USF Libraries
USF Digital Collections

Effect of catalyst preparation conditions on the performance of eggshell cobalt/sio2 catalysts for fischer-tropsch synthesis

MISSING IMAGE

Material Information

Title:
Effect of catalyst preparation conditions on the performance of eggshell cobalt/sio2 catalysts for fischer-tropsch synthesis
Physical Description:
Book
Language:
English
Creator:
Gardezi, Syed-Ali
Publisher:
University of South Florida
Place of Publication:
Tampa, Fla
Publication Date:

Subjects

Subjects / Keywords:
Silica
Calcination
Fixed bed reactor
Biomass
Diesel
Aviation fuel
Conversion
Selectivity
Dissertations, Academic -- Chemical & Biomedical Engineering -- Masters -- USF   ( lcsh )
Genre:
non-fiction   ( marcgt )

Notes

Abstract:
ABSTRACT: A highly selective eggshell Fischer-Tropsch catalyst has been fabricated via interaction of hydrophobic and hydrophilic molecules on thermally treated silica gel. The physical interactions of the mesoporous silica support and the effect of catalyst preparation conditions on the performance of the cobalt/SiO2 were explored. It was found that dispersion and performance of the FT cobalt/SiO2 catalyst were significantly affected by the preparation technique used. In this study we focus on two key variables: the solvent used during the precursor loading and the calcination atmosphere. Silanol groups on the silica surface and near-surface regions can alter morphology and dispersion of the supported active metals. Solvents used for precursor such as water or alcohol attach to these silanol sites in specific configurations and compete with metal salts during ion exchange and adsorption. By fine tuning the solvent attachments on heat treated silica we have fabricated a cobalt/silica catalyst with high dispersion and low metal loads. Additionally, since silica has affinity for both polar and non-polar molecule depending on the surface conditions; this property has been exploited in preparing an engineered eggshell profile. This together with simultaneous calcination/ reduction in a dynamic hydrogen environment has been shown to further enhance dispersion and reducibility. Characterization techniques including BET, XPS, XRD, H-chemisorption and FTIR were employed. Catalyst activity, product selectivity, distribution and conversion were studied using a bench scale fixed bed reactor fitted with a GC/MS instrument.
Thesis:
Thesis (MSCH)--University of South Florida, 2010.
Bibliography:
Includes bibliographical references.
System Details:
Mode of access: World Wide Web.
System Details:
System requirements: World Wide Web browser and PDF reader.
Statement of Responsibility:
by Syed-Ali Gardezi.
General Note:
Title from PDF of title page.
General Note:
Document formatted into pages; contains X pages.

Record Information

Source Institution:
University of South Florida Library
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
usfldc doi - E14-SFE0004599
usfldc handle - e14.4599
System ID:
SFS0027914:00001


This item is only available as the following downloads:


Full Text
xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2010 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0004599
035
(OCoLC)
040
FHM
c FHM
049
FHMM
090
XX9999 (Online)
1 100
Gardezi, Syed-Ali.
0 245
Effect of catalyst preparation conditions on the performance of eggshell cobalt/sio2 catalysts for fischer-tropsch synthesis
h [electronic resource] /
by Syed-Ali Gardezi.
260
[Tampa, Fla] :
b University of South Florida,
2010.
500
Title from PDF of title page.
Document formatted into pages; contains X pages.
502
Thesis (MSCH)--University of South Florida, 2010.
504
Includes bibliographical references.
516
Text (Electronic thesis) in PDF format.
538
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
3 520
ABSTRACT: A highly selective eggshell Fischer-Tropsch catalyst has been fabricated via interaction of hydrophobic and hydrophilic molecules on thermally treated silica gel. The physical interactions of the mesoporous silica support and the effect of catalyst preparation conditions on the performance of the cobalt/SiO2 were explored. It was found that dispersion and performance of the FT cobalt/SiO2 catalyst were significantly affected by the preparation technique used. In this study we focus on two key variables: the solvent used during the precursor loading and the calcination atmosphere. Silanol groups on the silica surface and near-surface regions can alter morphology and dispersion of the supported active metals. Solvents used for precursor such as water or alcohol attach to these silanol sites in specific configurations and compete with metal salts during ion exchange and adsorption. By fine tuning the solvent attachments on heat treated silica we have fabricated a cobalt/silica catalyst with high dispersion and low metal loads. Additionally, since silica has affinity for both polar and non-polar molecule depending on the surface conditions; this property has been exploited in preparing an engineered eggshell profile. This together with simultaneous calcination/ reduction in a dynamic hydrogen environment has been shown to further enhance dispersion and reducibility. Characterization techniques including BET, XPS, XRD, H-chemisorption and FTIR were employed. Catalyst activity, product selectivity, distribution and conversion were studied using a bench scale fixed bed reactor fitted with a GC/MS instrument.
590
Advisor: John T Wolan, Ph.D.
653
Silica
Calcination
Fixed bed reactor
Biomass
Diesel
Aviation fuel
Conversion
Selectivity
690
Dissertations, Academic
z USF
x Chemical & Biomedical Engineering
Masters.
773
t USF Electronic Theses and Dissertations.
4 856
u http://digital.lib.usf.edu/?e14.4599



PAGE 1

Effect of Catalyst Preparation Conditions on the Performance of Eggshell Cobalt/SiO2 Catalysts for Fische r-Tropsch Synthesis by Syed Ali Z. H. Gardezi A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Chemical Engineering Department of Chemical a nd Biomedical Engineering College of Engineering University of South Florida Co-Major Professor: John T. Wolan, Ph.D. Co-Major Professor: Babu Joseph, Ph.D. Vinay K. Gupta, Ph.D. Date of Approval: June 28, 2010 Keywords: silica, calc ination, fixed bed reactor, bi omass, diesel, aviation fuel, conversion, selectivity Copyright 2010, Syed Ali Z. H. Gardezi

PAGE 2

Dedication First and foremost, I want to dedicate my res earch to the energy deprived nations of this world, I sincerely hope that it will contribut e to the expansion of the alternate energy sector, and reduce the dependence on deple ting fossil resources, resulting in the preservation of mother ear th and its inhabitants. I would also like to dedicate this research to the friendly and loving people of the United States who have made me feel at home. I am optimistic that my research will play a definite role in achieving the goal of energy independenc e and will help create more domestic jobs.

PAGE 3

Acknowledgements Above all, I am heartily thankful for the patience, guidance and helpful ideas of my principal adviser, Dr. John T. Wolan, not to mention his support in allaying of my fears that arose during the course of my gradua te school experience. The good advice, and friendship of my second advisor, Dr Babu Joseph, has been invaluable on both an academic and a personal level, for which I am extremely grateful. I would also like to thank my committee member, Dr. Vinay K. Gupta for his “eye opening” suggestions. The expertise of Dr. Sesha Srinivasan from the Clean Energy Rese arch Center, with regard to hydrogen chemisorption experiment s came in very handy. I am thankful to Haitao “Eddie” Li of the USF Green Energy Systems Lab for his help with the BET measurements in this work, and Bijith Ma nkidy for his help with procuring the TPR results. I would also like to acknowledge my lab part ners, current and former: Ala'a Kababji, Chris Monteparo, Brad Ridder, William Bo sshart, Sandra Pettit, and Lucky Roy Landrgan. Their words of encouragement, inte lligent conversations, and helpful advice are a memorable part of my graduate school experience. Last, but by no means least, I want to th ank my wife Arusa Aleen Gardezi for her constant support. Like a guardian angel, sh e stands by my side and appeases me whenever I am in dire strait.

PAGE 4

i Table of Contents List of Tables ................................................................................................................ ..... iii List of Figures ............................................................................................................... ..... iv Abstract ...................................................................................................................... ........ vi Chapter 1: Introduction ....................................................................................................... .1 Chapter 2: Background ........................................................................................................4 2.1 Silica Structure ...................................................................................................4 2.2 Silica-Solvent Interaction ...................................................................................5 2.3 Silica-Solute Interaction in an Aqueous System ................................................6 2.4 Silica-Solute Interaction in a Non-Aqueous System .........................................7 2.5 Calcination Atmosphere .....................................................................................8 2.6 Impact of Water in Calcination Atmosphere .....................................................8 2.7 Impact of Oxygen Scavenger in Calcination Atmosphere .................................9 2.8 Impact of Dynamic Atmosphere ......................................................................10 2.9 Impact of Calcination Temperature .................................................................10 2.10 Catalyst Morphology .....................................................................................12 2.11 Conclusion .....................................................................................................12 Chapter 3: Design of Experiments .....................................................................................14 Chapter 4: Experimental Procedure ...................................................................................19 4.1 Catalyst Preparation .........................................................................................19 4.2 Experimental Setup ..........................................................................................20 4.3 Catalyst Characterization .................................................................................21 4.4 Catalyst Testing ..............................................................................................22 Chapter 5: Characterization Results and Discussion .........................................................25 5.1 Verification of Eggshell Thickness ..................................................................25 5.2 XPS ..................................................................................................................26 5.3 Nitrogen Physisorption ....................................................................................30 5.4 Hydrogen Chemisorption .................................................................................31 5.5 Temperature Program Reduction .....................................................................35 5.6 Fourier Transform Infrared Spectroscopy .......................................................38

PAGE 5

ii Chapter 6: Catalyst Performance .......................................................................................42 6.1 Activity Measurement Using Research Grade Gases ......................................42 6.2 Selectivity Assessment Using Research Grade Gases .....................................45 6.3 Selectivity Assessment Using Bi omass Derived Synthesis Gas ......................46 Chapter 7: Discussion of Experimental Findings ..............................................................50 7.1 Metal Deposition in an Aqueous Medium .......................................................50 7.2 Metal Deposition in a Non-Aqueous Medium .................................................51 7.3 Calcination in the Stagnant Air ........................................................................52 7.4 Calcination in the Dynamic Hydrogen ............................................................53 Chapter 8: Conclusions ......................................................................................................55 References .................................................................................................................... ......57

PAGE 6

iii List of Tables Table 1. Important variables that influence catalyst properties. ........................................16 Table 2. Matrix of experiments. .........................................................................................17 Table 3. Reduced set of variables. .....................................................................................18 Table 4. Eggshell thickness of samples prepared under different conditions. ...................25 Table 5. The surface composition of the catalyst based on XPS analysis. ........................27 Table 6. Binding energies and surface contributions by Co2p3/2 and Co2p1/2. ..................29 Table 7. Surface properties of 20% Co/SiO2 catalyst based on N2 physisorption .............31 Table 8. Properties of active metal ba sed on hydrogen dissocia tive adsorption. ..............34 Table 9. Catalytic performance of cobalt-si lica supported catalyst in FT synthesis… .....43

PAGE 7

iv List of Figures Figure 1. The interaction between the meta l and the support in the intermediate pH range. ..................................................................................................... .........6 Figure 2. A schematic representation of silica surface showing silanol group and hydrogen bonded water molecule. .....................................................................11 Figure 3. Silica surface surrounded by water molecule via multiple bonding ...................11 Figure 4. Immobilized single/ multiple a queous layer surrounding silanol groups ...........11 Figure 5. Vertical stacking of alcohol (ethanol) allowing direct contact between the salt and the support. ..................................................................................... 11 Figure 6. Fish Bone analysis of catalyst preparation conditions .......................................14 Figure 7. Magnified images of eggshell prof ile obtained using optical microscope. ........20 Figure 8. Experimental setup for catal yst preparation via precipitation ............................20 Figure 9. Bench scale reactor setup for car rying out Fischer-Tropsch Synthesis ..............24 Figure 10. (a) High resolution spectra of the Co2p level of all catalyst samples, (b) decomposition of Sample 2 spectra using Gaussian fit ..............................29 Figure 11. (a) Total hydrogen uptake at 373 K, (b) combined isotherm at 373K .............32 Figure 12. Temperature program reduction .......................................................................35 Figure 13. Infrared spectra of catalyst us ing attenuated total reflectance (ATR) ..............41 Figure 14. CO conversion with time. .................................................................................44 Figure 15. GC analysis of liquid produc t obtained from FTS of Sample 1. ......................47 Figure 16. GC analysis of liquid produc t obtained from FTS of Sample 2 .......................47 Figure 17. GC analysis of liquid produc t obtained from FTS of Sample 3 .......................48

PAGE 8

v Figure 18. GC analysis of liquid produc t obtained from FTS of Sample 4 .......................48 Figure 19. Random samples of FTS liquid product produced from the developed catalysts .................................................................................................... .........49 Figure 20. (a) Biomass derived liquid fuel sample, (b) result of third party GC analysis. ............................................................................................................49

PAGE 9

vi Effect of Catalyst Preparation Condi tions on the Performance of Eggshell Cobalt/SiO2 Catalysts for Fischer-Tropsch Synthesis Syed Ali Z. H. Gardezi Abstract A highly selective eggshell Fischer-Tropsch cat alyst has been fabricated via interaction of hydrophobic and hydrophilic molecules on ther mally treated silica gel. The physical interactions of the mesoporous silica suppor t and the effect of catalyst preparation conditions on the performance of the cobalt/SiO2 were explored. It was found that dispersion and performance of the FT cobalt/SiO2 catalyst were significantly affected by the preparation technique used. In this study we focus on tw o key variables: the solvent used during the precursor loading and the calcination atmosphere. Silanol groups on the silica surface and near-surface regions can alter morphology and dispersion of the supported active metals. Solvents used for precursor such as water or alcohol attach to these silanol sites in specific configurations and compete with metal salts during ion exchange and adsorption. By fine tuning the so lvent attachments on heat treated silica we have fabricated a cobalt/silica catalyst with high dispersion and low metal loads. Additionally, since silica has affinity fo r both polar and non-polar molecule depending on the surface conditions; this property has been exploited in preparing an engineered eggshell profile. This together with simu ltaneous calcination/ re duction in a dynamic hydrogen environment has been shown to fu rther enhance dispersion and reducibility. Characterization techniques including BET, XPS, XRD, H-chemisorption and FTIR were employed. Catalyst activity, product selectivit y, distribution and c onversion were studied using a bench scale fixed bed reacto r fitted with a GC/MS instrument.

