with the demonstrated correlation between acidity and activity for hydrogenolysis in HDS reactions (Laine et al., 1980,

Maternova, 1983), the Cs and Na-doped catalysts with the lowest acid site density (Table V-2) exhibited the lowest hydrogenolysis (metal deposition) activity in the HDM reaction. The base y-Al2 0 with an acid site density greater than both alkali-doped samples, exhibited higher hydrogenolysis activity. Similarly, the oxide CoMo/Al203 catalyst with yet higher acidity had the greatest hydrogenolysis activity among these. The limited data available on these catalysts suggests that the acid site density and hydrogenolysis (k6 + k7 ) activity in HDM are related.

The superior metal depositing activity of the oxide catalyst compared to the support indicates all hydrogenolysis activity is not associated with the alumina. Likewise, the contribution of the Co and Mo species to the total acidity of the catalyst is apparent by the higher acid density on the fresh catalyst compared to the alumina support.

This enhanced surface acidity has previously been reported (Ratnasamy et al., 1974). - 216 -

The high molar loadings of Cs, 2.20 mmole/g oxide cat. (2.68 mmole/ g Al 20 3) and Na, 1.95 mmole/g oxide cat. (2.37 mmole/g A1 2 03 ) are far in excess of the amount of alkali required to neutralize the alumina

(1.1 mmole/g Al2 03 , Lycourghiotis, 1980a) if no transition metals were present. It is hard to imagine, however, that a surface so highly covered with Mo (b50% coverage for HDS-16A) and Cs or Na does not result in interaction of the two. The alkali, present as Na 20 and Cs20' may also exist on Mo vacancies on the fresh catalyst and donate electrons to neutralize electron acceptor sites on these metals. This interpretation is verified by the substantial drop in surface acidity of the CoMo catalyst in the presence of Na or Cs which indicates interaction of these dopants with both the support and the transition metals. The larger effect seen with Cs is consistent with its greater electron donating ability (low electron affinity).

Evidence for the existence of hydrogenation sites with acidic properties on the oxide CoMo/Al203 catalyst was presented in Section IV.A with the discussion of the pyridine poisoning results. These sites were speculated to be Mo anion vacancies on the surface. The presence of strong hydrogenation activity on the catalysts with little or no surface acidity after sodium or cesium doping suggests one of two possibilities.

The dopants once deposited may not interfere with the in-situ generation of additional Mo vacancies. Hence, even though these acidic sites are poisoned on the fresh oxide catalyst, under the reducing reaction conditions new sites may emerge. Alternatively, a second type of hydrogenation site may exist. This non-acidic hydrogenation site may be associated with metallic species such as cobalt which has been detected on reduced

CoMo/ Al203 catalysts (Chin and Hercules, 1982) not exposed to air prior - 217 -

to analysis. Part of the promoting role of cobalt has been attributed to increased hydrogenation activity (Massoth, 1978). These sites maintain their integrity and are not poisoned or obstructed despite the high surface coverage of Cs and Na possible if it were to exist as a monolayer.

It is conceivable that some of this hydrogenation activity is associated with nickel from the reacting porphyrin which deposits over the partially deactivated sites. However, the activity of the base alumina which also developed a surface covered with nickel was only one third as active (compare k values) as the alkali-doped catalysts. Thus the majority of the hydrogenation activity on these alkali-doped catalysts must be due to Co and Mo species on the surface not deactivated by the dopant.

Quantitative comparison of the surface acidities of the S, I, and

Cl treated catalysts was difficult due to the trouble encountered with neutralization of ammonia by the desorption of acidic species. It is apparent, however, that the total number of sites on these catalysts is greater than on the oxide catalyst as the measured acid site densities even with this problem were comparable to the oxide catalyst. Qualita- tively this is consistent with the higher hydrogenolysis activity observed on these treated catalysts.

The enhanced metal deposition activity of the sulfided catalyst discussed in Section IV.B.1 was attributed to the presence of a high concentration of Bronsted acidity associated with hydrogenolysis activity.

The generation of these sites was proposed by Yang and Satterfield (1983) to be due to H 2S'adsorbing and dissociating on Mo vacancies. A similar phenomena may be occurring on the halogenated catalysts with the HI and

HU1 generated during pre-treatment also dissociating on No. This is - 218 -

especially likely when CHCl3 is in the feed as a constant supply of HC is then present to maintain an adsorption equilibrium with the catalyst.

Considering this adsorption and dissociation process to be responsible for the enhanced surface acidity of these catalysts, it is doubtful that a true measure of the catalyst acidity at reaction conditions can be obtained from the TPAD as performed. The acidity would obviously depend on the partial pressure of the dissociating species (H2 S, HI, HCl) and no such components are present in the gas phase during these measurements. Thus determining the absolute magnitude of the surface acidity increase on these treated catalysts over the oxide catalyst may be a difficult task.

The presence of chlorided alumina on the surface of the pre-chlorided catalyst suggested by the XPS results may also be contributing to the metal deposition activity observed on this catalyst. The high hydrocracking activity of chlorided alumina has been reported in asphaltene and model coal liquid degradation studies (Ignasiak et al., 1981, Salim and Bell, 1982). This additional bond cleavage activity is consistent with the high rate of metal deposition which results in Ni-PH2 as the only stable intermediate.

As demonstrated by the kinetic results in the previous section, the most significant impact due to the presence of iodine and chlorine on the catalyst was a dramatic reduction in hydrogenation activity compared to the sulfided catalyst. By treating the catalyst with these additives in a reducing environment, I and Cl interact with vacancies on molybdenum, normally associated with hydrogenation activity, much like the Mo-SH groups that form in the presence of H 2S (Yang and Satterfield, 1983).

The XPS results were consistent with the presence of molybdenum-halide - 219 -

species. For the in-situ chlorided catalyst, most representative of an equilibrium halogen loading, the Cl/Mo atomic ratio was 2.2 indicating there was, on the average, sufficient halogen present to interact with a vacancy site on every molybdenum.

Results reported by Kiskinova and Goodman (1981) on the chemi- sorptive and reactive properties of nickel surfaces modified with electro- negative adatoms (Cl,S,P) are qualitatively similar to the trend seen with the S,I,Cl-doped CoMo/Al203 catalysts. These investigators observed a reduction in the adsorption strength and capacity (number of sites) for hydrogen and CO on doped Ni. Catalytic activity for methanation decreased on the modified surfaces. The extent of poisoning increased with increasing electron affinity of the adsorbed atom (largest effect with

Cl) and was interpreted in terms of both an occupation or obstruction of the active site and an alteration in the surface electron density on the metal in the presence of these atoms.

Similar reasoning is useful in interpreting the decline in hydrogenation activity observed in proceeding from the S, to I, to Cl treated catalyst. The presence of I and Cl can be viewed as occupying active sites on Mo (generated by reduction) eliminating their availability for hydrogen adsorption and for coordination with double bonds. Sulfur

(present as SH after H2 S dissociation) also occupies these vacancies but is readily desorbed as evidenced by the hydrogenation activity on this catalyst.

The stronger inhibition seen with Cl than I is consistent with its greater electron affinity enabling it to form a more stable, more irreversibly bound molybdenum complex. Binding energies for desorption of halogens from molybdenum surfaces measured by Bolbach and Blais (1983) - 220 -

were progressively smaller moving down the Periodic Column (F>Cl>Br).

Iodine would be expected to follow this trend and exhibit an even lower binding energy. Based on these binding energy strengths, less of an inhibiting effect would be expected with I than Cl which is consistent with the observed results. The strength of the Mo-Cl interaction was experimentally demonstrated by the inability to restore the hydrogenation activity of a pre-chlorided catalyst back to the activity of the oxide catalyst after exposure to flowing hydrogen at 450*C for 4 hours.

The results of this doping study have shown how selective alteration of the relative hydrogenation/hydrogenolysis rates in the HDM network of Ni-T3MPP are possible enabling changes as dramatic as a shift in the overall rate limiting step to occur. The relationship between hydrogenolysis (metal deposition) activity and surface acidity has been demonstrated. Addition of Cs and Na to the CoMo/Al203 catalyst lowered the acidity and selectively reduced the metal deposition steps (k6 + k ) without inhibiting hydrogenation (k ). The high hydrogenation activity on these catalysts suggests the presence of non-acidic hydrogenation sites in addition to the hydrogenation activity associated with the Mo vacancies.

The overall impact of I or Cl treatment on the activity of the CoMo catalyst can be interpreted in terms of two effects resulting from the interaction of HI or HCl with vacancies on the molybdenum. The dissociation of these species on Mo sites generated by reduction, produces

Bronsted acidity which maintains a high level of hydrogenolysis activity

(k6 + k7 ) promoting metal deposition. The simultaneously generated molybdenum-halide bond eliminates a site otherwise available for hydrogenation lowering the total number present and the observed activity (k1 ). - 221 -

The strength of the Mo-X interaction contributes to the degree of inhibition.

Sulfiding had the unique effect of producing an enhancement in both the hydrogenation and hydrogenolysis activities over the oxide catalyst. The increase in hydrogenolysis (k6 + k7 ) activity was selectively larger and was attributed to the presence of an increase in

Bronsted acidity arising from H 2S dissociation on Mo vacancy sites.

Catalyst selectivity for Ni-X formation (k5 /k7 ) was diminished in the presence of all dopants. Only on the sulfided catalyst did the dominance of this species in the oil at high feed conversions approach that observed with the oxide catalyst. The slow demetallation rate (k6 of this non-porphyrinic compound suggests high selectivity for Ni-X is desirable from a standpoint of obtaining uniform metal deposits in catalyst pellets.

V.E. Reaction Engineering Implications

The implications of this ability to selectively alter one functionality of the catalyst and hence selectively change the relative reaction rates within the Ni-T3MPP demetallation network are revealed in the metal deposition profiles within catalyst pellets, obtained under dif-

fusion limited conditions. Presented in Figure V-15 are metal deposition

profiles in 1/16 inch pellets calculated with the model developed in

Section III.C.2 using the kinetic parameters presented in this chapter

for the different catalysts. The profile for the chlorided catalyst was

calculated using the model for Ni-Etio developed by Agrawal (1980) as

the kinetic reaction schemes were analogous. These profiles are calculated - 222 -

1 1.0 Cs doped I Na doped oxide

I-

0 ' 0 I 0 0 00 1 0 r/R r/R r/ R

3 3 3 sulfided iodized chlorided

2 2 2

I 1

0 0 0 1 0 r/R 0 r/R 1T/R

Figure V-15. Calculated nickel deposition profiles in 1/16" diameter catalyst pellets at the reactor entrance for Ni-T3MPP demetallation at 345 *C and 6.99 MPa H2 (1000 psig) on the modified catalysts. - 223 -

for the entrance of the bed (z=0). The oxide and sulfided catalyst profiles have been experimentally verified and discussed elsewhere (see

Section III.C.2 and Section IV.B.2). All non-kinetic parameters used in calculating the other plots (Cs, Na, I, and Cl-doped) were identical to those used for the oxide catalyst listed in Table 111-4.

The Thiele moduli, # , for the uncoupled system of pseudo components reacting by first order irreversible kinetics in the solution technique given by Wei (1962) characterize the extent to which diffusional limitations are important in the metal profiles. Listed in Table V-4 are the

Thiele moduli for each set of kinetic data calculated from the eigenvalues of the diffusion coefficient and rate coefficient matrices as discussed in Section III.C.2. These parameters are the characteristic decay rates of concentration versus radial distance in the terms summed to obtain the metal profiles:

I (4$ r/R) M(r,z) = { b.(z) 0 i i 1 4$.)

where M(r,z) is the concentration of deposited metal at a given radial, r, and axial, z, position. The relative magnitude of the coefficients b. also listed in Table V-4 indicates all terms contribute to the shape of the profiles. The large $. terms decay rapidly from maxima at r/R =

1.0 leaving the slower decaying components to dominate the region of small r/R.

Consistent with the slower metal deposition steps (k6 + k ) on the

Cs and Na-doped catalysts, the smallest Thiele moduli have values near one or lower, and, correspondingly, uniform metal profiles are seen on these catalysts. The continuous increase in the dominant 4 on - 224 -

TABLE V-4

Characteristic Thiele Moduli and Metal Distribution Parameters

for Modified Catalysts

0 Ni-T3MPP, 345 C 6.99 MPa H 2 (1000 psig)

1 (.r/R) M = lb.(z) 0

b 1 (0) pi b2 (0) $2 b3 (0) $ 3 b4 (0) p em

Cs 1.00 0.805 -1.48 16.41 -0.93 24.71 0 0 0.93

Na 1.00 1.42 -0.74 1.62 -0.49 17.2 0.23 25.10 0.92 oxide 1.00 2.13 0.56 4.81 -3.23 15.98 1.65 22.08 0.66

sulfide 1.00 4.29 1.91 8.43 -4.26 23.90 2.10 33.38 0.36

iodize 1.00 3.43 -0.62 3.94 -0.50 7.58 0.11 15.31 0.75

chloride 1.00 4.17 -1.00 12.07 0.50

* all b are normalized to b - 225 -

progressing from the Cs doped to the sulfided catalyst reflects an increasingly stronger diffusion limitation in the deposition reactions as is revealed in the metal profiles.

It is possible to compare the deposition patterns of metal in the catalyst extrudates quantitatively using the metal distribution parameter defined by Tamm et al. (1981):

fM(r) r dr

S= 0 m MmaxJr dr 0 where M(r) is the local wt% metal deposit, Mmax the maximum concentration, and r the fractional radius. This parameter is essentially the ratio of

the average metal concentration to the concentration at the maximum and

characterizes the effective utilization of the catalyst for metals deposits. A distribution parameter of unity corresponds to uniform metal profiles whereas a value approaching zero would be characteristic of a sharp spike at the pellet's edge. Values calculated at the entrance of

the bed for each catalyst are listed in Table V-4.

Catalysts (Cs, Na-doped) with small 4'. (slow deposition rates) were

calculated to have high distribution parameters. As metal deposition

activity increased (large '.), 0M decreased and less effective use was made of the metal loading capacity of the catalyst. Catalysts with

small 0m reach the plugging limit at Mmmax more quickly and with less total metal in the catalyst (i.e., e x M ) than is typical for more in max uniform distributions. The highly desirable uniform profiles are ob-

tained at the expense of somewhat lower total metal conversion at - 226 -

equivalent contact times. Hence the optimum conditions for metal deposition would be a balance between high conversion and most complete utilization of the catalyst.

An alternative interpretation of the data in Figure V-14 is to examine the relative hydrogenation/hydrogenolysis activities in the

Ni-T3MPP reaction network on the doped catalysts. This selectivity, S, defined as k /(k + k ) is plotted in Figure V-16 as a function of the 1 6 7 ) electron affinity of the dopant. The ability to correlate the observed variation in S over three orders of magnitude with a chemical property of the isolated dopant element is quite remarkable. Considering the diverse influences imparted by the Cs and Na dopants compared to the S,1, and

Cl additives on the catalytic sites, it would seem mechanistically more appropriate to quantify the selectivity in terms of a surface property of the catalyst. Possibly some catalytically sensitive surface property such as adsorption strength, charge density, or surface acidity does correlate with dopant electron affinity.

The correlation, while empirical and of unproven predictive use- fulness, does clearly reveal the transition in rate limiting step in the overall network. A selectivity greater than one corresponds to hydrogenolysis as the rate limiting step and conversely, a selectivity less than one indicates hydrogenation is rate limiting.

The demonstrated ability to vary to such an extent the relative reactivities of two different catalytic functions suggests control of the metal deposition location is possible. A more attractive demetallation processing scheme may be one where the two types of reactions are separated on two different catalysts, possibly in different reactors. On the first catalyst the metal bearing molecules would be activated for - 227 -

100[

(I)

0.1

Cs Na 0 S I CI 111 1 i1i 0.01 1 0 -20 -40 -60 -80 -100 ELECTRON AFFINITY, kCAL/MOLE

Figure V-16. Reaction selectivity in Ni-T3MPP demetallation at 345 *C and 6.99 MPa H2 (1000 psig) versus electron affinity of dopant on modified catalysts. - 228 -

deposition by hydrogenation to generate the intermediate species. A catalyst with low metal deposition activity, S >> 1 is desired. These products would then be passed on to a second catalyst where metal deposition would occur. In this configuration poisoning of the precious hydrogenating catalyst by metal deposition would be avoided. Excessive hydrogenation on the second catalyst is unwanted (S << 1). The observed correlation between high metal deposition activity and surface acid density suggests some acid character to this catalyst is required.

