<%BANNER%>

Design and Implementation of an Optical-Access Steam Methanol Reformer


PAGE 1

DESIGN AND IMPLEMENTATION OF AN OPTICAL-ACCESS STEAM METHANOL REFORMER By AMY L. TWINING A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2005 by Amy L. Twining

PAGE 3

To my parents.

PAGE 4

iv ACKNOWLEDGMENTS I have been extraordinarily lucky to work for Dr. David Hahn, an outstanding scientist, and an advisor of amazing op timism and generosity. Dr. Hahns philosophy towards experimental design fostered the i nnovation and creativity ne eded to develop a new means of quantifying certain effects in catalysis within stri ct budget constraints. Completion of this work would have been impossible without his help and encouragement. As a mentor, his guidance assisted me through some very trying circumstances. After seeing the issues inhere nt in the culture and framework of industry, having worked for him is likely to be the mo st heartening and influential experience of my career. It has been a privilege to be his student, and he has my utmost admiration and gratitude. I have also to thank Dr. Vernon Roan fo r giving me this opportunity, and for being my first advisor at the University of Fl orida. Research conducted by his lab group reinforced the interconnections among rese arch, industry, and policy, as well as the importance of maintaining balanced and objective scientific opinions. I thank my committee members, Dr. Willia m Lear and Dr. David Mikolaitis, for giving their time and advice on this work. I also thank Robert Schutt, my former boss at Honeywell, and Dr. Aaron Byerley, at the United States Air Force Academy, fo r believing in my capabilities, for their exemplary integrity, and for enabling me to pursue a graduate degree.

PAGE 5

v I thank my classmates, Tim Simmons, Daniel Betts, Jessica Rozzi-Ochs, and Patricia Alexander for being my friends. Most especially, I thank my parents.

PAGE 6

vi TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...........................................................................................................viii LIST OF FIGURES...........................................................................................................ix ABSTRACT....................................................................................................................... xi CHAPTER 1 OVERVIEW............................................................................................................1 1.1 Why Study Hydrogen Production fr om Hydrocarbons such as Methanol?....................................................................................................2 1.2 Overview of Catalyst Development.............................................................4 1.2.1 How the Catalyst Works..................................................................6 1.2.1.1 Adsorption............................................................................8 1.2.1.2 Surface reactions and deactivation.....................................10 1.2.2 Rate Laws, Mechanisms, and the Role of Constituents.................14 1.2.2.1 Contributions of Cu (111), ZnO and Al2O3 to activity, stability, and selectivity.....................................................17 1.2.2.2 Proposed reaction mechanisms and rate laws....................21 1.3 Heat Transfer Limitations in Pack ed Bed Reformers, Advantages in Monolith Reformers...................................................................................32 1.3.1 Non-Isothermality in Packed Bed Reactors...................................33 1.3.2 Monolith Reformers.......................................................................35 1.4 Raman Spectroscopy for Study of Catalysis..............................................39 2 RAMAN SPECTROSCOPY.................................................................................41 3 REACTOR DESCRIPTION..................................................................................45 3.1 Reactor.......................................................................................................45 3.2 Optical Configuration................................................................................47 4 EXPERIMENTAL METHODS.............................................................................51 4.1 Catalyst Bed Reduction..............................................................................51

PAGE 7

vii 4.2 Reformer Operation...................................................................................53 4.3 Reformer Operating Procedure..................................................................54 5 SPATIAL RESOLUTION.....................................................................................57 5.1 Spatial Resolution of the Optical Set-Up...................................................57 5.2 Calibration for Gas Concentration.............................................................59 6 DATA....................................................................................................................64 6.1 Variation of Species Concentrati on at Each Position, at Flow 1...............65 6.2 Variation of Species Concentrati on at Each Position, at Flow 2...............67 6.3 Variation of Species Concentra tion along Length and Height above Catalyst at Flow 1......................................................................................67 6.4 Variation of Species Concentra tion along Length and Height above Catalyst at Flow 2......................................................................................70 6.5 Comparison of Concentration Grad ients at Flow 1 and Flow 2................72 6.6 Species Gradients along Reactor Leng th and Height above the Catalyst..74 6.7 Species Concentration vs Position at Height 1..........................................80 7 SUMMARY AND CONCLUSIONS....................................................................83 8 RECOMMENDATIONS FOR FUTURE WORK................................................85 APPENDIX DATA SUMMARIES..................................................................................87 LIST OF REFERENCES...................................................................................................91 BIOGRAPHICAL SKETCH.............................................................................................98

PAGE 8

viii LIST OF TABLES Table page 1-1 Summary of Proposed Rate Laws............................................................................28 1-2 Summary of Steam Meth anol Reformer Studies......................................................29 2-1 Raman Shift..............................................................................................................43 3-1 Equipment and Specifications..................................................................................50 4-1 Reactant Flow Rates and Pump Settings..................................................................53 4-2 Flow Parameters.......................................................................................................54 A-1 Hydrogen Data Summary.........................................................................................88 A-2 Methanol Data Summary.........................................................................................89 A-3 Water Data Summary...............................................................................................90

PAGE 9

ix LIST OF FIGURES Figure page 1-1 Dehydration of Methanol on a Catalyst Surface....................................................12 1-2 Portions of Atoms on FaceCentered Cubic Crystal Faces....................................18 1-3 Example of a micro-reformer, adapted from (Pan et al)........................................38 2-1 Typical Raman Spectrum of Steam Methanol Reforming.....................................44 3-1 End View of Pellet Bed..........................................................................................45 3-2 Top View of Pellet Bed..........................................................................................46 3-3 Reformer assembly................................................................................................47 3-4 Experimental Set-Up for Raman Specrtroscopy....................................................49 4-1 Catalyst Reduction Raman Spectra........................................................................52 4-2 Water and Hydrogen Signal vs Time During Reduction.......................................53 5-1 End View of Reformer...........................................................................................57 5-2 Side View of Reformer..........................................................................................58 5-3 Spatial Resolution..................................................................................................59 5-4 Temperature Independence of Hydrogen Signal...................................................60 5.5 Hydrogen Calibration Curve..................................................................................61 5-6 Methanol Calibration Curve..................................................................................62 5-7 Water Calibration Curve........................................................................................63 6-1 Typical Raman Spectra Observed Along the Axis of the Reactor.........................65 6-2 Position 1, Flow 1 Species Concentration vs Height.............................................66 6.3 Position 2, Flow 2 Species Concentration vs Height.............................................66

PAGE 10

x 6-4 Position 3, Flow 3 Species Concentration vs Height.............................................67 6-5 Position 1, Flow 2 Species Concentration vs Height.............................................68 6-6 Position 2, Flow 2 Species Concentration vs Height.............................................68 6-7 Position 3, Flow 2 Species Concentration vs Height.............................................69 6-8 Flow 1 Methanol Concentra tion at 3 Positions vs Height.....................................70 6-9 Flow 1 Hydrogen Concentrati on at 3 Positions vs Height.....................................71 6-10 Flow 1 Water Concentration at 3 Positions vs Height...........................................71 6-11 Flow 2 Hydrogen Concentrati on at 3 Positions vs Height.....................................72 6-12 Flow 2 Methanol Concentra tion at 3 Positions vs. Height....................................73 6-13 Flow 2 Water Concentration at 3 Positions vs Height...........................................73 6-14 Hydrogen and Methanol, Flows 1 and 2, Position 1..............................................75 6-15 Hydrogen and Water, Flows 1 and 2, Position 1...................................................75 6-16 Hydrogen and Methanol, Flows 1 and 2, Position 2..............................................76 6-17 Hydrogen and Water, Flows 1 and 2, Position 2...................................................76 6-18 Hydrogen and Methanol, Flows 1 and 2, Position 3..............................................77 6-19 Hydrogen and Water, Flows 1 and 2, Position 3...................................................77 6-20 Hydrogen Flow 1...................................................................................................78 6-21 Methanol Flow 1....................................................................................................78 6-22 Water, Flow 1.........................................................................................................79 6-23 Hydrogen Flow 2...................................................................................................79 6-24 Methanol Flow 2....................................................................................................80 6-25 Water, Flow 2.........................................................................................................80 6-26 Hydrogen, Flow 1 and 2.........................................................................................81 6-27 Water, Flow 1 and 2...............................................................................................82 6-28 Methanol, Flow 1 and 2.........................................................................................82

PAGE 11

xi Abstract of Thesis Presen ted to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science DESIGN AND IMPLEMENTATION OF AN OPTICAL-ACCESS STEAM METHANOL REFORMER By Amy L. Twining May 2006 Chair: David W. Hahn Major Department: Mechanic al and Aerospace Engineering This study is an assessment of the feasib ility of Raman spectroscopic analysis of gas phase species in methanol steam reformi ng over a catalyst surface. Included in this work is a comprehensive literature review of control volume and surface science studies of steam methanol reforming, as well as re action mechanisms proposed in these studies, and sources of error in their measurement. Advantages of in situ study of a monolith reformer are discussed. Experimental data reveal fluid dynamic aspects of the reacting flow, specifically, the scale of the boundary layer and concen tration gradients as related to the flow rate. Recommendations for fu ture work are proposed, highlighting the successful use of Raman spectroscopy in st udy of gas-phase species, as well as the advantages to measuring directly the change s in concentration over time for accurate determination of rate laws.

PAGE 12

1 CHAPTER 1 OVERVIEW Fuel cell technology, which relies on gaseous hydrogen and oxygen, is in development as a more efficient and cleaner method of power generation. Considering that the transport of gase ous hydrogen adds significantly to its cost, many fuel cell applications are being developed to incor porate the reformation of more energy-dense liquid fuels. One such fuel, methanol has the best hydrogen to carbon ratio. Unfortunately, even 20 parts per million carbon monoxide can poison fuel cells (Purnama et al, 2004), so there is an effort to devel op more efficient reformers as well as reformate clean-up and separation tec hnology. Traditional packed bed reformers have shown diffusion limitations, nonisothermality, and pressu re gradients that ar e difficult to control and lead to inefficiencies in reform ing. For hydrogen production, several monolith reformers are in development, but none ar e currently available commercially. The reformer in this investigation se rves to provide a means of studying in-situ the gas phase and to determine two-dimensional scale f actors of steam methanol reforming over a catalytic surface, rather than through a packed bed, in order to aid in future reformer system design. The reformer uses Synetix 33-5 steam methanol reforming catalyst pellets closely packed in a single layer in an alum inum tray, which is situated within the reformer so that reactant gases pass over it There are three window s along the length of the reactor for in situ evaluation of gas phase composition via infrared Raman spectroscopy. The reactor itself can be raised or lowered; and c oncentration gradients were evaluated along length and height above the catalyst. Qualitative relationships

PAGE 13

2 between gas phase composition, flow rate, a nd reactor length were determined. In the future, other modified reforming reactions ma y be studied over the same type catalyst (combined SMR/POX), which are more thermo dynamically efficient. If adhesion of catalyst material can be achieved, another ca talyst tray may be configured above the original, to study the gas phase between two catalytic surfaces, possibly with a variable distance separating them. This Poisiuelle -type flow configuration would aid in determining the critical reactor lengths and diameters for fully developed plug flow, as well as the length and flow rates at which pr oduct gases begin to cau se reverse reactions. Such a configuration will allow, as closely as possible, in-situ study of a monolith reformer while minimally di sturbing reactor conditions. 1.1 Why Study Hydrogen Production from Hydrocarbons such as Methanol? The goal of fuel cell technology, incl uding hydrogen production from various feedstocks, is to provide more efficient power that emits fewer unrecoverable byproducts within the real economic constraints of limite d natural resources and the current state of technological development. Various greenhouse gas models provide a spectrum of the timescale and severity of impact of a CO2 accumulation in the atmosphere (Litynski et al, 2005). Without debating these, more generall y, exhausted heat and materials represent wasted energy and resources. For portable device s, fuel cells can provi de better efficiency and energy density (Hu et al, 2003) than a conventional lithium ion battery. The energy density of methanol is 5.5 kWh/kg co mpared to 0.5-1.0 kWh/kg for stored H2, and an estimated 0.3 kWh/kg for a lithium-ion battery (Hu et al, 2003). The required efficiency of fuel cell systems to match lithium-ion batteries is 5.5% for a methanol-based fuel cell system and 30-60% for a stored-hydrogen fuel cell system (Hu et al, 2003). Fuel cells, which exhaust only heat and water, are considered the new hope for clean power

PAGE 14

3 generation. However, established methods of hydrogen production themselves have a by product of CO2. Designs that rely on non-hydrocarbon feedstock, such as solar-powered electrolysis for hydrogen producti on, are still prohibitively expensive and less efficient (Padro et al,1999). These concepts will likely not be economically competitive until they progress more technologically, and the cheaper carbon-based feedstoc ks become scarcer and more expensive. (Padro et al,1999; Roan et al, 2004) In the mean time, hydrogen production from readily available hydrocarbon feedstocks is being improved, with the means of containing and sequestering CO2 in concurrent development. Such examples include power generation from coal gasifi cation, with hydrogen co-production, and an experimental gasifier th at also incorporates CO2 separation (Rizeq et al, 2003; Kubek et al 2002; Chen et al, 2002). A variety of methods of hydrogen production and delivery will enable tailoring to particular applications using fuels that are most readily available in that locale, for example, coal gasification where natural gas is unavailable, and liquid fuel transport where there is not yet a pipeline infrastruc ture. Studies have shown that transport of gaseous hydrogen over land, as opposed to via pi peline, greatly increases the final price (Amos et al,1998; Roan et al, 2004). If the infrastructure is available, hydrogen transmitted through pipelines may be ideal for large stationary fuel cell power generation, such as domestic power. However, extensive pipeline installation is costly, and in the interim, truck transport of liquid fuel to smaller local reformers will have an advantage (Myers et al, 2002). For small or portable applications, a fuel cell system incorporating a liquid fuel reformer may be a better option than one using a large compressed gas tank, considering the energy density of liquid fuels compared to gaseous hydrogen (Ogden

PAGE 15

4 2002). A compact fuel cell system using metha nol, operating at approximately 5% overall efficiency, still has a higher energy density than a lithium polymer battery (Hu et al, 2003). In contrast, a fuel cell system usi ng stored hydrogen must operate at 30-60% efficiency to equal the energy density of L ithium ion batteries (Hu 2003). Fuel processors for fuel cell applications stil l face the difficulty of transient reaction kinetics, especially the time required to provide heat to begi n reforming (Springmann et al, 2004). Leak detection and boil off in compressed gas a nd liquid hydrogen transport methods, as well as flammability, provide additional complicatio ns to portable fuel cell systems. These circumstances indicate the usefulness in optim izing reformation of liquid fuels such as methanol. 1.2 Overview of Catalyst Development Composition and spatial variation in catal ysts affect possible reaction pathways, resulting in many observed rates and pr oposed mechanisms. The development of catalysts for steam methanol reforming bega n with a study of the reverse reaction of methanol synthesis from the hydrogenation of CO2. However, it quickly became apparent that, given the effect gas phase chemistry ha s on catalyst activity and changes in catalyst structure during adsorption, steam methanol re forming is not simply a reversal of the better known methanol synthesi s reaction (Peppley et al,199 9). Competitive adsorption, as well as changes in adsorptive energies with coverage (dAl noncourt et al, 2003), are factors that contribute to the fact that steam methanol reforming is not simply the reverse of methanol synthesis. For example, the he at of adsorption of CO on a copper catalyst varies by almost 30 kJ/mol between low and hi gher coverages, with the higher coverage reducing the heat of adsorption of additional CO. In order for a species to adsorb on the catalyst, valence electrons in the catalyst have the mobility to stabilize the valence

PAGE 16

5 electron shell in the adsorbate. This results in a change in the material in immediate proximity to the adsorbate, the cumulative e ffect of which is shown by the changes in adsorptive energy with catalyst coverage (a mount of adsorbate covering active sites). In this way, the catalyst is like an additional reactant, and its ac tivity will change depending on the gas phase composition. Though overall mechanisms are proposed over certain types of sites, it becomes clear in the literatur e that the chemistry of the catalyst itself plays a significant role. Varying rate laws and mechanisms whic h consider the copper site to be the active site for react ion have been proposed for c opper catalysts augmented with a variety of constituents. Proposed rate laws and kinetics also vary with catalysts made from the same metal oxides but in different proportions, and in different preparation methods; and there is even debate over mech anisms and rate determining steps over the same catalyst. The next sections disc uss how catalysts work, proposed reaction mechanisms seen in literature, and changes in catalyst structur e and activity during reactions. These chemistry issues are the bus iness of surface science, however they are discussed here in order to explain how the differences in the apparent mechanisms and deactivation rates observed are affected by reactor design, catalyst composition, and certain surface and geometric factors. The e xperiments done on this reactor will serve to qualitatively illustrate the relationship between reactant flow rate, conversion, the scale of the reaction in two dimensions: height above the catalytic surface, and along the length of the reactor. While this study used commercia l components to assess the feasibility of Raman spectroscopy for a gas phase study, a more refined experiment may quantify scale factors for future use in monolith reactor design.

PAGE 17

6 1.2.1 How the Catalyst Works A gas reaction on a catalytic surface diffe rs fundamentally from a homogeneous gas phase reaction. Even though the overall re action is the same, the interaction of adsorbed gas species with the catalyst results in a different mechanism than in a purely gas phase reaction. While heating the catal yst and gas phase reactants can overcome barriers to adsorption and r eaction, the behavior follows the Arrhenius law through a narrower range than in homogeneous, thoroughly mixed chemistry. This is due to the fact that not all the participants in the reaction are equally mobile, and unlike other reactants, the catalyst must be regenerated by the end of the reaction process. The addition of heat in a homogeneous gas phase reaction increase s the rate by increasing the number and energy of collisions, resulting in improved chance s of collisions that result in a reaction. Adding heat to a heterogeneously catalyzed reaction will increase the rate at which particles collide with the su rface, but the speed at which they dissociate and recombine on the surface may not have a direct propor tionality with temperature. Increasing the temperature of the catalyst can result in reactants becoming more strongly bound to the surface, which reduces their mobility across th e surface and subsequent rates of reaction with other adsorbates. When the catalyst b ecomes too hot, adsorbates can approach the activation energy at which they become pe rmanently bound to the surface (Masel 2001), but prior to this occurrence, there is a sh arp leveling-off of the reaction rate with temperature. The reason for this could be due to the reduced mobility of adsorbates as they become more strongly bound to the catal yst, but another cause may be that the catalyst material properties are beginning to change. Metal oxides in catalysis are less mobile in their microstructure, and therefore less subject to change with temperature, but this is not necessarily the case for a bulk metal used in a catalytic process. Thermal

PAGE 18

7 production of point defects in solids, gas phase reactions, gas reacti ons on solid surfaces, even reactions in liquid solutions, all follo w an Arrhenius-type law (Shackelford, 1992) within a certain temperature range. The temp erature restrictions on the catalyst arise out of the closeness in temperature of each of the chemical processes: the reaction of adsorbates on the surface, the process of creating unfavorably strong bonds between adsorbate and catalyst, and thermally restru cturing the catalyst (sintering). The temperature at which atomic mobility is suffi cient to affect mechanical properties is approximately one-third to one-half times the absolute melting point. (Shackelford 1992). The melting point of copper is 1357K, a nd one-third of that is about 452K, 180oC. This is essentially the heat treatment temperature of the metal; the temperature at which point defects can migrate and grain boundari es can begin to restructure. Reactor temperatures range from 200-300 oC; therefore a copper catalyst must be designed to be structurally stable at these temperatures. Ot her metal oxides, such as zinc and zirconium oxides, have been used to stabilize the c opper material. The copper atoms are less mobile in contact with the metal oxides, which aid in maintaining the structure of the copper. Given heat transfer problems that result in non-isothermality in packed bed reformers, it is easy for a hot-spot to occur that can approach the critical temperature for restructuring copper crystal face active sites on the catalyst, even with additives to stabilize the copper material. Diffusion of defects through copper can occur in the bulk, on the surface, and at grain boundaries, at respectiv ely decreasing temperatures and depending on the amount of the diffusing region that is availabl e (Shackelford 1992). The desired reaction temperature is an intermediate one that allows strong enough adsorption for dissociation and reaction, while being low enough that the catalyst is regenerated without non-

PAGE 19

8 reversible reactions occurring on the active site s. The reaction rate for steam methanol reforming levels off around 220oC (Shen et al, 1997) and begins to show non-Arrhenius behavior around this temperature (227oC) (Agrell et al 2003) Mass transfer limitations become significant at temp eratures higher than 220oC, (Agrell et al, 2002) while the reaction generally shows Arrheniu s-type behavior between 175oC and 220oC. Generally, catalysts work to lower the act ivation energy in chemical reactions by stabilizing intermediates and positioning reactants in more ideal proximity and orientation in order to improve the odds of a reaction occurring. (Masel 2001) A catalytic process comprises adsorption, either dissoci atively or molecularl y, followed by reaction on the surface, then desorption. For catalyt ic reactions in this study, only monolayer adsorption occurs. Multilayer adsorption is a condensation process and occurs in water condensation and coking or soot formation, when the attractive interactions between adsorbates are strong compared to kBT, near the boiling point. Catalytic activity depends on valence shell configurations, differs funda mentally between meta l crystals and ionic solids, and varies within metals in the same group on the periodic chart, though the details are not fully understood. 1.2.1.1 Adsorption The process of adsorption of a gas on a me tal surface occurs when there is a net polarization as the gas approach es the surface, and an interaction with the free electrons in the conduction band of the metal. Adsorp tion of gases on oxide surfaces has been described as an equalization of electronegativ ities (Masel 1996). Polar molecules, such as water, bond more easily to insulating oxides; and cova lent molecules, like carbon monoxide, bond more strongly to metals. Th e presence of empty orbitals and bulk electron states on a metal surface enable mob ility of valence electrons, especially d-band

PAGE 20

9 electrons, to stabilize the adso rbate. Adsorption is most stable when the highest occupied orbitals interact with the lowe st unoccupied orbitals, and an electron density at one end of a covalent molecule can cause that end to ad sorb on the metal surface. In metals, there are empty orbitals very close in energy to the Ferm i level (the top of the filled levels), so very little energy is needed to excite the uppermost electrons (Atkins 2002). In transition metals, d-band electrons in partially filled outer shells can shift easily and lend an electron to the outer shell of the reacta nts. This weak bond changes the shape and orientation of the antibonding orbital, allowing it to break more easily. D-band electron mobility can also help to lo wer Pauli repulsions, thereby aiding in bond scission, though they contribute minimally to the actu al heat of adsorption (Masel 1996). Bonding on insulating oxides is mainly ioni c. Since electron mobility is restricted in metal oxides, the surface is fairly inert a nd bonding is dominated by structural defects. Aluminas are considered Lewis acid sites, containing an empty orbital to accept a charge. Oxide doping changes the electronegativity of defects, and th erefore their ability to accept charge. Masel gives a complex process of adsorption of gases on ionic compounds. To begin with, the adsorbate reacts first with a defect, and creates an ion, which may then migrate to a defect-free regi on of the surface. The heat of adsorption on defect-free surfaces is related to ionic and pol arization forces; therefore, neutral or nonpolar species interact fairly weakly on an insulating oxide while polar species interact fairly strongly and create ionic adsorbates. The scale of surface roughness and defects can also affect adsorptive energies. This is true for metal oxides, as has been describe d earlier, as well as for adsorption on metal crystals. There is greater electron density on an inside corner, for example, and a

PAGE 21

10 molecule that adsorbs there can dissociate more easily since the material accommodates its three-dimensional shape. If a molecule is large compared to the scale of surface roughness, there could be an adverse effect since it may be more difficult for the atom most likely to bond to the catalyst to get cl ose enough to the active site. On the other hand, surface defects need to st abilize adsorbates in close enough proximity to facilitate their reaction with each other. Strain in crystal lattices has also shown to improve catalytic activity via a similar effect (G unter et al 2001). Moderate ranges of surface roughness seem to provide the most benefit (Masel 1996). 1.2.1.2 Surface reactions and deactivation Dehydration of methanol over a catalyst can illustrate the difference between gasphase and surface reaction mechanisms. This pro cess is depicted in Figure 1-1. In the gas phase, molecules must approach each ot her with the right amount of energy and orientation in order to break the correct bonds with the weakest bonds expected to break first. In catalysis, one particular end of a molecule will bond to the ca talyst site, causing a bond distortion and possible dissociation betw een the bonded atom and the rest of the molecule, but this may not be the same bond br eakage that initiates reaction in the gas phase. Methanol at first adsorbs weakly on th e metal, then it dissociates into H+ and methoxy, which are each more strongly bonded. Fo r methanol in a gas phase reaction, the C-O bond breaks first, since it is the weakest at 84 kcal/mol On the metal catalyst site, the oxygen in methanol bonds closest to th e catalyst, causing a dissociation of one hydrogen atom. This is because the hydroxyl gr oup is highly polar, compared to the CH3end of the molecule, so this end bonds to the metal. The net polarizat ion that occurs when the methanol molecule approaches the conducti on band in the metal is such that the bond to copper of the H+ and the methoxy ends of the molecule strain the O-H bond enough to

PAGE 22

11 break it. As a result, the H-O bond is distorte d and the first to brea k, even though it is the strongest bond (103 kcal/mol) in the gas pha se molecule. On the catalyst, the C-O bond remains intact, but the relative polarity of the hydrogen atoms with each of the carbon and oxygen atoms causes C-H and O-H bonds to di stort and hydrogen to dissociate first. Methanol dehydrogenates one hydrogen at a tim e on the catalyst surf ace, resulting in adsorbed hydrogen atoms and an adsorbed carbon monoxide molecule, which then desorbs. As noted earlier, adsorbed species can migrate on the catalyst surface, and the adsorbed hydrogens would combine to form H2 before desorbing. This is not necessarily the same path that methanol takes in the st eam reforming reaction. In fact, several models include steam methanol reforming, as well as methanol decomposition and water gas shift reactions, due to probable interactions of adsorbates on catalyst sites. Reactant species can adsorb on copper, as well as ZnO and Al2O3 constituents of the catalyst. Additionally, reaction pathways may occur over crystalline co pper metal, and some have been proposed over a special CuZnO site (Waugh 2004). Depending on the size of constituents, whether they are in regions large enough for several molecules to adsorb and react, or homogeneously distributed at a molecular level, ther e exist opportunities for more than one reaction pathway to occur at the same time. Adsorbed species can interact directly with each other, or an adsorbate can form a bond with an electron donated from the surface, changing the electronic properties of the surface. In this way, bonding of additional mol ecules has been shown to be affected by the coverage. As a result, it is possible for more than one pathway to take place on a catalyst, though proportions of additives to c opper catalysts have been shown to alter the selectivity for certain re action pathways, even though the (111) copper phase is

PAGE 23

12 considered the predominant site for cataly tic activity. These phenomena are reflected in some of the proposed mechanisms and inte rmediates to be discussed later. Figure 1-1. Dehydration of Methanol on a Catalyst Surface Surface reactions do not exactly follow the Arrhenius law, where Rate= Aexp (Ea/kBT), and A is a pre-exponential, Ea is the activation energy, kB is the Boltzmann constant, and T is the temp erature in Kelvin. Many reac tion models have evaluated a global Arrhenius-type relationship that is valid through a range of conditions. The addition of heat will initially overcome the activation energy of adsorption, but further heating can cause adsorbed in termediates to stabilize on the surface, or to break down on the surface, resulting in decrea sed catalytic activity and reac tion rate. For example, if the catalyst surface is hot enough that carbon monoxide breaks up into carbon and oxygen rather than desorbing, an active site has been removed from the reaction by oxidation, and carbon deposition can begin the process of deactivation (Argawal et al, 2004).

