Synthesis Gas and Olefins from the Catalytic Autothermal Reforming of Volatile and Non-volatile Liquids. Bradon Justin Dreyer

Transcription

1 Synthesis Gas and Olefins from the Catalytic Autothermal Reforming of Volatile and Non-volatile Liquids A DISSERTATION SUBMITTED TO THE FACULTY OF THE GRADUATE SCHOL OF THE UNIVERSITY OF MINNESOTA BY Bradon Justin Dreyer IN PARTIAL FULLFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY Lanny D. Schmidt, Adviser September 2007

2 Bradon Justin Dreyer 2007

3 Acknowledgements Many people have helped me reach this point in my professional career. I would like to sincerely thank Professor Lanny D. Schmidt for allowing me to conduct research in his group and providing guidance and direction in my research. His enthusiasm for research and his encouragement for me to pursue my own ideas made my research both very challenging and rewarding. Additionally, I would like to thank the Schmidt Group for teaching me various experimental techniques, consultation in experiments, academic challenges, and their friendship. I appreciated the guidance from Dr. Jakob Krummenacher, Dr. Ivan Lee, and Gregg Deluga for my initial training and indoctrination into the lab. For this thesis, I would like to thank Dr. James Salge for his assistance with the reactive flash volatilization of liquid fuels, Paul Dauenhauer for the his efforts and guidance with the reforming of solid fuels, Nick Degenstein and Dr. Raimund Horn for their guidance in the spatial profiles of benzene, solids reforming, and mass-spectrometry related analysis, and Brian Michael for the acquisition of the methane spatial profile. I also would like to thank Professor Ulrike Tschirner for her assistance in obtaining biomass samples for the solids reforming experiments, Professor Raul Caretta for his advice in experimental design and equipment selection, and Paul Dauenhauer, Dave Rennard, and Dr. Edward Wanat for their consultation with my thesis and dossier. Furthermore, I appreciated the many discussions with Anders Bitsch-Larsen regarding experimental design and non-research related topics. Finally, I would like to thank my parents, Ronald and Joan, brother Jason, and sister Carla for their financial and emotional support. My parents installed a strong work-ethic early in my life and encouraged me to pursue my interests in science and engineering. i

4 Abstract The research presented in this thesis develops an understanding of a clean energy process technology, catalytic partial oxidation (CPO). CPO is a process in which a carbon containing fuel, such as a hydrocarbon, is passed over a noble metal catalyst (e.g. rhodium and platinum) to efficiently generate synthesis gas (H 2 and CO) and olefins (e.g. ethylene and propylene) in millisecond contact times. Chapter 1 introduces CPO and compares this technology with conventional methods for synthesis gas and olefin production. CPO has several advantages over the traditional synthesis gas and olefin production methods. One advantage includes autothermal operation, requiring no external heat input from furnaces or heat exchangers. Autothermal operation allows these reactors to be built compactly. The short contact-times associated with CPO further enable for high throughput in relatively small reactor systems, and more compact reactors typically translate to faster response times if transient operation is required. Nobel metal based CPO catalysts are also resistant to deactivation, resulting in less catalyst replacement, regeneration, and maintenance, and an increase in operating efficiency. An overview of the many applications of the chemicals produced from CPO is also presented in Chapter 1. The chemicals produced are crucial in generating valuable chemical intermediates that are eventually incorporated in consumer products, medical devices, building structures, and fertilizers. Additionally, H 2 can be used as a source of energy in mobile fuel applications. Fuel cells convert H 2 and O 2 into electricity and water at higher efficiencies than thermal engine generators. Due to the difficulties in H 2 storage, these more efficient energy generators are dependent on hydrogen obtained from synthesis gas production in compact, portable fuel reformers, such as CPO reactors. Furthermore, H 2 and CO can be used in reducing environmentally harmful emissions. Particularly, the implementation of NOx traps and hydrogen into diesel engines has shown potential in reducing NOx emissions into the environment. Both concepts are dependent on synthesis gas generated from portable, compact fuel reformers, such as CPO reactors. Chapter 1 also reviews previous research in CPO, along with several important experimental parameters, and outlines the remaining research directions in the remaining chapters. In Chapter 2, steam addition to the CPO of higher hydrocarbons was explored over rhodiumcoated ceramic foam supports at millisecond contact times. Steam addition to the CPO of n- decane and n-hexadecane in air produced considerably higher H 2 and CO 2 and lower olefin and CO selectivities than traditional CPO. For steam to carbon feed ratios from 0.0 to 4.0, the ii

5 reactor operated autothermally, and the H 2 to CO product ratio increased from ~1.0 to ~4.0, which is essentially the equilibrium product composition near synthesis gas stoichiometry (C/O ~1) at contact times of ~7 milliseconds. In fuel-rich feeds exceeding the synthesis gas ratio (C/O > 1), steam addition suppressed olefins, promoted synthesis gas and water-gas shift products, and reduced catalyst surface carbon. Furthermore, steam addition to the CPO of the military fuel JP- 8 was performed successfully, also increasing H 2 and suppressing olefins. Chapter 3 investigates the interactions of different fuel components commonly found in logistic and naphtha feedstocks in CPO reactors. The CPO of benzene and n-hexane components and mixtures in air over rhodium- and platinum- catalysts supported on alumina foams at contact times of ~10 milliseconds is investigated. On both rhodium and platinum catalysts, pure benzene produced primarily synthesis gas and combustion products, while pure n- hexane produced olefins, synthesis gas, and combustion products. The partial oxidation of pure benzene yielded considerably higher operating temperatures than pure n-hexane. At carbon to oxygen ratios (C/O) > 1.0, the addition of benzene to n-hexane increased the selectivity of ethylene and propylene formed from n-hexane. When compared to pure n-hexane, the addition of benzene to n-hexane also increased the operating temperature and conversion of the n- hexane in the mixture. Additionally, rhodium catalysts produced higher synthesis gas selectivities and lower operating temperatures than platinum catalysts, while platinum catalysts produced more water, carbon dioxide, and olefins selectivities and higher operating temperatures. In the benzene and n hexane mixtures, the conversion of benzene was higher on rhodium catalysts than platinum catalysts. Two catalyst support ceramic foam monoliths, 80 and 45 pores per linear inch (PPI), were also investigated. The smaller pore-size, higher surface to volume ratio, 80 PPI supports produced higher selectivities of synthesis gas, while larger pore-size, lower surface to volume ratio, 45 PPI supports produced high selectivities of olefin and combustion products. In the mixtures, the conversion of benzene was greater on the smaller pore-size, higher surface to volume ratio supports. The CPO of benzene in air on rhodium-coated foam monoliths is examined in Chapter 4 using spatial temperature and species profiles at atmospheric pressure and near adiabatic conditions. Upstream of the catalyst region, some homogeneous oxidation occurs, generating primarily CO, CO 2, and H 2 O. Within the catalyst region, O 2 is reacted within the initial 4 mm, generating primarily H 2 and CO. From these species profiles, the species of CO 2 and H 2 O remain relatively constant throughout the catalyst region, while downstream of the catalyst, the products species remain unchanged. For comparison, the CPO of methane was also performed under identical conditions. Methane partial oxidation displayed no significant homogeneous oxidation upstream of the catalyst region, leaving the majority of the oxidation chemistry to occur iii

6 within the catalyst region. With methane, O 2 is completely reacted within the initial 2 mm, generating primarily H 2, CO, CO 2, and H 2 O. Downstream of this oxidation zone, H 2 O and unreacted methane is consumed to form more H 2 and CO, while CO 2 remains relatively constant. Similar to benzene partial oxidation, the species remain constant upon exiting the catalyst region as well. In Chapter 5, detailed homogeneous and heterogeneous mechanisms for benzene reforming are proposed and compared with the species profiles presented in Chapter 4. The model is an extension of a 38-step reaction mechanism developed for the CPO of methane on Rh. The additional steps were generated from previously published surface science experiments and firstprinciple calculations of benzene and acetylene adsorption, desorption, and decomposition on single crystal Rh surfaces. The kinetic heterogeneous model coupled with a previously developed homogeneous oxidation model appears to capture the general trends observed in the experimentally obtained profile data. The extension of CPO to non-volatile liquids is discussed in Chapter 6. Droplets of nonvolatile fuels such as soy oil and glucose-water solutions were flash evaporated by CPO to produce hydrogen in high yields (~70%) with a total time in the reactor of less than 50 milliseconds. Pyrolysis, coupled with catalytic oxidation of the fuels and their fragments upon impact with a hot Rh-Ce catalyst surface, avoided the formation of deactivating carbon layers on the catalyst. The catalytic reactions of these products generate approximately 1 MW/m 2, which maintained the catalyst surface above 800 o C at high drop impact rates. At these temperatures heavy fuels were catalytically transformed directly into H 2, CO, and other small molecules in very short contact times without accumulation of carbon. Following the demonstration of reforming non-volatile liquids, the work in Chapter 7 shows that the reactive flash-volatilization technique can be further extended to solids, such as biomass. Small particles directly contacting a hot catalytic surface maintained by heat generated from partial oxidation underwent rapid decomposition without detectable char production to form a tarfree synthesis gas stream at millisecond reaction times. Considered solid fuels included cellulose, starch, wood chips from Aspen (Populus tremuloides), and polyethylene, an example of common municipal waste. This unique catalytic method converted non-volatile biomass polymers to synthesis gas without an external heat source at least an order of magnitude faster than existing systems. In Chapter 8, the results in this thesis are summarized and future research directions are proposed. Possible future research directions include further developing mechanistic details iv

7 through spatially resolved measurements and catalyst characterization. Additionally, research should likely be conducted into the development of homogeneous and heterogeneous kinetic parameters through a combination of first-principle theoretical calculations and experimental transient and spatial measurements. v

8 Table of Contents Acknowledgements... i Abstract... ii Table of Contents... vi List of Tables...x List of Figures...x General Nomenclature...xii List of Definitions...xiii Chapter 1: Introduction...1 Clean energy conversion processes...1 Catalytic partial oxidation and competing reactions...2 Traditional synthesis gas and olefin production Hydrogen and synthesis gas Olefins...4 Economic, environmental, and size advantages of CPO...5 Uses of synthesis gas and olefins Olefins for commodity chemicals Hydrogen and synthesis gas for commodity chemicals Hydrogen for mobile energy Synthesis gas for environmental emission reduction...10 Previous work with CPO Feedstocks Experimental parameters...12 Thesis Outline Tuning the H 2 /CO ratio through steam addition Effect of aromatics and linear alkanes on CPO Routes to synthesis gas in the CPO of methane and benzene Extension of CPO to non-volatile liquid and solid fuels...14 References...15 Chapter 2: Autothermal steam reforming of higher hydrocarbons: n-decane, n- hexadecane, and JP Introduction Experimental Reactor and procedure Start-up and shutdown Catalyst Product analysis Quantification of JP vi

9 2.2.6 Equilibrium estimates Results n-decane n-hexadecane Comparing n-hexadecane and n-decane at C/O = 0.8 and C/O = JP Heat addition Carbon burn-off Discussion Heterogeneous and homogenous reactions Temperature effects Effects of carbon Linear alkanes vs. JP Summary References...36 Chapter 3: Synthesis gas and olefins from the catalytic partial oxidation of benzene and n- hexane mixtures on rhodium and platinum Introduction Experimental Reactor system Catalyst Product analysis Results Pure components on Rh, 80 PPI Mixtures on Rh, 80 PPI Pure components on Pt, 80 PPI Mixtures on Pt, 80 PPI Rh, 45 PPI support Pt, 45 PPI support Olefins from n-hexane Conversion to oxidation products Homogeneous chemistry Discussion Routes to synthesis gas & combustion products Routes to olefins Effect of catalyst support: surface area and mass transfer Synthesis gas generation: Pt vs Rh Temperature effects Benzene and n-hexane reactivity: chemisorption Benzene reactivity on Pt and Rh: decomposition pathways Coke formation Summary References...68 Chapter 4: Comparison of spatial profiles in the catalytic partial oxidation of benzene and methane on rhodium Introduction Experimental...80 vii

10 4.2.1 Catalyst preparation Reactor and capillary system: methane Reactor and capillary system: benzene Data analysis: methane profiles Data analysis: benzene profiles Reactor start-up and operation time Results Methane: species Methane: temperature Benzene: species Benzene: temperature Discussion Direct vs. indirect mechanism: methane and benzene Temperature differences Upstream heat-shield: axial diffusion Upstream heat-shield: homogeneous chemistry Mass transfer limitations Benzene and methane oxidation zones Conclusions References...93 Chapter 5: A kinetic model for the catalytic partial oxidation of benzene and methane on rhodium Introduction Model Development Heterogeneous mechanism Catalytic methane oxidation mechanism Extension of methane partial oxidation mechanism to benzene Benzene adsorption, desorption, and decomposition Acetylene adsorption, desorption, and decomposition Homogeneous mechanism Reactor models Energy balance Modeling various sections of the reactor Results Model trends Comparison with experimental data: methane Comparison with experimental data: benzene Discussion Experimental uncertainty Upstream heat-shield: axial dispersion Upstream heat-shield: homogeneous chemistry Comparing benzene and methane oxidation zones Secondary steam reforming Mass transfer limitations Catalytic surface area & pore size Thermodynamic inconsistencies Benzene & acetylene decomposition pathway Conclusions References viii

11 Chapter 6: Synthesis gas from nonvolatile fuels by reactive flash volatilization Introduction Experimental Reactor system Catalyst Reactor start-up and shutdown Product analysis Glucose-water operation Results Soy-bean oil Biodiesel Aqueous glucose solution Pt Discussion The flash volatilization process Pyrolysis pathways Catalytic reforming pathways Biodiesel vs. soybean oil Glucose and other biomass solids Other fuels Summary References Chapter 7: Millisecond reforming of solid non-volatile fuels, an extension of reactive flash volatilization Introduction Experimental Control of Solid Particle Flow Reactor Set-up Experimental feedstock Equilibrium calculations Results & discussion Cellulose as a function of C/O Temperature profile Particle size Polyethylene Aspen Non-noble metal catalysts: Ni Conclusion References Chapter 8: Thesis summary and future directions Thesis summary Future directions Spatial profiles Modeling References ix

12 List of Tables Table 5-1. Proposed mechanism for the partial oxidation of benzene, an extension of the Deutschmann mechanism for the partial oxidation of methane Table 7-1. Selected experimental data for the millisecond reforming of solid particles List of Figures Figure 1-1. Schematic of a traditional catalytic partial oxidation reactor...17 Figure 1-2. Photograph of the catalytic partial oxidation of methane on a rhodium-coated ceramic foam support...18 Figure 2-1. Schematic of the reactor system Figure 2-2. Schematic of experimental set-up Figure ppi α-alumina foam support (left) and rhodium coated 80 ppi α-alumina foam support (right)...39 Figure 2-4. Effect of n-decane /oxygen (C/O) and steam/n-decane (S/C) ratio on the fuel conversions (left) and the catalyst back-face temperatures (right) at 4 SLPM and 1.1 atm over an ~5 weight % Rh, ~4 weight % γ-alumina wash-coat, 80 ppi α-alumina foam monolith Figure 2-5. Effect of n-decane /oxygen (C/O) and steam/n-decane (S/C) ratio on the product selectivities at 4 SLPM and 1.1 atm over an ~5 weight % Rh, ~4 weight % γ-alumina washcoat, 80 ppi α-alumina foam monolith Figure 2-7. Effect of n-hexadecane /oxygen (C/O) and steam/n-hexadecane (S/C) ratio on the product selectivities at 4 SLPM and 1.1 atm over an ~5 weight % Rh, ~4 weight % γ-alumina wash-coat, 80 ppi α-alumina foam monolith Figure 2-9. Effect of JP-8 /oxygen (C/O) and steam/jp-8 (S/C) ratio on the fuel and O 2 conversions, the catalyst back-face temperatures (top), and product selectivities (middle, bottom) at 4 SLPM and 1.1 atm over an ~5 weight % Rh, ~4 weight % γ-alumina wash-coat, 80 ppi α-alumina foam monolith...45 Figure Effect of n-decane /oxygen (C/O) and external heat input on the fuel conversions, the catalyst back-face temperatures (top), and product selectivities (middle, bottom) at 4 SLPM and 1.1 atm over an ~5 weight % Rh, ~4 weight % γ-alumina wash-coat, 80 ppi α- alumina foam monolith Figure Temperature profile of carbon oxidation at the catalyst surface for JP-8 at 4.3 SLPM N 2, 0.4 SLPM O 2 and 1.1 atm...47 Figure 3-1. Schematic of reactor...71 Figure 3-2. Catalytic partial oxidation of n-hexane and of benzene on ~5 wt% Rh on ~5 wt% γ - alumina wash-coat, 80 PPI α-alumina foam support at 4 LPM (GHSV ~ 10 5 h 1 ) and 1.1 atm Figure 3-3. Catalytic partial oxidation of n-hexane and benzene mixtures on ~5 wt% Rh on ~5 wt% γ -alumina wash-coat, 80 PPI α-alumina foam support at 4 LPM (GHSV ~ 10 5 h 1 ) and 1.1 atm...73 Figure 3-4. Catalytic partial oxidation of n-hexane and of benzene on ~5 wt% Pt on ~5 wt% γ alumina wash-coat, 80 PPI α-alumina foam support at 4 LPM (GHSV ~ 10 5 h 1 ) and 1.1 atm Figure 3-5. Catalytic partial oxidation of n-hexane and benzene mixtures on ~5 wt% Pt on ~5 wt% γ -alumina washcoat, 80 PPI α-alumina foam support at 4 LPM (GHSV ~ 10 5 h 1 ) and 1.1 atm...75 Figure 3-6. Catalytic partial oxidation of n-hexane and n-hexane-benzene mixtures on ~5 wt% Rh on 45 PPI α-alumina foam support at 4 LPM (GHSV ~ 10 5 h 1 ) and 1.1 atm Figure 3-7. Catalytic partial oxidation of n-hexane and n-hexane-benzene mixtures on ~5 wt% Pt on 45 PPI α-alumina foam support at 4 LPM (GHSV ~ 10 5 h 1 ) and 1.1 atm...77 x

13 Figure 3-8. Heated n-hexane in air over an un-coated 80 PPI α-alumina foam support at 4 LPM (GHSV ~ 10 4 h 1 ) and 1.1 atm...78 Figure 4-1. Scanning electron microscopy (SEM) images of a foam cross section at various magnifications Figure 4-2. Schematic of the axial sampling system showing the interface with the reactor and vacuum system to collect spatially-resolved temperature and species data Figure 4-3. Micrograph of the optical fiber inside a fused silica capillary...97 Figure 4-4. Spatial profiles of the catalytic partial oxidation of methane (panels A & B) at C/O = 1.1, 2 LPM (GHSV ~ 5x104 h 1 ), and 1.1 atm...98 Figure 4-5. Spatial profiles of the catalytic partial oxidation of benzene (panels A & B) at C/O = 1.1, 2 LPM (GHSV ~ 5x104 h 1 ), and 1.1 atm...99 Figure 5-1. Pore from a ~5 wt% Rh on ~5 wt% γ alumina washcoat, 80 PPI α-alumina foam support Figure 5-2. Spatial temperature profiles of the catalytic partial oxidation of methane (panels A) and benzene (panel B) at C/O = 1.1, 2 LPM (GHSV ~ 5x10 4 h 1 ), and 1.1 atm Figure 5-3. Experimental and predicted spatial profiles of the catalytic partial oxidation of methane at C/O = 1.1, 2 LPM (GHSV ~ 5x10 4 h 1 ), and 1.1 atm Figure 5-4. Experimental and predicted spatial profiles of the catalytic partial oxidation of benzene at C/O = 1.1, 2 LPM (GHSV ~ 5x10 4 h 1 ), and 1.1 atm Figure 5-5. Predicted species coverage of the catalytic partial oxidation of methane (Panel A) and benzene (Panel B) at C/O = 1.1, 2 LPM (GHSV ~ 5x10 4 h 1 ), and 1.1 atm Figure 5-6. Predicted spatial profiles of the homogeneous partial oxidation of methane (Panel A) and benzene (Panel B) at C/O = 1.1, 2 LPM (GHSV ~ 5x10 4 h 1 ), and 1.1 atm Figure 5-7. Scanning Electron Microscropy (A) and Back-Scattered Electron (B) image of an ~5 wt% Rh on ~5 wt% γ alumina washcoat, 80 PPI α-alumina foam support Figure 6-1. Diagram and photograph of reactor autothermally reforming soybean oil Figure 6-2. Schematic of reactor Figure 6-3. Soy oil (A) conversion and temperature and (B) selectivities to H 2 and carbon products by reactive flash volatilization Figure 6-4. Biodiesel (A) conversion and (B) selectivities to H 2 and carbon products by reactive flash volatilization Figure 6-5. Effect of methane/glucose ratio (C Methane /C Glucose ) on (A) the fuel conversion (X), the catalyst temperature (T), and (B) product selectivities (S) at 1.1 atm over Rh-Ce catalysts Figure 6-6. Sketches of possible configurations in (A) conventional film boiling of a volatile drop on a hot surface, (B) reactive volatilization on a hot catalyst surface, (C) sequence of drop impingement and breakup on a hot catalytic surface, and (D) sketch of possible configuration on a porous ceramic foam Figure 7-1. Schematic of solid feed and reactor system Figure 7-2. Particle size distribution of the different solid feedstocks Figure 7-3. Top) Measured temperature at 10 mm, T 10 ( ), and 30 mm, T 30 ( ), from the top of the catalytic bed during the processing of ~230 μm particles of cellulose Figure 7-4. Gas temperature of cellulose ( ) processing at C/O of 0.7 ( ) and 0.9 ( ) with reaction diagram of volatile organic compounds (VOC) undergoing oxidation, steam reforming (SR: VOC + H 2 O H 2 + CO), water-gas-shift (WGS: H 2 O + CO H 2 + CO 2 ), and cracking reactions Figure 7-5. Comparison of size reduction energy requirements on basis of specific area increase [ 20] Figure 7-6. Photograph of an operating reactor of Rh-Ce coated spheres (LEFT). The exothermic chemistry near the top of the reactor heats the catalyst and support, emitting visible orange radiation Figure 7-7. Scanning electon microscopy (SEM) image of ash obtained from flash volatilization of solid aspen wood particles (Panel A) Figure 7-8. Energy dispersive x-ray spectroscopy (EDS, 15 kv accelerating voltage) of ash obtained from flash volatilization of solid aspen wood particles xi

14 General Nomenclature CPO C/O 0 ΔH PPI S/C SLPM catalytic partial oxidation carbon to oxygen ratio standard heat of reaction (298K and 1atm) pores per linear inch steam to carbon ratio standard liters per minute (298K and 1atm) xii

15 List of Definitions Catalyst Contact Time Catalyst contact times are calculated assuming that vaporized fuel, water, and gases behave ideally. The gas velocity at standard temperature and pressure (P = 1 atm, T = 298 K) is the following: Q 0 =, A v STP where Q is the volumetric flow rate at 298 K and 1 atm and A is the cross sectional area of the reactor tube. Applying the continuity equation, the velocity v 0 at inlet conditions of the foam is estimated to be the following: v 1 T P =, 0 1atm STP 0 v 0 T 298K P ε 0 where ε is the void fraction, T 0 is the reactant temperature entering the monolith and P 0 is the reactor pressure. The superficial contact time is defined as the nominal monolith length divided by the entering gas velocity: L τ =, v 0 where L is the nominal monolith length. Gas Hourly Space Velocity (GHSV) GHSV is defined as the following: GHSV Q ε V =, where Q is the volumetric flow rate of reactants at 25 o C and 1 atm, and ε and V are the void fraction and volume of the catalyst monolith, respectively. Conversion Conversion (X) is the fraction of a reactant species that is consumed by reaction: Fj,0 Fj X j =, F j,0 xiii

16 where X j is the conversion of reactant j, F j,0 is the initial molar flow rate of reactant j, and is the molar flow rate of reactant j exiting the reactor. Selectivity Product selectivity (S) is defined as the number of carbon or hydrogen atoms in product i divided by the number of carbon or hydrogen atoms in the converted reactants: n F S i =, n j reactants i i ( F F ) where F i is the molar flow rate of product i, n i is the number of carbon (or hydrogen) atoms in product i, and n j is the number of carbon (or hydrogen) atoms in reactant species j. indicates that all reactant species are summed. j,0 j reactants xiv

17 Chapter 1: Introduction Clean energy conversion processes Among the most enduring challenges for scientists and engineers is the development of clean energy conversion processes [1]. These processes require the ability to continually meet the growing demands for energy while minimizing negative economic and environmental impacts. A positive relationship is perceived between energy use and standard of living, and as a result, energy consumption is increasing to raise the standard of living, especially in developing nations where energy use is not as efficient as in the developed world. Additionally, uncertainty in energy prices has generated a great need for more economically stable energy sources and processing technologies. Natural gas and petroleum prices have been volatile in recent years. Fluctuations have been attributed to several controllable and uncontrollable factors, such as environmental, political, economic, technological, and physical instabilities involved in supplying the raw material for the continually growing energy demand. Increasing and unpredictable energy prices greatly affect the affordability of products and services that people rely upon to maintain their standard of living. At the same time, environmental considerations require that energy be generated and used with minimal impact on our surroundings. Most conventional processes to convert fossil fuels to energy involve air-blown combustion processes, which generate NOx, SOx, volatile organic compounds, CO 2, and other greenhouse gases into the environment. Several research areas are dedicated to developing clean energy conversion processes. These areas include the development of lower emission technologies, (more usable Btu/lb emission), higher efficiency processes (more usable Btu/Btu fuel), and the transition to renewable energy feedstocks (usable Btu without consuming non-renewable fossil fuel). The research presented in this thesis develops an understanding of a clean energy conversion process technology, catalytic partial oxidation (CPO). CPO is a process in which a carbon containing fuel, such as a hydrocarbon, is passed over a noble metal catalyst (e.g. rhodium and platinum) to efficiently generate synthesis gas (H 2 and CO) and olefins (e.g. ethylene and propylene) at millisecond contact times. CPO has several advantages over the traditional synthesis gas and olefin production methods. One advantage includes autothermal operation, requiring no external heat input from furnaces or heat exchangers. Autothermal operation allows these reactors to be built compactly. 1

18 The short contact-times associated with CPO further enable for high throughput in relatively small reactor systems, and more compact reactors typically translate to faster response times if transient operation is required. Nobel metal based CPO catalysts are also resistant to deactivation, resulting in less catalyst replacement, regeneration, and maintenance, and an increase in operating efficiency. The chemicals produced from the CPO reactors are important to maintaining the current standard of living as well reducing environmentally harmful emissions. These chemicals are crucial in generating valuable chemical intermediates that are eventually incorporated in consumer products, medical devices, building structures, and fertilizers. Additionally, H 2 can be used as a source of energy in mobile fuel applications. Fuel cells convert H 2 and O 2 into electricity and water at higher efficiencies than thermal engine generators. Due to the difficulties in H 2 storage, these more efficient energy generators are dependent on hydrogen obtained from synthesis gas production in compact, portable fuel reformers, such as CPO reactors. Furthermore, H 2 and CO can be used in reducing environmentally harmful emissions. Particularly, the implementation of NOx traps and hydrogen into diesel engines has shown potential in reducing NOx emissions into the environment. Both concepts are dependent on synthesis gas generated from portable, compact fuel reformers, such as CPO reactors. Catalytic partial oxidation and competing reactions Figure 1-1 displays a schematic of a typical catalytic partial oxidation reactor. Several global reactions can be used to generally describe product evolution within CPO reactors, such as the global partial oxidation and combustion reactions for a generic hydrocarbon of x carbon atoms and y hydrogen atoms: Partial Oxidation: x y C x Hy + O2 H2 + xco (1.1) 2 2 Combustion or Complete Oxidation: y y C xhy + x + O 2 H2O + xco2 (1.2) 4 2 2

19 These reactions are exothermic and, under adiabatic conditions, require no external heating to maintain reaction conditions. Figure 1-2 displays a photograph of the catalytic partial oxidation of methane on a rhodium-coated ceramic foam support. The exothermic oxidation reactors heat the catalytic support to >800 o C, as indicated by the emission of orange thermal radiation. At synthesis gas stoichiometry (carbon to oxygen ratio equal to 1.0 (C/O = 1.0)), the partial oxidation reaction dominates, generating high selectivities of H 2 and CO [2-4]. At stoichiometries exceeding synthesis gas stoichiometry, pyrolysis reactions of the unoxidized fuel are also observed [5], as shown for n-hexane pyrolyzing to propylene: Pyrolysis: CH CH H 2 (1.3) These bond breaking reactions are endothermic and generate olefins from hydrocarbon feedstocks. Pyrolysis without any external heating requires exothermic partial oxidation and combustion. Additionally, water produced from combustion or added to the feed can react with un-oxidized fuel on the catalyst, generating H 2 and CO through the steam reforming reaction [2,6-7]: Steam Reforming: y C x Hy + xh2o + x H2 + xco (1.4) 2 Steam reforming is endothermic and relies on the exothermic chemistry provided by the oxidation reactions. Similarly, CO 2 produced from combustion or added to the feed can react with unoxidized fuel on the catalyst, generating H 2 and CO through the CO 2 reforming reaction [2,7]: CO 2 Reforming: y C x Hy + xco2 H2 + 2xCO (1.5) 2 Furthermore, since combustion and synthesis gas products are simultaneously present in the partial oxidation system, the water-gas shift reaction is important [8]: 3

20 Water-gas Shift: CO + H +, 2O CO2 H2 kj Δ H 0 = - 41 (1.6) mol Water from combustion or the inlet feed stream reacts with CO generated from the partial oxidation of fuel to produce H 2 and CO 2. At temperatures exceeding 600 C, this reaction typically reaches equilibrium in millisecond residence times, and the maximum H 2 production is often limited by water-gas shift equilibrium [8]. Traditional synthesis gas and olefin production Hydrogen and synthesis gas Over 95% of H 2 production is supplied by steam reforming of light hydrocarbons [3]. The steam reforming reaction requires natural gas, predominately methane, and steam to pass over a supported nickel catalyst to produce H 2 and CO [10-11]: CH + +, 4 H2O CO 3H Ni 2 kj Δ H 0 = +207 (1.7) mol Similar to partial oxidation, steam and CO are simultaneously present during reaction conditions, and thus, the water-gas shift reaction is important. The water-gas shift reaction usually proceeds to equilibrium in industrial reactors. Steam reforming temperatures typically range from 780 to 880 C and pressures range from 1.4 to 3.8 MPa [10-11]. Furnaces are used to supply heat for endothermic steam reforming, and the reaction rates are very slow, typically residence times of a second, resulting in large reactors. Long start-ups are also incorporated with these large systems. Additionally, a steam to carbon (S/C) ratio of 2.5 to 4.0 is used to favor H 2 in water-gas shift equilibrium and reduce coke formation Olefins Thermal cracking of large hydrocarbons is the principal route for industrial production of ethylene and propylene [12-13]. Endothermic thermal cracking is accomplished in large tubular furnaces, commonly known as cracking furnaces, and the reaction rates are slow, typically 0.1 to 0.6 second residence times. The furnace tubes provide energy to perform thermal cracking, or pyrolysis, of large hydrocarbons. Cracking is typically performed under low pressure and high temperature conditions. A naphtha reactant stream enters the furnace at temperatures and 4

21 pressures ranging from 500 to 700 C and from 160 to 475 kpa, respectively, and the exiting product stream is maintained between 775 and 950 C and 150 to 275 kpa. Coke is also formed during pyrolysis, and to minimize this coke formation, steam is added to the feed. The steam minimizes the side reaction of coke formation and improves selectivity to the desired olefins by lowering the hydrocarbon partial pressure. Even with steam addition, furnaces are periodically shut down every 40 to 100 days to de-coke the tubular furnace. Long start-ups (time to reach steady-state operation) are also incorporated with these large systems. Economic, environmental, and size advantages of CPO CPO has several advantages over the previously mentioned traditional H 2 and olefin producing methods. The majority of CPO reactions are exothermic, providing sufficient energy to sustain reactions in an insulated reactor. Also, the heat generated from the reactor can supply sufficient energy to vaporize and preheat the reactants, creating a completely autothermal system. Without required heat input from furnaces or external heat exchangers, these reactors can be built compactly, further reducing capital investment. Large furnaces in conventional processes are limited by the rate of heat flux from the furnace into a tubular reactor. These heat transfer limitations translate to long residence times on the order of 0.1 and 0.6 seconds. CPO reactors operate at high temperatures, exceeding 600 C. Since exothermic oxidation and endothermic reforming chemistry occur in relatively close proximity, no heat transfer limitations exist between the exothermic and endothermic chemistry. These high reaction temperatures and lack of heat transfer limitations translate to high reaction rates and consequently high fuel and O 2 conversion in millisecond contact times. These shortcontact times allow for high throughput in relatively small reactor systems, which is ideal for portable applications. More compact reactors typically translate to faster response times if transient operation is required. Additionally, temperatures in the large industrial furnaces that drive the endothermic steam reforming and cracking reactions can reach 1500 to 2000 o C [12-13]. These temperatures are high enough to produce significant quantities of harmful pollutants such as NOx [14]. CPO reactors typically operate at temperature below 1200 o C, where NOx formation is not favored. 5

22 Furthermore, when compared to conventional nickel-based catalysts, the application of noble metal catalysts greatly reduces deactivation [15], which increases the operating efficiency of the reactor system by reducing catalyst replacement, regeneration, and maintenance. Uses of synthesis gas and olefins CPO appears to be an efficient technology to produce synthesis gas and olefins. These products are very important to generating energy and reducing pollution emissions, and they are also crucial in generating valuable chemical intermediates that are eventually incorporated in consumer products, medical devices, building structures, and fertilizers. Several important uses for synthesis gas and olefins are reviewed in the following sections Olefins for commodity chemicals Olefins, such as ethylene and propylene, are the largest-volume petrochemicals produced worldwide [12-13]. Over 100 million tons of ethylene are produced annually. However, olefins have no direct end uses, being used almost exclusively as a chemical building block. For example, more than 80% of the ethylene consumed in 1993 was used to produce ethylene oxide, ethylene dichloride, and polyethylene. Significant amounts were also used to make ethylbenzene, oligomer products (e.g., alcohols), acetaldehyde, and vinyl acetate. The chemical intermediates generated from olefinic building blocks are further reacted to generate more complex materials, such as textiles, plastics, adhesives, pharmaceuticals, rubbers, and coatings Hydrogen and synthesis gas for commodity chemicals In 1999, the global consumption of H 2 was approximately 42 million tons [10-11]. H 2 is primarily used as an intermediate in the chemical process industry. For example, the synthesis of ammonia, methanol, and waxes requires H 2. Approximately 61% of H 2 becomes feedstock for ammonia production, 23% is used in petroleum refining, and another 9% is used to manufacture methanol. Ammonia is commercially produced by passing H 2 and N 2 over a magnetite (Fe 3 O 4 ) fused with potassium oxide (K 2 O), alumina, and calcium oxide (CaO), at temperatures exceeding 450 C and elevated pressures between 100 and 300 bar [10-11,15]. 3H 2 N2 2NH Fe O K O,CaO + (1.8) 2 6

23 Similar to olefins, ammonia is a very important chemical intermediate used to generate many consumer products and construction materials. Also, ammonia is used as a fertilizer source for food crops, such as corn. Methanol is commercially produced when a high pressure gas mixture of CO and H 2 is passed across a catalyst containing copper, zinc oxide (ZnO), and alumina at temperatures between 250 and 300 C [10-11,15]. CO 2H CH3OH + (1.9) 2 Cu ZnO Al O 2 3 Methanol is a very important chemical intermediate and is used as a reactant along with plant or animal-derived fatty acids to form methyl esters (biodiesel). These methyl esters can be used in diesel engines, serving as a replacement to conventional petroleum-based diesel fuels. Furthermore, clean synthetic waxes can be produced from Fischer-Tropsch synthesis using synthesis gas as the reactant at temperatures near 330 C and elevated pressures near 25 atm over a ruthenium, copper, or iron catalyst [10-11,15]. ( CH ) nh O nco + 2nH (1.10) Ru,Cu,Fe n These waxes are further reformed to liquid combustible fuels free of sulfur or nitrogen containing compounds Hydrogen for mobile energy H 2 is also used as a source of energy in mobile fuel applications. Fuel cells convert H 2 and O 2 into electricity and water at higher efficiencies than thermal engine generators. The current leading technologies are polymer electrolyte membranes (PEMFC), used mainly in automobile applications, and the solid oxide fuel cell (SOFC), used mostly in large-scale stationary power generation with some mobile applications [16]. These efficient energy generators are dependent on hydrogen obtained from synthesis gas reformed in compact, portable fuel reformers, such as CPO reactors Polymer electrolyte membrane fuel cell (PEMFC) The PEMFC consists of an anode, a cathode, a catalyst (typically platinum) on a carbon support, and a polymer electrolyte membrane (e.g. Nafion ), which is a specially treated material 7

24 that selectively conducts protons and not electrons [16]. Pressurized H 2 gas enters the fuel cell on the anode side. The H 2 adsorbs to the platinum catalysts, and dissociates into two protons (H+) and two electrons (e-): + H2 2H + 2e (1.11) Pt The electrons are conducted from the anode, through the electrical circuit, and return to the cathode side of the fuel cell. The protons are conducted through the polymer membrane to the cathode side. On the cathode side, O 2 adsorbs to the platinum catalysts, and dissociates into two oxygen atoms. An oxygen atom along with the two protons transported through the polymer membrane combines with two electrons from the electric circuit to form a water molecule: + 1 2H + 2e + O2 H2O Pt 2 (1.12) This reaction in a single fuel cell ideally produces 1.23 open cell voltage. To increase the voltage in the system, many separate fuel cells are combined to form a fuel cell stack. The PEMFC operates at low temperatures, approximately 80 C. Consequently, low temperature operation does not require expensive containment structures, and a PEMFC can be heated from ambient to operating conditions in a relatively short time. Unfortunately, PEMFCs require that CO concentration be less than 10 PPM [16]. To decrease the CO concentration, additional catalytic water-gas shift and preferential oxidation reactors are combined with the conventional reforming process. Recent research has also demonstrated that water-gas shift can be successfully performed on transition metals catalysts with cerium on alumina supports, such as nickel-, rhodium-, platinum-, and ruthenium-cerium, in millisecond reactors [8]. Preferential oxidation (PROx) of CO to CO 2 is performed on the watergas shift product stream and typically incorporates a noble metal catalyst such as platinum, ruthenium, or rhodium on alumina: 1 CO + O 2 CO Au,Pt,Rh,Ru 2, 2 kj Δ H 0 = (1.13) mol Gold supported on metal oxides has also been shown to preferentially oxidize CO at temperatures below 100 o C [16]. Preferential oxidation is very sensitive to temperature and CO to O 2 ratio, which makes a PROx system difficult to control. 8