PAGE 10

1 Chapter 1: Introduction The demand for clean transportation fuel conti nues to grow at an ever increasing pace while limited supplies of fossil continues to increase the fuel prices. Projected trends show that production of crude oil will peak around 2030, while, the global energy requirement is expected to rise by 60% [1 ]. Concerns about global warming are putting pressure on developing rene wable sources for transportatio n fuel. Raw materials like biomass, coal, natural gas, and even municipa l waste can be converted to clean fuel via a thermo-chemical route called Fischer-Trops ch synthesis (FTS), a catalytic route employing transition metals such as Fe, Co and Ru. FTS converts hydrogen and carbon monoxide on a catalyst as per fo llowing nominal reaction [2]: nCO + (2n+1) H2 CnH2n+2 + nH2O where “n” can be anywhere from 1(methane) to 30 or higher (waxes). Exploration of this chemistry by experimental and theoretical techniques reveals that ruthenium, cobalt and iron are the catalysts of choi ce for FTS [3]. The cobalt catalyst has some additional benefits e.g. lower water gas shift activ ity, higher per pass conversion, low rate of attrition, and production of mo re paraffinic hydrocarbons [4]. Ruthenium has the highest activity, but is not commercially used because of its high cost and limited availability. When compared to cobalt and ruthenium, iron has the lowe st cost but produces more waxes and olefins; iron also promotes the undesirable water gas shift reaction. The active catalyst metal is usually deposited on an inert support material like silica, alumina, titania, silicon carbide etc. Alt hough inactive to the reactant gases, these supports can affect the overall synthesis by controlling the morphology of the active catalytic specie, its disper sion, and metal-suppor t interactions. Si lica and alumina

PAGE 11

2 supports are most commonly used due to the st rength and porous nature of the material. Alumina, when used as support, decreases the reducib ility of the cataly st [4]. Silica, on the other hand does not appreciably produce irreducible compounds, however, the active metal dispersion is lower than that for alumina [4]. Iglesia et al. [5] showed that for large metal particles (above 7 nm) the reaction rate is proportional to th e cobalt surface sites (i.e. dispersion). Dispersion of a silica suppor ted catalyst can be improved by the addition of noble metals (Rh, Ru, Pt and Pd). These metals also facilitate the reduction of the metal oxides; however, cost is a limiting factor. Surface wetting of silica with different polar solvents (ethanol, 1 propanol and 1 bu tanol) is another method in practice to increase metal distribution [6]. For an active Co catalyst, CoO and Co3O4 are the desired phase because they easily reduce to cobalt me tal [4]. Synthesis of the active catalyst requires strong interaction between the metal and the support; howev er, if too strong, these interactions can lead to formation of irreducible mixed metal support oxides. Ho et al. [7] observed that the use of ethanol as a so lvent (instead of water) for a cobalt nitrate precursor resulted in metal suppo rt interaction of su fficient strength to increase active metal dispersion while retaini ng a high extent of reduction. A silica surface has affinity for both polar a nd non-polar molecule depending on surface conditions. This property was exploited in th is study in order to prepare an effective eggshell catalyst. Different polar molecules have different attachment patterns on the silica surface, thus by controlling the surf ace condition one can control the dispersion on silica supported particles. Th e solvent used during catalyst preparation also affects the deposition of the salt precursor via ion exchange. Thermal treatment of the catalyst precursor (drying, calcinations and reduction) also impacts the dispersion. The calcination atmosphere changes the arrangement of the surface molecules to some degree and affects the metal support interaction. Earnest et al. [8] showed that during the reduction process the reaction of CoO and SiO2 results in the formation of irreducible cobalt ortho-silicates. This formation of silicates increases the dispersion.

PAGE 12

3 This research work is focused on the devel opment of a highly active, selective, tunable and economic Co/SiO2 Fischer-Tropsch catalyst. To this end the following hypotheses are presented (i) silica surface posse ss both hydrophilic and hydrophobic properties depending on the pretreatment conditions (ii) presence of water on silica surface reduces active metal dispersion (iii) eggshell catalyst profile ensure tune-a bility depending on the thickness of shell (iv) space velocity plays primary role during thermal treatment of catalyst precursor. In order to assist the reader, this thesis is organized as follows. Prior research work is discussed in Chapter 2 in order to identi fy shortcomings, areas of improvement and direction of this study. The expe rimental strategy used in this work as a re sult of these findings has been laid out in Chapter 3. Fina l preparation of representative samples is presented in Chapter 4. Th eoretical comparison and veri fication of the hypotheses via several characterization techniques is presen ted in Chapter 5. Evaluation and discussion of sample performance under actual FTS condi tions in a fixed bed reactor is given in Chapter 6. Based on these results a theory has been developed and presented in Chapter 7 together with future studies.

PAGE 13

4 Chapter 2: Background 2.1 Silica Structure Much controversy exists regarding the natu re of silica surface. Fundamental studies by Stober [9], Meyer and Heckerman [10], and Be ring and Serpinski [11] indicate that a silica surface mainly co nsists of siloxane ne twork in the bulk whil e hydroxyl groups are attached on to the “Si” surface atoms. Howeve r these groups are not similar to each other in adsorption or reaction beha vior. Figure 2 represents a ge neral arrangement on a silica surface. Belyakova et al.[12] identifies that the number of hydroxyl (silanol) groups on different type of silica surfaces are the same i.e. 4 -5 SiOH groups per nm2. Lange [13] states that water associates with these silanol groups in two wa ys, by hydrogen bonding and physically adsorption. Dalton an d Iler [14] state that there is at least a monolayer of water immobilized on the surface silanol groups due to hydrogen bonding under normal conditions. This “glassy layer” tends to shie ld the underlying silica network from foreign species. Klier and Zettlemoyer [15] suggest that water sits “oxyge n down” on the silanol groups. De Boer and Vleeskens [16] argue that the silica loses this adsorbed water at around 120 oC in ambient air, unless it is present in micropores, where the loss is at 180oC. Drying under a vacuum, at low temperatur e is also an effective technique to remove all the surface adsorbed water. Armist ead and Hockey [17] found that removal of the silanol group take s place around 400-450 oC in ambient air where half of the hydroxyl groups leave the silica surface creating large s iloxane areas that do not rehydrate readily. Young and Bush [18] suggest that the silo xane network is essentially hydrophobic in nature, and it excessive ly slows down rehydration. Whenever a silica gel is loaded with an active metal under ambient condition it is not possible to have strong surface-metal intera ction because of the presence of a stagnant

PAGE 14

5 single or double aqueous laye r. In order to obtain suffi ciently strong metal-support interaction the silica surface must be dried at 200 oC or above prior to metal loading. Under ambient conditions, silica gel only attach es polar molecules to its surface. In order to make use of the polar-non polar system in tandem, the silica surface has to be heated to 500-600 oC, thus exposing the hydrophobic siloxa ne space while retaining sufficient “polar regions”. 2.2 Silica–Solvent Interaction On fully hydroxylated silica su rfaces, water first covers all silanol groups via multiple hydrogen bonding. This creates one or at most tw o layers of water in equilibrium with the surface. The key to this strong attachment is the inherent nature of the strong electron donor water which attaches itself with the vi cinal, isolated and even mutual hydrogen bonded silanol group. For organic compounds (e.g. et hanol), Hair et al. [19] as well as Clark-Monks and Ellis [20] cl aim that the best adsorption sites are the freely vibrating isolated silanol groups and th ere is little tendency of bonding on the mutually hydrogen bonded adjacent SiOH. Robert [21] finds that et hanol is adsorbed vert ically oriented on the hydroxyl group (hydrogen bonded to the OH groups of silanol). Furthermore, the number of alcohol molecules esterified on the silica surface depended on the size of the carbon chain and branching [ 22]. Around 3.7 ethoxy groups per nm2 attach on the surface leaving some silanol sites unoccupied [23]. Figure 3 and 5 illustrates the differing attach ment of water and ethanol molecules on a silica surface. Due to the strong affinity of water towards the surface, it is not possible for a solute molecule to anchor directly on the silica base; rather an immobile aqua monolayer will form as shown in Figure 4. It is also possi ble that this monolayer is already present on the gel due to the hygroscopic nature of silica. On the other hand, alcohol due to its vertical orie ntation can lead to random distri bution of vertical stacks on isolated silanol groups. The ga ps between these stacks presen t ideal location for solute anchorage on the silica surface. Due to steric hindrance, not every silica atom reacts with

PAGE 15

6 an alkoxy group, so the concentr ation of ethoxy groups can be less than predicted i.e. 3.7 per nm2. 2.3 Silica-Solute Interactio n in an Aqueous System For the case of a dilute aque ous solution, a classical mode l using the Nernest equation describes the interaction of solute ions with a silica surface [24]. Ac cording to this model a solute ion when in the vici nity of the surface, diffuses in to a “double layer” of ions which are kept in motion due to thermal phonons. Then a significant number of these ions based on their size enter the “stern layer”, a co mpact immobilized layer near the surface. Healy et al. [25] have performed pioneering work on the state of pr ecipitated cobalt ions and possible interaction with a silica surface. At low pH (6 or below) in low salt concentrations, the active specie is Co2+, swarming in a diffuse double layer (kept in motion by thermal phonons). At a pH of approximately 6.5 but well below the complete precipitation limit the aqueous salt near the surf ace is able to anchor; the interaction takes following form. Figure 1. The interaction between the metal a nd the support in the intermediate pH range Upon precipitation (i.e. pH =8) the adsorbed sp ecies exist in the form of polymeric chain of cobalt (II) hydroxide and it covers entire surface. This branched chain also consists of anions and water molecules. Rajamathi and Kama th [26] state that the composition of this specie depends on the temp erature. At around 80oC in an aqueous system a hydrotalcitetype structure is obtained having the formula Co (OH)2(NO3)0.27 .2/3 H2O. While close to 100oC it takes the form Co(OH)1.75(NO3)0.25 .2/3 H2O.

