1 INVESTIGATION OF IRON OXIDATION KINETICS FOR SOLAR FUEL PRODUCTION VIA CHEMICAL LOOPING By RICHARD CRAIG STEHLE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013
2 2013 Richard Craig Stehle
3 All great efforts derive fro m acts of passion. My passion is solely fueled by my b eautiful and extraordinary wife
4 ACKNOWLEDGMENTS I would like to express my upmost appreciation and gratitude to Dr. David Worthington Hahn, not only for his guidance in this dissertation as my advisor but for his incredible representation of an exceptional individual who demonstrates professionalism in all endeavors and compassion to those around him.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .............. 4 LIST OF TABLES ................................ ................................ ................................ ......................... 7 LIST OF FIGURES ................................ ................................ ................................ ....................... 8 ABSTRACT ................................ ................................ ................................ ................................ 11 CHAPTER 1 INTRODUCTI ON ................................ ................................ ................................ ................. 13 Introduction into Solar Thermal Energy Storage ................................ ............................ 13 Favoring Iron Based Materials ................................ ................................ .......................... 15 Hydrogen Production from the Steam Iron Process ................................ ...................... 17 Overview ................................ ................................ ................................ ....................... 17 Kinetics ................................ ................................ ................................ .......................... 19 Carbon Monoxide Production from CO 2 ................................ ................................ .......... 24 Syngas Production ................................ ................................ ................................ .............. 26 2 FUNDEMENTAL SCIENCE AND BACKGROUND ................................ ....................... 29 Mass Spectrometric Methods ................................ ................................ ............................ 29 Historical Development ................................ ................................ ............................... 29 System Layout and Vacuum Systems ................................ ................................ ...... 29 Ionization ................................ ................................ ................................ ....................... 31 Mass Analysis ................................ ................................ ................................ ............... 34 Ion Detection ................................ ................................ ................................ ................. 36 Raman Spectros copy ................................ ................................ ................................ ......... 38 3 EXPERIMENTAL METHODS ................................ ................................ ........................... 42 Low Temperature Hydrogen Production ................................ ................................ ......... 42 Experimental Design ................................ ................................ ................................ ... 42 Data Acq uisition System ................................ ................................ ............................. 45 Experimental Procedure ................................ ................................ ............................. 51 High Temperature Fuel Production (H 2 and CO) ................................ ........................... 52 Experimental Design ................................ ................................ ................................ ... 52 Data Acq uisition System ................................ ................................ ............................. 58 Experimental Procedure ................................ ................................ ............................. 63 4 CURRENT RESULTS AND DISCUSSION ................................ ................................ ..... 65 Hydrogen Production ................................ ................................ ................................ .......... 65
6 Low Temp erature Production ................................ ................................ ..................... 65 Production and kinetic rates ................................ ................................ ................ 66 Raman Spectroscopy species identification ................................ ..................... 74 Oxide layer development ................................ ................................ ..................... 77 High Temperature Hydrogen Production ................................ ................................ 82 High Temperature Syngas Production ................................ ................................ ............ 85 CO 2 Production ................................ ................................ ................................ ............ 86 Oxide Layer Morphology ................................ ................................ ............................. 93 Conclusions of Iron Oxidation from Species Splitting ................................ ................... 96 5 CONCLUSION OF STUDY AND CONCENTRATED SOLAR REACTOR IMPLEMENTATION ................................ ................................ ................................ ............ 99 Combined CO H 2 Production ................................ ................................ .......................... 10 0 Iron Ferrite Oxidation ................................ ................................ ................................ ........ 105 Concentrated Solar Fuel Production Economic Analysis ................................ ........... 109 Summary Thoughts ................................ ................................ ................................ .......... 114 LIST OF REFERENCES ................................ ................................ ................................ ......... 115 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ..... 123
7 LIST OF TABLES Table page 3 1 List of set flow rates during oxidation experiments. ................................ .................. 43 3 2 Calibration constants used to quantify measured species from mass spectrometer signal responses. ................................ ................................ ................... 49 3 3 Calibration constants used to quantify measured species from mass spectrometer signal responses. ................................ ................................ ................... 62 3 4 List of set flow rates during oxidation experiments at higher temperatures (800 1200C). ................................ ................................ ................................ .................. 64 4 1 Produced hydrogen volumetric flow rates for each of the 3 hours of reaction time. ................................ ................................ ................................ ................................ .. 67 4 2 Kineti c constants calculated from the Arrhenius forms displayed in Figure 4 4. ................................ ................................ ................................ ................................ ....... 71 4 3 Calculated oxide growth rates from measured hydrog en production rates. ......... 80 4 4 Produced CO volumetric flow rates for each of the 3 flow rates of CO 2 during a reaction time of 1hr. ................................ ................................ ................................ .... 87 4 5 Kinetic constants calculated from the Arrhenius forms displayed in Figure 4 17. ................................ ................................ ................................ ................................ ..... 90 5 1 Production rates of hydrogen and CO at subsequent temperature and flow rates. ................................ ................................ ................................ ............................... 101 5 2 Production rates of hydrogen and CO during simultaneous reaction within the same reactor. ................................ ................................ ................................ ................ 104 5 3 Produced CO volumetric flow rates for each of the 3 flow rates of CO 2 during a reaction time of 1hr. ................................ ................................ ................................ .. 106
8 LIST OF FIGURES F igure page 2 1 Layout of Mass Spectrometer system components and arrangement. ................. 30 2 2 A molecule is identified at a mass to charge ratio of 88 at different ionization energies with different fragmentation patterns at each ion ization energy. ............ 32 2 3 For different ionization energies absolute intensity may not change even though the relative intensity does. ................................ ................................ ............... 33 2 4 Quadrupole layout with concentric rods with applied RF and DC voltages. Adjacent rods are 180 out of phase with one another. ................................ ........... 35 2 5 Faraday Cup ion detector component schematic. ................................ .................... 36 2 6 Discrete Dynode Electron Multiplier that component layout that benefits from current gain for faster response times when compared to the Faraday Cup. ...... 37 2 7 Continuous Dynode Electron Multiplier component layout that is less sensitive than the discrete dynode and is more commonly us ed in applications today. ................................ ................................ ................................ ......... 38 2 8 Relationship between a Raman scattered Stokes shift and the incident radiat ion. ................................ ................................ ................................ .......................... 40 2 9 Raman Spectrometer schematic layout for the implemented instrument in this dissertation ................................ ................................ ................................ ..................... 41 3 1 Schematic of the Low Temperature thermal reactor implemented for steam iron kinetics investigation. ................................ ................................ ............................. 43 3 2 Iron rod supported within Inconel tubing by stainless steel insert (positioned vertically in reactor). ................................ ................................ ................................ ....... 44 3 3 Layout of thermocouple locations for reactor temperature control. ........................ 45 3 4 Signal ratios of hydrogen to nitrogen from mass spectrometer sampling plotted versus mass controller flow ratio. ................................ ................................ ... 47 3 5 Signal ratios of hydrogen to argon from mass spectrometer sampling plotted versus mass controller flow ratio. ................................ ................................ ................ 47 3 6 Signal ratios argon to nitrogen from mass spectrometer sampling plotted versus mass controller flow ratio. ................................ ................................ ................ 48 3 7 Hydrogen signal percent in relation to water signal that represents water fractionation effect. ................................ ................................ ................................ ......... 50