25 Other methods to reduce CO levels include palladium membrane purification and pressure swing adsorption [16]. Palladium membranes allow only H 2 to permeate and retain other gas components, such as N 2, CO 2, CO, and any trace impurities on the upstream side. Operating temperatures and pressures usually exceed 350 C and 20 bar to facilitate H 2 permeation Solid oxide fuel cell (SOFC) Another fuel cell technology is the SOFC. The design and operation is most suited to largescale stationary power generation; however, several automotive companies have been investigating this power generation technology in mobile applications [16]. SOFC have similar design to the PEMFC; however, the polymer electrolyte membrane is replaced with a hard ceramic material, typically zirconia (ZrO 2 ) doped with 8 to 10 mole percent yttria (Y 2 O 3 ). In the SOFC, oxygen ions (O -2 ) are conducted from the cathode to the anode through the ceramic material while electrons flow on the opposite direction through the external circuit. To enable oxidic electrolytes to become ions, SOFCs operate at temperatures exceeding 800 C. The anode contains nickel and yttria stabilized on a zirconia skeleton, and the cathode typically contains strontium (Sr) doped with lanthanum manganite (LaMnO 3 ) [3]. High operating temperatures with nickel also allow for internal fuel reforming; thus, CO and low molecular weight hydrocarbons, such as methane, can be used directly as fuels [16]. For example, the high temperatures allow the steam reforming of low molecular weight hydrocarbons, such as methane: CH 4 H2O CO + 3H 2 Y Ni 2O3 ZrO2 + (1.14) The H 2 produced from the reaction can then be oxidized in the anode by the following reaction: 2 H2 + O H2O + 2e Ni Y2O3 ZrO2 (1.15) CO produced can be oxidized in the anode by the following reaction: CO O CO + 2e Ni Y2O 3 ZrO2 (1.16) and O 2 is reduced at the cathode by the following reaction: 9

26 1 O e O Sr LaMnO3 + (1.17) The main advantage of a SOFC is that it can use CO as a fuel, whereas trace CO poisons a PEMFC. Low cost materials and low cost fabrication of ceramic structures are two technical challenges existing in commercializing this fuel cell [16] Synthesis gas for environmental emission reduction H 2 and CO can be used in reducing environmentally harmful emissions. Increasing usage of the internal combustion engine for transportation has elevated the emissions of environmentally harmful chemicals [1]. Several technologies have been developed over the years that rely on synthesis gas to reduce these emissions. Particularly, the implementation of NOx traps and hydrogen into diesel engines have shown potential in reducing NOx emissions. Both concepts require synthesis gas Catalytic converters and NOx traps Harmful emissions, including CO, NOx, hydrocarbons, and volatile organic compounds, are produced from gasoline and diesel engines. Vehicles with gasoline engines use a three-way catalytic converter to reduce these harmful emissions [15]. The converter uses two different catalysts, a reduction catalyst and an oxidation catalyst. Both types consist of a ceramic structure coated with a metal catalyst. The reduction catalyst uses platinum and rhodium to reduce NOx with CO or hydrocarbons in the exhaust stream: 2 CO + 2NO N + 2CO (1.18) Pt,Rh 2 2 y y y Cx Hy + 2x + NO x N2 + xco2 + H2O Pt,Rh + (1.19) In the exhaust steam, CO and hydrocarbons are also reacted with O 2 on the oxidation catalyst (platinum and/or palladium): 2CO + O 2 2CO 2 (1.20) Pt,Pd y y C x Hy + x + O 2 xco2 + H2O (1.21) Pt,Pd

27 Diesel engines have a higher thermal efficiency than gasoline engines [1]. Due to their increased efficiency, an incentive exists to operate diesel engines over gasoline engines. However, without emission controls, these engines produce unacceptably high levels of NOx and particulate matter, which will be severely restricted in the future. To help reduce the emission of particulate matter, diesel engines operate under fuel lean conditions. This fuel lean operation causes the engine to produce very small quantities of CO and unburned hydrocarbons, and elevated quantities of NO. With low concentrations of CO and hydrocarbons, low/no reducing agent is available to reduce NO with a three-way catalytic converter. To remove the NOx from the diesel exhaust gas, barium oxide NOx traps can be used [1]. NOx reacts with oxygen over Pt catalysts coated on barium oxide (BaO)/ α-alumina support to form NO 2. The NO 2 then reacts with the basic metal oxide (BaO) to generate barium nitrate (Ba(NO 3 ) 2 ) and nitrite (Ba(NO 2 ) 2 ). Periodically, the barium oxide is regenerated by passing synthesis gas over the Pt and barium nitrate or nitrite coated trap, releasing N 2, CO 2, and water. This synthesis gas is usually generated from the partial oxidation of some diesel fuel Hydrogen engines NOx abatement technology makes use of hydrogen. Experiments have shown that the addition of small amounts of H 2 to gasoline improves the overall performance of the gasoline internal combustion engine [17-19]. This improvement arises from many of the combustion properties associated with H 2. The addition of H 2 to the gasoline enables combustible mixtures at ratios below the lean flammability limit of pure gasoline/air mixtures. The lower temperatures and lean stoichiometry that result from this H 2 /gasoline/air mixture lead to less NOx formation and lower heat losses. Additionally, the increased burning velocity associated with H 2 -gasoline mixtures decreases flame-quenching distances and hydrocarbon emissions. Furthermore, mixture homogeneity and combustion efficiency are improved by the high molecular diffusivity of H 2 into air; when compared to standard operating conditions, a gasoline internal combustion engine that operates with 6% H 2 enrichment can have 15 to 20% reduced fuel consumption along with reduced NOx and CO emission. Previous work with CPO Feedstocks For the past 15 years, research in the Schmidt group has focused on CPO of several hydrocarbons, ranging from methane [4] to n-hexadecane [20], logistic fuel mixtures, such as gasoline [21] and diesel fuel [20], and oxygenates, such as ethanol [22] and glycerol [23]. The 11

28 reactors typically consisted of catalytic gauzes, ceramic foams, or spheres in a quartz tube surrounded by insulation as shown in Figure 1-2. These millisecond reactors have been used for CPO as well as hydrogen cyanide synthesis [24], ammonia oxidation [24], oxidative dehydrogenation [25], oxidative methane coupling [26], and catalytic combustion [27] Experimental parameters Several tunable parameters exist in the CPO process that control the performance in the reactor. The simplest variable to manipulate is the carbon to oxygen ratio (C/O) in the feed stream. The C/O ratio is defined as the ratio of moles of atomic carbon in the fuel to the moles of atomic oxygen in O 2. Using this definition, the stoichiometric ratio for partial oxidation is C/O = 1.0. The production of synthesis gas is maximized for C/O ~ 1.0. The production of combustion products increases as more O2 is added (C/O < 1.0), and the reactant mixture approaches combustion stoichiometry. When the reactant stream is more fuel rich than synthesis gas stoichiometry (C/O > 1.0), insufficient levels of oxygen exist to oxidize all of the fuel. The energy released from oxidation drives the thermal pyrolysis of un-oxidized fuel. For hydrocarbon feedstocks, olefins are often formed from the pyrolysis of un-oxidized fuel. Additionally, the catalyst metal is another variable in these CPO reactors. A variety of noble metal and transition metal catalysts, including rhodium, platinum, iridium, nickel, palladium, iron, cobalt, rhenium, and ruthenium, have been used in these reactors [28]. Catalysts containing rhodium have been shown to produce the highest selectivities of H 2 and CO [28-29]. For hydrocarbons, platinum based catalysts have the highest performance for lower olefin production, ethylene and propylene. Furthermore, bimetallic catalysts have also been shown to increase synthesis gas and olefin selectivities. Platinum-tin catalysts produced higher yields of ethylene from ethane than platinum catalysts [29], while rhodium-lanthanum and rhodium-cerium catalysts produced higher levels of synthesis gas than rhodium [23]. The catalyst support structure is also an important variable that greatly affects product distribution and fuel conversion. Ceramic foam monolith supports are typically an alumina or zirconia-based ceramic foam with a catalytic metal coated on the surface as a thin film [31-32]. The foam monoliths have been chosen primarily for their tortuous paths and small pores, which enable radially uniform heat transfer and species concentrations. Experiments have been performed on foam monoliths with a variety of pore diameters. Smaller pore diameters have been found to produce higher selectivity to synthesis gas and lower selectivity to pyrolysis products than supports with large pore diameters. Adding a γ-alumina wash-coat to the foam supports further decreases the pore size and enhances catalyst dispersion within the support. The addition of a wash-coat has been shown to increase synthesis gas and decrease pyrolysis 12

29 products. Ceramic spheres and metal gauzes have also been used as supports [33-34]. Gauzes maintain autothermal operation at microsecond contact times, yielding oxidation intermediates, such as oxygenates [33], while spheres maintain autothermal operation at very high flow rates, yielding synthesis gas and acetylene from methane [33]. Thesis Outline Tuning the H 2 /CO ratio through steam addition The compact size and high through-put of CPO is attractive for on-board fuel reforming of logistic fuels to synthesis gas. Synthesis gas can be used in pollution abatement technologies or efficient energy generation. The partial oxidation of linear alkanes typically generates maximum synthesis gas ratios of 2 moles H 2 to 1 mole CO. However, some applications require less CO and more H 2 than are currently produced by autothermal CPO. To increase H 2 and reduce CO concentrations, steam could be added to the inlet feed. The addition of steam potentially increases steam reforming and water-gas shift reactions to increase H 2 and decrease CO. Chapter 2 investigates the effect of steam addition on autothermal CPO of liquid alkanes in millisecond contact times, specifically the pure major components of liquid logistic fuels, n-decane and n-hexadecane, and the logistic fuel JP-8. These results can be used to assess the potential application of CPO technology in mobile fuel reforming of liquid logistic fuels Effect of aromatics and linear alkanes on CPO The results presented in Chapter 2 indicate that the products selectivities and fuel conversion from the autothermal reforming of linear alkanes are not reflective of the product selectivities and fuel conversion of actual logistic fuel mixtures. Logistic fuels are complicated mixtures of linear, branch, and cyclic alkanes, alkenes, and aromatic compounds. Each class of compounds may significantly affect the performance in the CPO reactor. Aromatics are a major component in logistic fuels and likely interact with linear alkanes in these CPO reactors. Chapter 3 investigates the autothermal reforming of aromatic benzene and the linear alkane n-hexane and their interactions within 20/80 and equimolar benzene and n-hexane mixtures Routes to synthesis gas in the CPO of methane and benzene The results presented in Chapters 2 and 3 incorporated integral species and temperature measurements. These measurements were used to assess the reforming ability of a variety of different fuels and mixtures. While these results provide insight into the fuel reactivities, the reaction pathways to the observed products cannot be conclusively determined. To develop 13

30 predictive kinetic models for the scale-up of this CPO technology, insight into species evolution within these short contact time reactors is required. Recently, Horn et. al developed a sampling technique that enables species profiles to be examined within 0.3 mm resolution in catalytic ceramic foams[35]. This technique has been demonstrated primarily for the CPO of methane and ethane fuels, and has not yet been applied to liquid feedstocks, such as benzene. In Chapter 4, spatially resolved ( 1 mm) measurements of composition and temperature profiles within ceramic foam catalysts are demonstrated for the CPO of liquid benzene. Insights into the species evolution within high temperature, short-contact time reactors are observed, which can be used to develop kinetic homogeneous and heterogeneous mechanisms of more complicated liquid feedstocks. In Chapter 5, detailed homogeneous and heterogeneous mechanisms for benzene reforming are proposed and compared with the species profiles presented in Chapter Extension of CPO to non-volatile liquid and solid fuels Production of synthesis gas from renewable feedstocks has become a focus of new research. Worldwide, renewable energy sources (including wood as well as other materials), account for about 19% of total energy usage [1]. In the United States, the corresponding value is roughly 11%, with biomass/solid waste and hydro power accounting for most of this total. Despite the potential of biomass gasification as a source of synthesis gas for the production of clean fuels, 98% of all energy from biomass is currently produced by direct combustion. A clear incentive and vast potential exists for the gasification of biomass to synthesis gas and finally to clean fuels (alkanes or alcohols) through Fischer Tropsch-type processes. Due to the compact nature of catalytic partial oxidation reactors, CPO technologies may provide a more economic, smaller scale route to synthesis gas from these biomass feedstocks. Also, CPO reforming routes have the potential to more effectively reform tars and other chemicals that are generated from traditional gasification technologies. These tars and chemicals negatively affect downstream fuel processing and elevate harmful environmental emissions. Extending CPO to processing biomass feedstocks is not trivial. To prevent catalyst deactivation and coke formation, CPO reactors traditionally required the fuel to be delivered to the catalyst in the vapor phase. Often times, this vaporization requirement restricted reforming to fuels that can be boiled. However, most biomass is non-volatile liquids and solids, which thermally decompose to carbon when heated by traditional vaporization methods. This carbon can quickly foul vaporizers and physically block catalytic surfaces, leading to deactivation. Chapters 6 and 7 reveal a technique to volatilize non-volatile fuels by using the energy generated 14

31 from the CPO reactor itself. This process will be explored on a variety of liquid and solid feedstocks to assess the applicability of this new technique. References 1. J.J. Spivey, Catalysis Today 100 (2005) A.P.E. York, T. Xiao, M.L.H. Green, Topics in Catalysis 22 (2003) A.T. Ashcroft, A. K. Cheetham, J. S. Ford, M. L. H. Green, C. P. Grey, A. J. Murrell, P. D. F. Vernon, Nature 344 (1990) D.A. Hickman, L. D. Schmidt, Science 259 (1993) A.G. Dietz III, A.F. Carlsson, L.D. Schmidt, Journal of Catalysis 176 (1998) E.J. Klein, S. Tummala, L.D. Schmidt, Studies in Surface Science and Catalysis 136 (2001) R. Horn, K.A. Williams, N.J. Degenstein, L.D. Schmidt, Journal of Catalysis 242 (2006) A.T. Ashcroft, A. K. Cheetham, M. L. H. Green, P. D. F. Vernon, Nature 352 (1991) Wheeler, A. Jhalani, E.J. Klein, S. Tummala, L.D. Schmidt, Journal of Catalysis 223 (2004) W.F. Baade, U.N. Parekh, V.S. Raman, Hydrogen, Kirk Othmer Encyclopedia of Chemical Technology, John Wiley and Sons, New York, P. Häussinger, R. Lohmüller, A.M. Watson, Hydrogen, Ullmann's Encyclopedia of Industrial Chemistry, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, K. M. Sundaram, M. M. Shreehan, E. F. Olszewski, Ethylene, Kirk-Othmer Encyclopedia of Chemical Technology, John Wiley and Sons, New York, H. Zimmermann, R. Walzl, Ethylene, Ullmann's Encyclopedia of Industrial Chemistry, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, S.R. Turns, An Introduction to Combustion, McGraw-Hill Publishing Co., New York, J.M. Thomas, W.J. Thomas, Principles and Practice of Heterogeneous Catalysis, Wiley VCH GmbH & Co. KGaA, Weinheim, Germany, G. Hoogers, ed., Fuel Cell Technology Handbook, CRC Press, New York, L.M. Das, International Journal of Hydrogen Energy 15 (1990) Y. Jamal, J. Wyszynski, International Journal of Hydrogen Energy 19 (1994) J.D. Naber, D.L. Siebers, International Journal of Hydrogen Energy 23 (1998) J.J. Krummenacher, K.N. West, L.D. Schmidt, Journal of Catalysis 215 (2003) R.P. O'Connor, E.J. Klein, L.D. Schmidt, Catalysis Letters 70 (2001) G.A. Deluga, J. R. Salge, L. D. Schmidt, X.E. Verykios, Science 303 (2004) P.J. Dauenhauer, J.R. Salge, L.D. Schmidt, Journal of Catalysis 244 (2006)

32 24. D.A. Hickman, M. Huff, L. D. Schmidt, Industrial and Engineering Chemistry Research 32 (1993) M. Huff, L. D. Schmidt, Journal of Physical Chemistry 97 (1993) K.L. Hohn, P.M. Witt, L. D. Schmidt, Catalysis Letters 54 (1998) C.T. Goralski Jr., L.D. Schmidt, Catalysis Letters 42 (1996) P.M. Torniainen, L. D. Schmidt, Journal of Catalysis 146 (1994) M. Huff, P. M. Torniainen, D. A. Hickman, L. D. Schmidt, Catalysis Today 21 (1994) A.S. Bodke, D. Olschki, L.D. Schmidt, E. Ranzi, Science 285 (1999) A.S. Bodke, S. Bharadwaj, L.D. Schmidt, Journal of Catalysis 179 (1998) N.J. Degenstein, R. Subramanian, L.D. Schmidt, Applied Catalysis A: General 305 (2006) K. L. Hohn, L. D. Schmidt, Applied Catalysis A: General 211 (2001) D.A. Goetsch, L. D. Schmidt, Science 271 (1996) R. Horn, N.J. Degenstein, K.A. Williams, L.D. Schmidt, Catalysis Letters 110 (3-4) (2006)

33 C x H y + O 2 ~10 ms Catalyst H 2 O + CO 2 C 2 H 4 + C 3 H 6 + H 2 +CO Figure 1-1. Schematic of a traditional catalytic partial oxidation reactor. 17

34 Figure 1-2. Photograph of the catalytic partial oxidation of methane on a rhodium-coated ceramic foam support. The exothermic oxidation reactions heat the catalytic support to >800 o C. The orange color is the thermal radiation emitted from the support. Photograph was taken by Scott Roberts and Paul Dauenhauer. 18

35 Chapter 2: Autothermal steam reforming of higher hydrocarbons: n-decane, n-hexadecane, and JP Introduction Reformation of heavy hydrocarbons and logistic fuels, such as diesel and JP-8 (similar to kerosene, used as a military fuel), to synthesis gas (H 2 and CO) is attractive for mobile applications, because synthesis gas can be used to generate electricity in solid oxide fuel cells and to abate pollutants in internal combustion engines [1-4]. Also, automotive onboard reforming of these liquid fuels is desirable, since it eliminates gaseous H 2 storage issues and avoids the need for H 2 refueling stations [1-3]. Higher hydrocarbons, such as isooctane, n-decane, n-hexadecane, and diesel fuel, have been successfully catalytically partially oxidized over Rh-coated monoliths, producing high yields of H 2 and CO (~80%) in autothermal millisecond reactors [6-10]. The catalytic partial oxidation (CPO) reactor is attractive for mobile applications, primarily due to its autothermal operation, high fuel and O 2 conversion, short contact times, fast light-off, and easily scalable design. Some applications require less CO and more H 2 than are currently produced by autothermal CPO. To increase H 2 and reduce CO concentrations, steam could be added to the inlet feed. Steam addition to the CPO of methane and ethanol over Rh and Rh-Ce coated monoliths in millisecond reactors has been successfully demonstrated [11-12]. The results propose that CO reacts with steam to form CO 2 and H 2 through the water-gas shift reaction kj CO + H2O CO2 + H2, Δ H 0 = - 41, (2-1) mol and the fuel also reacts with steam to form H 2 and CO through the steam reforming reaction, as shown for the hydrocarbon fuel n-decane, C + +, 10H22 10H2O 10CO 21H 2 kj Δ H 0 = (2-2) mol Chapter 2 is adapted from B.J. Dreyer, I.C. lee, J.J. Krummenacher, L.D. Schmidt, Autothermal steam reforming of higher hydrocarbons: n-decane, n-hexadecane, and JP-8, Applied Catalysis A: General 307 (2006) Elsevier, Inc. All rights reserved. 19

36 In addition, steam addition to the catalytic partial oxidation of diesel and gasoline fuel and their surrogates over various precious and transition metal catalysts at contact times typically greater than milliseconds was found to increase H 2 and CO 2 and reduce CO levels when compared to dry conditions [1, 3, 14-21]. These results further suggest that water-gas shift and steam reforming occur with the addition of steam to the reactor feed. Dry CPO could produce a maximum of 11 moles of H 2 per mole of n-decane, kj C 10H22 + 5O2 10CO+ 11H2, Δ H 0 = - 855, (2-3) mol while CPO with steam addition could perform water-gas shift on the CO to potentially increase the H 2 yield to a maximum of 21 moles per mole of n-decane C + +, 10H22 5O2+ 10H2O 10CO2 21H2 kj Δ H 0 = (2-4) mol Even though these reactions are exothermic, the addition of steam dilutes the reactants, lowering the adiabatic reactor temperature. With lower operating temperatures, the possibility of autothermal operation with steam addition in millisecond reactors is unclear. These species are also limited by equilibrium, and thus, the maximum level of H 2 produced is less than stoichiometrically indicated. This chapter investigates the effect of steam addition on autothermal CPO of liquid alkanes in millisecond contact times, specifically the pure major components of liquid logistic fuels, n-decane and n-hexadecane, and the logistic fuel JP-8. Also, olefin production, an undesirable competing reaction to partial oxidation, will be investigated. These results can be used to assess the potential application of this technology in mobile fuel-reforming of liquid logistic fuels. 2.2 Experimental Reactor and procedure The reactor consisted of a Y-shape 19mm inner diameter (ID) quartz tube as sketched in Figure 2-1, and a schematic of the experimental set-up is shown in Figure 2-2. The fuel, n- decane, n-hexadecane (Sigma-Aldrich, HPLC-grade 99+% purity), or United States Army East Coast JP-8 (Fort Belvoir Military Base), and steam were delivered into the reactor through two automotive fuel injectors (Delphi) from 5 psig pressurized liquid tanks. The injectors sprayed 20

37 liquid fuel on the heated reactor walls. The walls along the vaporization section were heated with Variac controlled resistive heating tape. High pressure cylinders of high purity (Airgas, %) gases fed the system with O 2 and N 2, and the flow rates were adjusted to air stoichiometry using Brooks 5850i mass flow controllers. Air entered the reactor through a small port located below the fuel injector. As the liquids evaporated on the walls of the reactor, the air convectively fed the steam and vaporized fuel toward the catalyst, preventing the reflux of reactants. An uncoated cylindrical foam monolith, 80 pores per linear inch (PPI), 17 mm outer diameter (OD) and 10 mm length α-alumina (Vesuvius Hi-Tech Ceramics, 92% α-al 2 O 3, 8% SiO 2 ), was wrapped in approximately 1 mm thick ceramic fiber paper (Fiberfrax) around the perimeter surface of the catalyst. This uncoated monolith provided a static, tortuous path for the uniform mixing of the fuel and air, and the paper prevented bypassing of gases. A thermocouple was placed below the static mixer to measure the temperature of the reactants upstream of the catalyst. Under steady-state operation, the upstream feed temperature was maintained at ~ 250 C. The upstream feed temperature of ~250 C is below the boiling point of n-hexadecane, 287 C, however greater than the dew point temperature of the resulting fuel mixture. Since the autoignition temperature of hexadecane is below its normal boiling point, the upstream temperature had to be maintained at ~250 C to prevent a cool flame from igniting. The heat radiated from the catalyst provided additional energy for vaporization, preventing the fuel from reaching the catalyst in the liquid phase. Also, the vaporization section was made appropriately long to achieve vapor liquid equilibrium prior to entering the mixer. From visual observations, no n-hexadecane condensed upstream of the catalyst or on the upstream radiation shield, verifying that n-hexadecane completely vaporized upstream of the catalyst. The catalyst support, an 80 PPI, 17 mm OD, and 10 mm length coated foam monolith, was placed between two uncoated 45 PPI α-alumina foam monoliths, which reduced axial radiation heat losses. To measure the back-face temperature of the catalyst, a thermocouple was placed between the bottom uncoated monolith and the catalyst. These three monoliths were wrapped around the perimeter surface with ceramic fiber paper to prevent bypassing of gasses and placed in the quartz reactor tube, below the static mixer and upstream thermocouple. The exterior of the quartz reactor containing the catalyst was wrapped with woven ceramic insulation to minimize radial radiation heat losses. The fuel injectors and mass flow controllers were controlled by LabView software. The software also recorded thermocouple readings at the catalyst back-face and in the pre-heat zone of the reactor tube. Product samples were taken from a port downstream of the reactor with a gas-tight syringe (Hamilton Samplelock, 50 microliter) and injected into a gas chromatograph 21

38 (HP 5890 Series II) for analysis. Typically, a sample was taken after continuously operating for 30 minutes at a specified inlet composition and flow rate. The continuous product stream was incinerated and vented to a fume hood. During operation, the reactor pressure was maintained near atmospheric conditions (1.1 atm), with no significant sources of pressure drop except through the monoliths. A total flow rate of 4 standard liters per minute (SLPM, 25 C and 1 atm) was also maintained. To keep the total flow rate constant, the air and fuel flow rates were reduced when steam was added to the reactor. This total flow rate corresponds to catalyst contact times of ~7 milliseconds assuming an entering reactant temperature of 800 C and pressure of 1.1 atm. The feed stream stoichiometry is reported as the steam to carbon ratio (S/C) and carbon to oxygen ratio (C/O). S/C is defined as the number of moles of water divided by the number of moles of carbon atoms in the mixture. C/O is defined as the number of moles of carbon atoms divided by the number of moles of oxygen atoms from O 2 in the mixture. By this definition, the stoichiometric feed composition for the partial oxidation reaction was C/O =1.0 for all hydrocarbon fuels. C/O, between 0.7 and 1.5, were combined with S/C, between 0.0 and 4.0. Steam rich conditions of S/C = 4.0 translate to 40 moles of water per mole of n-decane or 64 moles of water per mole of n-hexadecane (which ranges from ~53% to ~64% volume of the feed stream, depending on C/O). C/O and S/C where the fuel conversion was >65% were typically reported Start-up and shutdown For start-up, the reactor was heated with flowing N 2 until the catalyst back-face temperature reached ~250 C. The fuel was then admitted to the reactor through the injector, and O 2 was added to the N 2 at air stoichiometry. The catalyst typically ignited within ~15 seconds and reached the steady-state temperature within several minutes. Steam was then added while increasing the upstream heat input to maintain a constant feed temperature. The reactor was shutdown by first terminating the O 2, then the fuel and water flows. Occasionally, O 2 was reintroduced with N 2 over the hot catalyst to burn-off carbon deposited on the surface, as described later Catalyst The catalyst was Rh coated on a γ-alumina wash-coated, 80 PPI α-alumina foam monolith support. To prepare the catalyst, ~4 weight % (of monolith) γ alumina wash-coat (Alfa Aesar) was added to the α-alumina foam support, which covered the support in an ~30 to 50 μm thick 22

39 layer [2231]. The wash-coat and rhodium were applied using the incipient wetness technique, where a slurry of γ-alumina in distilled water was added drop-wise to both sides of the foam monolith and capillary forces distributed the γ-alumina throughout the monolith. The monolith was dried for 8 hours under ambient conditions and placed in a furnace at 600 C for 4 hours. Next, using the same incipient wetness technique, ~5% weight (of monolith) Rh was added to the wash-coated support by applying an aqueous solution of Rh(NO 3 ) 3 (Alfa Aesar, 13 wt% Rh in HNO 3 ) dropwise. The monolith was then dried and placed in a furnace at 600 C for 6 hours. Figure 2-3 shows a picture of an uncoated and a rhodium coated α-alumina foam support used in the experiments. Ten catalysts were used in these experiments, and each was operated for ~16 hours with several start-ups and shutdowns. No evident changes in catalyst activity were observed when the catalyst back-face temperature was maintained below 1100 C Product analysis Product gases for n-decane and JP-8 were analyzed using a single-column gas chromatograph equipped with a capillary column (J&W Scientific GASPRO, 60 m length, 0.32 mm ID) and thermal conductivity detector. Product gases for n-hexadecane were analyzed in a dual-column gas chromatography system, equipped with a capillary column (DB-5, 60 m length, 0.32 mm ID, 0.25 μm film thickness) and packed column (HAYESEP D 100/120, 9 m length, 2.2 mm ID) and thermal conductivity and flame ionization detectors, which is described elsewhere [7]. These systems were capable of detecting both permanent gases and higher hydrocarbons. In the reactor system, N 2 does not react at these relatively low reactor temperatures and can be considered an inert. Since N 2 was an inert species, it was the reference gas and incorporated into the calibration constant for mass balance calculations. The calibration constant or response factor for species j with respect to N 2, (R j ) is defined as the following: A N c 2 j R j =, (2-5) A j c N2 where A N 2 and A j are the measured peak areas for species N 2 and j respectively, and 23 c N 2 and c j are known concentrations of N 2 and j respectively in the injection sample. Response factors were determined by recording the peak areas of standard gas samples with known concentrations (Matheson Tri-Gas MICRO-MAT 14).

40 Species flow rates (F j ) were calculated from the integrated peak areas (A j ) in the resulting chromatogram using N 2 as an internal standard. The flow rates were determined with the following relationship: F j A j = R j FN. (2-6) 2 A N2 Since liquid compounds were not available in standard gas sample with N 2, desired liquids samples were mixed with a species of known response factor, such as pentane, and then injected into the gas chromatograph. This process enabled the response factors of unknown compounds to be referenced indirectly to N 2. Analyzing every product species from large fuel molecules, such as n-decane and n- hexadecane, became difficult at high C/O ratios [7]. Hundreds of potential olefins can be produced from n-hexadecane. Previous work was performed on analyzing the high molecular weight products of the partial oxidation of n-decane and n-hexadecane (>C 4 hydrocarbon backbone condensed from product stream) on a Gas Chromatograph with Mass Spectrometer (GC-MS). Greater than 99% of the high molecular weight products were identified as α-olefins, olefins with the double bond between the first and second carbon atoms. From n-decane and n- hexadecane partial oxidation product stream analysis, less than 0.1% internal olefins and less than 0.1% branched olefins of high molecular weight products were found in the product stream. Large alkanes were identified to be less than 0.5% of high molecular weight products and were identified as a small peak next to every α olefin peak in the chromatograph. The oxygen atom balance was closed to determine the water molar flow rate. To close the oxygen atom balance, the CO, CO 2, and O 2 were calculated, and then the water molar flow rate was calculated using the difference between the atomic oxygen in the specified feed conditions and in these measured compounds. Using this quantitative method, the carbon and hydrogen atom balances typically closed within +/- 7%. Conversion (X) is defined as the fraction of a reactant species that is consumed by reaction: X j Fj,0 Fj =, (2-7) F j,0 where F j,0 is the initial molar flow rate of reactant species j. 24

41 Product selectivity (S) is defined as the number of carbon or hydrogen atoms in product i divided by the number of carbon or hydrogen atoms in the converted reactants: n F S i =, (2-8) n j reactants i i ( F F ) j,0 j where F i is the molar flow rate of product i, n i is the number of carbon (or hydrogen) atoms in product i, and n j is the number of carbon (or hydrogen) atoms in reactant species j. indicates that all reactant species are summed. reactants Reaction products containing carbon are reported on a carbon atom basis using selectivities. H 2 is reported as selectivity on a hydrogen atom basis, using the moles of H 2 produced from the hydrocarbon reactants. For steam addition experiments, the hydrogen from water was not considered a reactant and not included when calculating the hydrogen atom selectivities. Thus, if H 2 is produced from the conversion of water, then the H 2 selectivity could exceed 100% Quantification of JP-8 When compared to single component fuels, analysis of a mixture has additional quantification challenges. JP-8 is a fuel mixture that typically contains 60% C 9 to C 16 linear and branched alkanes, 20% cyclic alkanes, 18 % aromatics, 2% alkenes, several ppm of fuel additives, 500 PPM of sulfur compounds, and has an average molecular formula of C 11 H 21 [23]. Quantifying all the possible GC responses for the components of this fuel mixture was not possible. Therefore, to estimate the conversion of JP-8, any molecule >C 8 was assumed to be JP-8 reactants, and the GC responses for these reactants were assumed to be identical to an abundant single component in JP-8, n-decane. Since some high molecular weight reactants in JP-8 may react to products in the C 9 to C 16 range, the above assumption provides a lower bound to the conversion, and the actual conversion may be higher. The average molecular formula was also used to calculate approximate C/O and S/C compositions of JP-8, steam, and air flowing through the reactor Equilibrium estimates Chemical equilibrium was estimated for various C/O and S/C factorial combinations of a pressure of 1.1 atm and the experimentally observed catalyst back-face temperature. These calculations are shown as dashed lines in Figure 2-8. The species included in the calculations were assumed to behave as an ideal gas mixture and to be species typically observed from gas 25

42 chromatograph analysis: N 2, O 2, H 2, CO, CO 2, H 2 O, CH 4, C 2 H 2, C 2 H 4, C 2 H 6, C 3 H 6, C 3 H 8, and the fuel. Additionally, a solid phase of carbon was also added to the calculations. To determine the equilibrium state, the CHEMKIN-III EQUIL subroutine was used, which iteratively minimized the Gibbs free energy subject to specified atomic constraints [24]. 2.3 Results n-decane Figure 2-4 displays n-decane conversion (left) and the catalyst back-face temperature (right). As the S/C increased, the catalyst back-face temperature decreased for all C/O. In addition, as the S/C increased, the n-decane conversion decreased when C/O > 0.8. For all S/C and C/O combinations, the O 2 conversion was >99%. Figure 2-5 shows the H atom selectivity for H 2 and also displays the C atom selectivities for CO, CO 2, and ethylene and propylene. H 2, H 2 O, CO, CO 2, and ethylene and propylene were the major products observed. Minor products included CH 4, C 2 H 6, C 3 H 8, and α-olefins ranging from 1-C 4 H 8 to 1-C 10 H 20. When C/O < 1.0, as the S/C increased, the H 2 and CO 2 selectivities increased, and the CO and ethylene and propylene selectivities decreased. The optimum H 2 was typically observed at C/O ~ 0.8. As the S/C increased from 0.0 to 4.0, the maximum H 2 selectivity increased from 82% to 132%, while the CO 2 selectivity increased from 7% to 60%, and the CO selectivity decreased from 89% to 38%. This change in selectivities translates to an increase in H 2 /CO from ~1 to ~4 when the S/C increased from 0.0 to 4.0. For C/O > 1.0, as the S/C increased, the selectivity of ethylene and propylene decreased. At C/O ~ 1.3, the maximum ethylene and propylene selectivity was typically observed. As the S/C increased from 0.0 to 2.0, the maximum ethylene and propylene selectivity decreased from 32% to 2%. An increase in synthesis gas and water-gas shift products were observed instead of olefins n-hexadecane Figure 2-6 displays n-hexadecane conversion (left) and the catalyst back-face temperature (right). Similar to n-decane, as the S/C increased, the catalyst back-face temperature decreased across all C/O. In addition, as the S/C increased, the n-hexadecane conversion decreased when C/O > 0.8. However, the decrease in n-hexadecane conversion was not as large as the decrease 26

43 in n-decane conversion. Furthermore, for all S/C and C/O combinations, the O 2 conversion was typically >99%. Figure 2-7 shows the H atom selectivity for H 2 and displays the C atom selectivities for CO, CO 2, and ethylene and propylene. H 2, H 2 O, CO, CO 2, and ethylene and propylene were the major products observed. Unlike n-decane, higher α-olefins ranging from 1-C 4 H 8 to 1-C 16 H 32 (not shown) were also significant at C/O > 1.0. Minor products included CH 4, C 2 H 6, and C 3 H 8. Similar to n-decane for C/O < 1.0, as the S/C increased, the H 2 and CO 2 selectivities increased, and the CO and ethylene and propylene selectivities decreased. At C/O ~ 0.8, the maximum H 2 was typically observed. As the S/C increased from 0.0 to 4.0, the maximum H 2 selectivity increased from 59% to 140%, while the CO 2 selectivity increased from 8% to 73%, and the CO selectivity decreased from 75% to 27% at the corresponding C/O. This change in selectivities translates to an increase in H 2 /CO from ~1 to ~5 when the S/C increased from 0.0 to 4.0. For C/O > 1.0, as the S/C increased, the selectivity of ethylene and propylene decreased. At C/O ~ 1.2, the maximum ethylene and propylene selectivity was typically observed. As the S/C increased from 0.0 to 2.0, the maximum ethylene and propylene selectivity decreased from 42% to 2%. When C/O >1.2, higher α-olefins, synthesis gas, and water-gas shift products were observed in place of ethylene and propylene Comparing n-hexadecane and n-decane at C/O = 0.8 and C/O = 1.2 Figures 2-8 (a) and (b) show the effect of increasing S/C at C/O = 0.8 for the two linear alkanes. The H atom selectivity for H 2 and C atom selectivities for CO and CO 2 are displayed for n-decane in (A) and for n-hexadecane in (B) with corresponding equilibrium values. The experimentally observed selectivities follow equilibrium selectivities for both fuels. Also, at C/O = 0.8, fuel and O 2 conversions exceed 95%, which are near equilibrium conversions of 100%. Figures 2-8 (C) and (D) display the effect of increasing S/C at C/O = 1.2. The C atom selectivities for ethylene, propylene, olefins smaller than C 6 H 12, and α-olefins larger than 1-C 5 H 10 are displayed in (C) for n-decane and in (D) for n-hexadecane. The figures show that selectivity of olefins less than C 6 H 12 decrease for both fuels as the S/C increases and that n-decane has a greater decrease in lower olefin selectivity than n-hexadecane. In addition, the higher α-olefin selectivity for n-hexadecane increased from S/C = 0.0 to 1.0 and then decreased from S/C = 1.0 to 2.0, while higher α-olefin selectivity decreased with increasing S/C for n-decane. 27

44 2.3.4 JP-8 Figure 2-9 (top) displays JP-8 conversion, O 2 conversion, and the catalyst back-face temperature. >99% O 2 conversion was observed for S/C = 1.0. However, in contrast to the n- decane and n-hexadecane experiments, some O 2 breakthrough was observed for S/C = 0.0 and C/O > 1.2. In addition, as the S/C increased, the catalyst back-face temperature decreased for all C/O. However, the average temperatures were ~100 C higher than n-hexadecane and n- decane at all C/O and S/C. Furthermore, as the S/C increased, the JP-8 conversion decreased for C/O > 1.3 and S/C = 1.0. Figure 2-9 shows the H atom selectivity for H 2 (middle) and displays the C atom selectivities for CO, CO 2, and ethylene and propylene (bottom). H 2, H 2 O, CO, CO 2, and ethylene and propylene were the major products observed. Minor products included CH 4, C 2 H 6, C 3 H 8, and olefins ranging from C 4 H 8 to C 7 H 14. As the S/C increased at C/O < 1.0, the H 2 and CO 2 selectivities increased, and CO and ethylene and propylene selectivities decreased. At C/O ~ 0.9, the highest H 2 selectivity was observed. As the S/C increased from 0.0 to 1.0, the H 2 selectivity increased from 84% to 90%, while the CO 2 selectivity increased from 9% to 15%, and the CO selectivity decreased from 85% to 81% at the corresponding C/O. However, the increase in H 2 and CO 2 selectivities and corresponding decrease in CO selectivity was not as large as for n-decane and n-hexadecane. At C/O > 1.0, as the S/C increased, the selectivity of ethylene and propylene decreased. At C/O = 1.3, as the S/C increased from 0.0 to 1.0, the ethylene and propylene selectivity decreased from 34% to 15%. An increase in synthesis gas and water-gas shift products was observed instead of olefins. Also, when steam was added, the suppression of olefins for JP-8 was not as large as the olefin suppression for n-decane and n-hexadecane at corresponding C/O and S/C Heat addition Experimental observations for n-decane, n-hexadecane, and JP-8 suggest that, when steam is added, temperature is positively correlated with fuel conversion and olefin production. To investigate this observation further, heat was supplied to the catalyst and heat shields for n- decane. This heat increased the reactor operating temperature, and the products were analyzed and compared to insulated conditions. In the experiment, three C/O feed ratios (1.1, 1.2, 1.3), where maximum lower olefins are typically observed, were investigated at S/C =1.0 and at an upstream stream temperature of ~250 C. The heat was supplied by a furnace and was 28