PAGE 16

7 During the precipitation process, if the pH of the support surface is above its isoelectric point, the direct contact between salt ions and the surface is possible. However, these salt ions can attach in many different forms de pending on the precipitat ion process. In the case of a well mixed, fully precipitated syst em a polymeric chain may attach to the support surface, otherwise the hydrated ions of Co2+ will be in close proximity to the silica. These ions may remain in thermal motion or attach to the support surface. 2.4 Silica-Solute Interaction in a Non-Aqueous System Generally, it is believed that an inorganic salt dissolved in an organic solvent can be adsorbed on a silica surface [27] For silica gel the order of the adsorption is as follows, alcohol > salt ions > acetone. Ru ssell et al. [27] state that the addition of a small amount of water (2-3%) increases the adsorption. They also ascertain that adsorption is preferred over ion exchange. Dugger et al. [28] suggest that alkali metal nitrates dissolved in ethanol react slightly with the surface of silica gel. However, when higher alcohols are used which concentrate the silanol by thei r hydrogen bonding (large width and branch number)) no adsorption is observed. For this reason, ethanol was used in this study. Adsorption sites for the salt and alcohol are th e polar silanol groups on the silica surface, as shown in Figure 5. Alcohol is not adsorbed on all silanol groups (only half of the available silanol groups are c overed), however, alc ohol is more preferentially adsorbed than salt. The strong contact between salt and support requir es a strong attractive force i.e. negative silanol site (SiO-) attracting positive cobalt ions (Co2+). It seems that around 3% by weight of water in salt-alcohol so lution can cause enough ionization required to renders sufficient negative ch arge to the silica surface. This promotes ion exchange between salt and silica resulting in coor dination bonding between them. Under these circumstances, the anchorage of salt on the silica surface will be ra ndom and far apart, reducing the possibility of agglomeration. Als o, for substantial loading, silica gel has to be in contact with a very concentrated salt-a lcohol solution to avoid the formation of the stagnant aqua layer by excess water molecules.

PAGE 17

8 2.5 Calcination Atmosphere Thermal treatment (calcination) of the catalyst precursor controls the final distribution of active metal. During calcination it is necessa ry to provide an environment conducive for smooth precursor decomposition to avoid sinter ing and to ensure high dispersion of the active metal. Other important parameters include the control of heating rate, an atmosphere that retards crystal growth, and fl ow rates that ensure swift removal of heat treatment product. Despite its importance, there is lack of constructiv e research in this area. 2.6 Impact of Water in Ca lcination Atmosphere Borg et al. [29] identify that under identical dynamic conditions, the presence or absence of steam in air impacts the active metal’s cr ystal size and the amount of residual nitrates in the final product. They concluded that th e decomposition of cobalt nitrate leads to the formation of water. If water is already pr esent in the calcination atmosphere, then the decomposition product cannot be easily rem oved. The long residence time of water on the surface favors crystal growth. Puskas et al [30] state that wate r reacts with cobalt oxide on catalyst surface to form a hydroxi de. Cobalt hydroxide fu rther reacts with migrating silicic acid to form hard to redu ce silicates which are inactive in FischerTropsch Synthesis. CoO + H2O Co(OH)2 Co(OH)2 + SiO(OH)2 CoSiO3 + 2H2O The equilibrium concentration of water regardless of the atmosphere seems to be the primary factor that controls the catalyst proper ties. However, claims are controversial. It appears that the presence of water during pr ecursor decomposition is detrimental; it gives rise to mixed metal-support oxide and decrea ses the extent of catal yst reduction. On the other hand, there are claims that support th e formation of agglomerated species, and indicate a significant reduction in metal support interaction in the presence of moist air.

PAGE 18

9 2.7 Impact of Oxygen Scavengers in Calcination Atmosphere Sietsma et al. [33] argue that the presence of NO/He in the calcination atmosphere leads to a gradual decomposition of the nitrate precursor following zero order kinetics. These results were confirmed by thermal and kineti c analysis. According to Sietsma, a sudden surge in endothermic heat transfer is obser ved during calcination in the static air. The inert NO/He environment improves repartit ion of the active metal on the support. Similarly, Linyang et al. [34] identify that the presence of nitr ogen during calcination leads to improved segregation between the active metal and added promoter. This can enhance the interaction between the active meta l and the underlying supp ort. Borg et al. [29] observe that the presence of a dynami c nitrogen atmosphere improves the metal dispersion and its precursor decomposition. Puskas et al. [30] provides an alternate expl anation about the impact of thermal treatment in a reducing atmosphere. Here cobalt silicate is formed dur ing calcination in a reducing or inert environment. The formation of water and metal hydroxides during the reduction process (or in an inert atmo sphere) provides a low potenti al energy pathway for the development of ortho as well as meta-silicates. He also states that the space velocity of the calcination gas plays an important role in controlling the reaction equilibrium during cobalt silicate formation. There exists a great deal of controversy ove r the possible advantages and disadvantages of thermal treatment using oxygen scaven ging gases. Some claim that under the scavenging environment the he at flux over the surface is gr adual; this avoids thermal sintering of metal particle s [33]. The surface rearrangem ent also leads to more segregation between the partic les and more contact with th e support. While others state that if the space veloc ity of the gases is kept low, there is no advantage of using scavenging or reducing atmosphere as the prod uct water will eventually give rise to a moist atmosphere [30]. It appears that the fl ow rate of calcination gases control the final properties of active catalyst, for favorable result s, the flow rate should be kept as high as possible.

PAGE 19

10 2.8 Impact of Dynamic Atmosphere Borg et al. [29] have compar ed the properties of a Co/SiO2 catalyst prepared under dynamic nitrogen and dynamic air. They are of the view that ambient air with a higher flow rate allows a smaller particle size distribution when compared to an inert atmosphere at low flow. Sietsma et al. [ 33] have found that as the gas hourly space velocity of NO is increased during calcinati on, the average crystal size of a NiO/SiO2 catalyst system is reduced. In conclusion th e degree of size reduction varies with the GHSV i.e. at low velocities th e variation is the highest. 2.9 Impact of Calcination Temperature The purpose of high temperatur e calcination is to decompos e the precursor to oxide states; however there are certai n factors that must be addre ssed. Borg et al. [29] have shown that the cobalt crystal size increa ses with increasing the temperature of calcinations from 400 to 700 oC due to sintering of meta l crystallites. The observed catalyst TPR profile shows a shift to a hi gher metal support interaction when the calcination temperature is incr eased. Conversely Coulter and Sault [34] observed the formation of metal-support oxides at low temperatures under vacuum annealing. Jablonski et al. [35] stat e that calcination around 800-900 oC (either in argon or oxygen) results in the decomposition of Co3O4 to CoO which further lead s to the reaction between CoO and SiO2 forming cobalt silicate. These cobalt silicates are only pa rtially reducible at high temperature (600 oC). Kababji et al. [31] state that high calcination temperature approximately 500 oC leads to the loss in surface area due to silica migration. Tao et al. [36] have reported that a slow calcin ations temperature ramp rate of 0.5-2 oC/min ensures superior textural properties of silica supported cobalt catalysts.

PAGE 20

F F igure 2. A s Figur e Figure 4 Figure 5. V e chematic re p e 3. Silica s u 4 Immobiliz e rtical stack i p resenation o b o n u rface surro u ed single/m u i ng of alcoh o 11 o f silica sur f n ded water m u nded by wa t u ltiple aque o o l (ethanol) and the sup p f ace showin g m olecule t er molecul e o us layer su r allowing di r p ort g silanol gr o e via multip l r rounding si l r ect contact b o up and hyd r l e bonding l anol group s b etween the r ogen s salt

PAGE 21

12 2.10 Catalyst Morphology Iglesia et al. [37] has argued that Fischer-T ropsch synthesis is diffusion limited process and the activity of the catalyst depends on th e arrival of reactants, and the removal of products from the active sites. Post et al. [38] showed that eggshell catalyst pellets of 2mm decouple the diffusion limitation from ot her reactor constraints e.g. pressure drop. Peluso et al. [39] have shown that an eggshe ll catalyst with 10% cobalt within the half radius of a 1.81 mm catalyst give s higher middle distillate selec tivity. Iglesia et al. [40] have shown that 1-3 mm diameter pellets are re quired to control pre ssure gradients across a packed bed reactor; however, to reach de sired conversion with these particles reactor volume has to be very large. According to Iglesia, this limitati on can be overcome by using an eggshell catalyst design. To this end, Iglesia has defined a parameter “ ” which is dependent on certain catalyst characterist ics. Based on their modeling results, this parameter can be used to determine the optimum thickness required in an eggshell catalyst to maximize activity. 2.11 Conclusion It is evident that silica gel possesses both the hydrophobic and the hydrophilic properties depending on the temperature and environment. This provides a unique opportunity to tailor interactions on the silica support base d on specific catalytic requirements. Also within a certain class of solv ents, the pattern of attachment with the silica surface varies appreciably. This phenomenon affects the uptak e (loading) of solute and its interaction with the silica surface. Thus, the final properties of the loaded catalyst depend on the type and proportion of the solvents used. There is controversy over the preferable conditions leading to smooth precursor decomposition. The role of inert calcinati on environment is not sufficiently understood. The resultant effect of water on the catalyst surface is still under debate. The impact of gas space velocity during ca lcination requires more inve stigation. The surface-gas interactions during the ca lcination are not known.

PAGE 22

13 In this study, the role of calcination environment (inert and oxygenated) will be investigated. It appears that the space velo city (of the calcination gas) is the primary factor that controls the surf ace properties after thermal treatm ent. On the other hand, the calcination environment may control the kine tics of decomposition. Also there is a possibility that calcination gas interferes w ith the surface atoms resulting in segregation of active metal atoms. All these factors requi re a thorough investigatio n in order to gain a strong insight into the th ermal decomposition process. The “eggshell” profile ensures high catalys t activity and product se lectivity. However, the optimum eggshell thickness has not yet been defined clearly. Much of the prior work is based on the single phase modeling during the Fischer-Tropsch reaction. The experimental evidence required to establish a general correlation between the selectivity and the eggshell thickness is not available. Th ese controversies and lack of specific data presents an extensive opportuni ty of research in order to prove a direct correlation between the product selectivit y, catalyst surface propertie s and eggshell thickness.

PAGE 23

14 Chapter 3: Design of Experiments The design of experiments presented in this wo rk is based on a fish bone analysis shown in Figure 6. From this analysis a matrix of possible experiments has been developed using Taguchi design [41]. Figure 6. Fish Bone analysis of catalyst preparation conditions Prior work indicates that the catalyst activity follows the order; Co2(CO)8 > Co(NO3)2 > Co(CH3COO)2 while chain growth probabili ty follows the sequence Co(NO3)2 > Co2(CO)8 > Co(CH3COO)2 [42]. Based on these findings Co(NO3)2 is a preferred precursor salt. The salt concentration is base d on the solubility in water (134 g/100 ml) and ethanol (9 g/100 mL). Typically a base is used as the precip itation agent for this system. Urea was chosen in this case due to rapid precipitation and slow hydrolysis [43] resulting in high dispersion.

PAGE 24

15 The choice of the solvent is governed by the de sire to develop an eggshell profile. The interactions of polar and nonpolar molecules with the si lica surface and their mutual repulsion provide a novel path to develop the eggshell profile. Ethano l is the preferred polar solvent due to its unique surface attachment and its relative short chain length, which leads to lesser overall branching on the silica surface. The solvent n-heptane was selected as a non polar solvent because it has not been cited in the literatu re for use in FTS catalyst synthesis. The drying rate is adjusted to ensure the stabil ity of egg shell profile. It has been reported that at low rate, drying proceeds slowly down the pores, thus concentrating the precursor towards the center of the pellets. However, if the drying rate is too fast, deposition occurs near the entrance [44]. Thus, the more rapid the evaporation, the grea ter is the chance of eggshell formation. Drying in a vacuum furnace, already heated to the drying temperature i.e. 100oC, can successfully achieve rapid drying. For a given catalyst sample, the TPR (tempera ture program reduction) spectra indicates the optimum temperature range for the reduction process. TPR also indicates the presence of different phases during the reduction pro cess and the optimum rate of reduction. The rate of reduction controls the attachment of cobalt on the silica; yet Ernst et al. [8] attribute this fixation to the reac tion of CoO with silica to form Co2SiO4 during the reduction as shown below. 2 CoO + SiO2 Co2SiO4 Activity is the primary indicator of catalyst effectiveness. However, as shown in the “Fish Bone”, the activity depends on the loadi ng of the catalyst, active metal distribution (dispersion) and percentage reduction. The obj ective is to choose the catalyst preparation and pre-treatment condition to achieve higher dispersion while maintaining the appreciable degree of reduction. Other propertie s like accessibility to catalyst sites (pore size distribution) and surface area have to be kept within acceptable limits. Keeping in