PAGE 24

13 Oxidized sites can be reduced again, but carbon is difficult to remove. Sometimes the catalyst can be heated with pure steam to react with the carbon and remove it, as has been done with cyclical steam methane refo rming (Choudhary, T.V. et al, 2000, 2001; Choudhary, V.R. et al 2001), but generall y carbon build up leads to coking and deactivation. In addition, reactions on catal yst surfaces are generally slower than gas phase collisions. Catalysts improve the ra te by lowering the ac tivation energy, putting molecules in proximity and creating stabili zed radicals, thereby improving the odds of reaction by avoiding non-productive collisions. As a result, heterogeneous reactions are spatially limited, and also have greater temper ature limitations than gas phase reactions. Elevated temperatures in steam methanol reforming generally do not contribute to carbon buildup, as they do in steam methan e reforming; however they can present problems in maintaining the catalyst struct ure. Fortunately, even at stoichiometric chemistry, the C-O bond in methanol steam refo rming generally stays intact and desorbs from the catalyst, so carbon buildup is not usually a problem. Since carbon monoxide has a triple bond, it takes much mo re energy to break it, and it is less likely to occur at the lower temperatures of steam methanol re forming. Even if reactants do not become permanently bonded to catalytic material, at elevated temperatures, sintering becomes a risk. This occurs when high temperature and pressure cause the dispersed copper to coalesce, and make copper masses that have less surface area and ar e less catalytically active. Sintering in copper can occur far belo w the melting point of the metal, as low as 300oC, and is related to vacancy diffusion, whic h is associated with the cohesive energy of the metal (Twigg et al, 2003). It ha s been observed at approximately 250 oC in water gas shift reactions over coppe r catalysts (Tanaka et al, 2003). Among catalyst materials,

PAGE 25

14 the copper is considered relatively unstable, and this is reflected in its low melting and point cohesive properties. Cha nges in metallic crystal structure also occur well below the melting point, for example, in the common solution heat treatments (quenching and tempering) of metals like steel and alumin um (Shackleford, 1992). When heat treating metals, stress concentrations occur at co rners, or in the case of small crystals, microscopic defects. A higher microscopic st rain can cause a stress that overcomes intermolecular forces, such that copper atoms will migrate to a more stable configuration, or, conversely, vacancy diffusion. Maintaining the structure of the (111) copper crystal face is important in catalytic activity because of the geometric factors that influence electron density and therefore adsorptive energi es. Defects of the right scale, on the right composition of catalyst, can improve catalytic activity, until the catalyst becomes so hot that those structures are not maintained. Catalysts can be poisoned by chemical impurities in the reactant feed, such as sulfur, which can occur in liquid fuels derive d from coal, and chlorine, which occurs in tap water. To combat this, some fuel cell sy stem designs incorporate guard beds that are separate from the reactor so that they can be changed. For example, ZnO can remove sulfur from the stream by creating ZnS which remains fairly stable (Huang et al, 2004). Removal of chlorine is much more difficu lt because it bonds easily with both copper and zinc and has highly mobile ions. A ZnCl2 compound is not stable enough to remove chlorine, but an aluminum oxide is recommended (Lindstrom et al, 2001). 1.2.2 Rate Laws, Mechanisms, and the Role of Constituents The influence of the constituents of catalyst materials is not reflected in rate laws for catalytic chemistry. Yet this can be misl eading, since different rate laws have been found for the same reactions over various cataly sts. Even if the active sites are the same

PAGE 26

15 chemically, the manner in which they are di spersed and their inte raction with other constituents effect how easily reactants adsorb, interact, and desorb, as well as the ability of the crystalline structure of the site to return to its or iginal state after desorption. Though pure metal crystals act as the active site s in many catalytic reactions, they have also been shown to improve with doping by small amounts of another metal or metal oxide. This can form a promoter which is an alloy that modifies the electronic structure and changes the interstitial electron dens ity, thereby strengthening or weakening interactions on the catalyst surface. Additives can improve the surface area of active sites by the manner in which they adhere to and disp erse the catalyst material. Because of this, the methods of preparation of catalysts are important and ofte n result in differences in activity, for example coprecipita tion versus homogeneous precipitation (Shishido et al, 2004). The discussion of why certain ionic so lutions precipitate a more homogeneous mixture of metal oxides than others is beyond the scope of this pape r. However, it bears consideration that ions overcome an energy of solvation when they are precipitated out of solution. Homogeneous precipitation of diffe rent metal oxides out of solution would require a solution in which the metal ions ha ve similar energies and concentrations to overcome their respective solvation shells so that they will precipitate at similar rates, and therefore more homogeneously. Takeguchi et al (2002) also found that preparation methods resulted in different catalytic activ ities when comparing coprecipitation methods to a glycothermal preparation method. The glycothermal method resulted in a higher catalyst surface area and higher conversion, wh ich supports the concept that surface area effect reactant exposure and turnover frequency from the catalyst site (Takeguchi et al 2002). Some additives have been found to make the catalyst more stable such that the

PAGE 27

16 surface returns to its original state more easily after products desorb. Most of the effects of additives are observed experimentally, rather than anticipated, and analysis of subsequent experiments over numerous reacti on conditions has served to provide the theories discussed in literature. More recently, there have been in situ studies of adsorbates and changes in the catalyst surface during reaction, but the most common means of studying catalysts has been prepara tion the catalyst, characterization of the surface for active sites, exposure reforming cond itions in a packed bed reactor, in various temperatures, residence times and gas concen trations, followed by gas chromatography of reactor products. From this, correlations between measured active sites, chemical composition, and reactor conditions are devel oped. Much of the data used to develop rates is handled this way. R eaction mechanisms and rate laws are often adjusted to match the experimental data. As will be explained la ter, there can be errors in apparent rate laws, since diffusion limitations, non-isot hermality, and residence time can cause inconsistencies and changes in catalytic activity in packed bed reactors. Because of this, studies of monolith reformers and adhesion of catalyst material to monolith substrates have been investigated in the literature, whic h will likely result in more consistent, more predictive algorithms for catalyst and reformer design in the future. The catalyst used in this investigation comprises copper oxide, whic h is reduced to copper metal in order to be active, ZnO, and Al2O3. It is arranged a single layer in the reactor so that the reactions occur on the outer surface of the pellets, thereby eliminating the complications arising from intraparticle diffusion and effectivene ss factors that would normally have to be calculated to estimate the actual flow behavi or, as well as some of the problems of

PAGE 28

17 nonisothermality. Studying the catalytic r eaction over the surface will allow correlations to a monolith-type design. 1.2.2.1 Contributions of Cu (111), ZnO and Al2O3 to activity, stability, and selectivity In catalysis, activity is how easily the catalyst converts reac tants to products, stability describes the long term behavior of the catalyst, an d selectivity is a measure of how well a catalyst enables reactants to fa vor a particular reaction pathway over other possible reactions. Recent literature invest igates the possible roles that th ese additives have in altering catalyst activity, selectivity, and stability. In steam methanol reforming, metallic copper sites are considered the active sites for the reaction mechanism. Comparison of Zndeposited Cu (111), (110), and (100) surfaces have shown minimal catalytic activity over Cu (100) and Cu(110), but much better perfor mance over Cu(111) in methanol synthesis from CO2 and H2 (Nakamura et al, 1998). If ZnO is used to distribute, stabilize, and increase the copper surface area, it stands to reason that the el ectrons in copper crystals in contact with the ZnO would be somewhat le ss mobile since ZnO is an ionic compound. Cu(111) has a higher planar de nsity of copper than Cu(100) and Cu(110). The atoms per crystal face area, where r is the atomic radius, are 0.1768/r2 for a (110) crystal, 0.25/r2 for (100), and 0.57/ r2 for (111). With the reduced electr on mobility due to its interaction with ZnO, the copper atoms per crystal face become important in providing the electron density for adsorption. This gives an indicati on of the importance of the structure of the active site since methanol steam reforming a nd methanol synthesis ar e essentially reverse reactions that take place within the same temperature ranges. It is important to keep in mind that they are not exactly reverse reactio ns since the gas phase species can compete

PAGE 29

18 for the same sites, and the species that ar e already adsorbed influence the adsorptive energies of sites nearby. Still, this illustrates the importan ce of controlling heat transfer to prevent sintering, as well as the structural sensitivity that the alcohol dehydration and synthesis reactions have. Consideration should be given to the fact that more closely packed crystal structures in metals will ha ve greater electron density in the conduction band, which can enable a three dimensional mol ecule to bond more ea sily, especially if the atom that bonds is not one of the outer-most. In fact, methanol it self weakly adsorbs, but has strong bonds when it dissociates into H+ and CH3O-, so the site has to accommodate the geometry of the in-tact methanol molecule, and then the conduction band must be such that its effect on the polarity of the O-H bond distorts it enough to break. Figure 1-2 gives an illustration of th e portions of atoms on crystal faces for a facecentered cubic crystal, such as copper. Figure 1-2. Portions of Atoms on Face-Centered Cubic Crystal Faces

PAGE 30

19 ZnO and Al2O3 additives have been shown to improve the activity and stability of copper catalysts. A correlation between activ ity and stability has been found (Kumari et al, 2002), which is reasonable, as an unstabl e catalyst would not ma intain active sites. Catalysts containing ZnO have a higher surf ace area of copper, and therefore more active sites, and those containing Al2O3 have improved stability (Agrell et al, 2003). Al2O3 is considered an insulating oxide, meaning th at it tends to crea te ionic bonds with adsorbates. ZnO is considered a semiconducti ng oxide and can form either covalent or ionic bonds. Metal surfaces have mainly cova lent bonding with adsorbates. CO tends to bond strongly to metals and weakly to insulating oxides, whereas H2O bonds weakly to most metals and strongly to insula ting oxides (Masel,1996). Addition of Al2O3 has been shown to increase Cu surface area as well. As a result, there is greater conversion of methanol at lower temperatures, however th ere is also a higher selectivity of CO (tendency for a reaction pathway that produces carbon monoxide) (Shi shido et al, 2004). Even though the copper crystal s ite is considered the active si te for catalytic reaction, the apparent influence of ZnO on the activity is not accounted for in terms of increased copper surface area, alone (Gunt er et al, 2001). This is unde rstandable considering that, according to Masel (1996), ZnO can bond eith er covalent or ionic molecules, for example, CO or H2O; and H2O should bond strongly to Al2O3, weakly to clean copper, and reasonably well to ZnO. So, methanol adsorbs dissociatively on copper, but the other reactant, steam, has minimal bonding on copper but good bonding on the other two constituents. Its tempting to infer that ZnO and/or Al2O3 play roles as active sites in the steps of the reaction, in addition to disper sing and stabilizing the Cu(111) sites, but studies on Cu/ZnO/ Al2O3 catalysts all assume an active copper site without examining

PAGE 31

20 adsorption on metal oxide constituents. One th eory regarding the Cu/ZnO interactions is that Cu may be incorporated into ZnO on interstitial and substitutional sites, assuming possible valence states of Cuo Cu+, Cu2+; and that hydrogenation of formate species on Cuo proceeds via hydrogen spillover from ZnO si tes (Agrell et al, 2003). Interestingly, several articles in literature discuss a possible symbiotic relationship between the ZnO and copper sites in steam meth anol reforming, and that copper catalysts with ZnO may have an adsorbed formaldehyde or formyl inte rmediate (Choi et al, 2003). Gunter et al (2001), Grunwaldt et al (2000), a nd Nakamura et al (1998) have pointed to the possibility that additives play a larger role in cataly tic activity, besides disp ersing and increasing the surface area of copper crystals, such as strain at the Cu/ZnO interface, and activity resulting from Zn being dispersed in the c opper bulk. With easily reducible oxides, such as ZnO and CuO, Zn atoms can migrate and form a copper surface alloy at around 300oC (Grunwaldt et al, 2000). The steam methanol reforming reaction has both structural and temperature sensitivity. For example, prep aration methods for dispersion of ZnO and Al2O3 in Cu(111) reflect improved activity over more homogeneously precipitated catalysts, than those in which there are larger conglomerations of each separate constituent (Shishido et al, 2004). Reactants can adsorb on more than one type of the constituents, the structure and crystal faces of which influence adsorptive energies. Simultaneously, there exists a narrow temperature range (175-220oC) where the reaction follows the Arrhenius law, above which the act ual structure of the catalyst can change. Though the precise pathways, and the influences of particular site s on selectivity, are unclear, there is a general trend that a roughl y 50/50 ratio of Cu/Zn results in the highest activity. If Al2O3 is included, up to 18% (Shishido et al, 2004) improves reactivity as

PAGE 32

21 well, if the constituents are dispersed hom ogeneously. Both the copper surface area and the supporting materials have a combined effect on catalytic activity. Although copperbased catalysts with SiO2 have improved over-all surface ar ea, there is reduced specific activity (Saito et al, 2002). Zinc has also b een shown to improve the reducibility of CuO to copper metal for catalytic activity (Fierro et al, 1996). Many fuel cell/reformer systems incorpor ate a selective oxidation catalyst to complete the oxidation of CO to CO2, as a trade off of being able to run the steammethanol reforming reactor at a lower te mperature for nearly the same hydrogen production. So far, ternary catalysts feat uring zinc, aluminum, and copper show good reforming rates and stability at moderate temp eratures, though other studies with Zr or Cr doped copper catalysts are showing improved activity with Zr, and improved stability with Cr (Huang et al, 2004; Loffler et al, 2003; Agrell et al, 2003). 1.2.2.2 Proposed reaction mechanisms and rate laws There have been generalized, proposed mode ls for types of reaction mechanisms in catalysis: Langmuir-Hnshelwood, Rideal-E ley, and Precursor mechanisms. The Langmuir-Hinshelwood reaction fo llows for the reaction A+B AB follows the path A+ S A (ad) B+S B(ad) A(ad) +B(ad) AB + 2 S where S is the catalyst site. This assumes that reaction follows the Langmuir adsorption isotherm in which assumes no competitive and no dissociative adsorption. This rate mechanism supposedly dominates at high v acuum, although many reactions occurring at moderate pressures appear to follow this m echanism as well. (Masel 2001) In the RidealEley mechanism, gas phase species collide with adorbed species and react at once. For

PAGE 33

22 the case of HBr formation, this occurs at mode rate pressures, but none of the pathways in literature for steam methanol reforming seem to follow it. In the Precursor mechanism, species A adsorbs, species B collides with th e surface and enters a mobile state until it encounters the adsorbed A species, forms A-B complex, then desorbs. This seems similar to the Langmuir-Hinshelwood mechanism, in that there are essentially adsorbed complexes that react on the site, rather in the single step of a gas molecule collision with an adsorbed species. In fact, most rate data appear to match Langmuir-Hinshelwood mechanisms. (Masel 2001) There are a numbe r of observed, apparent reaction pathways that can seem to match certain rate laws, t hough these can be skewed by errors related to diffusion limitations in packed bed reactors. The behavior that emerges from the influence of metal oxide additives on the cataly tic activity of a metal surface is difficult to verify experimentally, and even more di fficult to anticipate since there are so many factors. For example, the th eoretical mechanisms describe d above assume all sites are identical and do not account for competitive adsorption. They also do not account for the behavior of various metals or pairwise in teractions between adsorbed intermediates (Masel 2001, 1996). Despite this, apparent rate laws derived from experiments seem to mimic those derived from the Langmuir-Hin shelwood mechanism. Still, much progress has been made in the past few years to wards understanding the most likely reaction pathways in catalysis through experimentation. One thing to bear in mind is that the reaction mechanisms in literature vary wide ly, even though they describe the same or related reactions, because they were studied over catalysts of various compositions and preparations in reformers that were not all the same size or had the same residence times. The composition and preparation affect the na ture of the active sites, and gas phase

PAGE 34

23 composition affects adsorption energies. Since reactants can adsorb on active sites in any arrangement, when they dissociate and react, any two intermediates can react and desorb. This means that there are opportunities for mo re than one reaction pathway to occur and what is actually observed experimentally is the comb ined effect of these. In this section, some of the proposed mechanisms, rate s, and the rationale behind them will be discussed and compared. Init ially, the proposed reaction mechanism for steam methanol reforming was the sum of methanol decomposition and the water gas shift. CH3OH 2H2 + CO CO + H2O H2 +CO2 However, in experiments, CO levels were very small and thought to be either an attainment of equilibrium in the water gas shift, or the decomposition of carbon monoxide on the catalyst, which would lead to the oxidizing of cat alyst sites and carbon build-up. The latter of these explanations is unlikely, since it would quickly lead to catalyst deactivation (Lee et al, 2004). It was also determined from studies of methanol synthesis that methanol is pr oduced by the hydrogenation of CO2 from syngas, which is a mixture of CO, CO2, and H2, rather than by hydrogenation of CO (Peppley et al,1999). This prompted the inclusion the water gas shift reaction in the mechanisms for both methanol synthesis and steam methanol refo rming, presuming, to begin with, that steam methanol reforming was a reverse reaction of methanol synthesis (Peppley et al, 1999; Rozovskii, 2002). This, however, is not exactly true either. It turns out that the, even though the reaction occurs on the surface, th e composition of the gas phase effects catalytic activity. Methanol and CO competitively adsorb, and there is much more

PAGE 35

24 methanol in the gas stream in steam methanol reforming, than in the syngas of methanol synthesis. As a result, methanol competi tively adsorbs and inhibits adsorption and subsequent conversion of CO (Peppley et al,1999). Peppley et al proposed a kinetic model that involves two types of sites and includes all th ree reactions: steam methanol reforming in one step, methanol decom position, and the water gas shift reaction. CH3OH+ H2O CO2 + 3H2 (1) CH3OH 2H2 + CO (2) CO + H2O H2 +CO2 (3) Peppley developed an elaborate mechanism th at involves two types of sites, type 1 for reactions (2) and (3) and type 2 for reacti on (1), as well as type 1a and 2a for the adsorption of atomic hydrogen. Peppley explains the need for these designations because his literature review indicates that H2 does not adsorb on copper. However, according to Waugh, atomic hydrogen can be adsorbed to copper, and indeed it is when methanol dissociatively adsorbs. Perh aps the atomic hydrogen adso rbate migrates across the surface (Masel 1996) to the sort of inters titial site Peppley is referring to. Water dissociatively adsorbs on the Cu/ZnO/Al2O3 (Peppley 1999), possibly on the ZnO or Al2O3 (Masel 1996) or on copper, in which cas e there are scenario s in which atomic hydrogen can be bound to the copper site without requiring different types of sites than the adsorbed reactants. Though Masel (1996) states that water adsorbs on ZnO, no literature in methanol steam reforming, as far as could be found, considers it to be an active site. Since ZnO-enhanced activity is re ported to be insufficiently explained by improvement in copper surface area, alone, it seems worth mentioning here. In Peppleys FTIR spectra of adsorbed species, CO does not appear, which is hard to reconcile with

PAGE 36

25 methanol decomposition on the surface. Fully reversible adsorp tion of CO on clean, hydrogen-reduced copper was determined to be fully reversible at room temperature by dAlnoncourt et al, with heats of adsorption ra nging from 70kJ/mol at low coverages and 45 kJ/mol at high coverages. Adsorbates th at do appear in Peppleys DRIFTS study are hyroxy, methoxy, formate, CO2. Another reaction pathway, to be discussed next, suggests methyl formate as a major intermedia te. To explain this discrepancy, adsorbed methoxy groups have been observed to react to form formate groups in the absence of gas phase methanol. With gas-phase methanol, methoxy reacted to produce methyl formate. (Trimm et al, 2001) Peppleys Type 1 (for both reactions (2) and (3)) is a site that evolves from partial decomposition of a hydroxyl-carbonate phase that supports hydroxylation of the intermediate surface format es, and Type 2 (for (1)) is the Cu(111) face-centered cubic crystal that supports the decomposition reaction. Later studies have proposed reaction mech anisms based on the scheme that water reacts directly with methanol, with CO con centrations resulting from a reverse water gas shift reaction (Agrell et al 2002, 2003; Lee et al, 2004): 2CH3OH CH3OCHO + 2H2 (4) CH3OCHO + H2O CH3OH + HCOOH (5) HCOOH CO2 +H2 (6) The net of the above reactions (4)-(6) is CH3OH+ H2O CO2 + 3H2 and the reverse water gas shift reaction is CO2 +H2 CO +H2O (7)

PAGE 37

26 While Purnama and Agrell have develope d overall rate laws based on similar reaction schemes, Lee proposes a reaction mech anism involving two sites to explain their experimental data. Like Peppleys model, Lee proposes that adsorbed methoxy and adsorbed hydrogen do not occur on the sa me active sites. Since only hydrogen and methanol were shown to affect the reforming rate, partial pressures of other terms, and likewise, adsorption of other components on ot her types of sites, were disregarded as negligible towards the overall reaction. Cons idering the earlier di scussion of possible adsorption of reactants on more than one type of site, it would seem that these noncontributing pathways would contribute to thermal inefficiency, giving rise to the importance of homogeneous distribution of active sites and catalyst additives in the overall rate of hydrogen production, as shown by Shishido and Agrell. Yet another study shows a variety of the po ssible interactions between adsorbates, favoring a pathway that direc tly involves combination of water with methanol, rather than through a water gas shif t reaction (Huang et al, 2004). CH3OH HCHO + H2 (8) HCHO + H2O HCOOH + H2 (9) HCOOH CO2 + H2 (10) Equations 8, 9, and 10 sum to equation (1). Another pathway could be: 2CH3OH 2HCHO + 2H2 2HCHO CH3OCHO (11) CH3OCHO CH3OH +CO (12) CO + H2O CO2 + H2 (13)

PAGE 38

27 which also sum to equation (1). Huang sugge sts that reactions 8, 9, and 10 predominate, though the second scheme has intermediates more similar to those in the scheme used by Lee, et al. Studies by Agrell et al, Lee et al and Purnama et al all favor mechanisms in which CO appears in a secondary reaction, such as the revers e water-gas shift, since it tends to appear when the methanol has been consumed (Lee et al, 2004). Some general trends in the reforming studies indicate that higher temperat ures and longer contact times increase CO concentrations. In most cases, the rate of hydrogen production seems to level off with increasing temperature once 80% of the methanol has been consumed. For example, in Shishido et als study, the rate of hydrogen production was more than twice at 250oC as it was at 200oC, but only increases by about 3% at 300oC. Choi et al (2003, 2005) did studies of methanol decomposition without water, and with lower than stochiometric amounts of steam, and determined that, unless the feed contained less than 43 mol % water, no methyl formate was detect ed in the product gases. Neither were dimethyl ether, or methane detected, and thes e only appeared at less than 24 mol% water and less than 8 mol % water, respectively (C hoi et al, 2003). All st udies indicated that CO formation was suppressed with increased water concentrations, though there was slight variation between the amount of exce ss water and the level of CO suppression for various catalysts. Generally, it ranged from a mol ratio of 1.1 to 1.4 for steam to methanol to keep CO concentrations in the produc t gas below 2% at the temperature for 80% methanol conversion.

PAGE 39

28 Table 1-1 Summary of Proposed Rate Laws Reference rate law notes Lee -rM = ka exp (-Ea/RT) PM a (A + PH)b mol/kg/sec M = methanol, ka = 2.19E9, Ea= 103 kJ/mol, H=hydrogen, A= 11.6kPa, a = 0.564, b=0.647 P= partial pressure, -rM = rate for methanol decomposition Purnama rSR= k1 PCH3OH m PH2O n rSR = steam reforming rate m= 0.6, n= 0.4 Apparent Ea= 76 kJ/mol, Pre-exponential k01= 8.8E11/kg/sec k1= k01 exp (-Ea/RT) rRWGS= k2 PCO2 PH2 k-2PH2OPCO Apparent Ea = 108 kJ/mol Pre-exponential k02 = 6.5E9 bar-1sec-1gcat -1 k2= k02 exp (-Ea/RT) Agrell(Jiang) rSR= k PCH3OH 0.26 PH2O 0.03 (PH2< 7kPa) rSR= k PCH3OH 0.26 PH2O 0.03 PH2 .2 (PH2>7kPa) k(T)= 1.9E12 exp(-100.9kJ/molRT)

PAGE 40

29Table 1-2: Summary of Steam Methanol Reformer Studies Reference Agrell Peppley Jiang Lee Purnama Argawa l Shishido Activation Energy, kJ/mol 100.9 102.8 105.1 103 76 92 not given Temperature, C 175-270 160-260 170-260 160-260 120-325 260300 200-300 Catalyst Composition, Cu/ZnO/Al2O3 wt% 66/23/11 40/40/20 31.7/49.5/18. 8 64/24/10/, 2% MgO 50 wt% Cu, others not given 10/5/19 85 45/45/10 Feed Composition, steam:methanol 1:1 and 1.3:1 2:1, 1.5:1, 1:1 1:1 1.4:1 1.2: 1 Mass of Catalyst 50 mg 1 g 200 mg 200 mg Reactor Size 6mm dia 7.5mm 10mm % Methanol Conversion, equal molar reactants 80% at 260C 80% at 240C 80% at 235C 80% at 250C 78% at 300C 97.3% at 250C 2% CO Selectivity 275 C for 1:1 steam: methanol, 310C for 1.3:1 steam: methanol much less than 1% 1% at 250C

PAGE 41

30 Lees experiments concluded that only methanol and hydrogen partial pressure influenced the reforming rate. Their mechan ism is modeled on a methyl formate reaction route, shown by equations 4, 5, and 6 given above, with CO occurring only through a reverse water-gas shift reaction in a secondary step. CO was not detected at reforming temperatures lower than 200oC and much less than 1 mol % during experiments at higher temperatures. Steam to methanol ratios in their experiments were 1/1, 1.5 /1 and 2/1. Increases in CO concentration were linked to increases in methanol in the reactants, and it is suggested that the small amount was due to a reverse water-gas shift via increased production of CO2 and H2, rather than the amounts pred icted by methanol decomposition to CO in the Peppley model. The role of wate r appears to be mainly to convert the methyl formate species to methanol and HCOOH, the former repeating reactions 4 and 5 again, and the latter decomposing to CO2 and H2. Lee sites internal diffusion limitations for changes in catalytic behavior above 200oC for particle sizes of 0.3-0.42 mm diameter pellets. The model proposed by Purnama et al i nvolves the overall reactions of steam methanol reforming and reverse water gas re action, and like Lees model, considers the reverse water gas shift reaction to be secondary, occurring as a result of a reverse reaction involving the products of the steam reforming reaction. Steam reforming (SR): CH3OH +H2O 3H2 +CO2, KSR = k1/k-1 = (PH2 3PCO2 )/(PCH3OH PH2O) Reverse water gas shift (rWGS): CO2+ H2 CO+H2O, KrWGS = k2/k-2 = (PH2OPCO)/( PCO2 PH2)

PAGE 42

31 Purnama et al also determined that th ere were greater intraparticle diffusion limitations for catalyst particle sizes of 0.71-1.0 than for 0.45 0.50 mm diameter particles. CO concentrations increased with the increased catalyst particle size, and this effect has been linked to the difficulty of product gases diffusing through the pellet material at higher temperatur es. This may be an analogous effect to the increased CO concentration that occurs at higher residence times as we ll. The accumulation of products can force the reverse reaction if they rema in in proximity to the active sites. CO concentrations at all reaction temperatures were below those predicted by the water gas shift reaction, supporting other findings that CO2 does not occur via th e water gas shift, and that CO formation occurs in a consecuti ve reaction rather th an through methanol decomposition. (Purnama et al, 2004) Data comp lied much more closely with the reverse water gas shift following st eam methanol reforming reaction. The reasons for minimal involvement of the water gas shift reaction can be attributed to competitive adsorption between methanol and CO (Purnama et al, 2004). The investigation done by Agrell et al(2003) also supports a methyl formate route with CO occurring in a subsequent, reverse wa ter gas shift reaction. Agrell et al used catalyst pellets of size 0.12-0.25 mm diameter Like the other two models, CO production increased with temperature and contact time, suggesting a relationship of the mobility of product gases away from ac tive sites as a function of temperature and diffusion limitations. The increase in temperature can push the reaction rate to a quicker completion, but then this can re-enforce the tendenc y of reverse reactions since product gases can re-adsorb more strongly at the higher temperature.

PAGE 43

32 In all three of the more recent models, (Purnama et al 2004, Lee et al 2004, Agrell et al 2003), water participat es directly in a steam methanol reforming reaction on the catalyst, via a formate intermediate, rather than the two step meth anol decomposition and water gas shift reaction mechanis m indicated by previous models. If this is the case, in the scope of including a methanol steam refo rmer in a portable fuel cell system, which would include a water-gas shift reactor for converting CO (Zalc et al, 2002) may be unnecessary. It may be more efficient to se parate the hydrogen from other reformate gases via separation membranes rather than in vest in additional reformers for complete gas conversion. 1.3 Heat Transfer Limitations in Packed Bed Reformers, Advantages in Monolith Reformers Many heterogeneous catalyst applications rely on forcing reactant gases through porous catalyst pellets designed with a surface area and porosity to maximize the contact with active sites on the catalyst. However, te mperature and pressure gradients occur in a packed bed configuration, which can result in material wear, incomplete reactions in some parts of the reformer, and greater risk of sintering. This especia lly true of reactions and catalyst materials which have a narrow te mperature range of optimal performance, and which have shown sensitivity to catalys t surface geometry, such as reactions with alcohols. The diffusion limitations within cata lyst pores have been linked to increased CO production and catalyst deactivation (Purnama et al 2004, Lee et al 2004, Agrell et al 2003). Research has shown that the diffusion limitations can be reduced with smaller pellet sizes, meaning that fewer reactant gase s go through the pellets; instead, there is an increased flow around the pellets. However, these smaller pellet s also increase the pressure drop along the catalyst bed. The pr oblem with this is, not only that higher

PAGE 44

33 pressure at elevated temperat ures increases the likelihood of sintering and material wear, but that product gases are not qui ckly removed away from cataly st sites after they react. It has subsequently been suggested that a future design might be an eggshell type catalyst, meaning a thin-walled, honeycomb t ype catalyst, with channels through which the reactant gases would flow, rather than the microscopic diffusion flow through pellets (Agrell et al, 2003). 1.3.1 Non-Isothermality in Packed Bed Reactors Many packed bed reactors are modeled af ter a plug flow assumption, however, in reactors with small ratios of tube diameter to pellet diameters, temperature gradients can be significant enough to skew reactor data. (Karim et al, 2005) The models discussed above have crushed and sieved commercial pellets to approximately less than one millimeter diameter. Small tube diameter to pa rticle diameter ratios, as in micro reactors and the experimental pellet beds used in th e studies above, result in a variations in velocity and temperature. (Demirel et al 2000) Changes in velocity and temperature profiles caused by back mixing prevent accurate simulation of the temperature response by the traditional plug-flow model, and a hyperb olic model of axial dispersion in packed beds has been described. (Prasad et al, 2002) Catalyst dilution and th e change of reactor diameter from 4.1mm to 1.0 mm have shown higher apparent catalyst activity, indicating heat transfer limitations within the pellet bed (Karim et al 2005). Temperature gradients of up to 40 K and 22 K have been observ ed within the 4.1 mm and 1.0 mm reactors, respectively. (Karim 2005) For endothermic r eactions, temperature gradients can lead to lower catalyst activity; and for exothermic reactions, they can lead to reaction rates several thousand times greater at the axis than at the wall (Karim et al, 2005). Heat transfer at the wall cannot be neglected for react or tube to particle ratios of less than 100,

PAGE 45

34 and the ones sited in this paper range from 20 to 50. Karim et al tested a commercial Cu/ZnO/Al2O3 catalyst in packed beds of diam eter 4.1, 1.75 and 1 mm using 200 mg of catalyst in the first experiment and diluted w ith 175 mg of inert material in the second. The additional material effectively incr eased the bed length, th ereby increasing the external surface area for heat transfer. The effect was higher apparent catalyst activity, but reduced reaction rate per bed volume; that is to sa y, there was higher methanol conversion with the dilution, but a greater bed volume for the catalyst was needed. For 90% conversion, 25% more catalyst for the 1.75 mm diameter reactor than for the 1 mm diameter reactor, and 85% more for the 4.1 mm reactor for 1.1/1 steam to methanol ratio at 230oC. Karim et al assert that these observatio ns are linked to interpellet mass transfer limitations, rather than intrapellet or inte rphase mass transfer limitations. However, decreasing pellet size and reacto r diameter to reduce interpellet mass transfer issues increases the over all pressure gradient in th e reactor, and dilution of the catalyst is not feasible for commercial applications. Karim, Bravo et al then developed a wall coated reactor of the same catalyst used in their pa cked bed experiments and compared methanol conversion and CO selectivity. Their packed bed reactor was a 4.1 mm i.d. 8.55 mm long quartz tube packed with 100 mg of 0.10 mm diameter catalyst pellets. Their wall-coated reactor was coated with 75 mg catalyst on the inside of a 4.1 mm i.d. quartz tube 18 cm long. For 80% methanol conversion, the wall co ated reactor required a flow rate of 20 W/F (kg s/mol) compared to 34 W/F (kg s/mol) in the packed bed reactor, where W/F is the ratio of the weight of catalyst to the fu el flow in mol/s. 90% conversion was achieved at 27 W/F for the wall coated reactor and 47 W/F for the pellet bed reactor. The 2% CO selectivity occurred at 92.5% methanol conversion in th e wall coated reactor, and 85%

PAGE 46

35 conversion in the packed bed reactor. Br avo et al show that the methanol steam reforming reaction is catalyzed more efficiently in the wall-coated reactor, i.e. when mass transfer and thermal gradients are reduced. Additionally, their findi ngs support previous studies that CO formation o ccurs at higher methanol conversions, and increases with mass transfer limitations. If CO also occurs as a secondary reaction from the products of the steam methanol reaction as hypothesi zed above, then these findings reflect the importance of determining the right geometry a nd flow rate for the best contact time with the catalyst, as well as reducing back mixing and turbulence, which in crease contact time that products have with the cat alyst. It is important to not e that the methanol synthesis reaction has been studied over the same cat alyst and similar temperature range, so controlling reverse reactions is important. 1.3.2 Monolith Reformers Monolith catalyst configurations for hydrogen production from methanol have begun to appear in the past few years. Th eir advantages include improved temperature control and greater reliability and durability. For portable ap plications, a catalyst-coated monolith, made of either stainless steel or ceramic, is more structurally robust (Lindstrom et al, 2003). Howeve r, most commercially availa ble steam methanol catalysts are in pellet form. Examples of variou s configurations incl ude the honeycomb-like monolith used by Lindstrom, catalyst coated brass wires (Horny et al, 2004), and layers of stainless steel or glass etched plates that can be coated with catal yst (Park et al, 2005; Yu et al, 2005; Terazaki et al, 2005). Seve ral studies have been devoted to catalyst adhesion and the effect of the substrate on catalytic activity (K awamura et al, 2005; Bravo, 2004). Horny et al found that coating catalyst on brass wires and placing the wires close together like a cable created straight channels approximately 1 mm in diameter.

PAGE 47

36 They found that the laminar flow and short ra dial diffusion resulted in nearly plug flow with hydrodynamics similar to multichannel microreactors. In terestingly, they also found that Al2O3 was a necessary component for catalytic activity over the Cu/Zn-coated brass wires. This reflects an influence on the alloy microstructure on activity. Monolith reformer studies have so far shown developments in the surface science in adhering the catalyst to a monolith frame, such as a wash coated glass capillary, or catalyst coated wires or etched plates (Fukuha ra et al, 2004; Bravo et al, 2004; Horny et al, 2004). In addition to isothermality and better control of reaction pathways, monolith reformer designs have simpler fluid dynamics and can be used in a broader range of various scales. Several micro-reforming systems have been developed that in corporate catalytic combustors and vaporizors with the reformer. Th ese chambers are arranged in layers in a heat exchanger configuration. In the start-up of the micro-reformer system, hydrogen and air from a gas supply are introduced into the combustion chamber to heat up the catalytic combustor (Pan et al, 2005). The combustion chamber is adjacent to the vaporizer, and the thermal contact allows enough heat transfer to vaporize the reacta nts that are supplied to the methanol reformer. The catalytic co mbustor could also be designed to consume some of the methanol rather than part of the hydrogen product gas, af ter start up. In the combustion and reforming chambers, typica lly a commercial catalyst is ground, made into a slurry, and adhered to etched channels which make the layers of the reformer and combustor. In the case of Pan et al, actual pellets smaller than 1 mm were loaded into chambers that appear to be approximately 8 mm wide, according to their schematic. The coated surface approach has advantages in scalability. Researchers found comparable performance between the monolith and packed bed reformers (Lim et al, 2005; Lindstrom

PAGE 48

37 et al, 2003). Though one detractor of the monolith reformer is that th ere is less catalyst material per reactor volume, they are able to operate at higher temperatures (210-290oC) and achieve higher methanol conversion with less risk of sinter ing. Recall that many pellet bed studies found that sintering can beco me a problem at temperatures as low as 220 oC. There is higher methanol conversion at the higher temperature, but also, the catalyst coating is more stably distributed a nd less likely to develop hot spots than in the packed bed reformer. The effect of reduced catalyst material per volume in the monolith reformer is compensated for by the greater operating temperature ra nge and reduced back flow, which is considered undesirable (Demirel et al, 2000). Monolith and metallic heat exchangers offer advantages over pellet beds in terms of mechanical strength, which is more ideal for portable applications (Giroux et al, 20005), additionally, the heat transfer through the monolith structure is much more efficient than conduction from the walls through the pellets, which is an important adva ntage in start-up times for fuel reformers that supply fuel cell systems (deWild et al, 2000). CuO/ZnO pelle ts crumble easily, and over time, the interpellet diffusion discusse d by Bravo et al will become more of a problem in a mobile reform er experiencing vibration. It should be noted that the reformer sy stems in these monolith studies are very small, under 100 cm3 in total volume, with the dimensions of the methanol reformer section being less than 1mm in diameter. Being fully enclosed in the system, they do not lend themselves well to in situ studies. The pellet bed experiments were also conducted in small diameter tubes. Thermocouples can be employed for in situ temperature monitoring, but generally the catalyst is analyzed by conduc ting a spectroscopic measurement of copper surface area prior to reaction, then product gases are analyzed.