45 electronically controlled to maintain catalyst back-face temperatures ~200 C greater than insulated conditions. Figure 2-10 (top) displays the catalyst back-face temperatures and n-decane conversions. At S/C = 1.0, the addition of heat increased the conversion of n-decane at corresponding C/O. Thus, fuel conversion appears positively correlated with temperature when steam is added. Figure 2-10 also shows the H atom selectivity for H 2 (middle) and displays the C atom selectivity for CO, CO 2, and ethylene and propylene (bottom). As heat was added, the maximum H 2 and CO 2 selectivities decreased from 121% to 111% and from 30% to 15%, respectively, and this occurred at C/O = 1.1. The CO selectivity increased from 67% to 81%. Furthermore, as heat was added at S/C = 1.0, the selectivity of ethylene and propylene remained essentially constant, ~3% at C/O = 1.2. Therefore, from these observations, when steam is added, increasing the reactor operating temperature does not increase olefin selectivities Carbon burn-off The carbon burn-off experiment was designed to provide a method to qualitatively relate the catalyst back-face temperature profile to the steady-state surface carbon level and indicate how the surface carbon level changes with steam addition. In this experiment, a mixture of O 2 and N 2 was passed over a hot catalyst that had been operated for 30 minutes at a specified inlet composition. O 2 reacted with the carbon on the surface to form CO, CO 2, and H 2 O. The heat released from the oxidation increased the temperature of the monolith in the insulated system, and this temperature qualitatively indicates the reactive steady-state carbon level on the catalyst surface. The amounts of carbon deposited with steam addition and without steam addition were qualitatively compared. The catalyst was operated at C/O = 1.0, S/C = 1.0, and total flow rate of 5.26 SLPM for 30 minutes. The balance was N 2 to make an air stoichiometric mixture. After 30 minutes, the fuel and water were simultaneously stopped, and the N 2 and O 2 flow rates were changed to 4.30 and 0.40 SLPM respectively. A control system recorded the back-face temperature. This process was repeated for C/O = 1.0, S/C = 0.0, and a total flow rate of 5.26 SLPM. Fuel and O 2 flow rates remained the same as the previous trial with S/C = 1.0, and the balance of the flow was N 2. Thus, the fuel, O 2, and total flow rates remained constant for both trials. 29

46 Figure 2-11 displays the carbon burn-off temperature profiles for the fuel JP-8 at C/O = 1.0. The temperature profile for operation at S/C = 0.0 is greater than the profile for operation at S/C = 1.0. The larger temperature profile suggests that more carbon was deposited on the catalyst surface during the 30 minute steady state operation at S/C = 0.0 than at S/C = 1.0 and, therefore it indicates that steam addition to a CPO reactor suppresses steady-state surface carbon. Product analysis was taken ~5 seconds after initiating the burn-off conditions. The ratio of CO to CO 2 was ~ 9:1 for both burn-offs with no H 2 produced. 2.4 Discussion The results demonstrate that steam addition to CPO increases H 2 and CO 2 and decreases CO selectivities at C/O < 1.0, strongly suppresses olefins at C/O > 1.0, and reduces surface carbon. In this discussion, the mechanism will be proposed for the addition of steam in the CPO reactor. Additionally, even when excessive amounts of steam are added to the CPO reactor, reactions do not extinguish. However, steam addition does reduce temperature and fuel conversion at high C/O. The relationship between steam addition, temperature, and fuel conversion will be explained. Furthermore, the logistic fuel JP-8 also qualitatively display similar trends to the higher alkanes, however, less H 2 and more olefins are observed. Possible causes for this observation will also be discussed Heterogeneous and homogenous reactions The steam addition experiments primarily display the production of H 2, CO, CO 2, which are primarily produced from catalytic partial oxidation and combustion, steam reforming, and watergas shift reactions. However, these reactions occur simultaneously in the millisecond reactor and are difficult to decouple from product analysis, especially when compositions close to equilibrium are observed. In a general description of the process, catalytic partial oxidation and combustion are thought to dominate initially in the reactor until all the O 2 is consumed typically in the first millimeter of the catalyst [3, 11, 14, 16]. H 2 and CO are mostly produced with some H 2 O and CO 2, and the temperature typically is 100 to 200 C higher than the measured back-face temperature. 30

47 With the absence of O 2, the remaining unreacted and partially reacted fuel is then catalytically steam reformed to CO and H 2, and steam catalytically reacts with CO to produce CO 2 and H 2 [3, 11, 14, 16]. Endothermic steam reforming is assumed to dominate downstream of the oxidation zone, increasing the level of H 2 and CO and reducing the temperature to approximately the observed back-face temperature. Water-gas shift is believed to occur along side the steam reforming reactions downstream of the oxidation zone, decreasing CO and increasing H 2 and CO 2 levels. When steam is added, catalytic methanation also occurs simultaneously with water-gas shift and fuel steam reforming [14, 16]. However, methanation does not appear to be a dominant reaction, because less than 4% selectivity to methane was produced in these experiments when S/C > 0.1. Olefins are assumed to be produced mostly through homogenous cracking reactions that occur downstream of the surface reactions, typically in the last ~9 millimeters of the monolith where O 2 is not present [7]. High temperatures of the exothermic catalytic combustion and partial oxidation reactions provide energy for homolysis of the alkane fuel to produce two alkyl radicals. These radicals decompose by β scission to yield ethylene and a smaller radical or by β hydrogen elimination to yield a larger stable α-olefin and a hydrogen radical [7, 25]. Radical decomposition continues with each radical eliminating ethylene molecules until an ethyl or propyl radical remain. This residual ethyl or propyl radical then performs β hydrogen elimination to form ethylene or propylene. In addition, lower operating temperatures appear to favor β-hydrogen elimination to large α-olefins over complete β-scission to ethylene and propylene. From the experimental results at C/O between 1.1 and 1.5, steam addition reduces these homogenous and increase heterogeneous product selectivities. Previously, CPO with steam addition of isooctane and other liquid hydrocarbons has shown that a spectrum of hydrocarbons is produced and then reformed [14, 18, 26]. Perhaps, olefins are produced throughout the reactor and then steam reformed to H 2 and CO. Elevated steam levels may increase the heterogeneous reaction rates of steam reforming and water-gas shift and enable the heterogeneous reactions to dominate over the homogenous reactions of olefin production Temperature effects To sustain a total flow rate of 4 SLPM, the flow rates of fuel and O 2 were decreased as steam was added. Consequently, the fuel and O 2 generated less heat, lowering the reactor temperature. Steam addition up to ~64% volume of the feed also dilutes the reactants with a fluid of high 31

48 sensible heat, further reducing the adiabatic temperature. Lower operating temperatures typically reduce both heterogeneous and homogeneous reaction rates. Slower homogenous reaction rates usually result in incomplete cracking to higher α-olefins in millisecond reactors, as observed for n-hexadecane, and a decrease in fuel conversion occurs from the reduced homogeneous and heterogeneous reaction rates. In addition, in fuel rich feeds, C/O > 1.0, where insufficient levels of O 2 are available to completely perform exothermic oxidation reactions, endothermic cracking and steam reforming reactions dominate. The increase in endothermic and decrease in exothermic chemistry generate less heat under insulated conditions and, consequently, lower operating temperatures further, resulting in a greater decrease in fuel conversion. Thus, fuel rich feeds experience a larger decrease in conversion than the synthesis gas region, where exothermic chemistry dominates. Still, very steam rich conditions of S/C = 4.0 (which translate to 40 moles of water per mole of n-decane or 64 moles of water per mole of n-hexadecane) generated sufficient levels of heat under insulated conditions for autothermal operation. As noted previously, the n-hexadecane conversion and α-olefins selectivity were greater than n-decane. The higher conversion and α-olefin selectivity of n-hexadecane indicate that n- hexadecane is reacting by incompletely cracking to higher α-olefins as the S/C increases and reactor temperature decreases. As the S/C increases, n-decane does not crack under the ensuing lower temperatures, and, as a result, its conversion decreases more significantly than n- hexadecane with no observed increase in α-olefins. Adding heat to the catalyst and heat shields (Figure 2-10) increased the operating temperature and provided faster kinetics for the endothermic reactions. Apparently, under fuel rich conditions, the higher temperatures increased the steam reforming reaction rate, resulting in the complete steam reforming of the fuel and water-gas shift to CO, CO 2, and H 2. In the reactor system, homogenous reaction rates were less than the steam reforming rates at higher temperatures and steam addition. Thus, any homogenous olefins products were likely steam reformed in the reactor, and olefin selectivities did not increase when steam and heat were added. Furthermore, as heat is added to S/C=1.0 (Figure 2-10), the observed decrease H 2 and CO 2 and increase in CO selectivities was expected since the water gas shift-reaction is exothermic, and the operation was near equilibrium. Adding heat to the exothermic water-gas shift reaction at equilibrium reduces the products, H 2 and CO 2, and increases reactants, H 2 O and CO. At C/O=1.2, the ~200 C increase in back-face temperature corresponds to ~12% shifts in CO and 32

49 CO 2 and ~4% shift in H 2 equilibrium selectivities, while the experimental shifts in selectivity were ~12% for CO and CO 2 and ~5% for H Effects of carbon Even though carbon burn-off temperature profiles indicate the presence of carbon on the catalyst surface, the carbon deposited on the catalyst does not accumulate and deactivate the catalyst over time. During steady state operation, the rate of carbon deposition is most likely balanced by carbon gasification or oxidation. Also, the carbon burn-off temperature profiles qualitatively suggest that less steady-state surface carbon is formed when steam is added to the reactor system. Steam could reduce surface carbon through the carbon gasification reaction, C + ( S) + H2O CO H2, kj H 0 = mol Δ. (2-9) At the reaction temperatures, sufficient energy is provided by the partial oxidation and combustion reactions to promote the endothermic gasification reaction. When surface carbon is reduced, more catalytic sites are available for the fuel, O 2, and steam to react to synthesis gas and water-gas shift products. Therefore, reducing surface carbon promotes heterogeneous steam reforming, partial oxidation, and water-gas shift reactions. Increasing the heterogeneous product selectivities, H 2, CO, and CO 2, decreases the competing homogeneous reaction product selectivities, such as olefins. Experiments where less surface carbon was qualitatively identified showed a shift from homogeneous olefin to heterogeneous H 2, CO, and CO 2 product selectivities with steam addition, supporting this theory Linear alkanes vs. JP-8 As mentioned previously, the logistic fuel JP-8 was more resistant to water-gas shift than the linear alkanes of n-decane and n-hexadecane. At C/O < 1.0 and S/C = 0.0, JP-8 appears to produce synthesis gas at selectivities similar to n-decane and n-hexadecane. As water was added to S/C = 1.0, synthesis gas was still being produced from JP-8; however, CO was not reacting with excess water to form CO 2 and H 2 at selectivities similar to n-decane and n- hexadecane. JP-8 also displayed higher back-face temperatures than n-decane and n- hexadecane at corresponding C/O and S/C, possibly indicating that more exothermic or less 33

50 endothermic chemistry was occurring. Furthermore, when compared to the linear alkanes, steam addition to JP-8 suppressed less olefins. Many complexities arise from interpreting differences between the autothermal steam reforming of a logistic fuel mixture and pure component surrogates. First, the stoichiometry (C/O ratio) is not precisely described for a mixture with varying carbon atoms and variable C/H ratios. In this study, 11 carbon atoms and 21 hydrogen atoms was assumed to determine the C/O and S/C ratios. JP-8 is well characterized, and thus using this average molecular formula is most likely an appropriate approximation. However, variable compositions are contained within these fuels. Since H 2 have maxima for C/O near 1, this maximum could be broadened and lowered or shifted with a fuel mixture because some components are above and below this optimum stoichiometry for H 2 production. Second, the reactivities of different molecules may be quite different. Linear, branched, and cyclic alkanes should not react identically. Aromatic and polyaromatic molecules should have different reactivities than aliphatics. These reactivities can vary greatly in homogeneous and heterogeneous pyrolysis and oxidation reactions within the reactor. Homogeneous pyrolysis rates are typically related to C-C and C-H bond strengths since breaking the bonds to form radicals is the initiation step in these reactions. These reactivities are probably correlated with autoignition temperatures, which vary by as much as 100 o C between fuel components in this mixture. Homogeneous oxidation reactions have a different order of reactivities than pyrolysis reactions. These reactions depend on collisions with O 2, OH, O, O 2 H, and related alkyl species. In contrast to pyrolysis reactions, homogeneous gas phase oxidation reactions do not correlate with bond strength, however, they are more sensitive to molecular geometry and electronic structure. The electronic structure and molecular geometry vary greatly among the fuel components. Even less is known regarding the relative surface reactivities of these components and their variation with size and structure. Adsorption, decomposition, and oxidation are probably quite different on Rh than on the alumina support. Rh should have a much greater capability to adsorb oxygen, which promotes surface oxidation channels with adsorbed hydrocarbons. These channels produce synthesis gas and combustion products that readily desorb at reaction temperatures, leaving the Rh sites active. On alumina, the lower capability to adsorb oxygen may lead to mainly pyrolysis decomposition pathways. These pathways are expected to have a greater tendency to form coke, yielding these sites inactive. Also, the compounds electronic structure and molecular geometry are likely important on the catalyst surface. Compounds that contain π-bonds or oxygen atoms adsorb at higher probabilities than compounds that contain only 34

51 sigma bonds. Also, large or branched compounds may be sterically hindered to adsorb on a surface. Furthermore, a fundamental possible complication with partial oxidation of fuel mixtures is that reactivities are not simply an average over all reactant molecules. The most reactive molecule consumes the oxygen quickly, leaving the pyrolysis and potential steam reforming as the available pathways for the less reactive fuel components. If pyrolysis pathway dominates on the catalyst surface, surface sites may become predominately covered with carbon, preventing steam reforming. With these difficulties being stated, a possible explanation for the observed differences between autothermal reforming of JP-8 and single component hydrocarbons is that the oxidation of the most reactive hydrocarbon species occur, generating synthesis gas and combustion products. However, less reactive compounds in the JP-8 mixture, such as aromatics, sulfur, or fuel additive compounds, are selectively adsorbing onto the catalyst and depositing more carbon or sulfur on the catalyst surface [17, 20-21, 27-28]. The carbon or sulfur coated surface prevents water from adsorbing and reacting with CO and the fuel and promotes homogenous olefin production. A reduction in steam reforming pathways with the fuel and fragments would explain higher operating temperatures and increased olefin selectivities. With less water adsorbing on the surface, water-gas shift would be also reduced, yielding higher CO and lower H Summary n-decane, n-hexadecane, and JP-8 with air over Rh-coated monoliths were successfully partially oxidized to synthesis gas in the presence of large amounts of steam. H 2 and CO 2 levels were enhanced and CO levels were reduced with the addition of steam to the reactor feed stream, which indicates that water-gas shift is achieved in these millisecond reactors at high temperatures. Because equilibrium product compositions were observed, detailed reaction pathways are difficult to decouple in the CPO reactor system. Steam addition strongly suppressed olefins (from >30% to <3% ethylene and propylene selectivities for n-decane) and promoted synthesis gas and water-gas shift products in fuel-rich feeds exceeding the synthesis gas ratio. Steam addition also appears to reduce the level of carbon deposited on the catalytic surface. The reduction of carbon possibly promotes heterogeneous chemistry by exposing more reactive surface and, thus, further reduces homogenous products. Also, dilution from steam addition may reduce observed olefin products by reducing reaction temperature and homogenous olefin producing reaction rates. Further 35

52 experiments are required to determine the dilution effect of steam addition in this reactor system and to investigate in detail the role that surface carbon plays in the suppression of olefins and promotion of heterogeneous products. The catalytic partial oxidation of JP-8 with steam addition displayed similar trends to linear alkanes. However, less olefins were suppressed with steam addition, and higher back-face temperatures were observed. Also, when compared to the linear alkanes, JP-8 was more resistant to water-gas shift when steam was added. Excess surface carbon deposited by components in JP-8 is suspected to be the cause for these differences when compared to linear alkanes. Although, additional studies are required to quantify these differences. The logistic fuel mixture that was used in these experiments contained sulfur, aromatics and olefin compounds. These components could cause problems in higher concentrations. Thus, the results presented in this chapter do not indicate that these mentioned compounds can be reformed at high concentrations. 2.6 References 1. M. Krumpelt, T.R. Krause, J.D. Carter, J.P. Kopasz, S. Ahmed, Catalysis Today 77 (2002) G. Hoogers, ed., Fuel Cell Technology Handbook, CRC Press, New York, D-J. Liu, T.D. Kaun, H-K. Liao, S. Ahmed, International Journal of Hydrogen Energy 29 (2004) K. Liu, W.G. Wnuck, W.P. Leenhouts, US Patent Application No A1 (2005). 5. C. Tan, J.G. Weissman, J.V. Bonadies, J.E. Kirwan, US Patent Application No A1 (2005). 6. R.P. O Connor, E.J. Klein, L.D. Schmidt, Catalysis Letters 70 (2001) J.J. Krummenacher, K.N. West, L.D. Schmidt, Journal of Catalysis 215 (2003) L.D. Schmidt, E.J. Klein, C.A. LeClerc, J.J. Krummenacher, K.N. West, Chemical Engineering Science 58 (2003) J.J. Krummenacher, L.D. Schmidt, Journal of Catalysis 222 (2004) R. Subramanian, G. J. Panuccio, J. J. Krummenacher, I. C. Lee, L. D. Schmidt, Chemical Engineering Science 59 (2004) E.J. Klein, S. Tummala, L.D. Schmidt, Natural Gas Conversion VI, Studies in Surface Science and Catalysis 136 (2001) G.A. Deluga, J.R. Salge,L.D. Schmidt, X.E.Verykios, Science 303 (2004)

53 13. C. Wheeler, A. Jhalani, E.J. Klein, S. Tummala, L.D. Schmidt, Journal of Catalysis 223 (2004) M. Flytzani-Stephanopoulos, G.E Voecks, International Journal of Hydrogen Energy 8 (1983) D.J. Moon, K. Sreekumar, S.D. Lee, B.G. Lee, H.S. Kim, Applied Catalysis A: General 215 (2001) S. Springmann, G. Friedrich, M. Himmen, M.Sommer, G. Eigenberger, Applied Catalysis A: General 235 (2002) J.P. Kopasz, L.E. Miller, D.V. Applegate, SAE SP 1790 (2003) M. Pacheco, J. Sira, J.P. Kopasz, Applied Catalysis A: General 250 (2003) A. Tsolakis, A. Megaritis, International Journal of Hydrogen Energy 29 (2004) R.L. Borup, M.A. Inbody, T.A. Semelsberger, J.I. Tafoya, D.R. Guidry, Catalysis Today 99 (2005) J.P. Kopasz, D. Applegate, L. Miller, H.K. Liao, S. Ahmed, International Journal of Hydrogen Energy 30 (2005) A. Bodke, S. Bharadwaj, L.D. Schmidt, Journal of Catalysis 179 (1998) T. Edwards, L. Q. Maurice, Journal of Propulsion Power 17(2) (2001) R.J. Kee, et al., Chemkin Collection Release 3.7.1: EQUIL Application User Manual, Reaction Design, San Diego, CA, E. Ranzi, T. Faravellia, P. Gaffuri, E. Garavaglia, A. Goldaniga, Industrial Engineering and Chemistry Research 36 (1997) J. Kopasz, L.E. Miller, S. Ahmed, P. Devlin, M. Pacheco, 2002 Future Car Congress, Arlington, Virginia, June 3-5, A. Shamsi, J.P. Baltrus, J.J. Spivey, Applied Catalysis A: General 293 (2005), D. Shekhawat, T.H. Garner, D.A. Berry, M. Salazar, D.J. Haynes, J.J. Spivey, Applied Catalysis A: General 311 (2006) 8. 37

54 Fuel Injectors Fuel Water Air Heating Tape Static Mixer Catalyst Thermocouple Insulation Radiation Shields Sample Port Products Thermocouple Figure 2-1. Schematic of the reactor system. Hydrocarbon fuel and water are introduced into the reactor through low-flow automotive fuel injectors, vaporized, and mixed with air. The heatshields prevent thermal radiation heat losses in the axial direction, and the tube is wrapped in insulation to reduce radial heat losses. 38

55 Pressurized Water Tank Pressurized Fuel Tank Incinerator Fuel Injectors Mass Flow Controllers Reactor HP 5890 GC LabView N 2 O 2 Figure 2-2. Schematic of experimental set-up. N 2 and O 2 mass flow controllers and automotive fuel injectors are controlled by LabView software on the computer. The software also records back-face and upstream thermocouple readings. Product gases are sampled downstream from the center of the tube and injected into an HP 5890 GC for separation and analysis. All gases that are not sampled are incinerated. Figure ppi α-alumina foam support (left) and rhodium coated 80 ppi α-alumina foam support (right) 39

56 Conversion (%) S/C=4.0 n- C 10 H 22 S/C=2.0 Equilibrium S/C=0.0 S/C=0.1 S/C=0.5 S/C= C/O Temperature ( C) S/C=4.0 S/C=0.0 S/C=0.1 S/C=1.0 S/C=0.5 S/C= C/O Figure 2-4. Effect of n-decane /oxygen (C/O) and steam/n-decane (S/C) ratio on the fuel conversions (left) and the catalyst back-face temperatures (right) at 4 SLPM and 1.1 atm over an ~5 weight % Rh, ~4 weight % γ-alumina wash-coat, 80 ppi α-alumina foam monolith. 40

57 S/C=4.0 S/C=2.0 S/C= S/C=0.1 S/C=0.5 H Selectivity (%) S/C=0.5 S/C=0.0 S/C=0.1 C Selectivity (%) H S/C=1.0 S/C= S/C=4.0 S/C= C/O C/O CO C Selectivity (%) S/C=4.0 S/C=2.0 S/C=0.5 S/C=0.1 CO 2 S/C=1.0 S/C= C/O C Selectivity (%) C 2 H 4 & C 3 H 6 S/C=0.0 S/C=0.1 S/C=4.0 S/C=1.0 S/C=0.5 S/C= C/O Figure 2-5. Effect of n-decane /oxygen (C/O) and steam/n-decane (S/C) ratio on the product selectivities at 4 SLPM and 1.1 atm over an ~5 weight % Rh, ~4 weight % γ-alumina wash-coat, 80 ppi α-alumina foam monolith. 41

58 Conversion (%) S/C=4.0 S/C=0.5 Equilibrium S/C=0.1 S/C=0.0 S/C=1.0 S/C= C/O Temperature ( C) n- C 16 H S/C=0.0 S/C=0.5 S/C=4.0 S/C=0.1 S/C=1.0 S/C= C/O Figure 2-6. Effect of n-hexadecane /oxygen (C/O) and steam/n-hexadecane (C/O) ratio on the fuel conversions (left) and the catalyst back-face temperatures (right) at 4 SLPM and 1.1 atm over an ~5 weight % Rh, ~4 weight % γ-alumina wash-coat, 80 ppi α-alumina foam monolith. 42

59 H Selectivity (%) C Selectivity (%) S/C=4.0 S/C=0.5 S/C=0.1 S/C=0.0 S/C=2.0 S/C=1.0 H C/O S/C=2.0 S/C=1.0 S/C=0.1 S/C=4.0 S/C=0.5 S/C=0.0 CO C/O C Selectivity (%) C Selectivity (%) S/C=0.1 S/C=0.5 S/C=2. 0 S/C=4.0 S/C=1.0 S/C=0.0 CO C/O C 2 H 4 & C 3 H 6 S/C=4.0 S/C=0.0 S/C=0.1 S/C=0.5 S/C=1.0 S/C= C/O Figure 2-7. Effect of n-hexadecane /oxygen (C/O) and steam/n-hexadecane (S/C) ratio on the product selectivities at 4 SLPM and 1.1 atm over an ~5 weight % Rh, ~4 weight % γ-alumina wash-coat, 80 ppi α-alumina foam monolith. 43

60 A H or C Selectivity (%) C/O=0.8 H 2 CO CO 2 Equilibrium S/C H or C Selectivity (%) B C/O=0.8 H 2 CO CO 2 Equilibrium S/C C 50 C/O=1.2 D 50 C/O = 1.2 C Selectivity (%) Total Olefins Olefins <C 6 C 2 H 4 C 3 H 6 Olefins >C 5 C Selectivity (%) Olefins <C 6 Total Olefins C 2 H 4 Olefins >C5 C 3 H S/C S/C Figure 2-8. Effect of steam/fuel (S/C) ratio for n-decane and n-hexadecane at 4 SLPM and 1.1 atm over an ~5 weight % Rh, ~4 weight % γ-alumina wash-coat, 80 ppi α-alumina foam monolith. H 2, CO, and CO 2 selectivities with corresponding equilibrium values at C/O = 0.8 are displayed in A for n-decane and in B for n-hexadecane. Olefin product selectivities at C/O = 1.2 are displayed in C for n-decane and in D for n-hexadecane. Equilibrium values are calculated from experimentally observed back-face temperatures and 1.1 atm. 44

61 100 O 2, S/C= JP-8, S/C=0.0 Conversion (%) O 2, S/C=0.0 S/C=0.0 JP-8, S/C=1.0 S/C= Temperature ( C) C/O 800 H or C Selectivity (%) CO, S/C=1.0 H 2, S/C=1.0 CO, S/C=0.0 H 2, S/C= C/O 50 C Selectivity (%) C 2 H 4 & C 3 H 6, S/C=0.0 CO 2, S/C=1.0 C 2 H 4 & C 3 H 6, S/C=1.0 CO 2, S/C= C/O Figure 2-9. Effect of JP-8 /oxygen (C/O) and steam/jp-8 (S/C) ratio on the fuel and O 2 conversions, the catalyst back-face temperatures (top), and product selectivities (middle, bottom) at 4 SLPM and 1.1 atm over an ~5 weight % Rh, ~4 weight % γ-alumina wash-coat, 80 ppi α- alumina foam monolith. 45

62 Conversion (%) S/C=1.0 n- C 10 H 22 w/heat n- C 10 H 22 INSULATED w/heat INSULATED C/O Temperature ( C) H or C Selectivity (%) H 2 INSULATED H 2 w/heat CO w/heat CO INSULATED S/C= C/O C Selectivity (%) S/C=1.0 CO 2 INSULATED CO 2 w/heat C 2 H 4 & C 3 H 6 INSULATED C 2 H 4 & C 3 H 6 w/heat C/O Figure Effect of n-decane /oxygen (C/O) and external heat input on the fuel conversions, the catalyst back-face temperatures (top), and product selectivities (middle, bottom) at 4 SLPM and 1.1 atm over an ~5 weight % Rh, ~4 weight % γ-alumina wash-coat, 80 ppi α-alumina foam monolith. 46

63 S/C = 0.0 C/O=1.0 Temperature ( C) S/C = Time (s) Figure Temperature profile of carbon oxidation at the catalyst surface for JP-8 at 4.3 SLPM N 2, 0.4 SLPM O 2 and 1.1 atm. Prior to oxidation (t<0), reactor was operating at steady state for 30 minutes at C/O = 1.0 and a total flow rate of 5.26 SLPM. Carbon burn-offs were performed over ~5 weight % Rh. ~4 weight % γ-alumina wash-coat 80 ppi α-alumina foam monoliths. 47

64 Chapter 3: Synthesis gas and olefins from the catalytic partial oxidation of benzene and n-hexane mixtures on rhodium and platinum 3.1 Introduction Olefins, such as ethylene and propylene, are the largest volume intermediates produced in the petrochemical industry [1-4]. Olefins are produced from endothermic steam cracking of ethane or naphtha feeds (hydrocarbon mixtures containing many compounds) that typically is carried out in large tube furnaces with a residence of 1 s. This long residence time and external heat source translate to large reactor units of high capital costs. Alternatively, high yields of olefins from hydrocarbons (>70%) can be produced autothermally from the catalytic partial oxidation (CPO) of hydrocarbons at millisecond contact times [4-9]. In these reactors, O 2 is combined with the hydrocarbon feed, and some of the feed is partially oxidized to synthesis gas (H 2 and CO), releasing heat. This heat provides thermal energy for the endothermic cracking of the remaining hydrocarbons to olefins. The millisecond contact times and absence of external heat input typically translate to more compact, lower capital cost reactors, which is an economic benefit over conventional steam cracking reactors. However, predicting performance from these catalytic reactors requires knowledge of the chemical species involved in the reaction. The reactor performance from a complex composition of naphtha feeds cannot be easily modeled or described by a single representative component [9]. The catalytic reforming capabilities of naphtha-like fuel mixtures to synthesis gas (H 2 and CO) and olefins in CPO reactors have been shown to differ from single component aliphatic surrogates [7, 10-11]. The mixtures displayed higher operating temperatures than single component aliphatic compounds at corresponding inlet stoichiometries. Furthermore, when steam was added to the CPO reactor (autothermal steam reforming), the mixtures were more resistant to water-gas shift, producing less H 2, CO 2, olefins, and more CO than the reforming of the aliphatic surrogates without the addition of steam [11-14]. Naphtha streams often contain high levels aromatics (>20%) [3]. The high concentration level of these aromatics has been suspected to cause differences in observed operability in catalytic partial oxidation and autothermal steam reforming systems with a mainly aliphatic feedstock. This chapter investigates the effect of an aromatic compound, benzene, in the CPO of an aliphatic compound, n-hexane, in millisecond contact times over Rh and Pt in the regions favoring olefin production. Catalyst supports designed to enhance heterogeneous and 48

65 homogeneous chemistry in the millisecond reactor will be used to further reveal the relative reactivities of benzene and n-hexane in homogeneous and heterogeneous reaction pathways. This information provides insight in the engineering of CPO processes for the production of olefins or synthesis gas from complex fuel mixtures. 3.2 Experimental Reactor system The reactor consisted of a 19 mm inner diameter (ID) quartz tube as shown in Figure 3-1 and described elsewhere [7]. The fuel, n-hexane (HPLC-grade 95+% purity with ~5% C 5 and C 6 saturated hydrocarbon isomers, Sigma Aldrich) and benzene (99+% purity, Sigma Aldrich) were delivered into the reactor through an automotive fuel injector (Delphi) from a 5 psig pressurized tank. High pressure cylinders of high purity gases (99.9+ %, Airgas) fed the system with O 2 and N 2, and the flow rates were adjusted to air stoichiometry using mass flow controllers (Brooks 5850i). The walls along the vaporization section were heated with Variac controlled resistive heating tape. The catalyst supports were 45 and 80 pores per linear inch (PPI), cylindrical ceramic foam monoliths (92% Al 2 O 3, 8% SiO 2, 16.5 mm outer diameter (OD), 10 mm length, Vesuvius Hi-Tech Ceramics). The catalyst was positioned between two un-coated cylindrical 80 PPI ceramic foam monoliths. These two foams were used as heat-shields and mixers to reduce the axial radiation heat losses and promote reactant mixing upstream of the catalyst. To measure catalyst operating temperatures, a K-type thermocouple (Omega) was positioned on the backface of the cylindrical foam catalyst. Under steady-state operation, the upstream feed temperature was maintained at ~ 150 C. During operation, the reactor pressure was maintained near atmospheric conditions (1.1 atm), and a total flow rate of 4 standard liters per minute (SLPM at 25 C and 1 atm or GHSV~1x10 5 hr - 1 ) was also maintained. This total flow rate corresponds to catalyst contact times of ~7 milliseconds assuming an entering reactant temperature of 800 C and pressure of 1.1 atm. Typically, samples were taken after continuously operating for 30 minutes at a specified inlet composition and flow rate. The feed stream stoichiometry is reported as the carbon to oxygen ratio (C/O). C/O is defined as the number of moles of carbon atoms divided by the number of moles of oxygen atoms from O 2 in the mixture. Previous experimentation of hydrocarbon in CPO reactors suggest that the maximum ethylene and propylene occur in the fuel-rich regime where insufficient O 2 is supplied to partially oxidize the fuel, typically between C/O of 1.1 and 1.5 [6-8,11,15-16]. Therefore, C/O ratios between 0.8 and 2.0 were investigated to evaluate the olefin production 49

66 from the fuels. Results were reported where no observable catalyst deactivation was observed during experimental operation Catalyst 45 PPI and 80 PPI α-alumina (92% Al 2 O 3, 8% SiO 2 ) monoliths were used as catalyst supports, coated with ~5 weight % Rh and ~5 weight % (of monolith) Pt. The 80 PPI catalyst supports were also coated with ~4 weight % (of monolith) γ-alumina wash-coat (Alfa Aesar) using the incipient wetness technique, as previously described [7,11,31-32]. The pore diameters of the 45 PPI and 80 PPI wash-coated monoliths are ~1.00 mm and ~0.50 mm, respectively, which were observed through SEM micrographs of the foams. From previous experimentation, the larger pore size, 45 PPI monoliths promote less surface and more homogeneous chemistry than compared to the 80 PPI wash-coated monoliths [4,6,8,11, 31-32]. These two monoliths were chosen primarily to investigate the effect of the fuel in a reactor designed to enhance heterogeneous synthesis gas or to enhance homogeneous cracking reactions. The wash-coat also increased the dispersion of the metal on the catalyst surface, further increasing available catalytic surface on 80 PPI supports. Using the incipient wetness technique, the Rh metal was applied by dropping an aqueous solution of Rh(NO 3 ) 3 onto the ceramic foam monoliths, and Pt metal was applied by dropping an aqueous solution of H 2 PtCl 6. The Rh coated monoliths were then dried and placed in a furnace in the presence of air at 600 C for 6 hours, while the Pt coated monoliths were dried and placed in a furnace under flow of 10 mole % H 2 in N 2 at 500 C for 6 hours Product analysis Product gases for n-hexane and benzene were analyzed using a single-column gas chromatograph (HP 5890 Series II) equipped with a capillary column (J&W Scientific GASPRO, 0.32 mm id, 60 m length) and thermal conductivity detector. This column was capable of detecting both permanent gases and higher hydrocarbons. Reaction products containing carbon are reported on a carbon atom basis using selectivities. Product selectivity (S) is defined as the number of carbon or hydrogen atoms in product i divided by the carbon or hydrogen atoms in the converted fuel. H 2 and H 2 O are reported as selectivity on a hydrogen atom basis. The oxygen atom balance was closed to determine the water molar flow rate. Applying this quantitative method, the carbon and hydrogen atom balances typically closed within +/- 7%. 50

67 As shown later, benzene does not produce large selectivities to ethylene and propylene under the specified reaction inlet conditions. Since benzene does not produce these olefins, the production of these olefins is attributed to the linear alkane fuel n-hexane. Thus, the carbon atom selectivities of ethylene and propylene from a mixture of benzene and n-hexane are also reported on the number of carbon atoms that were reacted in the n-hexane only. 3.3 Results Pure components on Rh, 80 PPI Figure 3-2 displays conversion and the catalyst back-face temperature for benzene (Panel A), and n-hexane (Panel B) on Rh coated, 80 PPI foam supports. The catalyst back-face temperature for the pure component benzene was ~300 o C higher than the pure component n- hexane at corresponding C/O ratios, while the total conversion of benzene and n-hexane were similar. For both benzene and n-hexane, the O 2 conversion was >98% for all C/O ratios. Figure 3-2 also shows the hydrogen atom selectivity for H 2 and H 2 O and also displays the carbon atom selectivities for CO, CO 2, and ethylene and propylene for benzene (Panel C) and n- hexane (Panel D). H 2, H 2 O, CO, CO 2, CH 4 (not shown), ethylene and propylene, 1 to 3 carbon linear hydrocarbons (not shown), and 4 to 6 carbon linear olefins (not shown) were the major products observed for n-hexane. For benzene, H 2, H 2 O, CO, and CO 2 were the major products with traces (<0.5% selectivity) of methane (not shown), acetylene (not shown), and ethylene. As the C/O ratio increased, olefins (primarily ethylene and propylene) were produced in place of synthesis gas for n-hexane. For C/O > 1.0, H 2 from benzene was replaced with H 2 O, while CO remained constant, without significant olefin production. The maximum H 2 and CO selectivities were typically observed at C/O ~ 1.0, and were determined to be ~85% H 2 and ~90% CO selectivities for both benzene and n-hexane Mixtures on Rh, 80 PPI Figure 3-3 displays the operating catalyst back-face temperature and benzene and n-hexane conversions on Rh coated 80 PPI foam supports for a 20 mole% benzene:80 mole% n-hexane mixture (Panel A) and for a 50 mole% benzene:50 moles% n-hexane mixture (Panel B). At corresponding C/O, high benzene concentrations in the mixture resulted in elevated back-face temperatures. For example, the temperature of n-hexane was 800 C at C/O = 1.2, while the temperatures at this same C/O ratio for the 20:80 and 50:50 mixture were 840 C and 940 o C, respectively. Similar to the pure components, the O 2 conversion was >98% for all C/O ratios. 51

68 Additionally, the conversion of benzene and n-hexane in the mixtures did not reflect the single component mixture conversion at corresponding C/O. In the 20:80 mixture, the n-hexane conversion was similar to the single component n-hexane conversion at corresponding C/O ratios. However, the conversion of benzene in the 20:80 mixture was much higher than the conversion of pure benzene. Also, when comparing the reactivities of benzene and n-hexane in the 20:80 mixture, the conversion of benzene was higher than the conversion of n-hexane for C/O > 1.3. Benzene conversion remained at ~85% for C/O > 1.1, while n-hexane conversion decreased from 85% at C/O = 1.1 to 40% at C/O = 1.9. In the 50:50 mixture, the n-hexane conversion was greater than the single component n- hexane conversion at corresponding C/O ratios. However, the conversion of benzene in the mixture was similar to the conversion of pure benzene. Also, when comparing the reactivities of benzene and n-hexane in the mixture, the conversion of benzene was lower than the conversion of n-hexane for C/O > 1.3. Benzene conversion decreased from 70% at C/O = 1.1 to 55% at C/O =1.9, while the n-hexane conversion decreased from 95% at C/O = 1.1 to 60% at C/O = 1.9. Figure 3-3 also shows the hydrogen atom selectivity for H 2 and H 2 O and also displays the carbon atom selectivities for CO, CO 2, and ethylene and propylene for 20:80 (Panel C) and 50:50 (Panel D) mixtures. Generally for both mixtures, as the C/O ratio increased, olefins (primarily ethylene and propylene) were produced in place of synthesis gas. The optimum H 2 and CO selectivities were typically observed at C/O ~ 1.0: ~80% H 2 and ~90% CO selectivities for both mixtures. When compared to the pure components, the mixtures showed an increase in ethylene and propylene in place of synthesis gas. The ethylene and propylene selectivities increased from 5% for pure n-hexane to 15% and 30% for the 20:80 and 50:50 mixtures, respectively Pure components on Pt, 80 PPI Figure 3-4 displays conversion and the catalyst back-face temperature for benzene (Panel A), and n-hexane (Panel B) on Pt coated, 80 PPI foam monolith supports. Similar to the Rh results, the catalyst back-face temperature for the pure component benzene was ~250 o C higher than the pure component n-hexane at corresponding C/O ratios. However, contrary to Rh, the total conversion of benzene was lower than n-hexane. Also, at corresponding C/O ratios, the backface temperature was ~50 o C higher for Pt than Rh. For both benzene and n-hexane, the O 2 conversion was >98% for all C/O ratios. Figure 3-4 also shows the hydrogen atom selectivity for H 2 and H 2 O and also displays the carbon atom selectivities for CO, CO 2, and ethylene and propylene for benzene (Panel C) and n- hexane (Panel D). Similar to Rh, H 2, H 2 O, CO, CO 2, ethylene and propylene, 1 to 3 carbon linear 52