9 3 8 Dissociation temperature, enthalpy change and pumping loses as a function of absolute pressure for oxygen release from an iron oxid e matrix. ...................... 52 3 9 Gibbs free energy change with respect to temperature for Reactions 3 1 and 3 2. ................................ ................................ ................................ ................................ .... 54 3 10 Moles of produced hydrogen and CO with respect to temperature. ...................... 54 3 11 High temperature thermal reactor design for oxidation of iron that can be implemented for hydrogen production, CO 2 production or syngas production. ... 57 3 12 Iron monolith supported within high temperature furnace by stainless steel noose. ................................ ................................ ................................ ............................... 58 3 13 Signal ratios of hydrogen to helium from mass spectrometer sampling plotted versus mass controller flow ratio. ................................ ................................ ................ 59 3 14 Signal ratios of hydrogen to argon from mass spectrometer sampling plotted versus mass controller flow ratio. ................................ ................................ ................ 60 3 15 Signal ratios of CO to helium from mass spectrometer sampling plotted versus mass controller flow ratio. ................................ ................................ ................ 60 3 16 Signal ratios of CO to argon from mass spectrometer sampling plotted versus mass controller flow ratio. ................................ ................................ ............................. 61 3 17 Signal ratios of argon to helium from mass spectrometer sampling plotted versus mass controller flow ratio. ................................ ................................ ................ 61 3 18 CO signal percent in relation to CO 2 signal that represents CO 2 fractionation effect. ................................ ................................ ................................ ................................ 62 4 1 Species trends from real time mass spectrometer sampling of N 2 Ar, and H 2 ... 65 4 2 Production rate of hydrogen for each hour with respect to temperature. .............. 68 4 3 Arrhenius plot of first hour oxidation for hydrogen production. ............................... 69 4 4 Arrhenius form comparison of hydrogen production for extent of reaction hour to hour. ................................ ................................ ................................ ............................. 70 4 5 Hydrogen production rate with respect to water concentration at 800 K. ............. 73 4 6 Log log plot of hydrogen production rate with respect to water concentration. .... 74 4 7 Raman signal responses for iron oxide reference standards. ................................ 75 4 8 Resulting Raman spectra for oxidized species after hydrogen production. .......... 76
10 4 9 EDS images that present the produced oxide layer over the range of temperatures. ................................ ................................ ................................ .................. 78 4 10 Oxygen to iron ratios measured from EDS signals along the extent of the oxide layer. ................................ ................................ ................................ ...................... 79 4 11 SEM images of developed oxide layer with the presence of spallation. ............... 81 4 12 Hydrogen produced from Iron oxidation under the influence of steam. ................ 83 4 13 Arrhenius curves for the oxidation of iron under the influence of steam. .............. 84 4 14 Species trends from real time mass spectrometer sampling of He, Ar, CO and CO 2 Flow rates of CO 2 are labeled in mL/min. ................................ ................. 86 4 15 CO Produced from Iron Oxidation Under the Influence of CO 2 ............................. 88 4 16 Arrhenius curve for the oxidation of iron under the influence of CO 2 supplied at 12.5 mL/min. ................................ ................................ ................................ ............... 89 4 17 Arrhenius form comparison of CO production at different flow rates of CO 2 ....... 89 4 18 Comparison of Arrhenius curves for low temperature (700 temperature (1000 2 splitting. ................................ ................................ 92 4 19 Resulting Raman spectra for oxidized species after CO 2 splitting. ........................ 93 4 20 CO 2 splitting. ................................ ................................ ................................ ................... 94 4 21 CO 2 splitting. ................................ ................................ ................................ ................... 95 4 22 Oxygen to iron ratios measured from EDS signals along the extent of the oxide layer following CO 2 splitting. ................................ ................................ .............. 96 5 1 Gibbs free energy approximation for chemical equilibrium water gas shift. ....... 103 5 2 CO Produced from Iron Allo y Oxidation Under the Influence of CO 2 ................. 107 5 3 from CO 2 splitting. ................................ ................................ ................................ ........ 108 5 4 EDS mapping of Fe (top), Co (Bt. Right) and Ni (Bt. Left) to illustrate the oxide layer being primarily iron o xide. ................................ ................................ ....... 108 5 5 Resulting Raman spectra for an oxidized iron alloy specie after CO 2 splitting ................................ ................................ ................................ ...................... 109
11 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy INVESTIGATION OF IRON OXIDATION KINETICS FOR SOLAR FUEL PRODUCTION VIA CHEMICAL LOOPING By Richard Craig Stehle August 2013 Chair: David Worthington Hahn Major: Mechanical Engineering Solar driven production of fuels by means of a n intermediate reactive metal for species splitting has provided a practical and efficient pathway for disassociating molecules at significantly lower thermal energies. The fuels of interest are of or derive from the separation of oxygen from H 2 O and CO 2 t o form hydrogen and carbon monoxide. The concept of utilizing thermochemical processes as a means of storing solar energy in the form of an energy carrier demonstrates enormous potential for application in engineering systems. The following study foc uses on iron oxidation through water and CO 2 splitting to explore the fundamental reaction kinetics and k inetic rates that are relevant to these process es M onolith designed laboratory scaled reactors were implemented to investigate reaction temperatures that range from 600 K to 1400 K, In order to properly characterize the reactive metal potential and to optimize a solar reactor scaled up system. The formation of oxide layers on the iron monoliths is concluded to follow a Cabrera Mott model for oxidation of metals In addition, the oxide phase analysis from m icro Raman s pectroscopy is consistent with m agnetite (Fe 3 O 4 ) growth.
12 Kinetic rates where measured using real time mass spectrometry to calculate kinetic constants and estimate oxide layer thicknesse s. Activation energies of 47.3 kJ/mol and 32.8 kJ/mol were found for water splitting and CO 2 splitting respectively which are consistent with the disassociation energies for chemisorbed water and CO 2 The o xide layer structures were processed with high re solution SEM and Electron Dispersion Spectroscopy (EDS) for morphology considerations. The result revealed limitations for consistent oxide growth at su bstantial oxidation thickness ( 10 m) attributed to species spallation. This dependency on extent of oxide layer growth suggests short controlled oxidation steps during redox cycling for continued material integrity. The conclusions of the independent oxidation reactions where applied to experimental results for syngas (H 2 CO) production to demonstrate ideal process characteristics. Finally, an analysis of the economic impact of solar derived fuels is presented.
13 CHAPTER 1 INTRODUCTION To begin this investigation into iron oxidation kinetics for solar thermal energy storage applications, an introduction into the history of iron as an oxidative species w ill be presented that covers a range of necessary topics of discussion. The focus will be on the benefits of iron for solar thermal storage and the theoretical kinetics and kin etic models that have been associated with the multitude of chemical reactions that pertain to iron oxidation. The discussion that follows is presented in a literature review format that outlines the significant topics and results pertaining to the oxidat ion of iron and the general topic of solar thermal storage. Introduction into Solar Thermal Energy Storage The modern societies of the world benefit from the development of technologies and infrastructure that result in the comfort of daily life. The progr ession of technological discovery has been contingent upon our ability to harness the resources provided by the planet into usable energy sources. Fossil fuels have been the most substantial energy source over the past century, where today fossil fuel is r esponsible for more than 80% consumption . While fossil fuel reserves maintain a significant existence, concern has grown for the impact this source of energy has on geopolitical relations, the environment and economic prosperity. The energy demand p rojected over the next 50 years that takes into account the constantly increasing world popu lation along with the increase of living standards of u nder developed nations estimates the doubling of our energy usage over this time frame. Currently the world consumes between 14 16 TW of energy, however if the total became 30 TW by the middle of this century additional sources of energy to relieve the strain on fossil fuels
14 will be essential to deal with the concerns previously stated . While the impact CO 2 emissions have on the ecosystems of our planet continue s to be debated, it cannot be assumed that a significant increase in CO 2 emissions resulting from our need to meet the growing energy standards would have a neutral effect on the environment. Numerous alternative energy sources are being widely studied and tested but when it comes to effectiveness in creating a new energy infrastructure that would produce the necessary energy density to meet demands, solar technology appears to show strong promise. The sun supplies around 100,000 TW of energy to the earth which is nearly 7000 times more than our current need . In other words, 1 hour of direct sunlight contains enough energy potential to provide our yearly energy demand. The inhibiting factor in using solar energy as an energy supplier has to do with storage. Photovoltaic for a consistent energy supply poses a problem for direct conversion of solar energy during times of low solar inputs ( e g. night time) and variable atmospheric conditions ( e.g. weather). An alternative solar collection technique derives from using the process heat obtained from concentrated solar energy. The process heat can be used to drive chemical reactions in order to store the solar energy as a liquid fuel. The resultant fuel source could act as a synthetic hydrocarbon that can be utilized in the current infrastructure for electrical and mechanical energy devices. The benefit of the synthetic fuel or solar fuel would be the opportunity to create a carbon neutral economy. The chemical processes that have become most attractive for fuel production originate with CO 2 and H 2 O that act as oxidants when introduced to metal alloys that have been reduced to low oxide states by thermally concentrated solar heat. During the oxidative