PAGE 49

38 Figure 1-3. Example of a micro-refo rmer, adapted from (Pan et al). This was the predominant approach for both pellet bed and monolith reformers. Some recent studies using Raman spectroscopy have ve rified surface intermediates and changes in surface structure and chemistry, with mu ch emphasis on developing the investigative process itself (Gunter et al, 2001; Grunwaldt et al, 2000). In s itu studies have also been performed to assess the effect of the gas phase composition on ad sorbate species (KnopGericke et al, 2001), surface vibrational spectr oscopies of chemisorbates (Park et al, 2002), copper oxidation states (K udelski et al, 2004), among others. There has been an in situ study over a catalytic surface in which th e gas phase is sampled at various locations along the reactor (Springmann et al, 2002), but this approach has disadvantages in that the flow is disturbed during the sampling and the boundary layer cannot be observed. The monolith reformer concept has shown advantag es in durability and in thermal and mass transport issues; however, since residence ti me has been linked to undesirable secondary reactions in steam methanol reforming, the reactor must be scaled properly for the

PAGE 50

39 prescribed reforming rate. Understanding th e effect that flow rate has on the two dimensional gas phase composition (reactor length and height above catalyst) would benefit from in situ studies. The reformer in this study employed commercial steam methanol reforming catalyst pellets, however they were arranged on a tray so that reac tant gases primarily flowed over the catalytic surf ace, rather than through the pellets. This reactor was designed on a scale so that the spatial variation of the gas pha se over the catalyst surface could be studied using Raman spectroscopy and a readily available ca talyst, in order to develop some scale factors for reforming in a monolith configuration. Perhaps in the future, the adhesion processes described by G. -G Park, Lim, or Bravo et al will be developed on larger surfaces so that the sp acing between two cata lytic surfaces can be studied in this in situ configuration. 1.4 Raman Spectroscopy for Study of Catalysis Raman spectroscopy has become popul ar in the past 5 years for in situ study of catalytic surfaces (Weaver et al, 1999; Stair et al 2000, 2003; Wachs 1999; Knozinger et al,1999). The time scale for Raman spectroscopy allows simultaneous study of kinetics and molecular characterization (Wachs 1999) hence recent emphasis has been on the surface chemistry. Supported metal species have different vibrational energies depending on their arrangement; similarly, adsorbate species vary in vibrational energies from their gas phase counterparts. This is due to the bond distortion, and th erefore vibrational frequency, that occurs when a molecule adsorbs to a surface, as well as bond structures associated with crystal lattices or metall ic oxide surfaces (Haw 2002). When molecules adsorb on these, both the vibration of the bonds of the adsorbed molecule and the of the catalytic surface are changed. The gas phase composition gradients arise out of the

PAGE 51

40 combined effects of surface ch emistry, reactant molar ratio, and the free stream velocity. This Raman study over a catalys t differs from other surface studies by determining the relationship that the free stream velocity has on the two-dimens ional scale of the composition boundary layer.

PAGE 52

41 CHAPTER 2 RAMAN SPECTROSCOPY Electromagnetic radiation on incident a ny matter can be emitted, absorbed, or scattered; and most materials have some combination of absorption and scattering. In radiation scattering, a photon of light energy incident on a molecule will induce a dipole moment in the molecule or at om. The energy of a photon is h inc, where h is Plancks constant and inc is the frequency of the incident radiation. The frequency is = c/ where c is the speed of light and is the wave length. In elastic scattering, the emitted radiation has the same frequency, and theref ore the same photon en ergy, as the incident radiation, but the direction (i .e. distribution) of the s cattered energy depends on the incident angle, size of scattering, and polari zation of incident li ght. Types of elastic scattering are distinguished by the refractive index and the wavelength of the incident radiation as compared to the particle size. Raman scattering, however is inelastic; and emitted radiation shows a shift in frequenc y from the incident radiation, which is generally irrespective of scatte ring angle (Ingle et al, 1988). With Raman scattering, the incident ra diation and the induced dipole moment create a quantum shift in the vibrational modes of the molecule. In the case of Raman scattering, the excitation energy (h inc) does not exactly match th e difference between the vibrational quantum energy levels in the mo lecule. Therefore some of the energy is absorbed, and the remainder is emitted. Th e emitted photon is therefore shifted in frequency by the amount of vibrational energy. As described above, whenever an incident photon adds energy to a molecule such that its vibrational energy is increased, the emitted

PAGE 53

42 photon has a proportionately lower frequency. This is termed a Raman Stokes shift, and is more common than the reverse phenomenon, anti-Stokes shift. This is because the ground state is the most highly populated, hence it is more likely that the incident photons encounter a molecule in its ground vibrational state. Anti -Stokes shift can occur as molecules relax to a lower vibrational stat e during the scattering interaction. The Stokes shift has frequency of inc vibration, and the anti-Stokes shift frequency is inc + vibration. Both vibrational and rotational modes of a molecule can cause a Raman shift, although the rotational energies are usually much less th an the vibrational energies. A molecule is considered Raman-active if it can undergo a change in polar izability during one of its normal modes of vibration. Atoms are not Ra man active since thei r polarizability does not change with vibration. The theory and requirements for Raman scatte ring are as follows: An electric field E interacts with polarizab le electron clouds and induces a dipole moment in (Ingle et al, 1988). E = Em cos (2 p ex t) in= E where Em is the amplitude of the wave, in is the induced dipole moment, and is the polarizability in J-1C2m2. Polarizability varies with internuclear separation around the equilibrium polarizability, o, occurring at the equilibrium bond distance re. = o + (rre)( / r)e +. The change in nuclear distance is (rre)= rm cos (2p vt)

PAGE 54

43 where rm is the maximum internuclear separation, and v is the frequency of vibration. The criteria for Raman shift is that the following integral be non-zero: R = *i (rre) ( / r)e j d where and are the complex conjugates of the wave function, representing the probability distribution of molecu les in a particular state. Each type of Raman-active molecule cau ses a particular shift in the Raman spectrum resulting from the nature of the atom ic bonds in that molecule, which is what makes Raman spectroscopy useful for chemical identification. Molecule s with more than one type of bond can show more than one sp ectral line. Methanol, for example, has a spectral line associated with its O-H bond, that is distinct from, but close in wavelength, to the O-H bond in water. It can also s how spectra for the C-O bond and the C-H bonds. Table 2.1 gives typical Raman shifts of th e species investigated in this study. Table 2-1 Raman Shift Species Raman Shift (cm-1) Emission (nm) from Excitation of 355 nm H2 4156 416.0 H2O 3650 407.5 CH3OH 2955 396.2 In a sample Raman spectrum taken during reforming, there are several peaks that reflect the vibrational modes of different bonds within the same molecule. For example, near the water peak at 3650 cm-1 there is a methanol peak. These two peaks have similar spectra due to the vibrational modes of the OH bonds. However, they still differ owing to the interaction of other atoms, -CH3 in the case of methanol, and the H in water. The hydrogen peaks are located to th e right of water, near 4156 cm-1. There are two possibly because of the existence of ortho and para hydrogen. In ortho and para hydrogen, the

PAGE 55

44 atoms have either parallel or opposite electron spins, which could affect the induction of a dipole moment between the two atoms. On the far left are two additional methanol peaks, which are associated with th e C-H bonds in methanol (Figure 2-1). 50 100 150 200 25028003200360040004400Raman Scattering Intensity (a.u.)Raman Shift (cm-1) Figure 2-1. Typical Raman Spectru m of Steam Methanol Reforming H2 OH from H2O C-H bonds of methanol

PAGE 56

45 CHAPTER 3 REACTOR DESCRIPTION 3.1 Reactor In this study, the composition of the chem ical boundary layer above a catalyst in steam methanol reforming was evaluated as a function of reactor length and flow rate. Katalco 33-5 catalyst pellets manufacture d by Johnson Matthey were used for the catalyst. The pellets are comprised of 64 wt % CuO, 24 wt % ZnO, 10 wt % Al2O3, and 2 wt% MgO, and measure 5.2 mm diameter by 5.4 mm length. They were packed end-toend in a single layer in an aluminum pelle t bed, shown in Figures 3.1 and 3.2. The pellet bed was heated by cartridge heaters placed in each end of the bed. Three access ports along the length of the bed allow for thermo couples F, G, and 4 to measure the bed temperature, as shown in Figure 3-3. Figure 3-1: End View of Pellet Bed

PAGE 57

46 Figure 3-2. Top View of Pellet Bed The pellet bed is centered within the reac tor, and its temperature is regulated by routing thermocouples F and G, and thei r corresponding cartridge heaters F and G, through a temperature controller The reactor itself is constr ucted from 2.5 stainlesssteel vacuum housings purchased from Hunti ngton Laboratories. The housing is wrapped in heater tape E, which is routed to a temperature controller and thermocouple E, and covered in fiberglass tape insulation. The r eactor housing is stainle ss steel with stainless end caps, valve fittings, and view ports. The reactants are preheated to ensure complete vaporization by passing the reactants through a pre-heat se ction filled with brass pellet s to promote heat transfer. Methanol and water are delivered to their resp ective tubes via peristaltic pumps, and each tube has a nitrogen feed for carrier gas. The r eactant delivery tubes are wrapped in heater tapes (A and C for steam, B and D for metha nol) and insulation, and the temperature is controlled by thermocouples A and B and temperature controllers. See Figure 3.3. The temperature of the stream above the catalys t is measured by thermocouples 1, 2, and 3. Power for the heating elements is delivered through the temperature controllers. When

PAGE 58

47 centered within the reformer, the actual cata lyst surface is approximately 5.5 inches from each end. The reactant and product streams ente r and exit through fittings with an inside diameter of approximately 0.25 inches. The cat alyst surface is approximately level to the inlet stream, however, the gap between the ends of the pelle t bed can cause some mixing, which may result in non-ideal flow conditions. This will be discussed in later chapters. 5 channel temp indicator temperature controller flow controller, nitrogen methanol pump water p um p thermocouple po w e r to h eate r tapes A C, B D E a n d to ca r t ri dge h eate r s F a n d G Nitrogen Methanol vapor Steaminlet outlet 1 2 13 4 A,C B C D E F G E F G A B,D A B 1 2 3 4 Key E Flat Bed Steam Methanol Reformer Figure 3-3. Reformer assembly. 3.2 Optical Configuration The light source is a Q-switched, Neodimium:Ytrium-Aluminum-Garnet (Nd:YAG) laser firing at 5 Hz with a 8 mm beam diameter. For Raman measurement, frequency-tripled light at 355 nm was used. Based on earlier resear ch (Ball 2005), it was found that 355 nm excitation provided an optimal Raman signal as compared to 532 nm excitation. Theoretically, the Raman signal should increase as the inve rse fourth power of

PAGE 59

48 wavelength, hence the 355 nm should provide a better signal. One must also consider fluorescence with UV-excitation, however, fl uorescence was not found to be an issue with the current system. The optical configuration is shown in Figur e 3-4.The mirrors all have combinations of 532 and 1063 nm anti-reflection coatings to minimize the amount of residual laser energy at these two wavelengths. The laser beam is focused to a horizontal line using a 500 mm focal length cylindr ical lens. The resulting beam is approximately 500m high by 8 mm wide. Raman scattered light is collected in back scatter using a 4 diameter pierced mirror and then focused to a fiber optic using a 4 diameter, 160 mm focal length lens. Prior to entering the fiber optic, light was passed th rough a 375-nm high-pass filter (Filter 1) and 450-nm short-pass filter (Filter 2), which f unctioned to block residual 355-nm and 532nm light, respectively. Finally, a sharp-edge high-pass filter (F ilter 3) was used to further reduce 355-nm stray light. Light was fiber-coupled to a 0.275-m grating spectrometer and recorded with an intensified charge-coupled device (CCD) arra y detector. The detect or was synchronized to the Nd:YAG laser, with Raman spectra reco rded using a 300-ns gate centered on the laser pulse. In general, Raman scattering is prompt, hence coincident with the ~20-ns full-width laser pulse. However, the wider gate was used to ensure capture of the entire signal due to the ~150-ns of laser-pulse jitter. Ta ble 3.1 lists equipment and specifications. This reactor was constr ucted from off-the-shelf components.

PAGE 60

49 Figure 3-4: Experimental Se t-Up for Raman Specrtroscopy Adjustable Platform to raise and lower reactor, 1 at each end Cylindrical lens Beam Dump Beam Dump Beam Dump Beam Dump 1064 nm Nd:YAG Laser Beam Dump Aperture Pierced Mirror Collection Lens Filter 1 Filter 2 Filter 3 Fiber Optic Cable Scattered Light 355 nm turning mirror, 45o, passes 1064 nm 355 nm turning mirror, 45o, passes 532/1064 nm ( 2 mirrors )

PAGE 61

50 Table 3-1. Equipment and Specifications Equipment Specification Heater Tapes A, B, C, D, E A and B: 1"x8' Omegalux FGH101-080 120V, 830W, 6.92A C and D: 0.5"x8' Thermolyte B1H051-08, 120V, 420W, 3.5A E: 0.5"x4' Thermolyte B1H051-040, 120V, 210W, 1.75A, 482C Thermocouples A-G, 1-4 Type K Cartridge Heaters F and G 0.25"dia, 9.5" length McMaster-Carr #3618K197 Peristaltic Pumps (2) Fisher Scientific Low Flow 13-876-1 Nitrogen Flow Controller Alicat Sc ientific MC-5SLPM +/0.05 slpm Temperature Controllers (3) Omega CN872RTD-500C +/-5% temp. range View Ports VP-151 7056 glass Multichannel Temp. Indicator Omega Digicator 412B-K, 0.1deg resolution Collection Lens 4" Plano-convex, Comar, 160-PG-100 Pierced Mirror 100 mm x 100 mm, center pi erced 0.5"in +-0.2"dia, Rolyn Optics, 60.2475 Cylindrical Lens Plano-Convex s quare, L30.0/ W30.0/FL500, 022-2546 Filter 1 high pass filter, Comar, 375-GY-50 Filter 2 short pass filter, CVI Laser LLC, SPF-450-2.00 Filter 3 Comar, 685-IH-25 Big Sky Laser 1064 nm Q-switched (Nd:YAG) turning mirrors: 355nm blocks 1064, 1 CVI Laser LLC, BSR-31-2025 blocks 532/1064, 2 CVI Laser LLC, BSR-315-2025 peristaltic pumps, 2 Fisher Scientific Low Flow, 13-876-1

PAGE 62

51 CHAPTER 4 EXPERIMENTAL METHODS 4.1 Catalyst Bed Reduction To be active, the catalyst must be in the reduced state. That is, the pellets must be heated in a nitrogen/hydroge n environment to reduce (remove oxygen from) the pellets. The product of the reduction process is steam. Based on information from the manufacturer, the process for the ca talyst reduction is as follows: 1. Heat catalyst in an N2 atmosphere to around 130o C (~3o/min). Heating the pellet bed in air can oxidize the coppe r or zinc in such a way th at the pellets are sintered or deactivated, such that they canno t be reduced. Introduce hydrogen at a concentration of about 1% of total pressure. 2. Observe temperature profile down the bed. It is recommended to limit the maximum bed temperature to 240 oC during this operation. 3. After the temperature wave has progre ssed down the bed, increase the hydrogen concentration to 15%, watching for temperature spikes. 4. Maintain this flow rate and concentrati on until the temperature profile is steady. 5. Though the product data sheet states that operation can occur over the range of 200320 oC, the exothermic process will heat the catalyst to higher than the bed temperature. Therefore it is recommended to start at the lower temperature. Between test runs, the reformer was turned off for safety and power concerns. To shut down the reformer following a test, the reformer was flushed with nitrogen at the reaction temperature to remove any trace r eactants or products that may condense onto the catalyst bed at room temperature. Then the reactor inlet and exit valve will be closed, with a slight over-pre ssure of nitrogen. The catalyst may exhibit changes in activity over long run times, or differences in catalytic activity dependi ng on the initial reactant co mposition. A study of long term

PAGE 63

52 operation (over 100 hours) conducted by Dams et al (2001, 2002), et al indicated that temperatures between 200 oC and 220 oC resulted in the leas t amount of change in catalyst activity, as well as minimal coke bu ild up. Although this study is much shorter, the temperature range was limited to 200-220 oC in order to preclude the any complications involving deactivation or reve rse reactions discussed in earlier. Typical Raman spectra observed during reduc tion are shown in Figure 4-1 80 120 160 200 240 280 320034003600380040004200Intensity (a.u.)Raman Shift (cm-1)Initiation Mid-Point Completion H2O H2 Figure 4-1 Catalyst Reduction Raman Spectra Initially, there is a high concentration of hydrogen. As reduction proceeds, water appears as the hydrogen reacts with the CuO, leaving pure copper sites. As the reduction nears completion, the water is flushed out of the reactor and hydr ogen reappears, as shown in Figure 4-1.

PAGE 64

53 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 020406080 Hydrogen Signal Water SignalRaman Signal (a.u.)Time (min) Figure 4-2 Water and Hydrogen Si gnal vs Time During Reduction Figure 4-2 shows the changes in hydroge n and water signals during reduction. 4.2 Reformer Operation For the current study, it was desirable to r un the methanol reformation process at a steam/methanol molar ratio slightly greater than 1. The actual flow parameters are summarized in tables 4.1 and 4.2, noting that th e two flow conditions will be referred to as Flow 1 and Flow 2. The overall reformati on process is summarized in this section. Table 4-1. Reactant Flow Rates and Pump Settings Water flow rate Methanol flow rate pump dial setting (fast) liquid flow rate (ml/min) pump dial setting (fast) liquid flow rate (ml/min) 2.5 0.475 2.75 1.0 9 0.65 9.2 1.37

PAGE 65

54 Table 4-2 Flow Parameters Flow 1 Flow 2 Nitrogen 0.70 l//min 0.96 l/min Methanol 1 ml/min 1.37 ml/min Water 0.475 ml/min 0.65 ml/min Steam/Methanol molar ratio 1.079/1 1.054/1 Inlet velocity 7.0 cm/s 9.7 cm/s Residence time 6.3 s 8.4s 4.3 Reformer Operating Procedure Reformer operating and shut-down procedures were devised by the researcher. The laser operating and shut-down procedures were devised by the re searcher and D.W. Hahn. The procedure for operating the reform er and taking spectroscopic data is as follows: I. Start Up Procedure for Steam Methanol Reformer 1. Plug in power. Switch on power strips. Turn on all three temperatur e controllers and set to 200oC. Open nitrogen cylinder. Turn on nitrogen flow feed to steam and methanol pre-heat tubes. Verify that methanol and water supplies are full. Turn on fume hood. Open reactor inlet and e xhaust valves, connect exhaust to fume hood. Wait for bed to reach reforming temperature. Thermocouple 4 will show 220 oC and thermocouples 1, 2, and 3 will show a range of 160-185 oC. Set methanol and water pumps, and the nitr ogen flow controllers to either Flow 1 or Flow 2 conditions, as summa rized in tables and II. Start-Up Procedure for the Laser

PAGE 66

55 1. Wear laser goggles and turn on laser warning light. 2. This experiment uses a frequency-tripled Nd:YAG laser. 3. Turn on CCD detector: a. Turn on nitrogen purge for CCD. b. Check rotameter for full-scale nitrogen flow c. Turn on Neslab water chiller. d. Turn on CCD controller e. Turn on Pulse generator. 4. Turn on Nd:YAG laser: a. Turn on safety warning light for Nd:YAG laser b. Turn key to on position. c. Nominal front panel settings: i. F5.0 (frequency of flash lamps, 5 Hz) ii. 180 (Q-switch delay, microseconds) iii. P01 (pulse selector, =5Hz) iv. Push run for operating menu. v. Push start/stop button to initiate flash lamps. vi. Push shutter button to initiate laser beam. d. Set laser power to 12.5 J flash lamp pump (~4 mJ/pulse) i. c. Open external shutter to front of laser. 5. Spectroscopy software, Laser and CCD Settings: a. External Trigger: install 355 nm trig ger cable, set tri gger to 0.75 volts, positive. b. Detector Delay: 0.05 us c. DetectorWidth: 0.30 us d. Window: 406 nm center wavelength

PAGE 67

56 e. 250 shots per measurement f. 100 rows binned 6. Verify that all beam dumps, filters, and mirrors are in place according to diagram. 7. To shut down, turn off laser and CCD in reverse order, and run CCD purge gas for 20 minutes. III. Taking Spectroscopic Data: 1. Allow reforming gases to run for 15 minutes after start up to reach steady state. Verify that the reformer is at the initia l height: support screws will be adjusted to their maximum height. Verify that the re former is level along the bed length and across the bed at the reformer supports. Take spectroscopic data as a 250laser pulse averaged spectrum. Lower reactor to the next height below laser. After data is taken for the 5 heights reset the reactor to the original position. IV. Shut Down Procedure for Reformer Turn off water and methanol pumps, and close the methanol and water valves to the reformer. Continue to run nitrogen through the re former for 10 minutes and turn off temperature controllers. Close exit, then entrance valves on reform er, turn of nitrogen flow controllers and power strips, and unplug power Close nitrogen cylinders.

PAGE 68

57 CHAPTER 5 SPATIAL RESOLUTION 5.1 Spatial Resolution of the Optical Set-Up For the current study, the reactor was constr ucted from standard vacuum hardware, with the optical access provided via 6-way crossed, as shown in Figures 5.1 and 5.2. While this provided a flexible and low-cost design, optical resolution was compromised by the rather long optical path of 150 mm. Note that the r eactor bed has a width of only about 25 mm. To assess the optical resoluti on, the reactor was removed and the following measurements were recorded. Figure 5-1 End View of Reformer Optical Path Thermocouple Thermocouple Catalyst Pellets Catalyst Tray Reformer Housing Dead Air Space, x3 locations

PAGE 69

58 Figure 5-2 Side View of Reformer Spatial resolution across the reformer wi dth was measured by placing a jet of hydrogen, from a 4 mm diameter tube, at 4 lite rs/min. at the position of the center axis of the reactor. Spectra were taken at increments on either side of the center axis of the reactor. Negative values in position are locatio ns closer to the lase r source, and positive values are further away. The focal point as the laser is focused through the cylindrical lens aligns with the zero location, as shown below. Using this essentially point-source, Raman spectra of hydrogen were recorded while the jet was translated across a distance equal to the optical path of the reactor. The resulting signals are pres ented in Figure 5-3 Thermocouple Thermocouple Catalyst Pellets Catalyst Tray Reformer Housing Dead Air Space, x3 locations Reactant Flow

PAGE 70

59 Clearly the strongest signal corresponds to the location of the catalyst bed, however additional signal is also present from the dead space between the windows and the catalyst bed. Integration of th is plot reveals that approxi mately 1/3 of the recorded spectral signal during reformation corresponds to the catalyst bed position (between 10mm and +10 mm on Figure 5-3). -20 0 20 40 60 80 100 120 140 -80-60-40-20020406080Spatial ResolutionRaman Signal (a.u.)Position (mm) Figure 5-3 Spatial Resolution 5.2 Calibration for Gas Concentration Prior to analysis of the recorded spectral data, a calibration must be performed to convert the Raman signal to an equivalent molar concentration. This procedure was performed for all three species of interest: H2, H2O, and CH3OH. For all calibration, the catalyst bed was removed, but everything else was fixed in position. While, ideally, the Raman cross-section is independent of temp erature, measurements were recorded for

PAGE 71

60 both ambient and heated conditions to veryify temperature independence. The temperatures given for the calibrations are in the gas phase at the center of the reactor. Overall, the gas phase te mperatures ranged from 155oC to 190oC along the length of the reactor. The thermostat range for resetti ng the temperature cont rollers resulted in a variation of +5oC detected in the gas phase. Hydrogen was calibrated using a nitroge n flow rate of 5 liters/min for each hydrogen flow rate of 0.0, 0.25, 0.50, 1.0 and 2.0 liters/min at 190oC. These measurements were repeated at ambient temperature. The temperature independence and resulting calibration curve for hydrogen is shown in Figures 5.4 and 5.5. A highly linear curve results, with both ambient temperatur e and high temperature values falling on the curve. 0 5 10 15 20 25 30 35 0.0000.0020.0040.0060.0080.0100.012 298 K 450 KHydrogen Signal (P/B)Hydrogen Concentration (mol/l) Figure 5-4. Temperature Inde pendence of Hydrogen Signal

PAGE 72

61 0 5 10 15 20 25 30 35 0.0000.0020.0040.0060.0080.0100.012Hydrogen Signal (P/B)Hydrogen Concentration (mol/l) Y = M0 + M1*x + ... M8*x8 + M9*x9 0.021509 M0 2420.5 M1 20361 M2 0.99973 R Figure 5-5 Hydrogen Calibration Curve Methanol was calibrated at the 190oC with a feed of nitrogen and methanol in the following proportions: 1.5 l/min N2 and 1.4 m l/min liquid methanol,;3 l/min N2 and 1.4 ml/min liquid methanol; 5 l/min N2 and 1.4 m l/min liquid methanol; and 5 l/min N2 and 1 ml/min liquid methanol. As discussed earlier (see Figure 2.1), methanol is characterized by three distinct peaks in the current region of interest, namely, the one near water, labeled P/B A, and a double peak near the le ft end of the spectra, labeled P/B B. The resulting calibration response is shown in Fi gure 5-6 for peaks A and B. As seen in the figure, peak B yields a better response and also avoids n earing the water band. Ultimately, peak B was used for this study.

PAGE 73

62 0 20 40 60 80 100 120 0.0000.0020.0040.0060.0080.010 Meth P/B A Meth P/B B y = 0.7467 + 1160.5x R= 0.9955 y = 1.3468 + 11490x R= 0.99935 Methanol Signal (P/B)Methanol Concentration (mol/l) Figure 5-6 Methanol Calibration Curve Water was calibrated at the 190oC with a feed of nitrogen and water at standard temperature and pressure in th e following proportions: 1.5 l/min N2 and 0.55ml/min water,;3 l/min N2 and 0.55 ml/min water; 5 l/min N2 and 0.55 ml/min water; and 5 l/min N2 and 0.475ml/min water. Though our peristaltic pumps are supposed to be identical, the water pump yielded a non-linear calibration curve, although it was re peatable. Water also showed the greatest variation in calibration spectra, as well as in the reforming spec tra, although such behavior was difficult to li nk to a specific issue.

PAGE 74

63 0 5 10 15 20 0.0000.0020.0040.0060.0080.010 y = 1.5855 + 1968.2x R= 0.98171 Water Signal (P/B)Water Concentration (mol/l) Figure 5-7. Water Calibration Curve

PAGE 75

64 CHAPTER 6 DATA Spectral data during the reaction were take n at 5 heights and at each of the three axial locations. Position 1 refe rs to the location closest to the entrance (x/L =0.08), position 2 refers to the center of the reactor (x/L= 0.5), and position 3 is nearest the exit (x/L=0.92). Data were taken at 5 heights ( up to 6.4 mm) above the catalyst, which was deemed sufficient given the overall resoluti on. Ideally, the mass bala nce of the reaction would include the concentrations from the catalyst surface up to the top of the boundary layer. The concentrations at any specific he ight vary due to differences in rates of diffusion for each species, as well as any reci rculation that may be generated by the dead space near each optical region, as well as by surface roughness created by catalyst pellets that may have been dislodged out of plane. Th e data presented here are in concentration (mole/liter) at particular locations for given flow rates. For each position, height, and flow rate, Raman spectra were recorded in duplicate for each run, using a single spectrum corresponding to a 250-shot average. Each experiment was repeated a minimum of 4 times, for a total of 2000 shots averaged over multiple experimental runs. For the following discussion, results are presented as a function of axial position, height above the catalyst bed, and for the two different flow rates. Of all the experimental runs, data taken at height 5 were inconsistent with trends seen at other heights. It is unclear whether this occurrence is an artifact due to the fl ow and geometric conditions of the reactor. Figure 6-1. shows typical Ra man spectra observed during re action. From Position 1 to

PAGE 76

65 Position 3, the intensities of the spectra for methanol and water both decrease while that of hydrogen increases. 0 100 200 300 400 500 600 700 2500.0003000.0003500.0004000.0004500.000 Flow 2, Position 1 Flow 2, Position 2 Flow 2, Position 3relative intensityRaman Shift (cm-1) Figure 6-1 Typical Raman Spectra Observed Along the Axis of the Reactor 6.1 Variation of Species Concentratio n at Each Position, at Flow 1 The next three graphs (Figures 6-2, 6-3, 64) show the concentrations of the two reactant species, water and methanol, and one product gas, hydrogen, at one position and one flow rate. It can be shown that the con centrations of the reactants decrease along the bed length while hydrogen increases. H2 H-O CH3OH

PAGE 77

66 0 0.005 0.01 0.015 0.02 1.0002.0003.0004.0005.0006.0007.000 H2, position1, F1 CH3OH, position1, F1 H2O, position1, F1concentration, mole/literHeight (mm) Figure 6-2 Position 1, Flow 1 Species Concentration vs Height 0.005 0.006 0.007 0.008 0.009 0.01 0.011 1.0002.0003.0004.0005.0006.0007.000 H2, position2, F1 CH3OH, position2, F1 H2O, position2, F1concentration, mole/literHeight (mm) Figure 6.3 Position 2, Flow 2 Species Concentration vs Height

PAGE 78

67 0 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 1.0002.0003.0004.0005.0006.0007.000 H2, position3, F1 CH3OH, position3, F1 H2O, position3, F1concentration, mole/literHeight (mm) Figure 6-4 Position 3, Flow 3 Species Concentration vs Height 6.2 Variation of Species Concentratio n at Each Position, at Flow 2 At flow 2, which is 1.4 times faster than fl ow 1, there is a more dramatic change in concentrations away from the catalyst surface, especially at position 1. This would reflect the scale of a thinner boundary layer at higher velocities. In both flows 1 and 2, height 1 at position 1 shows similar concentrations of water and methanol, while at height 1 of positions 2 and 3 there is much less water than methanol. 6.3 Variation of Species Concentration alon g Length and Height above Catalyst at Flow 1 In the next three graphs (Figure 6-8, 69, 6-10) show the concentration of each species as it changes along th e length and height above the catalyst. Methanol predictably decreases along the catalyst bed, but shows th e greatest variation in concentration at Position 1.

PAGE 79

68 0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 1.0002.0003.0004.0005.0006.0007.000 H2, position1, F2 CH3OH, position1, F2 H2O, position1, F2concentration, mole/literHeight (mm) Figure 6-5 Position 1, Flow 2 Species Concentration vs Height 0.005 0.006 0.007 0.008 0.009 0.01 0.011 1.0002.0003.0004.0005.0006.0007.000 H2, position2, F2 CH3OH, position2, F2 H2O, position2, F2concentration, mole/literHeight (mm) Figure 6-6 Position 2, Flow 2 Species Concentration vs Height

PAGE 80

69 0 0.002 0.004 0.006 0.008 1.0002.0003.0004.0005.0006.0007.000 H2, position3, F2 CH3OH position3, F2 H2O, position3, F2concentration, mole/literHeight (mm) Figure 6-7 Position 3, Flow 2 Species Concentration vs Height Likewise, water also decreases as expected, sh owing a similar variability at Position 1 as methanol. This would seem to indicate an unsteady flow at Position 1, which may be attributed to the peristaltic pumps. The pumps essentially force increments (i.e. discrete drops) of water through with each turn of the pump wheel, so the flow could possibly have enough variation to be shown by the Raman spectra. There is a decrease in hydrogen concentrations at Position 3. Since bot h water and methanol concentrations are shown to continue decreasing at that position, it is likely that the apparent reduction in hydrogen concentration is a combined effect of back mixing with unreacted species at the end of the reactor and a higher speed of di ffusion for hydrogen. Prev ious studies have cited that reverse water-gas shift reaction ha s been associated with higher temperatures and longer contact times. However, if such a reaction were responsible for the reduced

PAGE 81

70 hydrogen, the water concentrations would have increased. This is not the case, as water and methanol concentrations both decrease mo notonically along the length of the reactor for both flows. Note that the diffusion co efficients of hydrogen, methanol, and water have ratios of 13.3: 2.8: 4.8 respectively, at 465 K, the reactor temperature. 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.018 1234567 CH3OH, position1, F1 CH3OH, position2, F1 CH3OH, position3, F1concentration, mole/literHeight (mm) Figure 6-8 Flow 1 Methanol Concentr ation at 3 Positions vs Height 6.4 Variation of Species Concentration alon g Length and Height above Catalyst at Flow 2 At the higher velocity flow, less back mixi ng occurs at the Position 3, and this is supported by the fact that the hydrogen concentrations at Positions 2 and 3 are much closer together in magnitude. If the flow were faster, it is lik ely that the hydrogen concentration at position 3 would have increas ed as it had from Position 1 to 2. Methanol and water concentrations also decrease from Position 2 to Position 3, but their values are much closer together than in flow 1.