PAGE 25

16 mind these constraints and our area of intere st, following variables (Table 1) influence the dispersion and the reducibility of the supported catalyst. Table 1. Important variables that influence catalyst properties The scope of this study has been limited to the parameters hypothesized to have the strongest impact on activity and selectivity of the catalyst system. As stated earlier the factors that greatly influen ce the catalyst preparation and pre-treatment include: (i) the choice of solvent during precursor loading, and, (ii) the thermal treatment of the precursor. These two areas have not yet been studied in tandem. Thus, the overall object of this work is to identify their relative impact on the final properties of Co/SiO2 eggshell catalyst for FTS. Based on these findings, the final set of expe riments are limited to the impact of changes in precursor solvent and calci nation environment (experiment 2, 5, 10 and 17 in Table 2). These experiments (shown in Ta ble 3) will be used to under stand various aspects leading to the desired catalytic propert ies of high selectivity and activity. The resu lts of this study will not only add to the knowledge base but also lead to the development of catalyst with superior properties. Variable Lower value (-1)Central value (0) Upper value (+1) Precipitation Solvent (PrSl) Alcohol Alcohol (1-2% water) Water Precipitation Environment (PrEv) Nitrogen Dry Air Moist Air Drying/Calcination rate (DrClRt) Slow Nominal Rapid Calcination Environment (CalEv) Hydrogen Nitrogen Static Moist Air Reduction rate (RdRt) 1 K/min 5 K/min 20 K/min

PAGE 26

17 In a broader prospective this work compares the situation where water was present in any form (solvent or moisture in atmosp here) to that where no water exists. Table 2. Matrix of experiments Experiment PrSl PrEv DrClRt CalEv RdRt 1 -1 -1 -1 -1 -1 2 -1 -1 +1 -1 -1 3 -1 -1 -1 +1 -1 4 -1 -1 +1 -1 +1 5 -1 -1 +1 +1 -1 6 -1 +1 -1 -1 -1 7 -1 +1 -1 +1 -1 8 -1 +1 +1 -1 +1 9 -1 +1 +1 +1 +1 10 +1 +1 +1 -1 -1 11 +1 -1 -1 -1 +1 12 +1 -1 -1 +1 +1 13 +1 -1 +1 -1 -1 14 +1 -1 +1 +1 -1 15 +1 +1 -1 -1 +1 16 +1 +1 -1 +1 +1 17 18 +1 +1 +1 +1 +1 +1 +1 +1 -1 +1

PAGE 27

18 Table 3. Reduced set of variables

PAGE 28

19 Chapter 4: Experimental Procedure 4.1 Catalyst Preparation In this study commercially available CA RiACT spherical silica gel pellets (FUJI SILYSIA CHEMICAL LTD, Japan, grade Q10, 5-10 mesh size, surface are 319 m2/g) were used as the catalyst support material. Th e support was first dehydrated in an oven at 450-500 oC. The dehydrated silica pellets were sa turated with n-hept ane solvent. The saturated pellets were then dr ied at a slow rate to 60-65 oC. This ensured that only the external periphery of the gel lost n-heptane. Cobalt nitrate hexahydrate was used as the metal salt precursor. Depending on the design of experiments (shown in Table 3) the me tal salt was either dehydrated at 180-200 oC or left as received. To avoid rehydration (where required) the catalyst samples were prepared in a nitrogen glove-box. The dehydr ated samples were then placed in ethanol (containing 3 – 4% water by weig ht), while the hydrated sample s were placed in water. The salt-ethanol solution was heated to 70oC while the salt-water system was heated to approximately 90 oC. The hot solution was then poured into a fritted funnel. Urea was used as the base for precipitation. Urea so lution was either made in ethanol or water depending on the system (dehydrated or hydrated). The surface dried pellets were immersed in the hot salt solution, and the urea solution was then added drop-wise. During this precipitation step, the entire mixture was continuously stirred to avoid bulk precipitation. The time of preci pitation was adjusted to co ntrol the amount of cobalt loaded (around 20% within eggshell). A vac uum pump was used to remove the excess solution. All loaded catalysts were subjected to rapid drying (pressure dropped from zero to 30 in-Hg gauge in one minute) in a vacuum furnace. After drying the samples were

PAGE 29

20 either calcined in air (hydrated samples) or simultaneously reduced/calcined in a packed bed reactor (dehydrated samples). The thic kness of the eggshell wa s based on previous work done by Iglesia et al. [37]. Factors th at controlled the thic kness of the eggshell included (i) drying rate and temperature duri ng the removal of n-heptane from the gel’s outer periphery, (ii) contact time of the gel with the precursor solution during precipitation, and (iii) the rate of vacuum drying. Magnified images of the eggshell catalyst are shown in Figure 7. Table 4 indicates the measured and required thickness of the eggshell. Conceptual diagram Sample-2 Sample-4 Figure 7. Magnified images of eggshell prof ile obtained using optical microscope 4.2 Experimental Setup Figure 8. Experimental setup for cat alyst preparation via precipitation Data Solution:Burette……………………Urea Funnel…………………….Cobalt nitrate Titration Temperature:Aqueous System…………90 oC Non Aqueous System……70 oC Titration Time:Aqueous System…………1min 20 s Non Aqueous System……1min 40 s Silica Cobalt

PAGE 30

21 4.3 Catalyst Characterization An optical microscope was used to determin e the thickness of the egg shell. Randomly selected samples were cut in half using a razor blade and then further sliced to get circular cross sections (showi ng egg shell and silica). Sand pa per (600 grit) was used to polish the samples. The eggshell thickness wa s calculated from magnified images of the 2mm catalyst samples. In order to characterize th e surface and near surface species present on the prepared samples, XPS (X-Ray photoele ctron spectroscopy) was carried out in a Perkin Elmer PHI 560 UHV XPS/ SAM system. Sample degass ing was performed for 24 hours under 10-6 Torr vacuum. XPS elemental analysis was done using both Al K and Mg K radiation based on the binding energy range and resolu tion requirement. A “Gaussian Curve Fit” was used to identify the chemical-state and relative concentration of elemental species. References for the binding energies were taken from the Handbook of Photoelectron Spectroscopy [45] and prior research work [8]. Note that XPS is a surface analysis technique limited to 10-1 monolayer fractions and is not representative of bulk composition. N2-physisorption was performed at -196 oC using the Quantachrome Autosorb gas sorption system. Each sample wa s degassed under UHV for 24 hours at 100 o C. BJH method was applied on the desorption branch of the isotherm in order to calculate the pore volume as a function of pore size. The pore diameter was taken as the one where maximum differential pore volume occurred. Hydrogen adsorption isotherm was taken on a Quantachrome Autosorb gas sorption unit at 373 K. The samples were reduced in hydrogen at 673 K; the flow of hydrogen was maintained to swiftly remove the wa ter formed during reduction. In situ calcination/reduction samples were directly tran sferred to the appara tus after drying in a vacuum. Following reduction, the samples we re evacuated at 673 K and then cooled down to 373 K. An adsorption isotherm wa s recorded from 80-560 mmHg (gauge). The

PAGE 31

22 amount of chemisorbed hydrogen was determin ed by extrapolating the straight-line portion of the isotherm to zero pressure. Fo r calculating dispersion, it was assumed that two cobalt sites were covered by one hydrogen molecule and all the exposed cobalt atoms were reduced to metallic cobalt. Hydrogen ch emisorptions provided strong insight into metal support interaction and dispersion. In a way the results of hydrogen chemisorption form the basis of our studies. Temperature programmed reduction (TPR) was al so done in the Quantachrome Autosorb gas sorption unit. The sample was exposed to the reducing gas (pure hydrogen) while the temperature was increased at 5 K/min from ambient to 973 K. The consumption of hydrogen was measured by analyzing the effl uent gas with the thermal conductivity detector. By plotting the output signal against the temp erature a catalyst reduction pattern was obtained, indicating metal-support interaction and the presence/absence of silicates. TPR results proved useful as they provided in situ re sults for the hydrogen calcined catalyst. A Bio-RAD Excalibur FTS3000 FTIR operated in ATR (attenuated total reflectance) mode was used to further characterize the different catalyst samples. For this purpose, a Pike Technologies diamond MIRacle single reflection horizontal attenuated total reflectance (HATR) unit designed for use in FTIR spectrometer was utilized. Two hundred and fifty scans were recorded fo r each sample at a resolution of 4 cm-1 and a sensitivity of 16. A permanently aligned gas cel l was used to analyze the effluent gases during Fischer-Tropsch runs. The 60 mm diameter gas cell pr ovided an infrared beam length of 2.4 m and consisted of a borosilic ate glass body with a KBr window and a total volume of 0.1 liter. 4.4 Catalyst Testing The different catalyst samples were tested in a fixed bed reacto r (BTRS Jr supplied by Autoclave Engineers) at conventional FT S conditions. Two mass flow controllers by Brooks instrument (Model 5850 EM and EC) were used to control the flow of reactant

PAGE 32

g o i n i n b c o m i n a r w d y o w a c t o r e C A m c o ases (CO a n f inert gas n side the m a n ternal dia m ed temperat u o nnected to m aintain the n ert quartz c r ound 6 m L w aiting to b e y namic hyd r f the bed w a w as carried o c hieved, W H o approxim a e action isot h C O conver s A bsorbance v m onoxide a n o nversion n d H2) and a (N2), Figur e a in assembl y m eter of 0.01 u re a therm o a PID temp system pre s c hips in the L (2.4 g). D e in situ ca l r ogen strea m a s the main c o ut at 200 O C H SV was m a a tely 230 oC h erm in the a s ion was c v ersus conc e n d methan e a n Omega fl o e 9. The m y The react 3 m fitted w o couple was erature con t s sure. The c ratio of 1:4 D ifferent cat a l cine d ) wer e m (flow rate c ause for su c C an d 20 b a a intained at The purp o a bsence of a n c alculated e ntration pl o e ). The fol l 23 o w meter (F m ixed gases w or itself co n w ith a jack e inserted th r t roller. A m a c atalyst bed by volume ; a lyst sampl e e reduced a t = 5 L/min) f c h a high fl o a r with H2/C 94 g/ (hr.gc a o se of the q u n external c o using Fou r o ts were de v l owing nor m MA 1818) w w ere pre-he n sisted of a e t heater. In r oughout the a nual backp r consisted o ; the total v o e s (those c a t atmospher i f or 16 hrs. T o w of the re d O ratio of 2 a t+inert) and t e u artz chips w o oling mech a r ier transf o v eloped for m alized ex p w as used to ated by an o 0.43 m SS 3 order to m o length of t h r essure reg u o f active cat a o lume of a c a lcined in a i i c pressure T he low the r d uction gas. :1, once th e e mperature w w as to effec t a nism. o rm infrare the individ u p ression w a monitor the o ven heate r 3 16 tube w i o nitor the ca t h e be d whic h u lator was u s a lyst mixed c tive catalys t i r & those d and 400 o C r mal conduc The FT re a e equilibriu m w as raised s l t ively contr o d spectros c u al gases (c a a s used fo r flow built i th an t alyst h was s ed to with t was d rie d C in a tivity a ction m was l owly o l the c opy. a rbon r CO

PAGE 33

24 Figure 9. Bench-scale reactor setup for carrying out Fischer-Tropsch synthesis

PAGE 34

25 Chapter 5: Characterization Results and Discussion 5.1 Verification of Eggshell Thickness Iglesia et al. [37] have identif ied that the eggshell thickness is related to the cobalt surface density and the pellet’s mean pore radius via parameter “ ”, as shown below. = (Ro –R c)2 M / rp Where Ro is the pellet radius, Rc is the radius of internal non impregnated core, M is the cobalt site density (atoms /m2) and rp is the pore radius. is the parameter that strictly depends on the structural propertie s of the catalyst. Iglesia et al. [37] has suggested that for optimal activity and selectivity of eggshell catalyst, the value of must lie between 200 – 2000 10-16, (more preferably 100 – 1000 10-16). Based on the hydrogen chemisorption results (for M) and the nitrogen physisor ption results (for rp), the optimum eggshell thickness has been calculated and compared with the results from optical microscopy as shown in Table 4. Table 4. Eggshell thickness of samples prepared under different conditions Sample ID Mean Pore Radius (109 *m) Cobalt Density On Surface (1017 Co.m-2) 10-16 (Ro –R c) Calculated (mm) (Ro –R c) Actual (mm) Sample 1 6.25 0.53 100-1000 0.47-1.00 0.50 Sample 2 4.75 3.48 100-1000 0.19-0.42 0.25 Sample 3 5.00 5.20 100-1000 0.15-0.34 0.25 Sample 4 6.00 1.80 100-1000 0.26-0.58 0.30