The significance of these doping results are in the realization that it is possible to alter the reactivity of the metallo-porphyrin in a hydrodemetallation environment. By modifying the chemical nature of the individual catalytic sites on CoMo/Al203 through judicious choice of dopant or possibly by designing new catalysts, a continuous variation in the degree of hydrogenation activity compared to hydrogenolysis activity is possible. The dopants chosen in this study may not represent the most convenient method of modification (corrosive nature of HI and

HC1, for example) but they do demonstrate the kinetic feasibility of selectively controlling the reactivity of metal compounds characteristic of those in residuum oil. - 229 -

CHAPTER VI

CONCLUSIONS

1) The reaction of nickel-porphyrins on CoMo/Al203 catalysts during

hydrodemetallation (HDM) proceeds via sequential pathways. The first

step involves hydrogenation of peripheral double bonds to "activate"

the porphyrin for the final hydrogenolysis step which fragments the

ring and deposits the metal on the catalyst surface. A dual function

catalyst with both hydrogenation activity associated with Co and Mo

sites and hydrogenolysis (metal deposition) activity associated with

acid sites on the support and the Mo is effective for HDM.

2) Structural differences on the periphery of the metallo-porphyrin

molecules are shown to influence the complexity of the reaction path-

ways and the relative rates of hydrogenation and hydrogenolysis. The

substituent groups at the beta-pyrrolic and methine bridge carbon

atoms in the porphyrin appear to exert a dominant role, both steric

and chemical, in determining these reactivity differences. The sub-

stituent groups at the beta-pyrrolic positions influence the initial

hydrogenation whereas the groups at the methine bridge positions

influence the final metal deposition step.

Ni-etioporphyrin is slow to hydrogenate due to the steric con-

straint and the higher reduction potential of the peripheral double

bonds attributable to the presence of methyl and ethyl groups at the - 230 -

beta-pyrrolic positions. Once the porphyrin is hydrogenated at one pyrrole position, the resulting enhancement in reactivity of adjacent methine bridges coupled with the accessibility of these bonds renders these positions highly susceptible for reaction. This leads to ring hydrogenolysis and metal deposition. Only one stable nickel intermediate, the hydrogenated porphyrin (Ni-EPH2 ), is present in the oil with the feed material.

Ni-tetra(3-methylphenyl)porphyrin (and Ni-TPP) rapidly hydro- genates due to the accessibility and low reduction potential of the beta-pyrrolic positions forming not one but two hydrogenated intermediates, Ni-PH2 and Ni-PB . The second hydrogenation occurs selectively at a pyrrole adjacent to the first pyrroline ring. The subsequent reaction of Ni-PH at the activated methine bridges is influenced in part by the presence of the tolyl (or phenyl for Ni-TPP) substituents at these positions. One route involves cleavage of the ring leading to destruction of the macrocycle and deposition of the metal. This behavior is similar to the reaction of the hydrogenated Ni-etioporphyrin although proceeds more slowly. A second pathway involves generation of a relatively stable non-porphyrinic nickel intermediate, Ni-X, which comprises the bulk of the metal in the oil at high feed conversions. This intermediate has been isolated and identified as a contracted ring structure (i.e., a nickel-corrin) formed by elimination of a methine bridge carbon as a xylyl group (tolyl for Ni-TPP) and subsequent reclosure of the ring. Both a hydrogenation function and an acid function appear necessary for Ni-X formation.

Variations in the overall rate limiting step for metal removal were observed for the different porphyrins on the oxide catalyst. - 231 -

Hydrogenation was rate limiting with Ni-etioporphyrin whereas metal

deposition (hydrogenolysis) was the slow step with Ni-T3MPP (and

Ni-TPP).

3) The reaction network for demetallation of Ni-etioporphyrin was

previously reported to consist of three major kinetic steps; hydro-

genation (k ), dehydrogenation (k2), and demetallation or hydro-

genolysis (k3).

The seven kinetic steps in the network for Ni-T3MPP can likewise

be grouped into these three types of reactions. The hydrogenation

steps (k1 , k3 ) produced intermediates which could then dehydrogenate

(k2, k ) or react via hydrogenolysis steps to produce Ni-X (k5 ) or

deposit metal (k6 , k7 ). All steps were first order with respect to

the nickel species reacting. Langmuir-Hinshelwood kinetics were not

required to interpret the kinetic behavior over the concentration

range examined. Hydrogenation steps (ki, k 3) had activation energies

of 16 to 18 kcal/mole and were first order in hydrogen pressure.

Dehydrogenation steps (k3 , k4 ) were characterized by activation ener-

gies of 20 to 23 kcal/mole and were zero order in hydrogen. Activation

energies for the hydrogenolysis steps (k5 , k 6, k7 ) ranged from 19 to

24 kcal/mole. These steps exhibited a second order hydrogen pressure

dependence.

The temperature and pressure dependence of the three kinetic

steps for Ni-etioporphyrin demetallation determined by Agrawal (1980)

were: hydrogenation (k1 ), activation energy of 16 to 18 kcal/mole,

first order in hydrogen; dehydrogenation (k2), activation energy of

22 to 24 kcal/mole, zero order in hydrogen; metal deposition (k3), - 232 -

activation energy of 28 to 30 kcal/mole, second order hydrogen pressure

dependence.

4) There was no change in the reaction pathway of Ni-T3MPP during

HDM in the presence of pyridine or on the sulfided catalyst compared

to that observed on the clean, oxide catalyst. The competitive ad-

sorption of pyridine did however inhibit all reaction rates.

Sulfiding produced an enhancement in all reaction rates compared

to the oxide catalyst. The metal deposition steps were selectively

enhanced (seven-fold increase) compared to the hydrogenation/dehydro-

genation steps (two-fold increase). This selectivity change was

reflected in a higher ratio of nickel-porphyrin (Ni-P) concentration

to total metal concentration in the oil. In terms of the ratio of the

hydrogenation rate coefficient (k ) / metal deposition rate coefficients

(k6 + k7 ), this ratio was 4.8 on the oxide catalyst and 1.2 on the

0 sulfided catalyst at 345 C and 6.99 MPa H2 (1000 psig).

The increase in hydrogenation activity was attributed to an in-

crease in Mo vacancies (hydrogenation sites) from the simultaneous

reduction of the catalyst. The high hydrogenolysis activity was con-

sistent with an increase in Bronsted acidity arising from the dis-

sociation of H 2S on Mo vacancies.

The temperature and pressure dependencies of the seven kinetic

steps for Ni-T3MPP on the sulfided catalyst were similar to those

determined on the oxide form.

5) Metal deposition profiles measured under diffusion limited condi-

tions in 1/16 inch catalyst extrudates reflect the intrinsic reactivity - 233 -

of the nickel-porphyrins. Profiles obtained at the entrance of the reactor were characterized by an internal maximum in the metal deposition pattern. These were generated as a result of the sequential reaction pathway of the porphyrins as metal deposition occurred from a species not originally in the oil. This maxima was observed to shift to the edge of the pellets by the end of the bed as the concentration of metal depositing species in the bulk increased. The position of the internal maxima was deeper at the reactor entrance with Ni-T3MPP (r/R = 0.75) compared to Ni-etioporphyrin (r/R = 0.90) on the oxide catalyst (345*C, 6.99 MPa H2) due to differences in their intrinsic reactivity.

The enhanced metal deposition activity of the sulfided catalyst compared to the oxide catalyst was apparent in the pattern of metal deposition. The profiles were more U-shape (larger 4.) with less deposit in the pellets' center.

Theoretical calculations based on a steady state, coupled diffusion and reaction model using rate parameters obtained in the intrinsic studies were used to interpret the metal deposition profiles.

Quantitative agreement between the theoretical and experimental profiles was obtained using diffusion coefficients for the nickel species -6 2 on the order of 10 cm /sec. Diffusivities for the hydrogenated porphyrinic intermediates were two times larger than the feed porphyrin.

These values were unchanged for both the oxide and sulfided systems.

The diffusivity of the non-porphyrinic intermediate, Ni-X, was 2 to

3.5 times larger than the feed porphyrin, Ni-P, depending on the system. - 234 -

6) The feasibility of varying the HDM reaction selectivity, S, of

Ni-T3MPP defined as the porphyrin hydrogenation rate constant / the

metal deposition rate constants k1I/(k6 + k7 )] was demonstrated. This

was accomplished by doping the CoMo/Al203 catalyst with additives

which were specific in altering one functionality of the catalyst.

Changes as dramatic as a shift in the overall rate limiting step were

possible.

The relationship between hydrogenolysis (metal deposition) activity

and surface acidity was demonstrated. Neutralization of catalyst

acidity by addition of basic dopants (Cs, Na) significantly reduced the

metal deposition activity with only a marginal reduction in hydrogena-

tion activity. The high hydrogenation activity on these catalysts

suggests the presence of non-acidic hydrogenation sites (possible Co)

in addition to the hydrogenation activity associated with the Mo

vacancies.

Treatment of the CoMo/Al203 catalyst with S, I, Cl in a reducing

environment resulted in major changes in the reactivity of Ni-T3MPP

compared to the oxide catalyst. All of the treated catalysts had high

surface acidities and high metal deposition activity. Sulfiding had

the unique effect of also enhancing the hydrogenation activity which

was attributable to the simultaneous reduction of the catalyst.

Iodine and chlorine treatment produced successively lower hydrogena-

tion activity compared to the sulfided catalyst (S>I>Cl) despite the

reduction of Mo which occurred. Hydrogenation became rate limiting

in the overall sequence on the I and Cl treated catalysts. This be-

havior was interpreted as arising from the dissociation of HI and

HCl on Mo vacancies which generated a high level of Bronsted acidity - 235 -

(desirable for metal deposition) but also resulted in strong, irreversible-like, interaction between the Mo and the halide which occupied a hydrogenation site.

The reaction on catalysts with large S was characterized by a low value for the ratio of the concentration of nickel-porphyrin to total metal in the effluent oil. This composition ratio was progressively larger on the sulfided and halided catalysts and approached unity on the chlorided catalyst.

Calculated metal deposition profiles in 1/16 inch catalyst pellets displayed significantly different behavior reflecting the selectivity variations in the Ni-T3NPP reaction network. Reaction in the presence of basic additives (Cs, Na) was characterized by small c1. and uniform profiles. On the other hand, the acidic components (S, I, Cl) added on the catalyst resulted in large $ and more sharp, U-shape profiles indicative of strongly diffusion limited reactions. The optimum condition is a balance between uniform metal deposition (long catalyst life) and high metals removal activity. - 236 -

NOMENCLATURE

b. ith coefficient of characteristic terms in metal profile summation

C initial nickel concentration, ppm Ni 0

Cl, concentration of Ni-P, Ni-PH2 , Ni-PH , and Ni-X, C2 ' 3 C4 ' ppm Ni

C concentration vector

D effective diffus vity of Ni-P, Ni-PH2, Ni-PH , 1 , D 2 , D , D 3 4 and Ni-X, cm /sec

D diffusion coefficient matrix of eq. (1)

E activation energy, kcal/mole

modified Bessel function of first kind of 0 zero order

modified Bessel function of first kind of first order k kinetic rate coefficient, ml oil/g cat. hr kM pseudo first order metal removal rate coefficient, ml oil/g cat. hr

K adsorption coefficient in Langmuir-Hinshelwood expression

rate coefficient matrix of eq. (1) and B.5 K diffusion disguised rate coefficient matrix of eq. (9)

L length of catalyst pellet, cm

M deposited metal, weight percent

Ni-EP Ni-etioporphyrin

Ni-EPH Ni-etiochlorin 2 Ni-P Ni-tetra(3-methylphenyl)porphyrin

Ni-PH Ni-tetra(3-methylphenyl)chlorin 2 - 237 -

Ni-PH 4 Ni-tetra(3-methylphenyl)isobacteriochlorin

Ni-X non-porphyrinic contracted ring Ni species

Q oil flow rate, ml/hr

r radial position in catalyst, cm

R catalyst pellet radius, cm

t time on stream, hrs

W weight of catalyst, g

X, - eigenvector matrix and inverse matrix of K

X+ +-l eigenvector matrix and inverse matrix of K xx= = - Y, Y eigenvector matrix and inverse matrix of p cD K

z reactor axial position, cm

th .=-J= 1th eigenvalue of p cD K 1c + th =+ I t eigenvalue of K .th 1 eigenvalue of K a.1

p oil density, g/ml

PC catalyst density, g/ml ith Thiele modulus = 1T R

1M I (i diagonal matrix of eq. (5) I4) e metal distribution parameter m

subscript s catalyst pellet surface, r = R - 238 -

BIBLIOGRAPHY

Agrawal, R., "Kinetics and Diffusion in Hydrodemetallation of Nickel and Vanadium Porphyrins," Sc.D. Thesis, M.I.T., (August, 1980).

Alder, R.W., Baker, R., and Brown, J.M., "Mechanism in Organic Chemistry," Chap. 2, Wiley, N.Y. (1971).

Alturki, Y.A., Eglinton, G., and Pillinger, C.T., Adv. Org. Geochem., 5, 135 (1971).

Andreu, P., Galiasso, R., Morales, A., Abrams, 0., Katan, L., Pantoja, L., Sojo, P., and Vergara, E., in "Studies in Surface Science and Catalysis," Vol. 7, Part B, T. Seiyama and K. Tanabe, eds., Elsevier, (1981), p. 877.

Angst, C., "New Hexahydroporphyrin Ligand Systems," Ph.D. Thesis, ETH- Zurich (1981), Diss. ETH No. 6783.

Angst, C., Kratky, C., and Eschenmoser, A., Angew. Chem. Int. Ed. Engl., 20, 263 (1981).

Baker, E.W., and Palmer, S.E., in "The Porphyrins," D. Dolphin, ed., Vol. I, Chap. 11, Academic, N.Y. (1978).

Baker, E.W., and Smith, G.D., ACS Prepr., Div. Petrol. Chem. 19 (4), 744 (1974).

Barwise, A.J.G., and Whitehead, E.V., ACS Prepr., Div. Petrol. Chem., 25 (2), 268 (1980).

Bearden, R., and Aldridge, C.L., Energy Progress, 1, 44 (1981).

Benesi, H.A., and Winquist, B.H.C., Adv. Catal., 27, 97 (1978).

Beuther, H., and Schmid, B.K., Proc. Sixth World Petrol. Congress, Section III, Paper 20 (1963).

Beuther, H., Larson, 0.A., and Perrotta, A.J., "Catalyst Deactivation: Studies in Surface Science and Catalysis," Vol. 6, B. Delmon and G.F. Froment, eds., Elsevier (1980), p. 271.

Bhinde, M.V., "Quinoline Hydrodenitrogenation Kinetics and Reaction Inhibition," Ph.D. Thesis, University of Delaware, (June, 1979).

Bolach, G., and Blais, J.C., Surf. Sci., 126, 405 (1983).

Boorman, P.M., Kriz, J.F., Brown, J.R., and Ternan, M., Proceedings Climax Fourth International Conference on the Chemistry and Uses of Molybdenum, H.F. Barry and P.C.H. Mitchell, eds., Climax Molybdenum Co., Ann Arbor (1982), p. 192. - 239 -

Brinen, J.S., and Armstrong, W.D., J. Catal., 54, 57 (1978).

Broderick, D.H., "High-Pressure Reaction Chemistry and Kinetic Studies of Hydrodesulfurization of Dibenzothiophene Catalyzed by Sulfided Coo-Moo 3 /y-Al 2 03 ," Ph.D. Thesis, University of Delaware, (May, 1980).

Buchler, J.W., in "Porphyrins and Metalloporphyrins," K.M. Smith, ed., Chap. 5, Elsevier (1975).

Buchler, J.W., in "The Porphyrins," D. Dolphin, ed., Vol. I, Chap. 10, Academic, N.Y. (1978).

Bunger, J.W., and Li, N.C., eds., "Chemistry of Asphaltenes," Adv. Chem. Series, No. 195, American Chemical Society, Washington, D.C. (1981).

Burger, H., in "Porphyrins and Metalloporphyrins," K.M. Smith ed., Chap. 11, Elsevier (1975).

Cecil, R.R., Mayer, F.Z., and Cart, E.N., Paper presented at AIChE Meeting, Los Angeles, 1968. Referenced in Gates et al. (1979).

Chin, R.L., and Hercules, D.M., J. Phys. Chem., 86, 3079 (1982).

Cocchetto, J.F., "Reaction Network and Kinetics of Vapor-phase Catalytic Hydrodenitrogenation of Quinoline," Ph.D. Thesis, M.I.T., (February, 1980).

Cukor, P.M., and Prausnitz, J.M., J. Phys. Chem., 76, 598 (1972).

Daly, F.P., J. Catal., 51, 221 (1978).

deBeer, V.H.J., and Bevelander, C., van Sint Fiet, T.H.M., Werter, P.G.A.J., and Amberg, C.H., J. Catal., 43, 68 (1976).

de Bruijn, A., Proceed. Sixth Int. Cong. on Catalysis, 2, 951 (1976).

DeVries, A.B., and Jellinek, F., Rev. Chim. Mineral. 11, 624 (1974).

Dolphin, D., ed., "The Porphyrins," Academic, N.Y. (1978).

Doraiswamy, L.K., and Tajbl, D.G., Cat. Rev. Sci. Eng., 10, 177 (1974).

Dorough, G.D., and Huennekens, F.M., J. Am. Chem. Soc., 74, 3974 (1952).