PAGE 82

71 0 0.002 0.004 0.006 0.008 0.01 0.012 1234567 H2, position1, F1 H2, position2, F1 H2, position3, F1concentration, mole/literHeight (mm) Figure 6-9 Flow 1 Hydrogen Concentr ation at 3 Positions vs Height 0 0.005 0.01 0.015 0.02 1234567 H2O, position1, F1 H2O, position2, F1 H2O, position3, F1concentration, mole/literHeight (mm) Figure 6-10 Flow 1 Water Concentra tion at 3 Positions vs Height

PAGE 83

72 At flow 2, it appears that the majority of the reaction occu rs between Positions 1 and 2, then the gradients between Positions 2 and 3 sharply decrease. At flow 1, on the other hand, the differences in concentrations betw een Positions 1 and 2 are more similar to those at Positions 2 and 3. At a lower flow rate, un-reacted species have more time to diffuse through to the surface of the catalyst. See Figures 6-11, 6-12, 6-13. 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 1234567 H2, position1, F2 H2, position3, F2 H2, position2, F2concentration, mole/literHeight (mm) Figure 6-11 Flow 2 Hydrogen Concentr ation at 3 Positi ons vs Height 6.5 Comparison of Concentration Gradients at Flow 1 and Flow 2 At position 1 and flow 2, the concentration of methanol increases with the distance away from the catalyst, but remains fairly cons tant with the height above the catalyst for flow 1 (Figure 6.14). Water shows the same tr end in the subsequent graph. The steepness of the concentration gradient indicates a thinner boundary layer a nd that mass transfer influenced by both the bulk velocity and th e diffusion process. Figure 6-16, of methanol and water at Position 2, shows both flows showing almost the same changes in concentration from one height to the next.

PAGE 84

73 0.004 0.008 0.012 0.016 0.02 0.024 0.028 0.032 1234567 CH3OH, position1, F2 CH3OH position3, F2 CH3OH, position2, F2concentration, mole/literHeight (mm) Figure 6-12 Flow 2 Methanol Concentr ation at 3 Positi ons vs. Height 0 0.01 0.02 0.03 0.04 0.05 0.06 1234567 H2O, position1, F2 H2O, position3, F2 H2O, position2, F2concentration, mole/literHeight (mm) Figure 6-13 Flow 2 Water Concentra tion at 3 Positions vs Height

PAGE 85

74 Both flows of methanol (blue and black markers) follow a similar trend, and the higher velocity flow methanol has a great er over all concentra tion due the reduced contact time, as compared to flow 1. Similarly, hydrogen for both flows follows the same gradient in height above the catalyst with the flow 2 hydrogen having a lower concentration overall. At position 3 (Figure 6-18 ), the difference the two flow rates have on reactant concentration is most apparent. At the lower velocity, methanol is about half the concentration at the higher velocity. 6.6 Species Gradients along Reactor Leng th and Height above the Catalyst The next section serves to illustrate how concentrations of products and reactants increase or decrease along the reactor. In order to include all heights, error bars were omitted to avoid obscuring data, and each species was s hown separately for each flow. For hydrogen at flow 1, the concentration reaches a maxi mum at Position 2, but generally decreases with height above the catalyst. In the methanol flows, where the concentration is greatest at the entrance, it is also l east at the height 1 where it is being decomposed on at the surface. In both flows, the methanol concen tration sharply decreases along the reactor length, as well as showing a more uniform c oncentration with height, near the exit. Water shows a similar trend to methanol, for both flows, except at the entrance where the concentration varies the most in th e cross section of the flow. As the water is consumed in the reaction, both flows go to a greatly reduced value, and become more uniform in the cross section. Hydrogen incr eases along the reactor length from position 1 to 2 for both flows, and in Positions 2 and 3 a ppears to level off (in Flow 2), or decrease (in Flow 1). Considering the much higher di ffusion coefficient of hydrogen, as well as the volume of dead space within the reactor, this phenomena is more likely due to the circulation caused by the physical geometry of the reactor.

PAGE 86

75 0 0.005 0.01 0.015 0.02 0.025 0.03 1.0002.0003.0004.0005.0006.0007.000 H2, position1, F2 CH3OH, position1, F2 H2, position1, F1 CH3OH, position1, F1concentration, mole/literHeight (mm) Figure 6-14 Hydrogen and Methanol, Flows 1 and 2, Position 1 0 0.005 0.01 0.015 0.02 0.025 0.03 1.0002.0003.0004.0005.0006.0007.000 H2, position1, F2 H2O, position1, F2 H2, position1, F1 H2O, position1, F1concentration, mole/literHeight (mm) Figure 6-15 Hydrogen and Water, Flows 1 and 2, Position 1

PAGE 87

76 0.005 0.006 0.007 0.008 0.009 0.01 0.011 0.012 1.0002.0003.0004.0005.0006.0007.000 H2, position2, F2 CH3OH, position2, F2 H2, position2, F1 CH3OH, position2, F1concentration, mole/literHeight (mm) Figure 6-16 Hydrogen and Methanol, Flows 1 and 2, Position 2 0.005 0.006 0.007 0.008 0.009 0.01 0.011 1.0002.0003.0004.0005.0006.0007.000 H2, position2, F2 H2O, position2, F2 H2, position2, F1 H2O, position2, F1concentration, mole/literHeight (mm) Figure 6-17 Hydrogen and Water, Flows 1 and 2, Position 2

PAGE 88

77 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 1.0002.0003.0004.0005.0006.0007.000 H2, position3, F2 CH3OH position3, F2 H2, position3, F1 CH3OH, position3, F1concentration, mole/literHeight (mm) Figure 6-18 Hydrogen and Methanol, Flows 1 and 2, Position 3 0 0.002 0.004 0.006 0.008 1.0002.0003.0004.0005.0006.0007.000 H2, position3, F2 H2O, position3, F2 H2, position3, F1 H2O, position3, F1concentration, mole/literHeight (mm) Figure 6-19 Hydrogen and Water, Flows 1 and 2, Position 3

PAGE 89

78 Decomposition of hydrogen through a subsequent reverse reaction is unlikely because of the low temperature. As mentioned earlier, if the reverse water-gas shift reaction were occurring, it is expected that water would ha ve increased at the end of the reactor, however this was not observed. 0 0.002 0.004 0.006 0.008 0.01 0.012 0.511.522.533.5 height 1, H2 height 2, H2 height 3, H2 height 4, H2 height 5, h2concentration, mole/literPosition Figure 6-20: Hydrogen Flow 1 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.511.522.533.5 height 1, methanol height 2, methanol height 3, methanol height 4, methanol height 5 methanolconcentration, mole/literPosition Figure 6-21: Methanol Flow 1

PAGE 90

79 0 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.5001.0001.5002.0002.5003.0003.5 0 H2O, height1, F1 H2O, height2, F1 H2O, height3, F1 H2O, height4, F1 H2O, height5, F1concentration, mole/literPosition Figure 6-22 Water, Flow 1 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.009 0.511.522.533.5 height 1, hydrogen height 2, hydrogen height 3, hydrogen height 4, hydrogen height 5, hydrogenconcentration, mole/literPosition Figure 6-23: Hydrogen Flow 2

PAGE 91

80 0.005 0.01 0.015 0.02 0.025 0.03 0.511.522.533.5Flow 2, Methanol height 1, methanol height 2, methanol height 3, methanol height 4, methanol height 5, methanolconcentration, mole/literPosition Figure 6-24: Methanol Flow 2 0 0.005 0.01 0.015 0.02 0.025 0.5001.0001.5002.0002.5003.0003.500 H2O, height1, F2 H2O, height2, F2 H2O, height3, F2 H2O, height4, F2 H2O, height5, F2concentration, mole/literPosition Figure 6-25: Water, Flow 2 6.7 Species Concentration vs Position at Height 1 The following figures compare the changes in concentration of each species at the two different flow rates, along axial position in the reactor.

PAGE 92

81 To further verify the validity of the measurements taken in this study, a mass balance of hydrogen atoms was conducted, comparing the mole fraction of hydrogen atoms at the inlet to the average of hydroge n mole fractions across the five heights at Position 3. Bear in mind that the measuremen ts taken above the catalyst surface do not encompass the entire boundary layer, that hydrogen diffuses much faster than other constituents, and that the signal over the ca talyst is diluted by the dead space in the optical regions of the reactor. As a result, at Flow 2, there was very good agreement, with only a difference of 7%: 0.051 mole H/l at th e inlet and 0.0474 mole H/liter average at Position 3. However, at Flow 1, which ha s a thicker boundary layer, there was a difference of 46%. 0 0.002 0.004 0.006 0.008 0.01 0.012 0.5001.0001.5002.0002.5003.0003.500 H2, height1, F1 H2, height1, F2concentration, mole/literPosition Figure 6-26: Hydrogen, Flow 1 and 2

PAGE 93

82 0 0.005 0.01 0.015 0.02 0.5001.0001.5002.0002.5003.0003.500 H2O, height1, F1 H2O, height1, F2concentration, mole/literPosition Figure 6-27: Water, Flow 1 and 2 0.002 0.004 0.006 0.008 0.01 0.012 0.014 0.016 0.5001.0001.5002.0002.5003.0003.500 CH3OH, height1, F1 CH3OH, height1, F2concentration, mole/literPosition Figure 6-28: Methanol, Flow 1 and 2

PAGE 94

83 CHAPTER 7 SUMMARY AND CONCLUSIONS This study was undertaken to explore the feasibility of in situ measurements of methanol reformation using Raman spectrosc opy, with a goal of a ssessing concentration boundary layers and the overall degree of r eaction completion with a suitable residence time-scale. To date, considerable research ha s been made of methanol reformation; such studies generally utilize a control-volume a pproach, with only input reactants and output products measured. Until now, other in situ studies have investigated the surface chemistry of catalysis, specifically the adsorpti ve energies of species and the interactions of the catalyst microstructure during reacti on. While much has been learned from these control volume and surface science studies, they have limited use for validation of detailed kinetic models of heterogeneous and gas-phase chemistry. This is especially true considering findings in literat ure that the longer residence times, or reactor length for a given flow rate, back mixing, and non-isothermality in packed bed reactors can lead to reverse reactions and skewed apparent rate laws. The efficacy of monolith reactors has been an effort for development in other studies, though they too are control volume studies of the chemistry alone. This investigation is unique in that it seeks to quantify some of the fluid dynamic aspects of the re actant flow through a monolith reactor, in addition to studying the chemistry of reforma tion. Reported herein are some of the first in situ Raman measurements of gas-phase species in a catalytic reactor. Key results of the current study are the su ccessful measurements of the reactants methanol and water and the product hydroge n via Raman spectroscopy. Calibration was

PAGE 95

84 used to convert measured spectra to absolute molar concentrations, which revealed a significant conversion of methanol to hydrogen over the residence times (6 to 8 s) used in this study. While the reactor design was not optimized for spatial resolution in spectroscopic measurement, spatial gradients (millimeter-scale) were observed above the catalyst surface under reforming conditions. Th e main findings of this study are as follows: 1. The Katalco 33-5 (Johnson Matthey) catalys t pellets were an excellent lowtemperature (~200oC) reforming catalyst. These pellets functioned well when configured in a single-layer pellet be d for a laminar flow reactor. This configuration enabled a two-dimensional assessment of the reforming process, providing information about the boundary layer via height measurements, and about the overall kinetics us ing the axial measurements. 2. Raman spectroscopy using 355-nm excitation was successfully implemented for the in situ measurement of hydrogen, methanol and water vapor during actual methanol reformation. While Raman spect roscopy is a techni que characterized overall by relatively low sensitivity as co mpared to other methods, such as laserinduced fluorescence, significant signal levels for the three targeted species were recorded under all reforming conditions. 3. The optical access design described here proved successful for in situ Raman analysis. The use of UV-grade fused s ilica optical windows successfully minimized background fluorescence with 355-nm excita tion. Since the current reactor was based on low-cost, low-build-time design criteria, spatial resolution was not optimized. As a result, Raman signals were diluted by a significant factor with respect to the expected actual gradients above the catalyst surface. That is to say, the scale of the crosses for the optical windows were such that there were significant volumes of non-reacting dead space that interfered with the Raman signal of the gas phase over the reactor. This was a major source of noise and prevented further refinement of using Raman spectroscopy as an in situ diagnostic in reforming. The fact that consistent Ra man spectra were detected despite these detractors in the reactor design only serv es to emphasize the feasiblity of Raman spectroscopy for gas phase analysis.

PAGE 96

85 CHAPTER 8 RECOMMENDATIONS FOR FUTURE WORK This study identified design and implementa tion considerations for a monolith-type reformer. Several improvements can be made in the design to improve resolution and accuracy and, were these improvements to be implemented, opportunities would exist for more accurate characterization the gas phase above catalysts, in conjunction with flow rates and geometric scales. 1. Spatial resolution along the optical line of s ite is integral to accurately probing the boundary layer, and this can be accomplishe d by eliminating the dead air space between the optical windows and the catal yst bed. To do this, a custom design would incorporate optical windows in the re actor housing wall, with a catalyst bed the same width as the reactor This would reduce the noise that occurs due to the volume of the optical window cr osses in the current design. 2. Steady flow of reactants would eliminate much of the variability seen at the entrance of the reactor. The discrete flow of the peristaltic pumps resulted in significant fluctuations at the entrance of the reactor. The flow variation was 1-2 seconds, depending on the flow rate setting, with the average flow rate over the residence time being the assigned flow ra tes given as Flow 1 and Flow 2. The variation in flow was caused by the rpm of the pump as the pump wheel pushes a discrete amount of fluid through the tube. 3. Effects of unsteady reactant flows, catalyst pellets that are out of plane, and end effects at the entrance and exit of the catalyst, though realistically occurring in commercial reactors, would each have to be studied separately. 4. A more ideal, large scale monolith reactor would be to have two catalyst-coated plates such that Pouisiuelle flow is cr eated between them. This would be more similar to the internal struct ure of an actual monolith. If the catalyst coated surfaces within the reactor were interchangeable, th en this configurati on could be used to characterize catalyst materials, as well as determine ideal separation of the surfaces, as well as flow rates and temperatures, fo r desired rates of re action. In this way, scale parameters for monolith designs could be developed. 5. No reverse reactions were observed in this study, however, a greater range of residence times and temperatures would have to be studied to determine an

PAGE 97

86 optimum range of operation. Reverse reacti ons have been associated with higher catalyst temperatures and longer residence times. The water-gas shift reaction is active over the same catalyst, and if the steam-methanol reforming reaction occurs via the methyl formate intermediate, then the occurrence of CO would be due to reverse water-gas sh ift reaction. Consequently, eith er additional steam would be needed to convert the CO with another water-gas shift, or the products of the reaction must be quickly removed from th e catalyst surface to prevent the reverse reaction in the first place. In this re gard, mass diffusion limitations and residence time contribute to reverse reactions. In the case of increased temperature, the higher temperature product gases collide with su rfaces with greater frequency, and as the product partial pressures increase with the reaction, there is increased probability of reverse reactions. It is impor tant to bear in mind that adsorption and desorption of species on the catalyst surface is a proce ss that requires minimal energy, which is why readsorption of product gases, and s ubsequent reverse reactions, occur more easily than in homogeneous gas-phase chemis try. Water gas shift, reverse water gas shift, methanol decomposition, and steam me thanol reforming can all occur on this type of catalyst, offering a variety of possible reaction pathways, so controlling mass transfer to and from surface sites is important. 6. More accurate rate laws for catalytic su rfaces can be determined with this design configuration. The concentrations measur ed at each position are essentially instantaneous molar ratios, and the positions along the reactor length can be resolved temporally, knowing the velocity and volume of the reactor. With a wellunderstood velocity profile, and with end c onditions resolved, concentrations taken along the axis of the reactor would de scribe the incremental change in concentration per time. In situ studies in this manner would be able to assess more accurately the onset of non-Arrhenius behavior in catalytic reactions.

PAGE 98

87 APPENDIX DATA SUMMARIES This appendix contains the data summaries for hydrogen, water, and methanol Raman spectra. The species concentration is give n as the average, in moles per liter, as determined from the raman spectral data and the calibration cu rves. The standard deviations and relative standard deviations are given. Flow 1 and Flow 2 values are given in Tables 4.1 and 4.2. Position 1 is located at x/L =0.09, Position 2 is at x/L= 0.5, and Position 3 is at x/L=0.82. The distance a bove the catalyst surface, Height 1, is approximately 1.25 mm; Height 2, 2.5 mm; He ight 3, 3.8 mm; Height 4, 5.1 mm; Height 5, 6.35 mm.

PAGE 99

88 Table A-1. Hydrogen Data Summary. Hydrogen (mol/l) Flow Height Position Average StDev RSD 1 1 1 1.63E-03 5.46E-04 33.5 1 2 1 1.82E-03 5.79E-04 31.8 1 3 1 1.81E-03 3.76E-04 20.7 1 4 1 1.77E-03 5.57E-04 31.5 1 5 1 1.41E-03 4.63E-04 32.9 2 1 1 1.99E-03 4.38E-04 22.0 2 2 1 2.39E-03 6.98E-04 29.2 2 3 1 2.83E-03 3.12E-04 11.0 2 4 1 2.73E-03 5.84E-04 21.4 2 5 1 2.08E-03 5.52E-04 26.5 1 1 2 1.02E-02 1.57E-03 15.3 1 2 2 9.18E-03 1.43E-03 15.6 1 3 2 8.07E-03 1.51E-03 18.8 1 4 2 8.40E-03 8.87E-04 10.6 1 5 2 8.11E-03 1.78E-03 22.0 2 1 2 8.08E-03 9.06E-04 11.2 2 2 2 7.49E-03 1.45E-03 19.3 2 3 2 6.58E-03 9.71E-04 14.8 2 4 2 6.37E-03 9.50E-04 14.9 2 5 2 7.07E-03 1.22E-03 17.3 1 1 3 6.57E-03 2.47E-03 37.6 1 2 3 6.15E-03 9.93E-04 16.2 1 3 3 5.50E-03 1.21E-03 21.9 1 4 3 4.35E-03 7.63E-04 17.5 1 5 3 3.85E-03 3.64E-04 9.5 2 1 3 7.68E-03 8.67E-04 11.3 2 2 3 7.71E-03 1.05E-03 13.7 2 3 3 7.43E-03 1.63E-03 22.0 2 4 3 7.42E-03 2.23E-03 30.1 2 5 3 5.89E-03 2.24E-03 38.0

PAGE 100

89 Table A-2. Methanol Data Summary Methanol (mol/l) Flow Height Position Average StDev RSD 1 1 1 0.0098180.00283528.87355 1 2 1 0.0139550.0041329.59285 1 3 1 0.0129460.00231217.86036 1 4 1 0.0138090.00298521.61706 1 5 1 0.0097580.00287829.49025 2 1 1 0.0149810.00183912.27357 2 2 1 0.0192090.00679935.39807 2 3 1 0.0267810.00405715.14849 2 4 1 0.0271970.00608222.36423 2 5 1 0.0215950.00367517.01606 1 1 2 0.009110.00184520.25148 1 2 2 0.008230.00163619.88287 1 3 2 0.0069990.00197228.17427 1 4 2 0.007820.00120615.41797 1 5 2 0.007480.00189425.31673 2 1 2 0.0093970.00100410.68457 2 2 2 0.0090650.00212123.39339 2 3 2 0.0079070.00153319.38787 2 4 2 0.0077130.00162521.06962 2 5 2 0.0081470.00151318.57486 1 1 3 0.0038330.00106327.74715 1 2 3 0.0037580.00072519.30516 1 3 3 0.0033710.00100129.69631 1 4 3 0.0029250.00072324.71729 1 5 3 0.0027290.00049818.2398 2 1 3 0.007280.000810.98548 2 2 3 0.0073050.00115715.83687 2 3 3 0.0070990.00154921.82156 2 4 3 0.0069510.00200328.80956 2 5 3 0.005540.00189434.19093

PAGE 101

90 Table A-3. Wate r Data Summary. Water (mol/l) Flow Height Position Average StDev RSD 1 1 1 0.0120 0.00493741.14 1 2 1 0.0149 0.00217214.61 1 3 1 0.0151 0.00711347.07 1 4 1 0.0149 0.00477632.14 1 5 1 0.0101 0.00241323.90 2 1 1 0.0136 0.00169412.46 2 2 1 0.0168 0.00499829.82 2 3 1 0.0223 0.00306413.73 2 4 1 0.0248 0.00478719.32 2 5 1 0.0196 0.00299915.31 1 1 2 0.0064 0.00210432.65 1 2 2 0.0066 0.00117217.81 1 3 2 0.0060 0.00145924.25 1 4 2 0.0072 0.00241133.28 1 5 2 0.0074 0.00200127.07 2 1 2 0.0067 0.00054 8.07 2 2 2 0.0067 0.00153522.80 2 3 2 0.0059 0.00105417.86 2 4 2 0.0061 0.00130521.30 2 5 2 0.0069 0.00122217.66 1 1 3 0.0030 0.003398113.89 1 2 3 0.0016 0.00044727.77 1 3 3 0.0014 0.00053338.26 1 4 3 0.0013 0.00043 34.04 1 5 3 0.0008 0.00025730.84 2 1 3 0.0026 0.00029811.61 2 2 3 0.0034 0.00160847.55 2 3 3 0.0028 0.00074927.06 2 4 3 0.0026 0.00087534.13 2 5 3 0.0027 0.002777103.05

PAGE 102

91 LIST OF REFERENCES Agarwal, V. ; Sanjay, P.; Pant, K. K. H2 Production by Steam Reforming of Methanol over Cu/ZnO/Al2O3 Catalysts: Transient Deactivation Kinetics Modeling. Applied Catalysis A: General. In Press. Agrell, J.; Birgersson, H.; Boutonnet, M. Steam Reforming of Methanol over a Cu/ZnO/Al2O3 Catalyst: a Kinetic Analysis and Strategies for Suppression of CO Formation. Journal of Power Sources. 106 (2002) 249-257. Agrell, J.; Boutonnet, M.; Me lian-Cabrera, I; Fierro, J. L.G. Production of Hydrogen from Methanol over Binary Cu/ZnO Cataly stsPart I. Catalyst Preparation and Characterization. Applied Cata lysis A: General 253 (2003) 201-211. Agrell, J.; Boutonnet, M.; Fierro, J.L.G.; Production of Hydrogen from Methanol over Binary Cu/ZnO CatalystsPart II. Catalytic Activity and Reaction Pathways. Applied Catalysis A: General 253 (2003) 213-223. Agrell, J.; Birgersson, H.; Boutonnet, M.; Me lian-Cabrera, I; Navarro, R.M.; Fierro, J.L.G. Production of Hydrogen from Meth anol over Cu/ZnO Catalysts Promoted by ZrO2 and Al2O3. Journal of Catalysis 219 (2003) 389-403. Amos, W. Costs of Storing and Trans porting Hydrogen. National Renewable Energy Laboratory. Golden, CO. 1998. Atkins, P.; de Paula, J. Physical Chemistry W.H. Freeman and Company. 7th ed. 2002. Ball, A.J. Investigation of Gaseous Hydr ogen Leak Detection Using Raman Scattering and Laser-Induced Breakdown Spectroscopy. Masters Thesis. University of Florida. 2005. Bravo, J.; Karim, A.; Conant, T.; Lopez, G.P; Datye, A. Wall Coating of a CuO/ZnO/Al2O3 Methanol Steam Refo rming Catalyst for Micro-Channel Reformers. Chemical Engineering Journal 101 (2004) 113-121. Chen, C.; Carolan, M.; Rynders, S. ITM S yngas and ITM H2: Engineering Development of Ceramic Membrane Reactor Systems for Converting Natural Gas To Hydrogen and Synthesis Gas for Liquid Transportati on Fuels, Air Products and Chemicals. National Renewable Energy Labor atory Progress Report, 2002. Choudhary, T.V.; Goodman, D. W. CO-Free Pr oduction of Hydrogen via Stepwise Steam Reforming of Methane. Journa l of Catalysis 192, 316-321 (2000)

PAGE 103

92 Choudhary, T. V.; Sivadinarayana, C.; Chusue i, C. C.; Klinghoffer, A.; Goodman, D.W. Hydrogen Production via Catalytic Deco mposition of Methane. Journal of Catalysis 199, 9-18 (2001). Choudhary, V. R.; Banerjee, S.; Rajput, A. M. Continuous Production of H2 at Low Temperature from Methane Decomposition over Ni-Containing Catalyst Followed by Gasification by Steam of the Carbon on the Catalyst in Two Parallel Reactors Operated in Cyclic Manner. Jour nal of Catalysis. 198, 136-141 (2001). Choi, Y.; Stenger, H.G. Water Gas Shift Reaction Kinetics and Reactor Modeling for Fuel Cell Grade Hydrogen. Journal of Power Sources 124 (2003) 432-439. Choi, Y.: Stenger, H.G. Kinetics, Simula tion and Optimization of Methanol Steam Reformer for Fuel Cell A pplications. Journal of Po wer Sources 142 (2005) 81-91. dAlnoncourt, R.N.; Kurtz, M.; Wilmer, H.; Lo ffler, E.; Hagen, V.; Shen, J.; Muhler, M. The Influence of ZnO on the Differen tial Heat of Adsorp tion of CO on Cu Catalysts: A Microcalorimetric Study. J ournal of Catalysis. 220 (2003) 249-253. Dams, R. A. J.; Hayter, P. R.; Moore, S. C. Continued Development of a Mathematical Model for A Methanol Reformer. ETS U/ F/02/00152REP. Contractor: Wellman CJB Ltd. Crown Copyright 2001. Dams, R. A. J.; Hayter, P. R.; Moore, S. C. Long Term Testing of a Methanol Fuel Processor. ETSU/F/02/00181/REP. DTI/ Pub URN 02/800. Contractor: Wellman Defense Ltd. Crown Copyright 2002. Demirel, Y.; Sharma, R.N.; Al-Ali, H.H. On th e Effective Heat Transfer Parameters in a Packed Bed. International Journal of Heat and Mass Transfe. 43, (2000) 327-332. Fierro, G.; Lo Jacono, M.; Inversi, M.; Porta, P.; Cioci, F.; Lavecchia, R. Study of the Reducibility of Copper in CuO-ZnO Ca talysts by Temperature-programmed Reduction. Applied Catalysis A: General. 137 (1996) 327-348. Fukuhara, C.; Ohkura, H.; Kamata, Y.; Murakami Y.; Igarashi, A. Catalytic Properties of Plate-Type Copper-Based Catalysts, for Steam Reforming of Methanol, on an Aluminum Plate Prepared by Electroless Pl ating. Applied Catalysis A: General 273 (2004) 125-132. Giroux, T.; Hwang, S.; Liu,Y.; Ruettinger, W. ; Shore, L. Monolithic Structures as Alternatives to Particulate Catalysts for the Reforming of Hdyrocarbons for Hydrogen Generation. Applied Catalysi s B: Environmental. 56 (2005) 95-100. Gunter, M.M.; Ressler, T.; Jentoft, R.E.; Be ms, B. Redox Behavior of Copper Oxide/Zinc Oxide Catalysts in the Steam Reforming of Methanol Studied by in Situ X-Ray Diffraction and Absorption Spectroscopy. Journal of Catalysis 203 (2001) 133-149.

PAGE 104

93 Grunwaldt, J.D.; Molenbroek, A.M.; Topsoe, N.Y.; Topsoe, H.; Clausen, B.S. In Situ Investigations of Structural Changes in Cu/ZnO Catalysts. Journal of Catalysis. 194. (2000) 452-460. Haw, J. F., ed In-Situ Spectroscopy in Heterogeneous Catalysis Weinheim: Wiley-VCH. 2002. Horny, C.; Kiwi-Minsker, L.; Renken, A. Micro-structured String Reactor for Autothermal Production of Hydrogen. Chemi cal Engineering Journal 101 (2004) 39. Huang, X.; Ma, L.; Wainwright, M.S. The Influence of Cr, Zn, and Co Additives on the Performance of Skeletal Copper Catalyst s for Methanol Synthesis and Related Reactions. Applied Catalysis A: General 257 (2004) 235-243. Hu, J.; Wang, Y.; VanderWiel, D.; Chin, C.; Palo, D. Rozmiarek, R.; Dagle, R.; Cao, J.; Holladay, J.; Baker, E. Fuel Processing fo r Portable Power Applications. Chemical Engineering Journal. 93 (2003) 55-60. Ingle, J.D. Jr.; Crouch, S. R. Spectroch emical Analysis. Prentice-Hall, Inc. 1988. Karim, A.; Bravo, J.; Datye, A. Nonisoth ermality in Packed Bed Reactors for Steam Methanol Reforming of Methanol. App lied Catalysis A: General 282 (2005) 101109. Kawamura, Y.; Yamamoto, K.; Ogura, N.; Katsumata, T.; Igarashi, A. Preparation of Cu/ZnO/Al2O3 Catalyst for a Micro Methan ol Reformer. Journal of Power Sources (2005). Knop-Gericke, A.; Havecker, M.; Schedel-Nied rig, T.; Schlogl, R. Characterization of Active Phases of Copper Catalyst fo r Methanol Oxidation Under Reaction Conditions: an In Situ X-Ray Absorption Spectroscopy Study in the Soft Energy Range. Topics in Catalysis. Vol. 15. No. 1 (2001) 27-34. Knozinger, H.; Mestl, G. Laser Raman Sp ectroscopyA Powerful Tool for In Situ Studies of Catalytic Materials. T opics in Catalysis. 8(1999) 45-55. Kudelski, A. Raman Study on Methanol Partial Oxidation and Oxidative Steam Reforming over Copper. Surface Science. 566-568 (2004) 1007-1011. Kubek, D., Sharp, C., Kuper, D., Clark, M., Di Dio, M., Whysall, M. Recent Selexol, PolySep, and PolyBed Operating Experience with Gasification for Power and Hydrogen. UOP Gas Processing Presen tation, Gasification Technologies Conference 2002. Los Angeles, CA, USA.