69 hydrocarbons (not shown), and 4 to 6 carbon linear olefins (not shown) were the major products observed for n-hexane. For benzene, H 2, H 2 O, CO, and CO 2 were the major products with traces (<0.5% selectivity) of methane (not shown), acetylene (not shown), and ethylene. As the C/O ratio increased, olefins (primarily ethylene and propylene) were produced in place of synthesis gas for n-hexane. For C/O > 1.0, H 2 from benzene was replaced with H 2 O, while CO remained relatively constant. The optimum H 2 and CO were typically observed at C/O ~ 1.0 and are determined to be ~80% H 2 and ~90% CO selectivities for both benzene and n-hexane, which were similar to results obtained on Rh. Olefins and H 2 O selectivities were greater on Pt than Rh catalysts Mixtures on Pt, 80 PPI Figure 3-5 displays the operating catalyst back-face temperature and benzene and n-hexane conversions in a 20 mole% benzene:80 mole% n-hexane mixture (Panel A) and in a 50 mole% benzene:50 moles% n-hexane mixture (Panel B). At corresponding C/O, increasing benzene concentrations in the mixture resulted in elevated back-face temperatures. For example, the temperature of n-hexane was 880 C at C/O = 1.2, while the temperature at this same C/O ratio for the 20:80 and 50:50 mixture were 880 C and 950 o C, respectively. Similar to the pure components, the O 2 conversion was >98% for all C/O ratios. The conversion of benzene and n-hexane in the mixtures did not reflect the single component mixture conversions at corresponding C/O. In the 20/80 mixture, the n-hexane conversion and benzene conversions were greater than their single component conversions at corresponding C/O ratios. Also, when comparing the reactivities of benzene and n-hexane in this mixture, the benzene conversion was lower than the n-hexane conversion for all C/O ratios studied. Benzene conversion decreased from ~95% at C/O = 0.85 to ~50% at C/O = 1.9, while n-hexane conversion decreased from ~99% at C/O = 0.85 to ~55% at C/O = 1.9. This result is the contrary to conversion trends observed for Rh coated catalysts. In the 50:50 mixture, the n-hexane conversion was greater than the single component n- hexane conversion at corresponding C/O ratios, and the conversion of benzene in the mixture was similar to the conversion of pure benzene. Also, when comparing the reactivities of benzene and n-hexane in this mixture, the conversion of benzene was lower than the conversion of n- hexane for all C/O ratios studied. Benzene conversion decreased from ~90% at C/O = 0.85 to ~40% at C/O = 1.9, while n-hexane conversion decreased from ~99% at C/O = 0.85 to ~75% at C/O =

70 Figure 3-5 also shows the hydrogen atom selectivity for H 2 and H 2 O and also displays the carbon atom selectivities for CO, CO 2, and ethylene and propylene for 20:80 (Panel C) and 50:50 (Panel D) mixtures. Generally for both mixtures, as the C/O ratio increased, olefins (primarily ethylene and propylene) were produced in place of synthesis gas. For both mixtures, the optimum H 2 and CO selectivities (~80% H 2 and ~90% CO) were typically observed at C/O ~ When compared to the pure components, the mixture showed an increase in ethylene and propylene in place of synthesis gas. The ethylene and propylene selectivities increased from 20% for pure n-hexane to 25% and 27% for the 20:80 and 50:50 mixtures, respectively. On Pt, pure n- hexane produced higher levels of ethylene and propylene than observed on Rh However, when comparing the mixtures, the levels of ethylene and propylene generated were similar Rh, 45 PPI support Figure 3-6 displays the operating catalyst back-face temperature and n-hexane and/or benzene conversions for pure n-hexane (Panel A), a 20 mole% benzene:80 mole% n-hexane mixture (Panel B), and a 50 mole% benzene:50 moles% n-hexane mixture (Panel C) over Rh coated 45 PPI foam supports. The CPO of pure benzene was also attempted on 45 PPI supports, however, several problems were encountered that prevented steady state operation. These problems included the formation of coke downstream, increased pressure drop through the reactor, and ignition of the fuel with unconverted oxygen downstream of the catalyst. Since steady-state operation was not achieved with pure benzene, no data regarding reforming pure benzene on 45 PPI supports is presented. At corresponding C/O, increasing benzene concentrations in the mixture results in elevated temperatures. For example, the temperature of n-hexane was 840 C at C/O = 1.2, while the temperature at this same C/O ratio for the 20:80 and 50:50 mixture were 900 C and 920 o C, respectively. Furthermore, the conversion of n-hexane was greater in the mixture than when n- hexane was partially oxidized separately. The conversion of pure n-hexane for C/O = 1.2 was 85%, while the conversion of n-hexane in the mixture was 98% for corresponding C/O. Benzene displayed a lower conversion than n-hexane in the mixture for all C/O. Benzene conversion remained ~20 to ~30% lower than n-hexane for all corresponding C/O ratios. For n-hexane, the O 2 conversion was >97% for all C/O ratios. For the 20:80 mixture, the O 2 conversion decreased from >98% at C/O = 0.85 to >93% at C/O = 1.9, while for the 50:50 mixture, the O 2 conversion decreased from >98% at C/O = 0.85 to >90% at C/O = 1.9. Figure 3-6 shows the hydrogen atom selectivity for H 2 and H 2 O and also displays the carbon atom selectivities for CO, CO 2, and ethylene and propylene. H 2, H 2 O, CO, CO 2, ethylene and propylene, and 4 to 6 carbon olefins were the major products observed for n-hexane. On 45 PPI 54

71 supports, synthesis gas selectivities were lower and the CO 2, H 2 O and olefin selectivities were higher than those observed 80 PPI supports. The optimum H 2 and CO selectivities were typically observed at the lowest measured C/O ratio (0.85). As the benzene concentration increased in the mixture, the maximum H 2 selectivity decreased from ~55% H 2 for pure n-hexane to ~40% H 2 for the 50:50 mixture. The maximum CO selectivities remained relatively constant at ~75% for both mixtures and pure n-hexane. When the C/O ratio increased, olefins, primarily ethylene and propylene, were produced in place of synthesis gas for n-hexane. When compared to the pure component n-hexane, the mixture displayed similar selectivities of ethylene and propylene in place of synthesis gas Pt, 45 PPI support Figure 3-7 displays the operating catalyst back-face temperature and n-hexane and/or benzene conversions for pure n-hexane (Panel A), a 20 mole% benzene:80 mole% n-hexane mixture (Panel B), and a 50 mole% benzene:50 moles% n-hexane mixture (Panel C) over Pt coated 45 PPI ceramic foam supports. While Rh coated catalysts were successfully operated with high benzene concentrations at all C/O ratios designed for this study, the addition of benzene into the feed greatly reduced the autothermal operability limits on Pt-coated 45 PPI foam supports. As the benzene concentration increased in the mixture from 0% to 50%, the fuel-rich limit in autothermal operability decreased from C/O ~ 1.9 to C/O ~1.0. With the 20:80 mixture, the reactor did not operate at steady-state autothermally above C/O ratios of ~1.5, while with the 50:50 n-hexane-benzene mixture, the reactor did not run at steady-state autothermally at C/O ratios > 1.0. Above these C/O ratios, the reactor temperature gradually decreased to temperatures that were ~200 o C greater than preheat temperatures, and the oxygen conversion decreased to below 50% conversion over a 90 minute operation period. On Pt coated 45 PPI foam supports, the pure n-hexane fuel conversion and C and H atom selectivities to combustion products and olefins were greater than those observed on Pt-coated 80 PPI supports. When benzene was added to the n-hexane, the fuel conversion of n-hexane remained relatively constant, while the benzene fuel conversion was lower than Pt-coated 80 PPI supports. On Pt-coated 45 PPI supports, the C and H atom selectivities to synthesis gas, combustion products, and olefins were similar to the selectivities observed for pure n-hexane at corresponding C/O ratios where autothermal operation was achieved. For n-hexane and the mixtures where autothermal operation was achieved, the O 2 conversion was >85% for all C/O ratios. 55

72 3.3.7 Olefins from n-hexane As mentioned previously, benzene does not appear to produce ethylene and propylene under these reactor conditions. Assuming that benzene does not produce ethylene and propylene, the ethylene and propylene selectivity can be attributed to the fuel n-hexane only. Applying this reasoning, Figures 3-3, 3-5, 3-6, and 3-7 display ethylene and propylene selectivity with respect to the entire fuel mixture and with respect to n-hexane. Considering only the carbon from n- hexane can produce ethylene and propylene, the maximum ethylene and propylene selectivity from n-hexane over 80 PPI Rh supports are ~20% and ~50% for the 20:80 and 50:50 mixture, respectively, while the maximum is <5% for pure n-hexane. On 80 PPI Pt catalysts, the ethylene and propylene selectivities from n-hexane are 40% and 60% for the 20:80 mixture and 50:50 mixture, respectively, while the maximum is 20% for pure n-hexane. The maxima in ethylene and propylene selectivity also occur at C/O ~ 1.9. On Rh coated 45 PPI supports, the ethylene and propylene selectivities from pure n-hexane was 40%, while for the mixture, the maximum increased to 50% and 60% for the 20:80 and 50:50 mixture, respectively. The maxima in ethylene and propylene selectivity occurred between C/O ~ 1.2 and 1.9. However, since benzene addition reduced the autothermal operation at C/O ratios where maximum olefin selectivities were observed, minimal increase in the maximum olefin selectivities generated from n-hexane was observed with benzene addition on Pt coated 45 PPI supports Conversion to oxidation products Assuming that n-hexane is the source of all olefins, such as ethylene and propylene, the total conversion of n-hexane to oxidation products CO and CO 2 can be determined. This conversion better indicates the relative conversion of benzene and n-hexane to oxidation products. The conversion to oxidation products is shown for the pure components (Panels A and B in Figures 3-2, 3-4), as well the 50:50 mixtures (Panels B in Figures 3-3, 3-5). On both Rh and Pt, when viewing pure component conversions to the oxidation products CO and CO 2, the benzene and n- hexane conversions were very similar at corresponding C/O ratios. For the 50:50 mixtures on both Rh and Pt, the oxidation product conversions indicate that benzene conversion to oxidation products is greater than the n-hexane conversion to oxidation products Homogeneous chemistry The selectivity and reactivity of homogeneous chemistry and heterogeneous chemistry from the catalyst support in the partial oxidation of n-hexane was also investigated. To isolate this 56

73 chemistry, the metal coated support was removed from the reactor and replaced with an uncoated 80 PPI α-alumina foam monolith. A furnace was used to maintain reaction temperatures similar to the temperatures observed with Pt and Rh coated 80 PPI supports. n-hexane and air were also delivered at the same flow rates used for the catalytic partial oxidation experiments (4 SLPM, 25 o C, 1 atm). The residence time within the furnace was ~100 ms, which is 10 times longer than the residence time within a Pt or Rh-coated catalyst. Figure 3-8 displays the conversion and operating temperature of n-hexane partial oxidation at various C/O ratios. Figure 3-8 also shows the C and H atom product selectivities from n-hexane. Similar products were observed from the un-coated supports and the metal-coated supports, however, the ratios of H 2 to H 2 O and CO to CO 2 were lower over the un-coated 80 PPI foam supports. At C/O = 1.3, ~1 mole H 2 to 1 mole H 2 O and ~9 moles CO to 1 mole CO 2 were observed for the un-coated α-alumina supports. ~8 moles H 2 to 1 mole H 2 O and ~18 moles CO to 1 mole CO 2 were detected for Rh coated 80 PPI foam supports, and ~5 moles H 2 to 2 moles H 2 O and ~14 moles CO to 1 mole CO 2 were obtained on Pt coated 80 PPI foam catalysts. Higher n-hexane conversion and olefin selectivities were also observed within the furnace than observed with the autothermal Pt and Rh catalytic partial oxidation; however, these higher conversions are likely the result of the longer residence time (longer exposure to high temperatures) within the furnace. Furthermore, autothermal operation was only obtained on the Pt and Rh coated foam supports. 3.4 Discussion The roles of benzene and n-hexane in homogeneous and heterogeneous reaction pathways will be proposed in the following sections from the investigation of the two catalyst supports which were designed to enhance heterogeneous and homogeneous chemistry in the millisecond reactor. Furthermore, the product selectivities and fuel reactivities differed for Rh and Pt catalysts. These differences in reactivities of benzene and n-hexane over Pt and Rh and resulting product formation pathways will be discussed Routes to synthesis gas & combustion products High resolution spatial profiles of species and temperatures have been measured for methane partial oxidation within similar 80 PPI ceramic foam supports [19-22]. From these experiments, the evolution of synthesis gas and combustion products from methane has been quantitatively identified. Results indicate that catalytic partial oxidation and combustion dominate initially in the catalyst until all the O 2 is consumed, which is typically within the first two millimeters 57

74 of the catalyst. For methane, H 2, CO, and H 2 O are mostly produced with small amounts of CO 2, and the temperature typically is 100 to 200 C higher than the measured back-face temperature. Further endothermic catalytic reforming of the unreacted fuel occurs with water, produced from complete fuel oxidation, to further increase H 2 and CO and reduce the operating temperature. Generally, a similar species evolution pathway that was observed with methane is suspected to occur for higher hydrocarbons, such as n-hexane and benzene. However, unlike methane partial oxidation, homogenous chemistry becomes important in the partial oxidation of larger hydrocarbon fuels. At experimental temperatures and pressures, the homogeneous chemistry for methane oxidation has been shown to be <10%, and often, the dynamics of catalytic partial oxidation of methane can be predicted well without including a homogeneous oxidation pathway [23-26]. For linear alkanes at similar residence times, homogeneous oxidation and pyrolysis have been experimentally observed at temperatures as low as 600 K [27] and for benzene, homogeneous oxidation and pyrolysis has been experimentally observed at 1000 K [28]. While heterogeneous chemistry have been observed to generate primarily H 2 and CO within milliseconds, these homogenous pathways primarily produce higher selectivities to H 2 O, CO 2, and olefins, and occur at slower rates than competing heterogeneous chemistry [23-31]. Figure 3-8 displays the species selectivities for n-hexane partial oxidation in a reactor system that does not contain Rh or Pt coated supports. At simulated catalytic partial oxidation reaction temperatures and pressures (1.1 atm, ~800 o C), homogeneous chemistry does occur and produces lower ratios of synthesis gas to combustion products than observed with Pt and Rh catalyzed partial oxidation. High olefin yields were also observed without the metal coated catalysts. These results further demonstrate that homogeneous chemistry is likely present at catalytic partial oxidation conditions, less selective toward the generation of synthesis gas, and more selective toward olefins and combustion products Routes to olefins Within these short-contact time reactors, olefins are assumed to be produced mostly through homogenous cracking reactions of aliphatic compounds, such as n-hexane [6-7, 15]. High temperatures of the exothermic catalytic combustion and partial oxidation reactions likely provide energy for homolysis of the aliphatic fuel to produce two alkyl radicals. These radicals decompose by β scission to yield ethylene and a smaller radical or by β hydrogen elimination to yield a larger stable α-olefin and a hydrogen radical [6,25]. Radical decomposition continues with each radical eliminating ethylene molecules until an ethyl or propyl radical remain. This residual ethyl or propyl radical then performs β-hydrogen elimination to form ethylene or propylene. In addition, lower operating temperatures appear to favor β-hydrogen elimination to large α-olefins 58

75 over complete β scission to ethylene and propylene. A previous study applying radical-induced homogeneous pyrolysis mechanisms for iso-octane and n-octane also predicted the yields and distribution of olefins experimentally observed in an identical catalytic reactor system, which further supports this pathway toward generating olefins [15]. Additionally, homogeneous cracking of linear alkanes through radicals typically yields linear hydrocarbons and olefins, whereas heterogeneous catalytic cracking through carbocations over acidic surfaces, such as zeolites, yield primarily branched olefins and hydrocarbons [33]. β- scission pathways are favored though radical catalyzed reactions, while carbocation intermediates favor isomerization of the carbon skeleton. Since branched hydrocarbon species are not typically observed as reaction products from linear hydrocarbon feedstocks, the reaction pathway most likely follows the proposed radical-induced homogeneous pathway. The generation of radicals from the catalytic surface or from oxygen molecules may also promote homogeneous pyrolysis [28-32,34]. As mentioned previously and observed for the pure component, benzene is not expected to produce olefins under CPO reactor conditions. Pyrolysis pathways for linear alkanes require the breaking of the parent molecule s carbon skeleton to yield olefins, and this back-bone typically requires a hydrogen to carbon ratio (H/C) greater than the ratio in the olefin product (H/C 2). However, benzene does not contain a sufficiently high internal H/C ratio in its structure to form ethylene or propylene through homolytic reactions. Also, the high carbon-carbon bond-strength and resonance stability of the aromatic ring reduces pyrolysis reaction rates that form small olefins. Previous homogeneous models and experiments on benzene pyrolysis and oxidation at these reactor conditions indicate minimal production of ethylene and propylene, further supporting the observation that benzene is not a major fuel source for the production of olefins [28-31]. From previous studies and commercial applications on Pt, benzene can be hydrogenated to cyclohexane [35], and previous studies within CPO reactors have shown that cyclohexane can successfully produce olefins [36], such as ethylene and propylene. Therefore, the pathway for benzene to hydrogenate to cyclohexane and then undergo thermal cracking to olefins may also be possible. However, this pathway is not expected to dominate. Hydrogenation reactions of benzene on Pt-type catalysts are typically thermodynamically controlled, and to thermodynamically favor cyclohexane formation, low temperatures (<300 o C) and high pressures (>1 bar) are required [35]. At CPO reactor conditions (1 bar, 1000 o C), benzene formation is strongly thermodynamically favored over cyclohexane. Thus, thermodynamics suggest that benzene does not generate olefins through this catalytic hydrogenation to cyclohexane pathway. 59

76 3.4.3 Effect of catalyst support: surface area and mass transfer Previous results on lower hydrocarbons have shown that catalyst supports with high specific surface areas produce larger amounts of synthesis gas and fewer olefins and water than supports with low specific surface areas [6,8,15-16]. In the current work, the comparison of the selectivities obtained from 80 PPI with wash-coat catalyst to 45 PPI without wash-coat catalyst with the same metal for a single fuel shows that this trend also holds for the CPO of these higher hydrocarbons. The ratio of the active surface area to the volume of the gas phase inside the catalytic foam (S/V ~ 4/d c ) is higher for the 80 PPI catalysts than the 45 PPI foams, because the average pore diameter is smaller (0. 5 vs mm, respectively). The addition of a wash-coat also increases the dispersion of catalyst particles within the catalyst, further increasing the catalyst surface area for the 80 PPI foams. Therefore, the ratio of heterogeneous to homogenous reactions taking place in the 80 PPI foams should be higher. Since H 2 and CO are believed to be formed heterogeneously, and ethylene and other olefins are believed to be formed largely in the gas phase, experiments performed on 80 PPI catalysts should generate higher selectivities of heterogeneously-derived synthesis gas, while reactions on 45 PPI foams produce greater amounts of homogeneously-generated olefins. Additionally, previous modeling of methane over 80 PPI ceramic foams and straight channel monoliths have indicated that the mass transfer of limiting reactants is also important under these reaction conditions, especially in the oxidation zone [22,37]. Because of the importance of foam catalysts for short contact time reactions, a correlation for the Sherwood (Sh) number in dependence of the Reynolds (Re) and Schmidt (Sc) numbers has been established by Giani et al. for pore dimensions of 1-5 mm [38]: k d Sh = 1.11Re D m 1 c =. (3-1) Sc where k m is the mass transfer coefficient, d c is the pore diameter, and D is the diffusion coefficient. Applying the definition of the Reynolds number, the mass transfer rate of the limiting reactant can be described as a function of the pore diameter: k m α d. (3-2) c

77 This correlation implies that larger pore diameter supports (45 PPI) should have lower mass transfer rates of the reactants and products to and from the catalytic surface. At a constant residence time, lower mass transfer likely results in the reactants remaining in the gas-phase and undergoing more homogeneous oxidation and pyrolysis reactions. From experiments, the 80 PPI supports typically achieved equilibrium concentrations of H 2, CO, CO 2, and H 2 O at low C/O (C/O < 1.0). At low C/O, sufficient levels of O 2 exist to partially oxidize all of the fuel to CO and H 2, and the surface reactions are more dominant than homogeneous reactions in the smaller pore size supports, enabling the fuel to be completely catalytically reformed without yielding olefin intermediates. For C/O > 1.0, insufficient levels of O 2 exist to partially oxidize all of the fuel, and as a result, unconverted fuel remains available to homogeneously crack to olefins. However, the heterogeneous reforming rates and mass transfer rates in these small pore size supports are very high, resulting in complete O 2 conversion and higher selectivities of heterogeneously produced H 2 and CO than homogeneously produced olefins and oxidation products. For 45 PPI catalysts where the reactants are exposed to less catalytic surface, the production of olefin intermediates was higher than the 80 PPI supports, even under fuel lean conditions (C/O <1.0). The products also yielded higher selectivities to homogeneously favored oxidation products, H 2 O, CO, and CO 2. These results further suggest that olefins and some oxidation products are generated in the gas phase of these catalytic reactors. Furthermore, incomplete O 2 conversion was also observed for these 45 PPI supports, which further suggests that lower reactant mass transfer and catalytic oxidation rates exist within the larger pore size, lower catalytic surface area supports Synthesis gas generation: Pt vs Rh For the reaction of the same fuel on the same support, Rh catalysts always generate higher selectivities to synthesis gas and lower water and olefin selectivities than Pt catalysts. These results suggest that consistent mechanistic differences exist among the heterogeneous reactions that occur on Rh and Pt catalysts for each fuel [16]. Hickman and Schmidt compared the elementary step surface reaction mechanisms of the direct formation of H 2 from the partial oxidation of methane on Rh and Pt [39]. They found that the primary difference that leads to the large discrepancy in the experimentally measured H 2 selectivities for Pt and Rh lies in the difference in the activation energy for formation of OH on the surface: 61

78 H (s) + O(s) OH(s) + *. (3-3) The activation energy for this reaction on Rh is significantly higher than on Pt. This energy barrier dictates that less adsorbed hydrogen atoms are going to combine with adsorbed O atoms and are therefore more likely to combine with other H atoms to form H 2 instead of forming water through OH. H 2 and CO can also be formed indirectly, beginning with complete oxidation of hexane or benzene to H 2 O and CO 2 followed by catalytic reforming of unreacted fuel to H 2 and CO through steam reforming and CO 2 reforming: 19 o KJ C6H14 + O2 6CO2 + 7H2O, ΔHR = 3170, (3-4) 2 mol C H o KJ + 6H 2O 6CO + 13H 2, ΔHR = 704, (3-5) mol o KJ C 6 H14 + 6CO2 12CO + 7H 2, ΔHR = (3-6) mol If the indirect formation of H 2 and CO is considered, the discrepancy in the performance of Rh and Pt catalysts with respect to synthesis gas production could be a function of the difference in the activities of the metals. Both Rh and Pt appear to be efficient oxidation catalysts, however Rh may be a much better reforming catalyst than Pt. Recently published spatial profiles for methane partial oxidation on Pt and Rh over 80 PPI foam supports indicate that both Rh and Pt produce similar levels of water and CO 2 in the presence of oxygen [22]. However, the reforming of this generated water with methane was significantly lower for Pt than Rh. These results suggest that Pt and Rh are both adequate oxidation catalysts, however, Rh is better at steam reforming. Furthermore, as mentioned previously, Pt catalysts display higher selectivities to olefins. Possibly, since olefin production occurs homogeneously, the olefin production rates are likely to be independent on the catalyst; however, the consumption of olefins is likely to be strongly dependent on the catalyst. The lower steam reforming activity on Pt catalysts may result in less olefins being steam reformed to synthesis gas, causing the olefin selectivities to be higher on Pt catalysts. 62

79 3.4.5 Temperature effects Aromatics, when partially oxidized to synthesis gas, produce more heat than partially oxidized aliphatics. This additional heat released from the aromatic compounds may yield higher adiabatic reforming temperatures in the fuel mixture when compared to the pure aliphatic feed. For the partial oxidation of aromatic benzene, the exothermic heat of reaction is approximately 1.5 times the heat released from the catalytic partial oxidation of aliphatic n-hexane: kj C 6H6 + 3O2 6CO + 3H2, Δ H o R = - 746, (3-7) mol kj C 6H14 + 3O2 6CO + 7H2, Δ H o R = (3-8) mol This additional heat released from benzene likely explains the higher adiabatic reforming temperatures in the fuel mixture when compared to pure n-hexane. Previous experiments with benzene and n-tetradecane pure components and mixtures over Ni have also shown similar temperature trends between catalytic aromatic and aliphatic reforming [14]. On the catalyst support designed to enhance heterogeneous reactions (80 PPI), higher temperatures were observed for the mixtures than pure n-hexane, suggesting that the benzene is partially oxidized. On the catalyst support designed to enhance homogeneous reactions (45 PPI), higher temperatures were also observed for the mixtures than pure n-hexane. However, the temperature difference between the mixtures and pure n-hexane on the 45 PPI support was not as large as the temperature difference between the mixtures and n-hexane on the 80 PPI support. Thus, larger temperature differences between the mixture and pure n-hexane appear to be an effect of catalytic reforming and not homogeneous reforming. However, temperature differences may not be completely dependent on the ability of benzene or n-hexane to oxidize. Endothermic chemistry, such as indirect steam reforming, can also alter operating temperatures. For specific fuel feeds, Pt catalysts typically operate at higher temperatures than Rh and generate higher selectivities of water. As discussed previously, Pt is not as active as Rh in catalyzing the steam reforming reaction, and this lower activity would result in lower endothermic and higher adiabatic operating temperatures. Furthermore, when benzene is added to n-hexane, the resulting elevated temperatures also appear to increase the endothermic homogeneous cracking rates of n-hexane in these millisecond reactors: 63

80 kj C 6H14 3C2H4 + H2, Δ H o R = (3-9) mol Higher operating temperatures in similar reactor systems have demonstrated elevated olefin yields from aliphatic fuels, which indicate increased cracking rates. Higher cracking rates increase the conversion of n-hexane to olefins in millisecond contact times. Thus, when benzene is added to n-hexane, the resulting elevated temperature from the partial oxidation of benzene likely explains the increase in n-hexane conversion and olefin selectivities. Additionally, the endothermicity of the cracking reactions is very small compared to the exothermic partial oxidation and endothermic steam reforming reactions, which suggests the heat load for this specific reaction does not appear to strongly affect the operating temperatures within the reactor. Therefore, high temperatures and olefins selectivities appear correlated within these adiabatic catalytic reactors Benzene and n-hexane reactivity: chemisorption On both Rh- and Pt-coated 80 PPI supports, when viewing pure component conversions to the oxidation products CO and CO 2, the benzene and n-hexane conversions are very similar at corresponding C/O ratios. These results suggest that benzene and n-hexane react through similar oxidation pathways to produce CO and CO 2. However, the reactivities to oxidation products differ greatly in the mixtures; the benzene conversion to oxidation products is greater than the n-hexane conversion. Even though the reforming of the pure components suggests similar catalytic pathways to CO and CO 2, the mixtures indicate that benzene preferentially reacts to oxidation products. Previous modeling and analysis of methane partial oxidation on Pt and Rh catalysts suggest that for high rates of interphase mass transfer, the adsorption of reactants is a limiting step in the catalytic partial oxidation system (flux-limited) [39,40]. Once adsorbed, the surface reactions and desorption occur more rapidly than the adsorption rates. Assuming that the adsorption of benzene and n-hexane were the limiting steps in the partial oxidation, relative reactivities of benzene and n-hexane in the mixtures could possibly be explained through competitive adsorption, where the adsorption rate of benzene at the catalyst surface was higher than n- hexane. On Rh catalyst supports where heterogeneous reactions were enhanced, a higher conversion of benzene in the presence of n-hexane was observed. Possibly, the conjugated π bonds in benzene may increase the ability to chemisorb on the Rh surface, resulting in a higher adsorption rate on the surface than n-hexane. From single crystal experiments on Rh (111), saturated 64

81 hydrocarbons displayed a lower adsorption probability (the statistical probability that a species will adsorb when it comes in contact with a surface) than aromatic compounds, such as benzene [42-43]. Once absorbed, benzene was shown to rapidly decompose to carbon and hydrogen at temperatures above 800K. This carbon and hydrogen coverage from benzene would likely rapidly consume co-adsorbed oxygen to yield oxidation products and heat. With benzene reacting rapidly with oxygen, less n-hexane would be oxidized and more n-hexane would remain in the homogeneous phase to react to olefins. Similar behavior suggesting that aromatic fuels have higher adsorption rates than saturated alkanes has been observed in low temperature, low concentration, catalytic oxidation of isooctane and benzene mixtures on Rh-coated, alumina wash-coated, straight channel cordierite monoliths (<100 ppm, <300 C) [44-45]. In these studies, the temperature at which 50% of isooctane converted was higher than the temperature at which 50% of benzene was converted. At these low operating temperatures and concentrations, homogeneous chemistry would be minimal inside the reactor, and thus, these studies correlated the 50% fuel conversion temperature with the catalytic activity of the iso-octane and benzene. The results suggest that a saturated hydrocarbon has a lower catalytic reactivity than benzene on Rh. Other experiments have been conducted with the partial oxidation of methylnaphthalene and n-tetradecane mixtures, as well as naphthalene and n-decane mixtures over Rh catalysts [9,47]. The addition of the aromatic compound increased the operating temperature and decreased the conversion of the linear hydrocarbon to synthesis gas and increased CO 2, olefin, and lower hydrocarbon formation. In these experiments, the observed inhibition toward reforming the aliphatic fuel to synthesis gas products and higher operating temperatures was also attributed to the higher adsorption rates of the aromatic on the Rh catalysts than saturated hydrocarbons. For Pt catalysts on supports where heterogeneous reactions were enhanced (80 PPI), the mixture displayed lower differences in reactivities to oxidation products between benzene and n- hexane than observed on similar Rh-coated supports. Possibly, Pt has the ability to chemisorb benzene and n-hexane at similar rates, resulting in a less biased adsorption rate toward the aromatic fuels on the catalyst surface [44-45]. Less biased adsorption rates would result in more n-hexane adsorbed and converted to oxidation products. Studies have suggested that benzene and saturated hydrocarbons have similar affinities for the Pt catalyst surfaces. In the low temperature, low concentration catalytic oxidation of isooctane and benzene mixtures on Pt-coated, alumina wash-coated, straight channel cordierite monoliths (<100 ppm, <300 C), the temperatures at 50% fuel conversion of isooctane and benzene were similar, indicating that benzene and isooctane have similar reactivities on Pt [44-65

82 45]. However, other studies have been performed on the catalytic oxidation of binary mixtures of n-hexane and benzene at low temperatures and concentrations on Pt (<5000 ppm, 150 o C), and have indicated that the competitive adsorption rate of benzene and consequently reactivity is significantly greater than n-hexane [46]. Additionally, experiments involving the partial oxidation of methylnaphthalene and n-tetradecane mixtures and the autothermal steam reforming of isooctane and xylene mixtures over Pt catalysts showed reduction in synthesis gas and enhancement in olefins and smaller hydrocarbons with the addition of the aromatic to the saturated hydrocarbon at short contact times [47]. In these experiments, the observed inhibition toward reforming to synthesis gas products and generation of olefins and hydrocarbons was credited to the higher adsorption rate of the aromatic for the Pt catalysts than saturated hydrocarbons Benzene reactivity on Pt and Rh: decomposition pathways The addition of benzene to n-hexane reduced the autothermal operability range on large pore size Pt catalysts; however, autothermal operability was not hindered on similar Rh-coated catalyst supports or on smaller pore size, larger surface area Pt-coated supports. From these experiments, benzene appears to react differently on Pt and Rh, and these differences could possibly be explained by the relative decomposition pathways of benzene on Pt and Rh. Previous work has shown that benzene decomposes on Rh(111) at temperatures as low as 400 K to form CH and C 2 H species [41-43], which suggests that the C-C bond breakage is favored over the dehydrogenation of chemisorbed benzene. The C-C bond breaking pathway may be a crucial step in explaining the difference between the reactivity of benzene on Pt and Rh. On Rh, breaking the C-C bonds likely yields mobile partially hydrogenated carbon species that can rapidly react with low concentrations of co-adsorbed oxygen to form CO and CO 2. Once formed, CO and CO 2 can easily desorb from the surface, providing vacant catalytic sites to facilitate additional surface reactions. If the surface and desorption reactions are sufficiently fast, the catalyst surface would be remain active to sustain autothermal operation. Conversely, studies of benzene decomposition on single crystal Pt (111) have shown that dehydrogenation (C-H bond breaking) is primarily favored over the C-C bond breakage [41-43]. Rapid dehydrogenation and hydrogen desorption would result in relatively immobile carbon clusters located on the catalyst surface [48]. To remove these carbon clusters from the surface, the carbon would likely have to be oxidized with co-adsorbed oxygen. However, immobility caused by the rapid dehydrogenation of clusters would slow the reaction with co-adsorbed oxygen, and to generate adequate oxidation rates to remove the clusters from the surface, a higher coverage of co-adsorbed oxygen would likely be required. 66

83 For small pore-size catalysts, the mass transfer rate of oxygen is sufficiently fast to provide the surface with a high coverage of oxygen to remove these carbon clusters. However, for the larger pore size catalysts, the oxygen mass transfer rate is too slow to provide this high surface oxygen coverage. Without high surface oxygen coverage, the clusters would grow, gradually covering the Pt catalyst with immobile carbon and block oxygen from adsorbing onto the catalyst surface. Once the growing clusters cover the majority of the catalytic surface area, the catalyst would deactivate and extinguish. These events could possibly explain the observed reduction in autothermal operation with high benzene concentrations on large pore size Pt catalysts Coke formation When compared to oxygen-free pyrolysis, the presence of oxygen can promote the extraction of a hydrogen atom from benzene, yielding very reactive benzene radicals [28-31]. If the radical concentrations reach high levels, these radicals can lead to the formation of polyaromatics and coke [14,31,49]. For pure benzene, oxygen breakthrough within the reactor could promote downstream coke formation, and therefore, rapid consumption of oxygen within these catalytic reactors appears to be essential for eliminating downstream polyaromatic and coke formation. From the experimental results, catalysts with high surface area, small pore-size catalysts were successful in rapidly consuming oxygen within the catalyst bed leaving the un-reacted benzene in an oxygen-free environment downstream of the catalyst zone. Due to the thermal stability of the benzene structure, benzene remained relatively un-reacted in this oxygen-free, pyrolysis environment, resulting in no coke formation. However, lower surface area, larger poresize catalysts (45 PPI) were unable to completely consume oxygen within the catalyst bed, exposing high concentrations of benzene to oxygen-catalyzed radicals for long residence times (seconds). Extended residence times in the presence of oxygen and high temperatures most likely promoted the coke formation observed for the lower surface area high pore size catalysts on Rh. 3.5 Summary Aromatic benzene and aliphatic n-hexane mixtures, along with the corresponding pure components, were partially oxidized in millisecond contact times over Pt and Rh coated monoliths. Varying the catalyst support further revealed heterogeneous and homogeneous pathways that the aromatic and aliphatic compounds encounter in CPO reactors. Over both Pt and Rh-coated catalysts, the reactivities of single component benzene and n-hexane are different than the reactivities of these respective compounds in a mixture. 67

84 The results presented in this chapter suggest that, within a benzene/n-hexane mixture, benzene competitively adsorbs onto the catalyst, reducing the catalytic oxidation of the n-hexane. As a result, less n-hexane reacts with oxygen and more remains in the gas phase where homogeneous cracking to olefins occurs. The benzene is partially oxidized to synthesis gas, which releases high quantities of heat and elevates the adiabatic operating temperature. Higher temperatures further increase the reforming and cracking rates in these millisecond reactors, resulting in increased n-hexane conversion. Benzene is a valuable commodity chemical and would likely be separated from the mixture and not be used as a fuel source in the generation of olefins. If aromatic containing mixtures were to be used to produce olefins, the aromatic within the feedstock would have to be of low value (not worth the cost of separation). Such low value aromatic feedstocks could be in form of lignin or poly-aromatic coal tars, however, further research in the CPO of these compounds is required. Additionally, in the experiments with benzene, the reactivity of the aromatic ring was investigated. However, the reactivity of functional groups associated with more complex aromatic compounds (e.g. toluene, phenol, xylene, cumene) may alter the homogenous and heterogeneous reaction pathways that were observed with benzene. Likely, these functional groups will increase homogeneous and heterogeneous reactivity of the aromatic compound 29-30,44-45]. Insight into the effect of these functional groups on aromatic compounds will require additional studies. Overall, this research has provided insight into the CPO of complex fuel mixtures for the suppression or promotion of olefins. In the next chapter, the reforming of benzene will be further investigated by measuring the evolution of species within a Rh-coated catalyst support. 3.6 References 1. R. Subramanian, L.D. Schmidt, Angewandte Chemie International Edition 44 (2005) R. Deng, F. Wei, Y Jin, Q. Zhang; Y. Jin, Chemical Engineering Technology 25 (2002) H. Zimmermann, R. Walzl, Ethylene, Ullmann's Encyclopedia of Industrial Chemistry, Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, G.J. Panuccio, L.D. Schmidt, Applied Catalysis A: General 313 (2006) A. Bodke, D. Olschki, L.D. Schmidt, E. Ranzi, Science 285 (1999) A.G. Dietz III, A.F. Carlsson, L.D. Schmidt, Journal of Catalysis 176 (1998)

85 7. J.J. Krummenacher, K.N. West, L.D. Schmidt, Journal of Catalysis 215 (2003) J.J. Krummenacher, L.D. Schmidt, Journal of Catalysis 222 (2004) R. Subramanian, G. Panuccio, J.J. Krummenacher, I.C Lee, L.D. Schmidt, Chemical Engineering Science 59 (2004) R.P. O'Connor, E.J. Klein, L.D. Schmidt, Catalysis Letters 70 (2001) B.J. Dreyer, I.C. lee, J.J. Krummenacher, L.D. Schmidt, Applied Catalysis A: General 307 (2006) M. Flytzani-Stephanopoulos, G.E Voecks, International Journal of Hydrogen Energy 8 (1983) J.P Kopasz, L.E. Miller, D.V. Applegate, SAE SP 1790 (2003) J.P. Kopasz, D. Applegate, L. Miller, H.K. Liao, S. Ahmed, International Journal of Hydrogen Energy 30 (2005) G. J. Panuccio, K. A. Williams, L. D. Schmidt, Chemical Engineering Science 61 (2006) G. J. Panuccio, B. J. Dreyer, L. D. Schmidt, AIChE Journal 51(1) (2006) A. Bodke, S. Bharadwaj, L.D. Schmidt, Journal of Catalysis 179 (1998) N.J. Degenstein, R. Subramanian, L.D. Schmidt, Applied Catalysis A: General 305 (2006) R. Horn, K.A. Williams, N.J. Degenstein, L.D. Schmidt, Journal of Catalysis, 242 (2006) R. Horn, N.J. Degenstein, K.A. Williams, L.D. Schmidt, Catalysis Letters, 110 (3-4) (2006) R. Horn, K.A. Williams, N.J. Degenstein, L.D. Schmidt, Chemical Engineering Science 62 (2007) R. Horn, K.A. Williams, N.J. Degenstein, A. Bitsch-Larsen, D. Dalle-Nogare, S.A. Tupy, L.D. Schmidt, Journal of Catalysis 249 (2007) K.A. Williams, R. Horn, L.D. Schmidt, AIChE Journal 53(8) (2007) O. Deutschmann, L.D. Schmidt, AIChE Journal 44 (1998) C.T. Goralski, R.P. O Conner, L.D. Schmidt, Chemical Engineering Science 55 (2000) G. Veser, J. Fraunhammer, Chemical Engineering Science 55 (2000) H.J. Curran, P. Gaffuri, W. J. Pitz, C. K. Westbrook, Combustion and Flame 114 (1998) M.U. Alzueta, P. Glaborg, K. Dam-Johansen, International Journal of Chemical Kinetics 32 (2000) Z.M. Djurisic, A.V Joshi, H. Wang, Second Joint Meeting of the U.S. Sections of the Combustion Institute, Oakland, CA, March 25-28,