Dorough, G.B., Miller, J.R., and Huennekens, F.M., J. Am. Chem. Soc., 73, 4315 (1951).

Ekstrom, A., Loeh, H., and Dale, L., ACS Prepr., Div. Fuel Chem., 28 (3), 166 (1983).

Filby, R.H., in "The Role of Trace Metals in Petroleum," T.F. Yen, ed., Chap. 2, Ann Arbor Science Publishers, Inc., Michigan (.1975). - 240 -

Fitz, C.W., Jr., and Rase, H.F., Ind. Eng. Chem. Prod. Res. Dev., 22, 40 (1983).

Forni, L., Catal. Rev. Sci. Eng., 8, 65 (1973).

Frost, D.C., McDowell, C.A., and Woolsey, I.S., Mole. Phys., 27, 1473 (1974).

Fuhrhop, J.H., Z. Naturforsch., 25b, 255 (1970).

Fuhrhop, J.H., in "The Porphyrins," D. Dolphin, ed., Vol. II, Chap. 5, Academic, N.Y. (1978).

Fuhrhop, J.H., and Subramanian, J., Phil. Trans. R. Soc. Lond., B273, 335 (1976).

Gajardo, P., Grange, P., and Delmon, B., J. Catal., 63, 201 (1980), see also J. Catal., 66, 469 (1980).

Galiasso, R., Blanco, R., Gonzalez, C., and Quinteros, N., Fuel, 62, 817 (1983).

Galiasso, R., and Morales, A., Appl. Catal., 7, 57 (1982).

Gates, B.C., Katzer, J.R., and Schuit, G.C.A., "Chemistry of Catalytic Processes," McGraw-Hill, N.Y. (1979).

Gissy, H., Bartsch, R., and Tanielian, C., J. Catal., 65, 150 (1980).

Gleim, W.K.T., and Gatsis, J.G., U.S. Patent 3,165,463 (1965); British Patent 1,327,635 (1973).

Gould, K.A., ACS Prepr., Div. Petrol. Chem., 25 (2), 288 (1980).

Gouterman, M., in "The Porphyrins," D. Dolphin, ed., Vol. III, Chap. 1, Academic, N.Y. (1978).

Grange, P., Catal. Rev. Sci. Eng., 21, 135 (1980).

Grigg, H., in "The Porphyrins," D. Dolphin, ed., Vol. II, Chap. 10, Academic, N.Y. (1978).

Hall, G., and Herron, S.P., in "Chemistry of Asphaltenes," J.W. Bunger and N.C. Li, eds., Chap. 8, Adv. Chem. Series, No. 195, American Chemical Society, Washington, D.C. (1981).

Hambright, P., personal communication, June, 1982.

Hardin, A.H., Packwood, R.H., and Ternan, M., ACS Prepr., Div. Petrol. Chem., 23 (4), 1450 (1978). - 241 -

Hirmnelblau, D.M., Jones, C.R., and Bischoff, K.B., Ind. Eng. Chem. Fundamentals, 6, 539 (1967).

Hoard, J.L., in "Porphyrins and Metalloporphyrins," K.M. Smith, ed., Chap. 8, Elsevier (1975).

Horvath, C., ed., "High Performance Liquid Chromatography - Advances and Perspectives," Vol. 1 - 3, Academic Press, N.Y. (1980-83).

Hou, P., and Wise, H., J. Catal., 78, 469 (1982).

Hung, C.W., "The Kinetics of Hydrodemetallation of Metalloporphyrins," Ph.D. Thesis, M.I.T., (August, 1979).

Hung, C.W., and Wei, X., Ind. Eng. Chem. Process Des. Dev., 19, 250, 257 (1980).

Ignasiak, T., Bimer, J., Samman, N., Montgomery, D.S., and Strausz, O.P., in "Chemistry of Asphaltenes," Adv. Chem. Series, No. 195, J.W. Bunger and N.C. Li, eds., American Chemical Society, Washington, D.C. (1981), p. 183.

Johansen, J.E., Angst, C., Kratky, C., and Eschenmoser, A., Angew. Chem. Int. Ed. Engl., 19, 141 (1980).

Johansen, J.E. Piermattie, V., Angst, C., Diener, E., Kratky, C., and Eschenmoser, A., Angew. Chem. Int. Ed. Engl. 20, 261 (1981).

Journal of Chromatography, 214, B237 (1981).

Kameyama, H., and Amano, A., J. Japan Petrol. Inst., 25, 118 (1982).

Kameyama, H., Sugishima, M., Yamada, M., and Amano, A., J. Japan Petrol. Inst., 24, 317 (1981).

Katzer, J.R., and Sivasubramanian, R., Catal. Rev. Sci. Eng., 20, 155 (1979).

Katzer, J.R., Stiles, A.B., and Kwart, H., US-DOE Report No. DOE/ET/ 03297-1, Washington, 1980.

Kincaid, J., Nakamoto, K., j. Inorg. Nucl. Chem., 37, 85 (1975).

Kiskinova, M., and Goodman, D.W., Surf. Sci., 108, 64 (1981).

Kordulis, C., Voliotio, S., and Lycourghiotis, A., J. Less-Comm. Metals, 84, 187 (1982).

Kwan, T., and Sato, M., Nippon Kagaku Zasshi, 91 (12), 1103 (1970).

Kwart, H., Schuit, G.C.A., and Gates, B.C., J. Catal., 61, 128 (1980).

Labrecque, J.J., Appl. Spectros., 30, 625 (1976). - 242 -

Laine, J., Brito, J., and Yunes, S., in "Proceedings Climax Third Inter- national Conference on the Chemistry and Uses of Molybdenum," H.F. Barry and P.C.H. Mitchell, eds., Climax Molybdenum Company, Ann Arbor, (1980), p. 111.

Larson, O.A., and Beuther, H., ACS Prepr., Div. Petrol. Chem., 11 (2), B95 (1966).

Long, R.B., in "Chemistry of Asphaltenes," J.W. Bunger and N.C. Li, eds., Chap. 2, Adv. Chem. Series, No. 195, American Chemical Society, Washington, D.C. (1981).

Longo, F.R., Finarelli, J.D., and Schmalzbach, E., J. Phys. Chem., 74, 3296 (1970).

Lycourghiotis, A., Defosse, C., Delannany, F., and Delmon, B., J. Chem. Soc. Faraday Trans. I, 76, 2052 (1980a).

Lycourghiotis, A., Defosse, C., and Delmon, B., Bull. Soc. Chim. Belg., 91, 613 (1982).

Lycourghiotis, A., Defosse, C., Delannany, F., Lemaitre, J., and Delmon, B., J. Chem. Soc. Faraday Trans. I, 76, 1677 (1980).

Lycourghiotis, A., Vattis, D., Karaiskakis, G., and Katsanos, N., J. Less-Comm. Metals, 86, 137 (1982a).

MacKay, K.M., and MacKay, R.A., "Introduction to Modern Inorganic Chemistry," International Textbook Company, London (1973).

Martinez, N.P., and Mitchell, P.C.H., in "Proceedings Climax Third International Congress on the Chemistry and Uses of Molybdenum," H.F. Barry and P.C.H. Mitchell, eds., Climax Molybdenum Company, Ann Arbor (1980), p. 105.

Massoth, F.E., Adv. in Catal., 27, 265 (1978).

Maternova, J., Appl. Catal., 3, 3 (1982).

Maternova, J., Appl. Catal., 6, 61 (1983).

McKinley, J.B., in "Catalysis," P.H. Emmett, ed., Vol. 5, Reinhold, N.Y. (1957), p. 405.

Mears, D.E., Ind. Eng. Chem. Process Des. Dev., 10, 541 (1971).

Mitchell, P.C.H., in "Catalysis," Vol. 4, C. Kemball and D.A. Dowden, eds., Chap. 7, The Royal Society of Chemistry, London (1980).

Mitchell, P.C.H., and Valero, J.A., React. Kinet. Catal. Lett., 20, 219 (1982).

Montagna, A.A., and Shah, Y.T., Ind. Eng. Chem. Process Des. Dev., 14, 479 (1975). - 243 -

Morales, A., and Galiasso, R., Fuel, 61, 13 (1982).

MuraliDhar, G., Massoth, F.E., and Shabtai, J., Paper #B-5, presented at Seventh North American Meeting of the Catalysis Society, Boston, MA, October, 1981.

Neel, K.R., "Gilsonite," in Kirk-Othmer Encyclopedia of Chemical Technology, Third Edition, 11, 802 (1983).

Nelson, N., and Levy, R.B., J. Catal., 58, 485 (1979).

Ofner, S., "Synthesis of Isobacteriochlorins and of C,D-Tetradehydrocorrins," Ph.D. Thesis, ETH-Zurich (1981), Diss. ETH No. 6939.

Ohtsuka, T., Catal. Rev. Sci. Eng., 16, 291 (1977).

Oil and Gas Journal, 78, No. 50, 34, December 15, 1980.

Oleck, S.M., and Sherry, H.S., Ind. Eng. Chem. Process Des. Dev., 16, 525 (1977).

Oxenreiter, M.F., Frye, C.G., Hoekstra, G.B., and Sroka, J.M., Paper presented at the Jap. Petrol. Inst. Fuel Oil Desulf. Symp., Tokyo, Japan, November, 1972.

Perry, R.H. and Chilton, C.H., eds., "Chemical Engineers' Handbook," Fifth Edition, McGraw-Hill, N.Y. (1973).

Plumail, J.C., Jacquin, Y., Martino, G., and Toulhoat, H., ACS Prepr., Div. Petrol. Chem., 28 (3), 562 (1983).

Ramaswamy, A.V., Sivasanker, S., and Ratnasamy, P., J. Catal., 42, 107 (1976).

Rankel, L.A., ACS Prepr., Div. Petrol. Chem., 26 (3), 689 (1981).

Rankel, L.A., and Rollmann, L.D., Fuel, 62, 44 (1982).

Rasetti, V., "Synthesis of Nickel (II) - A1 8 ,1 9-dehydrocorrinates and Isobacteriochlorins," Ph.D. Thesis, ETH-Zurich (1979), p. 77.

Ratnasamy, P., and Knozinger, H., J. Catal., 54, 155 (1978).

Ratnasamy, P., Sharma, D.K., and Sharma, L.D., J. Phys. Chem., 78, 2069 (1974).

Reid, R.C., Prausnitz, J.M., Sherwood, T.K., "The Properties of Gases and Liquids," McGraw-Hill, N.Y. (1977).

Reynolds, W.L., and Dittenhaffer, D., ACS Prepr., Div. Petrol. Chem., 25 (2), 280 (1980). - 244 -

Riley, K.L., ACS Prepr., Div. Petrol. Chem., 23, 1104 (1978).

Rousseau, K., and Dolphin, D., Tetrahedron Letters, No. 48, 4251 (1974).

Salim, S.S., and Bell, A.T., Fuel, 61, 745 (1982).

Sato, M., Takayama, N., Kurita, S., and Kwan, T., Nippon Kagaku Zasshi, 92, 834 (1971).

Satterfield, C.N., "Mass Transfer in Heterogeneous Catalysis," MIT Press, Cambridge, MA (1970).

Satterfield, C.N., AIChE J., 21, 209 (1975).

Satterfield, C.N., and Cocchetto, J.F., Ind. Eng. Chem. Process Des. Dev., 20, 53 (1981).

Satterfield, C.N., and Gultekin, S., Ind. Eng. Chem. Process Des. Dev., 20, 62 (1981).

Satterfield, C.N., Modell, M., and Wilkens, J.A., Ind. Eng. Chem. Process Des. Dev., 19, 154 (1980).

Satterfield, C.N., and Yang, S.H., J. Catal., 81, 335 (1983), see also Ind. Eng. Chem. Process Des. Dev., in press (1983).

Scheer, H., in "The Porphyrins," D. Dolphin, ed., Vol. II, Chap. 1, Academic, N.Y. (1978).

Scheer, H., and Inhoffen, H.H., in "The Porphyrins," D. Dolphin, ed., Vol. II, Chap. 2, Academic, N.Y. (1978).

Sebastian, H.M., Lin, H.M., and Chao, K.C., AIChE J., 27, 138 (1981).

Shah, Y.T., and Paraskos, J.A., Ind. Eng. Chem. Process Des. Dev., 14, 368 (1975).

Sidorov, A.N., Biophysics, 10, 250 (1965).

Sie, S.T., "Catalyst Deactivation: Studies in Surface Science and Catalysis," Vol. 6, B. Delmon and G.F. Froment, eds., Elsevier (1980), p. 545.

Silbernagel, B.G., J. Catal., 56, 315 (1979).

Sivasanker, S., Ramaswamy, A.V., and Ratnasamy, P., in "Proceedings Climax Third International Conference on the Chemistry and Uses of Molybdenum," H.F. Barry and P.C.H. Mitchell, eds., Climax Molybdenum Company, Ann Arbor (1980), p. 98.

Smith, K.M., ed., "Porphyrins and Metalloporphyrins," Elsevier (1975). - 245 -

Smith, K.M., in "Porphyrins and Metalloporphyrins," K.M. Smith, ed., Chap. 1, Elsevier (1975a).

Speight, J.G., "The Desulfurization of Heavy Oils and Residua," Marcel Dekker, Inc., N.Y. (1981).

Speight, J.G., and Moschopedis, S.E., in "Chemistry of Asphaltenes," J.W. Bunger and N.C. Li, eds., Chap. 1, Adv. Chem. Series, No. 195, American Chemical Society, Washington, D.C. (1981).

Spry, J.C., and Sawyer, W.H., Paper presented at 68th Annual AIChE Meeting, Los Angeles, November, 1975.

Stanulonis, J.J., and Pederson, L.A., ACS Prepr., Div. Petrol. Chem., 25, 255 (1980).

Stephenson, D.A., and Binkowski, J. Non-Crystalline Solids, 22, 399 (1976).

Subramanian, J., and Fuhrhop, J.H., in "The Porphyrins," D. Dolphin, ed., Vol. II, Chap. 8, Academic, N.Y. (1978).

Sugihara, J.M., Branthaver, J.F., Wu, G.Y., and Weatherbee, C., ACS Prepr., Div. Petrol. Chem., 15, C5 (1970).

Sychra, V., Lang, I., and Sebor, G., Prog. Analyt. Atom. Spectrosc.,4, 341 (1981).

Tamm, P.W., Harnsberger, H.F., and Bridge, A.G., Ind. Eng. Chem. Process Des. Dev., 20, 262 (1981).

Thomas, D.W., and Martell, A.E., Archives of Biochemistry and Biophysics, 76, 286 (1958).

Thomas, D.W., and Martell, A.E., J. Am. Chem. Soc., 81, 5111 (1959).

Todo, N., Kabe, T., Ogawa, K., Kurita, M., Sato, T., Shimada, K., Kuriki, Y., Oshima, T., Takematsu, T., and Kotera, Y., Kogyo Kayaku Zasshi, 74, 563 (1971).

Topsoe, N., J. Catal., 64, 235 (1980).

Towson, D., "Tar Sands," in Kirk-Othmer Encyclopedia of Chemical Technology, Third Edition, 22, 601 (1983).

Treibs, A., Angew. Chem., 49, 682 (1936).

Tsutsui, M., and Taylor, G.A., in "Porphyrins and Metalloporphyrins," K.M. Smith, ed., Chap. 7, Elsevier (1975).

Turnock, P.H., Sekiyu Gakkai Shi, 19, 151 (1976). - 246 -

Tuttle, M.L., Dean, W.E., and Parduhn, N.L., ACS Prepr., Div. Fuel Chem., 28 (3), 85 (1983). van Dongen, R.H., Bode, D., van der Eijk, H., and van Klinken, J., Ind. Eng. Chem. Process Des. Dev., 19, 630 (1980).

Vaughan, G.B., Tynan, E.C., and Yen, T.F., Chem. Geol., 6, 203 (1970).

Venuto, P.B., and Habib, E.T., Jr., "Fluid Catalytic Cracking with Zeolite Catalysts," Marcel Dekker, Inc., N.Y. (1979).

Vrinat, M.L., Appl. Catal., 6, 137 (1983).

Walton, R.A., J. Less-Comm. Metals, 54, 71 (1977).

Wei, J., J. Catal., 1, 526, 538 (1962).

Wei, J., and Prater, C.D., Adv. Catalysis, 13, 203 (1962).

Wei, J., and Wei, R.G., Chem. Eng. Commun., 13, 251 (1982).

Wiser, W.H., Report No. DOE/ET/14700-8 for DOE Contract No. DE-AC22- 79Etl4700 (1981).

Wiser, W.H., Massoth, F.E., and Shabtai, J., DOE Report No. DOE/ET/ 14700-4 to 9 (1980-81).

Wojtowicz, R.J., personal communication, December, 1981. (Nujol properties: Tc = 864.8*K; Pc = 146.42 psi; Vc = 1378 cm 3mole1 ; 6 = 7.41 (cal/ml)1/2) See also Wojtowicz, R.J., and Bigot, P.C., Oak Ridge National Laboratory/M.I.T. Report No. 344, June, 1982.