PAGE 35

26 These results show that the required thic kness depends on the dispersion. A highly dispersed catalyst (cobalt surface density) re quires a smaller thickness which results in the reduction of the overall coba lt loading and cost saving (a nd vice versa). The thickness of the shell also impacts the product sel ectivity. As explained by Iglesia [37] the hydrocarbon product selectivity depends on the arrival of the diffusion limited reactant (CO) at the active sites. As more active sites are available for CO adsorption, the probability of chain growth incr eases (in case of a thin shel l). On the other hand, the low inter-pellet concentration increases the required eggshell thickness which favors the formation of lighter hydrocarbons. The eggshell thickness of all the samples was w ithin the required limits (it is important to note that a random sampling technique has be en employed and the reported thickness is based on the analysis of five samples in each case). In line with the above mentioned hypothesis which relates CO adso rption to the number of active sites, the Sample 2, 3 and 4, have produced FTS liquid product in the ra nge of diesel and aviation fuel. This is shown by the GC analysis in Figures 17-19. Ho wever, the GC analysis of the Sample 1 (Figure 16) has shifted towards lower hydrocar bons owing to a thicke r shell. Apart from eggshell thickness, other factors al so influence the product selectivity. 5.2 XPS The surface and near surface elemental state of all the samples was analyzed by XPS as shown in Table 5. Clearly the sample devel oped in ethanol and calcined in situ in hydrogen has the highest Co su rface content. However, on the other hand, the catalyst synthesized in aqueous media a nd calcined in air showed the least Co concentration. For characterization these samples were cut in he mispheres and analyzed such that only the top hemispherical section was under the focus of the incident beam. Hence, the analysis is strictly quaitative, however, it indicates th e repartition of cobalt ions on silica surface. Khodakov [2] states that the ICO/ISi ratio represents the dispersi on of cobalt ions on a silica surface. In other word s higher ratios are characteris tic of higher dispersion, while

PAGE 36

27 lower ratios indicate that metal agglomeration has taken place. Table 5 shows the ratio for the different samples, which increases in the following order, Sample 3 > Sample2 > Sample4 > Sample1. Thus, the precursor deposition and decomposition conditions control the active metal distribution on the si lica support; however, the degree of impact is different for each case. The effect of th e solvent and the precipitation environment is greater than that of the calcinations atmosphere. This is demonstrated by the intensity ratios of Samples 2 and 3 (same preci pitation conditions, di fferent calcination environment) verses that between the Sa mples 3 and 4 (different precipitation conditions). The precipitation solvent controls the interaction between the salt and the silica surface. This interaction is the basis for the build ing block that controls metal crystallite distribution on the surface. It was observed th at an inert calcination environment during the precursor loading limited rehydration of th e silica surface resulti ng in reduced silicametal interactions. On the other hand, the calcination environment controls the decomposition of precursor (cobalt nitrate) a nd the formation of cobalt silicate. It has no impact on the initial interacti on developed during the salt deposition. This was consistent with Sample 1 prepared under least favorable theoretical conditions (water as precursor solvent, impregnation in ambient air and calc ination in stagnant ai r) having the lowest intensity ratio. Table 5. The surface composition of the catalyst based on XPS analysis S= Sensitivity factor; values available in literature [45] Catalyst Surface Atom Content (n%) nCo / nSi SCO / SSi ICO / ISi Co O Si Sample 1 2.5 70.0 27.5 0.1 10.6 1.1 Sample 2 15.0 66.6 18.4 0.8 10.6 8.5 Sample 3 27.0 62.5 10.5 2.6 10.6 27.6 Sample 4 8.0 66.1 25.9 0.3 10.6 3.2

PAGE 37

28 High resolution XPS spectrum of the Co 2p region is shown in Figure 10. The presence of Co3O4 is identified by the peaks located in the range of 779.5 to 780.2 eV [45]. In order to confirm this resu lt, decomposition of the Co2p spectrum has been performed using a “Gaussian fit” routine. The feature consists of CoIII (octahedral), CoII (tetrahedral) and CoII satellites as shown in Fig 10(b). The bi nding energies shown in Table 6 are in agreement with available values found in literature [45]. Earn st et al. [8] has found that the area ratios of Co 2p3/2 peaks of CoIII and CoII is around 2 which gi ves a corresponding formula Co+2(Co+3)2O4. Comparison of the high resolution spectra (Fig 10(a)) shows a relatively intense satellite structure (787.0 and 803.0 eV) for the Sample 1 a nd 2. Girardon et al. [46] argue that this is a characteristic of Co2+ ions associated with the residua l cobalt nitrate. Since, both of these samples were calcined in air, it can be co ncluded that calcination in static air lead to partial decomposition of cobalt nitrate. This observation is confirmed by the FTIR spectra shown in Figure 13. As motione d earlier, the decomposition in static air is rapid, accompanied by a surge in heat transfer, while the decomposition in hydrogen is gradual with a higher degree of completion. No mixed metal-support oxides are observed with in the first few atomic layers of the outer periphery via XPS analysis. However, bulk techniques are re quired to identify the extent of any metal support in teractions on different sample s. For this reason, XRD, TPR and FTIR-ATR were used in this work to access the presence of any metal-support interactions. This interaction can either be in the form of anchors (between salt ions and SiOH) developed during precipita tion or in the form of bulk metal-support oxides e.g. orthosilicates, a high temperat ure product formed in the pr esence of water. In this research the development of the former has been encouraged while the latter is suppressed to give rise to high dispersion and greater reducibility.

PAGE 38

29 Figure 10. (a) High resolution spectra of th e Co2p level of all catalyst samples, (b) decomposition of Sample 2 spectra using Gaussian fit Table 6. Binding energies and surface contributions by Co2p3/2 and Co2p1/2 (Fig 10(b)) Catalyst Sample Binding Energy (eV) Co2p3/2 Co+3 Co+2 Co+2 (satellite) Binding Energy (eV) Co2p1/2 Co+3 Co+2 Co+2 (satellite) Sample-1 780.0 781.0 787.0 795.0 797.0 803.0 Sample-2 780.0 781.5787.5 795.0 796.5 803.5 Sample-3 780.0 782.5 787.5 795.0 796.5 802.5 Sample-4 779.0 780.5 785.5 794.0 796.0 802.5 2p1/2 2p3/2

PAGE 39

30 5.3 Nitrogen Physisorption BET surface area measurements are presented in Table 7. The data indicate that the surface area is dependent on the calcination envi ronment and the solvent used. Sample 3, which was synthesized using ethanol as th e solvent under a nitrogen atmosphere and calcined in dynamic hydrogen (most favored theo retical route), has retained most of the original surface area (surface ar ea of CARiACT support is 319 m2/g). However, the presence of water (either as solvent or as moisture in atmosphere) reduces the surface area as shown by the Samples 1 and 4. It is evident that a transition in the catalyst loading conditi ons (from ethanol to water and nitrogen to ambient air) changes the su rface area by 20% (Sample 3 vs. Sample 4). However, the change in the calcination environment (from static air to dynamic hydrogen) has approximately half the effect (S ample 2 vs. Sample 3). Thus, like the case of active metal distribution, the surface area also depends on precursor loading and decomposition conditions, but the impact of the loading parameters is more significant than that of the calcination conditions. Pore size distribution plays an important role in selectivity of high molecular weight hydrocarbons. Song and Li [47] state that pore size in the range of 6-10 nm display optimal Fischer-Tropsch activity and higher C5+ selectivity. CARiACT Q-10 support has mean pore size of around 10 nm (as stated by Fujitech). As shown in Table 7, the presence of water increases the pore size, shifting the mean pore size distribution to higher values. The large pore si ze reduces the required residence time of gases for chain growth. This relative impact on C5+ distribution is shown by the gas chromatographic (GC) analysis of the Sample 1 (Figure 16). Samples 2 a nd 3 have pores within the desirable size range which favors chain growth due to optimum residence time within the pores according to the literature cited. The GC results (Figure 17 and 18) confirm this hypothesis; however, the distribution depends on other factors such as dispersion and eggshell thickness.

PAGE 40

31 Table 7. Surface properties of 20% Co/SiO2 catalyst based on N2 physisorption 5.4 Hydrogen Chemisorption Dissociative hydrogen chemisorption was used fo r an in depth analysis of metal surface and metal-support interac tion. Reuel and Bartholo mew [48] have reporte d that this is an activated process which requires high temperat ure to ensure irreversibility. Numerous studies show that 373 K (100 oC) is the temperature that provides optimum irreversible adsorption [49]. The total hydrogen uptake for different catalyst samples (at 373K) is shown in Figure 11 along with the actual isotherm. The chemisorptive monolayer was obtained by extrapolation to zero pressure. Total hydrogen uptake in Figure 11(a) represen ts both the reversible and irreversible uptake. However, since the isotherm was developed at 373 K (optimum for dissociative adsorption), this adsorption trend represents the activity trend of different catalyst samples. Bligaard et al. [3] has shown that the dissociative adsorption on the catalyst surface is the rate limiting step in Fischer-Tropsch synthesis. So, the rate of FTS should vary in the following order: Sample 3 > Sample 2 > Sample 4 > Sample 1. Actual runs under FTS conditions have confirmed this hypothesis as shown in Table 9. Monolayer volume at zero pressu re (extrapolation of the isot herm) determines the surface properties of different samples. These quant itative results not only serve the purpose of comparison but also correlate microscopi c properties with ca talyst activity. Catalyst ID Specific Surface Properties Pore Size (nm) BET Surface Area (m2/g) BJH Pore Volume (cm3/g) Sample 1 255.0 1.04 13.0 Sample 2 290.7 0.91 9.5 Sample 3 313.0 1.01 10.0 Sample 4 260.3 0.97 12.0

PAGE 41

F Figure 1 ollowing fo r Monolay e 1 1. (a) Total r mulae are u e r Uptake ( hydrogen u u sed to calc u mol/g) 32 u ptake at 37 3 u late the pro p 3 K (b) com b p erties of ac t Vm = M o b ined isoth e t ive surface : o nolayer Co v e rm at 373 K : v erage

PAGE 42

T a l t h u c a S ( c s e g r P s o Active S u Active M Average C T he obtaine d l cohol perm h e higher d i sing alcoho l a lcinations ample 2, w c alcined in e gregation o r adual dec o resence of w o lvent, effe c u rface Area ( etal Dispers C rystal Size d results are its the direc t i spersion o f l as the sol v favors crys w hich was c dynamic o f an active o mposition o w ater in the f c tively bloc k (m 2/g) ion (%) : (Angstro m shown in T t contact of f catalyst sa m v ent. As sta t tal growth c alcined in hydrogen). metal (as c i o f the nitra t f orm of wat e k s contact b e 33 m ) T able 8. As metal and s u m ple 2 and t ed by Borg and agglo m static air s The dyn a i ted earlier) t e salt. All e r of hydrati e tween the p = A d Am =Cr o ac t = 6. 6 M = mo l L = Per c m e C1= 94 [ hypothesiz e u pport. This 3, b oth of et al. [29] t m eration. I n s hows lowe r a mic hydro g and avoids these cond i on on the si l p recursor a n d sorption St o o ss-sectional t ive surface a 6 oA2 l ecular wei g c ent loadin g e tal [ 50] e d earlier, w anchorage i which hav e t he presenc e n line with r dispersio n g en enviro n the therma l i tions favor l ica surface, n d the supp o o ichiometry area of eac h a tom g ht = 59 g of support e w etting patte r i s responsib l e b een fabri e of water d this hypot h n than Sam p n ment imp r l sintering d high dispe r or as a pre c o rt. Thus, d e h e d r n by l e for cated d uring h esis, p le 3 r oves d ue to r sion. c ursor e spite