PAGE 105

94 Kumari, V.D.; Subrahmanyam, M.; Ratnamala, A.; Venugopal, D.; Srinivas, B.; Sharma, M.V.P. Madhvendra, S.S.; Bikshapathi, B; Venkateswarlu, K.; Krishnudu, T.; Prasad, K.B.S.; Raghavan, K.V. Correlation of Activity and Stability of CuO/ZnO/Al2O3 Methanol Steam Reform ing Catalysts with Cu/Zn Composition Obtained by SEM-EDAX Analysis. Ca talysis Communications. 3 (2002) 417-424. Lee, J.K.; Ko, J.B.; Kim, D.H. Methanol Steam Reforming over Cu/ZnO/Al2O3 Catalyst: Kinetics and Effectiveness Factor. Applie d Catalysis A: General 278 (2004) 25-35. Lim, M. S.; Kim, M. R.; Noh, J.; Woo, S.I. A Plate-Type Reactor Coated with Zirconiasol and Catalyst Mixture for Methanol Steam-Reforming. Journal of Power Sources 140 (2005) 66-71. Lindstrom, B.; Pettersson, L.J. ; Deactivation of C opper-based Catalysts for Fuel Cell Applications. Catalysis Letters 74 (2001) 27-30. Lindstrom, B.; Agrell, J.; Pettersson, L.J. Combined Methanol Reforming for Hydrogen Generation over Monolithic Catalysts. Ch emical Engineering Journal. 93 (2003) 91-101. Litynski, J.T.; Klara, S.M.; McIlvried, H. G.; Srivastava, R.D. The United States Department of Energys Regional Carbon Sequestration Partnerships Program: A Collaborative Approach to Carbon Manage ment. Environment International. In Press. Loffler, D.G.; McDermott, S.D. Renn, C.N. Ac tivity and Durability of Water-Gas Shift Catalysts Used for Steam Reforming of Methanol. Journal of Power Sources. 114 (2003) 15-20. Masel, R. I. Chemical Kinetics and Catalysis Wiley. 2001. Masel, R.I Principles of Adsorption and Reaction on Solid Surfaces Wiley. 1996. Myers, D., Ariff, G., James, B., Lettow, J, Thomas, C., Kuhn, R. Cost and Performance Comparison of Stationary Hydrogen Fueli ng Appliances. Directed Technologies, U.S. DOE Hydrogen Program Review 2002. Arlington, VA. Nakamura, I.; Fujitani, T.; Uchijima, T.; Naka mura, J. The Synthesis of Methanol and the Reverse Water-Gas Shift Reaction over Zn-deposited Cu (100) and Cu (110) Surfaces: Comparison with Zn/Cu(111). Surface Science. 400 (1998) 387-400. Ogden, J. M. Review of Sma ll Stationary Reformers for Hydrogen Production. Center for Environmental Studies, Princeton Univ ersity. International Energy Agency. 2002. Princeton, NJ. Padro, C., Putsche, V. Survey of the Economics of Hydrogen Technologies. National Renewable Energy Laboratory. 1999. Golden, CO.

PAGE 106

95 Pan, L.; Wang, S. Methanol Steam Reforming in a Compact Plate-Fin Reformer for Fuel Cell Systems. International Journa l of Hydrogen Energy. 30 (2005) 973-979. Park, G-G.; Yim, S.-D.; Yoon, Y.-G.; Lee, W. -Y.; Kim, C.-S.; Seo, D.-J.; Eguchi, K. Hydrogen Production with Integrated Microc hannel Fuel Processor for Portable Fuel Cell Systems. 145 (2005) 702-706. Park, S; Wasileski, S.A.; Weaver, M.J. So me Interpretations of Surface Vibrational Spectroscopy Pertinent to Fu el-cell Electrocatalysis. Electrochimica Acta 47 (2002) 3611-3620. Peppley, B.A.; Amphlett, J.C.; Kearns, L.M.; Mann, R.F. Methanol-steam Reforming on Cu/ZnO/Al2O3. Part 1: The Reaction Netw ork. Applied Catalysis A: General. 179 (1999) 21-29. Peppley, B.A.; Amphlett, J.C.; Kearns, L.M.; Mann, R.F. Methanol-steam Reforming on Cu/ZnO/Al2O3 Catalysts. Part 2: A Comprehensive Kinetic Model. Applied Catalysis A: General. 179 (1999) 31-49. Prasad, V.S.; Das, S.K. An Experimental and Theoretical Investigation into the Hyperbolic Nature of Axial Dispersion in Packed Beds. International Journal of Heat and Mass Transfer. 45 (2002). 3681-3688. Purnama, H.; Ressler, T.; Jentoft, R.E.; So erijanto, H.; Schlogl, R.; Schomacker, R. CO Formation/Selectivity for Steam Reform ing of Methanol with a Commercial CuO/ZnO/Al2O3 Catalyst. Applied Cata lysis A: General. 259 (2004) 83-94. Rizeq, G.; West, J.; Frydman, A.; Subia, S.; Zamansky, V.; Das, K. Advanced Gasification-Combustion Technology fo r Production of Hydrogen, Power and Sequestration-Ready CO2. GE Global Res earch, U.S. DOE NETL, Gasification Technologies Conference, 2003. San Francisco, CA. Roan, V.P.; Betts, D.; Twining, A.; Dinh, K.; Wassink, P.; Simmons, T. An Investigation of the Feasibility of Coal-Based Methanol for Application in Transportation Fuel Cell Systems. Submitted to Georgetown University. 2004. Rozovskii, A.Y. Mechanism and Kinetic Reactions of C1 Molecules on Cu-Based Catalysts. Kinetics and Cata lysis. Vol.44. No.3. (2003) 360-378. Saito, M.; Wu, J.; Tomoda, K.; Takahara, I.; Murata, K. Effects of ZnO Contained in Supported Cu-based Catalysts on Their Acti vities for Several Reactions. Catalysis Letters. Vol. 83. No 1-2. (2002) 1-4. Shackleford, J. F. Introduction to Material s Science for Engineers Prentice Hall. 1992.

PAGE 107

96 Shen, G.C.; Fujita, S.; Matsumoto, S.; Takezawa, N. Steam Reforming of Methanol on Binary Cu/ZnO Catalysts: Effects of Preparation Condition upon Precursors, Surface Structure, and Catalytic Activ ity. Journal of Molecular Catlysis A: Chemical. 124 (1997) 123-136. Shishido, T; Yamamoto, Y.; Morioka, H.; Ta kaki, K.; Takehira, K. Active Cu/ZnO and Cu/ZnO/Al2O3 Catalysts Prepared by Homogeneous Precipitation Method in Steam Reforming of Methanol. Applie d Catalysis A: Ge neral 263 (2004) 249-253. Springmann, S.; Friedrich, G.; Himmen, M.; Sommer, M.; Eigenberger, G. Isothermal Kinetic Measurements for Hydrogen Pr oduction from Hydrocarbon Fuels Using a Novel Kinetic Reactor Concept. Applied Catalysis A: General. 235 (2002) 101111. Springmann, S.; Bohnet, M.; Docter, A.; Lamm, A.; Eigenberger, G. Cold Start Simulations of a Gasoline Based Fuel Pro cessor for Mobile Fuel Cell Applications. Journal of Power Sources. 128 (2004) 13-24. Stair, P.C.; Chua, Y.T. A Novel Fluidi zed Bed Technique for Measuring UV Raman Spectra of Catalysts and Adorbates. Journal of Catalysis. 196. (2000) 66-72. Stair, P.C. Chua, Y.T. An Ultraviolet Rama n Spectroscopic Study of Coke Formation in Methanol to Hydrocarbons Conversion over Zeolite H-MFI. Journal of Catalysis. 213 (2003) 39-46. Tanaka, Y.; Utaka, T.; Kikuchi, R.; Sasaki, K.; Eguchi, K. Water Gas Shift Reaction over Cu-based Mixed Oxides for CO Remova l from the Reformed Fuels. Applied Catalysis A: General. 242 (2003) 287-295. Takeguchi, T.; Kani, Y.; Inoue, M.; Eguchi, K. Steam Reforming of Methanol on Copper Catalysts on Large-Surface-Area ZnAl2O 3. Catalysis Letters. Vol. 83. No 1-2. (2002) 49-53. Terazaki, T.; Nomura, M.; Takeyama, K.; Na kamura, O.; Yamamoto, T. Development of Multi-Layered Microreactor with Methanol Reformer for Small PEMFC. Journal of Power Sources. (2005) In Press. Trimm, D.L.; Onsan, Z.I. Onboard Fuel Conversion for Hydroge n-Fuel-Cell-Driven Vehicles. Catalysis Reviews. 43(1&2), 31-84 (2001). Twigg, M.V.; Spencer, M.S. Deactivation of Copper Metal Catalysts for Methanol Decomposition, Steam Methanol Reforming, and Methanol Synthesis. Topics in Catalysis. Vol 22. Nos 3-4. (2003) 191-203. Wachs, I.E. In Situ Raman Spectroscopy Studies of Catalysts. Topics in Catalysis. 8 (1999) 57-63.

PAGE 108

97 Waugh, K.C. The Absorption and Locking-in of Hydrogen in Copper. Solid State Ionics. 168 (2004) 327-342. Weaver, M.J. Raman and Infrared Spectrosc opies as In Situ Probes of Catalytic Adsorbate Chemistry at Electrochemical and Related Metal-gas Interfaces: Some Perspectives and Prospects. Topi cs in Catalysis. 8 (1999) 65-73. deWild, P.J.; Verhaak, M.J.F.M. Catalytic Production of Hydrogen from Methanol. Catalysis Today 60 (2000) 3-10. Yu, X.; Tu, S.T.; Wang, Z.; Qi, Y. On-board Production of Hydrogen for Fuel Cells over Cu/ZnO/Al2O3 Catalyst Coating in a Mi cro-Channel Reactor. Journal of Power Sources. (2005) Article in Press. Zalc, J.M.; Loffler, D.G. Fuel Processing fo r PEM Fuel Cells: Transport and Kinetic Issues of System Design. Journal of Power Sources. 111. (2002) 58-64.

PAGE 109

98 BIOGRAPHICAL SKETCH Amy Lynn Twining was born on August 19, 19 75, the third of four children to Cheryl and Richard Twining, in Honolulu, Hawaii. Shortly thereafter, the family relocated to Clearwater, Florida, and have lived there since. Amy earned her degree in mechanical engineering from Mercer Universi ty, then worked at Honeywell Strategic and Space Systems and Guidance and Navigation Op erations before pursuing her graduate degree at the University of Florida.


Permanent Link: http://ufdc.ufl.edu/UFE0013362/00001

Material Information

Title: Design and Implementation of an Optical-Access Steam Methanol Reformer
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013362:00001

Permanent Link: http://ufdc.ufl.edu/UFE0013362/00001

Material Information

Title: Design and Implementation of an Optical-Access Steam Methanol Reformer
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0013362:00001


This item has the following downloads:


Full Text












DESIGN AND IMPLEMENTATION OF AN OPTICAL-ACCESS
STEAM METHANOL REFORMER
















By

AMY L. TWINING


A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA


2006

































Copyright 2005

by

Amy L. Twining



































To my parents.















ACKNOWLEDGMENTS

I have been extraordinarily lucky to work for Dr. David Hahn, an outstanding

scientist, and an advisor of amazing optimism and generosity. Dr. Hahn's philosophy

towards experimental design fostered the innovation and creativity needed to develop a

new means of quantifying certain effects in catalysis within strict budget constraints.

Completion of this work would have been impossible without his help and

encouragement. As a mentor, his guidance assisted me through some very trying

circumstances. After seeing the issues inherent in the culture and framework of industry,

having worked for him is likely to be the most heartening and influential experience of

my career. It has been a privilege to be his student, and he has my utmost admiration and

gratitude.

I have also to thank Dr. Vernon Roan for giving me this opportunity, and for being

my first advisor at the University of Florida. Research conducted by his lab group

reinforced the interconnections among research, industry, and policy, as well as the

importance of maintaining balanced and objective scientific opinions.

I thank my committee members, Dr. William Lear and Dr. David Mikolaitis, for

giving their time and advice on this work.

I also thank Robert Schutt, my former boss at Honeywell, and Dr. Aaron Byerley,

at the United States Air Force Academy, for believing in my capabilities, for their

exemplary integrity, and for enabling me to pursue a graduate degree.









I thank my classmates, Tim Simmons, Daniel Betts, Jessica Rozzi-Ochs, and

Patricia Alexander for being my friends. Most especially, I thank my parents.
















TABLE OF CONTENTS



A C K N O W L E D G M E N T S ................................................................................................. iv

LIST OF TABLES .................................................... ........ .. .............. viii

LIST OF FIGURES ......... ......................... ...... ........ ............ ix

ABSTRACT .............. .................. .......... .............. xi

CHAPTER

1 O V ER V IEW ................................................ ......................... ....... ........

1.1 Why Study Hydrogen Production from Hydrocarbons such as
M ethanol? ....................................... .............
1.2 Overview of Catalyst Development......... .................................
1.2.1 H ow the C atalyst W orks ........................................ .....................6
1.2.1.1 A dsorption.................................................. .. .......... 8
1.2.1.2 Surface reactions and deactivation....................... ..10
1.2.2 Rate Laws, Mechanisms, and the Role of Constituents................14
1.2.2.1 Contributions of Cu (111), ZnO and A1203 to activity,
stability, and selectivity ...................................................17
1.2.2.2 Proposed reaction mechanisms and rate laws....................21
1.3 Heat Transfer Limitations in Packed Bed Reformers, Advantages in
M onolith Reformers......................................................... 32
1.3.1 Non-Isothermality in Packed Bed Reactors...............................33
1.3.2 M onolith R eform ers ................................................. ............... 35
1.4 Raman Spectroscopy for Study of Catalysis................... ............... 39

2 RAM AN SPECTROSCOPY ........................................ ........................... 41

3 REA CTOR D ESCRIPTION .................................................... .... ............. 45

3 .1 R e a c to r ................................................................................................. 4 5
3.2 O ptical C configuration ........................................ .......................... 47

4 EXPERIM ENTAL M ETHODS....................................... .......................... 51

4.1 C atalyst B ed R eduction.................................... ............................. ....... 51









4.2 Reform er Operation ............................................................................53
4.3 Reform er Operating Procedure............................................................... 54

5 SPATIAL RESOLU TION ......................................................... ............. 57

5.1 Spatial Resolution of the Optical Set-Up................................................ 57
5.2 Calibration for Gas Concentration......... ........................ ........ .......... 59

6 D A T A ................................................................6 4

6.1 Variation of Species Concentration at Each Position, at Flow 1 ...............65
6.2 Variation of Species Concentration at Each Position, at Flow 2 ..............67
6.3 Variation of Species Concentration along Length and Height above
C ataly st at F low 1 ........6........ ..... .... .... ........ ...... .. ..................67
6.4 Variation of Species Concentration along Length and Height above
C ataly st at F low 2 .............. ...... ...... .. ........ .......... ...... ...... ........... .. 70
6.5 Comparison of Concentration Gradients at Flow 1 and Flow 2 ................72
6.6 Species Gradients along Reactor Length and Height above the Catalyst..74
6.7 Species Concentration vs Position at Height 1 .......................................80

7 SUM M ARY AND CONCLUSION S ........................................ .....................83

8 RECOMMENDATIONS FOR FUTURE WORK ...............................................85

A PPEN D IX D A TA SU M M AR IES ......... ................. ..................................................87

L IST O F R E FE R E N C E S ......... .. ............... ................. ....................... ......................... 9 1

BIO GRAPH ICAL SK ETCH .................................................. ............................... 98
















LIST OF TABLES

Table page

1-1 Summary of Proposed Rate Laws ................................................................... 28

1-2 Summary of Steam Methanol Reformer Studies..........................................29

2 -1 R am an Shift ........................................................................ 4 3

3-1 Equipm ent and Specifications ........................................................ ............... 50

4-1 Reactant Flow Rates and Pum p Settings............................................................... 53

4-2 Flow P aram eters ................. .......................... .................... .... ..... 54

A-i Hydrogen D ata Sum m ary ........................................................... ............... 88

A -2 M ethanol D ata Sum m ary .............................................................. .....................89

A -3 W after D ata Sum m ary. ...................................................................... ..................90
















LIST OF FIGURES

Figure page

1-1 Dehydration of Methanol on a Catalyst Surface................ ...............12

1-2 Portions of Atoms on Face-Centered Cubic Crystal Faces............................... 18

1-3 Example of a micro-reformer, adapted from (Pan et al) .......................................38

2-1 Typical Raman Spectrum of Steam Methanol Reforming.............................. 44

3-1 E nd V iew of P ellet B ed ............................................................... .....................4 5

3-2 Top V iew of Pellet B ed...................................................................... 46

3-3 R reform er assem bly. ....................................... .......... ......... .... .. ... 47

3-4 Experimental Set-Up for Raman Specrtroscopy.................................................49

4-1 Catalyst Reduction Raman Spectra.......... ..... ................................ ............... 52

4-2 Water and Hydrogen Signal vs Time During Reduction............................... 53

5-1 End View of Reform er.............................................. ................. ............... 57

5-2 Side V iew of R eform er ...................................... ....... .................. ............... 58

5-3 Spatial R solution .......................... .............................................. ........ 59

5-4 Temperature Independence of Hydrogen Signal ................................................60

5.5 H ydrogen Calibration Curve............... ....................................... ................61

5-6 M ethanol Calibration Curve ............................................................................ 62

5-7 W ater Calibration Curve .................................... ........................ ............... 63

6-1 Typical Raman Spectra Observed Along the Axis of the Reactor........................65

6-2 Position 1, Flow 1 Species Concentration vs Height........................................66

6.3 Position 2, Flow 2 Species Concentration vs Height................. ......... .......66









6-4 Position 3, Flow 3 Species Concentration vs Height.................. ........ .......67

6-5 Position 1, Flow 2 Species Concentration vs Height ................... ...................68

6-6 Position 2, Flow 2 Species Concentration vs Height.................. ... .......... 68

6-7 Position 3, Flow 2 Species Concentration vs Height.................. ... .......... 69

6-8 Flow 1 Methanol Concentration at 3 Positions vs Height .................................70

6-9 Flow 1 Hydrogen Concentration at 3 Positions vs Height..............................71

6-10 Flow 1 Water Concentration at 3 Positions vs Height.......................................71

6-11 Flow 2 Hydrogen Concentration at 3 Positions vs Height..............................72

6-12 Flow 2 Methanol Concentration at 3 Positions vs. Height ...............................73

6-13 Flow 2 Water Concentration at 3 Positions vs Height ......................................73

6-14 Hydrogen and Methanol, Flows 1 and 2, Position 1 .......................................75

6-15 Hydrogen and Water, Flows 1 and 2, Position 1 ......................... ............... 75

6-16 Hydrogen and Methanol, Flows 1 and 2, Position 2.......................................76

6-17 Hydrogen and Water, Flows 1 and 2, Position 2 ..............................................76

6-18 Hydrogen and Methanol, Flows 1 and 2, Position 3 ........................................77

6-19 Hydrogen and Water, Flows 1 and 2, Position 3 ....................................... 77

6-20 H hydrogen F low 1 ........................... ...... ...................... .. ........ .... ...... ...... 78

6-21 M ethanol Flow 1 ......................... ......... .. .. ..... .. ............78

6-22 W after, Flow 1....................... .... ................ ............. ........... 79

6-23 H hydrogen F low 2 ....................................................... .. ........ .... ..... ...... 79

6-24 M ethanol F low 2 ................................................................. ........ .. ..... 80

6-25 W after, Flow 2....................... .... .................. .. .. ........... ........... 80

6-26 H ydrogen, Flow 1 and 2................................................ ............................. 81

6-27 W after, F low 1 and 2 ...................................................................... ...................82

6-28 M ethanol, F low 1 and 2 .............................................................. .....................82















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

DESIGN AND IMPLEMENTATION OF AN OPTICAL-ACCESS
STEAM METHANOL REFORMER

By

Amy L. Twining

May 2006

Chair: David W. Hahn
Major Department: Mechanical and Aerospace Engineering

This study is an assessment of the feasibility of Raman spectroscopic analysis of

gas phase species in methanol steam reforming over a catalyst surface. Included in this

work is a comprehensive literature review of control volume and surface science studies

of steam methanol reforming, as well as reaction mechanisms proposed in these studies,

and sources of error in their measurement. Advantages of in situ study of a monolith

reformer are discussed. Experimental data reveal fluid dynamic aspects of the reacting

flow, specifically, the scale of the boundary layer and concentration gradients as related

to the flow rate. Recommendations for future work are proposed, highlighting the

successful use of Raman spectroscopy in study of gas-phase species, as well as the

advantages to measuring directly the changes in concentration over time for accurate

determination of rate laws.














CHAPTER 1
OVERVIEW

Fuel cell technology, which relies on gaseous hydrogen and oxygen, is in

development as a more efficient and cleaner method of power generation. Considering

that the transport of gaseous hydrogen adds significantly to its cost, many fuel cell

applications are being developed to incorporate the reformation of more energy-dense

liquid fuels. One such fuel, methanol, has the best hydrogen to carbon ratio.

Unfortunately, even 20 parts per million carbon monoxide can poison fuel cells (Pumama

et al, 2004), so there is an effort to develop more efficient reformers as well as reformate

clean-up and separation technology. Traditional packed bed reformers have shown

diffusion limitations, nonisothermality, and pressure gradients that are difficult to control

and lead to inefficiencies in reforming. For hydrogen production, several monolith

reformers are in development, but none are currently available commercially. The

reformer in this investigation serves to provide a means of studying in-situ the gas phase

and to determine two-dimensional scale factors of steam methanol reforming over a

catalytic surface, rather than through a packed bed, in order to aid in future reformer

system design. The reformer uses Synetix 33-5 steam methanol reforming catalyst pellets

closely packed in a single layer in an aluminum tray, which is situated within the

reformer so that reactant gases pass over it. There are three windows along the length of

the reactor for in situ evaluation of gas phase composition via infrared Raman

spectroscopy. The reactor itself can be raised or lowered; and concentration gradients

were evaluated along length and height above the catalyst. Qualitative relationships









between gas phase composition, flow rate, and reactor length were determined. In the

future, other modified reforming reactions may be studied over the same type catalyst

(combined SMR/POX), which are more thermodynamically efficient. If adhesion of

catalyst material can be achieved, another catalyst tray may be configured above the

original, to study the gas phase between two catalytic surfaces, possibly with a variable

distance separating them. This Poisiuelle-type flow configuration would aid in

determining the critical reactor lengths and diameters for fully developed plug flow, as

well as the length and flow rates at which product gases begin to cause reverse reactions.

Such a configuration will allow, as closely as possible, in-situ study of a monolith

reformer while minimally disturbing reactor conditions.

1.1 Why Study Hydrogen Production from Hydrocarbons such as Methanol?

The goal of fuel cell technology, including hydrogen production from various

feedstocks, is to provide more efficient power that emits fewer unrecoverable byproducts

within the real economic constraints of limited natural resources and the current state of

technological development. Various greenhouse gas models provide a spectrum of the

timescale and severity of impact of a CO2 accumulation in the atmosphere (Litynski et al,

2005). Without debating these, more generally, exhausted heat and materials represent

wasted energy and resources. For portable devices, fuel cells can provide better efficiency

and energy density (Hu et al, 2003) than a conventional lithium ion battery. The energy

density of methanol is 5.5 kWh/kg compared to 0.5-1.0 kWh/kg for stored H2, and an

estimated 0.3 kWh/kg for a lithium-ion battery (Hu et al, 2003). The required efficiency

of fuel cell systems to match lithium-ion batteries is 5.5% for a methanol-based fuel cell

system and 30-60% for a stored-hydrogen fuel cell system (Hu et al, 2003). Fuel cells,

which exhaust only heat and water, are considered the new hope for clean power









generation. However, established methods of hydrogen production themselves have a by

product of CO2. Designs that rely on non-hydrocarbon feedstock, such as solar-powered

electrolysis for hydrogen production, are still prohibitively expensive and less efficient

(Padro et al,1999). These concepts will likely not be economically competitive until they

progress more technologically, and the cheaper carbon-based feedstocks become scarcer

and more expensive. (Padro et al,1999; Roan et al, 2004) In the mean time, hydrogen

production from readily available hydrocarbon feedstocks is being improved, with the

means of containing and sequestering CO2 in concurrent development. Such examples

include power generation from coal gasification, with hydrogen co-production, and an

experimental gasifier that also incorporates CO2 separation (Rizeq et al, 2003; Kubek et

al 2002; Chen et al, 2002).

A variety of methods of hydrogen production and delivery will enable tailoring to

particular applications using fuels that are most readily available in that locale, for

example, coal gasification where natural gas is unavailable, and liquid fuel transport

where there is not yet a pipeline infrastructure. Studies have shown that transport of

gaseous hydrogen over land, as opposed to via pipeline, greatly increases the final price

(Amos et al,1998; Roan et al, 2004). If the infrastructure is available, hydrogen

transmitted through pipelines may be ideal for large stationary fuel cell power generation,

such as domestic power. However, extensive pipeline installation is costly, and in the

interim, truck transport of liquid fuel to smaller local reformers will have an advantage

(Myers et al, 2002). For small or portable applications, a fuel cell system incorporating a

liquid fuel reformer may be a better option than one using a large compressed gas tank,

considering the energy density of liquid fuels compared to gaseous hydrogen (Ogden









2002). A compact fuel cell system using methanol, operating at approximately 5% overall

efficiency, still has a higher energy density than a lithium polymer battery (Hu et al,

2003). In contrast, a fuel cell system using stored hydrogen must operate at 30-60%

efficiency to equal the energy density of Lithium ion batteries (Hu 2003). Fuel processors

for fuel cell applications still face the difficulty of transient reaction kinetics, especially

the time required to provide heat to begin reforming (Springmann et al, 2004). Leak

detection and boil off in compressed gas and liquid hydrogen transport methods, as well

as flammability, provide additional complications to portable fuel cell systems. These

circumstances indicate the usefulness in optimizing reformation of liquid fuels such as

methanol.

1.2 Overview of Catalyst Development

Composition and spatial variation in catalysts affect possible reaction pathways,

resulting in many observed rates and proposed mechanisms. The development of

catalysts for steam methanol reforming began with a study of the reverse reaction of

methanol synthesis from the hydrogenation of CO2. However, it quickly became apparent

that, given the effect gas phase chemistry has on catalyst activity and changes in catalyst

structure during adsorption, steam methanol reforming is not simply a reversal of the

better known methanol synthesis reaction (Peppley et al, 1999). Competitive adsorption,

as well as changes in adsorptive energies with coverage (d'Alnoncourt et al, 2003), are

factors that contribute to the fact that steam methanol reforming is not simply the reverse

of methanol synthesis. For example, the heat of adsorption of CO on a copper catalyst

varies by almost 30 kJ/mol between low and higher coverages, with the higher coverage

reducing the heat of adsorption of additional CO. In order for a species to adsorb on the

catalyst, valence electrons in the catalyst have the mobility to stabilize the valence









electron shell in the adsorbate. This results in a change in the material in immediate

proximity to the adsorbate, the cumulative effect of which is shown by the changes in

adsorptive energy with catalyst coverage (amount of adsorbate covering active sites). In

this way, the catalyst is like an additional reactant, and its activity will change depending

on the gas phase composition. Though overall mechanisms are proposed over certain

types of sites, it becomes clear in the literature that the chemistry of the catalyst itself

plays a significant role. Varying rate laws and mechanisms which consider the copper site

to be the active site for reaction have been proposed for copper catalysts augmented with

a variety of constituents. Proposed rate laws and kinetics also vary with catalysts made

from the same metal oxides but in different proportions, and in different preparation

methods; and there is even debate over mechanisms and rate determining steps over the

same catalyst. The next sections discuss how catalysts work, proposed reaction

mechanisms seen in literature, and changes in catalyst structure and activity during

reactions. These chemistry issues are the business of surface science, however they are

discussed here in order to explain how the differences in the apparent mechanisms and

deactivation rates observed are affected by reactor design, catalyst composition, and

certain surface and geometric factors. The experiments done on this reactor will serve to

qualitatively illustrate the relationship between reactant flow rate, conversion, the scale of

the reaction in two dimensions: height above the catalytic surface, and along the length of

the reactor. While this study used commercial components to assess the feasibility of

Raman spectroscopy for a gas phase study, a more refined experiment may quantify scale

factors for future use in monolith reactor design.









1.2.1 How the Catalyst Works

A gas reaction on a catalytic surface differs fundamentally from a homogeneous

gas phase reaction. Even though the overall reaction is the same, the interaction of

adsorbed gas species with the catalyst results in a different mechanism than in a purely

gas phase reaction. While heating the catalyst and gas phase reactants can overcome

barriers to adsorption and reaction, the behavior follows the Arrhenius law through a

narrower range than in homogeneous, thoroughly mixed chemistry. This is due to the fact

that not all the participants in the reaction are equally mobile, and unlike other reactants,

the catalyst must be regenerated by the end of the reaction process. The addition of heat

in a homogeneous gas phase reaction increases the rate by increasing the number and

energy of collisions, resulting in improved chances of collisions that result in a reaction.

Adding heat to a heterogeneously catalyzed reaction will increase the rate at which

particles collide with the surface, but the speed at which they dissociate and recombine

on the surface may not have a direct proportionality with temperature. Increasing the

temperature of the catalyst can result in reactants becoming more strongly bound to the

surface, which reduces their mobility across the surface and subsequent rates of reaction

with other adsorbates. When the catalyst becomes too hot, adsorbates can approach the

activation energy at which they become permanently bound to the surface (Masel 2001),

but prior to this occurrence, there is a sharp leveling-off of the reaction rate with

temperature. The reason for this could be due to the reduced mobility of adsorbates as

they become more strongly bound to the catalyst, but another cause may be that the

catalyst material properties are beginning to change. Metal oxides in catalysis are less

mobile in their microstructure, and therefore less subject to change with temperature, but

this is not necessarily the case for a bulk metal used in a catalytic process. Thermal









production of point defects in solids, gas phase reactions, gas reactions on solid surfaces,

even reactions in liquid solutions, all follow an Arrhenius-type law (Shackelford, 1992)

within a certain temperature range. The temperature restrictions on the catalyst arise out

of the closeness in temperature of each of the chemical processes: the reaction of

adsorbates on the surface, the process of creating unfavorably strong bonds between

adsorbate and catalyst, and thermally restructuring the catalyst sinteringg). "The

temperature at which atomic mobility is sufficient to affect mechanical properties is

approximately one-third to one-half times the absolute melting point." (Shackelford

1992). The melting point of copper is 1357K, and one-third of that is about 452K, 180C.

This is essentially the heat treatment temperature of the metal; the temperature at which

point defects can migrate and grain boundaries can begin to restructure. Reactor

temperatures range from 200-300 C; therefore a copper catalyst must be designed to be

structurally stable at these temperatures. Other metal oxides, such as zinc and zirconium

oxides, have been used to stabilize the copper material. The copper atoms are less mobile

in contact with the metal oxides, which aid in maintaining the structure of the copper.

Given heat transfer problems that result in non-isothermality in packed bed reformers, it

is easy for a hot-spot to occur that can approach the critical temperature for restructuring

copper crystal face active sites on the catalyst, even with additives to stabilize the copper

material. Diffusion of defects through copper can occur in the bulk, on the surface, and at

grain boundaries, at respectively decreasing temperatures and depending on the amount

of the diffusing region that is available (Shackelford 1992). The desired reaction

temperature is an intermediate one that allows strong enough adsorption for dissociation

and reaction, while being low enough that the catalyst is regenerated without non-









reversible reactions occurring on the active sites. The reaction rate for steam methanol

reforming levels off around 220C (Shen et al, 1997) and begins to show non-Arrhenius

behavior around this temperature (227C) (Agrell et al 2003) Mass transfer limitations

become significant at temperatures higher than 220C, (Agrell et al, 2002) while the

reaction generally shows Arrhenius-type behavior between 175C and 220C.

Generally, catalysts work to lower the activation energy in chemical reactions by

stabilizing intermediates and positioning reactants in more ideal proximity and

orientation in order to improve the odds of a reaction occurring. (Masel 2001) A catalytic

process comprises adsorption, either dissociatively or molecularly, followed by reaction

on the surface, then desorption. For catalytic reactions in this study, only monolayer

adsorption occurs. Multilayer adsorption is a condensation process and occurs in water

condensation and coking or soot formation, when the attractive interactions between

adsorbates are strong compared to kBT, near the boiling point. Catalytic activity depends

on valence shell configurations, differs fundamentally between metal crystals and ionic

solids, and varies within metals in the same group on the periodic chart, though the

details are not fully understood.

1.2.1.1 Adsorption

The process of adsorption of a gas on a metal surface occurs when there is a net

polarization as the gas approaches the surface, and an interaction with the free electrons

in the conduction band of the metal. Adsorption of gases on oxide surfaces has been

described as an equalization of electronegativities (Masel 1996). Polar molecules, such

as water, bond more easily to insulating oxides; and covalent molecules, like carbon

monoxide, bond more strongly to metals. The presence of empty orbitals and bulk

electron states on a metal surface enable mobility of valence electrons, especially d-band









electrons, to stabilize the adsorbate. Adsorption is most stable when the highest occupied

orbitals interact with the lowest unoccupied orbitals, and an electron density at one end of

a covalent molecule can cause that end to adsorb on the metal surface. In metals, there are

empty orbitals very close in energy to the Fermi level (the top of the filled levels), so

very little energy is needed to excite the uppermost electrons (Atkins 2002). In transition

metals, d-band electrons in partially filled outer shells can shift easily and "lend" an

electron to the outer shell of the reactants. This weak bond changes the shape and

orientation of the antibonding orbital, allowing it to break more easily. D-band electron

mobility can also help to lower Pauli repulsions, thereby aiding in bond scission, though

they contribute minimally to the actual heat of adsorption (Masel 1996).