Worthington, P., Hambright, P., Williams, R.F.X., Reid, J., Burnham, C., Shamim, A., Turay, J., Bell, D.M., Kirkland, R., Little, R.G., Datta-Gupta, N., and Eisner, U., J. Inorg. Biochem., 12, 281 (1980).

Yang, S.H., "Catalytic Hydrodenitrogenation and Hydrodeoxygenation of Model Compounds in a Trickle Bed Reactor," Sc.D. Thesis, M.I.T., (October, 1982).

Yang, S.H., and Satterfield, C.N., J. Catal., 81, 168 (1983).

Yen, T.F., Energy Sources, 1, 447 (1974).

Yen, T.F., "The Role of Trace Metals in Petroleum," Ann Arbor Science Publishers, Inc., Michigan (1975).

Yen, T.F., in "The Role of Trace Metals in Petroleum," T.F. Yen, ed., Chap. 10, Ann Arbor Science Publishers, Inc., Michigan (1975a). - 247 -

Yen, T.F., Energy Sources, 3, 339 (1978).

Zingg, D.S., Makovsky, L.E., Tischer, R.E., Brown, F.R., and Hercules, D.M., J. Phys. Chem., 84, 2898 (1980). - 248 -

Appendix A

Design Considerations to Eliminate Heat and Mass Transfer

Limitations in the Reactors

To test for transport limitations in the reactors, physical proper-

ties of the oil and catalyst were needed. Oil properties were calculated

for Nujol at 350*C:

heat capacity, Cp = 0.73 cal/g*C, Perry and Chilton (1973) Eq. 3-43 viscosity, p = 0.00125 g/cm s, Perry and Chilton (1973) Eq. 3-119 -4 thermal conductivity, kt = 2.OX10 cal/cm s*C, Perry and Chilton (1973) Eq. 3-83

density, p = 0.65 g/cm , Reid et al. (1977) p. 67 -5 2 porphyrin diffusivity, D = 4.8X10 0 cm /s, Reid et al. (1977) p. 567 (based on molecular diameter of 15 A)

C 'P Prandtl No., N = - 4.56 pr kt

Schmidt No., N = = 40 sc pDAB -3 catalyst thermal conductivity,A = 0.33X10 cal/s cm*C, Satterfield (1970) p. 170 c

effective thermal conductivity of bed, k = lX10 4 cal/cm s*C

All of the following criteria are taken from the work of Mears (1971).

I) Calculations for the 0.52 cm I.D. reactor are based on a catalyst

particle size of d = 0.008 cm (170-200 mesh). Heat and mass transfer p coefficients are calculated from Satterfield (1970) p. 80, based on a

Reynolds number for the lowest flow used, 1 ml/br:

d G N = = 0.0074 Re P - 249 -

where G = QP/SA, superficial mass velocity, g/scm2 for this low flow, the correlation

= 1.09 D 2/3 NR is used to calculate JD from which k , the solid/liquid mass transfer coefficient is calculated (Satterfield, 1970)

J G k D - -2 c 2 /3 =.lXlO cm/sec

and h, the solid/liquid heat transfer coefficient is calculated assuming JH = JD

JHP C G -2 2 h - 2/3 - 2.2X10 cal/cm S*C Pr

The void space in the bed, c, is taken to be 0.4. a) Intraparticle mass transfer:

Based on the initial rate of Ni-porphyrin removal at 65 ppm Ni, the observed rate of reaction, R, is approximately 5.0 X 10-3 g Ni/cm3 hr. The following criteria

2 Rr2 = 0.48 < 1 s eff indicates no intraparticle mass transfer limitations. This has also been experimentally verified as shown in Figure A-1. Data obtained on

170-200 mesh (75-88 p diameter) and 325-400 mesh (38-43 p diameter) catalyst are identical demonstrating the absence of mass transfer artifacts. - 250 -

80 I I I I I

704 MESH SIZE lA 325/400 170/200 TOTAL Ni A A 60 Ni-P 0

50 - A

a.4 - 4 bJ

Z 30 A

20k A A A

10 0 0

0 0 a I I S6 I I* 0.0 0.1 0.2 0.3 0.4 0.5 0.6 G CAT. HR W/Q) ML OIL

Figure A-1. Demonstration of the absence of intraparticle diffusional limitations in the demetallation of Ni-T3MPP at 345 *C and 6.99 MPa H 2 (1000 psig) on the oxide catalyst. - 251 -

b) Intraparticle heat transfer:

An overall activation energy E = 35 kcal/mole and a heat of reaction AH = 1.0 kcal/g Ni (58.7 kcal/mole), are assumed. The following relation is calculated to be valid:

JAHJRr2 TR AT < E c

1 X 10 < 3.5 X 10 2 indicating no intraparticle heat transfer effects. c) Interphase mass transfer:

This criteria, expressed in terms of a Damkohler number

Rr _- Ck 1.2 X 10 < 0.15 C bbk c demonstrates no interphase resistance which is expected considering the absence of intraparticle resistance. d) Interphase heat transfer:

A similar conclusion can be drawn for heat transfer:

AHfRr RT < 0.15 E hT

3.8 X 10 7 < 5.3 X 10-3 e) Interparticle Transport

Interparticle transport effects, both radially and axially, generally become the limiting resistances in experimental reactors. Axial conduction can be neglected for beds with length to particle diameter ratios - 252 -

sufficiently large that plug flow is approached (L/d > 30). The 10 cm p bed length used is well above the 0.24 cm required for the 0.008 cm particles. In the radial direction it is sometimes necessary to enhance the transport by dilution of the catalyst with an inert. This effectively decreases the ratio of catalyst volume to wall area for heat transfer.

When heat transfer resistance at the wall is negligible (R0 /r > 100) as is approximately the case here, the criteria

IAHIR R2 _b 0 < 0.4 RT k T 0.E effE can be used where R is the reactor radius and Rb, the average reaction rate in the bed, is given by

- (1-sC) R b (1+b) in which C is the bed void fraction and b is the ratio of diluent to catalyst volume.

This criteria can be used to determine the bed dilution ratio, b, needed. The equation simplifies to

1+b > 0.06 which indicates dilution is not required from a radial heat transfer aspect.

To minimize axial dispersion the practical solution is to lengthen the reactor bed. -The most common form of the criteria for first order reaction is - 253 -

L 20 Co d Pe C p a f in which Pea is the axial Peclet number based on particle diameter calculated to be 0.4 and Co, Cf the inlet and outlet concentrations. Tak- ing Co = 65 and C = 1, the required reactor length is calculated to be

L > 1.7 cm. A reactor length of 10 cm will satisfy this criteria.

Dilution of the bed to minimize radial gradients and to insure no axial dispersion is not without hazard. Large dilution ratios can lead to bypassing in the reactor. The upper bound on b is calculated from

L b > 250 - d6 p where 6 is an order of magnitude less than the resulting experimental error (percent) in the conversion. For a 10 cm bed with 6 = 1, the maximum dilution ratio is 5. Hence the dilution ratio of 2 used does not lead to significant channeling in the reactor.

II) Similar calculations using the same criteria are made for the 1.75 cm

I.D. reactor packed with 1/16" extrudates. The equivalent spherical diameter of these cylindrical pellets (d = 0.152 cm, tP = 0.5 cm) is required and calculated according to Satterfield (1970):

d = (d t + 1/2 d2)1/2 s p p p

The value determined is d = 0.30 cm. Heat and mass transfer coefficients s for the solid/liquid interface at a flow of 10 cm3 /hr are calculated to be h = 1.9 X 10 3.cal/cm2 s*C and k = 9.3 X 10-4 cm/s. The effective- c ness factor for the diffusion limited reaction in the catalyst is - 254 -

estimated to be 0.5 yielding an observed rate of reaction for porphyrin removed of Rn = 2.5 X 10-3 g Ni/ml hr. a) The observed metal deposition profiles demonstrate there is definitely intraparticle mass transfer. b) The calculation for intraparticle heat transfer indicates a heat of reaction well in excess of 1000 kcal/mole would be required for limitations to exist. c) The Damkohler number for interphase mass transfer based on a flow of 10 cm 3/hr is of the same order of magnitude as 0.15. Flows greater than 10 cm 3/hr were always used in the 1.75 cm I.D. reactor to help aleviate this problem and it was assumed that this resistance was not important. d) The criterion for interphase heat transfer was easily met indicating no limiting resistances. e) To assure plug flow behavior of the reactor and avoid dispersion effects it was necessary to dilute the catalyst bed. Based on the criteria

20 Co L/d > 2 n - p Pe a C f for Pea = 1 and Co = 65, Cf = 15 ppm, it was determined that L > 9.0 cm.

Assuming a bed length of 12 cm, the error encountered in using a dilu-

tion ratio b = 2.5 was 1.5%. Calculations showed that this dilution ratio was sufficient to avoid limitations in radial heat transfer. - 255 -

Appendix B

Solution of the Kinetic Model for Ni-T3MPP Demetallation

The reaction network for Ni-T3MPP proposed in Section III.A.2 is

given by:

1 3 7 Ni-P Ni-PH2 Ni-PH4 -- deposit 2 4

5 Ni-X -- deposit 6

The kinetic rate expressions describing the concentrations of the

individual nickel species are:

dA1 A =-k A + k2A B.1 dt 1A1 k2A2

dA 2 k A dt -k2A2 - 3A2 + k4A3 B.2

dA 4 k A3 + k5A3 - k7A3 k3A2 4 B.3

dA dAt B.4 k5A3 - k6A4

where A, = Ni-P, A2 = Ni-PH2, A3 = Ni-PH4 and A4 = Ni-X.

This can be written as

A 0 0 1 A A d -(k 2+k ) 0 2 k 3 k4 A 2 dt B.5 A 0 3 0 3 k -(k 4+k 5+k7) A3 A 0 0 k A - 4 5 - 256 -

which simplifies to:

dt -A = K A dt B.6

The initial condition for eq. B.5 is the feed composition to the reactor

at t=O A1 C

A 2 B.7 A3 0

A4 0

The solution to eq. B.6 with initial condition given by eq. B.7 is obtained by the Wei-Prater (1962) technique as:

A exp ('n 1 t) 0 0 0 C 1 0 A 0 exp (n t) 0 0 0 2 2 1 B.8

A 3 0 0 exp(T t) 0 0 3 A 0 0 0 0 _ exp ( 4 t) where X and X are the eigenvector matrix and inverse matrix of the rate coefficient matrix K and T1i, n12 J3,I and n its characteristic eigenvalues.

With this solution, the method of least squares can be used to evaluate values for the seven rate parameters (used to calculate X, X, and ri.) from the experimental data. A computer program written for this purpose is listed in Appendix C.l. (This program actually uses total metal as the fourth species rather than Ni-X. The solution to these equations is analogous to the approach outlined here).

Alternatively, the seven rate parameters may be evaluated from eqs. B.l - B.4 directly using the Himmelblau, Jones, and Bischoff (1967) - 257 -

technique. The computer program for this is listed in Appendix C.

Concentration versus contact time profiles using the kinetic paramters determined by either approach were calculated from eq. B.8 using the computer program in Appendix C. - 258 -

APPENDIX C

COMPUTER PROGRAMS FOR KINETIC AND METAL

PROFILE CALCULATIONS

This Appendix contains the FORTRAN programs used to calculate kinetic rate coefficients from the experimental data, concentration profiles to model the data, and metal deposition profiles for the nickel porphyrins.

The following programs were used:

OXIDE FORTRAN - Calculates kinetic rate coefficients (k's) from Ni-EP

(Ni-P) and Ni-EPH 2 (Ni-PH2 ) data using the Himmelblau,

Jones, Bischoff (HJB) technique. Concentration profiles

are then calculated with these k's using the Wei-Prater

solution to the coupled set of rate equations.

NTMHJB7 FORTRAN - Calculates kinetic rate coefficients (k1 to k7 ) for

the Ni-T3MPP reaction sequence from Ni-P, Ni-PH Ni-PH ,

and Ni-X data using the HJB technique.

HDM FORTRAN - Calculates kinetic rate coefficients (k1 to k7 ) for the

Ni-T3NPP reaction sequence from Ni-P, Ni-PH Ni-PH and

Total Metal data using a least squares fitting routine to

the Wei-Prater solution to the coupled set of rate equations.

WPTWOX FORTRAN - Calculates concentration profiles for the Ni-T3MPP

reaction sequence using known values of the kinetic

rate parameters (k1 to k7 ) in the Wei-Prater solution

to the coupled set of rate equations. Calculates Ni-P, - 259 -

Ni-PH 2 , Ni-PH 4, and Ni-X. Total Metal is obtained

by summing these four species.

METALVD FORTRAN - Calculates metal distribution profiles in catalyst

pellets throughout the reactor and effluent oil

composition for the Ni-T34PP reaction scheme with

diffusion. Solution obtained using Wei's technique

with known values for the kinetic rate parameters

(k1 to k7 ) and diffusion coefficients (D1 to D 4 ). - 260 -

C OXIDE FORTRAN ROBERT WARE OXI00010 C THIS PROGRAM CALCULATES RATE CONSTANTS K1, K2, K3 BY THE HJB METHOD 0X100020 C FOR M-PORPHYRIN ON CHLORIDED FORM OF HDS-16A CATALYST. 0X100030 C TlHIS CAN ALSO BE USED FOR NI-ETIO ON THE OXIDE CATALYST. OXI00040 C INPUT M-PORPHYRIN AND M-CHLORIN IN PPM VS CONTACT TIME IN 0X100050 C GCATHR/ML. 0X100060 C ND=NUMBER OF DATA POINTS, IP-NUMBER OF TIME INTERVALS FOR INTEGRATION.OXIO0070 C L-M-P, 2=M-C. OXIO0080 C THIS PROGRAM USES ICSICU, DCSQDU, LINV2F, VMULFF PROGRAMS. THESE ARE OXIO0090 C IMSL PACKAGES. OXI00100 C ICSICU=INTERPOLATORY APPROXIMATION BY CUBIC SPLINES WITH ARBITRARY OXI00110 C SECOND DERIVATIVE END CONDITIONS. OX100120 C DCSQDU=INTEGRATE A CUBIC SPLINE BETWEEN LIMITS A AND B. 0X100130 C LINV2F=INVERSION OF A MATRIX-FULL STORAGE MODE-HIGH ACCURACY SOLUTION.0XI00140 C VMULFF=MATRIX MULTIPLICATION-FULL STORAGE MODE. OX100150

DIMENSION C(30,2),T(30),Y(30,2),W(2),BPAR(4),XI(30),YI(30) 0X100170 DIMENSION XX(30,2),A(3,3),G(3,1),X(30,2,3),AINV(3,3),WKAREA(70) 0X100180 DIMENSION B(3,1),C(30,3),C2(30,3) OXIQO190 READ(5,*) ND 0X100200 IP=ND-1 0X100210 DO 10 I=1,ND OX100220 READ(5,*) T(I),CO(I,1),CO(I,2) 0X100230 10 CONTINUE OXI00240 Y1=0.0 0X100250 Y2-0.0 OXI00260 DO 15 1=2,ND 0X100270 11=1-1 0X100280 Y(i1,1)=CO(I,1)-CO(1,1) OXI00290 Y(11,2)-CO(I,2)-CO(1,2) 0X100300 Y-Y1+Y(11,1) OXI00310 Y2=Y2+Y(I1,2) OXI00320 15 CONTINUE OXI00330 YS1=Y1/FLOAT(IP) OX100340 YS2=Y2/FLOAT(IP) 0X100350 W(1)=0.0 0X100360 W(2)=0.0 OXI00370 IC=30 OXI00380 BPAR(1)=O.0 0X100390 BPAR(2)=0.O 0X100400 BPAR(3)=0.0 0X100410 BPAR(4)=0.0 0X100420 DO 30 1-1,IP 0X100430 NX=I+1 0X100440 XI(1)=T(1) 0X100450 YI(1)=CO(1,1) OXI00460 DO 40 L-2,NX - OX100470 XI(L)=T(L) 0X100480 YI(L)-CO(L,1) 0X100490 40 CONTINUE OXIO0500 AL-XI(1) OXIO0510 - 261 -