15 process, CO 2 and H 2 O produce CO and H 2 respectively where subsequent water gas shift and Fischer in liquid form. These kind of chemical looping reactions have been studied since the early 20 th century with the inven tion of reactors like the Lane hydrogen p roducer. The driving sourc e for the chemical looping steps is the active metal alloy that must be constantly reduced under high thermal temperatures and re oxidized to reflect a regenerative pr ocess. Studies into materials that would oxidize efficiently while maintain ing their mate rial integrity have been the subject of recent efforts which have focused on a handful of transition metals. Iron oxides and iron ferrites are a popular material under current investigation and are the main focus of this dissertation. Favoring Iron Based Materials Conditions that the reactive material will be subject to are the basis for selection of the oxygen carrier to be used in a two step chemical looping process. The material evaluations are conducted based on oxygen carrying capacity, thermodynami c properties, kinetics, physical strength and melting points . The typical materials under study are metal oxides of Ni, Cu, Cd, Co, Mn, Sn, Cr, Zn and Fe [5,6]. Based upon thermodynamic limitations, certain materials would oxidize with difficulty or at a much slower rate (Ni, Cu, Co) while others would have issue during the reduction step (Mn, Cr) . Many studies have favored the production of iron oxides for this process, with the common iron oxides being magnetite (Fe 3 O 4 ) and hematite (Fe 2 O 3 ) [8, 9]. Iron oxide readily exits in our environment in economically feasible means, and theoretically contains a high redox capacity per mass [10, 11]. In addition to iron oxides, zinc oxides have also been presented as a useful material for these reactions . The difference between the two metals is that an iron oxide based cycle involves non complex reaction
16 steps allowing for less irreversibility, which affects cycle efficiency. Where as zinc is subject to significant irreversibilities due to its h ighly volatile oxide state [13, 14 ]. Though it appears iron is the most convenient metal metal oxide system, certain material issues do arise when exposed to reactor conditions above 750 C [15,16]. These include deactivation of the reactive material and sinter ing, which is a structural issue in the majority of metals. The deactivation of materials essentially limits the capability of a material to uptake and release oxygen due to the loss of specific surface area. Bleeker et al published a review on the effec ts of deactivation in the steam iron process. They elaborated on the relation of sintering to deactivation based on decreased surface are a which would have an increase e ffect with increasing temperature along with material swelling as oxygen is absorbed [1 7] Their results suggested that as relative conversion rates increase, tendency for increased deactivation is enhanced. An additional disadvantage of iron is related to soot formation, especially at low temperature reduction with CO as a reducing agent. In order to address these structural issues, mixed iron oxides or ferrites have been studied [18, 19, 20 ]. Ni, Co, Mg and Zn based ferrites have shown to not only reduce the necessary reactor temperatures but have also in creased melting temperatures [21 ]. The use of the term ferrite here can represent pure iron or a solid solution with iron as the main component. This includes pure iron oxides and mixed metal oxides. If higher reactor temperatures are preferred, mixed metal ferrites that support iron oxides on zirconia substrates can help decrease the impact of deactivation and sintering at these higher temperatures [2 2 ]. The discussion presented in this dissertation focuses on the
17 oxidation kinetics and mechanisms of pure iron but multiple ferrite ma terials are also investigated during high temperature pure thermal reduction applications. Hydrogen Production from the Steam Iron Process Overview Production methods that form hydrogen as an energy carrier have been identified over the past few decades as plausible pathways for renewable energy. The characterization of hydrogen as a renewable energy derives from the potential of obtaining it from water through numerous chemical processes  Electrolysis and nuclear processes have been effective in yiel ding cycle efficiencies on the order of 30% with efforts continuing to develop and improve upon the technologies [24, 25 ] Processes that utilize solar energy for the production of hydrogen follow three pathways ; photochemical, thermochemical and electroc hemical. Combinations of the pathways have been shown with multiple ways of capturing solar radiation [26 ] The interest in hydrogen as a fuel source is inherently clear in that the energy produced from hydro gen heat processes result in by products of H 2 O that will be cycled back through water splitting methods to produce more hydrogen. The interest in water splitting has moved from using electric power from heat s ources to using direct process heat to drive thermochemical reactions as developments in low cost large scale hydrogen production have become more favorable Direct decomposition of water driven by heat occur s at temperatures that exceed 4000C which is not feasible to contain in a reactor [27 ] However, using intermediary materials that act as participants to drive the water splitting in an endothermic chemical transformation would occur at temperatures well within material capabilities noting that 1500C is a common target as an upper limit The steam iron process is a traditional technique and has been used as a hydrogen
18 producer fr om gasified c oal since the beginning of the 20 th century [28 ] The regenerative cycle operates in two stages: an endothermic reduction step to reduce iron ore by a synthesis gas and subseque ntly re oxidizing the iron ore in an exothermic step yielding a highly rich hydrogen fuel. The pure hydrogen produced represents a fuel with low impurities, which is essential for hydrogen based energy systems such as fuel cells [2 9, 30 ]. The interest in quality of hydrogen with fuel cells is centered on anode catalyst poisoning from CO that is common in most PEM based fuel cells. There is some speculation of the possibility for the introduction of fuel cell technology coupled with a hydrogen economy for energy supply as a substitution for fossil fuels in the transportation sector, however there is much more development in fuel cell technology before that can be realized. There are several other uses for hydrogen as a potential energy carrier that inclu de hydrogen combustion and fuel reforming but the more interesting topic is on the actual process of obtaining a hydrogen fuel source. The steam iron process has traditionally be en related to chemical reduction through agents like CO or biogas but the re duction step can also be accomplished in a pure thermal process using concentrated solar energy to drive oxygen release although difficulties may arise with regard to the specific thermodynamic states and phases at re latively high temperatures Funk discu ssed the past efforts in the thermochemical process dating back to the 1960s and the current direction that research in thermochemical h ydrogen production has taken  More recently, Steinfeld wrote an in depth scientific review on solar thermochemical production that focuses on the aspects of concentrated solar radiation  While the overall process kinetics and cycle thermodynamics are important for optim a l water splitting, the present dissertation focuses on the oxidation
19 step and more importantly the fundamental reaction kinetics associated with efficient hydrogen and oxygen separation. Kinetics In order to develop a working model for the two step steam iron process in an actual solar reactor, accurate rate constants from the reaction kinetics shou ld be considered This dissertation investigates fundamental reaction kinetics through experimental observation of a laboratory scale reactor working under controlled oxidation conditions. Identification of reaction mechanisms and pathways is essential in beginning any study into accurate kinetic parameter measurements and is the primary focus in reaction rate mechanisms for this study The steam iron process may be represented as a catalysis like reaction under the Mar s van Krevelen modeling form [31 ], which invokes a surface redox mechanism to describe rates of oxidation reactions on h eterogeneous oxide catalysts [32 ]. Mars van Krevelen mechanisms correlate to gas molecules reacting with lattice oxygen of the catalyst to produce oxides. Their model de picts oxygen on the catalyst surface reacting with aromatic compounds to form oxidation products and leaving the catalyst site in a reduced state [33 ] The reduced catalyst is then re oxidized by gas phase oxygen in a separate step which replen ishes the l attice oxygen [34, 35 ]. Mars and van Krevelen wrote their reaction mechanism in the following form [36 ] : Aromatic compound + oxidized catalyst oxidation products + reduced catalyst (1 1) Reduced catalyst + oxygen oxidized catalyst (1 2)
20 This type of mechanism can be represented by the following proposed four step mechanism for the oxidation reaction step: H 2 O (g) H 2 O (1 3) Fe H 2 O + O OH + OH Fe 2OH + H 2(g ) (1 4) (1 5) (1 6) In the above equations, V represents vacancies in the lattice structure of the metal surface, and O represents atomic oxygen in the lattice of the metal oxide. This proposed mechanism can be backed by an electron spectroscopic study done by Roberts and Wood where they showed evid ence of OH and O species being formed simultaneously during the interaction of water vapor. Their results suggest that the first step of water dissociation would be followed by interactions between surface hydroxyls to form chemisorbed oxygen [3 7 ]. Between these two steps, the authors suggest that a discrete chemical phase would exist within the lattice as proposed by the above mentioned mechanism. Among the different types of kinetic models that have been represented for oxidation reaction kinetics JMAK and Cabrera Mott models have been referred to most frequently and show the best relation in regards to nucleation growth and potential difference. R esearchers have considered the Johnson Mehl Avrami Kolmogorov (JMAK) model for successive reduction o xidation cycles (steam i ron) for hydrogen generation [38, 39 ]. The JMAK theory has been widely used and applied to solid state processes
21 to relate chemical kinetics to nucleation and growth rates. The JMAK authors demonstrated dependence of reaction curv es to rate o f growth and rate of nucleation in their study with Avrami giving a more in depth look into the kinetics of phase change in a series of publications. It was shown how JMAK kinetics correlate to crystallization and precipitation in solid state phase transformation [40,341 ] More importantly in regards to this dissertation, others have gi ve n insight into heterogeneous chemical reactions like gas solid decompositions or oxidations based on the model [42 ] Such a model predicts the fraction of re acted material, f as an exponential function of time in the form: (1 7) where N is the Avrami exponent (generally between 0.5 and 4), and k is the global rate constant which is dependent on both nucleation pr ocesses a nd oxide growth Cabrera and Mott publicized their theoretical and experimental work on the theory of the oxidation of metals which focuses on oxide film development and the dependence it has on oxidation kinetics [43 ]. The Cabrera Mott model is dependent upon the presence of an electric field within the oxide film due to a contact potential difference between the adsorbed oxyge n ions and the metal surface [44 ]. As oxygen or water is dissociatively adsorbed (i.e. chemisorbed) onto a metal surface, the metal ac ts as an electron donor and a double charge layer is produced. The oxidation rate is then controlled by the diffusion of oxygen anions into the oxide layer from the surface and the diffusion of metallic cati ons away from the bulk metal [45, 46 ] The resul ting oxide