Bonding on insulating oxides is mainly ionic. Since electron mobility is restricted

in metal oxides, the surface is fairly inert and bonding is dominated by structural defects.

Aluminas are considered Lewis acid sites, containing an empty orbital to accept a

charge. Oxide doping changes the electronegativity of defects, and therefore their ability

to accept charge. Masel gives a complex process of adsorption of gases on ionic

compounds. To begin with, the adsorbate reacts first with a defect, and creates an ion,

which may then migrate to a defect-free region of the surface. The heat of adsorption on

defect-free surfaces is related to ionic and polarization forces; therefore, neutral or non-

polar species interact fairly weakly on an insulating oxide while polar species interact

fairly strongly and create ionic adsorbates.

The scale of surface roughness and defects can also affect adsorptive energies. This

is true for metal oxides, as has been described earlier, as well as for adsorption on metal

crystals. There is greater electron density on an inside corner, for example, and a









molecule that adsorbs there can dissociate more easily since the material accommodates

its three-dimensional shape. If a molecule is large compared to the scale of surface

roughness, there could be an adverse effect since it may be more difficult for the atom

most likely to bond to the catalyst to get close enough to the active site. On the other

hand, surface defects need to stabilize adsorbates in close enough proximity to facilitate

their reaction with each other. Strain in crystal lattices has also shown to improve

catalytic activity via a similar effect (Gunter et al 2001). Moderate ranges of surface

roughness seem to provide the most benefit (Masel 1996).

1.2.1.2 Surface reactions and deactivation

Dehydration of methanol over a catalyst can illustrate the difference between gas-

phase and surface reaction mechanisms. This process is depicted in Figure 1-1. In the gas

phase, molecules must approach each other with the right amount of energy and

orientation in order to break the correct bonds, with the weakest bonds expected to break

first. In catalysis, one particular end of a molecule will bond to the catalyst site, causing a

bond distortion and possible dissociation between the bonded atom and the rest of the

molecule, but this may not be the same bond breakage that initiates reaction in the gas

phase. Methanol at first adsorbs weakly on the metal, then it dissociates into H+ and

methoxy, which are each more strongly bonded. For methanol in a gas phase reaction, the

C-O bond breaks first, since it is the weakest at 84 kcal/mol. On the metal catalyst site,

the oxygen in methanol bonds closest to the catalyst, causing a dissociation of one

hydrogen atom. This is because the hydroxyl group is highly polar, compared to the CH3-

end of the molecule, so this end bonds to the metal. The net polarization that occurs when

the methanol molecule approaches the conduction band in the metal is such that the bond

to copper of the H+ and the methoxy ends of the molecule strain the O-H bond enough to









break it. As a result, the H-O bond is distorted and the first to break, even though it is the

strongest bond (103 kcal/mol) in the gas phase molecule. On the catalyst, the C-O bond

remains intact, but the relative polarity of the hydrogen atoms with each of the carbon

and oxygen atoms causes C-H and O-H bonds to distort and hydrogen to dissociate first.

Methanol dehydrogenates one hydrogen at a time on the catalyst surface, resulting in

adsorbed hydrogen atoms and an adsorbed carbon monoxide molecule, which then

desorbs. As noted earlier, adsorbed species can migrate on the catalyst surface, and the

adsorbed hydrogens would combine to form H2 before desorbing. This is not necessarily

the same path that methanol takes in the steam reforming reaction. In fact, several models

include steam methanol reforming, as well as methanol decomposition and water gas

shift reactions, due to probable interactions of adsorbates on catalyst sites. Reactant

species can adsorb on copper, as well as ZnO and A1203 constituents of the catalyst.

Additionally, reaction pathways may occur over crystalline copper metal, and some have

been proposed over a special CuZnO site (Waugh 2004). Depending on the size of

constituents, whether they are in regions large enough for several molecules to adsorb

and react, or homogeneously distributed at a molecular level, there exist opportunities for

more than one reaction pathway to occur at the same time.

Adsorbed species can interact directly with each other, or an adsorbate can form a

bond with an electron donated from the surface, changing the electronic properties of the

surface. In this way, bonding of additional molecules has been shown to be affected by

the coverage. As a result, it is possible for more than one pathway to take place on a

catalyst, though proportions of additives to copper catalysts have been shown to alter the

selectivity for certain reaction pathways, even though the (111) copper phase is









considered the predominant site for catalytic activity. These phenomena are reflected in

some of the proposed mechanisms and intermediates to be discussed later.










adsorbed methanol -H=> methoxy -H =>

O O
/H I H I
o=c{ c c
O-CC C


carbon
formaldehyde -H => formyl -H => monoxide

adapted from Masel, 1996.

Figure 1-1. Dehydration of Methanol on a Catalyst Surface

Surface reactions do not exactly follow the Arrhenius law, where Rate= Aexp

(Ea/kBT), and A is a pre-exponential, Eais the activation energy, kB is the Boltzmann

constant, and T is the temperature in Kelvin. Many reaction models have evaluated a

global Arrhenius-type relationship that is valid through a range of conditions. The

addition of heat will initially overcome the activation energy of adsorption, but further

heating can cause adsorbed intermediates to stabilize on the surface, or to break down on

the surface, resulting in decreased catalytic activity and reaction rate. For example, if the

catalyst surface is hot enough that carbon monoxide breaks up into carbon and oxygen

rather than desorbing, an active site has been removed from the reaction by oxidation,

and carbon deposition can begin the process of deactivation (Argawal et al, 2004).









Oxidized sites can be reduced again, but carbon is difficult to remove. Sometimes the

catalyst can be heated with pure steam to react with the carbon and remove it, as has been

done with cyclical steam methane reforming (Choudhary, T.V. et al, 2000, 2001;

Choudhary, V.R. et al 2001), but generally carbon build up leads to coking and

deactivation. In addition, reactions on catalyst surfaces are generally slower than gas

phase collisions. Catalysts improve the rate by lowering the activation energy, putting

molecules in proximity and creating stabilized radicals, thereby improving the odds of

reaction by avoiding non-productive collisions. As a result, heterogeneous reactions are

spatially limited, and also have greater temperature limitations than gas phase reactions.

Elevated temperatures in steam methanol reforming generally do not contribute to

carbon buildup, as they do in steam methane reforming; however they can present

problems in maintaining the catalyst structure. Fortunately, even at stoichiometric

chemistry, the C-O bond in methanol steam reforming generally stays intact and desorbs

from the catalyst, so carbon buildup is not usually a problem. Since carbon monoxide

has a triple bond, it takes much more energy to break it, and it is less likely to occur at the

lower temperatures of steam methanol reforming. Even if reactants do not become

permanently bonded to catalytic material, at elevated temperatures, sintering becomes a

risk. This occurs when high temperature and pressure cause the dispersed copper to

coalesce, and make copper masses that have less surface area and are less catalytically

active. Sintering in copper can occur far below the melting point of the metal, as low as

300C, and is related to vacancy diffusion, which is associated with the cohesive energy

of the metal (Twigg et al, 2003). It has been observed at approximately 250 C in water

gas shift reactions over copper catalysts (Tanaka et al, 2003). Among catalyst materials,









the copper is considered relatively unstable, and this is reflected in its low melting and

point cohesive properties. Changes in metallic crystal structure also occur well below the

melting point, for example, in the common solution heat treatments (quenching and

tempering) of metals like steel and aluminum (Shackleford, 1992). When heat treating

metals, stress concentrations occur at covers, or in the case of small crystals,

microscopic defects. A higher microscopic strain can cause a stress that overcomes

intermolecular forces, such that copper atoms will migrate to a more stable configuration,

or, conversely, vacancy diffusion. Maintaining the structure of the (111) copper crystal

face is important in catalytic activity because of the geometric factors that influence

electron density and therefore adsorptive energies. Defects of the right scale, on the right

composition of catalyst, can improve catalytic activity, until the catalyst becomes so hot

that those structures are not maintained.

Catalysts can be poisoned by chemical impurities in the reactant feed, such as

sulfur, which can occur in liquid fuels derived from coal, and chlorine, which occurs in

tap water. To combat this, some fuel cell system designs incorporate guard beds that are

separate from the reactor so that they can be changed. For example, ZnO can remove

sulfur from the stream by creating ZnS which remains fairly stable (Huang et al, 2004).

Removal of chlorine is much more difficult because it bonds easily with both copper and

zinc and has highly mobile ions. A ZnCl2 compound is not stable enough to remove

chlorine, but an aluminum oxide is recommended (Lindstrom et al, 2001).

1.2.2 Rate Laws, Mechanisms, and the Role of Constituents

The influence of the constituents of catalyst materials is not reflected in rate laws

for catalytic chemistry. Yet this can be misleading, since different rate laws have been

found for the same reactions over various catalysts. Even if the active sites are the same









chemically, the manner in which they are dispersed and their interaction with other

constituents effect how easily reactants adsorb, interact, and desorb, as well as the ability

of the crystalline structure of the site to return to its original state after desorption.

Though pure metal crystals act as the active sites in many catalytic reactions, they have

also been shown to improve with doping by small amounts of another metal or metal

oxide. This can form a promoter, which is an alloy that modifies the electronic structure

and changes the interstitial electron density, thereby strengthening or weakening

interactions on the catalyst surface. Additives can improve the surface area of active sites

by the manner in which they adhere to and disperse the catalyst material. Because of this,

the methods of preparation of catalysts are important and often result in differences in

activity, for example coprecipitation versus homogeneous precipitation (Shishido et al,

2004). The discussion of why certain ionic solutions precipitate a more homogeneous

mixture of metal oxides than others is beyond the scope of this paper. However, it bears

consideration that ions overcome an energy of solvation when they are precipitated out of

solution. Homogeneous precipitation of different metal oxides out of solution would

require a solution in which the metal ions have similar energies and concentrations to

overcome their respective solvation shells so that they will precipitate at similar rates, and

therefore more homogeneously. Takeguchi et al (2002) also found that preparation

methods resulted in different catalytic activities when comparing coprecipitation methods

to a glycothermal preparation method. The glycothermal method resulted in a higher

catalyst surface area and higher conversion, which supports the concept that surface area

effect reactant exposure and turnover frequency from the catalyst site (Takeguchi et al

2002). Some additives have been found to make the catalyst more stable such that the









surface returns to its original state more easily after products desorb. Most of the effects

of additives are observed experimentally, rather than anticipated, and analysis of

subsequent experiments over numerous reaction conditions has served to provide the

theories discussed in literature. More recently, there have been in situ studies of

adsorbates and changes in the catalyst surface during reaction, but the most common

means of studying catalysts has been preparation the catalyst, characterization of the

surface for active sites, exposure reforming conditions in a packed bed reactor, in various

temperatures, residence times and gas concentrations, followed by gas chromatography of

reactor products. From this, correlations between measured active sites, chemical

composition, and reactor conditions are developed. Much of the data used to develop

rates is handled this way. Reaction mechanisms and rate laws are often adjusted to match

the experimental data. As will be explained later, there can be errors in apparent rate

laws, since diffusion limitations, non-isothermality, and residence time can cause

inconsistencies and changes in catalytic activity in packed bed reactors. Because of this,

studies of monolith reformers and adhesion of catalyst material to monolith substrates

have been investigated in the literature, which will likely result in more consistent, more

predictive algorithms for catalyst and reformer design in the future. The catalyst used in

this investigation comprises copper oxide, which is reduced to copper metal in order to be

active, ZnO, and A1203. It is arranged a single layer in the reactor so that the reactions

occur on the outer surface of the pellets, thereby eliminating the complications arising

from intraparticle diffusion and effectiveness factors that would normally have to be

calculated to estimate the actual flow behavior, as well as some of the problems of









nonisothermality. Studying the catalytic reaction over the surface will allow correlations

to a monolith-type design.

1.2.2.1 Contributions of Cu (111), ZnO and A1203 to activity, stability, and
selectivity

In catalysis, activity is how easily the catalyst converts reactants to products,

stability describes the long term behavior of the catalyst, and selectivity is a measure of

how well a catalyst enables reactants to favor a particular reaction pathway over other

possible reactions.

Recent literature investigates the possible roles that these additives have in altering

catalyst activity, selectivity, and stability. In steam methanol reforming, metallic copper

sites are considered the active sites for the reaction mechanism. Comparison of Zn-

deposited Cu (111), (110), and (100) surfaces have shown minimal catalytic activity over

Cu (100) and Cu( 10), but much better performance over Cu( 11) in methanol synthesis

from CO2 and H2 (Nakamura et al, 1998). If ZnO is used to distribute, stabilize, and

increase the copper surface area, it stands to reason that the electrons in copper crystals in

contact with the ZnO would be somewhat less mobile since ZnO is an ionic compound.

Cu(l 11) has a higher planar density of copper than Cu(100) and Cu( 10). The atoms per

crystal face area, where r is the atomic radius, are 0.1768/r2 for a (110) crystal, 0.25/r2 for

(100), and 0.57/ r2 for (111). With the reduced electron mobility due to its interaction

with ZnO, the copper atoms per crystal face become important in providing the electron

density for adsorption. This gives an indication of the importance of the structure of the

active site since methanol steam reforming and methanol synthesis are essentially reverse

reactions that take place within the same temperature ranges. It is important to keep in

mind that they are not exactly reverse reactions since the gas phase species can compete









for the same sites, and the species that are already adsorbed influence the adsorptive

energies of sites nearby. Still, this illustrates the importance of controlling heat transfer to

prevent sintering, as well as the structural sensitivity that the alcohol dehydration and

synthesis reactions have. Consideration should be given to the fact that more closely

packed crystal structures in metals will have greater electron density in the conduction

band, which can enable a three dimensional molecule to bond more easily, especially if

the atom that bonds is not one of the outer-most. In fact, methanol itself weakly adsorbs,

but has strong bonds when it dissociates into H+ and CH30-, so the site has to

accommodate the geometry of the in-tact methanol molecule, and then the conduction

band must be such that its effect on the polarity of the O-H bond distorts it enough to

break. Figure 1-2 gives an illustration of the portions of atoms on crystal faces for a face-

centered cubic crystal, such as copper.







(111) face






(110) face








(100) face

Figure 1-2. Portions of Atoms on Face-Centered Cubic Crystal Faces









ZnO and A1203 additives have been shown to improve the activity and stability of

copper catalysts. A correlation between activity and stability has been found (Kumari et

al, 2002), which is reasonable, as an unstable catalyst would not maintain active sites.

Catalysts containing ZnO have a higher surface area of copper, and therefore more active

sites, and those containing A1203 have improved stability (Agrell et al, 2003). A1203 is

considered an insulating oxide, meaning that it tends to create ionic bonds with

adsorbates. ZnO is considered a semiconducting oxide and can form either covalent or

ionic bonds. Metal surfaces have mainly covalent bonding with adsorbates. CO tends to

bond strongly to metals and weakly to insulating oxides, whereas H20 bonds weakly to

most metals and strongly to insulating oxides (Masel,1996). Addition of A1203 has been

shown to increase Cu surface area as well. As a result, there is greater conversion of

methanol at lower temperatures, however there is also a higher selectivity of CO

(tendency for a reaction pathway that produces carbon monoxide) (Shishido et al, 2004).

Even though the copper crystal site is considered the active site for catalytic reaction, the

apparent influence of ZnO on the activity is not accounted for in terms of increased

copper surface area, alone (Gunter et al, 2001). This is understandable considering that,

according to Masel (1996), ZnO can bond either covalent or ionic molecules, for

example, CO or H20; and H2O should bond strongly to A1203, weakly to clean copper,

and reasonably well to ZnO. So, methanol adsorbs dissociatively on copper, but the other

reactant, steam, has minimal bonding on copper but good bonding on the other two

constituents. It's tempting to infer that ZnO and/or A1203 play roles as active sites in the

steps of the reaction, in addition to dispersing and stabilizing the Cu(l 11) sites, but

studies on Cu/ZnO/ A1203 catalysts all assume an active copper site without examining









adsorption on metal oxide constituents. One theory regarding the Cu/ZnO interactions is

that Cu may be incorporated into ZnO on interstitial and substitutional sites, assuming

possible valence states of Cu Cu+, Cu2+; and that hydrogenation of format species on

Cu proceeds via hydrogen spillover from ZnO sites (Agrell et al, 2003). Interestingly,

several articles in literature discuss a possible symbiotic relationship between the ZnO

and copper sites in steam methanol reforming, and that copper catalysts with ZnO may

have an adsorbed formaldehyde or formyl intermediate (Choi et al, 2003). Gunter et al

(2001), Grunwaldt et al (2000), and Nakamura et al (1998) have pointed to the possibility

that additives play a larger role in catalytic activity, besides dispersing and increasing the

surface area of copper crystals, such as strain at the Cu/ZnO interface, and activity

resulting from Zn being dispersed in the copper bulk. With easily reducible oxides, such

as ZnO and CuO, Zn atoms can migrate and form a copper surface alloy at around 300C

(Grunwaldt et al, 2000). The steam methanol reforming reaction has both structural and

temperature sensitivity. For example, preparation methods for dispersion of ZnO and

A1203 in Cu( 11) reflect improved activity over more homogeneously precipitated

catalysts, than those in which there are larger conglomerations of each separate

constituent (Shishido et al, 2004). Reactants can adsorb on more than one type of the

constituents, the structure and crystal faces of which influence adsorptive energies.

Simultaneously, there exists a narrow temperature range (175-220C) where the reaction

follows the Arrhenius law, above which the actual structure of the catalyst can change.

Though the precise pathways, and the influences of particular sites on selectivity, are

unclear, there is a general trend that a roughly 50/50 ratio of Cu/Zn results in the highest

activity. If A1203 is included, up to 18% (Shishido et al, 2004) improves reactivity as









well, if the constituents are dispersed homogeneously. Both the copper surface area and

the supporting materials have a combined effect on catalytic activity. Although copper-

based catalysts with SiO2 have improved over-all surface area, there is reduced specific

activity (Saito et al, 2002). Zinc has also been shown to improve the reducibility of CuO

to copper metal for catalytic activity (Fierro et al, 1996).

Many fuel cell/reformer systems incorporate a selective oxidation catalyst to

complete the oxidation of CO to C02, as a trade off of being able to run the steam-

methanol reforming reactor at a lower temperature for nearly the same hydrogen

production. So far, ternary catalysts featuring zinc, aluminum, and copper show good

reforming rates and stability at moderate temperatures, though other studies with Zr or Cr

doped copper catalysts are showing improved activity with Zr, and improved stability

with Cr (Huang et al, 2004; Loffler et al, 2003; Agrell et al, 2003).

1.2.2.2 Proposed reaction mechanisms and rate laws

There have been generalized, proposed models for types of reaction mechanisms in

catalysis: Langmuir-Hnshelwood, Rideal-Eley, and Precursor mechanisms. The

Langmuir-Hinshelwood reaction follows for the reaction A+B AB follows the path

A+ S4 A (ad)

B+S- B(ad)

A(ad) +B(ad)- AB + 2 S

where S is the catalyst site. This assumes that reaction follows the Langmuir adsorption

isotherm in which assumes no competitive and no dissociative adsorption. This rate

mechanism supposedly dominates at high vacuum, although many reactions occurring at

moderate pressures appear to follow this mechanism as well. (Masel 2001) In the Rideal-

Eley mechanism, gas phase species collide with adorbed species and react at once. For









the case of HBr formation, this occurs at moderate pressures, but none of the pathways in

literature for steam methanol reforming seem to follow it. In the Precursor mechanism,

species A adsorbs, species B collides with the surface and enters a mobile state until it

encounters the adsorbed A species, forms A-B complex, then desorbs. This seems similar

to the Langmuir-Hinshelwood mechanism, in that there are essentially adsorbed

complexes that react on the site, rather in the single step of a gas molecule collision with

an adsorbed species. In fact, most rate data appear to match Langmuir-Hinshelwood

mechanisms. (Masel 2001) There are a number of observed, apparent reaction pathways

that can seem to match certain rate laws, though these can be skewed by errors related to

diffusion limitations in packed bed reactors. The behavior that emerges from the

influence of metal oxide additives on the catalytic activity of a metal surface is difficult

to verify experimentally, and even more difficult to anticipate since there are so many

factors. For example, the theoretical mechanisms described above assume all sites are

identical and do not account for competitive adsorption. They also do not account for the

behavior of various metals or pairwise interactions between adsorbed intermediates

(Masel 2001, 1996). Despite this, apparent rate laws derived from experiments seem to

mimic those derived from the Langmuir-Hinshelwood mechanism. Still, much progress

has been made in the past few years towards understanding the most likely reaction

pathways in catalysis through experimentation. One thing to bear in mind is that the

reaction mechanisms in literature vary widely, even though they describe the same or

related reactions, because they were studied over catalysts of various compositions and

preparations in reformers that were not all the same size or had the same residence times.

The composition and preparation affect the nature of the active sites, and gas phase









composition affects adsorption energies. Since reactants can adsorb on active sites in any

arrangement, when they dissociate and react, any two intermediates can react and desorb.

This means that there are opportunities for more than one reaction pathway to occur and

what is actually observed experimentally is the combined effect of these.

In this section, some of the proposed mechanisms, rates, and the rationale behind

them will be discussed and compared. Initially, the proposed reaction mechanism for

steam methanol reforming was the sum of methanol decomposition and the water gas

shift.

CH3OH 42H2 + CO

CO + H20 H2 +C02

However, in experiments, CO levels were very small and thought to be either an

attainment of equilibrium in the water gas shift, or the decomposition of carbon

monoxide on the catalyst, which would lead to the oxidizing of catalyst sites and carbon

build-up. The latter of these explanations is unlikely, since it would quickly lead to

catalyst deactivation (Lee et al, 2004). It was also determined from studies of methanol

synthesis that methanol is produced by the hydrogenation of CO2 from syngas, which is a

mixture of CO, CO2, and H2, rather than by hydrogenation of CO (Peppley et al,1999).

This prompted the inclusion the water gas shift reaction in the mechanisms for both

methanol synthesis and steam methanol reforming, presuming, to begin with, that steam

methanol reforming was a reverse reaction of methanol synthesis (Peppley et al, 1999;

Rozovskii, 2002). This, however, is not exactly true either. It turns out that the, even

though the reaction occurs on the surface, the composition of the gas phase effects

catalytic activity. Methanol and CO competitively adsorb, and there is much more









methanol in the gas stream in steam methanol reforming, than in the syngas of methanol

synthesis. As a result, methanol competitively adsorbs and inhibits adsorption and

subsequent conversion of CO (Peppley et al,1999). Peppley et al proposed a kinetic

model that involves two types of sites and includes all three reactions: steam methanol

reforming in one step, methanol decomposition, and the water gas shift reaction.

CH3OH+ H20 4CO2 + 3H2 (1)

CH3OH 42H2 + CO (2)

CO + H20 H2 +CO2 (3)

Peppley developed an elaborate mechanism that involves two types of sites, type 1

for reactions (2) and (3) and type 2 for reaction (1), as well as type la and 2a for the

adsorption of atomic hydrogen. Peppley explains the need for these designations because

his literature review indicates that H2 does not adsorb on copper. However, according to

Waugh, atomic hydrogen can be adsorbed to copper, and indeed it is when methanol

dissociatively adsorbs. Perhaps the atomic hydrogen adsorbate migrates across the

surface (Masel 1996) to the sort of interstitial site Peppley is referring to. Water

dissociatively adsorbs on the Cu/ZnO/A1203 (Peppley 1999), possibly on the ZnO or

A1203 (Masel 1996) or on copper, in which case there are scenarios in which atomic

hydrogen can be bound to the copper site without requiring different types of sites than

the adsorbed reactants. Though Masel (1996) states that water adsorbs on ZnO, no

literature in methanol steam reforming, as far as could be found, considers it to be an

active site. Since ZnO-enhanced activity is reported to be insufficiently explained by

improvement in copper surface area, alone, it seems worth mentioning here. In Peppley's

FTIR spectra of adsorbed species, CO does not appear, which is hard to reconcile with









methanol decomposition on the surface. Fully reversible adsorption of CO on clean,

hydrogen-reduced copper was determined to be fully reversible at room temperature by

d'Alnoncourt et al, with heats of adsorption ranging from 70kJ/mol at low coverages and

45 kJ/mol at high coverages. Adsorbates that do appear in Peppley's DRIFTS study are

hyroxy, methoxy, format, CO2. Another reaction pathway, to be discussed next,

suggests methyl format as a major intermediate. To explain this discrepancy, adsorbed

methoxy groups have been observed to react to form format groups in the absence of gas

phase methanol. With gas-phase methanol, methoxy reacted to produce methyl format.

(Trimm et al, 2001) Peppley's Type 1 (for both reactions (2) and (3)) is a site that

evolves from "partial decomposition of a hydroxyl-carbonate phase that supports

hydroxylation of the intermediate surface formates", and Type 2 (for (1)) is the Cu(l 11)

face-centered cubic crystal that supports the decomposition reaction.

Later studies have proposed reaction mechanisms based on the scheme that water

reacts directly with methanol, with CO concentrations resulting from a reverse water gas

shift reaction (Agrell et al 2002, 2003; Lee et al, 2004):

2CH30H- CH3OCHO + 2H2 (4)

CH3OCHO + H2O CH3OH + HCOOH (5)

HCOOH CO2 +H2 (6)

The net of the above reactions (4)-(6) is

CH3OH+ H20 4CO2 + 3H2

and the reverse water gas shift reaction is

CO2 +H2 "CO +H20 (7)









While Purnama and Agrell have developed overall rate laws based on similar

reaction schemes, Lee proposes a reaction mechanism involving two sites to explain their

experimental data. Like Peppley's model, Lee proposes that adsorbed methoxy and

adsorbed hydrogen do not occur on the same active sites. Since only hydrogen and

methanol were shown to affect the reforming rate, partial pressures of other terms, and

likewise, adsorption of other components on other types of sites, were disregarded as

negligible towards the overall reaction. Considering the earlier discussion of possible

adsorption of reactants on more than one type of site, it would seem that these non-

contributing pathways would contribute to thermal inefficiency, giving rise to the

importance of homogeneous distribution of active sites and catalyst additives in the

overall rate of hydrogen production, as shown by Shishido and Agrell.

Yet another study shows a variety of the possible interactions between adsorbates,

favoring a pathway that directly involves combination of water with methanol, rather

than through a water gas shift reaction (Huang et al, 2004).

CH3OH HCHO + H2 (8)

HCHO + H20 HCOOH + H2 (9)

HCOOH- CO2 + H2 (10)

Equations 8, 9, and 10 sum to equation (1). Another pathway could be:

2CH30H4 2HCHO + 2H2

2HCHO CH3OCHO (11)

CH30CHO CH3OH +CO (12)

CO + H20 CO2 + H2 (13)









which also sum to equation (1). Huang suggests that reactions 8, 9, and 10 predominate,

though the second scheme has intermediates more similar to those in the scheme used by

Lee, et al. Studies by Agrell et al, Lee et al, and Pumama et al all favor mechanisms in

which CO appears in a secondary reaction, such as the reverse water-gas shift, since it

tends to appear when the methanol has been consumed (Lee et al, 2004). Some general

trends in the reforming studies indicate that higher temperatures and longer contact times

increase CO concentrations. In most cases, the rate of hydrogen production seems to level

off with increasing temperature once 80% of the methanol has been consumed. For

example, in Shishido et al's study, the rate of hydrogen production was more than twice

at 250C as it was at 200C, but only increases by about 3% at 300C. Choi et al (2003,

2005) did studies of methanol decomposition without water, and with lower than

stochiometric amounts of steam, and determined that, unless the feed contained less than

43 mol % water, no methyl format was detected in the product gases. Neither were

dimethyl ether, or methane detected, and these only appeared at less than 24 mol% water

and less than 8 mol % water, respectively (Choi et al, 2003). All studies indicated that

CO formation was suppressed with increased water concentrations, though there was

slight variation between the amount of excess water and the level of CO suppression for

various catalysts. Generally, it ranged from a mol ratio of 1.1 to 1.4 for steam to methanol

to keep CO concentrations in the product gas below 2% at the temperature for 80%

methanol conversion.









Table 1-1 Summary of Proposed Rate Laws
Reference rate law
Lee -rM = ka exp (-Ea/RT) PMa (A + PH)b mol/kg/sec
ka = 2.19E9, Ea= 103 kJ/mol,
A= 11.6kPa, a = 0.564, b=0.647


notes
M = methanol,
H=hydrogen,
P= partial pressure,
-rM = rate for
methanol


decomposition
Purnama rsR= ki PCH30H m PH20 n rSR = steam
reforming rate
m= 0.6, n= 0.4
Apparent Ea= 76 kJ/mol,
Pre-exponential ko1= 8.8E11/kg/sec
kl= kol exp (-Ea/RT)
rRWGS= k2 Pco2 PH2 k-2PH20PcO
Apparent Ea = 108 kJ/mol
Pre-exponential k02 = 6.5E9 bar- sec- gcat1
k2= k02 exp (-Ea/RT)
Agrell(Jiang) rsR= k PCH30H 0.26 PH20 0.03 (PH2< 7kPa)
rSR= k PCH30H 0.26 PH20 0.03 PH2 -0.2 (PH2>7kPa)
k(T)= 1.9E12 exp(-100.9kJ/molRT)















Table 1-2: Summary of Steam Methanol Reformer Studies
Reference Agrell Peppley Jiang Lee Pumama Argawa Shishido
I


Activation Energy, kJ/mol
Temperature, C

Catalyst Composition,
Cu/ZnO/A1203 wt%

Feed Composition,
steam:methanol
Mass of Catalyst
Reactor Size
% Methanol Conversion,
equal molar reactants
2% CO Selectivity


100.9
175-270


66/23/11


1:1 and 1.3:1

50 mg
6mm dia
80% at 260C

275 C for 1:1
steam:
methanol,
310C for 1.3:1
steam:
methanol


102.8 105.1
160-260 170-260


103
160-260


76 92
120-325 260-
300


40/40/20 31.7/49.5/18. 64/24/10/, 2% 50 wt%
8 MgO Cu, others
not given
2:1, 1.5:1, 1:1 1:1


10/5/19
85


1.4:1


80% at
240C


Ig
7.5mm
80% at 235C

much less than


200 mg
10mm
80% at
250C


78% at
300C


not given
200-300

45/45/10


1.2: 1

200 mg


97.3% at 250C


1% at 250C









Lee's experiments concluded that only methanol and hydrogen partial pressure

influenced the reforming rate. Their mechanism is modeled on a methyl format reaction

route, shown by equations 4, 5, and 6 given above, with CO occurring only through a

reverse water-gas shift reaction in a secondary step. CO was not detected at reforming

temperatures lower than 200C and much less than 1 mol % during experiments at higher

temperatures. Steam to methanol ratios in their experiments were 1/1, 1.5 /1 and 2/1.

Increases in CO concentration were linked to increases in methanol in the reactants, and

it is suggested that the small amount was due to a reverse water-gas shift via increased

production of CO2 and H2, rather than the amounts predicted by methanol decomposition

to CO in the Peppley model. The role of water appears to be mainly to convert the methyl

format species to methanol and HCOOH, the former repeating reactions 4 and 5 again,

and the latter decomposing to CO2 and H2. Lee sites internal diffusion limitations for

changes in catalytic behavior above 200C for particle sizes of 0.3-0.42 mm diameter

pellets.

The model proposed by Purnama et al involves the overall reactions of steam

methanol reforming and reverse water gas reaction, and like Lee's model, considers the

reverse water gas shift reaction to be secondary, occurring as a result of a reverse reaction

involving the products of the steam reforming reaction.

Steam reforming (SR): CH3OH +H20 *- 3H2 +CO2,

KSR = kl/k1 = (PH23Pco2 )/(PCH30H PH20)

Reverse water gas shift (rWGS): CO2+ H2 _' CO+H20,

KrWGS = k2/k-2 = (PH20Pco)/( Pco2 PH2)









Purnama et al also determined that there were greater intraparticle diffusion

limitations for catalyst particle sizes of 0.71-1.0 than for 0.45 0.50 mm diameter

particles. CO concentrations increased with the increased catalyst particle size, and this

effect has been linked to the difficulty of product gases diffusing through the pellet

material at higher temperatures. This may be an analogous effect to the increased CO

concentration that occurs at higher residence times as well. The accumulation of products

can force the reverse reaction if they remain in proximity to the active sites. CO

concentrations at all reaction temperatures were below those predicted by the water gas

shift reaction, supporting other findings that CO2 does not occur via the water gas shift,

and that CO formation occurs in a consecutive reaction rather than through methanol

decomposition. (Purnama et al, 2004) Data complied much more closely with the reverse

water gas shift following steam methanol reforming reaction. The reasons for minimal

involvement of the water gas shift reaction can be attributed to competitive adsorption

between methanol and CO (Purnama et al, 2004).