BU-XI(NX) OXI00520 CALL ICSICU(XIYI,NX,BPAR,Cl,IC,IER) OXI00530 CALL DCSQDU(XI,YI,NX,Cl,IC,AL,BU,Ql,IER2) OXIO0540 XX(I, 1)=Q1 OXIO0550 W(l)-W(I)+(Y(I,1)-YS1)**2 OXI00560 YI(l)-CO(1,2) OXI0O570 DO 41 L-2,NX OXI0O580 YI(L)-CO(L,2) 0X100590 41 CONTINUE OXI00600 CALL ICSICU(XIYINX,BPAR,C2,IC,IER3) 0XI00610 CALL DCSQDU(XIYINXC2,ICALBUQ2,IER4) OXI00620 XX(I,2)=Q2 0XI00630 W(2)=W(2)+(Y(I,2)-YS2)**2 0XI0O640 30 CONTINUE 0XI00650 W(1)-1.O/W(1) OX100660 W(2)=1.0/W(2) 0XI00670 DO 45 I=1,IP OXI00680 X(I,1,1)=-XX(I,1) 0XI00690 X(I,1,2)=XX(I,2) OXIO0700 X(I,1,3)=0.0 OXI00710 X(I,2,1)-XX(I,1) 0X100720 X(1,2,2)--XX(I,2) OXI00730 X(I,2,3)=-XX(I,2) OXI00740 45 CONTINUE 0X100750 DO 50 K-1,3 OXI00760 DO 60 L=1,3 0X100770 SUMAO0.0 OX100780 DO 70 1=1, ip OXI0790 SUMA=SUMA+W(1)*X(I,1,K)*X(IJL)+W(2)*X(I,2,K)*X(I,2,L) 0X100800 70 CONTINUE 0XI00810 A(KL)-SUMA OXI00820 60 CONTINUE OXIO0830 SUMG=0.0 0XI00840 DO 80 I-1,IP OXI00850 SUMG-SUMG+W(1)*Y(I,1)*X(I,1,K)+W(2)*Y(1,2)*X(I,2,K) OXI00860 80 CONTINUE 0XI00870 G(K,1)=SUMG 0XI00880 50 CONTINUE 0X100890 WRITE(6,130) 0X100900 130 FORMAT(' INITIAL CONC. IN PPM METAL') OX100910 WRITE(6,135)(CO(1,J),J=1,2) 0X100920 135 FORMAT(lH,' NI-P=',F6.2,/,1H.,' NI-C=',F6.2) OXI00930 N=3 0X100940 IA=3 0XI00950 IDGT-3 OXI00960 CALL LINV2F(ANIAAINVIDGT,WKAREA,IER3) OX100970 IC-3 OXI00980 IB-3 0X100990 N-1 oXi01000 M=3 OXIolo]o L-3 0X101020 CALL VMULFF(AINV.,G , L ,M, NIA ,IB,B,IC, IER4) 0X101030 WRITE(6,110) B(1,1),B(2,1),B(3,1) OX101040 - 262 - 110 FORMAT(' Klu',E12.4,/,' K2=',E12.4,/, ' K3-',E12.4) OX101050 C CALCULATION OF CONCENTRATION VS W/Q USING ANALYTICAL SOLUTION OXIO1060 C OF WEI-PRATER TECHNIQUE. OX101070 AK B(1.,1) 0X101080 AK2-B(2,1) OXI01090 AK3=B(3,1) 0X101100 SUMK=AK1+AK2+AK3 OXIo 110 R-SQRT(SUMK**2-4.0*AKI*AK3) OXI01120 AL1=(-SUMK+R)/2.0 OXIO1130 AL2=(-SUMK-R)/2 .0 OXI01140 WRITE(6,140) ALI, AL2 OX101150 140 FORMAT(' LAMBDA1 = ',E12.4,/' LAMBDA2 - ',E12.4) OXI01160 A10=CO(1,1) OX101170 A20=CO(1,2) 0X101180 CC=AK2 /(AK1+AL1) OXIO 190 DD=(AK2+AK3+AL2)/AK1 0X101200 B10=(A1O-A20*DD)/(CC-DD) OXIO1210 B20=A20-B1O OXI01220 EP=0.0 OX101230 EC=0.0 0X101240 WRITE(6,150) OXI01250 WRITE(6,155) 0X101260 WRITE(6,160) T(1),CO(1,1),CO(1,1),CO(1,2),CO(1,2),CO(1,1) OXIO1270 DO 85 I=2,ND 0X101280 AP=B10*CC*EXP (ALI*T (I))+B20*DD*EXP (AL2*T (I)) 0X101290 A2=B10*EXP(AL1*T(I))+B20*EXP(AL2*T(I)) 0X101300 TM-Al+A2 0X101310 WRITE(6,160) T(I),CO(I,1),A,CO(I,2),A2,TM 0X101320 EP=(CO(I,1)-Al)**2 + EP OXI01330 EC=(CO(I,2)-A2)**2 + EC OXI01340 85 CONTINUE OXI01350 EP=EP/IP 0X101360 EC-EC/IP OXI01370 EP=SQRT(EP) OXI01380 EC=SQRT(EC) 0X101390 WRITE(6,165) EP,EC 0X101400 150 FORMAT(' SPACE TIME' ,7X,'PORPHYRIN' ,7X, 'CHLORIN' ,7X, 'TOTAL METAL')0XI01410 155 FORMAT(' GCATHR/ML',6X,'EXP',5X,'CAL',7X,'EXP',5X,'CAL',8X,'CAL') oX101420 160 FORMAT(4X,F6.4,5X,F5.2,3X,F5.2,5X,F5.2,3X,F5.2,6XF5.2) 0X101430 165 FORMAT(' EP-',E15.6,5X,'EC=',E15.6) 0X101440 STOP 0X101450 END OXI01460 - 263 -

C NTMHJB7 FORTRAN ROBERT WARE NTM000J0 C THIS PROGRAM CALCULATES RATE CONSTANTS K] TO K7 BY THE HJB METHOD NTMOO020 C FOR NI-T3MPP REACTION SEQUENCE ON HDS-16A CATALYST. NTMOO030 C ND - NO. OF DATA POINTS, IP - NO. OF TIME INTERVALS NTMOO040 C 1=NI-PORPHYRIN, 2-NI-CHLORIN, 3=NI-ISOB. CHLORIN, 4-NI-X NTMOO050 C CONC. IN PPM NI, TIME IN G CAT HR/ML OIL. NTMOO060 C THIS PROGRAM USES ICSICUDCSQDULINV2FAND VMULFF PROGRAMS. NTMOO070 C THESE ARE INTERNATIONAL MATHEMATICAL AND STATISTICAL NTM00080 C LIBRARIES, INC. (IMSL) PACKAGES. NTMOO090 C ICSICU - INTERPOLATORY APPROXIMATION BY CUBIC SPLINES WITH NTMOO100 C ARBITRARY SECOND DERIVATIVE END CONDITIONS. NTM00110 C DCSQDU = INTEGRATE A CUBIC SPLINE BETWEEN LIMITS A AND B NTMOO120 C LINV2F = INVERSION OF A MATRIX-FULL STORAGE MODE-HIGH NTM00130 C ACCURACY SOLUTION. NTMOO140 C VMULFF = MATRIX MULTIPLICATION-FULL STORAGE MODE. NTMOO150

DIMENSION Y(30,4),CO(30,4),W(4),T(30),XI(30),YI(30) NTMOO170 DIMENSION XX(30,4),X(30,4,7),A(7,7),G(7,1),AINV(7,7) NTMOO180 DIMENSION WKAREA(150),B(7,1),BPAR(4),Cl(30,3),C2(30,3),C3(30,3) NTMOO190 DIMENSION C4(30,3) NTM00200 NK=7 NTM00210 READ(5,*) ND NTM00220 IP = ND - 1 NTM00230 DO 10 I = 1,ND NTM00240 READ(5,*) T(I),CO(I,1) ,CO(I,2),CO(1,3),CO(I,4) NTM00250 10 CONTINUE NTM00260 Y1=0.0 NTM00270 Y2=0.0 NTM00280 Y3=0.0 NTM00290 Y4=0.0 NTM00300 DO 15 I=2,ND NTM00310 I-I=-1 NTM00320 Y(I1,1)=CO(I,1)-CO(1,1) NTM00330 Y(I1,2)=CO(1,2)-CO(1,2) NTM00340 Y(11,3)=CO(I,3)-C(1,3) NTM00350 Y(11,4)=CO(I,4)-CO(1,4) NTM00360 YL=Y1+Y(I1,1) NTM00370 Y2=Y2+Y(Il,2) NTM00380 Y3=Y3+Y(II,3) NTM00390 Y4=Y4+Y(11,4) NTM00400 15 CONTINUE NTM00410 YS1=Yl/FLOAT(IP) NTM00420 YS2=Y2/FLOAT(IP) NTM00430 YS3=Y3/FLOAT(IP) NTM00440 YS4=Y4/FLOAT(IP) NTM00450 W(1)=0.0 NTM00460 W(2)-0.0 NTM00470 W(3)-0.0 NTM00480 W(4)-0.0 NTM00490 IC=30 NTM00500 BPAR(1)nO.0 - NTM00510 BPAR(2)-0.0 NTM00520 - 264 - BPAR(3)-O.0 NTM00530 BPAR(4)-0.0 NTM00540 DO 30 1-1,1P NTM00550 NX=I+l NTM00560 XI(1)=T(1) NTM00570 YI(1)-CO(1,1) NTM00580 DO 40 L-2,NX NTM00590 XI(L)-T(L) NTM00600 YI(L)=CO(L,1) NTM00610 40 CONTINUE NTM00620 AL=XI(1) NTM00630 BU-XI(NX) NTM00640 CALL ICSICU(XIYINXBPARCl,IC,IER1) NTM00650 CALL DCSQDU(XIYINX,Cl,IC,ALBUQlIER2) NTM00660 XX(I,1)=Ql NTH00670 W(1)=W(1)+(Y(I,1)-YS.)**2 NTM00680 YI(1)nCO(1,2) NTM00690 DO 41 L-2,NX NTM00700 YI(L)=CO(L,2) NTM00710 41 CONTINUE NTM00720 CALL ICSICU(XI,YI,NX,BPAR,C2,IC,IER3) NTM00730 CALL DCSQDU(XIYINXC2,ICALBUQ2,1ER4) NTM00740 XX(I,2)-Q2 NTH00750 W(2)=W(2)+(Y(I,2)-YS2)**2 NTM00760 YI(1)-CO(1,3) NTM00770 DO 42 L=2,NX NTM00780 YI(L)=CO(L,3) NTM00790 42 CONTINUE NTMOO800 CALL ICSICU(XI,YI,NX,BPAR,C3,IC,IER5) NTMOO810 CALL DCSQDU(XIYINXC3,ICALBUQ3,IER6) NTMOO820 XX(I,3)=Q3 NTM00830 W(3)-W(3)+(Y(I,3)-YS3)**2 NTMOO840 YI(1)-CO(1,4) NTMOO850 DO 43 L=2,NX NTMOO860 YI(L)-CO(L,4) NTM00870 43 CONTINUE NTM00880 CALL ICSICU(XI,YI,NX,BPARC4,IC,IER7) NTM00890 CALL DCSQDU(XI,YI,NXC4,ICALBUQ4,IER8) NTM00900 XX(1,4)-Q4 NTM00910 W(4)=W(4)+(Y(I,4)-YS4)**2 NTM00920 30 CONTINUE NTM00930 W()-1.0/W(.) NTM00940 W(2)=1.0/W(2) NTM00950 W(3)=1.0/W(3) NTM00960 W(4)=1.O/W(4) NTM00970 DO 45 I-1, IP NTM00980 X(i,1,1)- -XX(I,1) NTM00990 X(I,12)- XX(I,2) NTMO1000 X(I,2,1)- XX(I1) NTMO1010 X(I,2,2)- -XX(I,2) NTH01020 X(I,2,3)- -XX(I,2) NTMO1030 X(I,2,4)- XX(1,3) NTMO1040 X(I,3,3)- XX(I, 2) NTMO1050 X(I,3,4)- -XX(I,3) NTMO1060 - 265 -

X(I,3,5)- -XX(I,3) NTMO1070 X(I,3,7)--XX(I,3) NTMO1080 X(I,4,5)=XX(I,3) NTMO1090 X(I,4,6)--XX(I,4) NTMO1100 DO 46 K-3,NK NTMO1110 X(I,1,K)-0.0 NTMO1120 46 CONTINUE NTM01130 DO 47 K-5,NK NTMO1140 X(I,2,K)-O.0 NTMOI 150 47 CONTINUE NTMO1160 X(I,3,1)= 0.0 NTMO1170 X(I,3,2)- 0.0 NTMO1 180 X(I,3,6)-0.0 NTMO1190 DO 49 K=1,4 NTM01200 X(I,4,K)=O.0 NTMO1210 49 CONTINUE NTMO1220 X(I,4,7)=0.0 NTM01230 45 CONTINUE NTM01240 DO 50 K-1,NK NTMO1250 DO 60 L-1,NK NTMO1260 SUMA-0 .0 NTMO1270 DO 70 1-1, IP NTM01280 CC=W(1)*X(I,1,K)*X(I,1qL) + W(2)*X(I,2,K)*X(I,2,L) NTMO1290 DD-W(3)*X(I,3,K)*X(I,3,L) + W(4)*X(I,4,K)*X(I,4,L) NTM01300 SUMA - SUMA + CC + DD NTMO1310 70 CONTINUE NTMO1320 A(KL)-SUMA NTMO1330 60 CONTINUE NTMO1340 SUMG = 0.0 NTMO1350 DO 80 I=1,IP NTMO1360 EE=W(1)*Y(I,1)*X(I,1,K) +W(2)*Y(I,2)*X(I,2,K) NTMO1370 FF=W(3)*Y(I,3)*X(I,3,K) + W(4)*Y(I,4)*X(I,4,K) NTM01380 SUMG=SUMG + EE + FF NTMO1390 80 CONTINUE NTMO1400 G(K,1)=SUMG NTM01410 50 CONTINUE NTM01420 WRITE(6,130) NTMO1430 130 FORMAT(' INITIAL CONC. IN PPM METAL') NTM01440 WRITE(6,135) (CO(1,J),J-1,4) NTMO1450 135 FORMAT(lH,'NI POR - ',F6.2,/,1H,'NI-C = ',F6.2,/, NTM01460 &1H,'NI-ISOBC = ',F6.2,/,1H,'NI-X ',F6.2) NTM01470 N=NK NTMO1480 IA=NK NTMO1490 IDGT=3 NTMO1500 CALL LINV2F(A,NIA,AINV,IDGT,WKAREAIER3) NTM01510 IC=NK NTMO1520 IB=NK NTMO1530 N= NTMO1540 M=NK NTM01550 L=NK NTMO1560 CALL VMULFF(AINV,GL,M,N,IA,IB,B,IC,IER4) NTMO1570 WRITE(6,110) B(1,1),B(2,1),B(3,1),B(4,1),B(5,1),B(6,1),B(7,1) NTMO1580 110 FORMAT(' Kl=',E12.4,/,' K2-',E12.4,/,' K3-',E12.4,/, ' K4=' NTMO1590 &,E12.4,/,' K5=' .E12.4./.' K6=',E12.4/,' K7=',E12.4) NTMO1600 STOP NTMO1610 END NTMO1620 - 266 -

C HDM FORTRAN ROBERT WARE HDMOOOI0 C THIS PROGRAM CALCULATES KINETIC PARAMETERS KI TO K7 FOR THE HDMOO020 C NI-T3MPP REACTION SEQUENCE FITTING DATA WITH A LEAST HDMOO030 C SQUARES FITTING ROUTINE TO THE WEI-PRATER SOLUTION TO THE HDMOO040 C COUPLED SET OF RATE EQUATIONS. HDMOO050 C INPUT DATA 1=NI-P, 2=NI-PH2, 3=NI-PH4, 4-TOTAL METAL IN PPM HDMOO060 C NI VERSUS CONTACT TIME IN G CAT HR/ML OIL. HDMOO070 C ND = NUMBER OF INPUT DATA POINTS. HDMOO080 C THIS PROGRAM USES ZXSSQ, EIGRF, LINV2F, AND VMULFF PROGRAMS. HDMOO090 C THESE ARE IMSL PACKAGES. HDMO100 C ZXSSQ = LEAST SQUARES MINIMIZATION ROUTINE. HDMOO110 C EIGRF = EIGENVALUES AND EIGENVECTORS OF A REAL GENERAL HDMOO120 C MATRIX IN FULL STORAGE MODE. HDMOO130 C LINV2F = INVERSION OF A MATRIX-FULL STORAGE MODE-HIGH HDMOO140 C ACCURACY SOLUTION. HDMOO150 C VMULFF = MATRIX MULTIPL ICATION-FULL STORAGE MODE. HDMOO160 C** ************************************************************** HDMOO170 REAL K HDMOO180 DIMENSION X(7),PARM(4),F(80),XJAC(80,7),XJTJ(28) HDMOO190 DIMENSION WORK(271),K(4,4),WK(24),RW(8),RZ(32) HDM00200 DIMENSION ZR(4,4),ZINV(4,4),WKAREA(30),AO(4),BO(4),A(4) HDM00210 COMPLEX W(4),Z(4,4) HDM00220 EQUIVALENCE (W(1),RW(1)),(Z(1,1),RZ(1)) HDM00230 COMMON/DATA/ CO(30,4),T(30),ND HDM00240 EXTERNAL FUNC HDM00250 READ(5,*) ND HDM00260 DO 15 -1=,ND HDM00270 READ(5,*) T(I),CO(I,1),CO(I,2),CO(I,3),CO(I,4) HDM00280 CO(I,4)-CO(I,1)+CO(I,2)+CO(1,3)+CO(I,4) HDM00290 15 CONTINUE HDM00300 X(1)=29.0 HDM00310 X(2)=35.6 HDM00320 X(3)=21.7 HDM00330 X(4)-33.0 HDM00340 X(5)=6.4 HDM00350 X(6)=0.2 HDM00360 X(7)=8.0 HDM00370 N=7 HDM00380 M=80 HDM00390 NSIG=2 HDM00400 EPS-0.05 HDM00410 DELTA-0.01 HDM00420 MAXFN=100 HDM00430 IOPT=1 HDM00440 IXJAC-M HDM00450 CALL ZXSSQ(FUNC,M,N,NSIG,EPSDELTAMAXFNIOPTPARM,XSSQ,F, HDM00460 &XJAC,IXJAC,, XJTJ ,WORK, INFER, IER) - HDM00470 WRITE(6,100) HDM00480 100 FORMAT(' CALCULATED RATE COEFFICIENTS') HDM00490 WRITE(6,105) X(1.),X(2),X(3),X(4),X(5),X(6),X(7) HDM00500 105 FORMAT(' K1=',E12.4,/,' K2=',E12.4,/,' K3-E12.4,/,' K4-' HDM00510 &,E12.4,/,' K5=',E12.4,/,' K6=',E12.4,/,' K7-',E12.4) HDM00520 - 267 -