22 formation then generally takes place within the bulk oxide layer. This process is initiated by an extremely rapid oxidation to establish the oxide film and electric field. The initial oxidation process is transient and continues until a critica l thickness is achieved (~100 A) where growth slows or nearly stops When the surface of the oxide is exposed to available atomic oxygen, electrons pass through the oxide from the metal. This movement can be characterized by thermionic emission or tunnel ing effect mechanisms and should occur much more rapid than ionic motion. The absorbed oxygen atoms would then be converted to O ions forming a f ield across the oxide layer which would be maintained until quasi equilibrium is achieved [47 ] The potentia l formed across the layer should be independent of oxide layer thickness allowing electrons to pass through the layer until equilibrium is reached. A multi step mechanism that reflects the prescribed Cabrera Mott model above for the oxidation of reduced ir on b y water vapor can be expressed in the following form [48 ] : H 2 O (ads ) + V O + H 2(g ) (1 8) 4O + 8e 4O 2 Fe (bulk) +2 + 2e 2 Fe (bulk) 2 Fe +3 + 6 e 4O 2 + 2Fe +3 + Fe +2 3 O 4 (1 9) (1 10) (1 11) (1 12) Reaction (1 8 ) represents the global reaction (see for example 1 3 through 1 6) for the chemiso rption of water and the hydrogen production step. For the current study as detailed below, reaction (1 8 ) is considered the rate limiting step; hence k eff = k 8 for
23 the reaction rate data measured over the investigated temperature range  At the typical reacti on temperatures of interest (>1000 C ), the mobility of the oxygen and metallic ions is expected to be very high, hence diffusion limitations are not considered significant except for oxides thickness exceeding several microns, as discussed below. For the rate limiting condition of water chemisorption, an effective one step reaction rate may be expressed in the form : (1 13) where hydrogen production is measured in the units mols/cm 2 s, and the water concentration is in mols/cm 3 [49 ] Hydrogen production is per unit surface area, which is considered as the actual surface area of the reactive metal oxide (i.e. after the initial rap id oxidation of the native reduced metal). The effective rate constant is given by the Arrhenius form below, (1 14) with the pre exponential constant A and activation energy E a For the rate limiting step given by water chemis orption, one may consider k eff = k 8 with the expectation of an activation energy consistent with water dissociation on the iron oxide surface. For water splitting in a wstite/magnetite cycle presented by Charvin et al it was speculated that hydrogen pr oduction rates may be limited by the growing oxide thic kness due to steam transfer  This is consistent with Cabrera Mott kinetics
24 which state that thermodynamic equilibrium exists at the oxide metal interface and at the oxide surface resulting in the diffusion of oxygen through the oxide layer under a concentration gradient. This gradient would be inversely proportional to oxide layer thickness and wou ld grow based on parabolic law rate kinetics. The competing effects of kinetics and diffusion have been noted in other studies as well. Nieses et al. explored steam oxidation for hydrogen production using a ceramic substrate coated with zinc ferrite [50 ]. They found that the rate controlling step changes with reaction time, with evidence of internal diffusion through the product playing a role, while further concluding that the reaction is controlled by the chemical reaction initiating the water splitting before the formation of a continuous product layer. An Arrhenius expression was found to describ e the temperature dependency of the peak reaction rate. Svoboda and co workers examined hydrogen production for reduced iron and steam based on thermodynamic equilibrium  From 600 to 800 K, they reported an overall conversion to magnetite with 94.8 t o 80% hydrogen con version rates, respectively For moderately low temperatures, they suggest that the steam react ion is limited by kinetics. This dissertation continues the study into effective reaction kinetics for the oxidation reaction step of the ste am iron process for kinetic constant calculations and reaction mechanism identification Carbon Monoxide Production from CO 2 The debate on the stabilization of greenhouse gas concentration in the atmosphere with the focus on climate change and the burning of fossil fuels has led to development efforts in CO 2 capture for sequestration or conversion. Sequestration methods propose capturing CO 2 and storing the gas in geological formations underground or within the oceans [51 ] These techniques still have high energy
25 consuming needs and the permanent storage capabilities cannot be guaranteed. The conversion of CO 2 to CO would lead to synthesis gas or liquid fuel production and would present a much more promising alternative to direct CO 2 sequestration. Not only does this demonstrate the possibility to limit carbon emissions but aims to producing synthetic fuels that would alleviate dependence on fossil fuel consumption. CO 2 decomposition with metals has been studied extensively over the past half cent ury using chemical, photochemical and biological methods that pertained to full CO 2 conversion into C and O 2 or forming CO for reforming into synthesis gas or methano l [52, 53, 54 ] The direction of these studies was in improving surface activity of the r eacting site for increased conversio n rates. At low temperatures (< 300 C) it has been reported that full conversion of CO 2 into C and O 2 would occur with 100% conversion efficiency [55,56 ] Converting CO 2 into O 2 and a solid form of carbon would have pos itive effects on [57 ] The formation of CO from the chemisorption of CO 2 on metal surfaces has been given greater attention due to the interes t in the water gas shift and it s relation to chemical fuel production. The research over the past decade has moved away from the studies in low temperature CO 2 decomposition toward high temperature CO 2 splitting with an emphasis on solar thermochemical processes, much like the before mentioned section on hydrogen production. When compared to water splitting, CO 2 splitting is much more thermodynamically favorable at higher temperature (> 800C) but the production of CO has also been shown to be kinetically slow [58 ] The kinetic mechanisms for this type of oxidation should be consistent with what has been previously discussed with oxidation of metals with water [59 ] A study by Menecier et al has shown formation of magnetite being the only possible oxide product
26 from full oxidation under CO 2 with the kinetic limitations being driven by the di ffusion of iron ions through the oxide layer due to potential differences between the oxide surface and iron iron oxide interface [60 ] Loutzenhiser et al have demonstrated thermodynamic and kinetic analyses for sy stems where the active metal is zinc oxid es and compared them to an iron oxide system. The thermodynamic results showed conversion to CO being favorable to similar temperatures for water splitting and the kinetic limitations being consistent with common oxide growth rates [61, 62 ] Venstrom et a l have also done kinetic studies of metal oxidation based on water splitting and CO 2 splitting which has resulted in similar characteristics. Their studies focused on imp roving thermodynamic limitations of the Zn/ZnO process by implementing gaseous Zn in place of solids and kinetics in ceria oxidation through the enhancement of porosity and surface area [63, 64 ]. Coker et al demonstrated in two studies the fast conversion of reduced metal oxides to Fe 3 O 4 upon exposure to CO 2 at 1100C [65, 66 ] Their exa mination was associated wi th the influence of iron present in a solid solution of 8YS Z. It was also realized in their study however, that even though the iron can readily be oxidized by CO 2 the total amount of CO produced was small in comparison to hydrog en production rates. These studies and their understanding give basis to combin in g water splitting and CO 2 splitting as a simultaneous process to improve upon the independent chemical reactions for solar fuel production. Syngas Production The continued research in storing sola r energy as fuel has led to the popular concept of producing syngas with the potential of processing it into liquid hydrocarbon fuels that can be implemented into the current fossil fuel economy. The utilization of solar energy has two competing disciplines for fuel production that derive from
27 photochemistry and thermochemistry. The benefit in thermochemistry over photochemistry can be realized from the ability of a single solar concentrating reactor being implemented for simultane ous CO and H 2 production which in contrast, pho tovoltaic and photo catalysis are limited [ 67 ]. Smestad and Steinfeld did a review on cost analysis and thermodynamic efficiencies for photochemical and thermochemical fuel production techniques and disused t he potential for thermochemical cycles being much more efficient and less costly with the greatest benefit being related to liquid hydrocarbon fuel production for the transportation sector  An additional study done by Loutzenhiser et al has shown thr ough second law analysis, indications for achieving high solar to chemical energy conversion efficiencies that demonstrate economic dominance over the competing solar fuel production techniques [ 68 ]. The discussion in the previous sections of this chapter focused on hydrogen production and CO production as independent processes which eludes to the possibility of combining the products as synthesis gas. The benefits in combining these two processes as a single step go beyond the positive implications for r eactor implementation. There have been experimental studies done on the interactions of water vapor when exposed to gas mixtures contacting CO 2 in the presence of alloys [69 ]. Kim and his colleagues demonstrated in their study on system analysis for fuel production from splitting both CO 2 and H 2 O together definite advantages over splitti ng the species independently [70 ]. Through their analysis, process efficiencies would increase in the combined process along with the possibility of increased fuel yields Loutzenhiser and Steinfeld also highlighted this consideration in their thermodynamic cycle analysis for syngas production with the main focus being on the effects of CO 2 concentrations [71 ].
28 Stamatiou, Loutzenhiser and Steinfeld have also illustrated the consistency in reaction dependencies over cycle time for the simultaneous oxidation step with the independent reactions where initially quick surface reactions occur upon exposure followed by a slower ionic diffusion controlled regime [72, 73 ] The fo cus of the w ork done in this dissertation on syngas production is associated with how the competing reaction mechanisms of the individual oxidation processes along with possible side reactions will influence the overall process. The produced syngas molar ratios between H 2 :CO 2 has significant interest for efficient Fischer Tropsch conversion and it is apparent that varying CO 2 /H 2 O flow ratios will have an effect on production concentrations. This concludes the literature review on solar thermochemical oxidation for H 2 CO 2 and syngas production. The following chapters will now diverge from the discussion of thermochemical processes by giving an in depth introduction in the physical science behind experimental techniques used for completion of this diss ertation followed by application of experimental concepts and designs.