The investigation done by Agrell et al(2003) also supports a methyl format route

with CO occurring in a subsequent, reverse water gas shift reaction. Agrell et al used

catalyst pellets of size 0.12-0.25 mm diameter. Like the other two models, CO production

increased with temperature and contact time, suggesting a relationship of the mobility of

product gases away from active sites as a function of temperature and diffusion

limitations. The increase in temperature can push the reaction rate to a quicker

completion, but then this can re-enforce the tendency of reverse reactions since product

gases can re-adsorb more strongly at the higher temperature.









In all three of the more recent models, (Purnama et al 2004, Lee et al 2004, Agrell

et al 2003), water participates directly in a steam methanol reforming reaction on the

catalyst, via a format intermediate, rather than the two step methanol decomposition and

water gas shift reaction mechanism indicated by previous models. If this is the case, in

the scope of including a methanol steam reformer in a portable fuel cell system, which

would include a water-gas shift reactor for converting CO (Zalc et al, 2002) may be

unnecessary. It may be more efficient to separate the hydrogen from other reformate

gases via separation membranes rather than invest in additional reformers for complete

gas conversion.

1.3 Heat Transfer Limitations in Packed Bed Reformers, Advantages in Monolith
Reformers

Many heterogeneous catalyst applications rely on forcing reactant gases through

porous catalyst pellets designed with a surface area and porosity to maximize the contact

with active sites on the catalyst. However, temperature and pressure gradients occur in a

packed bed configuration, which can result in material wear, incomplete reactions in

some parts of the reformer, and greater risk of sintering. This especially true of reactions

and catalyst materials which have a narrow temperature range of optimal performance,

and which have shown sensitivity to catalyst surface geometry, such as reactions with

alcohols. The diffusion limitations within catalyst pores have been linked to increased

CO production and catalyst deactivation (Purnama et al 2004, Lee et al 2004, Agrell et al

2003). Research has shown that the diffusion limitations can be reduced with smaller

pellet sizes, meaning that fewer reactant gases go through the pellets; instead, there is an

increased flow around the pellets. However, these smaller pellets also increase the

pressure drop along the catalyst bed. The problem with this is, not only that higher









pressure at elevated temperatures increases the likelihood of sintering and material wear,

but that product gases are not quickly removed away from catalyst sites after they react.

It has subsequently been suggested that a future design might be an "eggshell" type

catalyst, meaning a thin-walled, honeycomb type catalyst, with channels through which

the reactant gases would flow, rather than the microscopic diffusion flow through pellets

(Agrell et al, 2003).

1.3.1 Non-Isothermality in Packed Bed Reactors

Many packed bed reactors are modeled after a plug flow assumption, however, in

reactors with small ratios of tube diameter to pellet diameters, temperature gradients can

be significant enough to skew reactor data. (Karim et al, 2005) The models discussed

above have crushed and sieved commercial pellets to approximately less than one

millimeter diameter. Small tube diameter to particle diameter ratios, as in micro reactors

and the experimental pellet beds used in the studies above, result in a variations in

velocity and temperature. (Demirel et al, 2000) Changes in velocity and temperature

profiles caused by back mixing prevent accurate simulation of the temperature response

by the traditional plug-flow model, and a hyperbolic model of axial dispersion in packed

beds has been described. (Prasad et al, 2002) Catalyst dilution and the change of reactor

diameter from 4.1mm to 1.0 mm have shown higher apparent catalyst activity, indicating

heat transfer limitations within the pellet bed (Karim et al, 2005). Temperature gradients

of up to 40 K and 22 K have been observed within the 4.1 mm and 1.0 mm reactors,

respectively. (Karim 2005) For endothermic reactions, temperature gradients can lead to

lower catalyst activity; and for exothermic reactions, they can lead to reaction rates

several thousand times greater at the axis than at the wall (Karim et al, 2005). Heat

transfer at the wall cannot be neglected for reactor tube to particle ratios of less than 100,









and the ones sited in this paper range from 20 to 50. Karim et al tested a commercial

Cu/ZnO/Al203 catalyst in packed beds of diameter 4.1, 1.75 and 1 mm using 200 mg of

catalyst in the first experiment and diluted with 175 mg of inert material in the second.

The additional material effectively increased the bed length, thereby increasing the

external surface area for heat transfer. The effect was higher apparent catalyst activity,

but reduced reaction rate per bed volume; that is to say, there was higher methanol

conversion with the dilution, but a greater bed volume for the catalyst was needed. For

90% conversion, 25% more catalyst for the 1.75 mm diameter reactor than for the 1 mm

diameter reactor, and 85% more for the 4.1 mm reactor for 1.1/1 steam to methanol ratio

at 230C. Karim et al assert that these observations are linked to interpellet mass transfer

limitations, rather than intrapellet or interphase mass transfer limitations. However,

decreasing pellet size and reactor diameter to reduce interpellet mass transfer issues

increases the over all pressure gradient in the reactor, and dilution of the catalyst is not

feasible for commercial applications. Karim, Bravo et al then developed a wall coated

reactor of the same catalyst used in their packed bed experiments and compared methanol

conversion and CO selectivity. Their packed bed reactor was a 4.1 mm i.d. 8.55 mm long

quartz tube packed with 100 mg of 0.10 mm diameter catalyst pellets. Their wall-coated

reactor was coated with 75 mg catalyst on the inside of a 4.1 mm i.d. quartz tube 18 cm

long. For 80% methanol conversion, the wall coated reactor required a flow rate of 20

W/F (kg s/mol) compared to 34 W/F (kg s/mol) in the packed bed reactor, where W/F is

the ratio of the weight of catalyst to the fuel flow in mol/s. 90% conversion was achieved

at 27 W/F for the wall coated reactor and 47 W/F for the pellet bed reactor. The 2% CO

selectivity occurred at 92.5% methanol conversion in the wall coated reactor, and 85%









conversion in the packed bed reactor. Bravo et al show that the methanol steam

reforming reaction is catalyzed more efficiently in the wall-coated reactor, i.e. when mass

transfer and thermal gradients are reduced. Additionally, their findings support previous

studies that CO formation occurs at higher methanol conversions, and increases with

mass transfer limitations. If CO also occurs as a secondary reaction from the products of

the steam methanol reaction as hypothesized above, then these findings reflect the

importance of determining the right geometry and flow rate for the best contact time with

the catalyst, as well as reducing back mixing and turbulence, which increase contact time

that products have with the catalyst. It is important to note that the methanol synthesis

reaction has been studied over the same catalyst and similar temperature range, so

controlling reverse reactions is important.

1.3.2 Monolith Reformers

Monolith catalyst configurations for hydrogen production from methanol have

begun to appear in the past few years. Their advantages include improved temperature

control and greater reliability and durability. For portable applications, a catalyst-coated

monolith, made of either stainless steel or ceramic, is more structurally robust

(Lindstrom et al, 2003). However, most commercially available steam methanol catalysts

are in pellet form. Examples of various configurations include the honeycomb-like

monolith used by Lindstrom, catalyst coated brass wires (Horny et al, 2004), and layers

of stainless steel or glass etched plates that can be coated with catalyst (Park et al, 2005;

Yu et al, 2005; Terazaki et al, 2005). Several studies have been devoted to catalyst

adhesion and the effect of the substrate on catalytic activity (Kawamura et al, 2005;

Bravo, 2004). Horny et al found that coating catalyst on brass wires and placing the wires

close together like a cable created straight channels approximately 1 mm in diameter.









They found that the laminar flow and short radial diffusion resulted in nearly plug flow

with hydrodynamics similar to multichannel microreactors. Interestingly, they also found

that A1203 was a necessary component for catalytic activity over the Cu/Zn-coated brass

wires. This reflects an influence on the alloy microstructure on activity. Monolith

reformer studies have so far shown developments in the surface science in adhering the

catalyst to a monolith frame, such as a washcoated glass capillary, or catalyst coated

wires or etched plates (Fukuhara et al, 2004; Bravo et al, 2004; Horny et al, 2004). In

addition to isothermality and better control of reaction pathways, monolith reformer

designs have simpler fluid dynamics and can be used in a broader range of various scales.

Several micro-reforming systems have been developed that incorporate catalytic

combustors and vaporizors with the reformer. These chambers are arranged in layers in a

heat exchanger configuration. In the start-up of the micro-reformer system, hydrogen and

air from a gas supply are introduced into the combustion chamber to heat up the catalytic

combustor (Pan et al, 2005). The combustion chamber is adjacent to the vaporizer, and

the thermal contact allows enough heat transfer to vaporize the reactants that are supplied

to the methanol reformer. The catalytic combustor could also be designed to consume

some of the methanol rather than part of the hydrogen product gas, after start up. In the

combustion and reforming chambers, typically a commercial catalyst is ground, made

into a slurry, and adhered to etched channels, which make the layers of the reformer and

combustor. In the case of Pan et al, actual pellets smaller than 1 mm were loaded into

chambers that appear to be approximately 8 mm wide, according to their schematic. The

coated surface approach has advantages in scalability. Researchers found comparable

performance between the monolith and packed bed reformers (Lim et al, 2005; Lindstrom









et al, 2003). Though one detractor of the monolith reformer is that there is less catalyst

material per reactor volume, they are able to operate at higher temperatures (210-290C)

and achieve higher methanol conversion with less risk of sintering. Recall that many

pellet bed studies found that sintering can become a problem at temperatures as low as

220 C. There is higher methanol conversion at the higher temperature, but also, the

catalyst coating is more stably distributed and less likely to develop hot spots than in the

packed bed reformer. The effect of reduced catalyst material per volume in the monolith

reformer is compensated for by the greater operating temperature range and reduced back

flow, which is considered undesirable (Demirel et al, 2000). Monolith and metallic heat

exchangers offer advantages over pellet beds in terms of mechanical strength, which is

more ideal for portable applications (Giroux et al, 20005), additionally, the heat transfer

through the monolith structure is much more efficient than conduction from the walls

through the pellets, which is an important advantage in start-up times for fuel reformers

that supply fuel cell systems (deWild et al, 2000). CuO/ZnO pellets crumble easily, and

over time, the interpellet diffusion discussed by Bravo et al will become more of a

problem in a mobile reformer experiencing vibration.

It should be noted that the reformer systems in these monolith studies are very

small, under 100 cm3 in total volume, with the dimensions of the methanol reformer

section being less than 1mm in diameter. Being fully enclosed in the system, they do not

lend themselves well to in situ studies. The pellet bed experiments were also conducted in

small diameter tubes. Thermocouples can be employed for in situ temperature

monitoring, but generally the catalyst is analyzed by conducting a spectroscopic

measurement of copper surface area prior to reaction, then product gases are analyzed.









reforming



40 mm








125 nun
combustion 12 m
gases f



vaporization

Figure 1-3. Example of a micro-reformer, adapted from (Pan et al).

This was the predominant approach for both pellet bed and monolith reformers. Some

recent studies using Raman spectroscopy have verified surface intermediates and changes

in surface structure and chemistry, with much emphasis on developing the investigative

process itself (Gunter et al, 2001; Grunwaldt et al, 2000). In situ studies have also been

performed to assess the effect of the gas phase composition on adsorbate species (Knop-

Gericke et al, 2001), surface vibrational spectroscopies of chemisorbates (Park et al,

2002), copper oxidation states (Kudelski et al, 2004), among others. There has been an in

situ study over a catalytic surface in which the gas phase is sampled at various locations

along the reactor (Springmann et al, 2002), but this approach has disadvantages in that

the flow is disturbed during the sampling and the boundary layer cannot be observed. The

monolith reformer concept has shown advantages in durability and in thermal and mass

transport issues; however, since residence time has been linked to undesirable secondary

reactions in steam methanol reforming, the reactor must be scaled properly for the









prescribed reforming rate. Understanding the effect that flow rate has on the two

dimensional gas phase composition (reactor length and height above catalyst) would

benefit from in situ studies.

The reformer in this study employed commercial steam methanol reforming

catalyst pellets, however they were arranged on a tray so that reactant gases primarily

flowed over the catalytic surface, rather than through the pellets. This reactor was

designed on a scale so that the spatial variation of the gas phase over the catalyst surface

could be studied using Raman spectroscopy and a readily available catalyst, in order to

develop some scale factors for reforming in a monolith configuration. Perhaps in the

future, the adhesion processes described by G.-G Park, Lim, or Bravo et al will be

developed on larger surfaces so that the spacing between two catalytic surfaces can be

studied in this in situ configuration.

1.4 Raman Spectroscopy for Study of Catalysis

Raman spectroscopy has become popular in the past 5 years for in situ study of

catalytic surfaces (Weaver et al, 1999; Stair et al 2000, 2003; Wachs 1999; Knozinger et

al, 1999). The time scale for Raman spectroscopy allows simultaneous study of kinetics

and molecular characterization (Wachs 1999), hence recent emphasis has been on the

surface chemistry. Supported metal species have different vibrational energies depending

on their arrangement; similarly, adsorbate species vary in vibrational energies from their

gas phase counterparts. This is due to the bond distortion, and therefore vibrational

frequency, that occurs when a molecule adsorbs to a surface, as well as bond structures

associated with crystal lattices or metallic oxide surfaces (Haw 2002). When molecules

adsorb on these, both the vibration of the bonds of the adsorbed molecule and the of the

catalytic surface are changed. The gas phase composition gradients arise out of the






40


combined effects of surface chemistry, reactant molar ratio, and the free stream velocity.

This Raman study over a catalyst differs from other surface studies by determining the

relationship that the free stream velocity has on the two-dimensional scale of the

composition boundary layer.














CHAPTER 2
RAMAN SPECTROSCOPY

Electromagnetic radiation on incident any matter can be emitted, absorbed, or

scattered; and most materials have some combination of absorption and scattering. In

radiation scattering, a photon of light energy incident on a molecule will induce a dipole

moment in the molecule or atom. The energy of a photon is hvinc, where h is Planck's

constant and Vine is the frequency of the incident radiation. The frequency is yvc/k,

where c is the speed of light and X is the wave length. In elastic scattering, the emitted

radiation has the same frequency, and therefore the same photon energy, as the incident

radiation, but the direction (i.e. distribution) of the scattered energy depends on the

incident angle, size of scattering, and polarization of incident light. Types of elastic

scattering are distinguished by the refractive index and the wavelength of the incident

radiation as compared to the particle size. Raman scattering, however, is inelastic; and

emitted radiation shows a shift in frequency from the incident radiation, which is

generally irrespective of scattering angle (Ingle et al, 1988).

With Raman scattering, the incident radiation and the induced dipole moment

create a quantum shift in the vibrational modes of the molecule. In the case of Raman

scattering, the excitation energy (hvinc) does not exactly match the difference between the

vibrational quantum energy levels in the molecule. Therefore some of the energy is

absorbed, and the remainder is emitted. The emitted photon is therefore shifted in

frequency by the amount of vibrational energy. As described above, whenever an incident

photon adds energy to a molecule such that its vibrational energy is increased, the emitted









photon has a proportionately lower frequency. This is termed a Raman Stokes shift, and

is more common than the reverse phenomenon, anti-Stokes shift. This is because the

ground state is the most highly populated, hence it is more likely that the incident photons

encounter a molecule in its ground vibrational state. Anti-Stokes shift can occur as

molecules relax to a lower vibrational state during the scattering interaction. The Stokes

shift has frequency of vinc Vvibration, and the anti-Stokes shift frequency is vine + Vvibration.

Both vibrational and rotational modes of a molecule can cause a Raman shift, although

the rotational energies are usually much less than the vibrational energies. A molecule is

considered Raman-active if it can undergo a change in polarizability during one of its

normal modes of vibration. Atoms are not Raman active since their polarizability does

not change with vibration.

The theory and requirements for Raman scattering are as follows: An electric field

E interacts with polarizable electron clouds and induces a dipole moment [t in (Ingle et al,

1988).

E= Em cos (2 pv ex t)

[t in= aE

where Em is the amplitude of the wave, pt in is the induced dipole moment, and a is

the polarizability in J C2m2. Polarizability varies with internuclear separation around the

equilibrium polarizability, o, occurring at the equilibrium bond distance re.

a= ao + (r- re)( 6a/6r)e +....

The change in nuclear distance is

(r- re)= rm cos (2pvvt)









where rm is the maximum internuclear separation, and v is the frequency of vibration.

The criteria for Raman shift is that the following integral be non-zero:


R = f. *i (r- re) (6a/S6r)e Wj dr

where Y and Y are the complex conjugates of the wave function, representing the

probability distribution of molecules in a particular state.

Each type of Raman-active molecule causes a particular shift in the Raman

spectrum resulting from the nature of the atomic bonds in that molecule, which is what

makes Raman spectroscopy useful for chemical identification. Molecules with more than

one type of bond can show more than one spectral line. Methanol, for example, has a

spectral line associated with its O-H bond, that is distinct from, but close in wavelength,

to the O-H bond in water. It can also show spectra for the C-O bond and the C-H bonds.

Table 2.1 gives typical Raman shifts of the species investigated in this study.

Table 2-1 Raman Shift
Emission (nm)
Raman Shift from Excitation
Species (cm-1) of 355 nm
H2 4156 416.0
H20 3650 407.5
CH3OH 2955 396.2

In a sample Raman spectrum taken during reforming, there are several peaks that

reflect the vibrational modes of different bonds within the same molecule. For example,

near the water peak at 3650 cm-1 there is a methanol peak. These two peaks have similar

spectra due to the vibrational modes of the OH bonds. However, they still differ owing to

the interaction of other atoms, -CH3 in the case of methanol, and the -H in water. The

hydrogen peaks are located to the right of water, near 4156 cm-1. There are two possibly

because of the existence of ortho and para hydrogen. In ortho and para hydrogen, the










atoms have either parallel or opposite electron spins, which could affect the induction of

a dipole moment between the two atoms. On the far left are two additional methanol

peaks, which are associated with the C-H bonds in methanol (Figure 2-1).



250





S200
% H2
Co


SOH from
150 C-H bonds of H
n.o H20
Smethanol









5I I I I ,I
100 -





50"
2800 3200 3600 4000 4400
Raman Shift (cm1)


Figure 2-1. Typical Raman Spectrum of Steam Methanol Reforming















CHAPTER 3
REACTOR DESCRIPTION

3.1 Reactor

In this study, the composition of the chemical boundary layer above a catalyst in

steam methanol reforming was evaluated as a function of reactor length and flow rate.

Katalco 33-5 catalyst pellets manufactured by Johnson Matthey were used for the

catalyst. The pellets are comprised of 64 wt % CuO, 24 wt % ZnO, 10 wt % A1203, and 2

wt% MgO, and measure 5.2 mm diameter by 5.4 mm length. They were packed end-to-

end in a single layer in an aluminum pellet bed, shown in Figures 3.1 and 3.2. The pellet

bed was heated by cartridge heaters placed in each end of the bed. Three access ports

along the length of the bed allow for thermocouples F, G, and 4 to measure the bed

temperature, as shown in Figure 3-3.


End View 0.676 m.



/


0.125m. in 0.10
LI_ 0.10 lOin
0.356 in.



\ -1.375 m. dia

0.138 in.- P- 1.075 in. -4 i. dia

Figure 3-1: End View of Pellet Bed









Top View

24.00 in.


0.18 in. o0.18in.
1 0.138 in.



________1.075 in.
1.352 in.




Figure 3-2. Top View of Pellet Bed

The pellet bed is centered within the reactor, and its temperature is regulated by

routing thermocouples F and G, and their corresponding cartridge heaters F and G,

through a temperature controller. The reactor itself is constructed from 2.5" stainless-

steel vacuum housings purchased from Huntington Laboratories. The housing is wrapped

in heater tape E, which is routed to a temperature controller and thermocouple E, and

covered in fiberglass tape insulation. The reactor housing is stainless steel with stainless

end caps, valve fittings, and view ports.

The reactants are preheated to ensure complete vaporization by passing the

reactants through a pre-heat section filled with brass pellets to promote heat transfer.

Methanol and water are delivered to their respective tubes via peristaltic pumps, and each

tube has a nitrogen feed for carrier gas. The reactant delivery tubes are wrapped in heater

tapes (A and C for steam, B and D for methanol) and insulation, and the temperature is

controlled by thermocouples A and B and temperature controllers. See Figure 3.3. The

temperature of the stream above the catalyst is measured by thermocouples 1, 2, and 3.

Power for the heating elements is delivered through the temperature controllers. When










centered within the reformer, the actual catalyst surface is approximately 5.5 inches from

each end. The reactant and product streams enter and exit through fittings with an inside

diameter of approximately 0.25 inches. The catalyst surface is approximately level to the

inlet stream, however, the gap between the ends of the pellet bed can cause some mixing,

which may result in non-ideal flow conditions. This will be discussed in later chapters.

5 channel temp indicator
1; l l Flat Bed Steam Methanol
Reformer


Key
Temperature
controller
flow controller, nitrogen
methanol pump
water pump
S ower to heater taoes A. C. B. D. E and to cartridae heaters F and G
thermocouple


Figure 3-3. Reformer assembly.


3.2 Optical Configuration

The light source is a Q-switched, Neodimium:Ytrium-Aluminum-Gamet

(Nd:YAG) laser firing at 5 Hz with a 8 mm beam diameter. For Raman measurement,

frequency-tripled light at 355 nm was used. Based on earlier research (Ball 2005), it was

found that 355 nm excitation provided an optimal Raman signal as compared to 532 nm

excitation. Theoretically, the Raman signal should increase as the inverse fourth power of









wavelength, hence the 355 nm should provide a better signal. One must also consider

fluorescence with UV-excitation, however, fluorescence was not found to be an issue

with the current system.

The optical configuration is shown in Figure 3-4.The mirrors all have combinations

of 532 and 1063 nm anti-reflection coatings to minimize the amount of residual laser

energy at these two wavelengths. The laser beam is focused to a horizontal line using a

500 mm focal length cylindrical lens. The resulting beam is approximately 500-[tm high

by 8 mm wide.

Raman scattered light is collected in backscatter using a 4" diameter pierced mirror

and then focused to a fiber optic using a 4" diameter, 160 mm focal length lens. Prior to

entering the fiber optic, light was passed through a 375-nm high-pass filter (Filter 1) and

450-nm short-pass filter (Filter 2), which functioned to block residual 355-nm and 532-

nm light, respectively. Finally, a sharp-edge high-pass filter (Filter 3) was used to further

reduce 355-nm stray light.

Light was fiber-coupled to a 0.275-m grating spectrometer and recorded with an

intensified charge-coupled device (CCD) array detector. The detector was synchronized

to the Nd:YAG laser, with Raman spectra recorded using a 300-ns gate centered on the

laser pulse. In general, Raman scattering is prompt, hence coincident with the -20-ns

full-width laser pulse. However, the wider gate was used to ensure capture of the entire

signal due to the -150-ns of laser-pulse jitter. Table 3.1 lists equipment and

specifications. This reactor was constructed from off-the-shelf components.










Beam Dump


Beam Dump


Beam Dump







355 nm turning mirror,
450, passes 532/1064
nm (2 mirrors)

Cylindrical

Aperti






Pierced Mirror



Scattered Light-


lens


ure


1064 nm Nd:YAG Laser


Beam Dump


3355 nm turning mirror,
45, passes 1064 nm


Collection


Filter 1

Filter 2

I Filter 3


Fiber Optic Cable


Beam Dump


Adjustable Platform
to raise and lower reactor,
1 at each end


Figure 3-4: Experimental Set-Up for Raman Specrtroscopy









Table 3-1. Equipment and Specifications
Equipment Specification
A and B: 1"x8' Omegalux FGH101-080 120V, 830W,
Heater Tapes A, B, C, D, E 6.92A
C and D: 0.5"x8' Thermolyte B1H051-08, 120V, 420W,
3.5A
E: 0.5"x4' Thermolyte B1H051-040, 120V, 210W,
1.75A, 482C

Thermocouples A-G, 1-4 Type K

Cartridge Heaters F and G 0.25"dia, 9.5" length McMaster-Carr #3618K197

Peristaltic Pumps (2) Fisher Scientific Low Flow 13-876-1

Nitrogen Flow Controller Alicat Scientific MC-5SLPM +/- 0.05 slpm

Temperature Controllers (3) Omega CN872RTD-500C +/-5% temp. range

View Ports VP-151 7056 glass

Multichannel Temp. Indicator Omega Digicator 412B-K, 0. Ideg resolution

Collection Lens 4" Plano-convex, Comar, 160-PG-100

100 mm x 100 mm, center pierced 0.5"in +-0.2"dia,
Pierced Mirror Rolyn Optics, 60.2475

Cylindrical Lens Piano-Convex square, L30.0/ W30.0/FL500, 022-2546

Filter 1 high pass filter, Comar, 375-GY-50

Filter 2 short pass filter, CVI Laser LLC, SPF-450-2.00

Filter 3 Comar, 685-IH-25

Big Sky Laser 1064 nm Q-switched (Nd:YAG)

turning mirrors: 355nm
blocks 1064, 1 CVI Laser LLC, BSR-31-2025
blocks 532/1064, 2 CVI Laser LLC, BSR-315-2025

peristaltic pumps, 2 Fisher Scientific Low Flow, 13-876-1














CHAPTER 4
EXPERIMENTAL METHODS

4.1 Catalyst Bed Reduction

To be active, the catalyst must be in the reduced state. That is, the pellets must be

heated in a nitrogen/hydrogen environment to reduce (remove oxygen from) the pellets.

The product of the reduction process is steam. Based on information from the

manufacturer, the process for the catalyst reduction is as follows:

1. Heat catalyst in an N2 atmosphere to around 1300 C (-3/min). Heating the pellet
bed in air can oxidize the copper or zinc in such a way that the pellets are sintered
or deactivated, such that they cannot be reduced. Introduce hydrogen at a
concentration of about 1% of total pressure.

2. Observe temperature profile down the bed. It is recommended to limit the
maximum bed temperature to 240 C during this operation.

3. After the temperature wave has progressed down the bed, increase the hydrogen
concentration to 15%, watching for temperature spikes.

4. Maintain this flow rate and concentration until the temperature profile is steady.

5. Though the product data sheet states that operation can occur over the range of 200-
320 C, the exothermic process will heat the catalyst to higher than the bed
temperature. Therefore it is recommended to start at the lower temperature.

Between test runs, the reformer was turned off for safety and power concerns. To

shut down the reformer following a test, the reformer was flushed with nitrogen at the

reaction temperature to remove any trace reactants or products that may condense onto

the catalyst bed at room temperature. Then the reactor inlet and exit valve will be closed,

with a slight over-pressure of nitrogen.

The catalyst may exhibit changes in activity over long run times, or differences in

catalytic activity depending on the initial reactant composition. A study of long term










operation (over 100 hours) conducted by Dams et al (2001, 2002), et al indicated that

temperatures between 200 C and 220 C resulted in the least amount of change in

catalyst activity, as well as minimal coke build up. Although this study is much shorter,

the temperature range was limited to 200-220 C in order to preclude the any

complications involving deactivation or reverse reactions discussed in earlier. Typical

Raman spectra observed during reduction are shown in Figure 4-1 .

280
H
2

240 H



j 200


SMid-Point
2 160



120 Initiation



80 .
3200 3400 3600 3800 4000 4200
Raman Shift (cm-1)
Figure 4-1 Catalyst Reduction Raman Spectra

Initially, there is a high concentration of hydrogen. As reduction proceeds, water

appears as the hydrogen reacts with the CuO, leaving pure copper sites. As the reduction

nears completion, the water is flushed out of the reactor and hydrogen reappears, as

shown in Figure 4-1.







53


7.0
-*- Hydrogen Signal
--- Water Signal
6.0
*
S5.0


1 4.0
*0) 0


CU
3.0 *


2.0


1.0


0.0 I I I
0 20 40 60 80
Time (min)
Figure 4-2 Water and Hydrogen Signal vs Time During Reduction

Figure 4-2 shows the changes in hydrogen and water signals during reduction.

4.2 Reformer Operation

For the current study, it was desirable to run the methanol reformation process at a

steam/methanol molar ratio slightly greater than 1. The actual flow parameters are

summarized in tables 4.1 and 4.2, noting that the two flow conditions will be referred to

as Flow 1 and Flow 2. The overall reformation process is summarized in this section.

Table 4-1. Reactant Flow Rates and Pump Settings
Water flow rate Methanol flow rate

pump dial liquid flow pump dial liquid flow
setting rate setting rate
(fast) (ml/min) (fast) (ml/min)
2.5 0.475 2.75 1.0
9 0.65 9.2 1.37









Table 4-2 Flow Parameters
Flow 1 Flow 2
Nitrogen 0.70 1//min 0.96 1/min
Methanol 1 ml/min 1.37 ml/min
Water 0.475 ml/min 0.65 ml/min
Steam/Methanol
molar ratio 1.079/1 1.054/1

Inlet velocity 7.0 cm/s 9.7 cm/s

Residence time 6.3 s 8.4s


4.3 Reformer Operating Procedure

Reformer operating and shut-down procedures were devised by the researcher. The

laser operating and shut-down procedures were devised by the researcher and D.W.

Hahn. The procedure for operating the reformer and taking spectroscopic data is as

follows:

I. Start Up Procedure for Steam Methanol Reformer

1. Plug in power.

* Switch on power strips.

* Turn on all three temperature controllers and set to 2000C.

* Open nitrogen cylinder.

* Turn on nitrogen flow feed to steam and methanol pre-heat tubes.

* Verify that methanol and water supplies are full.

* Turn on fume hood. Open reactor inlet and exhaust valves, connect exhaust to fume
hood.

* Wait for bed to reach reforming temperature. Thermocouple 4 will show 220 C
and thermocouples 1, 2, and 3 will show a range of 160-185 C.

* Set methanol and water pumps, and the nitrogen flow controllers to either Flow 1
or Flow 2 conditions, as summarized in tables and

II. Start-Up Procedure for the Laser









1. Wear laser goggles and turn on laser warning light.

2. This experiment uses a frequency-tripled Nd:YAG laser.

3. Turn on CCD detector:

a. Turn on nitrogen purge for CCD.

b. Check rotameter for full-scale nitrogen flow

c. Turn on Neslab water chiller.

d. Turn on CCD controller

e. Turn on Pulse generator.

4. Turn on Nd:YAG laser:

a. Turn on safety warning light for Nd:YAG laser

b. Turn key to "on" position.

c. Nominal front panel settings:

i. F5.0 (frequency of flash lamps, 5 Hz)

ii. 180 (Q-switch delay, microseconds)

iii. P01 (pulse selector, =5Hz)

iv. Push "run" for operating menu.

v. Push start/stop button to initiate flash lamps.

vi. Push shutter button to initiate laser beam.

d. Set laser power to 12.5 J flash lamp pump (-4 mJ/pulse)

i. c. Open external shutter to front of laser.

5. Spectroscopy software, Laser and CCD Settings:

a. External Trigger: install 355 nm trigger cable, set trigger to 0.75 volts,
positive.

b. Detector Delay: 0.05 us

c. DetectorWidth: 0.30 us

d. Window: 406 nm center wavelength









e. 250 shots per measurement

f. 100 rows binned

6. Verify that all beam dumps, filters, and mirrors are in place according to diagram.

7. To shut down, turn off laser and CCD in reverse order, and run CCD purge gas for
20 minutes.

III. Taking Spectroscopic Data:

1. Allow reforming gases to run for 15 minutes after start up to reach steady state.

* Verify that the reformer is at the initial height: support screws will be adjusted to
their maximum height. Verify that the reformer is level along the bed length and
across the bed at the reformer supports.

* Take spectroscopic data as a 250-laser pulse averaged spectrum.

* Lower reactor to the next height below laser.

* After data is taken for the 5 heights reset the reactor to the original position.

IV. Shut Down Procedure for Reformer

* Turn off water and methanol pumps, and close the methanol and water valves to
the reformer.

* Continue to run nitrogen through the reformer for 10 minutes and turn off
temperature controllers.

* Close exit, then entrance valves on reformer, turn of nitrogen flow controllers and
power strips, and unplug power. Close nitrogen cylinders.