WRITE(6,110) SSQ, INFER HDM00530 110 FORMAT(' SUM OF SQUARES-' ,E1O.3,/,' INFER-' ,12) HDM00540 DO 10 1=1,5 HDM00550 WRITE(6,115) IWORK(I) HDM00560 115 FORMAT(' WORK' ,I1,'i',E12.3) HDM00570 10 CONTINUE HDM00580 STOP HDM00590 END HDM00600 SUBROUTINE FUNC(XMNF) HDM00610 REAL K HDM00620 DIMENSION K(4,4),X(7),F(80),WK(24),RW(8),RZ(32),ZR(4,4),ZINV(4,4) HDM00630 DIMENSION WKAREA(30),AO(4),BO(4),A(4) HDM00640 COMPLEX W(4),Z(4,4) HDM00650 EQUIVALENCE (W(1),RW(1)),(Z(1,1),RZ(1)) HDM00660 COMMON/DATA/ CO(30,4),T(30),ND HDM00670 AO(1)=CO(1, 1) HDM00680 AO(2)=C0(1,2) HDM00690 AO(3)=C0(1,3) HDM00700 AO(4)=CO(1,4) HDM00710 K(1,1)--ABS(X(1)) HDM00720 K(1,2)=ABS(X(2)) HDM00730 K(J,3)=0.0 HDM00740 K(1,4)=0.0 HDM00750 K(2, 1)-ABS(X(1)) HDM00760 K(2,2)--(ABS(X(2))+ABS(X(3))) HDM00770 K(2,3)=ABS(X(4)) HDM00780 K(2,4)=O.0 HDM00790 K(3,1)=O.0 HDM00800 K(3,2)-ABS(X(3)) HDM00810 K(3,3)--(ABS(X(4))+ABS(X(5))+ABS(X(7))) RDM00820 K(3,4)-0.O HDMOO830 K(4,1)-ABS(X(6)) HDMOO840 K(4,2)=ABS(X(6)) HDM00850 K(4,3)=ABS(X(6))-ABS(X(7)) HDM00860 K(4,4)=-ABS(X(6)) HDMOO870 N1-4 HDM00880 IA=4 HDM00890 IJOB=2 HDM00900 IZ=4 HDM00910 CALL EIGRF(KN1, IA, IJOB,W ,Z,IZ,WKIER) HDM00920 DO 20 I=1,4 HDM00930 DO 20 J=1,4 HDM00940 ZR(I ,J)-REAL(Z(I,J)) HDM00950 20 CONTINUE HDM00960 IDGT=3 HDM00970 CALL LINV2F(ZR,N1,IA,ZINV,IDGTWKAREA,IER) HDM00980 L1-4 HDM00990 ML=4 HDMO1000 NN=1 HDMO1010 IB-4 HDMO1020 IC-4 HDMO1030 CALL VMULFF(ZINV,AO,L1,M1,NN,IA,IB,BO,IC,IER) HDMO1040 F(1)-0.0 HDMO1050 F(2)=0.0 HDMO1060 - 268 -

F(3)-O=.0 HDMO1070 F(4)=O.0 HDMO1080 JI-5 HDM01090 DO 25 I-2,ND HDMO1100 C-REAL(W(1))*T(I) HDMO1110 IF(ABS(C).GT.40.0)GO TO 41 HDM1120 C=EXP(C) HDMO1130 GO TO 42 EDMO1140 41 C-0.0 HDMO1150 42 Sl=BO(1)*C HDMO1160 C'REAL(W(2))*T(I) HDM01170 IF(ABS(C).GT.40.0)GO TO 43 HDMO1180 C=EXP(C) HDM01190 GO TO 44 HDMO1200 43 C=0.0 HDMO1210 44 S2=BO(2)*C HDM01220 C-REAL(W(3))*T(I) HDMO1230 IF(ABS(C).GT.40.0)GO TO 45 HDM01240 C=EXP(C) HDMO1250 GO TO 46 HDMO1260 45 C-0.0 HDM01270 46 S3=BO(3)*C HDM01280 C=REAL(W(4))*T(I) HDMO1290 IF(ABS(C).GT.40.0)GO TO 47 HDMO1300 C-EXP(C) HDM01310 GO TO 48 HDMO1320 47 C-0.0 HDMO1330 48 S4=Bo(4)*C HDM01340 DO 30 L=1,4 HDM01350 Ji=J1+ HDMO1360 J=J -i HDM01370 A(L)=S1*Z(Li)+S2*Z(L,2)+S3*Z(L,3)+S4*Z(L,4) HDMO1380 F(J)-CO(I,L)-A(L) HDM01390 30 CONTINUE HDMO1400 25 CONTINUE HDM01410 RETURN HDM01420 END HDMO1430 - 269 -

C WPTWOX FORTRAN ROBERT WARE WPTOOO1O C THIS PROGRAM CALCULATES CONCENTRATION VS CONTACT TIME KINETIC WPT00020 C RESULTS FOR NI-T3MPP BASED ON WEI-PRATER SOLUTION TECHNIQUE WPT00030 C FOR 7 PARAMETER MODEL WITH TWO METAL DEPOSITION STEPS WPT00040 C WITH KNOWN VALUES FOR THE RATE PARAMETERS. WPT00050 C 1-NI-P, 2-NI-PH2, 3-NI-PH4, 4=NI-X WPT00060 C THIS PROGRAM USES EIGRF, LINV2F, AND VMULFF PROGRAMS. WPT00070 C THESE ARE IMSL PACKAGES. WPT00080 C EIGRF - EIGENVALUES AND EIGENVECTORS OF A REAL GENERAL MATRIX WPT00090 C IN FULL STORAGE MODE. WPTO0100 C LINV2F - INVERSION OF A MATRIX-FULL STORAGE MODE- HIGH WPT00110 C ACCURACY SOLUTION. WPT00120 C VMULFF - MATRIX MULTIPLICATION-FULL STORAGE MODE. WPT00130 WPT00140 DIMENSION WKAREA(50),CO(30,5),BO(4),AO(4),A(4),ZINV(4,4) WPT00150 DIMENSION K(4,4),T(30),ZR(4,4),RW(10),RZ(50),WK(40) WPT00160 COMPLEX W(4),Z(4,4) WPT00170 EQUIVALENCE (W(1),RW(1)),(Z(1,1),RZ(1)) WPT00180 REAL K,K ,K2,K3,K4,K5,K6,K7,K8 WPT00190 READ(5,*) ND WPT00200 DO 5 I-1,ND WPT00210 READ(5,*) T(I),CO(I,1),CO(I,2),CO(I,3),CO(I,4) WPT00220 CO(I,5)=CO(I,1)+CO(I,2)+CO(I,3)+CO(I,4) WPT00230 5 CONTINUE WPT00240 READ(5,*) K1,K2,K3,K4,K5,K6,K7 WPT00250 WRITE(6,95) K1,K2,K3,K4,K5,K6,K7 WPT00260 95 FORMAT(' K1-',F8.2,/,' K2-',F8.2,/,' K3-',F8.2,/,' K4-',F8.2,/, WPT00270 &' K5=',F8.2,/,' K6-',F8.2,/,' K7-',F8.2) WPT00280 AO(1)=C0(1,1) WPT00290 AO(2)=CO(1,2) WPT00300 AO(3)=CO(1,3) WPT00310 AO(4)=C(1,4) WPT00320 K(1,1)--K1 WPT00330 K(1,2)=K2 WPT00340 K(1,3)-O.0 WPT00350 K(1,4)=0.0 WPT00360 K(2,1)=K1 WPT00370 K(2,2)--(K2+K3) WPT00380 K(2,3)-K4 WPT00390 K(2,4)-0.0 WPT00400 K(3,1)=0.0 WPT00410 K(3,2)-K3 WPT00420 K(3,3)--(K4+K5+K7) WPT00430 K(3,4)=0.0 WPT00440 K(4,1)-O.O WPT00450 K(4,2)-O.0 WPT00460 K(4,3)-K5 WPT00470 K(4,4)--K6 WPT00480 N-4 WPT00490 IA-4 WPT00500 IJOB-2 WPT00510 IZ-4 WPT00520 J L

- 270 -

CALL EIGRF(K,N,IA,IJOBWZIZ,WK,IER1) WPT00530 WRITE(6,100) (W(J),J-1,4),WK(1) WPT00540 100 FORMAT(' LAMl',2E12.4,/,' LAM2-',2E12.4,/, ' LAM3-',2E12.4,/, WPT00550 &' LAM4.',2E12.4,/,/,' PFIX-',F6.2) WPT00560 WRITE(6,105) WPT00570 105 FORMAT(' MATRIX OF EIGENVECTORS FROM K MATRIX') WPT00580 DO 10 1-1,4 WPT00590 WRITE(6,110) (Z(IJ),J-1,4) WPT00600 110 FORMAT(lH,8E10.3) WPT00610 10 CONTINUE WPT00620 DO 15 1-1,4 WPT00630 DO 15 J=1,4 WPT00640 ZR(I,J)=REAL(Z(I,J)) WPT00650 15 CONTINUE WPT00660 IDGT=3 WPT00670 CALL LINV2F(ZR,NIA,ZINV,IDGT,WKAREA,IER2) WPT00680 WRITE(6,112) WPT00690 112 FORMAT(' INVERSE MATRIX OF EIGENVECTORS') WPT00700 DO 17 1-1,4 WPT00710 WRITE(6,115)(ZINV(I,J),J=1,4) WPT00720 115 FORMAT(lH,4E12.4) WPT00730 17 CONTINUE WPT00740 L-4 WPT00750 M=4 WPT00760 N-1 WPT00770 IB-4 WPT00780 IC-4 WPT00790 CALL VMULFF(ZINV,AO,LM,N,IA,IB,BO,IC,IER3) WPT00800 WRITE(6,120) BO(1),BO(2),BO(3),BO(4) WPT00810 120 FORMAT(' B01=',E12.4,/,' B02=',E12.4,/, WPT00820 &' B03-',E12.4,/,' B04-',E12.4) WPT00830 WRITE(6,130) WPT00840 130 FORMAT(' SPACE TIME',6X,'PORPHYRIN',8X,'CHLORIN',9X, WPT00850 &'ISOBACT.CHLORIN',5X,'UNKNOWN',7X,'TOTAL MET') WPT00860 WRITE(6,135) WPT00870 135 FORMAT(' GCATHR/ML',7X,'EXP',4X,'CAL',7X,'EXP',4X,'CAL',8X,'EXP' WPT00880 &,5X,'CAL',5X,'EXP',3X,'CAL',6X,'EXP',4X,'CAL') WPT00890 WRITE(6,140) T(1),AO(1),AO(1),AO(2),AO(2),AO(3),AO(3),AO(4),AO(4) WPT00900 &,AO(1),AO(1) WPT00910 SSQ0.0 WPT00920 DO 20 I=2,ND WPT00930 F-REAL(W(1))*T(I) WPT00940 IF(ABS(F).GT.40.)GO TO 21 WPT00950 F=EXP(F) WPT00960 GO TO 22 WPT00970 21 F-0.0 WPT00980 22 SL-BO(1)*F WPT00990 F-REAL(W(2))*T(I) WPT01000 IF(ABS(F).GT.40.)GO TO 23 WPTO1OIO F-EXP(F) WPT01020 GO TO 24 WPT01030 23 F-0.0 WPT01040 24 S2=BO(2)*F WPT01050 F-REAL(W(3))*T(I) WPT01060 - 271 -

IF(ABS(F).GT.40.)GO TO 25 WPT01070 F=EXP(F) WPT01080 GO TO 26 WPT01090 25 F-0.0 WPT01ooo 26 S3-BO(3)*F WPT01110 F-REAL(W(4 ))*T(I) WPT01120 IF(ABS(F).GT.40.)GO TO 27 WPT01130 F-EXP(F) WPT01140 GO TO 28 WPT01150 27 F-0.0 WPT01160 28 S4-BO(4)*F WPT01170 Do 35 L-1,5 WPT01180 A(L)=Sl*Z(L,1)+S2*Z(L,2) +S3*Z(L,3)+S4*Z(L,4) WPT01190 35 CONTINUE WPT01200 TM=A(1)+A(2)+A(3)+A(4) WPT01210 WRITE(6,140) T(I),CO(I,1),A(1),CO(I,2),A(2),CO(I,3),A(3),CO(I,4), WPT01220 &A(4),Co(I,5),TM WPT01230 20 CONTINUE WPT01240 140 FORMAT(4X,F5.3,5X,F5.2,3X,F5.2,5X,F5.2,2X,F5.2,6X,F5.2, WPT01250 &3X,F5.2,4X,F5.2,3X,F5.2,4X,F5.2,3X,F5.2) WPT01260 STOP WPT01270 END WPT01280 (itt

- 272 -

C METALVD FORTRAN ROBERT WARE MET00010 C THIS PROGRAM CALCULATES THE EFFLUENT OIL COMPOSITION FROM THE MET00020 C ONE INCH REACTOR AND CALCULATES CONCENTRATION DISTRIBUTIONS IN MET00030 C CATALYST PELLETS AT VARIOUS AXIAL POSITIONS ENABLING METAL MET00040 C DEPOSITION PROFILES AS A FUNCTION OF RADIAL POSITION WITHIN A MET00050 C CATALYST PELLET TO BE EVALUATED. SOLUTION IS FOR THE NI-T3MPP MET00060 C REACTION SEQUENCE, OBTAINED BY USING WEI'S TECHNIQUE. REQUIRED TO MET00070 C INPUT THE SEVEN KINETIC PARAMETERS (Kl TO K7) AND THE FOUR SPECIES MET00080 C (NI-P,NI-PH2,NI-PH4,NI-X) DIFFUSION COEFFICIENTS. MET00090 C RHOC = CATALYST DENSITY, G/ML MET00100 C RHO = OIL DENSITY, G/ML MET00110 C R = PELLET RADIUS, CM MET00120 C TIME = TIME ON STREAM, HRS MET00130 C WCAT = WEIGHT OF CATALYST, G MET00140 C Q = OIL FLOW RATE, ML/HR MET00150 C NL NUMBER OF AXIAL POSITIONS FOR CALCULATIONS MET00160 C ASIO = NI-P INLET CONC MET00170 C AS20 - NI-PH2 INLET CONC MET00180 C AS30 = NI-PH4 INLET CONC MET00190 C AS40 = NI-X INLET CONC MET00200 C D = SPECIES DIFFUSION COEFFICIENT, CM2/SEC MET00210 C K = RATE PARAMETER, ML OIL/G CAT HR MET00220 C THIS PROGRAM USES LINV2F, VMULFF,EIGRF, MMBSIO, AND MMBSI1. MET00230 C THESE ARE IMSL PACKAGES. MET00240 C LINV2F = INVERSION OF A MATRIX-FULL STORAGE MODE-HIGH MET00250 C ACCURACY SOLUTION. MET00260 C VMULFF = MATRIX MULTIPLICATION-FULL STORAGE MODE. MET00270 C EIGRF = EIGENVALUES AND EIGENVECTORS OF A REAL GENERAL MATRIX MET00280 C IN FULL STORAGE MODE. MET00290 C MMBSIO = MODIFIED BESSEL FUNCTION OF FIRST KIND OF ZERO ORDER. MET00300 C MMBSI = MODIFIED BESSEL FUNCTION OF FIRST KIND OF FIRST ORDER. MET00310 MET00320 REAL K,K1,K2,K3,K4,K5,K6,K7,KPMMBSIO,MBSI. MET00330 COMPLEX PHIC(4),Y(4,4),W(4),XP(4,4) MET00340 EQUIVALENCE (PHIC(1) ,RPHIC(1)),(Y(1,1),RY(1)) MET00350 EQUIVALENCE (W (1) ,RW(J)) , (XP(1,1) ,RXP(1)) MET00360 DIMENSION ASO(4),AS1(10),AS2(10) ,AS3(10),AS4(10),A(4),AAIO(4) MET00370 DIMENSION B(4),BP1(10),BP2(10),BP3(10),BP4(10),BPO(4) MET00380 DIMENSION DINV(4,4),DK(4,4),D(4,4),DYR(4,4),DYRT(4,4 ),ETA(4) MET00390 DIMENSION KP(4,4), K(4,4),PHI(4),RW(8),RPHIC(8),RY(32),RXP(32) MET00400 DIMENSION T(10), THETA(4,4),WKAREA(28),WKI(24),WK2(24),XPR(4,4) MET00410 DIMENSION AS(4),BS(4), XPINV(4,4) ,YR(4,4) ,YINV(4,4) MET00420 READ (5, *) K1,K2,K3,K4,K5,K6,K7 MET00430 READ(5,*) AS10 , AS20 , AS30 ,AS40 MET00440 READ(5,*) RHOCRHORTIMEWCATQNL MET00450 READ(5,*) D1,D2,D3,D4 MET00460 DO 12 I=1, NL MET00470 READ(5,*) T(I) MET00480 12 CONTINUE MET00490 WRITE(6,80) AS10 MET00500 80 FORMAT(' REACTOR INLET CONC=',F6.2,' PPM') MET00510 WRITE(6,82) MET00520 - 273 -