29 CHAPTER 2 FUNDEMENTAL SCIENCE AND BACKGROUND The following chapter discusses the theories and chemistry of mass spectrometry which is the analytical technique used to quantify s urface reaction rates and identify reaction mechanisms during the oxidation process of iron by H 2 O and CO 2 The chapter concludes with a short introduction on the physical process of Raman s pectroscopy, which is utilized as a species identifier in this di ssertation. Mass Spectrometric Methods Historical Development Mass s pectrometric theories have historically focused on the detection of species through identification of mass to charge ratio The basic science in mass spectrometry is to pass a sample throu gh an ionization process that may result in multiple fragmentations of ions. The ions pass through a mass analyzer that prepares the ions for mass detection that can be quantified in a mass spectrum. Following his discovery of the electron, J. J. Thompson developed the first mass spectrometer in 1907 which preceded electron ionization first reported by Dempster in 1918 [74 ] Atomic weight analysis using mass spectrometry first began in 1919 by Aston, for which he was awarded with the Nobel Prize in 1922 [ 75 ] The second half of the 20h century saw the profound evolution of mass spectrometric methods which included the development of time of flight mass analyzers, quadrupole mass filters, ion trap analyzers and tandem mass spectrometry [76, 77 ] System L ayout and Vacuum Systems The standard layout of a mass spectrometer system can be seen in Figure 2 1 with the driving force behind the technique being the vacuum system. The vacuum
30 system acts a means to minimize contamination in the resultant mass spectr a, to help control the high voltages on the detector, and most importantly to prevent ions from colliding with gas molecules. Figure 2 1 Layout of Mass Spectrometer system components and arrangement. The mean free path, defined by below in E quation 2 1, dictates the average distance an ion will travel before colliding with another molecule and this distance must be much less than the distance from the source, through the analyzer and to the detector. (2 1) T represe nts temperature, k is the Bolt zman n constant, is the effective cross sectional area for collision and P is pressure with typical operating pressures on the order of 10 6 torr. Some quadrupole and ion trap instruments can operate at higher pressure levels be tween 10 3 10 4 torr while more sensitive techniques require much lower pressures
31 (10 10 torr) in order to achieve the highest of resolution [78 ] The pumping system operates under differential pumping with a high vacuum pump (10 6 torr) that tends to be a turbo molecular or diffusion pump followed by a second mechanical pump (10 2 torr). A diffusion pump has its advantages by being cheap and quiet but takes a long time to achieve vacuum conditions. A turbo molecular pump is the common pump in most m ass s pectrometer systems today despite its potential for being quite expensive. The turbo pump design allows for fast pump down by deflecting gas molecules through the pump by rotating and fixed blades. The initial function of the pumping system is to draw the sample through the inlet into the ionization source. Ionization The ionization process that takes place will result in producing molecular ions (M + M ), protonated molecules ([M+H] + ) and fragment ions. The ionization techniques include chemical ionizat ion that forms ions through proton transfer such as [M+H] + and [M H] and desorption ionization from molecules in solids and solutions for techniques such as Matrix Assisted Laser Desorption Ionization (MALDI) [79 ] However, the historically traditional and most common technique used in most applications today is electron ion ization, defined in Equation 2 2 below. (2 2) The electron kinetic energy for the ionization process tends to be on the order of 70 eV which far exceeds ionization energies of most molecules that tend to be around 10 ev. This results in significant fragmentation, which can be beneficial in obtaining structural information for larger molecules but could eliminate all M + ions from the
32 spectrum. The other ionization techniques stated previously allow for a more gentle ionization, so fragmentation is limited. The kinetic energy of e controls the degree of fragmentation hence when the voltage is lowered fragmentation can be minimized For example, successful utiliza tion is shown in Figure 2 2 to identify an unknown molecule with a mass to charge ratio of 88. As the ionization energy is lowered the mass to charge ratio at 88 begins to increase and fragmentations at 42 and 61 decrease. This allows for identification of the species mass to charge ratio and the characteristics of the species structure based on fragmentation patterns. Figure 2 2 A molecule is identified at a mass to charge ratio of 88 at different ionization energies with different fragmentation patterns at each ionization energy Source: Yost 2010. However, the ionization energy cannot just be lowered to slightly above the comp ound s ionization energy because the amount of ions produced during electron
33 ionization is also dependent upon the electron energy. This is why 70 eV is the typical operating voltage and fragmentation is a necessary by product. Another way of illustratin g this point can be seen in Figure 2 3, which demonstrates that as e energy is decreased, the relative intensity of the M + ion may increase but the absolute intensity will not increase. Figure 2 3 For different ionization energies absolute intensity may not change even tho ugh the relative intensity does Source: Yost 2010. Though electron ionization could result in significant fragmentation, predictable fragmentation patterns lead to structural identification and because of the simplicity of the tec hnique it is the most popular ionization process resulting in large amounts of electron ionization spectra for identifying results. This is the ionization process utilized in the mass spectrometer associated with this dissertation that focuses on samplin g inert gas species with traces of H 2 O, H 2 CO and CO 2
34 Mass Analysis Following the ionization process, the ions enter the mass analyzer which is essentially an interaction between ions and electromagnetic fields. These interactions separate the ions as to charge ratio and can be achieved in a numbe r of different ways. Magnetic s ector analyzers distribute ions along a p ath in a magnetic field, while quadrupole and ion t rap analyzers rely on setting path stability in an AC fi eld [80 ] There are also time of f light mass analyzers and mass analyzers that utilize orbital frequencies in a magnetic field (FTMS). All mass analyzers are designed to detect elemental species that range in mass to charge ratio from 2 200, but the detec tion of higher mass molecules is dependent upon the ionization technique used. Electron Ionization and Chemical Ionization allow for detection of molecules from a mass to charge ratio of 10 to about 1000. Hi gher protein like molecules can be detected foll owing Desorption Ionization with mass to charge ratios being measured up to around 100,000. The most common commercially sol d mass spectrometers utilize a q uadrupole mass analyzer and the system implemented in this dissertation is a quadrupole. The q uadr upole is essentially 4 parallel hyp erbolic or round rods of equal length and diameter positioned symmetrically in line with one another, as shown in Figure 2 4. An RF voltage, V, with a superimposed D.C. voltage, U, is applied to the rods with adjacent ro ds being 180 out of phase to one anothe r. As ions pass through the q uadrupole path way, they begin to oscillate due to the applied voltage and ions with unstable oscillations are filtered out.
35 Figure 2 4 Quadrupole layout with concentric rods with applied RF and DC voltages. Adjacent rods are 180 out of phase with one another. Ions with stable oscillat ions are defined by Equation 2 3 where the RF voltage, V, is adjusted for the scanning of specific masses, m, while the field radius, r o are held constant. (2 3) As the voltage is swept, the superimposed D.C. voltage, U, is also swept while maintaining a constant U/V ratio. This ratio can be represented by a slope called an asses that make it through the q uadrupole mass filter are those that exist on this operating line and fall within the stable regime expressed in Equation 2 3 The mass range that the mass filter can analyze is dependent upon the relation of voltage to the field radius and circular o is required. There is a limit physically in obtaining higher voltages
36 and r o resolution in detecting specific masses and decreasing r o will lower the transmission ability of the mass filter. The mass spectrometer model used in thi s dissertation can operate between 800 2000 eV and has been optimized by the manufacturer for greatest detection resolution and transmission based on system requirements. Ion Detection The ions that survive through the mass filter are measured in t wo way s, either with a Faraday cup ion detector or an electron m ultiplier detector. The physical difference in the detection techniques is that the Faraday c up detection is independent of ion mass because it directly measures 1 ion charge per ion, but there is a long time constant during detection due to the small io n currents measured. Figure 2 5 depicts a schematic of the Faraday c up detector in line with a signal amplifier that leads to a data acquisition system. Figure 2 5 Far a day Cup ion detector component schematic. In order to obtain the necessary output voltage into the data acquisition system, the resistance across the amplifier must be large enough to compensate for the low ion
37 current. The large resistance results in large time scales during ion detection. The electron multiplier in contrast utilizes ion current gain to increase the current across the amplifier, which in turn will lower the resistance and decrease the time constant for detection. This gain can be achieved with a discrete dy node electron multiplier or a continuous dynode electron multiplier. Figure 2 6 shows the schematic of the detection process with a discrete dynode where the gain is achieved as ions pass through the dynode in a similar fashion to a photomultiplier tube. Figure 2 6 Discrete Dynode Electron Multiplier that component layout that benefits from current gain for faster response times when compared to the Far a day Cup. The more common technique today is the continuous dynode channel electron multiplier and c an be represented in Figure 2 7 The continuous dynode has its advantages over the discrete because it is less sensitive in air and has a more predictable lifetime. The downside to the electron mult iplier compared to the Faraday c up is that the ion signa l from the electron multiplier is a function of ion mass and ion stru cture. The Faraday c up detector is the more accurate technique for measuring ion ratios due to this reason. The detection technique used for this disserta tion implements
38 both a Faraday c up detector and a continuous dynode electron multiplier. The final step in the mass spectrometer system is data acquisition, which is done today with computer interfaces and is discussed in Chapter 3 of this dissertation. Figure 2 7 Continuous Dynode Electron Multiplier component layout that is less sensitive than the discrete dynode and is more commonly used in applications today. Raman Spectroscopy Raman spectroscopy has become a popular technique for chemical analysis since the invention of the las issertation, Raman s pectroscopy is applied for species identification an d structural reference. Raman s pectroscopy derives from Raman scattering theory or the Raman Effect In 1928, Sir C.V Raman discovered the Ra man Effect for which he describes as a type of scattering light that has a different wavelength than the incident radiation. Scattering of light occurs naturally when incident radiation encounters a non homogeneity and electrons within the object oscillat e at the same frequency as the incident wave [81 ] The oscillations are known as an induced dipole moment that in turn become a source of
39 electromagnetic radiation and release light. The majority of the released light is scattered at the same frequency as the incident radiation and is identified as elastic scattering. What Raman discovered was the possibility of inelastic scattering with an emission wavelength different to the incident light. What occurs to drive inelastic scattering are quantum shifts in the vibrational modes of a molecule that can lower the photon frequency or increase it with the classification of decreased frequency as a Stokes shift and the increased frequency as anti Stokes shift. The magnitudes of the inelastic shifts are define d in the equations below in wave number (cm 1 ) that include the incident wavelength ( ) and the wavelength of the scattered light ( ). (2 4) (2 5) These shifts in frequency are species dependent which enti tles Raman s pectroscopy as a unique species identification technique. Raman scattering results from the change in polarizability of atoms with vibrational motion in which atoms that are farthest apart and have the least interaction with one another would have the greatest polarizability change Likewise, atoms closest together would have the greatest interaction with one another and have the lowest polarizability. The oscillation of atoms is what causes the chang e in polarizability which signifies a species as being Raman active; hence Raman activity requires that the polarizability changes with the active vibrational mode. Both
40 Stokes and anti Stokes shifts can be generated from inelastically scattered light b ut Stokes scattering is much more likely to occur due to the probabi lity of finding a molecule in a reduced vibrational state. Because anti Stokes scattering is scattered light at a higher frequency than the incident radiation, the molecule before interac tion with the incident light would have to be at an excited energy state and would release energy to the emitted photons as it decays after interaction. So even though it is typical for both Stokes and anti Stokes to occur simultaneously, the intensity of the Stokes scattering would be much greater than the anti Stokes scattering. For this reason Stokes lines are usually adopted for detection using Raman s pectroscopy. Figure 2 8 is a graphical representation of Raman scattering technique that emphasizes the Stoke shift. Figure 2 8 Relationship between a Raman scattered Stokes s hift and the incident radiation Source: Hahn DW 2010.
41 The implementation of Raman s pectroscopy in this dissertation utilizes a helium neon laser with a wavelength of 632.8 nm that analyzes samples of iron species after oxidation for oxide layer cha ra cterization. The iron samples are sectioned at specific lengths and positions along the rod and a re positioned within the Raman s pectr ometer as depicted in Figure 2 9 Figure 2 9 Raman Spectrometer schematic layout for the implemented instrument in this dissertation Source: Hahn DW 2010. This concludes the introduction into the physical science behind the experimental techniques used in this study of the fundamental oxidatio n kinetics and speciation of iron oxides The following chapter depicts the implementation of experimental design and processes utilized in this work.