CHAPTER 5
SPATIAL RESOLUTION

5.1 Spatial Resolution of the Optical Set-Up

For the current study, the reactor was constructed from standard vacuum hardware,

with the optical access provided via 6-way crossed, as shown in Figures 5.1 and 5.2.

While this provided a flexible and low-cost design, optical resolution was compromised

by the rather long optical path of 150 mm. Note that the reactor bed has a width of only

about 25 mm. To assess the optical resolution, the reactor was removed and the following

measurements were recorded.


Thermocouple


Dead Air Space, x3
Catalyst Pellets lctions




Optical Path


Reformer Housing


Thermocouple


Figure 5-1 End View of Reformer













Thermocouple



Catalyst Pellets Dead Air Space, x3
lootio ns




Reactant Flow




Catalyst Tray Reformer Housing









Thermocouple



Figure 5-2 Side View of Reformer

Spatial resolution across the reformer width was measured by placing a jet of

hydrogen, from a 4 mm diameter tube, at 4 liters/min. at the position of the center axis of

the reactor. Spectra were taken at increments on either side of the center axis of the

reactor. Negative values in position are locations closer to the laser source, and positive

values are further away. The focal point as the laser is focused through the cylindrical

lens aligns with the zero location, as shown below. Using this essentially point-source,

Raman spectra of hydrogen were recorded while the jet was translated across a distance

equal to the optical path of the reactor. The resulting signals are presented in Figure 5-3







59


Clearly the strongest signal corresponds to the location of the catalyst bed, however


additional signal is also present from the dead space between the windows and the


catalyst bed. Integration of this plot reveals that approximately 1/3 of the recorded


spectral signal during reformation corresponds to the catalyst bed position (between -


10mm and +10 mm on Figure 5-3).


Spatial Resolution
140

*
120 -
*
100 -


80 -




) 40
15 6*





20 *

*
20
0





-20 ., , ,
-860-60 -40 -20 0 20 40 60 80
Position (mm)


Figure 5-3 Spatial Resolution

5.2 Calibration for Gas Concentration

Prior to analysis of the recorded spectral data, a calibration must be performed to


convert the Raman signal to an equivalent molar concentration. This procedure was


performed for all three species of interest: H2, H20, and CH30H. For all calibration, the


catalyst bed was removed, but everything else was fixed in position. While, ideally, the

Raman cross-section is independent of temperature, measurements were recorded for







60


both ambient and heated conditions to veryify temperature independence. The

temperatures given for the calibrations are in the gas phase at the center of the reactor.

Overall, the gas phase temperatures ranged from 155C to 190C along the length of the

reactor. The thermostat range for resetting the temperature controllers resulted in a

variation of +- 5C detected in the gas phase.

Hydrogen was calibrated using a nitrogen flow rate of 5 liters/min for each

hydrogen flow rate of 0.0, 0.25, 0.50, 1.0 and 2.0 liters/min at 190C. These

measurements were repeated at ambient temperature. The temperature independence and

resulting calibration curve for hydrogen is shown in Figures 5.4 and 5.5. A highly linear

curve results, with both ambient temperature and high temperature values falling on the

curve.




35
*298 K
S450 K
30


25


I 20
U)

0) 15
0
-D
10 I


5 *


0

0.000 0.002 0.004 0.006 0.008 0.010 0.012
Hydrogen Concentration (mol/1)


Figure 5-4. Temperature Independence of Hydrogen Signal







61





35 Y = MO + M1*x + ... M8*x + M9*x9
MO 0.021509
30 M1 2420.5

R~ 0.99973
S25


c 20

0



10 -
c
15





10





0.000 0.002 0.004 0.006 0.008 0.010 0.012

Hydrogen Concentration (mol/l)


Figure 5-5 Hydrogen Calibration Curve

Methanol was calibrated at the 190C with a feed of nitrogen and methanol in the

following proportions: 1.5 1/min N2 and 1.4 ml/min liquid methanol,;3 1/min N2 and 1.4

ml/min liquid methanol; 5 1/min N2 and 1.4 ml/min liquid methanol; and 5 1/min N2 and

1 ml/min liquid methanol. As discussed earlier (see Figure 2.1), methanol is characterized

by three distinct peaks in the current region of interest, namely, the one near water,

labeled P/B A, and a double peak near the left end of the spectra, labeled P/B B. The

resulting calibration response is shown in Figure 5-6 for peaks A and B. As seen in the

figure, peak B yields a better response and also avoids nearing the water band.

Ultimately, peak B was used for this study.












120 y = 0.7467 + 1160.5x R= 0.9955

S y = 1.3468 + 11490x R= 0.99935
100
I M Meth P/B A
SMeth P/B B
n 80


0)
U) 60

c
4-'
( 40



20



0

0.000 0.002 0.004 0.006 0.008 0.010
Methanol Concentration (mol/l)


Figure 5-6 Methanol Calibration Curve

Water was calibrated at the 190C with a feed of nitrogen and water at standard

temperature and pressure in the following proportions: 1.5 1/min N2 and 0.55ml/min

water,;3 1/min N2 and 0.55 ml/min water; 5 1/min N2 and 0.55 ml/min water; and 5 1/min

N2 and 0.475ml/min water.

Though our peristaltic pumps are supposed to be identical, the water pump yielded

a non-linear calibration curve, although it was repeatable. Water also showed the greatest

variation in calibration spectra, as well as in the reforming spectra, although such

behavior was difficult to link to a specific issue.







63



20 -
0-- y= 1.5855 + 1968.2x R= 0.98171





15





10 -





5 -





0

0.000 0.002 0.004 0.006 0.008 0.010
Water Concentration (mol/l)
Figure 5-7. Water Calibration Curve














CHAPTER 6
DATA

Spectral data during the reaction were taken at 5 heights and at each of the three

axial locations. Position 1 refers to the location closest to the entrance (x/L =0.08),

position 2 refers to the center of the reactor (x/L= 0.5), and position 3 is nearest the exit

(x/L=0.92). Data were taken at 5 heights (up to 6.4 mm) above the catalyst, which was

deemed sufficient given the overall resolution. Ideally, the mass balance of the reaction

would include the concentrations from the catalyst surface up to the top of the boundary

layer. The concentrations at any specific height vary due to differences in rates of

diffusion for each species, as well as any recirculation that may be generated by the dead

space near each optical region, as well as by surface roughness created by catalyst pellets

that may have been dislodged out of plane. The data presented here are in concentration

(mole/liter) at particular locations for given flow rates. For each position, height, and

flow rate, Raman spectra were recorded in duplicate for each run, using a single spectrum

corresponding to a 250-shot average. Each experiment was repeated a minimum of 4

times, for a total of 2000 shots averaged over multiple experimental runs. For the

following discussion, results are presented as a function of axial position, height above

the catalyst bed, and for the two different flow rates. Of all the experimental runs, data

taken at height 5 were inconsistent with trends seen at other heights. It is unclear whether

this occurrence is an artifact due to the flow and geometric conditions of the reactor.

Figure 6-1. shows typical Raman spectra observed during reaction. From Position 1 to









Position 3, the intensities of the spectra for methanol and water both decrease while that

of hydrogen increases.


-- Flow 2, Position 1
S Flow 2, Position 2
-- Flow 2, Position 3


700


600


500


400


300


200


100


H2


CH3OH


H-O


0 1 I I I 1 1 1 I
2500.000 3000.000 3500.000 4000.000 4500.000

Raman Shift (cm-1)
Figure 6-1 Typical Raman Spectra Observed Along the Axis of the Reactor

6.1 Variation of Species Concentration at Each Position, at Flow 1

The next three graphs (Figures 6-2, 6-3, 6-4) show the concentrations of the two

reactant species, water and methanol, and one product gas, hydrogen, at one position and

one flow rate. It can be shown that the concentrations of the reactants decrease along the

bed length while hydrogen increases.


--11













* H2, position, F1
* CH30H, position, F1
* H20, position, F1


0.02




0.015




0.01


0.005


1.000 2.000 3.000 4.000 5.000 6.000 7.000

Height (mm)

Figure 6-2 Position 1, Flow 1 Species Concentration vs Height


* H2, position, F1
* CH30H, position, F1
* H20, position, F1


0.011


0.01


0.009


0.008


0.007 -


0.006


0.005 I
1.000 2.000 3.000


4.000 5.000 6.000 7.000

Height (mm)


Figure 6.3 Position 2, Flow 2 Species Concentration vs Height


I







67



H2, position, F1
U CH30H, position, F1
H20, position, F1

0.008

0.007

0.006 -

0.005
0
E
o 0.004 -
E -- TT
8 0.003 -

0.002

0.001

0 I I
1.000 2.000 3.000 4.000 5.000 6.000 7.000

Height (mm)
Figure 6-4 Position 3, Flow 3 Species Concentration vs Height

6.2 Variation of Species Concentration at Each Position, at Flow 2

At flow 2, which is 1.4 times faster than flow 1, there is a more dramatic change in

concentrations away from the catalyst surface, especially at position 1. This would reflect

the scale of a thinner boundary layer at higher velocities. In both flows 1 and 2, height 1

at position 1 shows similar concentrations of water and methanol, while at height 1 of

positions 2 and 3 there is much less water than methanol.

6.3 Variation of Species Concentration along Length and Height above Catalyst at
Flow 1

In the next three graphs (Figure 6-8, 6-9, 6-10) show the concentration of each

species as it changes along the length and height above the catalyst. Methanol predictably

decreases along the catalyst bed, but shows the greatest variation in concentration at

Position 1.












* H2, position, F2
* CH30H, position, F2
* H20, position, F2


0.01 h


1.000 2.000 3.000


4.000 5.000 6.000 7.000


Height (mm)
Figure 6-5 Position 1, Flow 2 Species Concentration vs Height


H2, position, F2
CH30H, position, F2
H20, position, F2


0.011


0.009 -


0.008


0.007

{


0.006 -


0.005
1.000 2.000 3.000 4.000 5.000 6.000 7.000

Height (mm)


Figure 6-6 Position 2, Flow 2 Species Concentration vs Height


0.035


0.03


0.025


0.02


0.015


0.005











H2, position, F2
M CH30H position, F2
H20, position, F2




0.008



0.006 -
o

0
0.004 -



0.002



0
1.000 2.000 3.000 4.000 5.000 6.000 7.000

Height (mm)


Figure 6-7 Position 3, Flow 2 Species Concentration vs Height

Likewise, water also decreases as expected, showing a similar variability at Position 1 as

methanol. This would seem to indicate an unsteady flow at Position 1, which may be

attributed to the peristaltic pumps. The pumps essentially force increments (i.e. discrete

drops) of water through with each turn of the pump wheel, so the flow could possibly

have enough variation to be shown by the Raman spectra. There is a decrease in

hydrogen concentrations at Position 3. Since both water and methanol concentrations are

shown to continue decreasing at that position, it is likely that the apparent reduction in

hydrogen concentration is a combined effect of back mixing with unreacted species at the

end of the reactor and a higher speed of diffusion for hydrogen. Previous studies have

cited that reverse water-gas shift reaction has been associated with higher temperatures

and longer contact times. However, if such a reaction were responsible for the reduced











hydrogen, the water concentrations would have increased. This is not the case, as water

and methanol concentrations both decrease monotonically along the length of the reactor

for both flows. Note that the diffusion coefficients of hydrogen, methanol, and water

have ratios of 13.3: 2.8: 4.8 respectively, at 465 K, the reactor temperature.



CH30H, position, F1
CH30H, position, F1
I CH30H, position, F1

0.018

0.016

0.014 -

0.012
0

0.01

8 0.008

0.006

0.004

0.002 .
1 2 3 4 5 6 7

Height (mm)


Figure 6-8 Flow 1 Methanol Concentration at 3 Positions vs Height

6.4 Variation of Species Concentration along Length and Height above Catalyst at
Flow 2

At the higher velocity flow, less back mixing occurs at the Position 3, and this is

supported by the fact that the hydrogen concentrations at Positions 2 and 3 are much

closer together in magnitude. If the flow were faster, it is likely that the hydrogen

concentration at position 3 would have increased as it had from Position 1 to 2. Methanol

and water concentrations also decrease from Position 2 to Position 3, but their values are

much closer together than in flow 1.


















0.012



0.01



0.008



0.006



0.004



0.002



0


2 3


5 6


Height (mm)



Figure 6-9 Flow 1 Hydrogen Concentration at 3 Positions vs Height


H20, position, F1
H20, position, F1
H20, position, F1


0.015




0.01


0.005


3 4 5 6 7

Height (mm)


Figure 6-10 Flow 1 Water Concentration at 3 Positions vs Height


H2, position, F1
H2, position, F1
i H2, position, F1







-{
-{



-I











At flow 2, it appears that the majority of the reaction occurs between Positions 1 and 2,

then the gradients between Positions 2 and 3 sharply decrease. At flow 1, on the other

hand, the differences in concentrations between Positions 1 and 2 are more similar to

those at Positions 2 and 3. At a lower flow rate, un-reacted species have more time to

diffuse through to the surface of the catalyst. See Figures 6-11, 6-12, 6-13.



0.009

0.008 -

0.007 -

0.006 -
0

E 0.005 H2, position, F2
.2 H2, position, F2
S- H2, position, F2
8 0.004
0
0.003

0.002 -

0.001 I I I
1 2 3 4 5 6 7

Height (mm)

Figure 6-11 Flow 2 Hydrogen Concentration at 3 Positions vs Height

6.5 Comparison of Concentration Gradients at Flow 1 and Flow 2

At position 1 and flow 2, the concentration of methanol increases with the distance

away from the catalyst, but remains fairly constant with the height above the catalyst for

flow 1 (Figure 6.14). Water shows the same trend in the subsequent graph. The steepness

of the concentration gradient indicates a thinner boundary layer and that mass transfer

influenced by both the bulk velocity and the diffusion process. Figure 6-16, of methanol

and water at Position 2, shows both flows showing almost the same changes in

concentration from one height to the next.
















0.032

0.028

0.024

0.02

0.016

0.012

0.008

0.004


CH30H, position, F2
I CH30H position, F2
V CH30H, position, F2



sI ^


4 5 6 7


Height (mm)

Figure 6-12 Flow 2 Methanol Concentration at 3 Positions vs. Height


* H20, position, F2
* H20, position, F2
* H20, position, F2







}


V
U


I



U
ii ii ii


I f

1 1 1 1 ,mI I I


1 2 3 4 5 6

Height (mm)


Figure 6-13 Flow 2 Water Concentration at 3 Positions vs Height


I~~~ I .









Both flows of methanol (blue and black markers) follow a similar trend, and the

higher velocity flow methanol has a greater over all concentration due the reduced

contact time, as compared to flow 1. Similarly, hydrogen for both flows follows the same

gradient in height above the catalyst, with the flow 2 hydrogen having a lower

concentration overall. At position 3 (Figure 6-18), the difference the two flow rates have

on reactant concentration is most apparent. At the lower velocity, methanol is about half

the concentration at the higher velocity.

6.6 Species Gradients along Reactor Length and Height above the Catalyst

The next section serves to illustrate how concentrations of products and reactants increase

or decrease along the reactor. In order to include all heights, error bars were omitted to

avoid obscuring data, and each species was shown separately for each flow. For hydrogen

at flow 1, the concentration reaches a maximum at Position 2, but generally decreases

with height above the catalyst. In the methanol flows, where the concentration is greatest

at the entrance, it is also least at the height 1 where it is being decomposed on at the

surface. In both flows, the methanol concentration sharply decreases along the reactor

length, as well as showing a more uniform concentration with height, near the exit.

Water shows a similar trend to methanol, for both flows, except at the entrance

where the concentration varies the most in the cross section of the flow. As the water is

consumed in the reaction, both flows go to a greatly reduced value, and become more

uniform in the cross section. Hydrogen increases along the reactor length from position 1

to 2 for both flows, and in Positions 2 and 3 appears to level off (in Flow 2), or decrease

(in Flow 1). Considering the much higher diffusion coefficient of hydrogen, as well as the

volume of dead space within the reactor, this phenomena is more likely due to the

circulation caused by the physical geometry of the reactor.















H2, position, F2
CH30H, position, F2
H2, position, F1
CH30H, position, F1
0.03


0.025 -


0.02


0.015


0.01


0.005


0


1.000 2.000 3.000


4.000 5.000 6.000 7.000


Height (nm)

Figure 6-14 Hydrogen and Methanol, Flows 1 and 2, Position 1


H2, position, F2
p H20, position, F2
SH2, position, F1
A H20, position, F1


0.03


0.025


0.02


0.015


0.01


0.005


0


1.000 2.000 3.000


4.000 5.000 6.000 7.000


Height (mm)

Figure 6-15 Hydrogen and Water, Flows 1 and 2, Position 1


I I


AT t





I I I
'''' ''''' ''' ''' 'a














0.012


0.011


0.01


0.009


0.008


0.007


0.006


0.005 L
1.000


* H2, position, F2
* CH30H, position, F2
* H2, position, F1
k CH30H, position, F1


I2.00 3I .0 4.I00 5.0I0 6 .0 7I .0
2.000 3.000 4.000 5.000 6.000 7.000


Height (mm)

Figure 6-16 Hydrogen and Methanol, Flows 1 and 2, Position 2


* H2, position, F2
I H20, position, F2
* H2, position, F1
A H20, position, F1


30 2.000 3.000


4.000 5.000 6.000 7.000

Height (mm)


Figure 6-17 Hydrogen and Water, Flows 1 and 2, Position 2


0.011


0.01


0.009


0.008


0.007


0.006


0.005
1.0(


~I












* H2, position, F2
* CH30H position, F2
* H2, position, F1
k CH30H, position, F1


0.009


0.008

0.007

0.006

0.005

0.004

0.003


0.002 '
1.000


2.000 3.000 4.000 5.000 6.000 7.000


Height (mm)

Figure 6-18 Hydrogen and Methanol, Flows 1 and 2, Position 3


* H2, position, F2
* H20, position, F2
* H2, position, F1
A H20, position, F1


0.008


0.006




0.004


0.002




0


1.000 2.000 3.000


4.000 5.000 6.000 7.000


Height (mm)
Figure 6-19 Hydrogen and Water, Flows 1 and 2, Position 3


T
-


4 1

I I I








78



Decomposition of hydrogen through a subsequent reverse reaction is unlikely because of


the low temperature. As mentioned earlier, if the reverse water-gas shift reaction were


occurring, it is expected that water would have increased at the end of the reactor,


however this was not observed.


0.012 I


0.008



0.006



0.004


0.002



0





Figure 6-20:


0.016


0.014


0.012


0
E 0.01
c)
0

0.008

8
0.006


0.004


0.002
C


).5 1 1.5 2 2.5 3 3.

Position

hydrogen Flow 1




height 1, methanol
height 2, methanol
height 3, methanol
height 4, methanol
o height 5 methanol







0
_V




N


.5 1 1.5 2

Position


2.5 3


Figure 6-21: Methanol Flow 1


* height 1, H2
* height 2, H2
I height 3, H2
* height 4, H2
o height 5, h2






0
Y


-












0.016


0.014


0.012


0.01


0.008


0.006


0.004


0.002


0
0.


500 1.000 1.500


Position


Figure 6-22 Water, Flow 1


0.009


0.008


0.007


0.006


0.005


0.004


0.003


0.002


0.001


2.000 2.500 3.000 3.5(


0.5 1 1.5


2 2.5 3 3.5


Position


Figure 6-23: Hydrogen Flow 2


SH20, height, F1
H20, height, F1
o H20, height, F1
* x H20, height, F1
+ H20, height, F1

+




*







+
n n n i .


* height 1, hydrogen
9 height 2, hydrogen
- height 3, hydrogen
* height 4, hydrogen
o height 5, hydrogen








80




Flow 2, Methanol


Height 1, methanol
Height 2, methanol
* height 3, methanol
* height 4, methanol
o height 5, methanol


I


0.5 1 1.5 2 2.5 3 3.5

Position


Figure 6-24: Methanol Flow 2


0.025




0.02



0.015




0.01




0.005



0


0.500 1.000 1.500 2.000 2.500 3.000 3.500

Position

Figure 6-25: Water, Flow 2


6.7 Species Concentration vs Position at Height 1


The following figures compare the changes in concentration of each species at the


two different flow rates, along axial position in the reactor.


0.03


0.025


0.02


0.015 h


0.01



0 nn0


x
A

0 H20, heights, F2
W H20, height, F2
+ o H20, height, F2
x H20, height, F2
+ H20, height, F2











*(


0.00










To further verify the validity of the measurements taken in this study, a mass

balance of hydrogen atoms was conducted, comparing the mole fraction of hydrogen

atoms at the inlet to the average of hydrogen mole fractions across the five heights at

Position 3. Bear in mind that the measurements taken above the catalyst surface do not

encompass the entire boundary layer, that hydrogen diffuses much faster than other

constituents, and that the signal over the catalyst is diluted by the dead space in the

optical regions of the reactor. As a result, at Flow 2, there was very good agreement, with

only a difference of 7%: 0.051 mole H/1 at the inlet and 0.0474 mole H/liter average at

Position 3. However, at Flow 1, which has a thicker boundary layer, there was a

difference of 46%.


0.012


0.01


0.008


0.006


0.004


0.002


0.500 1.000 1.500 2.000

Position
Figure 6-26: Hydrogen, Flow 1 and 2


2.500 3.000 3.500


* H2, height, F1
* H2, height, F2



I










0.02



0.015


0.005



0


0.500 1.000 1.500 2.000 2.500 3.000 3.500

Position
Figure 6-27: Water, Flow 1 and 2


0.016 -
CH30H, heights, F1
k CH30H, height, F2
0.014


0.012

0.01

0.008

0.006


0.004

0.002 I I '
0.500 1.000 1.500 2.000 2.500 3.000 3.500

Position
Figure 6-28: Methanol, Flow 1 and 2


S* H20, heights, F1
m H20, heights, F2


} I














CHAPTER 7
SUMMARY AND CONCLUSIONS

This study was undertaken to explore the feasibility of in situ measurements of

methanol reformation using Raman spectroscopy, with a goal of assessing concentration

boundary layers and the overall degree of reaction completion with a suitable residence

time-scale. To date, considerable research has been made of methanol reformation; such

studies generally utilize a control-volume approach, with only input reactants and output

products measured. Until now, other in situ studies have investigated the surface

chemistry of catalysis, specifically the adsorptive energies of species, and the interactions

of the catalyst microstructure during reaction. While much has been learned from these

control volume and surface science studies, they have limited use for validation of

detailed kinetic models of heterogeneous and gas-phase chemistry. This is especially true

considering findings in literature that the longer residence times, or reactor length for a

given flow rate, back mixing, and non-isothermality in packed bed reactors can lead to

reverse reactions and skewed apparent rate laws. The efficacy of monolith reactors has

been an effort for development in other studies, though they too are control volume

studies of the chemistry alone. This investigation is unique in that it seeks to quantify

some of the fluid dynamic aspects of the reactant flow through a monolith reactor, in

addition to studying the chemistry of reformation. Reported herein are some of the first in

situ Raman measurements of gas-phase species in a catalytic reactor.

Key results of the current study are the successful measurements of the reactants

methanol and water and the product hydrogen via Raman spectroscopy. Calibration was









used to convert measured spectra to absolute molar concentrations, which revealed a

significant conversion of methanol to hydrogen over the residence times (6 to 8 s) used in

this study. While the reactor design was not optimized for spatial resolution in

spectroscopic measurement, spatial gradients (millimeter-scale) were observed above the

catalyst surface under reforming conditions. The main findings of this study are as

follows:

1. The Katalco 33-5 (Johnson Matthey) catalyst pellets were an excellent low-
temperature (-200C) reforming catalyst. These pellets functioned well when
configured in a single-layer pellet bed for a laminar flow reactor. This
configuration enabled a two-dimensional assessment of the reforming process,
providing information about the boundary layer via height measurements, and
about the overall kinetics using the axial measurements.

2. Raman spectroscopy using 355-nm excitation was successfully implemented for the
in situ measurement of hydrogen, methanol, and water vapor during actual
methanol reformation. While Raman spectroscopy is a technique characterized
overall by relatively low sensitivity as compared to other methods, such as laser-
induced fluorescence, significant signal levels for the three targeted species were
recorded under all reforming conditions.

3. The optical access design described here proved successful for in situ Raman
analysis. The use of UV-grade fused silica optical windows successfully minimized
background fluorescence with 355-nm excitation. Since the current reactor was
based on low-cost, low-build-time design criteria, spatial resolution was not
optimized. As a result, Raman signals were diluted by a significant factor with
respect to the expected actual gradients above the catalyst surface. That is to say,
the scale of the crosses for the optical windows were such that there were
significant volumes of non-reacting dead space that interfered with the Raman
signal of the gas phase over the reactor. This was a major source of noise and
prevented further refinement of using Raman spectroscopy as an in situ diagnostic
in reforming. The fact that consistent Raman spectra were detected despite these
detractors in the reactor design only serves to emphasize the feasibility of Raman
spectroscopy for gas phase analysis.














CHAPTER 8
RECOMMENDATIONS FOR FUTURE WORK

This study identified design and implementation considerations for a monolith-type

reformer. Several improvements can be made in the design to improve resolution and

accuracy and, were these improvements to be implemented, opportunities would exist for

more accurate characterization the gas phase above catalysts, in conjunction with flow

rates and geometric scales.

1. Spatial resolution along the optical line of site is integral to accurately probing the
boundary layer, and this can be accomplished by eliminating the dead air space
between the optical windows and the catalyst bed. To do this, a custom design
would incorporate optical windows in the reactor housing wall, with a catalyst bed
the same width as the reactor. This would reduce the noise that occurs due to the
volume of the optical window crosses in the current design.

2. Steady flow of reactants would eliminate much of the variability seen at the
entrance of the reactor. The discrete flow of the peristaltic pumps resulted in
significant fluctuations at the entrance of the reactor. The flow variation was 1-2
seconds, depending on the flow rate setting, with the average flow rate over the
residence time being the assigned flow rates given as Flow 1 and Flow 2. The
variation in flow was caused by the rpm of the pump as the pump wheel pushes a
discrete amount of fluid through the tube.

3. Effects of unsteady reactant flows, catalyst pellets that are out of plane, and end
effects at the entrance and exit of the catalyst, though realistically occurring in
commercial reactors, would each have to be studied separately.

4. A more ideal, large scale monolith reactor would be to have two catalyst-coated
plates such that Pouisiuelle flow is created between them. This would be more
similar to the internal structure of an actual monolith. If the catalyst coated surfaces
within the reactor were interchangeable, then this configuration could be used to
characterize catalyst materials, as well as determine ideal separation of the surfaces,
as well as flow rates and temperatures, for desired rates of reaction. In this way,
scale parameters for monolith designs could be developed.

5. No reverse reactions were observed in this study, however, a greater range of
residence times and temperatures would have to be studied to determine an









optimum range of operation. Reverse reactions have been associated with higher
catalyst temperatures and longer residence times. The water-gas shift reaction is
active over the same catalyst, and if the steam-methanol reforming reaction occurs
via the methyl format intermediate, then the occurrence of CO would be due to
reverse water-gas shift reaction. Consequently, either additional steam would be
needed to convert the CO with another water-gas shift, or the products of the
reaction must be quickly removed from the catalyst surface to prevent the reverse
reaction in the first place. In this regard, mass diffusion limitations and residence
time contribute to reverse reactions. In the case of increased temperature, the higher
temperature product gases collide with surfaces with greater frequency, and as the
product partial pressures increase with the reaction, there is increased probability of
reverse reactions. It is important to bear in mind that adsorption and desorption of
species on the catalyst surface is a process that requires minimal energy, which is
why readsorption of product gases, and subsequent reverse reactions, occur more
easily than in homogeneous gas-phase chemistry. Water gas shift, reverse water gas
shift, methanol decomposition, and steam methanol reforming can all occur on this
type of catalyst, offering a variety of possible reaction pathways, so controlling
mass transfer to and from surface sites is important.

6. More accurate rate laws for catalytic surfaces can be determined with this design
configuration. The concentrations measured at each position are essentially
instantaneous molar ratios, and the positions along the reactor length can be
resolved temporally, knowing the velocity and volume of the reactor. With a well-
understood velocity profile, and with end conditions resolved, concentrations taken
along the axis of the reactor would describe the incremental change in
concentration per time. In situ studies in this manner would be able to assess more
accurately the onset of non-Arrhenius behavior in catalytic reactions.















APPENDIX
DATA SUMMARIES

This appendix contains the data summaries for hydrogen, water, and methanol

Raman spectra. The species concentration is given as the average, in moles per liter, as

determined from the raman spectral data and the calibration curves. The standard

deviations and relative standard deviations are given. Flow 1 and Flow 2 values are given

in Tables 4.1 and 4.2. Position 1 is located at x/L =0.09, Position 2 is at x/L= 0.5, and

Position 3 is at x/L=0.82. The distance above the catalyst surface, Height 1, is

approximately 1.25 mm; Height 2, 2.5 mm; Height 3, 3.8 mm; Height 4, 5.1 mm; Height

5, 6.35 mm.







88


Table A-1. Hydrogen Data Summary.
Hydrogen
(mol/I)
Flow Height Position Average StDev RSD
1 1 1 1.63E-03 5.46E-04 33.5
1 2 1 1.82E-03 5.79E-04 31.8
1 3 1 1.81E-03 3.76E-04 20.7
1 4 1 1.77E-03 5.57E-04 31.5
1 5 1 1.41E-03 4.63E-04 32.9
2 1 1 1.99E-03 4.38E-04 22.0
2 2 1 2.39E-03 6.98E-04 29.2
2 3 1 2.83E-03 3.12E-04 11.0
2 4 1 2.73E-03 5.84E-04 21.4
2 5 1 2.08E-03 5.52E-04 26.5
1 1 2 1.02E-02 1.57E-03 15.3
1 2 2 9.18E-03 1.43E-03 15.6
1 3 2 8.07E-03 1.51E-03 18.8
1 4 2 8.40E-03 8.87E-04 10.6
1 5 2 8.11E-03 1.78E-03 22.0
2 1 2 8.08E-03 9.06E-04 11.2
2 2 2 7.49E-03 1.45E-03 19.3
2 3 2 6.58E-03 9.71E-04 14.8
2 4 2 6.37E-03 9.50E-04 14.9
2 5 2 7.07E-03 1.22E-03 17.3
1 1 3 6.57E-03 2.47E-03 37.6
1 2 3 6.15E-03 9.93E-04 16.2
1 3 3 5.50E-03 1.21 E-03 21.9
1 4 3 4.35E-03 7.63E-04 17.5
1 5 3 3.85E-03 3.64E-04 9.5
2 1 3 7.68E-03 8.67E-04 11.3
2 2 3 7.71E-03 1.05E-03 13.7
2 3 3 7.43E-03 1.63E-03 22.0
2 4 3 7.42E-03 2.23E-03 30.1
2 5 3 5.89E-03 2.24E-03 38.0










Table A-2. Methanol Data Summary
Methanol
(mol/I)
Flow Height Position Average StDev RSD
1 1 1 0.009818 0.002835 28.87355
1 2 1 0.013955 0.00413 29.59285
1 3 1 0.012946 0.002312 17.86036
1 4 1 0.013809 0.002985 21.61706
1 5 1 0.009758 0.002878 29.49025
2 1 1 0.014981 0.001839 12.27357
2 2 1 0.019209 0.006799 35.39807
2 3 1 0.026781 0.004057 15.14849
2 4 1 0.027197 0.006082 22.36423
2 5 1 0.021595 0.003675 17.01606
1 1 2 0.00911 0.001845 20.25148
1 2 2 0.00823 0.001636 19.88287
1 3 2 0.006999 0.001972 28.17427
1 4 2 0.00782 0.001206 15.41797
1 5 2 0.00748 0.001894 25.31673
2 1 2 0.009397 0.001004 10.68457
2 2 2 0.009065 0.002121 23.39339
2 3 2 0.007907 0.001533 19.38787
2 4 2 0.007713 0.001625 21.06962
2 5 2 0.008147 0.001513 18.57486
1 1 3 0.003833 0.001063 27.74715
1 2 3 0.003758 0.000725 19.30516
1 3 3 0.003371 0.001001 29.69631
1 4 3 0.002925 0.000723 24.71729
1 5 3 0.002729 0.000498 18.2398
2 1 3 0.00728 0.0008 10.98548
2 2 3 0.007305 0.001157 15.83687
2 3 3 0.007099 0.001549 21.82156
2 4 3 0.006951 0.002003 28.80956
2 5 3 0.00554 0.001894 34.19093