82 FORMAT(' INTRINSIC RATE PARAMETERS, ML/GCATHR') MET00530 WRITE(6,84) K1,K2,K3,K4,K5,K6,K7 MET00540 84 FORMAT(' K1=',F6.2,/,' K2=',F6.2,/,' K3=',F6.2,/ MET00550 &,' K4=',,F6.2,/,' K5=',F6.2,/,' K6=',F6.2,/,' K7=',F6.2) MET00560 WRITE(6,86) WCATQTIME MET00570 86 FORMAT(' AMOUNT OF CATALYST=',F5.2,' GMS',/,' OIL FLOW MET00580 &RATE=',F6.2,' ML/HR',/,' TIME ON STREAM=',F6.1,' HRS') MET00590 WRITE(6,88) D1,D2,D3,D4 MET00600 88 FORMAT(' D1=',E1O.4,/,' D2=',E10.4,/,' D3=',E10.4,/ MET00610 &,' D4=',E1O.4) MET00620 WRITE (6,90) R,RHOCRHO MET00630 90 FORMAT(' PELLET RADIUS=',F7.3,' CM',/,' CATALYST DENSITY=', MET00640 &F6.2,' G/CC',/,' OIL DENSITY=',F6.2,' G/CC') MET00650 10PT=1 MET00660 ASO(1)=AS10 MET00670 ASO(2)=AS20 MET00680 ASO(3)=AS30 MET00690 ASO(4)=AS40 MET00700 D1=D1*3600.0/RHOC MET00710 D2=D2*3600.0/RHOC MET00720 D3=D3*3600.0/RHOC MET00730 D4=D4*3600.0/RiHOC MET00740 C SET UP D MATRIX MET00750 D(1,1)=Dl1 MET00760 D(1,2)=0.0 MET00770 D(1,3)=0.0 MET00780 D(1,4)=O.0 MET00790 D(2,1)=0.0 MET00800 D(2,2)=D2 MET00810 D(2,3)=0.0 MET00820 D(2,4)=0.0 MET00830 D(3,1)=0.0 MET00840 D(3,2)=0.0 MET00850 D(3,3)=D3 MET00860 D(3,4)=0.0 MET00870 D(4,1)=0.0 MET00880 D(4,2)=0.0 MET00890 D(4,3)=0.0 METOO900 D(4,4)=D4 MET00910 C SET UP K MATRIX MET00920 K(1,1)=K1 MET00930 K(1,2)=-K2 MET00940 K(1,3)=0.0 MET00950 K(1,4)=0.0 MET00960 K(2,1)=-Kl MET00970 K(2,2)=K2+K3 METOO980 K(2,3)=-K4 MET00990 K(2,4)=0.0 MET01000 K(3,1)=0.0 MET01010 K(3,2)=-K3 MET01020 K(3 ,3)=K4+K5+K7 MET01030 K(3,4)=0.0 MET01040 K(4,1)=0.0 MET01050 K(4,2)=0.0 MET01060 - 274 -

K(4,3 )=-K5 MET01070 K(4,4)=K6 MET01080 C INVERT D MATRIX MET01090 N=4 MET01100 IA=4 MET01110 IDGT=3 MET01120 CALL LINV2F(DN,IA,DINVIDGT,WKAREA,IER) MET01130 C MULTIPLY DINV AND K MET01140 L=4 MET01150 M=4 MET01160 IB=4 MET01170 IC=4 MET01180 CALL VMULFF(DINV,K,L,M,NIA,IB,DK,IC,IERI) MET01190 C SOLVE FOR EIGENVALUES AND EIGENVECTORS OF DK MET01200 IJOB=2 MET01210 IY=4 MET01220 CALL EIGRF(DK,N,IA,IJOB,PHIC,Y,IY,WK1,IER2) MET01230 WRITE(6,100) (PHIC(J),J=1,4),WKl(1) MET01240 WRITE(6,105) MET01250 DO 10 I=1,4 MET01260 WRITE(6,110) (Y(IJ),J-1,4) MET01270 10 CONTINUE MET01280 DO 15 =1,4= MET01290 DO 15 J=1,4 MET01300 YR(I,J)=REAL(Y(I,J)) MET01310 15 CONTINUE MET01320 C INVERT EIGENVECTORS OF DK MET01330 CALL LINV2F(YR,N,IAYINV,IDGT,WKAREA,IER) MET01340 WRITE(6, 115) MET01350 DO 20 1=1,4 MET01360 WRITE(6,120) (YINV(I ,J),J=1,4) MET01370 20 CONTINUE MET01380 C EVALUATE PHI,THETA,ETA FOR EACH CHARACTERISTIC SPECIES MET01390 DO 30 1=1,4 MET01400 PHI(I)=R*SQRT(REAL(PHIC(I))) MET01410 ARG=PHI(I) MET01420 AAIO(I)=MMBSIO(IOPT ,ARG, IER) MET01430 AI1=MMBSI1(IOPT,ARG,IER) MET01440 THETA(I , I)=2.0*ARG*AIl/AAIO(I) MET01450 ETA(I)=THETA(I,I)/ARG**2 MET01460 WRITE(6,125) I,PHI(I),I,THETA(I,I),I,ETA(I) MET01470 30 CONTINUE MET01480 C COMPLETE ELEMENTS IN THETA MATRIX MET01490 DO 31 J=2,4 MET01500 THETA(1,J)=0.0 MET01510 31 CONTINUE MET01520 THETA(2,1)=0.0 MET01530 THETA(2,3)=0.0 MET01540 THETA(2,4)=O.0 - MET01550 THETA(3,1)=0.0 MET01560 THETA(3,2)=0.0 MET01570 THETA(3,4)=0.0 - MET01580 DO 32 J=1,3 MET01590 THETA(4,J)=0.0 MET01600 W- 6 -

0171Z0O13W ocsv=( Ofsv 0cz0law ozsv=U-)Zsvt 0z1zo~lsw oisv=( Oisv 011z0Iaw SNOIIISOa 1lODV9-a SflXVA IV 19TIga NI MI'itoHa ONOD UI~flD~IvD D O01zol~w SflNINOD 09 060ZOlaR (Ii v'(x)CSV'(I)Zsv'(I)ISV'(1XL (9'19)2ix1Ia

ot~ozolaw V )-8

ovozoaaw (~ 076Z 8

ooozo~l~w (x)i w ( t,)m)qv-) axg*( V~)oaq=( 0)Va 0661019aw

0961019N rmIN'Z=I 09 00 096101HW ovsV' 0csvg ozsv' olsv c(1)l (S91 49) gldxm 0176101UN (091g'9)liI1m 0C61019 ss~~gWa

016101UN (-gx'Dlgoaq'ql'Vx ENNgN'r1g0SVgANIcJX)aA1lNwA 71V), 00610131N I=NN 06810l31N (IwINlILV SSIDUJS OIISIM1IVWHO) -9019A 0J9 uivflTr'A9 D 098101aw glNILN0) S9 OLSI019N ('V'I-r'(r'I)ANITx) (ozlg9)Jlixm 0991013w V6'1=1 gg oc 09910I1R4vg~lu 0'791019N (-g'SX'0lA~~~ggET)ZNr TIVD 0c91OI3w V) AO SH)101AN9I0 L11MANI D 0ze1013w 9filNND 09 0191013w (r'gx) ax) ri=sm( f'x) Hax 0091013W vz'I=f 0g 00 06 1013W4 '7'1=1 0S 00

09L101314 '17 9 00 '7

01IOR 1134 fNIIN0C) Oti 00 13 (VT'=c'(r'i)ay) (oz1'9)3IlhIm 0691013W v7'I=I 07 00 0991013W (ou '9) slx~ 0L91012N (i'DI'a~gc'wt(Ilg~Ngr1gANIx(I1Ix)wIfNwA TIVO 0991013w T rrVD 099101ZW (xrM'x'Iv N'' 11~)~~lATVO 0i791019N 3flNILNOD SC 0C91019N Z**Y)/ (I' )V13111-(I ')VI9H1 0Z91013w 47'91=1 9c 00 0191013W 3fltIN0O ZE

- 9LZ- - 276 -

AS4(1)=AS40 MET02150 DO 65 I=1,NL MET02160 WRITE(6,170) T(I) MET02170 WRITE(6,175) MET02180 AS(1)=AS1(I) MET02190 AS(2)=AS2(I) MET02200 AS(3)=AS3(I) MET02210 AS(4)=AS4(I) MET02220 CALL VMULFF(YINVAS,L,M,NN,IA,IB,BSIC,IER) MET02230 DR=1.1 MET02240 DO 70 J=1,11 MET02250 DR=DR-0 .1 METO2260 ARG=PHI (1)*DR METO2270 AIO=MMBSIO(IOPT ,ARG, IER) MET02280 B(1)=BS(1)*AIO/AAIO(1) MET02290 ARG =PHI(2)*DR METO2300 AIO=MMBSIO(IOPT,ARG,IER) MET02310 B(2)=BS(2)*AIO/AAIO(2) MET02320 ARG=PHI(3)*DR MET02330 AIO=MMBSIO(IOPT,ARG, IER) MET02340 B(3)=BS(3)*AIO/AAIO(3) MET02350 ARG=PHI (4) *DR METO2360 AIO=MMBSIO(IOPT,ARG,IER) MET02370 B(4)=BS(4)*AIO/AAIO(4) MET02380 CALL VMULFF(YRB,L,M,NN,IA,IB,A,ICIER) MET02390 C CALCULATE WEIGHT PERCENT METAL ON CATALYST AFTER TIME HOURS METO2400 C ON STREAM AT EACH RADIAL POSITION MET02410 WM=(K6*A(4)+K7*A(3))*RHO*TIME*1.OE-4 MET02420 WRITE(6,180) DR,A(1),A(2),A(3),A(4),WM MET02430 70 CONTINUE MET02440 65 CONTINUE MET02450 100 FORMAT(' PHICI=',2E12.4,/,' PHIC2=',2E12.4,/,' PHIC3=',2E12.4,/, MET02460 &' PHIC4=',2E12.4,/,' PFIX=',F6.2) MET02470 105 FORMAT(' MATRIX OF EIGENVECTORS FROM DK MATRIX') MET02480 110 FORMAT(lX,8E12.4) MET02490 115 FORMAT(' INVERSE MATRIX OF EIGENVECTORS FROM DK') MET02500 120 FORMAT(lX,4E12.4) MET02510 125 FORMAT(' PHI',11,'=',E12.4,' THETA',Il,'=',E12.4,' ETA',Il1,'=', MET02520 &E12.4) MET02530 130 FORMAT(' MATRIX OF DIFFUSION DISGUISED RATE PARAMETERS') MET02540 135 FORMAT(' LPI=',2E12.4,/,' LP2=',2E12.4,/,' LP3=',2E12.4,/, METO2550 &' LP4=',2E12.4,/,' PFIX=',F6.2) MET02560 140 FORMAT(' MATRIX OF EIGENVECTORS FROM KP MATRIX') MET02570 145 FORMAT(' INVERSE MATRIX OF EIGENVECTORS FROM KP') MET02580 150 FORMAT(' BP1O=',E12.4,/,' BP20=',E12.4,/,' BP30=',E12.4,/, MET02590 &' BP40=',E12.4) METO2600 155 FORMAT(' SPACETIME' ,3X,'PORPHYRIN',3X,'CHLORIN' ,3X, METO2610 &'ISOBACT.CHLORIN',3X,'UNKNOWN') MET02620 160 FORMAT(' GCATHR/ML',6X,'PPM',8X,'PPM',10X,'PPM',12X,'PPM') MET02630 165 FORMAT(3XF5.3,8X,F5.2,6X,F5.2,8X,F5.2,11X,F5.2) MET02640 170 FORMAT(' SPACETIME=',F5.3) MET02650 175 FORMAT(' RADIAL POSITION' ,3X, 'PORPHYRIN' ,3X, 'CHLORIN' ,3X, MET02660 &'ISOBACT.CHLORIN' ,3X,'UNKNOWN ',3X,'WT% METAL') MET02670 180 FORMAT(5X,F4.2,8X,F5.2,6X,F5.2,8XF5.2,11X,F5.2,4X,E8.3) [ETD2630 STOP MET02690 END METO2700 - 277 -

APPENDIX D

EXPERIMENTAL DATA FROM INTRINSIC KINETIC RUNS

In this Appendix the kinetic data for the nickel porphyrin runs are tabulated. Included for each data set are the kinetic rate parameters which have been obtained to model the data set.

The data are coded with run numbers, i.e., MANTM21, and a description of the pre-treatment of the catalyst. The letter M of the run number stands for the microreactor (0.52 cm I.D.); A for

American Cyanamid HDS-16A catalyst; NTM for Ni-tetra(3-methylphenyl) porphyrin; and the number corresponds to the catalyst batch.

The units of temperature, hydrogen pressure, nickel species, and contact time (W/Q) in the tables are: 'C, psig, ppm Ni by weight, and g cat. hr/ml oil, respectively. The total metal (M) was determined by atomic absorption. The Ni-P, Ni-PH and Ni-PH were determined by UV-visible spectrophotometry. The Ni-X was determined by the difference of total metal and porphyrinic metal (i.e., Ni-P +

oil/g cat hr. Ni-PH2 + Ni-PH4 ). The units of the rate coefficients are ml - 278 -

MANTM6C OXIDE CATALYST

TEMPERATURE = 345 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 26.0 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 26.0 25.0 1.0 0.0 0.0

0.013 24.9 12.6 6.6 3.0 2.7

0.022 23.9 9.3 5.8 2.9 5.9

0.046 20.0 5.1 3.4 2.0 9.5

0.072 17.0 2.7 2.0 1.0 11.3

0.163 10.2 1.0 0.6 0.3 8.3

Ki = 85.00 K2 = 90.00 K3 =130.00 K4 =150.00 K5 = 66.00 K6 = 4.60 K7 = 25.00

MANTM5A OXIDE CATALYST

TEMPERATURE = 320 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 26.7 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 26.7 25.0 1.7 0.0 0.0

0.024 24.8 12.8 6.6 2.9 2.5

0.052 23.7 7.7 5.9 3.2 6.9

0.072 22.5 7.4 5.8 3.1 6.2

0.104 20.7 3.6 3.1 1.8 12.2

0.145 18.7 2.5 2.0 1.3 12.9

0.181 15.8 1.7 1.2 0.8 12.1

0.242 16.9 1.2 0.4 0.4 14.9

0.362 6.6 0.6 0.3 0.3 5.4

Ki = 46.00 K2 - 39.00 K3 = 60.00 K4 = 58.00 K5 = 34.00 K6 = 2.60 K7 = 12.00 - 279 -

MANTM5C OXIDE CATALYST

TEMPERATURE = 300 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 26.7 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 26.7 25.0 1.7 0.0 0.0

0.024 26.7 15.7 7.5 3.5 0.0

0.052 25.6 11.3 7.4 4.0 2.9

0.072 20.6 8.7 6.7 3.9 1.3

0.104 23.8 6.4 5.4 3.5 8.5

0.145 21.8 4.2 4.0 2.7 10.9

0.181 20.2 3.3 2.9 2.1 11.9

0.242 17.6 1.9 1.9 1.4 12.4

KI = 24.60 K2 = 18.60 K3 = 40.00 K4 = 30.00 K5 = 19.00 K6 = 1.48 K7 = 7.40

MANTM6A OXIDE CATALYST

TEMPERATURE = 285 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 26.9 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 26.9 25.2 1.7 0.0 0.0