86 30. Z.M. Djurisic, MS Thesis, University of Delaware, Newark, Delaware, U.S.A., I.D. Costa, R. Fournet, F. Billaud, F. Battin-Leclerc, International Journal of Chemical Kinetics 35 (2003) E. Ranzi, T. Faravelli, P. Gaffuri, E. Garavaglia, A. Goldaniga, Industrial & Engineering Chemistry Research 36 (1997) M.M. Green, H.A. Wittcoff, Organic Chemistry Principles and Industrial Practice, Wiley- VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, K.L. Hohn, P.M. Witt, L. D. Schmidt Catalysis Letters 54 (1998) M.L. Campbell, Cyclohexane, Ullmann s Encyclopedia of Industrial Chemistry, Wiley- VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, R.P. O Conner, E.J. Klein, D. Henning, L.D. Schmidt, Applied Catalysis A: General 238 (2003) O. Deutschmann, R. Sweidernoch, L.I. Maier, D. Chatterjee, Natural Gas Conversion VI, Studies in Surface Science and Catalysis 136 (2001) L. Giani, G. Groppi, E. Tronconi, Industrial & Engineering Chemistry Research 44 (2005) D.A. Hickman, L.D. Schmidt, AIChE Journal 39(7) (1993) K. A. Williams and L. D. Schmidt, Applied Catalysis A: General 299 (2006) M.E. Viste, K.D. Gibson, S.J. Sibener, Journal of Catalysis 191 (2000) B.E. Koel, J.E. Crowell, B.E. Bent, C.C. Mate, G.A. Somorjai, Journal of Physical Chemistry 90 (1986) B.E. Koel, J.E. Crowell, C.M. Mate, G.A. Somorjai, Journal of Physical Chemistry 88 (1984) M.J. Patterson, D. E. Angove, N. W. Cant, Applied Catalysis B: Environmental 26 (2000) M.J. Patterson, D.E. Angove, N.W. Cant, Applied Catalysis B: Environmental 35 (2001) S. Ordonez, L. Bello, H. Sastre, R. Rosal, F.V. Diez, Applied Catalysis B: Environmental 38 (2002) D. Shekhawat, T.H. Gardner, D.A. Beery, M. Salazar, D.J. Haynes, J.J. Spivey, Applied Catalysis A: General 311 (2006) B.J. McIntyre, M. Salmeron, G.A. Somorjai, Journal of Catalysis 164 (1996) H. Richter, J.B. Howard, Physical Chemistry Chemical Physics 4 (2002)

87 Fuel Injector Air Heating tape g Static mixer Thermocouple Catalyst Heat shields Insulation Products Figure 3-1. Schematic of reactor. Hydrocarbon fuel is introduced into the reactor through a lowflow automotive fuel injector, vaporized, and mixed with air. The heat-shields prevent thermal radiation heat losses in the axial direction, and the tube is wrapped in insulation to reduce radial heat losses. 71

88 A B X (%) O 2 C 6 H T ( o C) X (%) O 2 n-c 6 H 14, 80 PPI, Rh * n-c H n-c H T ( o C) 20 T T 800 C 6 H 6, 80 PPI, Rh C/O C/O C 100 CO D 100 CO 80 H 2 80 H 2 S H or S C (%) C H, PPI, Rh H 2 O S H or S C (%) n-c 6 H 14, 80 PPI, Rh 20 CO 2 C 2 H 4 &C 3 H 6 20 H 2 O C 2 H 4 &C 3 H 6 CO C/O C/O Figure 3-2. Catalytic partial oxidation of n-hexane and of benzene on ~5 wt% Rh on ~5 wt% γ - alumina wash-coat, 80 PPI α-alumina foam support at 4 SLPM (GHSV ~ 10 5 h 1 ) and 1.1 atm. A pure benzene feed is presented in panels A & C, while pure n-hexane feed is presented in panels B & D. Operating back-face temperature (T, ) and fuel conversion (X, :C 6 H 6, :n-c 6 H 14, :O 2 ) are displayed. Selectivities (S C or S H ) are defined as (C or H atoms in product)/(c or H atoms in converted fuel) and are presented for H 2 ( ), H 2 O ( ), CO ( ), CO 2 ( ), and C 2 H 4 & C 3 H 6 ( ). n- C 6 H 14 * ( ) indicates conversion of C in n-hexane to oxidation products (CO and CO 2 ). 72

89 A 100 O B 100 O C 6 H C 6 H 6 n-c 6 H X (%) 60 n-c H T H:20 B, Rh, 80 PPI C/O T ( o C) X (%) H:50 B 80 PPI, Rh * n-c H C/O T T ( o C) C CO H 2 D CO 50 H: 50 B 80 PPI, Rh S H or S C (%) H:20 B, Rh, 80 PPI * C H &C H C H &C H H O 2 S H or S C (%) H 2 C 2 H 4 &C 3 H 6 C 2 H 4 &C 3 H 6 * H 2 O 0 CO C/O 0 CO C/O Figure 3-3. Catalytic partial oxidation of n-hexane and benzene mixtures on ~5 wt% Rh on ~5 wt% γ alumina wash-coat, 80 PPI α-alumina foam support at 4 SLPM (GHSV ~ 10 5 h 1 ) and 1.1 atm. A 80:20 moles% n-hexane:benzene feed (80 H:20 B) is presented in panels A & C, while a 50:50 moles%:n-hexane:benzene feed (50 H:50 B) is presented in panels B & D. Operating back-face temperature (T, ) and fuel conversion (X, :n-c 6 H 14, :C 6 H 6, :O 2 ) are displayed. Selectivities (S C or S H ) are defined as (C or H atoms in product)/(c or H atoms in converted fuel) and are presented for H 2 ( ), H 2 O ( ), CO ( ), CO 2 ( ), and C 2 H 4 & C 3 H 6 ( ). n-c 6 H 14 * ( ) indicates conversion of C in n-hexane to oxidation products (CO and CO 2 ). C 2 H 4 & C 3 H 6 * ( ) indicates selectivity from carbon atoms in n-hexane. 73

90 A B C 6 H 6 O O 2 n-c 6 H X (%) T 1200 T ( o C) X (%) n-c 6 H 14 * 1000 T ( o C) C H, 80 PPI, Pt C/O T n-c 0 6 H 14, 80 PPI, Pt C/O C 100 CO D 100 CO 80 H 2 80 H 2 S H or S C (%) C 6 H 6, 80 PPI, Pt H O 2 C H &C H CO 2 S H or S C (%) n-c 6 H 14, 80 PPI, Pt H O 2 C H &C H CO C/O C/O Figure 3-4. Catalytic partial oxidation of n-hexane and of benzene on ~5 wt% Pt on ~5 wt% γ alumina wash-coat, 80 PPI α-alumina foam support at 4 SLPM (GHSV ~ 10 5 h 1 ) and 1.1 atm. A pure benzene feed is presented in panels A & C, while pure n-hexane is presented in panels B & D. Operating back-face temperature (T, ) and fuel conversion (X, :C 6 H 6, :n-c 6 H 14, :O 2 ) are displayed. Selectivities (S C or S H ) are defined as (C or H atoms in product)/(c or H atoms in converted fuel) and are presented for H 2 ( ), H 2 O ( ), CO ( ), CO 2 ( ), and C 2 H 4 & C 3 H 6 ( ). n- C 6 H 14 * ( ) indicates conversion of C in n-hexane to oxidation products (CO and CO 2 ). 74

91 A 100 O B 100 O n-c 6 H n-c 6 H X (%) H: 20 B, 80 PPI, Pt C 6 H T ( o C) X (%) T C 6 H T ( o C) 20 T C/O H: 50 B, 80 PPI, Pt 900 * n-c H C/O C S H or S C (%) CO H 2 C 2 H 4 &C 3 H 6 * 80 H: 20 B, 80 PPI, Pt C 2 H 4 &C 3 H 6 D S H or S C (%) CO H 2 50 H: 50 B, 80 PPI, Pt C H &C H * C H &C H H 2 O 20 H 2 O 0 CO C/O 0 CO C/O Figure 3-5. Catalytic partial oxidation of n-hexane and benzene mixtures on ~5 wt% Pt on ~5 wt% γ -alumina washcoat, 80 PPI α-alumina foam support at 4 SLPM (GHSV ~ 10 5 h 1 ) and 1.1 atm. A 80:20 moles% n-hexane:benzene feed (80 H:20 B) is presented in panels A & C, while a 50:50 moles%:n-hexane:benzene feed (50 H:50 B) is presented in panels B & D. Operating back-face temperature (T, ) and fuel conversion (X, :n-c 6 H 14, :C 6 H 6, :O 2 ) are displayed. Selectivities (S C or S H ) are defined as (C or H atoms in product)/(c or H atoms in converted fuel) and are presented for H 2 ( ), H 2 O ( ), CO ( ), CO 2 ( ), and C 2 H 4 & C 3 H 6 ( ). n-c 6 H 14 * ( ) indicates conversion of C in n-hexane to oxidation products (CO and CO 2 ). C 2 H 4 & C 3 H 6 * ( ) indicates selectivity from carbon atoms in n-hexane. 75

92 A B C 100 O O 2 n-c 6 H O 2 n-c 6 H n-c 6 H X (%) D S H or S C (%) T n-c H, 45 PPI, Rh C/O n-c 6 H 14, 45 PPI, Rh CO C H &C H H O 2 H 2 T ( o C) X (%) E S H or S C (%) H:20 B, 45 PPI, Rh C/O CO H 2 T C 6 H 6 80 H:20 B, 45 PPI, Rh C 2 H 4 &C 3 H 6 * H 2 O C 2 H 4 &C 3 H 6 T ( o C) X (%) F S H or S C (%) H:50 B, 45 PPI, Rh C 6 H C/O CO H 2 H 2 O T C 2 H 4 &C 3 H 6 * C 2 H 4 &C 3 H 6 50 H:50 B, 45 PPI, Rh T ( o C) 0 CO C/O 0 CO C/O 0 CO C/O Figure 3-6. Catalytic partial oxidation of n-hexane and n-hexane-benzene mixtures on ~5 wt% Rh on 45 PPI α-alumina foam support at 4 SLPM (GHSV ~ 10 5 h 1 ) and 1.1 atm. A pure n-hexane feed is presented in panels A & D; a 80:20 moles% n-hexane:benzene feed (80 H:20 B) is presented in panels B & E, while a 50:50 moles% n-hexane:benzene feed (50 H:50 B) is presented in panels C & F. Operating back-face temperature (T, ) and fuel conversion (X, :n-c 6 H 14, :C 6 H 6, :O 2 ) are displayed. Selectivities (S C or S H ) are defined as (C or H atoms in product)/(c or H atoms in converted fuel) and are presented for H 2 ( ), H 2 O ( ), CO ( ), CO 2 ( ), and C 2 H 4 & C 3 H 6 ( ). C 2 H 4 & C 3 H 6 * ( ) indicates selectivity from carbon atoms in n-hexane. 76

93 A X (%) D S H or S C (%) 100 O n-c H T n-c H, 45 PPI, Pt C/O n-c 6 H 14, 45 PPI, Pt CO C H &C H H O 2 CO 2 T ( o C) B X (%) E S H or S C (%) n-c 6 H 14 C 6 H 6 O C/O T T ( o C) X (%) H:20 B, 45 PPI, Pt 0 80 H:20 B, 45 PPI, Pt CO C 2 H 4 &C 3 H 6 * C 2 H 4 &C 3 H 6 H 2 O C F S H or S C (%) n-c 6 H 14 O 2 C 6 H C/O T 50 H:50 B, 45 PPI, Pt CO H 2 O 50 H:50 B, 45 PPI, Pt C 2 H 4 &C 3 H 6 * H 2 T ( o C) CO 2 10 CO 2 0 H 2 0 H 2 0 C 2 H 4 &C 3 H C/O C/O C/O Figure 3-7. Catalytic partial oxidation of n-hexane and n-hexane-benzene mixtures on ~5 wt% Pt on 45 PPI α-alumina foam support at 4 SLPM (GHSV ~ 10 5 h 1 ) and 1.1 atm. A pure n-hexane feed is presented in panels A & D; a 80:20 moles% n-hexane:benzene feed (80 H:20 B) is presented in panels B & E, while a 50:50 moles% n-hexane:benzene feed (50 H:50 B) is presented in panels C & F. Operating back-face temperature (T, ) and fuel conversion (X, :n-c 6 H 14, :C 6 H 6, :O 2 ) are displayed. Selectivities (S C or S H ) are defined as (C or H atoms in product)/(c or H atoms in converted fuel) and are presented for H 2 ( ), H 2 O ( ), CO ( ), CO 2 ( ), and C 2 H 4 & C 3 H 6 ( ). C 2 H 4 & C 3 H 6 * ( ) indicates selectivity from carbon atoms in n-hexane. 77

94 A B 50 n-c 6 H 14, 80 PPI 80 O 2 n-c 6 H CO C 2 H 4 &C 3 H 6 X (%) T 800 T ( o C) S H or S C (%) H 2 O n-c 6 H 14, 80 PPI C/O 10 0 H 2 CO C/O Figure 3-8. Heated n-hexane in air over an uncoated 80 PPI α-alumina foam support at 4 SLPM (GHSV ~ 10 4 h 1 ) and 1.1 atm. Operating temperature (T, ) and fuel conversion (X, :n-c 6 H 14, :O 2 ) are displayed. Selectivities (S C or S H ) are defined as (C or H atoms in product)/(c or H atoms in converted fuel) and are presented for H 2 ( ), H 2 O ( ), CO ( ), CO 2 ( ), and C 2 H 4 & C 3 H 6 ( ). Reaction was not autothermal; a tube furnace supplied the heat to maintain operating temperatures. 78

95 Chapter 4: Comparison of spatial profiles in the catalytic partial oxidation of benzene and methane on rhodium 4.1 Introduction The oxidation of aromatic fuels is fundamentally and practically important in fuel reforming and pollution abatement, since most practical hydrocarbon fuel blends or chemical feedstocks consist of large amounts of aromatics [1]. Also, aromatic compounds are known to be harmful to the environment, and the emission of these species from a number of oxidation systems is a significant concern. Several studies of the catalytic oxidation of aromatic compounds have been attempted, including some kinetic studies [2-5]. Many of these studies have been made on various noble metal catalysts at many different conditions. A major drawback to these studies is that they were not conducted at the high temperature and atmospheric pressure conditions involved in commercial processes. Additionally, most of the kinetic studies have been carried out at low concentrations (less than 5000 ppm). However, these low concentrations are not representative of the high concentrations that are often applied in industrial processes. Furthermore, numerous studies have been performed to determine axial temperatures (gas and surface) within working catalysts. However, high-resolution composition profiles are lacking [6-9]. Temperatures have been measured in extruded monoliths, fixed beds, and gauzes with IR thermography and thermocouples, although these experiments may introduce radiation and conduction losses compared to insulated catalysts [10-14]. Axial species profiles have been estimated by measuring exit compositions with different thickness monoliths or with sphere bed of different lengths [15-16]. However, due to the coupling of heat, mass, and momentum transfer within these reactors, these variable or differential bed-length reactors do not necessarily duplicate the profiles within a single catalyst [6-9]. Recently, Horn et. al developed an elegant sampling technique that enables species profiles to be examined within 0.3 mm resolution within catalytic foams [6-9]. This technique has been demonstrated primarily for the CPO of gaseous methane and ethane fuels, and has not yet been applied to liquid feedstocks, such as benzene. In this chapter, spatially resolved ( 1 mm) measurements of composition and temperature profiles within autothermally operated ceramic foam catalysts will be demonstrated for both liquid benzene and gaseous methane hydrocarbon fuels. The capillary technique developed for this 79

96 purpose perturbs the concentration, flow, and temperature fields only minimally. Insights into the species evolution within high temperature, short-contact time reactors will be observed, which can be used to develop kinetic homogeneous and heterogeneous mechanisms of more complicated liquid aromatic feedstocks. 4.2 Experimental Catalyst preparation The catalyst examined in this experiment was Rh coated on a γ-alumina wash-coated, 80 PPI α-alumina cylindrical foam monolith support. To prepare the catalyst, ~4 weight % (of monolith) γ alumina wash-coat (Alfa Aesar) was added to the α-alumina foam support, which covered the support in an ~30 to 50 μm thick layer [17-18]. The wash-coat was applied using the incipient wetness technique, wherein a slurry of γ-alumina in distilled water was added drop-wise to both sides of the foam monolith and capillary forces distributed the slurry throughout the foam. The monolith was dried for 8 hours under ambient conditions and placed in an oven at 600 C for 4 hours. Next, using the same incipient wetness technique, ~5% weight (of monolith) Rh was added to the wash-coated support by applying an aqueous solution of Rh(NO 3 ) 3 (Alfa Aesar, 13 wt% Rh in HNO 3 ) dropwise. The monolith was then dried and placed in a furnace at 600 C for 6 hours. A channel for the capillary sampling (diameter = 0.74 mm) was drilled through the center z- axis of the α-alumina foams. The channel was drilled prior to impregnating the α-alumina foam support with wash-coat and catalyst. Figure 4-1 displays scanning electron microscopy (SEM) images of a cross section of the porous support along with the drilled hole. The foam was cut and imaged after the profile experiment was performed Reactor and capillary system: methane The reactor and capillary system used in this study is a further development of the set-up used to measure species and temperature profiles for the methane CPO in adiabatic foams reported in the literature [6-9]. A schematic of the set-up for methane partial oxidation is shown in Figure 4-2. Reactant gases CH 4, O 2, and Ar were fed from high-pressure cylinders (99.9%+, Airgas) to the reactor and metered through calibrated mass flow controllers (Brooks 5850i). Feed stoichiometry was specified as the ratio of the carbon atoms fed to oxygen atoms fed (C/O ratio). In methane experiments, argon and oxygen were fed at a ratio of 3.76:1 to simulate air. The 80

97 steady state experiment was performed at a total flow rate of 1.8 liters per minute (20 o C, 1 atm). Upstream of the catalyst, the gases were heated to ~150 o C by adjusting a variac connected to resistive heating tape wrapped around the inlet tubing. The catalyst foam was sandwiched between front and back heat shields (uncoated foams), wrapped tightly in alumina-silicate paper and placed in a cylindrical quartz reaction tube (ID = 19 mm, length = 10 cm). The channels drilled along the centerline of each monolith were aligned to form a channel ~30 mm in length and 0.75 mm in diameter. The sampling capillaries used were fused silica gas chromatography (GC) capillary columns (OD = 0.65 mm, Deactivated FSOT, Alltech) having a polyimide coating on the outside that made the capillary flexible for handling but burned off quickly during the first use. An orifice was cut into the side of the capillary through which the gas sampling occurred, and the end of the capillary was sealed. The size of the orifice in the side of the fused silica tubing was <0.3 mm in diameter. The capillary exited the reactor by passing through the capillary port, a small stainless steel tube, ~0.8 mm ID, which entered the lower end of the reactor through a septum. The fused silica column was connected to the top of an axial positioning system used to move the capillary/temperature measurement device assembly along the reactor axis. The axial positioning system consisted of a stainless-steel micro-volume tee fitting (1/16, bore, Valco) mounted to a linear translation stage. Data was collected by moving the translation stage stepwise in either the positive or negative direction which yielded equivalent reactor temperature and species profiles along the centerline of the catalyst. For methane experiments, a stepper motor (T-LA60, Zaber Technologies) mounted underneath the reactor provided the movement of the translational stage with high precision. A port on the tee was connected to a stainless steel capillary (1/16, bore) that discharged into the inlet valve of the mass spectrometer. A pump generated a vacuum at the end of the stainless steel capillary (~100 Pa), drawing the gases from the sampling orifice into the vacuum chamber. The samples passed through a leak valve to a high vacuum chamber containing a quadrupole mass spectrometer (QMS). The sampling flow rate was determined largely by the choked flow in the stainless steel capillary and is on the order of 10 ml/min, which was a negligible disturbance in the flow through the reactor considering the total flow rate of 1.8 L/min. From previous experimental testing, the placement of the temperature measuring device, such as an Inconel-sheathed thermocouple, in the sampling tube provided an additional catalytic surface. This catalytic surface could provide species profiles that are not directly reflective to the 81

98 reaction occurring within the foam monolith. To eliminate this issue, temperature profiles were measured independently of species profiles. Two temperature measurement devices were applied in this system. First, a K-type thermocouple (0.254 mm OD, Omega) was placed in the capillary tube, aligned with the sample hole. The measured temperature from the thermocouple primarily reflected the gas temperature as the thermocouple is in thermal contact with the flowing gas but not with the surface of the catalyst. However, radiation and conduction from the catalyst to the thermocouple body also affected the temperature measurement. Second, to measure the temperature of the catalytic surface, a 0.33 mm diameter optical fiber was placed within the capillary. Thermal radiation emitted from the hot catalytic surface entered the tip of a quartz fiber and attenuated to an optical pyrometer (MI0GA 5-LO, Mikron Infrared), which accurately converted the radiation to temperatures between 300 and 1300 o C. This radiation measurement primarily indicated the surface temperature of the catalyst. Since the thermal conduction and convection could not be detected, the pyrometer reading was not convoluted with these other forms of heat transfer. Furthermore, the fiber wall contained a reflective coating, preventing radiation from entering along the radial wall of the fiber. To avoid radiation from transmitting from far above the pyrometer fiber tip, an α-alumina bead was placed above the fiber in the capillary. This bead was assumed to be in thermal equilibrium with the surrounding surface temperature of the catalyst. With this set-up, the fiber detects radiation within +/- 0.5 mm from the tip position within the catalyst. Figure 4-3 displays the fiber within the capillary Reactor and capillary system: benzene For benzene experiments, the experimental reactor was slightly modified. Liquid benzene was delivered though an ISCO Series D syringe pump to a vaporizer. The vaporizer consisted of a vertically placed, 19 mm diameter, 10 cm long quartz tube packed with 1 mm diameter quartz beads. The perimeter of the tube was wrapped with resistive heating tape. The liquid was delivered into the top of the vaporizer from the syringe pump. High purity O 2 and N 2 were also delivered at this position from high-pressure cylinders (99.9%+, Airgas) and were controlled through calibrated mass flow controllers (Brooks 5850e). Feed stoichiometry was specified as the ratio of the carbon atoms fed to oxygen atoms fed (C/O ratio). Nitrogen and oxygen were fed at a ratio of 3.76:1 to simulate air. The steady state experiment was performed at a total flow rate of 2.0 standard liters per minute (SLPM, 25 o C, 1 atm). The liquid benzene was vaporized and gases heated to ~150 o C by adjusting a variac connected to the resistive heating tape. The mixture of gaseous benzene, N 2, and O 2 exiting the vaporizer bottom was connected to the 82

99 reactor in the same position that is shown for the methane reactor. A K-type thermocouple (Omega) was placed ~15 mm upstream of the catalyst system to monitor the vaporizer temperature. Also, an additional ceramic foam monolith was placed ~20 mm upstream of the catalyst system to help mix the fuel stream exiting the vaporizer section. A micrometer with 1 mm resolution, which was manually adjusted, was used to translate the sampling position within the catalyst instead of a stepper motor. Instead of sucking through a vacuum pump, a back-pressure regulator valve was placed at the reactor exit and was adjusted to generate ~1.5 psig back-pressure on the system. This back pressure was sufficient to flow ~ 10 ml/min of gas through the capillary tube. The gas passed through the stainless steel tee to an additional sampling tee, which consisted of a side septum sample port. At each position, samples were taken through this port with a 1 ml syringe (Hamilton Gastight Samplelock) and injected into the GC for analysis. At each sampling position, gas continuously flowed through the tee for 20 minutes prior to sampling. The exiting stream from the tee was submerged in water to visually indicate that samples were not obtained from the exiting source and obtained from the gas flow from the reactor Data analysis: methane profiles The quadrupole mass spectrometer (QMS) used for data acquisition was a UTI Model 100C with electron impact ionization of 70 ev. During data collection, the pressure inside the QMS vacuum chamber was kept at ~10-7 torr. The QMS unit was computer-controlled via Labview. In the steady-state experiment about 100 spectra were averaged to determine the molar flow rates at each location in the catalyst. The Ar signal (m/z = 40 amu) was used as an internal standard to correct for changes in the total flow rate by chemical reactions. Binary mixtures of each individual component with argon were fed into the QMS to determine argon normalized sensitivity values at each m/z value of interest (H 2 at 2 amu, CH 4 at 15 amu, CO at 28 amu, O 2 at 32 amu, Ar at 40 amu, CO 2 at 44 amu). This array of sensitivities was used to calculate species flow rates as is standard for quantitative analysis of hydrocarbon mixtures. The water flow was determined by closing the O balance. Using this calibration and analysis technique the carbon and hydrogen balances typically close to +/- 5% and +/- 5%, respectively. Analysis was also performed through gas chromatography (GC) to verify the inlet and exit species molar flow rates. The GC analysis utilized an HP 5890 Series II with He carrier, 83

100 HayeSep D Packed Column, and thermal conductivity detector. The GC was calibrated in a similar manner, including binary mixtures, as described for the QMS Data analysis: benzene profiles Analysis was performed through gas chromatography (GC) to determine the molar flow species flow rates. The GC analysis utilized an HP 6890 with Ar carrier, J&W Scientific PLOT-Q Capillary Column (30 m length, 0.32 mm OD), and thermal conductivity detector. The GC was calibrated with known mixtures of permanent gases, olefins, and hydrocarbons from Matheson TRI-GAS. Since the carrier gas was Ar, N 2 was used as the reference gas for the GC sample run. The water flow was determined by closing the O balance. Using this calibration and analysis technique the carbon balances typically close to +/- 5%, hydrogen balance close to +/- 9% Reactor start-up and operation time For the methane experiments, the reactor was started by introducing H 2 and O 2 at a 2 to 1 feed ratio across the catalyst at ~50 o C. This stream was diluted with Ar at a synthetic air stoichiometry. The catalyst would ignite and the reactor temperature would increase to 600 o C, where methane was introduced and H 2 was removed. The flow of methane, Ar and O 2 were then adjusted to the flow rates applied in the steady-state experiment. Once light-off occurred, the catalyst was aged for 2 hours at reaction conditions before profiles were measured. The reactor was shut-down by first removing oxygen and then methane from the system and allowing Ar to convectively cool the reactor. Multiple catalyst runs were performed with ~20 hours of operations. After aging the catalyst for 2 hours, the spatial product and reactant profiles remained constant over multiple runs. Approximately 10 species profiles and 10 temperature profiles were obtained. Using a QMS and GC for analyzing the species enabled a species profile to be obtained within 1 hour, while a temperature profile was obtained in <10 minutes. For the benzene experiments, the reactor was started by introducing the vaporized benzene and air at 150 o C across the catalyst. A heater was placed around the catalyst wall to increase the catalyst temperature to >250 o C, which led to reactor light-off. Once light-off occurred, the heater was removed and a 2-inch thick woven ceramic fiber insulation was wrapped around the perimeter of the quartz reactor tube. The catalyst was then aged for 2 hours under reaction conditions. The reactor was shut-down by first removing oxygen and then benzene from the system and allowing N 2 to convectively cool the reactor. Multiple catalyst runs were performed with ~40 84

101 hours of operations. After aging the catalyst for 2 hours, the spatial product and reactant profiles remained constant over multiple runs. Approximately 3 species and 10 temperature profiles were obtained for benzene. Due to using GC for analyzing the species, 1 species profile required ~8 hours, while 1 temperature profile required < 10 minutes. 4.3 Results Methane: species Figure 4-4 displays experimental spatial profiles of the catalytic partial oxidation of methane at reaction conditions. Some oxygen and methane appear converted to synthesis gas and combustion products within a millimeter of the heat-shield. Within 2 mm of the catalyst region, all of the oxygen was consumed, and water, CO 2, H 2, and CO are produced. Water and synthesis gas products are the most dominant, with H 2 and CO produced in a 2 to 1 ratio and H 2 O and H 2 produced in a 1 to 1 ratio. A maximum in water was observed at the end of this oxidation zone. Downstream of this oxidation zone, water that was produced in the presence of oxygen was reacted with methane generating additional H 2 and CO in a 3 to 1 ratio. Water and methane consumption were observed through the entire length of the catalyst section downstream of the oxidation zone. Also, CO 2 produced within the presence of oxygen remained relatively constant downstream of the oxidation zone. Downstream of the catalyst, the experimental profile displayed a change in H 2. However, after close examination, no other species were changing; thus, the H 2 change was suspected to be an experimental artifact. With the exception of H 2, constant molar flow rates were experimentally observed downstream of the catalyst zone Methane: temperature In the upstream heat-shield, the surface temperature (obtained from the pyrometer) increases from 400 o C to 850 o C as the catalytic section is reached. At the beginning of the catalyst, a maximum in surface temperature (850 o C) is observed. Downstream of the front-face of the catalyst, the catalyst surface temperature decreases to 600 o C, and upon exiting the catalyst region, the surface temperature remains near 600 o C. 85

102 At the beginning of the front heat-shield, the thermocouple temperature, which indicates the gas phase temperature, is 180 o C and equilibrates with the surface temperature near the frontface of the catalyst. Since equilibration occurs prior to the observed maximum in surface temperature, the maximum temperature for the gas and surface appear to be similar. Once the gas phase has equilibrated with the surface temperature, the gas phase follows a similar trend as observed with the surface temperature measurement. Furthermore, the maximum temperature within the catalyst bed corresponds with the position where oxygen was completely consumed Benzene: species Figure 4-5 displays experimental spatial profiles of the catalytic partial oxidation of benzene at reaction conditions. Throughout the entire length of the heat-shield, oxygen and benzene conversion is experimentally observed. In the heat-shield, the consumption of fuel produces primarily CO, H 2 O, and CO 2. Product ratios are 4 to 1 and 2 to 1 for CO to CO 2 and H 2 O to CO, respectively. Traces of methane, acetylene, and ethylene were also observed but were not reported in the figures since they accounted for <0.5% of the total carbon fed into the reactor. Inside the catalyst region, the oxygen is completely consumed in the first 4 mm, which is 2 mm greater than methane. H 2, and CO generation with a ratio of 1 to 2 is experimentally observed with the consumption of oxygen and benzene. A small amount of H 2 O is generated in the oxidation zone, while CO 2 remains constant. After the oxygen is consumed in the catalyst, CO 2 remains relatively constant, while water slightly decreased. The consumption in water corresponds with the generation of H 2 and CO in a 1 to 1 ratio. Constant molar flow rates exiting the catalyst are observed upon exiting the catalyst Benzene: temperature In the upstream heat-shield, the surface temperature (obtained from the pyrometer) increases from 700 o C to 1000 o C as the front-face of the catalyst is reached. A maximum in surface temperature (1130 o C) is observed at 3 mm from the front-face of the catalyst. Downstream of the maximum, the catalyst surface temperature decreases to 1000 o C, and upon exiting the catalyst, the surface temperature further decreases to 920 o C. The maximum surface temperature corresponds well with the position where the oxygen is consumed. Generally, the autothermal operating surface temperature for benzene is 300 o C greater than for methane at each axial position At the beginning of the front heat-shield, the thermocouple temperature (gas phase) is 350 o C and equilibrates with the surface temperature near the front-face of the catalyst. Similar to 86

103 methane, since temperature equilibration occurs at the beginning of the catalyst region, the maximum in temperature for the gas and surface are similar within the catalyst. Once the gas phase has equilibrated with the surface temperature, the gas temperature follows a similar trend to the surface temperature. 4.4 Discussion Direct vs. indirect mechanism: methane and benzene An important and open question in CPO research is the reaction mechanism and product development in the catalyst bed under autothermal, high temperature conditions [6,19-19]. Direct and indirect mechanisms have been proposed. The direct mechanism assumes that H 2 and CO are primary reaction products formed by partial oxidation in the presence of gas-phase O 2. The direct mechanism for methane partial oxidation can be represented as the following: 1 CH 4 + O2 CO + 2H 2, 2 kj Δ H o R = (4-1) mol and for benzene partial oxidation: C + +, 6H6 3O2 6CO 3H2 kj Δ H o R = (4-2) mol The indirect mechanism postulates a two-zone model with strong exothermic fuel combustion to H 2 O and CO 2 at the catalyst entrance, followed by strongly endothermic steam- and CO 2 - reforming downstream, as shown for CH 4 : CH CH o KJ + 2O 2 CO2 + 2H2O, ΔHR = 803 (4-3) mol 4 o KJ + H2O CO + 3H2, ΔHR = 206 (4-4) mol 4 + o KJ CH4 + CO2 2CO + 2H2, ΔHR = + 247, (4-5) mol 87

104 and for benzene: C H 15 o KJ + O2 6CO2 + 3H2O, ΔHR = 3170 (4-6) 2 mol 6 6 C C H o KJ + 6H2O 6CO + 9H2, ΔHR = 704 (4-7) mol H o KJ + 6CO2 12CO + 3H2, ΔHR = 952 (4-8) mol Even though the terms direct and indirect are commonly used in the literature to discuss the mechanism of CPO, the equations above describe only a stoichiometrically observed product development in the catalyst bed and give no information regarding the elementary reaction steps occurring at the surface. In the experiments presented, H 2 and CO are formed in the presence of gas-phase O 2 at the entrance section of the catalyst, which follows the stoichiometric partial oxidation according to Equation 4-1 and 4-2. Although the partial oxidation reaction is observed, the H 2 formation may not be the primary event after CH 4 or C 6 H 6 dissociation at the catalyst surface. Possibly, surface hydrogen atoms H(s) are involved in rapid surface reactions with oxygen surface atoms O(s) before they leave the surface as H 2 or H 2 O. In this case, the reaction would be indirect even though it follows Equation 4-1. The actual surface reaction steps cannot be inferred from spatial profiles, thus, the reaction mechanism is discussed in terms of Equations 5-1 through 5-8 without using the terms direct and indirect. From examining Figures 4-4 and 4-5, the experimental spatial profiles reveal or exclude several of the identified stoichiometric reactions. For methane, in the presence of oxygen, a 2 to 1 production ratio of H 2 and CO indicate formation through a stoichiometric partial oxidation reaction. Downstream of the oxidation zone, a 3 to 1 production ratio of H 2 and CO and 1 to 1 consumption ratio of water and methane indicates species evolution through the stoichiometric steam reforming reaction. >50% of the total synthesis gas produced is generated from this secondary steam reforming. Furthermore, CO 2 remains constant downstream of the oxidation zone, suggesting that the stoichiometric CO 2 reforming reaction is not occurring. For benzene, a 2 to 1 production ratio of CO and H 2 and 3 to 1 consumption ratio of oxygen and benzene indicate formation through a stoichiometric partial oxidation reaction, while downstream of the oxidation zone, a ~1.5 to 1 production ratio of H 2 and CO and 1 to 1 88

105 consumption to production ratio of water and CO, respectively, indicates species evolution through the stoichiometric steam reforming reaction. Similar to methane, CO 2 remains constant, suggesting that the stoichiometric CO 2 reforming reaction is not occurring. Since little secondary steam reforming is occurring, the majority (>80%) of the synthesis gas is produced within the oxidation zone Temperature differences The experimental autothermal operating temperatures were much greater for benzene than for methane. This observation mostly can be attributed to additional heat released from the aromatic compound during the partial oxidation reaction. Reviewing Equations 5-1 and 5-2, the heat of reaction for benzene partial oxidation is much larger than for methane. Normalizing these reaction enthalpies to single carbon basis yields -124 KJ/mol and -36 KJ/mol for benzene and methane respectively. On this normalized basis, the exothermic heat of reaction of benzene partial oxidation is approximately 3.5 times the heat of reaction released from the partial oxidation of methane. Thus, when partially oxidizing, the excess heat released by benzene with respect to methane would significantly increase the adiabatic temperature of the reactor Upstream heat-shield: axial diffusion Some methane and oxygen appear to be converted in the upstream heat shield, ~1 mm before the catalyst front-face. This observed conversion appears to be attributable to the axial diffusion of species within the reactor. Previous 2D modeling accounting for axial diffusion in these reactors has shown that the slight observed fuel conversion near the front-face of the catalyst can be attributed to diffusion [6,19]. Even though axial diffusion is experimentally observed, the dispersion is limited to the ~1 mm upstream of the catalyst for methane. For benzene, axial dispersion may also exist. However, this dispersion should be limited to similar distances as observed with methane. Additionally, for benzene, H 2 and CO are the predominant species in the catalyst zone. The lowest molecular weight compound H 2, which is present in large concentrations, should axially diffuse at the fastest 1 2 rate ( 1 Dij α M, where D ij is the binary diffusion coefficient of species i into j, M is the molecular i weight of species i). Even though H 2 should be the fastest species to axially diffuse into the upstream zone, the levels of CO, CO 2, and H 2 O are much greater than the H 2 species, indicating that axial diffusion cannot completely account for the high levels of these compounds upstream of the catalyst zone. To account for the observed species evolution in the upstream heat-shield, chemical reactions are likely occurring. 89

106 4.4.4 Upstream heat-shield: homogeneous chemistry Unlike methane partial oxidation, homogenous chemistry is probably important in the partial oxidation of larger hydrocarbon fuels, such as benzene. At experimental temperatures and pressures, the homogeneous chemistry for methane oxidation has been shown to be <10%, and often, the dynamics of catalytic partial oxidation of methane can be predicted well without including a homogeneous oxidation pathway [19]. For benzene, homogeneous oxidation and pyrolysis has been experimentally observed at 1000 K in millisecond residence times [21]. With homogenous chemistry likely important for catalytic oxidation of benzene, artifacts in the experimental sampling technique may over-predict the conversion of fuel through homogeneous pathways. As mentioned previously, the sampling technique withdraws a sample of gas through a capillary placed along the z-axis of the cylindrical catalyst. Withdrawing the sample through the capillary, the reactants experience a longer residence time than occur within the reactor. This longer residence time enables more homogeneous chemistry than experimentally observed. Withdrawing the sample through upstream flow should also cool the reactants to pre-heat conditions quite rapidly, minimizing the homogeneous chemistry in the sampling capillary. The upstream alumina support appears non-catalytic for methane partial oxidation, however, for benzene, the alumina support may alter radical formation, propagation, and termination, potentially promoting the homogeneous pathways and rates of benzene oxidation observed within the reactor. Additionally, intermediate reactive oxygenate and hydrocarbon radicals from benzene oxidation could exist at low concentrations within the upstream heat-shield as well as the catalyst. However, with the relative long residence time (ms) in the sampling system, these unstable intermediates likely react to the final stable products observed in the GC analysis, which primarily were H 2, CO, CO 2, H 2 O, and C 6 H 6. Even with these possible experimental artifacts, a key observation from these experimental profiles is that for a relatively stable compound, benzene, oxidation pathways become important at these autothermal reaction temperatures. Experiments with benzene oxidation typically predict that homogeneous oxidation initiates at ~1000 K, which corresponds well with the upstream heatshield surface temperature [21]. Most likely, homogeneous oxidation chemistry plays even larger roles in the catalytic partial oxidation of more reactive aliphatic compounds such as n-heptane, iso-octane or n-hexadecane, where oxidation initiates at temperatures as low as 500 K [40-27]. 90

107 Also, the upstream chemical reactions observed for benzene produced primarily, CO, CO 2, and H 2 O. While CO is typically a desired product in partial oxidation products, CO 2 and H 2 O are not. Most of the time, the catalytic region is expected to react these intermediate and undesired products into the desired H 2 and CO, within equilibrium limitations. However, the CO 2 generated in the upstream heat-shield does not reform, which lowers the yield of desired CO. Previous studies of the catalytic partial oxidation of linear C 1 to C 16 hydrocarbons indicate that the CO 2 selectivity increases with hydrocarbon chain length [24]. Possibly, the increased homogenous reactivity of larger compounds, such as n-hexadecane, generates more homogeneous chemistry upstream of the catalyst. This upstream chemistry produces high selectivities to CO 2, which cannot be reformed within the downstream catalyst region. Furthermore, the complete oxidation of some hydrocarbon reactant prior to the catalyst region changes the actual fuel to oxygen ratio that was originally designated with upstream flow rates. For example, the C/O ratio with respect to benzene entering the upstream heat-shield was specified and observed to be ~1.1, and upon entering the catalyst region, the C/O ratio with respect to benzene was ~1.4. This change in hydrocarbon fuel to oxygen ratio could affect predictions in catalytic surface coverage and chemistry. Thus, the homogeneous chemistry observed upstream of the catalyst could make developing theoretical models of the catalytic partial oxidation performance of large hydrocarbons more difficult Mass transfer limitations Previous profile experiments were performed with the partial oxidation of methane on Rhcoated 80 PPI ceramic foam supports where the C/O ratio was varied [9]. From these experiments, the zone where oxygen was consumed remained constant, indicating that this zone was mass transfer limited on Rh. The length of the oxidation zone depends on the interplay of how quickly the limiting reactant (oxygen for methane partial oxidation) is converted at the catalyst surface and how quickly it can be transported to the catalyst surface from the bulk gas phase. In the random porous network of a foam catalyst, the bulk gas phase is well mixed, and high axial velocities render axial species diffusion is mostly negligible compared with convection (as observed with methane). However, radial diffusion cannot be neglected. Using the analysis provided by Horn et al., the mass transfer rate for the porous catalyst can be extrapolated from the data using a one-dimensional model with mass-transport resistance at the catalyst surface and assuming full mass transport control [9]: 91