42 CHAPTER 3 EXPERI M E NTAL METHODS The focus of the experimental techniques centered on the constr uction and implementation of two thermal reactors. The first reactor was designed for low temperature (670 875C) application into hydrogen production. The second reactor was for high temperature oxidation (700 1200C) for the production of hydrogen, CO 2 and syngas. The following procedures outline the construction and implementation of the two reactors, respectively. Low Temperature Hydrogen Production Experimental Design The intended application for the low temperature thermal reactor was to develop a redox cycle that produces hydrogen from waste heat or process solar energy with the by product being CO 2 as a result of chemical reduction with CO as an oxygen releasing agent. The design and experimental procedure for the study into low temperature hydr ogen production was noted in a recent publication by Stehle, Bobek and Hahn [ 48 ]. The objective of the design was to construct and instrument a lab oratory scale d reactor to measure fundamental chemical kinetic rates and process efficiency for the steam iro n oxidation reaction. The reactor was designed to carefully control the flow rates of water and bulk gases, the temperature of the reactor, and the surface area of the reactant metal. A monolithic reactor was developed in order to control these parameters and to represent a steady state reaction condition. The resultant design contains a steam generator a pre heat section, a main reactor core, and a post reaction exhaust heater as illustrated in Figure 3 1
43 Figure 3 1 Schematic of the Low Temperatu re thermal reactor implemented for steam iron kinetics investigation. Liquid water is supplied to the system by means of a syringe pump (BASi) that is passed through a well mixed water bath heated at 90C. This allows the water to reach a temperature close to the boiling temperature of liquid water before entering the steam generation process. The supply water is then passed through a 1.59 mm I.D. stainless steel tube that is heated at an outer wall temperature of 250C by heating t ape. The liquid water is vaporized to steam, and is then introduced to a flow of a preheated inert gas mixture of nitrogen and argon. The inert gases are supplied to the system by mass flow controllers. Table 3 1 lists the gas and water flow rates, with all gas volumetric flow rates corresponding to a state of 298 K and 1 atm pressure. Table 3 1 List of set flow rates during oxidation experiments. Species Nitrogen (5.0 UHP Grade) (cc/min) Argon (4.8 UHP Grade) (cc/min) Liquid Water ( l/min) Flow Rate 100 200 12.5
44 The steam gas mixt ure enters a 12.7 mm I.D. stainless steel section that is heated on the outer surface to the same temperature as the reactor section temperature by a 400 W heater coil The pre heater tubing section is packed with stainless steel beads to maximize heat transfer to the inert gas/steam mixture and to ensure a uniform mixture This preheating section allows for the steam g as mixture to enter the reactor section at t he desired reaction temperature. The reactor test section contains a 6.35 mm O.D. elemental iron rod (Surepure, 99.5% Fe) that is positioned uniaxially within a 12.7 mm I.D tube ( 705 Inconel) to create an annular reactor space with a hydraul ic diameter of 6.35 mm. The entire length of the reaction zone is also heated on the outer surface by two 400 W heater coils. A stainless steel fitting is inserted at the inlet of the reactor section, which functions to center the iron rod in the annular reactor space. As shown in Figure 3 2 the fitting is press fit into the Inconel housing, and has an inner hole that equals the iron rod diameter with a small tolerance for easy insertion and removal of the rod. Between the inner and outer diameters of the fitting, an array of uniformly spaced holes w as machined to allow for a uniform flow of the gas/steam mixture into the annular reactor space. The gas mixture reacts with the iron rod to form an iron oxide layer and pure hydrogen gas. Figure 3 2 Iron rod supported within Inconel tubing by stainless steel insert (positioned vertically in reactor). The produced hydrogen gas along with the inert gases and unreacted steam leave the reactor and pass through a heated exhaust section that acts as a therm al
45 buffer downstream before the gases leave the system. This section ensures that the temperature of the rod in the main reactor maintains a uniform temperature throughout The ent ire reactor is wrapped with high temperature ceramic insulation for thermal stability, such that the overall reactor vessel functions as an isothermal furnace. A series of type K thermocouples (TC) are positioned along the reactor for accurate temperature readings at each of the main sections, as shown in Figure 3 3 The thermocouples are used in conjunction with PID controllers to maintain the reactor temperature at the desired value. Figure 3 3 Layout of thermocouple locations for reactor temperature control. Data Acquisition System The reactor region is approximately 35.6 cm long, and at the exit of the reactor section a heated quartz capillary tube is inserted to sample and transport a portion of the product gases to a mass spectr ometer for real time analysis ( 16 cc/min sample rate ) An electron ionization quadrupole mass spectrometer (Hiden model HPR 20) TC TC TC
46 setting of 800 V. The resultant gas mi xture from the reaction is measured by the mass spectrometer and the resultant real time, mass to charge ratios of each species are recorded. This allows for the measurement of hydrogen produced in volumetric flow rate per unit area. The analytical measur ement of hydrogen is made from the ratio of the hydrogen signal to that of the nitrogen signal, with the nitrogen at a known volumetric flow rate. A calibration curve was conceived for mass to charge signal ratios of hydrogen to nitrogen signals at known volumetric flow ratios of hydrogen to nitrogen. From the curve, a calibration constant was found that relates the signal ratios to volumetric flow ratios for hydrogen to nitrogen. This calibration process was repeated for hydrogen to argon and argon to n itrogen, which was then used as an on line quality control for all experiments. Namely, the recorded argon to nitrogen signal mu st be consistent with the known argon to nitrogen flow rates, as set in the mass flow controllers. The calibration data was ob tained by flowing hydrogen, nitrogen and argon through the thermal reactor at known flow rates. The reactor temperature during these tests was 300 C and each data point was recorded by averaging the resultant signal over 2 minutes. The calibration for hy drog en was accomplished by flowing argon and nitrogen at constant flow rates of 200mL/min and 100mL/min respectively while adjusting the flow of hydrogen from .5mL/min to 12mL/min at specific intervals Figure 3 4 shows hydrogen/nitrogen signal ratios vers us flow ratios, with the slope of the plotted curve corresponding to the calibration constant. Figure 3 5 displays the same result for hydrogen/argon and was done as a check to ensure accuracy in the calculated hydrogen signal from experiment.
47 Figure 3 4 Signal ratios of hydrogen to nitrogen from mass spectrometer sampling p lotted versus mass controller flow ratio Figure 3 5 Signal ratios of hydrogen to argon from mass spectrometer sampling p lotted versus mass controller flow ratio 0 0.5 1 1.5 2 2.5 3 3.5 4 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 H 2 /N 2 Signal Ratio H 2 /N 2 Flow Ratio 0 0.5 1 1.5 2 2.5 3 0 0.025 0.05 0.075 0.1 0.125 0.15 0.175 H 2 /Ar Signal Ratio H 2 /Ar Flow Ratio
48 Calibration betw een argon and nitrogen was also done for internal data acquisition consistency and is displayed in Figure 3 6 Both gases were contr olled at flow rates that alternated betwee n 50mL/min and 500mL/min that en sured a broad spectrum of flow ratios. Before each experiment, the signal ratios measured by the mass spectrometer were checked for appropriate relation based on the calibration data at the set flow rates. This would help prevent system error during expe rimental tests and to ensure no mass spectrometer component failure. Figure 3 6 Signal ratios argon to nitrogen from mass spectrometer sampling p lotted versus mass controller flow ratio The calibration constants obtained are displayed in Table 3 2 an d have been checke d for accuracy periodically to e nsure no changes between the relative signals measur ed from the mass spectrometer occur. The calibration constants represent the relative sensitivity of the mass spec system between any two species For e xample, as 0 0.5 1 1.5 2 2.5 3 3.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Ar/N 2 Signal Ratio Ar/N 2 Flow Rate Ratio
49 shown in Table 3.2, equal flow rates of H 2 and N 2 would result in a H 2 signal that is 11 times greater than the N 2 signal. Table 3 2 Calibration constants used to quantify measured species from mass spectrometer signal responses. Species Calibrated H2 N2 H2 Ar Ar N2 Calibration Constant 11.0 3 14.82 0 .69 Additional calibration is needed to take into account the signal response of hydrogen due to the fragmentation or f ractionation of water in to hydrogen ions within the mass spectrometer. During the ionization process, species can fractionate into smaller compounds due to the high ionization energies associated with the process. Because water will fractionate readily into H and OH, creating H 2 in the process, the recorded H 2 signal needs to be corrected for the additional signal of hydrogen originating from actual H 2 O. This was done by obtaining a fractionation constant between the signal of water and hydrogen for a flow containing only inert gas and steam under non reacting conditions (i.e. no iron rod present). Figure 3 7 represents average signals of hydrogen as a ratio to signals of water. Sample tests were monitored with the mass spectrometer where water was passed through the reactor at reaction temperatures without an iron sample present. The purpose of the procedure was to eliminate the presence of hydrogen in order to determine how hydrogen signal relates to water independent of reaction. A wide verity of water concentrations were evaluated to show that the signal r esponse of hydrogen is independent of water concentration but instead an artifact in the mass spectrometer ionization process. This theory holds true for concentrations above two percent and satisfies our experimental concentration of 4%. The calculated fr actionation constant (~4%), which was reevaluated periodically, was then used to
50 determine the amount of fractionated H 2 corresponding to the amount of detected water, which was then subtracted from the total H 2 signal. In practice, the contribution of th e hydrogen signal from fractionated water ranged from about 10% of the total hydrogen signal at 750 K production rates to about 50% of the detected hydrogen signal at the lowest production rates (600 K). Figure 3 7 Hydrogen signal percent in relation t o water signal that represents water fractionation e ffect. An interesting phenomenon worth noting does occur for concentrations below 2% where the fractionation e ffect increases significantly and suggests that a low concentration of a species that could fr agment may display a near zero signal if not accounted for properly. This has no concern for species used in this dissertation but holds merit for species that have low disassociation energies and high fragmentation signals at moderate specie s concentrati ons. 0.00 1.00 2.00 3.00 4.00 5.00 6.00 7.00 0 2 4 6 8 10 12 14 16 18 Hydrogen Signal/Water Signal (%) Water Concentration (%)
5 1 Experimental Procedure A temperature range of 670 K to 875 K was investigated as the temperature for the oxidation process because of the thermodynamic favorability of the reaction in this range. The flow rates for the inert gases were kept constant for all experimental tests (see Table 3 1) with argon at 200 mL/min and nitrogen at 100 mL/min. Liquid water in more detail in Chapter 4). Following vaporization of all water, the reactant stream mole fractions are 0.0536, 0.316, and 0.631 for the water vapor, nitrogen and argon, respectively, entering the annular reactor section. Six experiments were done at each of four temperatures within the investigated te mperature range: 670, 735, 800 and 875 K. For each operating temperature, the experiments ran under oxidation conditions for 3 hours each. Prior to each experiment, a pristine iron rod was sanded using a successive series of grit papers, namely 320, 600 and finall y 1200, and then carefully cleaned with acetone to produce a polished rod free of any surface oxidation. This procedure was done just prior to insertion of the rods into the room temperature reactor. Inert gas flow was then established, and the reactor w as then heated to the desired operating temperature. Once the steady state reactor conditions were realized, a process that generally required in excess of 1 hour, the mass spectrometer was initiated and the experiments were started by initiating the flow of water through the syringe pump. An initial transient in the species concentrations was observed that lasted for only several minutes, after which steady state conditions were realized as discussed below. The mass spectrometer recorded the signals of hydrogen, nitrogen, argon, and water with a temporal resolution of about 4.6 s, and the resultant signals are tabulated
52 for further data analysis. After 3 hours, the water flow was stopped and the entire reactor was cooled back to room temperature under the flow of inert gas only. The iron rod was then removed from the reactor and immediately stored under inert gas for future investigation into chemical phase and oxide layer thickness using Raman spectroscopy, electron dispersive spectroscopy (EDS), and h igh resolution SEM analysis. High Temperature Fuel Production (H 2 and CO) Experimental Design The means to utilize concentrated solar energy for creating synthetic fuels focuses on temperatures that exceed 1000C. Figure 3 8 is a plot created by J rg Petr asch that displays thermal reduction temperatures of iron oxide for different environment pressures at thermodynamic equilibrium. Figure 3 8 Dissociation temperature, enthalpy change and pumping loses as a function of absolute pressure for oxygen release from an iron oxi de matrix Source: Petrasch 2011.