0.016 26.9 22.2 3.7 1.0 0.0

0.046 26.9 18.1 6.3 2.4 0.1

0.108 26.6 13.5 7.6 4.1 1.4

0.325 19.1 5.4 4.5 3.2 6.0

0.650 12.4 1.9 1.7 1.5 7.3

Ki = 18.00 K2 = 15.00 K3 = 24.00 K4 = 19.00 K5 = 9.00 -.K6 = 0.80 K7 = 2.80 - 280 -

MANTM38 OXIDE CATALYST

TEMPERATURE = 345 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 16.4 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 16.4 16.4 0.0 0.0 0.0

0.007 16.0 10.0 4.5 1.5 0.0

0.014 15.5 8.7 5.0 1.8 0.0

0.035 12.6 6.0 4.3 1.8 0.5

0.070 11.6 3.3 2.3 1.1 4.9

0.140 10.0 1.4 1.3 0.8 6.5

0.180 9.3 1.0 1.0 0.5 6.8

0.234 8.6 0.8 0.9 0.5 6.5

Ki = 86.00 K2 =100.00 K3 =120.00 K4 =145.00 K5 = 55.00 K6 = 4.20 K7 = 24.00

MANTM12 OXIDE CATALYST

TEMPERATURE = 345 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 90.0 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 90.0 90.0 0.0 0.0 0.0

0.019 82.5 22.8 24.6 14.0 21.1

0.043 77.5 15.7 16.3 9.9 35.6

0.077 65.0 9.5 7.4 4.9 43.2

0.128 54.3 6.4 3.3 2.6 42.0

0.192 38.7 3.7 1.5 1.2 32.3

0.256 30.5 1.2 1.2 1.1 27.0

0.383 18.2 0.8 0.8 0.7 15.9

KI = 94.00 K2 = 84.00 K3 =130.00 K4 =155.00 K5 = 60.00 K6 = 4.25 K7 = 19.50 - 281 -

MANTM33A OXIDE CATALYST

TEMPERATURE = 345 C PRESSURE = 650 PSIG INLET CONCENTRATION = 63.0 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 63.0 63.0 0.0 0.0 0.0

0.014 61.6 28.7 22.9 10.0 0.0

0.035 61.1 23.8 20.4 9.6 7.3

0.070 60.4 13.2 16.2 8.1 22.9

0.140 53.9 9.4 10.9 6.1 27.5

0.233 47.1 6.4 7.5 4.6 28.6

0.350 42.9 4.3 5.5 3.6 29.5

0.538 38.9 2.7 4.2 2.9 29.1

Kl = 84.00 K2 = 85.00 K3 = 73.00 K4 =140.00 K5 = 21.80 K6 = 1.20 K7 = 6.89

MANTM33B OXIDE CATALYST

TEMPERATURE = 345 C PRESSURE = 1450 PSIG INLET CONCENTRATION = 57.0 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 57.0 57.0 0.0 0.0 0.0

0.007 55.7 25.4 19.9 10.4 0.0

0.014 53.6 17.2 15.9 9.2 11.3

0.035 49.9 6.9 6.6 4.3 32.1

0.070 43.8 2.5 2.0 1.6 37.7

0.140 31.7 1.2 1.2 1.0 28.3

0.240 18.0 0.8 0.8 0.8 15.7

0.350 5.4 0.4 0.4 0.3 4.3

Ki =175.00 K2 =115.00 K3 =170.00 K4 =180.00 K5 =142.00 K6 = 5.10 K7 = 28.00 - 282 -

MANTM21 OXIDE CATALYST

TEMPERATURE = 345 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 63.0 PPM NI

W/o M NI-P NI-PH2 NI-PH4 NI-X

0.000 63.0 63.0 0.0 0.0 0.0

0.014 62.2 20.2 22.2 12.9 6.9

0.035 55.6 14.0 16.5 10.1 15.0

0.070 46.8 9.7 10.1 6.5 20.5

0.140 38.0 5.3 4.7 3.2 24.8

0.233 29.5 2.4 2.3 1.8 23.0

0.349 19.9 1.3 1.3 1.1 16.2

0.538 11.0 0.8 0.7 0.6 8.9

Ki =105.00 K2 = 94.00 K3 =120.00 K4 =150.00 K5 = 49.00 K6 = 3.80 K7 = 18.00

MANTM21A OXIDE CATALYST

TEMPERATURE = 320 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 66.0 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 66.0 66.0 0.0 0.0 0.0

0.014 66.5 33.6 21.9 11.0 0.0

0.035 63.5 22.6 20.4 12.0 8.5

0.070 61.6 14.7 15.7 10.5 20.7

0.140 53.8 7.6 8.8 6.5 30.9

0.233 43.9 4.3 4.1 3.3 32.2

0.349 36.4 1.9 1.7 1.5 31.3

0.538 30.1 1.2 1.0 0.9 27.0

Ki = 55.00 K2 = 43.00 K3 = 62.00 K4 = 55.00 K5 = 26.00 K6 = 1.90 K7 = 7.50 - 283 -

MANTM21B OXIDE CATALYST

TEMPERATURE = 285 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 66.0 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 66.0 66.0 0.0 0.0 0.0

0.014 66.0 42.5 17.5 6.0 0.0

0.035 65.7 31.5 21.0 13.2 0.0

0.070 65.6 23.4 19.3 14.8 8.1

0.140 62.6 15.2 15.3 14.1 18.0

0.233 59.0 9.4 10.7 11.3 27.6

0.349 52.2 4.9 6.1 7.0 34.2

0.538 47.4 2.1 1.9 2.4 41.0

0.698 39.6 1.3 1.6 1.9 34.8

Ki = 24.00 K2 = 17.00 K3 = 28.00 K4 = 21.00 K5 = 9.00 K6 = 0.62 K7 = 2.30

MANTM30 SULFIDED CATALYST

TEMPERATURE = 345 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 60.0 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 60.0 60.0 0.0 0.0 0.0

0.005 51.9 28.8 16.1 7.0 0.0

0.010 45.1 17.3 12.9 6.2 8.7

0.025 39.5 9.1 7.2 4.0 19.2

0.050 24.6 3.9 2.5 1.7 16.5

0.067 15.1 2.2 1.2 0.9 10.8

0.100 8.5 1.2 0.5 0.4 6.3

Ki =183.00 K2 =164.00 K3 =250.00 K4 =400.00 K5 =171.00 K6 = 27.00 K7 =120.00 - 284 -

MANTM31A SULFIDED CATALYST

TEMPERATURE = 320 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 60.0 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 60.0 60.0 0.0 0.0 0.0

0.010 52.5 32.6 14.2 5.7 0.0

0.025 44.9 19.3 12.0 5.7 7.9

0.050 39.0 10.4 7.9 4.2 16.5

0.100 34.2 5.9 4.7 2.8 20.8

0.200 24.0 2.7 2.3 1.5 17.5

0.333 11.2 0.8 0.7 0.6 9.1

Ki = 82.40 K2 = 92.50 K3 =130.00 K4 =190.00 K5 = 46.00 K6 = 5.50 K7 = 60.00

MANTM31B SUL FIDED CATALYST

TEMPERATURE = 285 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 63.0 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 63.0 63.0 0.0 0.0 0.0

0.010 63.0 50.0 10.0 3.0 0.0

0.025 61.5 43.0 13.4 5.1 0.0

0.050 54.4 33.4 14.8 6.2 0.0

0.100 40.1 20.5 11.8 5.7 2.1

0.200 29.8 13.0 8.0 4.0 4.8

0.333 15.0 5.6 3.8 2.0 3.6

Ki = 22.00- K2 = 24.30 K3 = 45.50 K4 = 58.70 K5 = 4.80 K6 - 2.00 K7 = 26.00 - 285 -

MANTM32 SULFIDED CATALYST

TEMPERATURE = 345 C PRESSURE = 600 PSIG INLET CONCENTRATION = 63.0 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 63.0 63.0 0.0 0.0 0.0

0.010 59.4 32.2 18.9 7.1 1.2

0.025 57.6 25.2 16.7 6.9 8.8

0.050 54.5 21.1 13.9 6.3 13.2

0.100 46.3 15.0 9.6 5.0 16.7

0.200 36.8 4.7 5.8 3.6 22.7

0.333 22.6 1.9 2.9 2.0 15.8

KI = 96.40 K2 =122.00 K3 =145.00 K4 =400.00 K5 = 50.00 K6 = 4.90 K7 = 25.00

MANTM32A SULFIDED CATALYST

TEMPERATURE = 320 C PRESSURE = 600 PSIG

INLET CONCENTRATION - 63.0 PPM. NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 63.0 63.0 0.0 0.0 0.0

0.025 58.7 32.6 18.3 7.2 0.6

0.050 57.5 27.1 18.2 7.6 4.6

0.100 55.6 22.4 16.1 7.0 10.1

0.200 52.7 17.8 12.9 6.0 16.0

0.333 49.0 16.3 11.3 5.3 16.1

0.500 31.8 8.8 6.0 3.1 13.9

Ki = 42.00- K2 = 53.00 K3 = 72.00 K4 =195.00 K5 = 12.50 K6 = 1.62 K7 = 10.00 - 286 -

MANTM32B SULFIDED CATALYST

TEMPERATURE = 285 C *PRESSURE = 600 PSIG INLET CONCENTRATION = 65.0 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 65.0 65.0 0.0 0.0 0.0

0.025 65.0 51.2 10.2 3.6 0.0

0.050 62.6 44.2 13.2 5.2 0.0

0.100 57.4 35.9 15.1 6.4 0.0

0.200 52.4 27.1 15.7 7.1 2.5

0.333 48.5 21.5 14.7 6.8 5.5

0.500 34.3 14.9 10.6 4.6 4.2

Kl = 12.00 K2 = 15.80 K3 = 34.90 K4 = 65.20 K5 = 2.20 K6 = 0.72 K7 = 4.70

MANTM32F SULFIDED CATALYST

TEMPERATURE = 345 C PRESSURE = 1500 PSIG INLET CONCENTRATION = 60.0 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 60.0 60.0 0.0 0.0 0.0

0.005 37.0 20.2 9.9 4.3 2.6

0.010 30.0 10.6 6.3 3.0 10.1

0.025 15.0 2.9 1.8 1.0 9.3

0.050 6.9 0.5 0.3 0.3 5.8

Ki =240.00 K2 =15C'.00 K3 =400.00 K4 =450.00 K5 =246.00 K6 = 44 .00 K7 =560.00 - 287 -

MANTM32E SULFIDED CATALYST

TEMPERATURE = 320 C PRESSURE = 1500 PSIG INLET CONCENTRATION = 60.0 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 60.0 60.0 0.0 0.0 0.0

0.005 50.8 33.3 12.6 4.9 0.0

0.010 42.3 23.9 11.3 4.7 2.4

0.025 22.4 10.8 6.3 2.7 2.6

0.050 9.2 3.1 1.9 0.9 3.4

0.067 5.3 1.6 1.0 0.5 2.2

KI =118.00 K2 =100.00 K3 =208.00 K4 =200.00 K5 = 50.00 K6 = 10.50 K7 =380.00

MANTM32D SULFIDED CATALYST

TEMPERATURE = 285 C PRESSURE = 1500 PSIG INLET CONCENTRATION = 60.0 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 60.0 60.0 0.0 0.0 0.0

0.010 57.3 41.4 11.6 4.3 0.0

0.025 49.9 31.4 12.8 5.3 0.4

0.050 36.7 20.5 10.8 4.9 0.5

0.100 18.8 9.0 5.6 2.7 1.5

0.200 1.6 0.7 0.4 0.2 0.3

KI = 35.00 K2 = 24.0 0 K3 = 72.80 K4 = 60.00 K5 = 7.00 K6 = 3.50 K7 = 97.50 - 288 -

MANTM27 100 PPM N (PYRIDINE)

TEMPERATURE = 345 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 71.0 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 71.0 71.0 0.0 0.0 0.0

0.014 71.0 44.4 23.3 3.3 0.0

0.070 62.5 31.3 22.0 8.7 0.5

0.233 44.1 14.7 13.0 6.1 10.3

0.350 39.8 11.3 10.6 5.5 12.4

0.538 25.0 5.6 6.3 3.6 9.5

Ki = 43.80 K2 = 53.20 K3 = 42.00 K4 = 78.30 K5 = 7.20 K6 = 2.30 K7 = 14.00

MANTM25 REDUCED CATALYST

TEMPERATURE = 345 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 60.0 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 60.0 60.0 0.0 0.0 0.0

0.010 55.0 14.2 17.6 10.7 12.5

0.025 47.4 8.6 9.5 6.2 23.1

0.050 38.3 4.4 4.3 3.0 26.6

0.100 26.7 1.6 1.5 1.2 22.4

0.167 14.9 0.6 0.5 0.5 13.3

Kl =264.00 K2 =197.00 K3 =225.00 K4 =202.00 K5 =123.00 K6 = 8.48 K7 = 40.40 - 289 -

MANTM40 CS DOPED CATALYST

TEMPERATURE = 345 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 63.0 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 63.0 63.0 0.0 0.0 0.0

0.035 62.9 32.8 22.1 8.0 0.0

0.070 62.9 33.0 22.0 7.9 0.0

0.140 58.4 30.3 20.6 7.5 0.0

0.233 57.5 29.7 20.3 7.5 0.0

0.350 55.3 28.6 19.3 7.4 0.0

0.538 52.8 28.7 17.6 6.5 0.0

Ki = 89.00 K2 =126.00 K3 =130.00 K4 =250.00 K5 = 0.00 K6 = 0.00 K7 = 2.70

MANTM37 NA DOPED CATALYST

TEMPERATURE = 345 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 56.0 PPM NI

W/Q M NI-P NI-Pd2 NI-PH4 NI-X

0.000 56.0 56.0 0.0 0.0 0.0

0.035 55.2 27.1 20.1 8.0 0.0

0.070 54.7 26.2 20.1 8.4 0.0

0.140 50.9 23.7 19.0 8.2 0.0

0.234 46.2 20.5 16.4 7.2 2.1

0.350 40.7 17.7 13.9 6.2 2.9

0.540 37.7 15.8 11.8 5.3 4.8

Ki =102.00 K2 =134.00 K3 =131.00 K4 =300.00 K5 = 3.30 K6 = 2.20 K7 = 5.00 - 290 -

MANTM39 IODIZED CATALYST

TEMPERATURE = 345 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 63.6 PPM NI

W/Q M NI-P NI-PH2 NI-PH4 NI-X

0.000 63.6 63.6 0.0 0.0 0.0

0.035 60.2 47.0 10.4 2.8 0.0

0.070 49.0 36.4 9.5 3.1 0.0

0.140 34.7 21.8 6.3 2.7 3.9

0.230 27.8 16.1 4.4 2.1 5.2

Ki = 12.00 K2 = 17.00 K3 = 45.00 K4 = 82.00 K5 = 20.00 K6 = 13.00 K7 = 66.00

MANTM34 CHLO RIDED CATALYST

TEMPERATURE = 345 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 61.0 PPM NI

W/Q M NI-P NI-PH2

0.000 61.0 61.0 0.0

0.035 48.7 46.2 2.5

0.070 37.3 34.9 2.4

0.140 22.2 20.4 1.9

0.234 12.9 11.5 1.4

0.350 6.0 5.2 0.8

Ki = 7.80 K2 = 7.00 K3 =114.00 - 291 -

MANTM34A CHLORIDED CATALYST

TEMPERATURE = 320 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 61.0 PPM NI

W/Q M NI-P NI-PH2

0.000 61.0 61.0 0.0

0.035 56.1 52.4 3.7

0.070 46.7 43.4 3.3

0.140 32.2 30.0 2.2

0.234 21.2 19.7 1.5

0.350 13.6 12.6 1.0

0.468 8.7 8.0 0.7

K1= 5.00 K2= Li.10 K3 = 66.30

MANTM34B CHLORIDED CATALYST

TEMPERATURE = 285 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 61.0 PPM NI

W/Q M NI-P NI-PH2

0.000 61.0 61.0 0.0

0.070 54.2 51.1 3.1

0.140 45.7 43.0 2.7

0.234 35.1 33.1 2.0

0.350 28.7 27.1 1.6

0.468 22.0 20.7 1.3

0.585 16.7 15.6 1.1

Ki = 2.60 K2 = 2.10 K3 = 30.20 - 292 -

NI-ETIO OXIDE CATALYST

TEMPERATURE = 343 C PRESSURE = 1000 PSIG INLET CONCENTRATION = 27.8 PPM NI

W/Q M NI-P NI-PH2

0.000 27.8 27.8 0.0

0.015 22.7 19.9 2.8

0.024 20.4 17.9 2.5

0.036 17.1 15.0 2.1

0.057 15.4 13.6 1.8

0.095 10.2 9.0 1.2

0.114 6.6 5.8 0.8

0.190 3.2 2.8 0.4

KI = 25.00 K2 =100.00 K3 = 80. 00