108 1 V u z = ln (9) f S m k O2 where z is the axial coordinate within the catalyst section (z=0, front-face of catalyst), f is the fractional conversion of oxygen at z, u is the average gas velocity in m/s, S/V is the surface to volume ratio of the catalytic active surface to pore volume in m -1 m, and k O 2 is the oxygen mass transfer coefficient in m/s. z has also a weak temperature dependence, ~T 0.3, which can often be neglected. Experimentally, for an oxygen fractional conversion of 50%, f~0.50, the position for methane correlates to ~0.38 mm. With a u of ~0.55 m/s (T = 800 o C, P= 1 atm) and S/V of 8.0 x 10 3 m -1, m k O 2 is estimated to be ~0.13 m/s. Further profile experiments with benzene at various C/O ratios will be required to determine if the oxidation zone is kinetically or mass transfer limited in oxygen or benzene. However, by performing the Horn et al. analysis on the benzene profile, benzene yields a smaller oxygen mass transfer coefficient value than CH 4. If the analysis with methane indicates the oxygen mass transfer coefficient value at reactor conditions, the lower value extracted for benzene possibly indicates that oxygen mass transfer rates from the bulk to the surface are not the dominant limititations in the oxidation zone. Possibly, kinetic limitations contribute to the longer oxidation zone observed for benzene Benzene and methane oxidation zones The difference in the experimentally observed oxidation zones for benzene and methane may be related to the species ability to adsorb onto the catalyst surface with respect to oxygen. Experimental studies on single crystal Rh (111) have shown that benzene has a very high sticking coefficient (the statistical probability that a species will adsorb when it comes in contact with a surface) [26] while methane has a lower probability than oxygen [27-31]. A high oxygen sticking coefficient with respect to the carbon containing fuel may result in a surface predominately covered with oxygen molecules. When a methane molecule adsorbs onto an oxygen covered surface, the high concentration of surface oxygen promotes rapid decomposition of methane and production of synthesis gas and combustion products. These products easily desorb at reaction temperatures, enabling the oxidation process to continue with a large area of un-occupied sites. Often the rapid kinetic conversion leads to mass transfer limitations with respect to the limiting reactant oxygen as previously discussed. 92

109 In contrast, a high benzene sticking coefficient when compared with oxygen may cause the surface to be rapidly covered with benzene molecules. At high reaction temperatures, benzene should rapidly decompose to adsorbed carbon and hydrogen [5,26,32]. Surface hydrogen should also rapidly desorb from the surface, leaving the surface highly covered with non-volatile surface carbon. This high carbon coverage reduces the number of available sites for oxygen to adsorb. With fewer sites available, the oxidation reactions are significantly slowed, resulting in a longer oxidation zone dominated by kinetic limitations. 4.5 Conclusions The heterogeneous partial oxidation of benzene and methane has been examined on rhodium-coated ceramic foam monoliths at atmospheric pressure. While only one profile has been presented for each species, these studies have revealed the importance of homogeneous and heterogeneous chemistry within short-contact time high-temperature catalytic reactors. Results indicate that homogeneous and heterogeneous chemistry are important in the partial oxidation of benzene, while only heterogeneous chemistry is important for methane. Larger contributions of homogeneous chemistry in these reactors will likely complicate the mechanistic modeling of the catalytic reforming of higher hydrocarbons, such as benzene. In the next chapter, a reaction model will be developed that will attempt to capture both the homogeneous and heterogeneous chemistry observed within these reactors for benzene. 4.6 References 1. Z.M. Djurisic, A.V Joshi, H. Wang, Second Joint Meeting of the U.S. Sections of the Combustion Institute, Oakland, CA, March 25-28, M.J. Patterson, D.E. Angove, N.W. Cant, Applied Catalysis B: Environmental 35 (2001) M.J. Patterson, D. E. Angove, N. W. Cant, Applied Catalysis B: Environmental 26 (2000) S. Ordonez, L. Bello, H. Sastre, R. Rosal, F.V. Diez, Applied Catalysis B: Environmental 38 (2002) M.E. Viste, K.D. Gibson, S.J. Sibener, Journal of Catalysis 191 (2000) R. Horn, K.A. Williams, N.J. Degenstein, L.D. Schmidt, Journal of Catalysis 242 (2006) R. Horn, N.J. Degenstein, K.A. Williams, L.D. Schmidt, Catalysis Letters 110 (3-4) (2006)

110 8. R. Horn, K.A. Williams, N.J. Degenstein, L.D. Schmidt, Chemical Engineering Science 62 (2007) R. Horn, K.A. Williams, N.J. Degenstein, A. Bitsch-Larsen, D. Dalle-Nogare, S.A. Tupy, L.D. Schmidt, Journal of Catalysis 249 (2007) F. Basile, G. Fornasari, F. Trifiro, A. Vaccari, Catalysis Today 64 (2001) L. Basini, K. Aasberg-Peterson, A. Guarinoni, M. Ostberg Catalysis Today 64 (2001) M. Bizzi, L. Basini, G. Saracco, V. Specchia, Chemical Engineering Journal (Amsterdam, Netherlands) 90 (2002) S. Marengo, P. Comotti, G. Galli, Catalysis Today 81 (2003) B. Li, K. Maruyama, M. Nurunnabi, K. Kunimori, K. Tomishige Applied Catalysis A: General 275 (2004) E.J. Klein, S. Tummala, L.D. Schmidt, Natural Gas Conversion VI, Studies in Surface Science and Catalysis 136 (2001) D.A. Henning, L.D. Schmidt, Chemical Engineering Science 57 (2002) A. Bodke, S. Bharadwaj, and L. D. Schmidt, Journal of Catalysis 179 (1998) N.J. Degenstein, R. Subramanian, L.D. Schmidt, Applied Catalysis, A: General 305 (2006) K.A. Williams, R. Horn, L.D. Schmidt, AIChE Journal 53(8) (2007) A.P.E. York, T. Xiao, M.L.H. Green, Topics in Catalysis 22 (2003) M.U. Alzueta, P. Glaborg, K. Dam-Johansen, International Journal of Chemical Kinetics 32 (2000) K. A. Williams and L. D. Schmidt, Applied Catalysis A: General 299 (2006) Curran, H. J., P. Gaffuri, W. J. Pitz, C. K. Westbrook, Combustion and Flame 114 (1998) G. J. Panuccio, B. J. Dreyer, L. D. Schmidt, AIChE Journal 51(1) (2006) L. Giani, G. Groppi, E. Tronconi, Industrial & Engineering Chemical Research 44 (2005) B.E. Koel, J.E. Crowell, C.M. Mate, G.A. Somorjai Journal of Physical Chemistry 88 (1984) D.A. Hickman, L.D. Schmidt, AIChE Journal, 39(7) (1993) O. Deutschmann, R. Sweidernoch, L.I. Maier, D. Chatterjee, Natural Gas Conversion VI, Studies in Surface Science and Catalysis 136 (2001) R. Sweidernoch, S. Tischer, C. Correa, O. Deutschmann, Chemical Engineering Science, 58 (2003) C.T. Campbell, J.M. White, Journal of Catalysis 54 (1978) P.A. Thiel, J.T. Yates, Jr., W.H. Weinberg, Surface Science 82 (1979)

111 32. B.E. Koel, J.E. Crowell, B.E. Bent, C.C. Mate, G.A. Somorjai, Journal of Physical Chemistry 90 (1986)

112 A B C Figure 4-1. Scanning electron microscopy (SEM) images of a foam cross section at various magnifications. The foam was cut and imaged after the profile experiment was performed. In Panels A and B, the cross-section of the capillary sampling channel (~800 μm in diameter with ~500 μm pores) can be observed. Panel C displays the support surface coated with γ-alumina wash-coat and rhodium. 96

113 Figure 4-2. Schematic of the axial sampling system showing the interface with the reactor and vacuum system to collect spatially-resolved temperature and species data. Inset A and B show the channel in the monolith, the fused silica capillary, the sampling orifice, and the thermocouple. Diagram and pictures were adapted from Horn et. al [7]. Figure 4-3. Micrograph of the optical fiber inside a fused silica capillary. An α-alumina bead is placed above the fiber tip to prevent light from attenuating from above. Picture provided was courtesy of N.J. Degenstein μm

114 A Flow Rate (mmol/s) Upstream Heat-Shield T Pyrometer T Thermocouple CH 4 O 2 Catalyst Downstream Heat-Shield Position (mm) Temperature ( o C) B 0.7 Upstream Heat-Shield Catalyst Downstream Heat-Shield Flow Rate (mmol/s) CO 2 H 2 CO H 2 O Position (mm) Figure 4-4. Spatial profiles of the catalytic partial oxidation of methane (panels A & B) at C/O = 1.1, 2 SLPM (GHSV ~ 5x10 4 h 1 ), and 1.1 atm. Positions between 0 and 10.8 mm correspond to the upstream heat shield, comprised of an 80 PPI α-alumina foam. Positions between 10.8 and 20.4 mm position correspond to the catalyst, comprised of ~5 wt% Rh on ~5 wt% γ -alumina wash-coat, 80 PPI α-alumina foam support mm+ positions correspond to the downstream heat shield, comprised of an 80 PPI α-alumina foam. Molar flow rates of the reactants, O 2 ( ) and CH 4 ( ), are presented in panel A, as well as temperature measurements from a pyrometer ( ) and thermocouple ( ). Molar flow rates of products, H 2 ( ), H 2 O ( ), CO ( ), and CO 2 ( ), are presented in panel B. 98

115 A 0.5 Upstream Heat-Shield Catalyst Downstream Heat-Shield 1200 Flow Rate (mmol/s) T Pyrometer T Thermocouple O C H Position (mm) Temperature ( o C) B 0.5 Upstream Heat-Shield Catalyst Downstream Heat-Shield Flow Rate (mmol/s) CO H 2 H 2 O CO Position (mm) Figure 4-5. Spatial profiles of the catalytic partial oxidation of benzene (panels A & B) at C/O = 1.1, 2 SLPM (GHSV ~ 5x10 4 h 1 ), and 1.1 atm. Positions between 0 and 10.8 mm correspond to the upstream heat shield, comprised of an 80 PPI α-alumina foam. Positions between 10.8 and 20.4 mm position correspond to the catalyst, comprised of ~5 wt% Rh on ~5 wt% γ -alumina wash-coat, 80 PPI α-alumina foam support mm+ positions correspond to the downstream heat shield, comprised of an 80 PPI α-alumina foam. Molar flow rates of the reactants, O 2 ( ) and C 6 H 6 ( ) are presented in panel A, as well as temperature measurements from a pyrometer ( ) and thermocouple ( ). Molar flow rates of products, H 2 ( ), H 2 O ( ), CO ( ), and CO 2 ( ), are presented in panel B. 99

116 Chapter 5: A kinetic model for the catalytic partial oxidation of benzene and methane on rhodium 5.1 Introduction The oxidation of aromatic fuels is fundamentally and practically important in fuel reforming and pollution abatement, since most practical hydrocarbon fuel blends or chemical feedstocks consist of large amounts of aromatics [1]. Also, aromatic compounds are known to be harmful to the environment, and the emission of these species from a number of oxidation systems is a significant concern. Several studies of the catalytic oxidation of aromatic compounds have been attempted, including some kinetic studies. Most kinetic studies have been carried out at low concentration of the organic compound (less than 5000 ppm), although concentrations are often much greater in industrial reactors. Ordonez et al. conducted catalytic oxidation of benzene and toluene over platinum catalysts and fitted their data to Mars-VanKrevelen kinetic expressions for the aromatic fuels [2]. However, these models only predicted the consumption of the fuel, not the evolution of products. Similarly, Patterson et al. studied the catalytic oxidation of benzene and toluene over platinum, palladium, and rhodium, without providing any kinetic models [3-4]. Several other studies have been made on various precious and transition metal catalysts; however, the evolution of products has not been modeled [5]. Sibener et al. also attempted to develop a kinetic model to capture the experimentally observed product evolution in benzene oxidation on Rh single crystals under ultra-high vacuum conditions [6]. The study assumed that the decomposition of benzene and oxygen on rhodium is quite rapid and the kinetics of CO and CO 2 production are dominated by reactions of surface carbon and oxygen. The model further assumed a rate-limiting step of the oxidation of surface carbon to surface carbon monoxide, followed by desorption of surface carbon monoxide or further oxidation of surface carbon monoxide to surface carbon dioxide. These assumptions provided appropriate fits to their data. However, these assumptions did not provide pathways for the evolution of H 2 and H 2 O or were not validated for more practical conditions, such as atmospheric pressure and high species fluxes to the catalyst. In the previous chapter, experimental spatial profiles of species and temperature were revealed for benzene and methane partial oxidation in air at high temperature and atmospheric conditions. These profiles showed the relative importance of homogeneous and heterogeneous chemistry that occur within these catalytic partial oxidation reactors. In this chapter, unlike the previously described models, a detailed catalytic partial oxidation model developed for methane 100

117 will be extended to involve the partial oxidation of benzene. This model will be compared to the spatial profiles presented in the previous chapter. 5.2 Model Development Heterogeneous mechanism The surface chemistry reaction mechanism utilized in the reactor model discussed in this chapter is shown in Table 1. The mechanism adds 14 addition reversible reactions, with 6 additional surface species, to a previously developed methane oxidation model [7], which contains 38 reversible reactions with 11 surface species. The mean field approximation, assuming that the adsorbates are randomly distributed on a uniform surface, was applied [8-9]. The state of the catalytic surface is described by the temperature T and a set of surface coverage θ i, both depending on the macroscopic position in the reactor and averaging over microscopic local fluctuations. Table 1 also displays the mechanism parameters in terms of pre-exponentials, activation energies, and sticking coefficients. The forward reaction rate coefficient, k fi, for any given species i is determined through an Arrhenius rate expression: Ns β E μ ε θ = i ai ji ji j k θ fi A it exp j exp (5-1) RT j= 1 RT where A i is the pre-exponential term, βi is the temperature exponent, E ai is the activation energy, θ j is the surface coverage of species j, and μ ij, and ε ij describe the dependence of the rate coefficients on the surface coverage of species j. When compared to the exponential temperature dependence associated with the activation energy, the effect of the pre-exponential temperature dependence β on the overall rate constant is negligible. Thus, in the mechanism presented, the pre-exponential temperature dependence was set to 0 (β=0). For adsorption reactions, the Arrhenius rate expression parameters are replaced by sticking coefficients, which are converted to forward reaction rates through the following relationship based on kinetic theory: k ads,i = si RT m Γ 2πM, (5-2) 101 i

118 where s i is the sticking coefficient (the statistical probability that a species will adsorb when it comes in contact with a surface), Γ is the surface site density, m is the number of sites occupied by the adsorbing species, and M i is the molar mass of species i. Implicit in the sticking coefficient description is an assumption that the sticking coefficient is relatively small (s i <<1) [9]. In this case, the molecular motion of gas molecules in the vicinity of the solid surface is random and the collision frequency of gas-phase species with the surface is not affected by the surface reaction. However, when the sticking coefficient is large (s i ~1), then the molecular velocity distribution becomes non-maxwellian. Species whose random motion causes a non-maxwellian velocity distribution at the molecular scale that alters the net species flux near the surface. Motz and Wise analyzed this situation and provided a correction factor that modifies Equation 5-2 with the following: k ads,i s 1 RT i =. (5-3) m si Γ 2πMi 1 2 The details of the mechanistic steps are discussed in the following sections Catalytic methane oxidation mechanism A 38-step surface chemistry mechanism for methane oxidation on Rh was used to capture the dynamics of the C 1 decomposition and oxidation steps for both benzene and methane [7]. This model has been validated against integral steady-state and transient experimental data. The mechanism in its current form is an improved version of the original mechanism for high temperature oxidation [10-11]. The mechanism was revised to include steam reforming and water-gas shift, along with CO 2 re-adsorption, which agreed well with integral steady-state data [8]. Coverage-dependent desorption energies for CO and O 2 were later included to extend the model to light-off conditions [7]. Intrinsic catalyst activity is included through the assumed site density for the Rh surface, which was 2.72 x 10-9 mol/cm 2 for the mechanism Extension of methane partial oxidation mechanism to benzene Previous work has shown that benzene decomposes on Rh(111) at temperatures as low as 400 K to form CH and C 2 H species [12-13]. Since the current study is undertaken at temperatures above 1000 K, this decomposition process is expected to be rapid relative to the adsorption rate of benzene. In addition, temperature programmed desorption (TPD) studies have 102

119 shown that hydrogen continuously desorbs between 500 and 700 K, indicating that the carbon is dehydrogenated in the temperature range of the current study [12]. Furthermore, oxygen (O 2 ) has been shown to dissociatively chemisorb to adsorbed oxygen (O(s)) on Rh (111) single crystals [14]. From these experimental observations, the overall surface reaction mechanism appears to involve the dissociative adsorption of oxygen to adsorbed oxygen (O(s)) and the adsorption of benzene followed by a rapid series of reaction steps which break down the benzene reagent to adsorbed carbon (C(s)) and hydrogen (H(s)). C(s), H(s), and O(s) are primarily expected to be the surface adsorbents formed from these rapid decomposition steps. Once formed, these adsorbents further react to produce OH, H 2, H 2 O, CO, and CO 2, which is captured in the methane partial oxidation mechanism. To capture this generalized adsorption-decomposition scheme described for benzene, several adsorption, desorption, and decomposition steps were added to the methane mechanism. The decomposition pathways were identified from previously published surface science experiments and theoretical calculations on single crystals of Rh (111) and (100). The adsorption and desorption rates were obtained from previously published surface science experiments, while the surface reaction activation energies and pre-exponentials were estimated from published first principle modeling studies, generalized correlations, and transition state theory. None of these parameters were adjusted to fit to the experimental profile data. The identification and estimation of the reaction kinetics and pathways will be described in detail within the following sections Benzene adsorption, desorption, and decomposition The adsorption and desorption of benzene have been experimentally studied on several single crystals of Rh under ultra-high vacuum conditions. Somorjai et al, observed through AES (auger electron spectroscopy), TPD, and high resolution electron energy loss spectroscopy (HREELS) that benzene chemisorbs molecularly on Rh (111) below 400 K and determined the sticking coefficient for benzene to be approximately unity at low coverage [13]. This value was assigned as the sticking coefficient of benzene in this model. Benzene was also assumed to adsorb onto three Rh sites. The assignment of three sites originates from dynamic low energy electron diffraction (LEED) calculations and angle-dependent HREELS studies that indicate chemisorbed benzene, when co-adsorbed with an equal amount of CO, on Rh(111) bonds in a three fold hollow site [12]: C ,C 6 H6 H + 3Rh(s) C H (s); S 1.0. (5-4) 6 = 103

120 Furthermore, the enthalpy of adsorption was determined to be a function of surface coverage on several transition metals. Through TPD, the adsorption enthalpy of benzene on Rh(111) was determined to be 110 KJ/mol at low coverage [12]. Through single crystal adsorption calorimetry (SCAC), adsorption enthalpy of benzene on single crystals has been shown to significantly decrease as the coverage increases. At a high coverage on Pt(111), the heat of adsorption is ~1/3 the enthalpy of adsorption at a low coverage [15]. Assuming that benzene decomposes rapidly to surface carbon, the change in enthalpy was assumed to be a linear function of carbon coverage instead of adsorbed benzene. This coverage dependence was incorporated into the enthalpy of adsorption for benzene, assuming at full coverage the heat of adsorption is 1/3 the heat of adsorption of low coverage. Furthermore, in the model, the enthalpy of adsorption was assumed to be equal to the activation energy of desorption: KJ C6H6(s) C6H6 + Rh(s); Ea = θc. (5-5) mol Through HREELS, TPD, and LEED, Koel and Somorjai have shown that when benzene decomposes on Rh(111) at 400 K, the decomposition fragments (CH and C 2 H) are the same as for decomposed acetylene at this temperature [12]. With these results, they proposed that benzene decomposes at 400 K by C-C bond scission to give three acetylenes as intermediates, which immediately decompose to CH and C 2 H species. Further support for this proposal that acetylene is an intermediate in benzene decomposition over Rh surfaces comes from LEED, Raman Spectroscopy, H/D exchange studies, and comparison to other metal/hydrocarbon systems. Additionally, Raman spectroscopy studies of Rh/Al 2 O 3 show that at room temperature chemisorbed benzene can be induced to decompose to acetylene by introducing a strong π acceptor ligand such as CO or NO. To account for the benzene decomposition to acetylene within the model, the reversible reactions were approximated using the linear relationship between activation energies and enthalpy changes for dissociation reactions at surfaces for Class II Hydrocarbon Cracking reaction, described by Michaelides et al [16]: E dissociation a + = 0.97ΔH 1.69, (5-6) where ΔH is the enthalpy of reaction. ΔH was determined using the enthalpy of reaction of the gas phase enthalpies of formation for benzene and acetylene at standard temperature adjusted with the low coverage enthalpy of adsorption of each species. 104

121 0 0 ( ΔHf, + ΔHads ) ( ΔHf + ΔHads ) C H C6 6 Δ H 3 (5-7) H 2 2 The activation energy for association was determined with the following relationship: E association a dissociation a = E ΔH. (5-8) All pre-exponentials in the reaction rate constants were approximated using transition state theory, assuming that the temperature dependence was negligible (1 x s -1 ) [17] Acetylene adsorption, desorption, and decomposition The adsorption and desorption of acetylene have been experimentally studied on several single crystals of Rh(111) and Rh(100) under ultra-high vacuum conditions [18-19]. By applying a molecular beam method, King et al, determined the sticking coefficient for acetylene on Rh (100) to be approximately 0.8 at low coverage [18]. This value was assigned as the sticking coefficient of acetylene in this model. Also, acetylene is assumed to adsorb onto one Rh site: C2 H2 + Rh(s) C2H2(s); s0,c 2 H = 0.8. (5-9) 2 Furthermore, through single crystal adsorption calorimetry (SCAC), the enthalpy of adsorption of acetylene on Rh(100) was determined to have a strong dependence on coverage, resulting in 210 KJ/mol at low coverage to 80 KJ/mol for full coverage [18]. Similar to benzene, acetylene has been shown to rapidly decompose to surface carbon at high temperatures. Due to this rapid decomposition, the change in enthalpy was assumed to be a linear function of carbon coverage instead of adsorbed acetylene. In the model, the enthalpy of adsorption was assumed to be equal to the activation energy of desorption, and this coverage dependence was incorporated into the activation energy for acetylene desorption: KJ C2H2(s) C2H2 + Rh(s); Ea = θc. (5-10) mol Once adsorbed, the acetylene decomposition was assumed to follow the pathway outlined by Nieskens et al [20]. By combining the density functional theory (DFT) based theoretical results with several experimental results (secondary ion mass spectrometry, TPD, LEED), Nieskens et al. proposed the likely decomposition pathways for acetylene, along with the activation energy for each reversible decomposition step. These steps and energies are outlined in Table

122 The first step is either the C-C bond breaking, leading to two C-H fragments, or the breaking of the C-H bond, leading to the ethynyl species. DFT results show that breaking of the C-H bond has lower activation energy compared to breaking of the C-C bond. However, the reaction energy change is in favor of the C-C bond breaking. These results suggest that C-H bond breaking is kinetically preferred, while C-C bond breaking is thermodynamically preferred. From experimental evidence, the first step in the decomposition process is the breaking of one of the C- H bonds to form the ethynyl species. Both experimental and theoretical results indicate that the most likely second step is the breaking of the C-C bond, requiring lower activation energy than the breaking of the second C-H bond. Furthermore, the energy for the second C-C bond breaking is now highly favored over that of the C-H bond breaking. After the second step, the surface is covered with a mixture of methylidyne and carbon species. The surface steps for these remaining species are addressed in the previously developed methane mechanism. Similar to the benzene reaction pathways, all pre-exponentials in the reaction rate constants for acetylene steps were approximated using transition state theory, assuming that the temperature dependence was negligible (1 x s -1 ) [17] Homogeneous mechanism The GRI-Mechanism 3.0 with 325 reversible reactions involving 53 species was used to model the homogeneous chemistry within the methane partial oxidation reactor [21]. This mechanism has been widely used and experimentally verified to capture the homogeneous oxidation chemistry of C 1 to C 3 alkanes. At experimental temperatures and pressures, the homogeneous chemistry for methane oxidation has been show to generate <10% of oxidation products, and often, the dynamics of catalytic partial oxidation of methane can be predicted well without including a homogeneous oxidation pathway [22-26]. However, with larger hydrocarbons and aromatic compounds, such as benzene, homogeneous chemistry becomes more relevant under experimental conditions. Thus, accounting for the homogeneous chemistry of benzene should be crucial to understanding the dynamics that occur within these catalytic partial oxidation reactors. For benzene, several homogeneous oxidation mechanisms were initially applied to predict the homogeneous chemistry within the reactor, each yielding similar results [27-30]. The Alzueta mechanism was used to predict the homogeneous chemistry discussed in the following sections [28]. 106

123 The Alzueta mechanism consists of 108 species and over 500 reversible elementary reactions. For most species, the reaction pathways and rate coefficients were mostly obtained form literature data and compilations. This reaction mechanism was validated against experimental data, including plug flow and mixed flow reactors, and was capable of predicting the concentration profiles of major species during benzene oxidation under a variety of different reactor conditions (900 to 1450 K, <150 ms, lean and stoichiometric combustion ratios) Reactor models As a first approximation, the system was modeled as a plug-flow reactor. The plug-flow assumption is an extremely useful tool for building a kinetic mechanism considering the tradeoff between the extremely low computing time (seconds) and the accuracy of the model [22,31]. A computational study of methane surface oxidation in short contact time reactors showed that the predicted differences between plug-flow and a complete 2D model are limited to entrance effects for honeycomb monoliths [8]. Thus, the validity of the assumptions of infinitely fast heat and mass transfer is critical only in a short fraction of the modeled reactor length. Furthermore, the ideal Reynolds number (<100) and the very high length-to-diameter ratio (~40 for 80 pore per linear inch (PPI) foam monolith, d c ~0.25mm, and L~10mm) suggest that axial dispersion is extremely low. In reality, foam monoliths unlike honeycomb monoliths contain pore tortuosity that enhances turbulence and radial transport which limit boundary layer development [22,31]. These considerations led to the conclusion that the plug flow model was valid for the initial estimates of the mechanism. The foam monolith reactor was described through an idealized single pore model (laminar flow) since surface area per volume is the major parameter in determining reactor performance. Final tuning of the surface mechanism should be carried out with a model that takes into account a detailed flow description, such as a complete 2D flow model. However, the foam monolith geometry is ill-defined for a rigorous computational fluid dynamics study. Numerical simulations were performed using Chemkin PLUG code [32] for a single pore of an 80 PPI, 92 wt% Al 2 O 3, 8% wt% SiO 2, wash-coated foam monolith (d c =0. 25 mm), neglecting tortuosity. The pore size of 0.25 mm was assessed by examining the foam support through SEM, as shown in Figure 5-1. The gas hourly space velocity was 5 x 10 4 h -1, corresponding to a total flow rate of ~2 liters per minute at 20 o C and 1 atm for an entire monolith diameter of 17 mm. The inlet pressure was 1.1 atm, dilution was air stoichiometry (N 2 /O 2 or Ar/O 2 = 3.76), and the C/O ratio was 1.1 for both fuels. Thermodynamic data for the homogeneous species was provided 107

124 through the Chemkin 3.7 library, along with values provided from published homogeneous mechanisms [28, 32] Energy balance Temperature measurements indicate that axial heat transfer through the conductance of the foam support is important. To model the temperature profile appropriately, a 1D or 2D model that includes the axial conductance within the support is required. However, the inclusion of this axial conductance increases the computational difficulty from seconds to hours, and for this experiment, the gas-phase and surface chemistry in the reactor are the primary focus. Therefore, if temperatures throughout the reactor bed are known, the energy balance that incorporates axial conductance is not required to determine the kinetic rate parameters. Through a new experimental technique developed by Horn et al. and Degenstein, the surface temperatures can be measured with 300 μm resolution [33-37]. This surface temperature profile was placed into the plug-flow reactor model, which eliminated the need for the inclusion of the axial conductance term. In the plug-flow model, no heat transfer limitations exist between the gas and the surface, thus, the gas-phase temperature is equal to the surface temperature. The experimentally determined surface temperatures allow the gas-phase and surface kinetics to be modeled within a 1D boundary layer model that excludes axial conductance within the energy balance. Intrinsic to the plug-flow assumption, the gases are thermally accommodated (isothermal) with the catalyst surface at each axial position. The experimental surface measurements used in this simulation are obtained from the pyrometer and are displayed as T pyrometer in Figure 5-2, along with the experimentally observed species profiles Modeling various sections of the reactor The chemistry in the upstream and downstream heat-shields was modeled by applying the previously described homogeneous mechanisms, while the chemistry in the catalyst section incorporated both the homogeneous mechanisms and the developed surface mechanism. Inlet species for the upstream heat-shield section were set to the experimentally measured species at the upstream heat-shield front-face (0 mm); inlet species for the catalyst section were set to the experimentally measured species at the catalyst front-face (~11 mm); and the inlet species for the downstream heat-shield were set to the experimentally measured species at the downstream heat shield front-face (~21 mm). The inlet velocity for each section was determined by applying the continuity equation based on the initial inlet molar flow rate and temperature. 108

125 Additionally, to access the contribution of homogeneous chemistry within the reactor, all the reactor sections (heat-shields and catalyst) were modeled with only the homogeneous mechanism. 5.3 Results Model trends Figures 5-3 and 5-4 display the predicted profiles (solid lines) for both methane and benzene throughout the upstream heat-shield, catalyst, and downstream heat-shield. Several differences should be noted from these profiles. First, as discussed previously, homogeneous chemistry is predicted within the upstream heat-shield for benzene, while no homogeneous chemistry is predicted for methane. For benzene, water and CO are the main products predicted in the upstream heat-shield. This upstream homogeneous chemistry also alters the hydrocarbon fuel to oxygen ratio entering the catalyst region. Within the catalyst section, CO 2 and water production is not predicted for benzene. However, from methane, water and CO 2 production occurs within the beginning 0.25 mm of the catalyst section. The produced water is then reacted with methane to produce CO and H 2 (stoichiometric steam reforming reaction). In contrast, H 2 and CO production is primarily predicted with the benzene model. In the downstream heat-shield, the products remain constant for both benzene and methane. Once formed, CO 2 remains constant throughout the remaining portion of the reactor. Furthermore, the methane model predicts oxygen consumption within the initial 0.25 mm of the catalyst section. However, the oxidation zone for benzene partial oxidation is >10 times longer than that of methane. Predicted catalyst surface coverage for methane and benzene are displayed in Figure 5-5. For methane, the majority of the surface is vacant (Rh(s) ~0.85) with some oxygen coverage (O(s) ~0.15) in the initial mm, while carbon (C(s)) and carbon monoxide (CO(s)) coverage increases and surface vacancies (Rh(s)) decrease when the gas-phase oxygen is consumed within the reactor. In contrast, for benzene, high carbon coverage (C(s)) is predicted for the entire length of the catalyst with some vacant catalytic sites (Rh(s) ~0.20 to ~0.10) within the initial 4 mm. 109

126 Figure 5-6 shows the profile without the inclusion of the surface mechanism. The product selectivities favor CO and H 2 O, while CO 2 and H 2 are also predicted. The homogeneously predicted fuel conversion is also much lower than the heterogeneously predicted fuel conversion Comparison with experimental data: methane Figure 5-3 displays both experimental and predicted spatial profiles of the catalytic partial oxidation of methane at reaction conditions. While no oxygen or methane conversion is predicted, some oxygen and methane appear converted within the millimeter of the heat-shield. Inside the catalyst region, the predicted oxygen consumption is much faster than experimentally observed. In fact, the oxygen is consumed within the initial 0.5 mm in the model, while experimentally it is consumed within 2 mm. With a smaller oxidation zone, water, CO 2, H 2, and CO production is also predicted to occur much sooner than experimentally observed. The model also under predicts the secondary steam reforming that is experimentally observed. Experimentally, the steam reforming continues through the entire length of the catalyst section, downstream of the oxidation zone, while the model predicts the steam reforming rates to significantly slow approximately 5 mm into the catalyst section. Upon exiting the catalyst, the experimental profile displays a change in H 2. However, after close examination, no other species are changing; thus, the H 2 change is suspected to be an experimental artifact. Experimentally observed constant molar flow rates downstream of the catalyst correspond well with the predicted model Comparison with experimental data: benzene Figure 5-4 displays both experimental and predicted spatial profiles of the catalytic partial oxidation of benzene at reaction conditions. Within the heat-shield, oxygen and benzene conversion is predicted and experimentally observed. However, the experimental consumption of benzene and oxygen is observed throughout the entire length of the heat-shield, while in the model, the consumption of reactants is predicted only near the front-face of the catalyst (~8 mm). In the upstream heat-shield, the homogeneous model predicts similar ratios of H 2 O to H 2 that were experimentally observed, however, the ratio of CO to CO 2 was much lower than experimentally observed. Inside the catalyst region, the predicted oxygen consumption is much faster than experimentally observed. The oxygen is consumed within the initial 4 mm for both the model and experiment. However, 50% oxygen conversion occurs in different locations. The model predicts 110

127 that 50% of the oxygen is consumed in the initial 1mm, while experimentally 50% of the oxygen is converted in the initial 2 mm. With higher initial oxygen consumption rates, H 2 and CO generation is also predicted much sooner than experimentally observed. In the model catalytic section, CO 2 remains relatively constant, which agrees well with the experimental results. However, the model over-predicts water consumption. Water consumption is predicted within the oxidation zone, while experimentally, water is consumed downstream of the oxidation zone. Predicted constant molar flow rates exiting the catalyst region correspond well with the experimental results. 5.4 Discussion Experimental uncertainty As with any experiment, species and temperature measurements incorporate experimental uncertainties. For example, a measure for experimental accuracy is the atom balance error at each axial location. At steady-state, species accumulation does not occur and atom balance errors should be zero. As mentioned in the previous chapter, the elemental carbon and hydrogen balances (Elements out Elements in ) typically closed with +/-5% and +/-5% for methane and +/- 5% and +/-9% for benzene. Also, the exact micrometer positioning and area/position that is measured with the pyrometer contain some uncertainty +/- 0.5 mm. Since the temperature readings from the pyrometer are used to determine the kinetics within the model, any pyrometer uncertainty would ultimately affect the predictions within the model. Reproducibility of the experimental profiles has been verified by performing independent profile measurements on different catalysts and with different reactor set-ups. Axial measurements (e.g. length of the oxidation zone) can be reproduced to within 1 mm, while species concentrations at the reactor outlet were reproducible to ~5% [22,33-37]. Pockets or blocked pores along the capillary channels or even back mixing effects caused by the pore structure around the channel can lead to features in the profile that can never be captured by the model, which assumes a well defined or idealized flow profile. Therefore, when comparing experimental data with model predictions, this experimental uncertainty should be considered to determine the validity of the proposed model. The model should not be fitted to this experimental data; instead the experimental data should be used as a general guideline in the trends observed within the model. Some discrepancies between the general experimental trends and model predictions will be discussed in the upcoming sections. 111

128 5.4.2 Upstream heat-shield: axial dispersion Even though the homogeneous model does not predict methane conversion in the upstream heat-shield, some methane and oxygen experimentally appear to be converted in the upstream heat shield, ~1 mm from the catalyst front-face. This observed conversion is likely attributed to the axial diffusion of species within the reactor. The reactor model used neglected axial dispersion, while in reality, some axial diffusion exists. Previous 2D modeling accounting for axial dispersion in these reactors has shown that the slight observed fuel conversion near the frontface of the catalyst can be attributed to axial diffusion. Even though axial dispersion is predicted and experimentally observed, the dispersion is limited to the ~1 mm upstream from the catalyst region for methane. For benzene, axial dispersion may also exist. However, this dispersion should be limited to similar distances as observed with methane. Additionally, in the benzene catalyst zone, H 2 and CO are the predominant species. The lowest molecular weight compound H 2, which is present in large 2 concentrations, should axially diffuse at the fastest rate ( 1, where D Dij α M ij is the binary diffusion i coefficient of species i into j, M is the molecular weight of species i). Even though H 2 should be the fastest species to axial diffuse into the upstream zone, the levels of CO, CO 2, and H 2 O are much greater than the H 2 species, indicating that axial diffusion cannot completely account for the high levels of these compounds upstream of the catalyst zone. Therefore, these results suggest that the fuel and oxygen conversion is most likely caused by oxidation reactions and not axial diffusion Upstream heat-shield: homogeneous chemistry As homogenous chemistry is important for the catalytic oxidation of benzene, artifacts in the experimental sampling technique may over-predict the conversion of fuel through homogeneous pathways. As mentioned in the previous chapter, the sampling technique withdraws a sample of gas through a capillary placed along the z-axis of the cylindrical catalyst. Withdrawing the sample through the capillary, the reactants experience a longer residence time than occurring within the reactor. This longer residence time enables more homogeneous chemistry than is observed. Withdrawing the sample through upstream flow should also cool the reactants to pre-heat conditions quite rapidly, minimizing the homogeneous chemistry in the sampling capillary. 112

129 As mentioned previously, several homogeneous oxidation models exist for benzene. While most of the models display similar results, these models were not validated for the fuel rich conditions used in this experiment. In addition, the alumina support may alter radical formation, propagation, and termination, potentially changing the homogeneous pathways and rates of benzene oxidation observed within the reactor Comparing benzene and methane oxidation zones The difference in the oxidation zone appears to be related to the species coverage. For methane, Rh(s) and O(s) coverage are dominant in the oxidation zone, while for benzene, C(s) and Rh(s) coverage are dominant. The difference in coverage is attributed to the difference in sticking coefficients of methane and benzene in relation to oxygen. In the model, the sticking coefficient of oxygen is greater than methane, while the sticking coefficient for benzene is much larger than oxygen. A high oxygen coefficient results in a surface covered with oxygen molecules. When a methane molecule adsorbs onto a predominately oxygen covered surface, the high concentration of surface oxygen promotes rapid decomposition of methane and production of synthesis gas and combustion products. These products easily desorb at reaction temperatures, enabling the oxidation process with a large area of un-occupied sites. In contrast, a high benzene sticking coefficient followed by rapid decomposition causes the surface to be rapidly covered with non-volatile C(s). This high C(s) coverage reduces the number of available sites for oxygen to adsorb. With fewer sites available, the oxidation reactions are significantly slowed, resulting in a longer oxidation zone Secondary steam reforming High C(s) and CO(s) coverage predicted downstream of the oxidation zone appear to reduce secondary steam reforming. In both the methane and benzene models, the coverage of C(s) and CO(s) was very high downstream of the oxidation zone. With the high C(s) and CO(s) coverage, the number of sites available for water to adsorb and react is greatly impeded, resulting in reduced steam reforming. The experimental data display more steam forming than predicted downstream of the oxidation zone. Potentially, the model over predicts the C(s) and CO(s) coverage, which could explain the under-prediction of steam reforming Mass transfer limitations Previous profile experiments were performed with the partial oxidation of methane partial on Rh-coated 80 PPI ceramic foam supports where the C/O ratio was varied [36]. From these experiments, the zone in which oxygen was consumed remained constant, indicating this zone 113