PAGE 43

34 the favorable calcinations condition, Sample 4 (prepared in aqueous media) shows lower dispersion than Sample 2 (prepared in alcoho l). Sample 1, being prepared in water and calcined in air, shows greater agglomeration of active metal evident by large crystal size and low dispersion. For the case of cobalt/silica catalyst, crystal growth follows an opposite trend to that of dispersion. High support interact ions result in reduced agglomeration, and smaller average crystal size. As shown in Table 8, cr ystal size increases in the following order, Sample 1 > Sample 4 > Sample 2 > Sample 3. Thermodynamic analysis of the catalyst kinetics [51] shows that nano particles in th e size range of 6 -11 nm show a maxima in Fischer-Tropsch synthesis activity. In acco rdance with this hypothesis, Sample 3 is expected to show optimum properties for Fisc her-Tropsch synthesis. Conversely, Sample 1 is expected to show minimum conversion. Table 8. Properties of active metal based on hydrogen dissociative adsorption The results of hydrogen chemisorption are inco mplete without estima ting the extent of reduction. There is the possibili ty that interactions betw een the metal and the support hamper this process. Temperature program reduction (TPR) gives a qualitative measure of the ease of reduction process. Although be ing qualitative in natu re it effectively differentiates between positive metal support in teraction (giving rise to high dispersion) Sample ID Actual Loading Monolayer Volume Vm (cc/g) Monolayer Uptake Nm ( mol/g) Active Surface Area ASA (m2/g) Dispersion D (%age) Average Crystal Size d (nm) Sample 1 16.0 0.2 10.7 0.9 1.0 94.0 Sample 2 10.0 1.0 50.0 4.0 6.0 15.0 Sample 3 10.0 1.7 76.0 6.1 8.9 10.5 Sample 4 10.0 0.6 25.0 2.0 3.0 31.3

PAGE 44

35 and the irreducible ortho-sili cates. TPR results confirm the hypothesis for calcination environment when viewed in conjunction w ith the hydrogen chemisorptions results. 5.5 Temperature Program Reduction Temperature program reduction is a valuable technique for evaluating the extent of reduction. The reducibility of a supported c obalt catalyst depends on the preparation condition, reduction conditions, the promoter a nd the choice of support [4]. Presence of strong metal support interaction can also impact the reduction process; if the interaction exists to the extent that hydro silicates ar e formed, the reduction process can extend up to 1100 K. However, under normal conditions supported and unsupported Co3O4 spinal follows a two step reduction process as shown below [52]. Co3O4 + H2 CoO + H2O 3CoO + 3 H2 3Co+ 3H2O During TPR two distinct, but slightly overlapping peaks are usually observed. These peaks can individually be assigned to the two step reduction process if their hydrogen consumption ratio is around 1/3 (stoichiometric ratio). However, deviations have also been observed especially in the case of bimodal particle distribution [8] in which, individual peaks represent reduction of a certain co llection of particle sizes. Figure 12. Temperature program reduction. Ramp rate= 5 K/min

PAGE 45

36 Figure 12 shows the TPR of the different catalyst samples. Due to the tight control of parameters during precipitation de position, the possibility of bimodal particle distribution has been ignored. Steen et al. [6] have ex tensively studied the reduction profile of different catalyst samples prepared by varying the solvents, the silica supports, the degree of cobalt nitrate dehydration, the drying time and the calcination temperature. They assign the following temperature range for re duction of different species; 540-560 K for reduction of trivalent to divalent cobalt, 570620 K for reduction of divalent with hardly any interaction with carrier, and around 700 -770 K for species with very strong interaction with the support surface. Formation of the cobalt silicates leads to broad reduction peaks ranging between 800 – 1100K. The TPR profile of Sample 1 shows the firs t peak centered at 500 K while the second distinct peak has a maxima at 580 K. Th ese peaks represent the two step reduction process with an area ratio of 0.35 relative to ea ch other. Castner et al. [53] reports that stabilization of the oxide phase due to metal support interaction results in broadening of the second peak. Thus, the narrow second peak of Sample 1 (as compared to that of Sample 2), represents a lesser interacti on between the metal and the support. In comparison to Sample 2 and 3 there is also a shift in the entire spectrum of Sample 1 towards lower temperature. This shift is al so reported to be a m easure of metal-support interaction [54]. These results confirm that the presence of wa ter during precipitation-deposition prevents direct contact of the active metal with the silica surface. However, the presence of water during calcination may form some bulk meta l-support oxides. However, the impact of loading conditions is much stronger than that of calcination environment. Hydrogen chemisorption results are in line with TPR results, Sample 1 shows strong agglomeration and a very low dispersion, an indica tor of low metal s upport interaction. The TPR spectrum of Sample 2 shows a sm all peak at around 400K, this represents decomposition of residual cobalt nitrate. As discussed earlier, the XPS spectrum of this sample shows a strong satellite peak whic h has already been attributed to the non-

PAGE 46

37 decomposed nitrates. The second and third p eak centered at 520 and 610 K represents the two step reduction process w ith an area ratio of 0.38. The broadening of second peak well beyond 750 K indicates a relatively strong metal support interaction. In fact the instrument limitation restricts analysis to 873K, preventing the identification of any cobalt silicates. The use of ethanol permits interaction of the metal and the support due to its specific attachment on the silica surface. This is c onfirmed by the broadening of the second TPR peak. During the calcination in static air ther e is a possibility of forming irreducible oxides as well as that of aggl omeration (cited earlier). The working limitations restrict experimental observation; otherwise one may have observed a peak at around 1070 K attributed to hydro silicates. Hydrogen chemisorptions results confirm the increase in dispersion and the decrease in crystal size as compared to Sample 1 indicative of enhanced interactions. However, as compare to Sample 3, the dispersion is low, due to agglomeration caused by rapid heat surges and insufficient removal of calcinations product in static air. TPR results of Sample 3 look remarkably diffe rent from other two catalysts analyzed. As this catalyst is only vacuum dried, the sharp peak appear ing at around 480K represents the decomposition of nitrate ions. During the preparation of the cat alyst, cobalt nitrate salt is dehydrated at around 180 oC, it is possible that some cobalt oxides are formed at that high temperature. Thus, the two relativ ely small peaks correspond either to the reduction of those oxides or to the reducti on of intermediate products formed during nitrate decomposition. Previous studies show that under nitrogen and hydrogen glow discharge plasma, cobalt nitrate on alumin a decomposes at low temperature producing very fine oxide crystals [2]. The TPR profile of Sample 3 is following the same trend i.e. the decomposition temperature is much lo wer than that of a typical calcination temperature (i.e. 673K) and produces small crys tallites, as shown in Table 8. TPR spectra of Sample 4 (not shown here) consist of nitrate decomposition peak at approximately same temperature (that sample is also in s itu calcined in hydrogen). The two smaller peaks are also present but they are shifted to a lower temperature. This shift can be

PAGE 47

38 attributed to the use of water as a solvent that diminishes metal support contact. As a result the dispersion goes down but reduction process becomes easier. 5.6 Fourier Transform Infrared Spectroscopy In order to confirm the qualitative results of temperature program reduction and gain a thorough insight into the nature of surface intera ction and the formation of cobalt silicate, IR spectra of various samples were obtained. Based on the previous research work [55], following are IR visible, vi bration by silica at 1100, 975, 800 cm-1, by nitrate ions at 1610, 1365, 842 cm-1 by hydroxyl groups at 1640 cm-1 and by Co3O4 at 667 cm-1 CoO is known to be IR inactive specie. As shown in Figure 13, the IR spectra cont ain ordered variation that depends on the preparation and treatment conditions of various samples. In order to elucidate the impact of loading, a pre-calcined por tion of Sample 3 (synthesized in alcohol) has also been included. The IR spectrum of sample 3 shows vibration ba nds at around 670 cm-1. This peak either represents Co3O4 (produced during the dehydration of cobalt nitrate) or Co-O associated with Co(OH)2 produced during titration with urea [56]. At around 800 and 1100 cm-1 peaks corresponding to Si-O are visible. An interesting feature ha s appeared around 1030 cm-1 which corresponds to Si-O-Co bond formed as indicated in prev ious research work [31]. As stated earlier, the use of alcohol during loading on dehydrated and part ially dehydroxylated silica permits the direct contact of cobalt with the support surf ace. Such a contact gi ves rise to Si-O-Co bonding. Two visible peaks around 1320 and 1610 cm-1 corresponds to the nitrate ions present on the surface of the dried catalyst. Additionally there are two low intensity peaks centered around 1420 and 1500 cm-1. Literature survey has revealed that NH4 + ions have an absorption region of 1390-1485 cm-1 corresponding to deformation [57]. Therefore these two peaks correspond to th e ammonium ion form ed during the urea decomposition. However, XPS analysis did not reveal the presence of any nitrogenous specie on the surface of the catalyst.

PAGE 48

39 Sample 2 gives a spectrum similar to the drie d catalyst. This sample was prepared in alcohol and calcined in air. As shown by th e XPS high resolution spectrum of sample 1 and 2, the nitrate ions are not entirely decomposed in stat ic air. The IR spectra confirm this finding, by showing the vibration bands around 1320 and 1610 cm-1. Presence of Co3O4 on the surface is c onfirmed by the vibration band at around 640 cm-1. As expected from its broad TPR peak (well beyond 700 oC), presence of metal support interaction is confirmed in the sample by a feature appearing at around 1020 cm-1. However, it is important to identify the subtle difference in the nature of Si-O-Co peak that appears in the dried sample and the air calcined sample. As, discussed earlier, the metal support interaction at the time of salt deposition is essential to de velop a building block for high dispersion. On the other hand, cal cination in static air keeps the product (including water) in contact with the surface for a longer period of time. In the presence of migrating silica, the presence of water can result in the forma tion of cobalt silicate, a product that severely reduces the catalyst reducibility. However, without a TPR features at around 1000-1100 K it is not possible to differentiate the two. Si-O absorbance peak is visible at 800 and 1100 cm-1. The presence of NH4 + ion is confirmed by a single peak centered at around 1420 cm-1, additional proof of incomplete decomposition in static air. Nitrate peaks are missing in the spectrum of Sample 3 due to complete decomposition in dynamic hydrogen which is consistent to what has been observed in the XPS high resolution spectrum. No cobalt silica interaction is observed. This is very interesting as it seems that the dynamic hydrogen may cause a su rface rearrangement th at retains the high dispersion of cobalt crystallites (as evinced by chemisorptions results of the Sample3), but diminishes the coordination between th e cobalt and the silica (i.e. Si-O-Co). The Co3O4 peak at 670 cm-1 is more intense owing to the presence of small crystallites (also shown in hydrogen chemisorptions resu lts). The silica peaks at 800 and 1100 cm-1 are also more pronounced. A new peak has appeared at around 1650 cm-1. Previous research work [31] associates this peak to silanol groups (SiOH) on the silica surface. It is hypothesized that in situ calcination has cl eansed the surface as a result unassociated silanol groups have become visible.

PAGE 49

40 Since both Sample 1 and 4 were prepared in aqueous medium, it is hypothesized that there is little to no po ssibility of a direct contact betw een the metal and the support. Thus, the Si-O-Co interaction peak is not present in the IR spectra. This promotes the agglomeration as shown by a low dispersion of samples 1 and 4. The complete decomposition of the precursor is evident for the Sample 4 (calcined in situ in hydrogen) due to the disappearance of vibration peaks a ssociated to the nitrates and the ammonium ions. However, Sample 1 shows a smaller peak centered at 1350 a nd another one at 1500 cm-1. The Co-O coupled vibration (associated to Co3O4) is relatively more intense for Sample 4 due to the smaller crystal size when compare to Sample 1. The presence of silanol groups in Sample 4 is indicated by the peak at 1650 cm-1, it is more intense with respect to that of Sample 3.

PAGE 50

Figure 1 1 3. Infrared s s pectra of c a 41 a talysts usin g g attenuated total reflec t t ance (ATR)

PAGE 51

42 Chapter 6: Catalyst Performance 6.1 Activity Measurement Using Research Grade Gases In order to assess the performance of different catalyst samples, FTS has been conducted for 72 hrs in a fixed bed reac tor (detail of which has b een explained earlier). The performance parameters are given Table 9; a ll parameters have been evaluated after 24 hrs at which time steady state is assumed. As shown in Table 9, Sample 3 exhibits the highest CO conversion. This can be explai ned in part by the dissociative hydrogen adsorption results given in Table 8, where Sa mple 3 shows the highest concentration of active metal on the surface (i.e. dispersion). This high active surface area provides more reaction sites and aids in the chain growt h. The opposite holds true for Sample 1. Based on the “ ” requirement an alternate explanation can also be provided. Sample 1 has the thickest “eggshell”, this reduces th e number of sites available per m2 surface area for the diffusion limited reactant i.e. carbon monoxid e for a given percentage of loading (especially when the dispersion is low). While in the case of Sample 3, the active metal is concentrated at the outer periphery, be ing more accessible to carbon monoxide. This results in higher conversion at relatively low loading. The activity of samples 2 and 4 follow the expected trend based on their surf ace properties i.e. Sample 2 > Sample 4. The impact of pore size on the sel ectivity is evident; Samples 2 and 3 have a mean pore size within the desirable range of 6-10 nm, thus they provide optimum inter-pellet residence time for chain growth as shown by the C5+ selectivity. However, the mean pore size of the samples 1 and 4 are greater than 10 nm. Within the pores of these samples, the reactant and products have a less residen ce time than optimum which retards chain growth. The use of cobalt as an active ingredient keeps CO2 production in check. So, despite low temperature (around 503 K), water gas shift reaction does not proceed to an appreciable degree in all cases.