53 At 1 atm, the equilibrium temperature is 2000C and as pressure decreases, equilibrium temperatures decrease but still exceed 1000C. For a thermal re dox cycle to occur, reduction experiments will have to occur at reduced pressures so as to bring the reduction temperature below 1500C. This is due to the melting temperature of iron being 1370C and the melting temperature of iron oxides being on the sa me order set generally less than 1500C. Furthermore, for a reasonable redox cycle to occur in the proper time scale, oxidation temperatures need to be increased as well. If reduction occurs at 1300C, the oxidation temperature should not be so low as t o have substantial down time between reaction steps. In contrast to this, reactor temperatures must also be restricted in exceeding thermodynamically favorable oxidation temperatures. Figure 3 9 shows the change in Gibbs Free Energy for the chemical proce sses displayed below in Equations 3 1 and 3 2 for a range of temperatures. 4 H 2 O (g) +3Fe (bulk) 4 H 2(g) + Fe 3 O 4 4CO 2(g) +3Fe (bulk) 4CO (g) + Fe 3 O 4 (3 1) (3 2 ) As temperature increases, the change in the Gibbs Energy for hydrogen production increases and beco mes greater than zero around 1250 K. CO production favors higher temperatures during oxidation and becomes more thermodynamically favorable as temperature increases with its Gibbs Energy bec oming less than zero around 900 K. Figure 3 10 uses the Gibbs results of Figure 3 9 to represent the extent of reaction for each oxidation process and displays the product gases in moles.
54 Figure 3 9 Gibbs free energy change with respect to temperature for Reactions 3 1 and 3 2. Figure 3 10. Moles of produced hydrogen and CO with respect to temperature. -100 -80 -60 -40 -20 0 20 40 60 400 600 800 1000 1200 1400 Gibbs Free Energy (kJ/mol) Temperature (Kelvin) Oxidation of Iron with Water Oxidation of Iron with CO2 Chemical Equalibrium 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 400 600 800 1000 1200 1400 Amount of Product (mols) Temperature (Kelvin) Mols of Hydrogen Mols of Carbon Monoxide
55 The amount of hydrogen produced at equilibrium conditions will decrease as temperature increases while CO production will increase with increasing temperature. These figures give basis for the implementation of the thermal reactor during oxidation to investigate a range between 800C to 1200C. A Sentro Tech STT 1700C high temperature vertical tube furnace coupled with a Sentro Tech STT 1200C high temperature horizontal furnace represent the main components of the high temperature reactor. The vertical furnace has an operating maximum temperature of 1700C with an optimum operating temperature of 1500C for extended use. This furnace is sufficient for both oxidation and reduction conditions as the main reactor housing. The heating length of the furnace is 18 inches and supports a 1.5 inch diameter alumina tube that is 3 ft. in length. The furnace uses MoSi 2 heating elements and measures the temperature with a B type thermocouple connected to a 30 segment programmable temperature controller. The horizontal furnace has a maximum operating temperature of 1200C and can continuously run at 1100C. This heating length is 12 inches and supports a 2 inch diameter alumina tube that is 3 ft in lengt h. This furnace uses high temperature metallic heating elements and measures temperature with a K type thermocouple that is also connected to a 30 segment programmable temperature controller. The horizontal furnace is used as a preheater for inert gases and as the heating source for steam generation during oxidation. The system design for the oxidation step differs slightly from the reduction step and the focus of the experimental set up for this dissertation will only be on the oxidation procedure. Ju st as in the low temperature furnace, t he objective of the design was to construct and instrument a lab oratory scale d reactor to measure fundamental chemical
56 kinetic rates and process efficiency for the oxidation of iron. This design also was intended for oxidation conditions with supplied CO 2 along with conditions for a flow of both steam and CO 2 simultaneously Also staying consistent with the low temperature furnace, the high temperature reactor was designed to carefully control the flow rates of water and bulk gases, the temperature of the reactor, and the surface area of the reactant metal. The monolith reactor design was implemented again for parameter control and to represent a steady state reaction condition. The reactor components are still made up of a steam generator main reactor core, and a post reaction exhaust heater. The pre heater section has been replaced by the initial length of the 1700C furnace heating zone because only 10 inches of the 18 inches are utilized for reaction length. The o ther 8 inches are used to increase the temperature of the gas mixture just before contact with the iron surface. An additional length between both furnaces has been added in order to properly transition the flow through the pre heating furnace, into the m ain reactor. This length is heated with heating tape at 300C and is insulated efficiently to maintain thermal stability throughout the length. Figure 3 11 represents the system layout for the oxidation experiment during syngas production. Liquid water is supplied to the system by means of a syringe pump (BASi) that is passed through a 1.59 mm I.D. stainless steel tube. The tube continues into the horizontal furnace set at 1000C wh ere liquid water is vaporized to steam. The tube opens up into a 12.7 mm I.D. Inconel 720 tube while still within the heating zone of the horizontal furnace and is then introduced to a flow of a preheated CO 2 and an inert gas mixture of helium and argon. The CO 2 and inert gases are supplied to the system by mass flow contr ollers. The steam gas mixt ure continues through the connecting section between the furnaces,
57 which is a 12.7 mm I.D. stainless steel tube that is heated at 300C. The steam gas mixture temperature may decrease slightly during this transition between furn aces but the impact should be limited. The mixture enters another 12.7 mm I.D. Inconel 720 tube within the vertical furnace and will travel 8 inches at reactor temperatures before coming into contact with the iron surface. Figure 3 11. High temperature thermal reactor design for oxidation of iron that can be implemented for hydrogen production, CO 2 production or syngas production. The iron monolith support as shown in Figure 3 12, was designed to function similarly to that of a noose. A 1/16in stainless steel tube is wrapped around the iron monolith and is hung down within the Inconel tube and positioned 10 inches into the
58 vertical furnace heating zone. The back end of the stainless steel noose is held in place and sealed by Swagelok fittings. The stea m gas mixture will pass over the hanging iron and react with the surface to produce a H 2 CO and a resultant oxide layer. The produced gases along with the supplied inert gases and left over steam and CO 2 will exit the furnace. T he exhaust species pass th rough the same heated exhaust section from the low temperature reactor that acts as a thermal buffer downstream before the gases leave the system. Figure 3 1 2. Iron monolith supported within high temperature furnace by stainless steel noose. Data Acquis ition System At the exit of the reactor section a heated quartz capillary tube is again inserted to sample and transport the product gases to a mass spectrometer for real time analysis
59 ( 16 cc/min sample rate ) The electron ionization quadrupole mass spect romet er (Hiden model HPR 20) was used fo r select ion monitoring, and the operating emission was set increase from the previous setting was due to a change in ion detection from the electron multiplier to t he Faraday detector. The resultant real time, mass to charge ratios of each species are recorded to produce the resulting volumetric flow rates per unit area in the same fashion as previously done, but because of the existence of new species, additional c alibration was needed. The calibration was reiterated for flows of hydrogen under the presence of argon and h elium and then duplicated for CO under the same conditions. The curves for both cases a re plotted below in Figures 3 13 and 3 14 for hydrogen ratios and Figures 3 15 and 3 16 for CO. The Ar/He signal ratios vs. known flow ratios are also displayed in Figure 3 17 with the resultant calibration constants for all species in Table 3 3. Figure 3 13. Signal ratios of hydrogen to helium from mass spectrometer sampling p lotted versus mass controller flow ratio 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 0 0.05 0.1 0.15 0.2 0.25 0.3 H 2 /He Signal Rato H 2 /He Flow Ratio
60 Figure 3 14. Signal ratios of hydrogen to argon from mass spectrometer sampling p lotted versus mass controller flow ratio Figure 3 15. Signal ratios of CO to helium from mass spectrometer sampling p lotted versus mass controller flow ratio 0.00 0.50 1.00 1.50 2.00 2.50 3.00 0 0.05 0.1 0.15 H 2 /Ar Signal Rato H 2 /Ar Flow Ratio 0 0.05 0.1 0.15 0.2 0.25 0.3 0 0.2 0.4 0.6 0.8 CO/He Signal ratio CO/He Flow Ratio
61 Figure 3 16. Signal ratios of CO to argon from mass spectrometer sampling p lotted versus mass controller flow ratio Figure 3 17. Signal ratios of argon to helium from mass spectrometer sampling p lotted versus mass controller flow ratio 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0 0.1 0.2 0.3 0.4 CO/Ar Signal Ratio CO/Ar Flow Ratio 0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 0 2 4 6 8 10 Ar/He Signal Ratio Ar/He Flow Ratio