130 was mass transfer limited on Rh. These experiments, along with previous models of catalytic partial oxidation of methane suggest that accounting for mass transfer limitations (O 2 for methane partial oxidation) from the bulk to the catalyst surface is important in modeling the oxidation zone [22,33]. However, intrinsic to the plug-flow model, mass transfer limitations from bulk phase to the surface inside the porous foam are neglected. Therefore, the model most likely will predict faster transfer of the limiting reactant to the bulk surface, resulting in a shorter oxidation zone for methane partial oxidation. While mass transfer limitations with respect to oxygen have been shown previously for methane partial oxidation, further profile experiments with benzene will be required to determine if the oxidation zone is kinetically or mass transfer limited in oxygen or benzene. However, if mass transfer is important in the oxidation zone, the diffusion rate of O 2 is greater than the diffusion rate 2 of C 6 H 6 from the bulk to the catalyst surface ( 1 ). From the model, high carbon coverage Dij α M i (C(s)) appears to reduce the oxidation rates, resulting in a long oxidation zone. However, by including mass transfer in the model, the higher diffusion rates of O 2 relative to C 6 H 6 may reduce the high carbon surface coverage (C(s)), resulting in increased oxidation rates and a shorter predicted oxidation zone. 1 From the methane model, when all the O 2 has been consumed, the surface C(s) and CO(s) coverage increases relatively fast, greatly reducing available sites for additional reactions to occur. Since mass transfer limitations likely require longer residence times for complete conversion of O 2, the presence of O(s) further into the catalyst would reduce the deactivating C(s) and CO(s) coverage and allow more time for H 2 O to adsorb and provide additional O(s) for the oxidation of C(s). Since the model neglects mass transfer limitations, the underestimation of steam reforming in the latter portion of the catalyst could be explained by the shorter oxidation zone Catalytic surface area & pore size A key assumption in the developed models is the active catalytic (active) surface area is represented by the geometric surface area. By examining the catalyst surface with Back- Scattered Electron (BSE) and Scanning Electron Microscopy (SEM), displayed in Figure 5-7, the dense catalyst particles (bright objects in the BSE image) are not distributed along the entire support surface and actually only occupy a fraction of the catalyst wall. This variation in catalytic surface may explain the under-estimation of the oxidation rates observed in both methane and benzene partial oxidation. 114

131 Additionally, the straight pore model applied in this chapter neglected porosity. Similar foams have been measured with porosities ranging from 1.5 to 5 [22]. High porosity would affect the geometric surface to volume ratio, and with a higher porosity, reactant flow could encounter different temperature gradients than assigned within the model. Previous modeling in 2D systems, which assign a higher porous block, has displayed better agreement with experimentally observed data [22]. Furthermore, the reactor model required the specification of a single pore diameter. In this work, a pore diameter of 0.5 mm was used, which is the mean pore diameter from experimental visualization of an 80 PPI foam. However, a foam possesses a large range of pore diameters, which are typically normally distributed for moderate PPI foams (10-65 PPI) [22]. Being restricted to a single channel diameter in these simulations does not account for this distribution. Since radial mass transfer and heat transfer are strongly dependence on this pore diameter specification, an assumption of a single representative pore size within the catalyst may greatly affect these transfer rates, resulting in some of the observed discrepancies with experimental data Thermodynamic inconsistencies The Deutschmann Model, upon which the benzene model is built, displays violations of equilibrium at high temperatures and at low C/O ratios (C/O < 1.0). Specifically, the Deutschmann model overpredicts the extent of water-gas shift at all temperatures, provided sufficient time/length for reaction, which indicates this mechanism is not thermodynamically consistent [22,33]. However, this violation of water-gas shift equilibrium is often observed at C/O <1.0. At C/O > 1.0, where product compositions begin to diverge from equilibrium values, the model appears to capture the species evolution in both methane partial oxidation and benzene oxidation. Since the Deutschmann model exhibits thermodynamic inconsistencies, caution should be issued for applying this benzene partial oxidation model in un-validated regions, especially C/O < Benzene & acetylene decomposition pathway The decomposition pathway was assumed to follow pathways experimental observed for acetylene and benzene under ultra high vacuum (UHV) and on Rh single crystals. Even though the model captures the general trends experimentally observed, the model does not prove that this decomposition pathway is actually occurring within the reactor. 115

132 The benzene and acetylene decomposition pathway and stable surface fragments observed from UHV experiments on single crystal Rh surfaces may be significantly different from those under catalytic reaction conditions. Hydrocarbon fragment stability is affected by the surface H atom concentration which is determined by ambient H 2 pressure. Studies in ultrahigh vacuum favor dehydrogenation accompanied by hydrogen desorption, while catalytic hydrocarbon conversion reactions are carried out in high pressure of H 2, increasing the stability of hydrogenated surface fragments. In addition, imperfections in crystal structures may also energetically favor different pathways that have not been included in this model. Also, as previously mentioned, the benzene surface mechanism does not account for reactions with intermediate compounds produced in the gas phase. Most likely these intermediate compounds have similar reactivities to benzene. However, when compared to benzene, the differences in geometry, electronic structure and bond strengths of these intermediate compounds may alter their ability to adsorb and react on the catalyst surface, generating different results. 5.5 Conclusions A model for the catalytic partial oxidation of methane was extended to benzene decomposition by applying reaction pathways previously described in literature. The kinetic heterogeneous model coupled with a previously developed homogeneous oxidation model appears to capture the general trends observed in the profile data. Additional refinements to the model are required to account for intermediate products produced through the homogeneous oxidation pathways. Also, to model properly, the homogeneous mechanism requires refinement to address the differences in reactivity that were observed experimentally and numerically. Finally, to validate and further the model, the application of 2D CFD is required to account for the axial and radial diffusion and heat transfer that occurs within the reactor system. 5.6 References 1. Z.M. Djurisic, A.V Joshi, H. Wang, Second Joint Meeting of the U.S. Sections of the Combustion Institute, Oakland, CA, March 25-28, S. Ordonez, L. Bello, H. Sastre, R. Rosal, F.V. Diez, Applied Catalysis B: Environmental 38 (2002)

133 3. M.J. Patterson, D.E. Angove, N.W. Cant, Applied Catalysis B: Environmental 35 (2001) M.J. Patterson, D. E. Angove, N. W. Cant, Applied Catalysis B: Environmental 26 (2000) S.K. Gangwal, M.E. Mullins, J.J. Spivey, O.R. Caffrey, B.A. Tichenor, Applied Catalysis 36(1-2) (1998) M.E. Viste, K.D. Gibson, S.J. Sibener, Journal of Catalysis 191 (2000) R. Sweidernoch, S. Tischer, C. Correa, O. Deutschmann, Chemical Engineering Science, 58 (2003) O. Deutschmann, R. Sweidernoch, L.I. Maier, D. Chatterjee, Natural Gas Conversion VI, Studies in Surface Science and Catalysis 136 (2001) R.J. Kee, M.E. Coltrin, P. Glaborg, Chemically Reacting Flow Theory and Practice, Wiley- Interscience, Hoboken, N.J., D. A. Hickman, L. D. Schmidt, Science 259 (1993) D.A. Hickman, L.D. Schmidt, AIChE Journal, 39(7) (1993) B.E. Koel, J.E. Crowell, B.E. Bent, C.C. Mate, G.A. Somorjai, Journal of Physical Chemistry 90 (1986) B.E. Koel, J.E. Crowell, C.M. Mate, G.A. Somorjai, Journal of Physical Chemistry 88 (1984) C.T. Campbell, J.M. White, Journal of Catalysis 54 (1978) H. Ihm, H. M. Ajo, J.M. Gottfried, P. Bera, C.T. Campbell, Journal of Physical Chemistry B 108 (2004) A. Michaelides, Z-P Liu, C.J. Zhang, A. Alavi, D.A. King, P. Hu, Journal of the American Chemical Society 125 (2003) J.A. Dumesic, D.F. Rudd, L.M. Aparicio, J.E. Rekoske, A.A Trevino, The Microkinetics of heterogeneous Catalysis, American Chemical Society, Washington, D.C., R. Kose, W.A. Brown, D.A. King, Chemical Physical Letters, 311 (1999) C.M. Mate, C.T. Kao, B.E. Bent, G.A. Somorjai, Surface Science 197(1-2) (1988) D.L.S. Nieskens, F. Ample-Navarro, M.M.M. Jansen, D. Curulla-Ferre, J. Ricart, H. Niemantsverdriet, Chemical Physics Physical Chemistry 7 (2006) G.P. Smith, D.M. Golden, M. Frenklach, N.W. Moriarty, B. Eiteneer, M. Goldenberg, C.T. Bowman, R.K. Hanson, S. Song, W.C. Gardiner, Jr., V.V. Lissianski, and Z. Qin, GRI- Mech 3.0, K.A. Williams, R. Horn, L.D. Schmidt, AIChE Journal 53(8) (2007) O. Deutschmann, L.D. Schmidt, AIChE Journal 44 (1998) C.T. Goralski, R.P. O Conner, L.D. Schmidt, Chemical Engineering Science 55 (2000)

134 25. G. Veser, J. Fraunhammer, Chemical Engineering Science 55 (2000) J.C. Slaa, R.J. Berger, G.B. Martin, Catalysis Letters 99 (2005) Z.M. Djurisic, MS Thesis, University of Delaware, Newark, Delaware, U.S.A., M.U. Alzueta, P. Glaborg, K. Dam-Johansen, International Journal of Chemical Kinetics 32 (2000) I.D. Costa, R. Fournet, F. Billaud, F. Battin-Leclerc, International Journal of Chemical Kinetics 35 (2003) A. Schobel-Ostertag, M. Braun-Unkhoff, C. Wahl, L. Krebs, Combustion and Flame 140 (2005) F. Donsi, K.A. Williams, L.D. Schmidt, Industrial Engineering Chemistry Research 44 (2005) R.J. Kee, F.M. Rupley, J.A. Miller, M.E. Coltrin, J.F. Grear, E. Meeks, H.K. Moffat, A.E. Lutz, G. Dixon-Lewis, M.D. Smooke, J. Warnatz, G.H. Evans, R.S. Larson, R.E. Mitchell, L.R. Petzold, W.C. Reynolds, M. Caracotsios, W.E. Stewart, P. Glaborg, C. Wang, O. Adigun, W.G. Houf, C.P. Chou, S.F. Miller, CHEMKIN Collection Release (2003), Reaction Design Inc., San Diego, CA. 33. R. Horn, K.A. Williams, N.J. Degenstein, L.D. Schmidt, Journal of Catalysis 242 (2006) R. Horn, N.J. Degenstein, K.A. Williams, L.D. Schmidt, Catalysis Letters 110 (3-4) (2006) R. Horn, K.A. Williams, N.J. Degenstein, L.D. Schmidt, Chemical Engineering Science, 62 (2007) R. Horn, K.A. Williams, N.J. Degenstein, A. Bitsch-Larsen, D. Dalle-Nogare, S.A. Tupy, L.D. Schmidt, Journal of Catalysis 249 (2007) N.J. Degenstein, Ph.D. Thesis, University of Minnesota-Twin Cities, Minneapolis, Minnesota, U.S.A.,

135 Table 5-1. Proposed mechanism for the partial oxidation of benzene, an extension of the Deutschmann mechanism for the partial oxidation of methane. (s) represents a surface species. Reactions s o A [cm,mol,s] E a [KJ/mol] Reference Adsorption H 2 + 2Rh(s) 2H(s) 1.0E-02 6 O 2 + 2Rh(s) 2O(s) 1.0E-02 6 CH 4 + Rh(s) CH 4 (s) 8.0E-03 6 H 2 O + Rh(s) H 2 O(s) 1.0E-01 6 CO 2 + Rh(s) CO 2 (s) 1.0E-05 6 CO + Rh(s) CO(s) 5.0E-01 6 C 2 H 2 + Rh(s) C 2 H 2 (s) 8.0E C6H6 + 3Rh(s) C 6 H 6 (s) 1.0E Desorption H(s) + H(s) H 2 3.0E O(s) + O(s) O 2 1.3E θ O 6 H 2 O(s) H 2 O 3.0E CO(s) CO 3.5E θ CO 6 CO 2 (s) CO 2 1.0E CH 4 (s) CH4 1.0E C 2 H 2 (s) C 2 H 2 1.0E θ C 16 C 6 H 6 (s) C 6 H 6 1.0E θ C 12 Surface Reacions H(s) + O(s) OH(s) + Rh(s) 5.0E OH(s) + Rh(s) H(s) + O(s) 3.0E H(s) + OH(s) H 2 O(s) + Rh(s) 3.0E H 2 O(s) + Rh(s) H(s) + OH(s) 5.0E OH(s) + OH(s) H 2 O(s) + O(s) 3.0E H 2 O(s) + O(s) OH(s) + OH(s) 3.0E C(s) + O(s) CO(s) + Rh(s) 3.0E CO(s) + Rh(s) C(s) + O(s) 2.5E CO(s) + O(s) CO 2 (s) + Rh(s) 1.4E CO 2 (s) + Rh(s) CO(s) + O(s) 3.0E CH 4 (s) + Rh(s) CH 3 (s) + H(s) 3.7E CH 3 (s) + H(s) CH 4 (s) + Rh(s) 3.7E CH 3 (s) + Rh(s) CH 2 (s) + H(s) 3.7E CH 2 (s) + H(s) CH 3 (s) + Rh(s) 3.7E CH 2 (s) + Rh(s) CH(s) + H(s) 3.7E CH(s) + H(s) CH 2 (s) + Rh(s) 3.7E CH(s) + Rh(s) C(s) + H(s) 3.7E C(s) + H(s) CH(s) + Rh(s) 3.7E CH 4 (s) + O(s) CH 3 (s) + OH(s) 1.7E CH 3 (s) + OH(s) CH 4 (s) + O(s) 3.7E CH 3 (s) + O(s) CH 2 (s) + OH(s) 3.7E CH 2 (s) + OH(s) CH 3 (s) + O(s) 3.7E CH 2 (s) + O(s) CH(s) + OH(s) 3.7E CH(s) + OH(s) CH 2 (s) + O(s) 3.7E CH(s) + O(s) C(s) + OH(s) 3.7E C(s) + OH(s) CH(s) + O(s) 3.7E C 2 H 2 (s) + Rh(s) CH(s) + CH(s) 3.7E CH(s) + CH(s) C 2 H 2 (s) + Rh(s) 3.7E C 2 H 2 (s) + Rh(s) C 2 H(s) + H(s) 3.7E C 2 H(s) + H(s) C 2 H 2 (s) + Rh(s) 3.7E C 2 H(s) + Rh(s) CH(s) + C(s) 3.7E CH(s) + C(s) C 2 H(s) + Rh(s) 3.7E C 2 H(s) + Rh(s) C 2 (s) + H(s) 3.7E C 2 (s) + H(s) C 2 H(s) + Rh(s) 3.7E C2(s) + Rh(s) C(s) + C(s) 3.7E C(s) + C(s) C 2 (s) + Rh(s) 3.7E C 6 H 6 (s) 3C 2 H 2 (s) 1.0E C 2 H 2 (s) C 6 H 6 (s) 1.4E

136 250μm Figure 5-1. Pore from a ~5 wt% Rh on ~5 wt% γ -Al 2 O 3 washcoat, 80 PPI α-alumina foam support 120

137 A 1000 Upstream Heat-Shield Catalyst Downstream Heat-Shield Flow Rate (mmol/s) T Pyrometer T Thermocouple Position (mm) B 1200 Upstream Heat-Shield Catalyst Downstream Heat-Shield Flow Rate (mmol/s) T Pyrometer T Thermocouple Position (mm) Figure 5-2. Spatial temperature profiles of the catalytic partial oxidation of methane (panels A) and benzene (panel B) at C/O = 1.1, 2 SLPM (GHSV ~ 5x10 4 h 1 ), and 1.1 atm. Positions between 0 and 10.8 mm correspond to the upstream heat shield, comprised of an 80 PPI α- alumina foam. Positions between 10.8 and 20.4 mm position correspond to the catalyst, comprised of ~5 wt% Rh on ~5 wt% γ -Al 2 O 3 washcoat, 80 PPI α-alumina foam support mm+ positions correspond to the downstream heat shield, comprised of an 80 PPI α-alumina foam. Temperature measurements from a pyrometer ( ) and thermocouple ( ) are shown. 121

138 A 0.5 Upstream Heat-Shield B 0.5 Catalyst C 0.14 Downstream Heat-Shield Flow Rate (mmol/s) CH 4 O Position (mm) Flow Rate (mmol/s) CH 4 O Position (mm) Flow Rate (mmol/s) CH 4 O Position (mm) D Flow Rate (mmol/s) Upstream Heat-Shield H 2 CO 2 H O 2 CO Position (mm) E Flow Rate (mmol/s) Catalyst CO H O 2 CO Position (mm) H 2 F Flow Rate (mmol/s) Downstream Heat-shield H 2 CO H O 2 CO Position (mm) Figure 5-3. Experimental and predicted spatial profiles of the catalytic partial oxidation of methane at C/O = 1.1, 2 SLPM (GHSV ~ 5x10 4 h 1 ), and 1.1 atm. Results are displayed for the upstream heat shield region (Panels A and D) applying the homogneous mechanism, for the catalyst region (Panels B and E) applying the homogeneous and heterogeneous mechanisms, and for a portion of the downstream heat shield applying the homogeneous mechanism (Panels C and F). Molar flow rates of the reactants, O 2 ( ) and CH 4 ( ), are presented in panels A through C, while molar flow rates of products, H 2 ( ), H 2 O ( ), CO ( ), and CO 2 ( ), are presented in panels D through F. The model uses the experimentally measured pyrometer temperatures, shown in Figure

139 A 0.30 Upstream Heat-Shield B 0.20 Catalyst C 0.05 Downstream Heat-Shield Flow Rate (mmol/s) O 2 C H Position (mm) Flow Rate (mmol/s) O 2 C 6 H Position (mm) Flow Rate (mmol/s) C 6 H 6 O Position (mm) D Flow Rate (mmol/s) Upstream Heat-Shield CO H 2 O CO 2 H Position (mm) E Flow Rate (mmol/s) Catalyst CO H O CO 2 2 H Position (mm) F Flow Rate (mmol/s) Downstream Heat-Shield CO CO 2 H O Position (mm) H 2 Figure 5-4. Experimental and predicted spatial profiles of the catalytic partial oxidation of benzene at C/O = 1.1, 2 SLPM (GHSV ~ 5x10 4 h 1 ), and 1.1 atm. Results are displayed for the upstream heat shield region (Panels A and D) applying the homogneous mechanism, for the catalyst region (Panels B and E) applying the homogeneous and heterogeneous mechanisms, and for a portion of the downstream heat shield applying the homogeneous mechanism (Panels C and F). Molar flow rates of the reactants, O 2 ( ) and C 6 H 6 ( ), are presented in panels A through C, while molar flow rates of products, H 2 ( ), H 2 O ( ), CO ( ), and CO 2 ( ), are presented in panels D through F. The model uses the experimentally measured pyrometer temperatures, shown in Figure

140 A Coverage Catalyst CO(s) C(s) H(s) O(s) Rh(s) Position (mm) B Coverage H(s) Catalyst Rh(s) CO(s) C(s) Position (mm) Figure 5-5. Predicted species coverage of the catalytic partial oxidation of methane (Panel A) and benzene (Panel B) at C/O = 1.1, 2 SLPM (GHSV ~ 5x10 4 h 1 ), and 1.1 atm. The model uses the experimentally measured pyrometer temperatures, shown in Figure

141 A 0.50 Upstream Heat-Shield Catalyst Downstream Heat-Shield Flow Rate (mmol/s) CH 4 O 2 CO CO 2 H 2 H 2 O Position (mm) B 0.30 Upstream Heat-Shield Catalyst Downstream Heat-Shield Flow Rate (mmol/s) O 2 C 6 H 6 CO H O 2 CO Position (mm) H 2 Figure 5-6. Predicted spatial profiles of the homogeneous partial oxidation of methane (Panel A) and benzene (Panel B) at C/O = 1.1, 2 SLPM (GHSV ~ 5x10 4 h 1 ), and 1.1 atm. The model uses the experimentally measured pyrometer temperatures, shown in Figure

142 A B Figure 5-7. Scanning Electron Microscropy (A) and Back-Scattered Electron (B) image of an ~5 wt% Rh on ~5 wt% γ -Al 2 O 3 washcoat, 80 PPI α- alumina foam support. The white spots in the back-scattered image likely represent high molecular weight Rh particles. 126

143 Chapter 6: Synthesis gas from nonvolatile fuels by reactive flash volatilization 6.1 Introduction The direct conversion of nonvolatile hydrocarbons into synthesis gas (a mixture of H 2 and CO) and chemicals is an important process for using fuels such as vegetable oils and liquids produced by hydrolysis or pyrolysis of biomass [1]. Hydrogen is needed for fuel cells and for on-board combustion in vehicles for enhanced performance and reduced emissions, and synthesis gas is used for production of synthetic liquid fuels, chemicals, and fertilizers. Conversion of gaseous and volatile fuels to H 2 is possible through pyrolysis [1] steam reforming [2], and partial oxidation [3-5], with or without catalysts. However, the direct processing of nonvolatile fuels such as vegetable oils, residual petroleum fuels, and liquid and solid biomass is more complicated because of their tendency to form solid carbon that interferes with process equipment and rapidly plugs pores in heterogeneous catalysts. Such heavy fuels decompose chemically before evaporation to form hydrogen, olefins, aromatics, and solid carbon. Flash pyrolysis (reaction times typically 1 second) of heavy liquids and solid biomass has been shown [1] to produce primarily gases (synthesis gas) and volatile liquids (bio-oils). Reaction times in these processes are limited by heat transfer into biomass particles to decompose reactants. Additionally, at least ~10% of the reactant biomass is reported to form a solid char that must be separated and removed. Nonvolatile solid biomass pellets have been shown to volatilize without the formation of carbon when exposed to very high heat fluxes (~10 6 W/m 2 ) [5] of focused radiation from a xenon lamp. Similar high heat fluxes can also occur by catalytic partial oxidation of biomass to H 2 and CO, providing a comparable steady state environment where nonvolatile biomass can be decomposed without the production of carbon. This chapter demonstrates that, by flash heating small drops of heavy fuels in the presence of small amounts of O 2 impinging on a catalyst coated ceramic foam maintained at ~800 o C by the reaction, steady state catalytic reforming can be achieved with refined soy oil, biodiesel (the volatile methyl ester of soy oil), and sugar-water solutions with no external heat supplied. This process produces ~70% yield of H 2 with >99% conversion of the fuel. Carbon formation does not occur because the presence of O 2 produces rapid oxidation of decomposition products, and the Chapter 6 is adapted from J.R. Salge, B.J Dreyer, P.J. Dauenhauer, L.D. Schmidt, Renewable Hydrogen from Nonvolatile Fuels by Reactive Flash Volatilization, Science 314 (2006) American Association for the Advancement of Science. All Rights Reserved. 127

144 resulting heat of reaction maintains a surface temperature of 800 o to 1000 o C that prevents quenching of the process that would lead to rapid carbon formation. 6.2 Experimental Reactor system A photograph and schematic of the reactor system are shown in Figures 6-1 and 6-2. The reactor consisted of a quartz tube of 19 mm inner diameter. A single holed automotive fuel injector (Delphi Part # ) sprayed pressurized liquid fuel (soybean oil, biodiesel, or an aqueous 35 wt% glucose solution) onto the front-face of the catalyst. The glucose was 96%, anhydrous, α-d (+) glucose from Sigma-Aldrich. The soybean oil was Crisco All Natural Pure Vegetable Oil from J.R. Smucker Company, and the biodiesel was soybean derived, SoyPower Biodiesel from InterWest, LC. N 2, O 2, and CH 4 were fed into the top of the reactor. These high purity (99.9+ %) gases were fed into the reactor system from high-pressure cylinders (Airgas), and the flow rates were adjusted using mass flow controllers (Brooks 5850i). Three cylindrical catalyst-coated foams (80 pores per linear inch, 92% Al 2 O 3, 8% SiO 2, 16.5 mm outer diameter (OD), 10 mm length, Vesuvius Hi-Tech Ceramics) followed by an uncoated foam were wrapped around the perimeter with a ceramic fiber paper and placed in the quart tube. The ceramic fiber paper (Fiberfrax) held the foams in the reactor and prevented bypassing of gases around the foams. The uncoated ceramic monolith reduced axial radiation heat losses downstream of the catalyst. K-type thermocouples (Omega) were placed after the first (10 mm downstream) and after the third (30 mm downstream) catalyst-coated foam. The exterior of the quartz reactor was wrapped with 1-inch thick ceramic fiber insulation (Fiberfrax) to minimize radial heat losses. Products from the reactor flowed through a condenser. Samples of the uncondensed product stream were taken from a port downstream of the condenser with a gas-tight syringe (50 μl Hamilton Samplelock) and injected into a gas chromatograph (HP 5890 GC) for analysis. Typically, a sample was taken after continuously operating for ~30 min at a specified inlet composition and flow rate. During operation, the reactor pressure was maintained near atmospheric (~1.1 atm), and a total liquid flow rate of ~25 grams-per-hour was delivered to the catalyst from the fuel injector at 5 Hz with < 400 μm diameter droplets. To vary the fuel-to-oxygen ratio, the air flow rate was varied from 1.6 to 2.4 standard liters per minute (25 C and 1 atm). This total flow rate corresponds to 128

145 calculated catalyst contact times of ~40 to 60 ms at an entering reactant temperature of 800 C and pressure of 1.1 atm. The feed stream is reported as the carbon-to-oxygen ratio (C/O). C/O is defined as the number of moles of carbon atoms entering with the fuel divided by the number of moles of oxygen atoms from O 2 in the mixture Catalyst The catalyst support was an 80 pores per linear inch (PPI), 17 mm OD, and 10 mm length cylindrical α-alumina foam (92% Al 2 O 3, 8% SiO 2, Vesuvius Hi-Tech Ceramics). The catalyst foams were coated with ~2.5 wt % Rh and 2.5 wt % Ce. The Rh-Ce coated foams were coated using the wet impregnation method as described previously, where an aqueous solution of Rh(NO 3 ) 3 (Alfa Aesar, 13 wt% Rh in HNO 3 solution) and Ce(NO 3 ) 3* 6H 2 O (Sigma Aldrich) was dropped onto the foam [7]. The Rh-Ce coated foams were then dried and calcined in air at 600 C for 6 h. The Pt metal was applied by dropping an aqueous solution of H 2 PtCl 6 onto the foam. The Pt coated monoliths were then dried and calcined in 20% H 2 /80% N 2 at 500 C for 6 h Reactor start-up and shutdown To initiate the autothermal reforming of the nonvolatile biomass, CH 4 and air were passed over the catalyst at 350 C and reacted to form synthesis gas (H 2 and CO), releasing high levels of heat. Once the catalyst surface reached temperatures > 1000 C, the automotive fuel injector dispersed droplets of liquid fuel, mixed in air, directly onto the hot surface. Gradually, the inlet stoichiometries were adjusted to enable the liquid fuel to autothermally reform in air without the addition of CH 4. When the reactor was shut-down, CH 4 was reintroduced into the reactor, and the injection of liquid fuel was terminated. The stoichiometry of the CH 4 and air was adjusted to produce backface temperatures of ~1000 C, and the reactor was operated at this temperature for ~10 min. Then, the O 2 and CH 4 were removed from the reactor, and the reactor was cooled by ambient N Product analysis Product gases were analyzed using an HP 5890 GC equipped with a capillary column (J&W Scientific GASPRO, 60m length, 0.32 outer diameter) and thermal conductivity detector (TCD). This system was capable of separating and detecting permanent gases, higher hydrocarbons, olefins, aromatics, ketones, and aldehydes. Compounds that condensed in the condenser were considered unreacted fuel and reflected in the fuel conversion. Some of these condensed 129

146 compounds may be reacted fuel, and thus, the actual fuel conversion may be higher than reported. Soybean oil is a triglyceride that primarily contains 5 fatty acids: methyl palmitate (~12%), methyl stearate (~5%), methyl oleate (~25%), methyl linoleate (~52%), and methyl linolenate (~5%) and has an average molecular formula of C 56.3 H 99.7 O 6 [8-10]. Biodiesel is the methyl ester of these fatty acids and has an average molecular formula of C 18.8 H 34.6 O 2. The average molecular formula of these compounds was used to calculate approximate C/O compositions flowing through the reactor. Reaction products containing carbon are reported on a carbon atom basis using selectivities. Product selectivity (S) is defined as the number carbon or hydrogen atoms in product i divided by the carbon or hydrogen atoms in the converted fuel: For the aqueous glucose solution, water was not considered fuel. Thus, if H 2 is produced from the conversion of water and fuel, the H 2 selectivity could exceed 100%. The carbon atom balance was closed to determine the fuel conversion (X), and the oxygen atom balance was closed to determine the water molar flow rate Glucose-water operation Glucose was dissolved in distilled water near maximum solubility in ambient conditions (~35 wt%) and sprayed onto the catalyst with the automotive fuel injector. To operate autothermally, CH 4 was simultaneously fed with the aqueous glucose solution in N 2 and O 2 over the catalyst. The ratio of carbon atoms in CH 4 to oxygen atoms in O 2 (C Methane /O) was set at 0.60, and the N 2 /O 2 was decreased from 3.76 (air stoichiometry) to 0.55 to provide a high surface temperature without igniting CH 4 and O 2 upstream of the catalyst. The ratio of carbon atoms from CH 4 to carbon atoms in glucose (C Methane /C Glucose ) was then varied by adjusting the rate that fuel was sprayed onto the catalyst. 6.3 Results Soy-bean oil Results for refined soy oil (the triglyceride of C 18 and C 16 fatty acids) as fuel are shown in Figure 6-3A for conversion and temperature. The temperature 10 mm downstream from the front face, which is generally cooler than the front face where oxidation reactions occur, decreased from 1100 o C (glowing bright orange) to 800 o C (dull red) as C/O increased from 0.8 to 1.2. The lower C/O limit is set by thermal destruction of the catalyst (sintering of Rh), and the higher C/O limit is set by carbon formation that shuts down the process. The catalyst temperature further 130

147 downstream (30 mm) was ~ 200 C lower than the upstream temperature (10 mm). O 2 conversion was > 99% for all C/O, and soybean conversion was > 95% at C/O < 1.0 and > 85% at C/O > 0.9. Figure 6-3B shows the hydrogen atom selectivity for H 2 and also displays the carbon atom selectivities for CO, CO 2, and ethylene and propylene. H 2, H 2 O, CO, CO 2, ethylene, propylene, and 4 to 6 carbon olefins were the major products observed for the non-condensed stream. The maximum H 2 and CO selectivities were observed at C/O ~ 0.8 and were 65% and 60%, respectively. At C/O < 0.8, H 2 was displaced by the production of H 2 O, while CO remained constant. When C/O increased, olefins (primarily ethylene and propylene) were produced in place of synthesis gas. Maximum ethylene and propylene were observed at the highest stable C/O, 1.1, and was ~18%. This process has been operated continuously for more than 20 hours on a given catalyst and the performance has been repeated on several catalysts. In all cases, no deactivation was observed (<2% changes in conversion or selectivities over 20 hours) as long as the catalyst was not overheated. Higher C/O ratios (C/O >1.1) caused deactivation through carbon formation in the catalyst, but the activity could be restored quickly by decreasing C/O to burn off this carbon or by adding CH Biodiesel Results for a similar experiment with biodiesel instead of soy oil are shown in Figure 6-4. Biodiesel (the methyl ester of the fatty acids from the triglyceride ester made by transesterification of soy oil) boils without decomposition at >300 o C, so vaporization of biodiesel without carbon formation is possible. Previous experiments have shown [4] that biodiesel could be converted into H 2 and CO or into olefins in a similar reactor system where the biodiesel was vaporized by heating the walls of the reactor above the catalyst to 300 o C, and a heat shield-mixer was inserted between the vaporized fuel and the catalyst. In the present experiment, no external heat was added, so fuel and air were fed at room temperature. Figure 6-4A displays biodiesel conversion and the catalyst temperature at 10mm and 30 mm from the front-face. Similar to soybean oil, the catalyst began to coke and deactivate when C/O > 1.2. Catalyst temperatures 10 mm and 30 mm downstream of the front-face were > 800 C and >550 C, respectively. The operating temperature was ~100 C lower than soybean oil at corresponding C/O. O 2 conversion was > 99% for all C/O, and biodiesel conversion was > 95% at C/O < 1.1 and > 85% at C/O >

148 Figure 6-4B shows the hydrogen atom selectivity for H 2 and also displays the carbon atom selectivities for CO, CO 2, and combined for ethylene and propylene. Water and 4 to 6 carbon olefins were also observed in the non-condensed stream. The maximum H 2 and CO selectivities were observed at C/O ~ 0.7 and were ~78% and ~73%, respectively. Similar to soybean oil, at C/O > 0.7, ethylene and propylene were produced in place of synthesis gas. Maximum ethylene and propylene were observed at the highest stable C/O, 1.2, and was ~17%. When compared to soybean oil, biodiesel produced more synthesis gas and less olefins at a given C/O. Additionally, the conversions and selectivities obtained from flash volatilization are similar to results obtained from pre-vaporization of biodiesel Aqueous glucose solution In another experiment, a glucose-water solution (35% by weight glucose in water) was fed through the fuel injector. (The solubility of glucose in water at 20 o C is 38% by weight.) Steady state autothermal operation was not maintained without adding CH 4 along with the sugar, and the lowest CH 4 we were able to maintain at C/O = 0.8 was 3 carbon atoms from methane per carbon atom from glucose. Figure 6-5A displays conversions for the aqueous glucose solutions and the catalyst temperature at 10mm and 30 mm from the front-face. Catalyst temperatures 10 mm and 30 mm downstream of the front-face were > 900 C and > 600 C respectively, and the catalyst temperature further downstream (30 mm) was ~ 200 C lower than the upstream temperature (10 mm). Adding the aqueous glucose solution to CH 4 reduced the catalyst temperature by ~100 C. O 2 conversion was > 99% for all C/O, and glucose conversion was > 95% when the glucose solution was added. The CH 4 conversion decreased from > 95% in the partial oxidation of CH 4 to 80% when the glucose solution was added to CH 4. Figure 6-5B also shows the hydrogen atom selectivity for H 2 and also displays the carbon atom selectivities for CO and CO 2. H 2, H 2 O, CO, and CO 2 were the major products observed for the noncondensed stream. The maximum H 2 selectivity was observed at C Methane /C Glucose = 4 and was 79%. Lower olefins were not observed for the aqueous glucose mixture. Performance was comparable to that for soy oil or biodiesel with >99% glucose conversion and ~70% H 2 selectivity Pt Autothermal reforming of soybean oil and biodiesel on Pt-coated monoliths was also attempted. However, stable operation was not achieved in the designed system. For soybean oil, fluctuations in temperature and product selectivities were observed at all C/O. H 2 O, CO, CO 2, 132

149 and olefins were primarily produced, while the H 2 selectivity was less than 10%. Biodiesel primarily produced CO, CO 2, H 2 O, while the H 2 selectivity was greater than 10%. While more synthesis gas was observed with biodiesel on Pt than with soybean oil on Pt, the synthesis gas selectivities produced from biodiesel on Pt were still 40% lower than the synthesis gas selectivities observed from biodiesel on Rh-Ce at corresponding C/O. 6.4 Discussion The results shown are from preliminary experiments in which little optimization of catalyst of conditions was attempted. Preheat in the volatilization zone or in the fuel injector should increase stability and performance, and smaller drop sizes should also increase the range of operation. This process produces no more than a few percent of higher hydrocarbons (primarily ethylene and propylene) for any of these fuels at C/O = 0.8, so these products are suitable for use in a fuel cell with minimal cleanup. ~20% carbon atom selectivity to ethylene and propylene at C/O = 1.2, and operation at higher C/O should produce much higher yields of olefins, comparable to the 80% total olefins reported for conventional catalytic partial oxidation of biodiesel [4] and n-hexadecane [5]. Operation at higher C/O would require adjustment of preheat, flow conditions, and perhaps use of sacrificial CH 4 to avoid carbon formation. A patent describing a similar experiment has been reported where heavy petroleum was co-fed with light hydrocarbons or fed periodically in order to produce ethylene and higher olefins [10]. In the following sections, the volatilization and reforming within this reactor will be discussed The flash volatilization process The processes by which nonvolatile liquids can be converted to H 2 and other small molecules without carbon formation in this reactor require examination. When a drop of a volatile liquid hits a hot surface, the vaporization at the interface can be fast enough that a gaseous layer is formed which insulates the drop from the surface. This regime is called "film boiling" or "water droplets on a hot frying pan," as sketched in Figure 6-6A, and the heat transfer and mechanisms of this process have been considered extensively [12-15]. Nonreactive drop breakup upon impact has been studied [16], as has nonreactive drop impact upon heated porous surfaces [17]. The present situation is considerably different than conventional film boiling. Without chemical reaction, the surface would cool rapidly by heat transfer and boiling of the liquid. Instead, the process relies on the chemical reaction of the volatile products (the parent liquid as well as H 2 and other smaller fragments) to continuously maintain the surface temperature high enough to sustain steady state impact and decomposition of drops. The possible configuration for a drop with reaction occurring on a hot catalyst surface is sketched in Figure 6-6B. The 133

150 process probably relies on reaction in the gaseous layer between the drop and the hot catalyst to continuously generate gaseous products and heat, and the process can continue until all fuel is volatilized, either in the initial drop or in successive smaller drops that form from the primary drop. The sequence of surface and homogeneous reaction steps in reactive flash volatilization is unknown. The initial reaction step is suspected to be partial vaporization and pyrolysis upon impact of the cold drop with the hot surface. The thermal radiation and conduction from the catalyst surface is suspected to rapidly heat the droplets to 1000 C within milliseconds on impingement on the catalyst surface. The rapid heating rate provides sufficient energy to decompose the fuel to lower molecular weight compounds, such as olefins, aldehydes, and ketones, without producing high levels of carbon. If carbon is produced, the carbon is sufficiently consumed through oxidation near the surface. For small droplets (~10-3 m) as generated in this system, volatilization is controlled by chemical kinetics, and no internal temperature gradient exists [6]. With the presence of O 2 and high temperatures, the chemical kinetics to homogeneously oxidize and crack the heavy fuel to volatile components is sufficiently fast to volatilize in milliseconds [5,18]. Additionally, rebound of droplets allows oxidation reactions, and film boiling permits gas and surface reactions. Once the drop or its fragments enter the ceramic foam, reactive decomposition should be very rapid. However, the amounts of reaction in each of these stages are unknown. The overall process is extremely complicated because of possible drop breakup dynamics and the structure of the porous catalyst surface on which the process occurs. Some of these issues are sketched in Figure 6-6C and D. The velocity of the drop above the surface is calculated to be ~1 m/s, so the impact of the drop on the surface could promote rapid breakup into smaller drops by momentum transfer. The initial impact probably involves the cold liquid making direct contact with the surface. These events are estimated to involve microsecond times where gradients and sequences of events are extremely large and difficult to predict. Previous studies have also noted the difficulties in accurately describing the dynamics of droplets which simultaneously boil and crack when impacting a hot surface [15] or which impact a hot porous substrate [17] Pyrolysis pathways To flash vaporize soybean oil and glucose, bonds in these compounds are most likely broken to form lower molecular weight compounds that can vaporize. Glucose is believed to initially dehydrate to form levoglucosan [19-20]. Subsequent dehydration and C-C bond cleavage reactions are then expected to occur, generating volatile, lower molecular weight compounds, such as small acids, aldehydes, ketones, hydrocarbons, olefins, and aromatics. 134