PAGE 52

43 The variation of activity for all the catalys t samples has been studied; Figure 15 shows CO conversion as a function of time. The CO conversion shows a gradual increase for the first 20-24 hrs. This is attributed to the d ecrease in the supply of inert nitrogen gas. Nitrogen is initially s upplied to suppress cataly st sintering. Usually at around 24 hrs, the catalyst bed is filled with liquid product whic h significantly improves the heat transfer across and along the bed and diminishes the risk of sintering. Another shift in conversion is observed around 40-45 hrs, this is due to the adjustment in CO/H2 ratio from 1/3 to 1/2 as required by stoichiometry. It is interesting to note that the gradual rise for the first 2024 hrs is more than that observed at 40-45 hrs. This can be due to the filling of pores with hydrocarbon liquid which blocks the active si tes preventing furthe r hydrogenation [31]. Sample 4 shows anomalous behavior during the first 24 hrs as conversion essentially remained constant. In general activity follows the adsorption trend shown in Figure 11 i.e. Sample 3 > Sample 2 > Sample 4 > Sample 1. The impact of metal support inte raction (MSI) can be correlat ed with activity. Sample 2 exhibited the presence of cobalt silicates dur ing the time of its preparation. Also during actual runs silicates form in the presence of wa ter, so their concentration increases as the run time progresses. These silicates are respon sible for the drop in activity of Sample 2 after 60 hrs of continuous operation. On the other hand, samples 1, 3 and 4 shows relative constant activities. For samples 3 and 4 one can attribute the sustained activity to the in situ treatment in hydrogen. Table 9. Catalytic performance of cobalt-s ilica supported catalysts in FT synthesis Reaction condition 2 MPa, 503 K, WHSV = 94 h-1 Activity: Gram hydrocarbon product/ gram catalyst used per 1 h operation Catalyst Activity CO conv. Selectivity (%) ID hr-1 (%) C1-4 CO2 C5+ Sample 1 1.3 55.0 38.2 11.6 50.1 Sample 2 2.5 73.0 31.1 8.2 60.0 Sample 3 2.8 85.0 27.8 9.2 63.0 Sample 4 1.8 63.0 27.4 15 57.5

PAGE 53

44 Figure 14. CO conversion with time. Reacti on conditions: 2 MPa, 503 K, WHSV= 94 h-1, FTS time: 80 h

PAGE 54

45 6.2 Selectivity Assessment Using Research Grade Gases The interest in an eggshell profile for FTS has developed due to an ever growing demand for a product specific catalyst and redu ced metal loading. Such a catalyst will tremendously reduce the operating cost which would otherwise be required for additional refinement. The product selectivity is fine tuned using a so called “ ” parameter (discussed earlier). This parameter has a st rong bearing on the activity of the catalyst as well. As shown in Table 4, the calculated th ickness of the egg shell (based on “ ”) depends on the surface concentration of cobalt crystallites. Sample 1, due to lesser per m2 active metal concentration, requires thicker shell (for a given loading) to ensure the availability of “a sufficient amount” of sites for ch ain growth and activity. However, carbon monoxide being the diffusion limited reactant ma y not effectively reach the sites located deep inside the pores, resulting in chain termination at low car bon numbers. This is confirmed by the shift in GC profile of Sample 1 towards lower hydrocarbons. Sample 3 has the highest dispersion and its shell thickness lies around the mean of the desired limit. All these parameters provide a suitabl e condition for tight control on hydrocarbon product distribution. This is c onfirmed by the narrow GC prof ile within the diesel and aviation fuel range. For samples 2 and 4 the thickness of the shell is closer to the lower limit. This favors the formation of longer hydrocarbon chains. As compare to Sample 2, Sample 4 has lesser dispersion; as a result products shift toward s higher hydrocarbons as seen in Table 4. Based on the above discussion, the selectivity of the product depends on the thickness of the eggshell, which in itself is a function of active metal disper sion. This kind of correlation is of immense importance in modern catalysis, a product sp ecific synthesis is highly beneficial especially for the devel opment of alternate en ergy industry e.g. conversion biomass to liquid fuels. If the refini ng cost of the fuel is reduced, the alternate energy route becomes economically more feasible.

PAGE 55

46 6.3 Selectivity Assessment Using Biomass Derived Synthesis Gas The performance of Sample 3 has also been a ssessed with biomass derived synthesis gas. This gas has been produced by a “Pearson Biomass Reformer”, a tubular free flowing hollow reactor which uses crushed pine chips as the primary feed stock. The gasifier outlet gas has the following composition; 30% H2, 30% CO, 30% CH4 and 10 % CO2, it is supplied in pressured cylinders for the liquefaction process. The FTS reaction was carried under normal c onditions, however, the additional hydrogen was provided in order to bring CO and H2 ratio to 1:2, The hydrocarbon product is essentially free of sulfur (around 0.5 ppm). This clean green fuel has strong market demand especially in the aviation industry. The produced fuel has a cloud point around 40 oC, which is much lower than conventiona l biodiesel, and the gr avimetric energy is also higher than ethanol making it a preferential alternate fuel. It is important to note that Sample 3 is tunable based on its eggshell thic kness, thus it can meet customer specific demands based on ever-changing market conditions.

PAGE 56

47 Figure 15. GC analysis of liquid prod uct obtained from FTS of Sample 1 Figure 16. GC analysis of liquid prod uct obtained from FTS of Sample 2 5.00 10.0015.0020.00 25.0030.0035.00 40.00 Time C28 C24 C20 C17 C16 C15 C14 C12 C11 C7 C5 Arbitrar y Units 5.0010.0015.0020.0025.0030.0035.0040.00 Time C28 C24 C20 C18 C17 C15 C12 C13 C14 C10 Arbitrar y Units

PAGE 57

48 Figure 17. GC analysis of liquid prod uct obtained from FTS of Sample 3 Figure 18. GC analysis of liquid prod uct obtained from FTS of Sample 4 Time 5.0010.00 15.00 20.00 25.00 30.0035.0040.00 C28 C24C20 C18 C17C16 C14 C15 C12 C7 Arbitrar y Units 5.0010.00 15.00 20.00 25.00 30.0035.0040.00 Time C24C20 C18 C17 C16 C15 C14 C13 C12 C11 C10 C7 Arbitrar y Units

PAGE 58

F F igure 19. R a Figure 20. ( a ndom sam p ( a) Biomass p les of FTS l derived liq u Arbitrar y Units 49 l iquid produ c u id fuel sam p 5.0010. 0 C 8 c t produced p le, (b) resu l 0 015.002 C1 3 C10 C9 8 T from the de v l t of third pa r 0.0025.00 C20C18 C16 3 T ime v eloped cat a r ty GC anal y 30.0035. 0 C24 a lysts y sis 0 040.00

PAGE 59

50 Chapter 7: Discussion of Experimental Findings 7.1 Metal Deposition in an Aqueous Medium Under ambient conditions, a silica surface is covered with immobilized single or multiple layers of water molecules. This is due to the strong affinity of different types of silanol groups towards water molecules via hydrogen bondi ng. It is very difficult to have a direct interaction between the silica su rface and a foreign molecule in the presence of this layer. Therefore, in this study the silica gel pellets were he ated to around 500 oC, under this condition the silica surface loses all the water of hydration and approximately half of its silanol groups. This thermally treated silica gel po sses both the hydrophobic and the hydrophilic properties. Thus, upon soaking in the non-po lar n-heptane, thes e pellets absorb a sufficient amount of this solvent. After dr ying, these pellets were transferred to an aqueous salt solution; both the water and th e salt compete for the at tachment to silica surface. At the start, the pH of the aque ous paste (salt & water; 1 g/mL) is around 3.0 which is closer to the silica ge l’s isoelectric point. This ne utral surface has strong affinity for water and forms an immobilized aqua layer. There is no direct contact between salt and the silica surface. The penetration depth of the solution within the pores depends on the availability of the dried surface area essentially free of n-heptane. The dropwise addition of aqueous urea soluti on in the cobalt nitrate solution at 95 oC during precipitation increases the pH to 5 via the following decomposition reaction [26] CO (NH2)2 NH3 + HNCO

PAGE 60

51 In an acidic aqueous environment HNCO is converted to CO2, and ammonia reacts with proton to give NH4 + [26]. HNCO + H+ + H2O NH4 + + CO2 NH3 + H+ NH4 + Both these reactions consume H+ which leads to an increase in the pH. Under such conditions Co2+ forms hydroxide [26]. If the surface has exposed silanol groups these hydroxides undergo an ion exchange with the surface and creating direct contact between SiOand Co2+. However, due to the low pH at the start, these silanol groups are covered with water before any ion exch ange can take place. So the c obalt will adsorb above this stagnant water layer in the form of CoX (OH)Y (NO3)Z .w(H2O) [26]. 7.2 Metal Deposition in a Non-Aqueous Medium Since the initial treatment of the silica gel is the same as it was in the case of aqueous solution, the same theory applies here. However, the interaction of et hanol with the silica gel at the time of immersion is very different than that of water. As discussed earlier roughly half of the available silanol groups w ill be covered by the alcohol leaving the rest for the solute ions. The initial pH of the cobalt nitrate-ethanol system is around 2.5 and contains 3% water, by weight. This amount of water is sufficient for the ionization of the cobalt nitrate, but, the sili ca surface does not ha ve a charge. Under this condition these cobalt ions will swarm near the silica surface (kept in thermal motion by phonons) without any direct contact. The drop wise addition of the urea-ethanol so lution in the cobalt nitrate solution at 70 oC will results in partial thermal decomposition of urea [26] as shown below. CO (NH2)2 NH3 + HNCO However, due to the presence of ethanol, some ethyl carbamate will also form via following mechanism [59]

PAGE 61

52 CO (NH2)2 + CH3CH2OH CH3CH2OCONH2 + NH3 In the presence of water HNCO is converted to CO2, and ammonia reacts with proton to give NH4 + [26]. HNCO + H+ + H2O NH4 + + CO2 NH3 + H+ NH4 + These reactions increase the pH of the system to ~ 4.0. Due to this ri se in pH, the silica surface becomes negatively charged and unde rgoes coordination covalent bonding with the swarming cobalt ions. However, only a fr action of the available silanol groups will ionize; this promotes dispersion and avoids agglomeration. This bonded cobalt-silica structure serve as the base for a highly disp ersed building block of metal crystallites. 7.3 Calcination in the Stagnant Air Calcination in stagnant air is a complex pr ocess that involves th e decomposition of the salt precursor. This decomposition pathwa y depends on the calci nation media and the temperature. Sietsma et al. [33] have proposed the decomposition of Ni3(NO3)2(OH)4 at 450 oC in terms of Ni(NO3)2. According to Sietsma, the overall process is explained by the following reaction. Ni (NO3)2 (s) NiO (s) + 2 NO2 (g) + O2 (g) Based on these finding, the following decomp osition reaction is proposed for the air calcined cobalt samples. 3 CoX (OH)Y (NO3)Z x Co3O4 + 3z NO2 + 3/2 y H2O + (3/2 z +3/4 y-2x) O2 No CoO peak is observed by XPS or XRD an alysis. Also TPR analysis has shown two reduction peaks assigned Co2+ and Co3+. All these findings c onfirm that only Co3O4 is formed during the calcination process.