151 Soybean oil is suspected to volatilize through homogeneous cracking of the glycerin component and the fatty acid chains. In the pyrolysis of soybean oil, initial dissociation should occur where the bond energy is the lowest. The weakest bonds are the C-O ester bonds that bind glycerin to the fatty acid chains. These bonds are likely to break to form di-glycerides and a fatty acid [21]. Once fatty acids are formed, several decomposition pathways exist. In the unsaturated fatty acid molecules, the C=C and C-O bonds are stronger than C-C bonds, and thus, a C-C bond should next dissociate. The C-C bond energy is the lowest where the allylic radical stabilizes one of the radicals as a resonance structure (R=C-C* and R-C=C*, where R is a hydrocarbon chain) [4]. About 90% of soybean oil contains a C-C double bond, and this double bond occurs between the tenth and eleventh carbon atom in the fatty acid (1 st carbon = carboxylated carbon). Thus, 90% of the oil molecules are likely to break between the eight and ninth carbon atoms on the fatty acid chain, forming two addition radicals: a smaller fatty acid and alkyl radical. At high reaction temperature gradients, β-scission probably occurs on the C-C bond of the alkyl and fatty acid radicals to produce ethylene and a smaller alkyl radical. This reaction continues producing ethylene until the parent fuel chain volatilizes. Additionally, the fatty acid radicals likely de-carboxylate to form CO 2 and a smaller alkyl radical Catalytic reforming pathways The lower molecular weight compounds produced from the volatilization zone are believed to adsorb and react on the catalyst surface with oxygen to form synthesis gas and combustion products and to release the heat that promoted the endothermic pyrolysis reactions upstream of the catalytic reforming reactions. Catalytic partial oxidation and combustion are thought to dominate initially in the catalytic reforming section until all the O 2 is consumed, typically in the first several millimeters of the catalyst [22]. H 2 and CO are mostly produced with some H 2 O and CO 2, and the temperature typically is 100 to 200 C higher than the measured temperature 10 mm downstream of the catalyst front-face. With the absence of O 2, the remaining unreacted and partially reacted fuel is then catalytically steam reformed to CO and H 2, and then the steam catalytically reacts with CO to produce CO 2 and H 2. Most likely endothermic steam reforming dominates downstream of the oxidation zone, increasing the level of H 2 and CO and reducing the temperature to approximately the observed temperature 30 mm downstream of the catalyst front-face. Olefins are most likely produced mainly through homogenous cracking reactions that occur upstream of the surface and simultaneously with surface reactions inside the catalytic foam [5]. The high temperatures of the exothermic combustion and partial oxidation reactions provide 135

152 energy for the vaporized fuel to further decompose by the thermal cracking mechanism as described in the flash volatilization section. Biomass molecules are composed of aldehyde, alcohol, ketone, ester and ether functional groups. The adsorption and decomposition of compounds containing these functional groups can be quite different. For example, on Rh (111), ethanol is believed to initially adsorb on the metal and form an ethoxide species [23-25]. A bridged oxametallacycle is then formed, and the C-C bond in the oxametallacycle is rapidly broken and quickly decomposes to adsorbed methylidyne and carbon monoxide. These atoms further recombine on the surface with adsorbed O to form synthesis gas and combustion products. In contrast, acetaldehyde is believed to adsorb on two adjacent Rh (111) sites to form an η 2 -acetalhyde species [23-25]. A η 1 -acyl is then formed, and the C-C bond in the η 1 -acyl is rapidly broken and quickly decomposes to adsorbed methyl and carbon monoxide. The methyl group can decompose to adsorbed C and H atoms and recombine on the surface with adsorbed O to form synthesis gas and combustion products, or it can combine with an adsorbed H to form methane. A decomposed biomass compound of glucose would also contain hydroxyl and aldehyde functional groups. These hydroxyl and aldehyde groups are believed to adsorb and react similarly to the hydroxyl group in ethanol and aldehyde group in acetaldehyde, producing CO and H adatoms. No obvious channels to produce olefins from glucose appear to exist. The cracked products in biodiesel and soybean oil have similar molecular structures to cracked linear alkenes and alkanes. Thus, cracked biodiesel and decomposed vegetable oil could adsorb to the surface similarly to the adsorption of light olefinic hydrocarbons, such as ethylene. On Rh(111) single crystals at ultra-high vacuum, ethylene has been observed to bond to the catalytic surface through its π-bonds and dissociate to C and H at temperatures >400 K[26]. Since the surface temperature is much greater than 400 K, the rapid decomposition of ethylene to adsorbed C and H is expected, and this surface C and H likely combine with adsorbed O to form combustion and synthesis gas products that desorbs from the surface. Steady operation was observed for Rh catalysts and not Pt catalysts. Rh is believed to form less carbon and has been shown to be a superior reforming catalyst of higher alkanes to synthesis gas when compared to Pt [27-28]. When added to a noble metal catalyst, Ce also has been shown to enhance noble metal dispersion at reaction conditions, which further promotes catalytic reforming of non-volatile fuels [29]. 136

153 6.4.4 Biodiesel vs. soybean oil The results from the biodiesel reforming agree well with previous results at corresponding C/O and flow rates [4]. However, the autothermal operability range for biodiesel was greatly reduced from C/O ~2.2 to ~1.2. The difference in operability is probably due to the coupling of the vaporization and reaction sections in the reactor system. Previous experiments were performed with the upstream preheated to ~300 C, which completely vaporized the fuel, and the catalyst was only used for catalytic partial oxidation and additional cracking. In contrast, the current reactor system uses the catalyst to provide energy for the initial cracking and vaporization of the biomass compounds as well as partial oxidation and further cracking. The additional energy load on the catalyst reduces the operability to regions where exothermic oxidation is dominant, C/O < 1.1. When C/O > 1.0, insufficient exothermic chemistry exists to provide fast cracking of biomass in millisecond contact times, causing the catalyst to coat with nonvolatile biomass. When the catalyst is coated with biomass, O 2 cannot adsorb on the catalyst surface and exothermically react with the fuel, and without the exothermic chemistry, the catalyst extinguishes. When compared to biodiesel reforming, soybean oil produced less H 2 and CO and more olefins. Olefins are believed to be produced homogeneously and are produced simultaneously with heterogeneous synthesis gas reactions [4-5]. Thus, the higher concentration of olefins and lower concentration of synthesis gas products suggest that soybean oil does not catalytically reform as well as biodiesel in the reactor system. Soybean oil is a larger molecule than biodiesel and is not volatile. To volatilize soybean oil, several bonds have to be broken. Once these bonds are broken, the cracked oil can vaporize, adsorb onto the catalyst surface, and finally react with adsorbed oxygen to produce synthesis gas. The breaking of these extra bonds requires more energy and potentially more time before the molecule can potentially catalytically reform with oxygen. Thus, at a constrained residence time in the reactor, soybean oil is expected to reform less than biodiesel to synthesis gas Glucose and other biomass solids The flash vaporization and reforming of biomass oils was successful in producing H 2 autothermally. However, the aqueous glucose solution did not produce H 2 autothermally without the addition of CH 4. The water solvent did not supply exothermic energy to the system to sustain autothermal operation. Instead, the water increased the endothermic steam reforming reactions and diluted the exothermic chemistry. To overcome the heat load of the water vaporization, CH 4 was added to reduce the fuel-to-water ratio and make the overall reaction exothermic. Potentially, other renewable solvents, such as alcohols, ketones, or aldehydes produced from fermentation of 137

154 biomass, could exothermically oxidize and enable the autothermal reforming of solid biomass without the addition of CH Other fuels This process appears to be general for any fuel, because the hot catalyst surface will pyrolyze and oxidize the liquid at the interface into products that are easily oxidized. Heavy fuels such as residual petroleum fractions, yellow grease (used cooking oil), and crude soy oil should be processable with little or no pretreatment. At the observed reactor temperatures of 800 to 1000 C, the presence of small amounts of impurities such as phosphorous, potassium, nitrogen, and other small organics present in renewable feedstocks will probably not be detrimental to the catalyst because they are volatile. Varying amounts of moisture within feedstocks should actually improve the H 2 yield and further suppress carbon formation. The process should also be scalable over a wide range of capacities. The present system processes approximately 0.6 kg/day of fuel using 150 mg of Rh, and a catalyst disc 5 cm in diameter would process ~5.2 kg/day under identical conditions. We had to use small single orifice automotive fuel injectors to obtain sufficiently low flows, but larger systems could use larger multiport injectors, multiple injectors, or different methods for uniform drop formation over the entire catalyst surface. Many technologies require fast drop volatilization without carbon formation. Diesel engines require rapid combustion of nonvolatile fuels, but impact of drops on walls is generally avoided to prevent coke formation. New engine technologies could use drop volatilization at a catalytic surface to improve diesel combustion and reduce pollutants. Heavy oils such as residual petroleum fractions and biomass derived liquids can be pyrolyzed and combusted in fixed or fluidized beds, but this generally involves a reducing zone where carbon forms followed by an oxidizing zone where the carbon is burned off. A single zone catalytic process would be much smaller and simpler, and use of catalysts would allow tuning of selectivities not possible with flame combustors. Catalytic processes also eliminate or strongly reduce pollution associated with flame combustors. 6.5 Summary Reactive drop volatilization appears to be a simple and readily adaptable method to convert nonvolatile fuels into H 2 or chemicals for large as well as small scales of production such as onboard vehicle reforming. It allows intensification of the gasification process into millisecond time scales and suggests that conversion of other nonvolatile biomass mixtures such as emulsions, slurries, and powders is possible. The process also requires further experiments, long-term evaluation, and modeling to optimize catalyst performance and exact mechanisms of reactive 138

155 flash volatilization. 6.6 References 1. A.V. Bridgwater, Chemical Engineering Journal 91 (2003) L. Garcia, R. French, S. Czernik, E. Chornet, Applied Catalysis A: General 201 (2000) G.A. Deluga, J.R. Salge, L.D. Schmidt, X.E. Verykios, Science 303 (2004) R. Subramanian, L.D. Schmidt, Angewante Chemie International Edition 44 (2005) J.J. Krummenacher, K.N. West, L.D. Schmidt, Journal of Catalysis 215 (2003) O. Boutin, M. Ferrer, J. Lede, Chemical Engineering Science 57 (2002) A. Bodke, S. Bharadwaj, L.D. Schmidt, Journal of Catalysis 179 (1998) National Biodiesel Board Chemical Weight and Formula, National Biodiesel Board, M. Marquevich, X. Farriol, F. Medina, D. Montane, Industrial Engineering and Chemistry Research, 40 (22) (2001) A. Demirbags, Fuel, 77(9-10) (1998) D.C. Griffiths, K.W. Palmer, I.A.B. Reid, US Patent No. 5,663,473 (1997), Assigned to BP Chemicals Limited, London, United Kingdom. 12. S. Deb, S.-C. Yao, International Journal of Heat Mass Transfer 32 (1989) L.H.J. Wachters, N.A.J. Westerling, Chemical Engineering Science 21 (1966) Y. Ge, L.-S. Fan, Physics of Fluids 17 (2005) B.S. Gottfried, C.J. Lee, K.J. Bell, International Journal of Heat Mass Transfer 9 (1966) M. Bussman, S. Chandra, J. Mostaghimi, Physics of Fluids 12 (2000) S. Chandra, C.T. Avedisian, International Journal of Heat Mass Transfer 35 (1992) R. Fournet, F. Battin-Leclerc, P. A. Glaude, B. Judenherc, V. Warth, G. M. Côme, G. Scacchi, A. Ristori, G. Pengloan, P. Dagaut, M. Cathonnet, International Journal of Chemical Kinetics 33 (10) (2001) R.J. Evans, T.A. Milne, Energy and Fuels 1(2) (1987) R.J. Evans, D. Wang, F.A. Agblevor, H.L. Chum, S.D. Baldwin, Carbohydrate Research 281 (1996) K.D. Maher, D.C. Bressler, Bioresource Technology 98 (2007) R. Horn, K.A. Williams, N.J. Degenstein, L.D. Schmidt, Journal of Catalysis, 242 (2006) C.J. Houtman, M.A. Barteau, Journal of Catalysis 130 (1991) E.C. Wanat, B. Suman, L.D. Schmidt, Journal of Catalysis 235 (2005)

156 25. P.J. Dauenhauer, J.R. Salge, L.D. Schmidt, Journal of Catalysis 244 (2006) L.H. Dubois, D.G. Castner, G.A. Somorjai, Journal of Chemical Physics 72(9) (1980) J.J. Krummenacher, L.D. Schmidt, Journal of Catalysis 222 (2004) D. Shekhawat, T.H. Gardner, D.A. Beery, M. Salazar, D.J. Haynes, J.J. Spivey, Applied Catalysis A: General 311 (2006) N.J. Degenstein, R. Subramanian, L.D. Schmidt, Applied Catalysis A: General 305 (2006)

157 Fuel Fuel Air Air Air Air 2 cm Insulation Catalyst Catalyst Heat shield Heat shield Figure 6-1. Diagram and photograph of reactor autothermally reforming soybean oil. An automotive fuel injector sprays soybean oil onto a Rh-Ce coated ceramic foam, while oxygen flows into the reactor around the fuel injector. Insulation surrounding the perimeter of the reactor tube is removed to show the catalyst surface in the photograph, and CH 4 is also added to help maintain high surface temperatures in the volatilization zone with the removal of the insulation. 141

158 Liquid Fuel N 2, O 2, CH 4 Fuel Injector N 2, O 2, CH 4 Catalyst Heat Shield Thermocouple (10mm) Thermocouple (30 mm) Insulation Condenser Sample Port (Non-Condensable Gases) To Incinerator Condensate Collection Figure 6-2. Schematic of reactor. 142

159 A X (%) Soybean Oil T (10 mm) T (30 mm) C/O T ( o C) B S C or S H (%) H 2 60 CO 40 CO 2 20 C 2 H 4 & C 3 H C/O Figure 6-3. Soy oil (A) conversion and temperature and (B) selectivities to H 2 and carbon products by reactive flash volatilization. All gases and liquids enter the reactor at room temperature, and no process heat is added. The inset shows the reactor configuration in which liquid drops of soy oil from an automotive fuel injector impact a Rh-Ce catalyst on an alumina foam to vaporize the drops in ~1 millisecond without carbon formation. A X (%) Biodiesel T (10 mm) T (30mm) C/O T ( o C) B S C or S H (%) 100 H 2 80 CO CO 2 20 C H & C H C/O Figure 6-4. Biodiesel (A) conversion and (B) selectivities to H 2 and carbon products by reactive flash volatilization. Conversion is high and selectivities are comparable to those from soy oil. 143

160 A B C 6 H 12 O 6 CH 4 CH H 2 CO X (%) T (10mm) T (10mm) T ( o C) S C or S H (%) CO CO 2 H 2 20 T (30mm) T (30mm) CO No Glucose No Glucose C Methane /C Glucose C Methane /C Glucose Figure 6-5. Effect of methane/glucose ratio (C Methane /C Glucose ) on (A) the fuel conversion (X), the catalyst temperature (T), and (B) product selectivities (S) at 1.1 atm over Rh-Ce catalysts. Catalyst temperatures were measured at 10 mm and 30 mm downstream of the catalyst frontface. To vary the C Methane /C Glucose, an aqueous glucose solution (35 wt % glucose) was added to a 0.87 SLPM (25 C, 1 atm) CH 4, 0.73 SLPM O 2, and 0.4 SLPM N 2 gas mixture. 144

161 A B C D Cold drop Cold drop O 2 Cold drop O 2 Cold drop O 2 Vapor Oxidation products Oxidation products Oxidation products Hot surface Hot catalytic surface Hot catalytic surface Hot porous catalytic surface Figure 6-6. Sketches of possible configurations in (A) conventional film boiling of a volatile drop on a hot surface, (B) reactive volatilization on a hot catalyst surface, (C) sequence of drop impingement and breakup on a hot catalytic surface, and (D) sketch of possible configuration on a porous ceramic foam 145

162 Chapter 7: Millisecond reforming of solid non-volatile fuels, an extension of reactive flash volatilization 7.1 Introduction Dependence on petroleum and continued carbon emissions have led to a focus on methods of utilizing a large supply of biomass in the form of grasses, trees, and agricultural residue [1] However, biomass presents a significant processing challenge, because it is a complex mixture of biopolymers dispersed across the countryside. Current techniques to produce synthesis gas for liquid fuels such as fast pyrolysis or gasification are complicated, and require long residence times and significant transportation to the processing location [2-3]. In this chapter, a unique catalytic method to convert non-volatile biomass polymers to synthesis gas without an external heat source at least an order of magnitude faster than existing systems. Small particles directly contacting a hot catalytic surface maintained by heat generated from partial oxidation undergo rapid decomposition without detectable char production to form a tar-free synthesis gas stream at millisecond reaction times. Considered solid fuels include cellulose, starch, wood chips from Aspen (Populus tremuloides), and polyethylene, an example of common municipal waste. Conversion by this technology has the potential to permit production of synthesis gas from solid biomass in small, simple processes. Direct thermochemical conversion of biomass to a stream of synthesis gas (H 2 and CO) is an attractive route to transportation fuels without extensive pre-processing of biomass. Clean, conditioned synthesis gas can be converted into diesel fuel or mixed alcohols through the Fischer Tropsch process or to methanol or dimethyl ether allowing high efficiency end use in modern diesel engines without significant changes in the current transportation infrastructure [4] While the thermochemical route to synthesis gas can convert a solid mixture of biopolymers, this process lacks an effective catalytic method that is easily scalable and sufficiently simple. A major challenge with direct catalytic conversion of solid biomass is to avoid the formation of solid char that can cover catalyst surface sites and block surface reactions. Slow heating of biomass such as cellulose, (C 6 H 10 O 5 ) n, at low temperatures can result in a significant fraction converting to solid char similar to charcoal production from wood. Chapter 7 is adapted from P.J. Dauenhauer, B.J. Dreyer, N.J. Degenstein, L.D. Schmidt, Millisecond Reforming of Solid Biomass for Sustainable Fuels, Angewandte Chemie International Edition, 119 (31) (2007) WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany. All rights reserved. 146

163 (C 6 H 10 O 5 ) n Char + H 2 O (7-1) Δ Global homogeneous models such as the Shafizadeh model describing this conversion predict significantly less char production above 400 C with most of the biomass being converted to volatile organic compounds (VOC) at 500 C in about a second [6-7]: (C 6 H 10 O 5 ) n VOC (i.e. hydroxyacetaldehyde), (7-2) Δ (C 6 H 10 O 5 ) n Gases (i.e. CO + H 2 ). (7-3) Δ At even higher temperatures, conversion occurs much faster with higher selectivity to gases and little selectivity to char [8] In the previous chapter, non-volatile liquids such as soy oil and sugar-water droplets were converted to synthesis gas without any carbon formation by reactive flash volatilization in which cold drops impinge on a hot catalyst surface [9]. In this chapter, this flash volatilization process is extended to solid particles of starch, cellulose, Aspen, and polyethylene ranging in size from 10 μm to 1 mm. The results demonstrate that solids carbon-containing fuels can be converted to synthesis gas on a hot Rh surface of a 30 mm catalytic bed without detectable deactivation from carbon formation. This process occurs at a total gas residence time of less than 70 milliseconds, which is more than ten times faster (and thus ten times smaller) than reported biomass gasification processes [3] Biomass can currently be converted to synthesis gas in several different types of gasifiers that oxidize and pyrolyze biomass particles in large systems. At shorter residence times, a technique called fast pyrolysis heats biomass particles for about one second to produce a predominately liquid product, bio-oil, that can be catalytically reformed to synthesis gas using Rh or Ni catalysts [3,10] This concept has been demonstrated as a complex integrated fast pyrolysis and catalytic reforming system at moderate temperatures, however, it still requires external heating and residence times of approximately one second to operate [11-12] 147

164 7.2 Experimental Control of Solid Particle Flow Figure 7-1 displays a schematic of the experimental set-up. Solid particles were delivered to the reactor from a hopper consisting of a 4 diameter acrylic tube. Particles were metered with a ¼ diameter auger running axially through the hopper to an exit tube attached to flexible tubing that connected to the top of the reactor. The entrance point of the auger to the hopper was sealed using a 3/8 fitting sealed with an o-ring, and the turning rate was maintained at 3-70 rpm using a variable speed mixer motor (Cole-Parmer). Solid particles were agitated to sufficiently fill the auger by impact of a plastic tube fastened to the front of a 10 subwoofer operating at 30 Hz (Panasonic). The top of the reactor consisted of an annulus of 7 mm OD through which solid particles and air flowed down to the front face of the catalyst foam bed. Gases were metered using mass flow controllers (Brooks 5850i) for N 2, O 2, and CH 4 at startup. During operation, radiation from the catalyst surface heated the annulus wall. For some feedstocks, such as polyethylene, this heat was sufficient to cause the feedstock to soften, agglomerate, and adhere to the inner wall of the annulus. Gradually the reactor feed tube would plug from the material adhering to the tube. To prevent adhesion and agglomeration, a cooling jacket was placed around the tube. Cold water (~10 o C) maintained the annulus wall temperature below the softening point of the solid feedstock, which prevented plugging upstream of the catalyst Reactor Set-up The reactor tube consisted of a 19-mm inner diameter, 20-cm-long quartz tube. The metal catalysts were loaded on a ceramic foam or sphere fixed bed by the wet impregnation technique described previously [13]. In Table 7-1, "Foam" consisted of 2.5 wt% (~0.05 g) Rh and 2.5 wt% Ce on 45 PPI α-al 2 O 3 foams for the first 10 mm. The support foam from 10 mm to 30 mm consisted of 2.5 wt% Rh and 2.5 wt% Ce on a 5 wt% γ-al 2 O 3 washcoat supported on 80 PPI α- Al 2 O 3 foams. "Spheres" consisted of a bed 30 mm in length of 1 mm diameter spheres with 5 wt% γ-al 2 O 3 washcoat and the same amount of Rh and Ce as used in the foam. The catalyst was initially heated by methane partial oxidation to 800 C after which a continuous flow of solid particles was immediately applied to the front catalyst surface. Methane was removed from the system and the flow rate of air was adjusted to satisfy 0.6 < C/O < 1.0. No external heating was required as solid particles and air entered the reactor at room temperature. The quartz reactor tube was wrapped in 1 inch of insulation (Fiberfrax) to prevent radial heat 148

165 losses. Gas samples were collected with a gas syringe (Hamilton Gastight Samplelock) after a water-cooled condenser and injected into a gas chromatograph (HP 6890, J&W Scientific PLOT- Q column, 30 m length, 0.32 mm outer diameter). Column response factors and retention times were determined by injecting quantities of known species relative to N 2 (Matheson Tri-Gas). Mass balances on carbon and hydrogen typically closed within +5%. Figure 7-4 shows the temperature of the catalyst reactor system processing cellulose at 25 g/hr with air on a foam support at C/O=0.7 and C/O=0.9. This data was collected with 1 mm resolution using a capillary technique developed by Horn, et al. to insert a K-type thermocouple (Omega) into the fixed catalytic bed during operation [14]. Data points were collected by averaging the observed temperature for 30 seconds at each fixed location. Data collection was completed twice for each C/O ratio, and the two data sets were averaged Experimental feedstock Cellulose in experiment 1 of Table 7-1 was obtained from Sigma-Aldrich. Cellulose in experiments 2 and 3 of Table 7-1 was obtained from FMC biopolymer. Starch in experiment 4 of Table 7-1 was obtained from Bob s Red Mill Natural Foods, Inc. as palletized starch prepared from the Cassava plant. Aspen chips obtained from trees growing in Rosemount, MN, were dried and ground over a size 20 mesh. Polyethylene in experiments 6 and 7 of Table 7-1 was obtained from Alfa Aesar. The mass average MW of polyethylene was ~50,000 determined by GPC. Mass average particle sizes were determined from particle size distributions measured by light scattering. The particle size distributions for each feedstock are shown in Figure 7-2. Moisture and ash content were determined gravimetrically Equilibrium calculations Equilibrium was calculated using HSC software [15]. Carbon equilibrium lines of Figure 7-3 were calculated using the feed ratios of C, H, and O and permitting a solid phase of C along with a gas phase of N 2, O 2, CO, CO 2, CH 4, H 2 O, C 2 H 2, C 2 H 4, C 2 H 6, C 3 H 6, C 3 H 8, C 4 H 10, C 4 H 8, C 5 H 12, C 5 H 10, CH 3 OH, C 2 H 5 OH, CH 2 O, C 2 H 4 O. Equilibrium selectivity to CO, H 2, and CH 4 in Figure 7-3 were calculated permitting the same species at the temperature measured 30 mm down in the catalytic bed. 149

166 7.3 Results & discussion Cellulose as a function of C/O Figure 7-3 shows the results of the catalytic processing of ~230 μm cellulose particles at residence times < 70 ms in a fixed foam bed with a Rh catalyst (inset) described elsewhere. [13] In this experiment we varied the ratio of cellulose to air feed rate defined as C/O (carbon atoms from fuel / oxygen atoms from air). The temperatures, measured with thermocouples 10 mm and 30 mm from the top of the catalytic bed, never decreased below 600 C into the region at which surface carbon becomes a thermodynamic product (dashed line). For these high temperatures, only H 2, H 2 O, and single-carbon-atom products are thermodynamically predicted, and the observed products followed this behaviour well with selectivity to H 2 and CO of ~50% near equilibrium Temperature profile In this experiment, sketched in Figure 7-4, solid particles directly contact a glowing hot surface at C to rapidly heat and avoid significant char formation. We postulate that volatile organic compounds produced by solid decomposition from rapid heating mix with air and undergo surface oxidation reactions within millimeters of the reactor front surface. These surface oxidation reactions are highly exothermic and produce a rapid increase in temperature as shown. The accompanying photo in Figure 7-4 shows that the front catalyst face remains bright orange when operating with a continuous flow of cellulose (white particles) in air. The gases produced should then undergo surface chemistry such as water-gas-shift or steam reforming in the last 20 mm of the catalytic bed before exiting as predominately synthesis gas Particle size The size of solid particles did not significantly affect processing characteristics. This agrees with models of cellulose particles contacting a hot surface which show that temperature gradients within the particle should become insignificant for particle diameters 1 mm or less [16]. However, larger particles of 1-5 mm have been shown to exhibit high temperature only at the hot surface interface, preventing a large fraction of the particle from slowly heating to form char [17]. This behavior was observed occasionally in experiments as aggregated particles of cellulose as large as 5 mm exhibited steady continuous processing without significant char formation. The particle size required to convert biomass to synthesis gas within these reactors is considerably smaller than particles obtained from harvested biomass. Likely, the biomass or 150

167 solid raw material will have to be mechanically ground into smaller particles prior to processing within these reactors. Figure 7-5 shows a comparison the energy required to generate small particles from wood [18]. Particle sizes used in these experiments were typically <20 U.S. Standard Mesh, and from the figure, the energy required to grind the particles to this size is <10% of the chemical energy contained in the feedstock. Thus, the energy input requirement to grinding biomass to small particles for processing in these autothermal reactors is minimal Polyethylene Other results, shown in Table 7-1, demonstrate that millisecond processing can be extended to other sources such as starch, or the saturated hydrocarbon polymer, low density polyethylene, which produces high selectivity to H 2 (S H ~69%) and CO (S C ~71%). The processing of lowdensity polyethylene exhibited higher selectivity to H 2 and CO than cellulose, starch, and Aspen. The absence of oxygen atoms in the polymer permits significantly higher equilibrium selectivity to H 2 and CO than the oxygen-containing polymers. Another key difference is that highly volatile oxygenated compounds exhibit very high conversion in millisecond reactors [13], while polyethylene produces small hydrocarbons, including olefins, which do not react completely to synthesis gas even at low C/O ratios. Under high operating temperatures, these olefins are believed to be formed from the gas phase decomposition of the parent polymer to radicals and subsequent beta-scission or beta-hydrogen elimination [19-20] Aspen Additionally, a source of wood chips considered for millisecond conversion was Aspen (Populus tremuloides), a fast growing hardwood tree in North America which is about ~2/3 cellulose and hemicellulose, and ~1/4 lignin, with the remaining fraction consisting of uronic acids and extractives, and ash (~0.5%) [21]. Table 7-1 shows that the processing of Aspen particles about 1 mm in diameter exhibited selectivity to H 2 of 51% for 8 hrs using a bed of ~1 mm spheres impregnated with Rh and Ce catalyst. This corresponds to 0.2 g of Rh processing 0.5 kg of biomass of which 2.5 g is ash. Figure 7-6 displays a photograph of an operating reactor of Rh-Ce coated spheres. The exothermic chemistry near the top of the reactor heats the catalyst and support, emitting visible orange radiation. Accumulation of ash occurred during the 8 hour operational period above the Rh-Ce coated spheres. However, even with visible accumulation of ash, the reactor maintains autothermal operation. 151

168 Figure 7-7A shows a scanning electron microscopy (SEM) image of ash obtained from flash volatilization of solid aspen wood particles. In the micrograph, the ash appears to be crystalline, with ~ 1μm crystal planes. Figure 7-7B displays X-Ray Diffraction (XRD) obtained from flash volatilization of solid aspen wood particles. The sharp peaks from the XRD indicate fairly large crystal planes are associated within the ash particles. Additionally, these crystals appear to be comprised of CaO, MgO, and CaCO 3. Figure 7-8 displays energy dispersive x-ray spectroscopy (EDS) of the ash obtained from flash volatilization of solid aspen wood particles. k-band emission of Mg, Ca, O, K, and P elements are mapped with respect to the scanning electron microscopy image (SEM). The relative counts obtained from the EDS maps qualitatively suggest that Ca, Mg, and O are the main elements within the ash, while K and P are present in lower quantities. Furthermore, the spatially mapped count intensities qualitatively indicate that the elements are not homogeneously dispersed within the ash sample. A significant problem in implementing catalytic gasification of biomass involves managing and removing the solid ash which would otherwise accumulate in the reactor. Biomass sources commonly contain the impurities N, S, Cl, K, Na, P, Si, Mg, and Ca, many of which are volatile as elements or compounds at these high temperatures. However, Aspen contains the minerals Ca, K, and Mg which make up over 90% of the produced ash in the form of oxides and carbonates such as CaO, MgO, and CaCO 3 which have very low vapor pressure [22]. EDS and XRD identified these compounds within the accumulated ash above the Rh-Ce coated catalyst, which further supports that these compounds likely do not volatilize. Short-term accumulation of these non-volatile components does not shut down the process, because the ash conducts heat from the catalytic oxidation region to the upper surface where biomass decomposition occurs. However, long-term operation will probably require a process such as a moving (non-fluidized) catalytic sphere bed that continuously removes catalyst from the reactor, separates the spheres and non-volatile ash, and returns the catalyst to the front of the bed. Volatile impurities passing through the catalyst can be removed by adsorption techniques downstream Non-noble metal catalysts: Ni Due to the lower costs, transition metal based catalysts, such as Ni, are attractive substitutes for noble metal catalysts in these CPO systems. Experiments were replicated with cellulose substituting 5 wt% NiCe for 5 wt% RhCe on alumina spheres. H 2 and CO selectivities similar to RhCe were observed for as long as 10 hours. However, Ni-based catalysts have been shown to accumulate carbon during the reforming of oxygenates, and further testing is required to show that Ni is a viable catalyst for long-term biomass processing [23]. 152

169 7.4 Conclusion This method has the potential for smaller, simpler production of clean synthesis gas. Reactor systems operating with millisecond residence times are at least an order of magnitude smaller than conventional systems and exhibit high power densities of ~5 kw/l of catalytic reactor volume (calculated for cellulose at C/O~1.0 producing synthesis gas for a fuel cell operating with 50% efficiency). At these conditions, approximately 2/3 of the fuel value of the biopolymer is retained as synthesis gas. This process appears to be robust with respect to particle size and type of biomass, and the reactor effluent does not contain tars and organics observed from fluidized bed gasifiers. Additionally, operation can occur in air with rapid start-up times of <5 minutes on Rh catalysts that have been operated for 20 hrs without significant evidence of deactivation. However, catalytic oxidation with air does not provide the optimum feed for this process or secondary processing to synthetic fuels. Dilution with N 2 cools the catalyst due to increased convection and increases the size of equipment downstream of the reactor. Solid processing with pure O 2, as well as other considerations such as steam addition or preheat, could permit operation at higher C/O resulting in an effluent stream with higher selectivity to synthesis gas that is more adaptable to secondary processing. Further research of key parameters as well as a more-detailed understanding of the process mechanism should have the potential to improve direct millisecond processing of biomass. 7.5 References 1. A.J. Ragauskas, C.K. Williams, B.H. Davison, G. Britovsek, J. Cairney, C.A. Eckert, W.J. Frederick, Jr., J.P. Hallett, D.J. Leak, C.L. Liotta, J.R. Mielenz, R. Murphy, R. Templer, T.Tschaplinski, Science 311 (2006) A.V. Bridgwater, Chemical Engineering Journal 91 (2003) J.P. Ciferno, J. Marano, Benchmarking Biomass Gasification Technologies for Fuels, National Energy Technology Laboratory, US Department of Energy, Pittsburgh, PA, USA, T. Miyazawa, T. Kimura, J. Nishikawa, K. Kunimori, K. Tomishige, Science and Technology of Advanced Materials 6 (2005) J.R. Rostrup-Nielsen, Science 308 (2005) A.G.W. Bradbury, Y. Sakai, F. Shafizadeh, Journal of Applied Polymer Science 23 (1979) A.L. Brown, D.C. Dayton, J.W. Daily, Energy & Fuels 15 (2001)

170 8. D.S. Scott, J. Piskorz, M.A. Bergougnou, R. Graham, R.P. Overend, Industrial & Engineering Chemistry Research 27 (1988) J.R. Salge, B.J Dreyer, P.J. Dauenhauer, L.D. Schmidt, Science 314 (2006) D. Wang, S. Czernik, D. Montane, M. Mann, E. Chornet, Industrial & Engineering Chemistry Research 36 (1997) M. Inaba, K. Murata, M. Saito, I. Takahara, Energy & Fuels 20 (2006) M. Asadullah, K. Tomishige, K. Fujimoto, Catalysis Communications 2 (2001) P.J. Dauenhauer, J.R. Salge, L.D. Schmidt, Journal of Catalysis 244 (2006) R. Horn, K.A. Williams, N.J. Degenstein, L.D. Schmidt, Journal of Catalysis 242 (2006) Outokumpu Research Oy, HSC Chemistry, Version 4.1, C. Di Blasi, Polymer International 49 (2000) C. Di Blasi, Chemical Engineering Science 51(10), (1996) M. T. Holtzapple, A. E. Humphrey, J. D. Taylor, Biotechnology and Bioengineering, 33 (1989) H. Sinn, W. Kaminsky, J. Janning, Angewante Chemie International Edition 88 (1976) W. Kaminsky, M. Predel, A. Sadiki, Polymer Degradation and Stability 85 (2004) M. Heitz, E. Capek-Menard, P.G Koeberle, J. Gagne, E. Chornet, R.P. Overend, J.D. Taylor, E. Yu, Bioresource Technology 35 (1991) M.K. Misra, K.W. Ragland, A.J. Baker, Biomass & Bioenergy 4 (1993) D.K. Liguras, K. Goundani, X.E. Verykios, Journal of Power Sources 130 (2004)

171 Table 7-1. Selected experimental data for the millisecond reforming of solid particles. Cellulose Starch Aspen Polyethylene Experiment # Fuel Properties Avg. Particle Size (μm) Ash (wt %) < 0.01 Water (wt %) < 0.01 Exp. Conditions Support Foam Foam Spheres Foam Spheres Foam Carbon/oxygen Mass Flow (g/hr) Residence Time (ms) T at 10 mm ( C) T at 30 mm ( C) H Selectivity (%) H H 2 O C Selectivity (%) CO CO CH C 2 H 4 and C 3 H 6 < 0.1 < 0.1 < 0.1 < H 2 /CO Selectivity was defined as (C or H atoms in product)/(c or H atoms in converted fuel). Conversion was >99%. All experiments were conducted at 1 atm with air stoichiometry, N 2 /O 2 =

172 Air Solids Hopper Water Water Cooling Jacket Subwoofer Servo Motor Thermocouple (10mm) Catalyst Thermocouple (30 mm) Heat Shield Insulation Water Condenser Sample Port (Non-Condensable Gases) Water To Incinerator Condensate Collection Figure 7-1. Schematic of solid feed and reactor system. 156

173 10 8 Cellulose (230 μm) Volume Fraction (%) Cellulose (20 μm) Polyethylene (370 μm) Aspen (780 μm) Starch (690 μm) Diameter (μm) Figure 7-2. Particle size distribution of the different solid feedstocks. Particle size was determined using light scattering. ( ) indicate the mass averaged particle size. 157

174 1200 T T T ( o C) 600 No Carbon Carbon C/O Figure 7-3. Top) Measured temperature at 10 mm, T 10 ( ), and 30 mm, T 30 ( ), from the top of the catalytic bed during the processing of ~230 μm particles of cellulose. Bottom) Selectivity to H 2 ( ), CO ( ), and CH 4 ( ) from cellulose. All solids enter the reactor in air at room temperature converting within 70 ms of gas residence time, and no process heat is added. S C and S H are defined as (C or H atoms in product)/(c or H atoms in converted fuel). Dashed lines represent thermodynamic equilibrium calculations based on T 30. Error bars represent 95% confidence intervals. 158

175 Figure 7-4. Gas temperature of cellulose ( ) processing at C/O of 0.7 ( ) and 0.9 ( ) with reaction diagram of volatile organic compounds (VOC) undergoing oxidation, steam reforming (SR: VOC + H 2 O H 2 + CO), water-gas-shift (WGS: H 2 O + CO H 2 + CO 2 ), and cracking reactions. Photograph: Front face (0 mm) of millisecond reforming of ~230 μm particles of cellulose in air. Photograph was taken by Paul Dauenhauer and Scott Roberts. 159

176 Figure 7-5. Comparison of size reduction energy requirements on basis of specific area increase [18]. Particle sizes used in these experiments were typically <20 U.S. Standard Mesh. From the graph, the energy required to grind the particles to this size were typically <10% of the chemical energy contained in the feedstock. 160

177 10 mm 10 mm Ash Exothermic Reforming Zone Endothermic Reforming Zone 10 mm Figure 7-6. Photograph of an operating reactor of Rh-Ce coated spheres (LEFT). The exothermic chemistry near the top of the reactor heats the catalyst and support, emitting visible orange radiation. Photographs (RIGHT) of ash obtained from the reactive flash volatilization of aspen wood particles over Rh-Ce coated spheres (top and side view of reactor). Ash accumulates above the top of the Rh-Ce coated spheres. 161

178 A B CaO Counts Per Second CaCO 3 CaO CaCO 3 MgO CaO MgO MgO CaO CaO θ (Degree) Figure 7-7. Scanning electon microscopy (SEM) image of ash obtained from flash volatilization of solid aspen wood particles (Panel A). Panel B displays an X-Ray Diffraction (XRD) obtained from flash volatilization of solid aspen wood particles. 162

179 Figure 7-8. Energy dispersive x-ray spectroscopy (EDS, 15 kv accelerating voltage) of ash obtained from flash volatilization of solid aspen wood particles. k-band emission of Mg, Ca, O, P, K elements are mapped with respect to the scanning electron microscopy image (SEM). Relative counts are indicated in the upper right corner of each mapped element. 163