PAGE 62

53 In stagnant moist air this water remains in contact with the surface and reacts back to form cobalt hydroxide. As disc ussed earlier, this cobalt h ydroxide further reacts with migrating silica to form cobalt-silicate. This hypothesis is confirme d by the FTIR spectra of Sample 2. However, no silicates are observe d in the spectrum of Sample 1. This is due to the initial deposition of the salt in the sta gnant aqua layer. In th e absence of a strong interaction with the silica surface, the meta l precursors are loosely absorbed on the immobilized water layer, very close to each other. Decomposition in air is accompanied by a rapid thermal surge. This heat flux prom otes the agglomeration of cobalt crystallites which are already under thermal motion and in close proximity. No significant surface rearrangement is observed in this case. 7.4 Calcination in the Dynamic Hydrogen Calcination-reduction is carried out in the hydrogen gas is a gradual and slow process with uniform heat transfer. Si nce dried catalysts are direc tly subjected to the hydrogen treatment, they will first undergo the d ecomposition reaction. Sietsma et al. [33] identified that treatment of Ni(NO3)2 in an oxygen scavenging atmosphere They have not observation the formation of oxygen during this reaction. Since hydrogen is also an oxygen scavenger, we can deduce a parall elism between the two mechanisms. The decomposition of Ni(NO3)2 proceeds as follows [33] Ni(NO3)2(s) + NO NiO(s) + 3NO2(g) Based on this reaction, the following decom position reaction is proposed for the cobalt nitrate precursor developed in an aqueous solution. 3CoX (OH)Y (NO3)Z + (-4x+3/2y+3z) H2 xCo3O4+3z NO2 + (-4x+3z+3y) H2O However, the catalyst precursor synthesi zed in ethanol may contain some unionized cobalt nitrate. If such is the case deco mposition in hydrogen will proceed as follows

PAGE 63

54 3Co (NO3)2 + 2H2 Co3O4 + 6NO2 + 2H2O Co3O4 further reduces to metallic cobalt via the two step reduction process, explained in the TPR analysis. The two small peaks in th e TPR analysis of Sample 3 confirm the formation of Co3O4. However, it cannot be conclusi vely stated whether any CoO is formed during the decomposition or not. Since the decomposition in dynamic hydrogen leads to the formation of water, space velocity of the gas mu st be kept high. This avoids the formation of irreducible cobalt silicate. Due to the non-availability of complete data for CoX (OH)Y (NO3)Z complex, one cannot compare the heat of formation data for the tw o calcination processes. This data can reveal the amount of heat a surface absorbs during the decomposition and is suggested for future work. It is suspected that in the stagnant ai r, the surface absorbs more heat, than in the case of hydrogen. This added h eat contributes to crystal growth and agglomeration. Treatment in hydrogen is suspected of causi ng some surface rearrangements. This is confirmed by the FTIR spectrum of the dried catalyst and the spectrum of the Sample 3 (Figure 13). The dried catalyst (developed in a non-aqueous solution) shows the presence of cobalt-silica interaction. However, when the catalyst is treated in the hydrogen, this interaction is unobservable. This may be the result of hydrogen inco rporation within the silica structure or some other form of rea rrangement. In order to confirm this hypothesis we need sophisticated charac terization techniques such as scanning tunneling microscopy and/or neutron scattering is required.

PAGE 64

55 Chapter 8: Conclusions A highly active and selective FTS Co/SiO2 eggshell catalyst has been developed using hydrophobic-hydrophilic interactions between n-heptane and et hanol (or water). Active Co metal has been impregnated in the si lica support either under ambient conditions using aqueous cobalt nitrate solution or under nitrogen atmos phere using cobalt nitratealcohol solution. Synthesis of an active catalyst requires a ba lance of strong interaction between the active metal and support withou t formation of irreducible mixed metal support oxides. Calcination atmosphere also impacts the final redi stribution of active metal on support. Presence of water in the calcination en vironment enhances metalsupport interaction. Silanol gr oups on silica can alter mo rphology and dispersion of active metals on the support. Solvents used for precursor such as water or alcohol attach to these silanol sites in specific configuration and compete with metal salts during ion exchange and adsorption. By fine tuning the solvent attachments on h eat treated silica we have fabricated a cobalt/silica catalyst with high dispersion. Silica has affinity for both polar and non-polar molecule depending on the surface c onditions. This property is exploited in preparing an eggshell prof ile. Simultaneous calcinations/reduction in dynamic hydrogen environment was used to further enhance the dispersion and reducibility. The parameter “ ” from previous work [37] is adopted in order to evaluate catalyst performance. Catalyst samples synthesized in water did no t show significant meta l-support interactions and when calcined in air, undergo significant sintering. This sintering reduces their active surface area and dispersion, and the average cr ystal size increases significantly. The samples synthesized in ethanol show consider able cobalt-silica interactions during the initial loading. When thermally treated in st agnant air, irreducible cobalt silicates are formed in these samples. The formation of sili cates significantly decreases their degree of

PAGE 65

56 reduction. The ethanol, stagnant air samples exhibited an increased dispersion and relatively smaller crystal size than the water, stagnant air samples. On the other hand, thermal treatment of these non-aqueous samples in a dynamic hydrogen environment produces very fine crystallites and no irreducible cobaltsilicates. For a given loading, the thickness of the e ggshell required for op timum selectivity and activity is dependent on the dispersion of th e active metal.. The lower the dispersion, the greater the thickness required in order to ensu re the availability of requisite sites (for high activity and desired selectivity). On th e other hand, a highly dispersed catalyst achieves the required activity and sel ectivity within a narrow eggshell. Eggshell thickness has a direct impact on pr oduct selectivity. Carbon monoxide is the diffusion limited reactant and its dissociation is the rate limiting step in Fischer-Tropsch Synthesis [3]. By exploiting this fact a nd synthesizing a catalytic system with low dispersion and a thick eggshell, the diffu sion limited arrival of carbon monoxide is reduced to all active sites. This results in an early termination of hydrocarbon chain growth in the range of gasoline and avia tion fuel. While, a thinner eggshell and high concentration of active sites promotes chain growth. However, due to the selected range of “ ” the overall distribution lies within na rrow hydrocarbon cut. The activity of a catalyst depends on its dispersion and reducibi lity. The change in solvent and calcination atmosphere varies the catalyst activity. Howeve r, the relative impact of solvent is more than that of calcination atmosphere. The catal yst activity pattern para llels the dispersion, i.e. high dispersion gives more conversion. Pr oduct selectivity has been controlled by eggshell thickness; current focus is in the production of diesel and aviation fuel. Gas chromatograph analyses of the resulting li quid fuel products show a very narrow hydrocarbon distribution

PAGE 66

57 References 1. World Energy Outlook 2004. Retrieved from http://www.iea.org/textbase/nppdf /free /2004/weo2004.pdf International Energy Agency, Paris, 2004, p. 29. 2. A. Y. Khodakov, Braz. J. Phys. 39, 171 (2009). 3. T. Bligaard et al., J. Catal 224, 206 (2004). 4. A. Y. Khodakov, W. Chu, P. Fongarland, Chem. Rev. 107, 1692 (2007). 5. E.Iglesia, S. C. Reyes, R. J. Madon, S. L. Soled, Adv. Catal. 39, 221 (1993). 6. E. van steen et al., J. Catal. 162, 220 (1996). 7. S. W. Ho, M. Houalla, D. M. Hercules, J. Phys. Chem. 94, 936 (1990). 8. B. Earnest, S. Libs, P. Cahumette, A. Kiennemann, Appl. Catal. A 186, 145 (1999). 9. W. Stober, Beitr. Silikose-Forsch. H-89, 87 (1966). 10. D. E. Meyers, N. Hackerman, J. Phys. Chem 70, 2077 (1966). 11. M. M. Dubinin, B. P. Bering, V. V. Serpinski, Recent Progress in Surface Science Academic, New York, 1964, p. 42. 12. L. D. Belyakova, O. M. Dzhigit, A. V. Kiselev, Zh. Fiz. Khim 31, 1577 (1957). 13. K. R. Lange, J. Colloid. Sci. 20, 231 (1965). 14. R. L. Dalton, R. K. Iler, J. Phys. Chem. 60, 995 (1956). 15. K. Kleir, A. C. Zettlemoyer, J. Colloid Interface Sci 58, 216 (1977). 16. J. H. De Boer, J. M. Vleeskens, K. Ned. Akad. Wet. Proc. Ser. B 60, 23 (1957). 17. C. G. Armistead, J. A. Hockey, Trans. Faraday Soc 48, 58 (1974). 18. G. J. Young, J. Colloid Sci. 13, 67 (1958). 19. M. L. Hair, W. Hertl, J. Phys. Chem. 73, 4269 (1969). 20. C. Clark-Monks, B. Ellis, J. Colloid Interface Sci 44, 37 (1973). 21. L. Robert, C. R. Acad. Sci. 234, 2066 (1952). 22. R. K. Iler, The Chemistry of Silica John Wiley and Sons, New York, 1979, p. 689 23. R. K. Iler, The Chemistry of Silica John Wiley and Sons, New York, 1979, p. 690

PAGE 67

58 24. R. K. Iler, The Chemistry of Silica John Wiley and Sons, New York, 1979, p. 663 25. T. W. Healy, R. O. James, R. Cooper, Advance in Chemistry Series 79, American Chemical Society, Washington, D.C., 1968, p. 62. 26. M. Rajmathi, P. V. Kamath, Int. J. Inorg. Mater. 3, 901 (2001). 27. R. Matmaan et al., J. Phys. Chem. 72, 97 (1968). 28. D. L. Dugger et al., J. Phys. Chem. 68, 757 (1964). 29. O. Borg et al., Top. Catal. 45, 39 (2007). 30. Puskas et al., Appl. Catal. A. 311, 146 (2006). 31. A. H. Kababji, B. Jospeh, J. T. Wolan, Catal. Lett. 130, 72 (2009). 32. B. Linyang et al., Catal. Commun. 10, 2013 (2009). 33. J. R. A. Sietsma et al., J. Catal. 260, 227 (2008). 34. K. E. Coulter, A. G. Sault, J. Catal 154, 56 (1995). 35. J. M. Jablonski, M. Wolcyrz, L. Krajczyk, J. Catal. 173, 530 (1998). 36. Z. Tao et al., Catal. Lett., 117, 130 (2007). 37. E. Iglesia et al., J. Catal. 153, 108 (1995). 38. M. F. Post et al., AIChE J 35, 1107 (1989). 39. E. Peluso et al., Chem. Eng. Sci. 56, 1239 (2001). 40. E. Iglesia et al., Top. Catal. 2, 17 (2005). 41. M. Nele et al., Appl. Catal. 178, 177 (1999). 42. M. K. Niemela et al., Top. Catal. 2, 45 (1995). 43. C. H. Bartholomew, R. J. Farrauto, Fundamental of Industrial Catalytic Processes, John Wiley and Sons New Jersey, 2006, p. 98. 44. C. H. Bartholomew, R. J. Farrauto, Fundamental of Industrial Catalytic Processes, John Wiley and Sons New Jersey, 2006, p. 95. 45. J. F. Moulder, W. F. Stickle, P. E. Sobol, K. D. Bomben, Handbook of X-ray Photoelectron Spectroscopy Physical Electronic In c, Eden Prairie, 1992, p. 83. 46. J. S. Giradon et al., J. Catal. 248, 143 (2007). 47. D. Song, J. Lee, J. mol. Catal. A 247, 206 (2006). 48. R. C. Reuel, C. H. Bartholomew, J. Catal. 85, 63 (1984). 49. C. H. Bartholomew, Catal. Lett. 7, 27 (1990).

PAGE 68

59 50. C. H. Bartholomew, R. J. Farrauto, Fundamental of Industrial Catalytic Processes, John Wiley and Sons New Jersey, 2006, p. 83. 51. D. Y. Murzin, Chem. Eng. Sci. 64, 1046 (2009). 52. A. M. Hilmen, D. Schanke, A. Holmen, Catal. Lett. 38, 143 (1996). 53. D. G. Castner, P. R. Watson, I. Y. Chan, J. Phys. Chem. 94, 819 (1990). 54. L. Zeng-xi et al., Chin J. Process Eng 6, 656 (2006). 55. I. Puskas et al., Appl. Catal. A 316, 197 (2007). 56. NIST Standard Reference Database Number 69 Retrieved from http://webbook. nist. gov/ chemistry/. 57. H. Gunzler, G. Han-Ulrich, IR Spectroscopy Wiley-VCH, Weinheim, 2002. 58. Reference Library XPert HighScore, PANanlytical B. V, Almelo. 59. D. Wang, D.Yang, X. Zhai, L. Zhou, Fuel Process. Technol 88, 807 (2007).