62 Table 3 3 Calibration constants used to quantify measured species from mass spectrometer signal responses. Species Calibrated H2 He H2 Ar CO He CO Ar Ar He Calibration Constant 2.8 20.0 0.3 7 0. 76 0.45 The same water fractionation data was used during these experiments but data for the fractionation e ffect due to CO 2 needed to be applied as well, because CO 2 readily fractionates into CO and O creating a false signal of CO. The calculated fractionation signal was determined from Figure 3 18 below and removed from the measured CO signal during data analysis The contribution of the CO signal from fractionated CO 2 ranged from 40% of total CO signal at 1373 K to nearly 90% for production done at 973 K. Figure 3 18. CO signal percent in relation to CO 2 signal that represents CO 2 fractionation e ffect. 0.00E+00 2.00E-02 4.00E-02 6.00E-02 8.00E-02 1.00E-01 1.20E-01 1.40E-01 1.60E-01 1.80E-01 0 0.05 0.1 0.15 0.2 0.25 0.3 CO/CO 2 Signal Ratio (%) CO 2 Concentration (%)
63 Experimental Procedure A temperature range of 700 C to 1200 C was investigated for three sets oxidation processes with the first being the steam iron process. The flow rates for the inert gases were kept constant for all experimental tests with argon at 200 mL/min and helium at 100 mL/min. Liquid water was supplie altered throughout each experimental run between 12.5 for concentration d ependence data. Two experiments were done at each of five temperatures within the investigated temperature range: 70 0, 800, 900, 1000 and 1100 C. For each operating temperature, the experiments ran under oxidation conditions for 1 hour. Prior to each experiment, a pristine iron rod was sanded using a successive series of grit papers, namely 320, 600 and finally 1200, and then carefully cleaned with acetone to produce a polished rod free of any surface oxidation. This procedure was done just prior to insertion of the rods into the room temperature reactor. Inert gas flow was then established, and the reactor was then heated to the desired operating temperature. Once the steady state reactor conditions were realized, a process that generally required in excess of 1 hour, the mass spectrometer was initiated and the experiments were started by initiating the flow of wate r through the syringe pump. An initial transient in the species concentrations was observed that lasted for only several minutes, after which steady state conditions were realized as discussed below. The mass spectrometer recorded the signals of hydroge n, helium, argon, and water with a temporal resolution of about 4.6 s, and the resultant signals are tabulated for further data analysis. The water flow rate was increased to 25 mL/min after 20 minutes and increased again to 50 mL/min at the 30 minute mar k. The flow rate was
64 lowered back to 25 mL/min and then 12.5 mL/min in 10 minute increments. After 1 hour, the water flow was stopped and the entire reactor was cooled back to room temperature under the flow of inert gas only. The iron rod was then remov ed from the reactor and immediately stored under inert gas for future investigation into chemical phase and oxide layer thickness using Raman spectroscopy, electron dispersive spectroscopy (EDS), and high resolution SEM analysis. This experimental process was reiterated twice more for CO and syngas production under the flow of CO 2 and steam CO 2 respectively for temperatures of 800, 900, 1000, 1100 and 1200C under the same inert gas flow conditions. Table 3 4 lists the gas and water flow rates that were us ed during experimentation, with all gas volumetric flow rates corresponding to a state of 298 K and 1 atm pressure. Table 3 4. List of set flow rates during oxidation experiments at higher temperatures (800 1200C) Species Helium (4.5 UHP Grade) (cc/min) Argon (4.8 UHP Grade) (cc/min) Liquid Water ( l/min) CO 2 ( 4 .0 UHP Grade) (cc/min) Flow Rate 100 200 12.5 12.5 100 200 25 25 100 200 50 50
65 CHAPTER 4 CURRENT RESULTS AND DISCUSSION Hydrogen Production Low Temperature Production The experimental study into hydrogen production for reactor temperatures between 450 K 850 K followed the experimental procedure presented in Chapter 3 that monitored four specific temperatures over a reaction progression of 3 hours. The discussion of results that follows is an elaboration of the low temperature oxidation kinetics and morphology published by Stehle, Bobek and Hahn  A typical 3 hour oxidation process monitored by the mass spectrometer is displayed in Figure 4 1 for a reactor temperature of 700 K. There are three production trends that can be identified from each experimental run and illustrated in Figure 4 1. Figure 4 1. Species trends from real time mass spectrometer sampling of N 2 Ar, and H 2 1.00E-10 1.00E-09 1.00E-08 1.00E-07 0 1 2 3 Signal Response (Torr) Time (hr) Nitrogen Argon Hydrogen
66 The first observed trend is a rapid increase in production that achieves a brief local maximum that is followed by a quick decay for a few minutes (less than ~5 minutes) down to a near steady state production rate. This sudden initial spike of hydrogen production is due to the rapid oxidation of the p ristine, reduced iron surface area, which has been reported previously [82 ]. Consistent with the Cabrera Mott theory, this rapid initial oxidation is generally believed to form an oxide layer of several monolayers (~few angstroms), after which the electri c field is established and the typical field assisted oxidation is established. This establishment signifies steady state production as the second production rate trend portrayed in Figure 4 1. As the experiment continues, the rate of production for hydr ogen remains quite steady, but shows a slight downward trend with time, presumably due to the formation of an iron oxide layer of sufficient thickness to impose some diffusion limitations of the oxygen and iron ion sp ecies within the oxide layer [83 ]. Th is final trend is presumed to continue until diffusion limitations prevent any fu rther surface reactions however this regime was never observed in this experimental procedure Following all experiment s the mass spectrometer data and the iron samples we re analyzed using the techniques described in detail in Chapter 3. The following discussion outlines the process of quantifying kinetic rates from mass spectrometric signals, identifying species from Raman m icroscope signals and oxide layer development fr om the tandem of SEM and EDS analysis. Production and kinetic rates The hydrogen production rate for each experiment was calculated for each 1 hr increment during the 3 hr experiment. Hydrogen production was quantified by calculating the true hydrogen rate using the hydrogen to nitrogen calibration curve and
67 correcting fo r hydrogen signal resulting from water fractionation within the mass spectrometer. The production was then normalized to the available surface area of the reaction zone using the dimensions of the rod. The total available surface area ( DL) was used for this calculation, with L the length of the rod from inlet until the location of the mass spectrometer sampling port, which assumes excess reactant water throughout the entire reactor, and therefore uniform hydrogen generation under a kinetically limited regime. Only 1 to 5% of t he reactant water was consumed within the reacto r over the operating range of 650 to 875 K, which is consistent with this assumption. Given the initial polishing of the rod to a high luster, no corrections are made for effective surface area beyond the si mple geometric surface area. Table 4 1 shows the average production of hydrogen, in volumetric flow rate per unit area, for each reactor temperature. It is readily seen that the production rate at the lowest tem perature investigated, namely 67 0 K, is ess entially constant over the entire 3 hr experiment. Table 4 1 Produced hydrogen volumetric flow rates for each of the 3 hours of reaction time Average Production Rate (mL/min cm 2 ) (RSD) Hour 1 Hour 2 Hour 3 67 0 K 0.0012 (12%) 0.0011 (29%) 0.00098 (19%) 735 K 0.0041 (49%) 0.0035 (46%) 0.0035 (35%) 8 00 K 0.0057 (26%) 0.0050 (24%) 0.0051 (17%) 875 K 0.0097 (10%) 0.0082 (12%) 0.0069 (8%) Volumetric flow defined at 298 K and 1 atm. For these relatively low production rates, the expected oxide film thickness linear growth rate, as elaborated below, is about 350 nm per hour. It is concluded that such a thickness presents only a negligible barrier to diffusion of the reactive ions when
68 compared to the kinetic rate limiting step, namely water dissociation, during the first hour. As expected, however, as the reaction temperature increases the hydrogen produc tion rate increases, with the 875 K rate more than 8 fold greater than the 67 0 K v alue. T he results are shown in Figure 4 2 for all experimental conditions. Figure 4 2 Production rate of hydrogen for each hour with respect to temperature. Of further significance, fo r the highest temperature of 875 K and to a lesser degree at 8 00 K, there is a noticeable decrease in the hydrogen production rate for the second and third hours of the experiments. For example, the hydrogen production rate during hour 3 is 29% less than t he hour 1 production rate at 875 K, while the hour 3 rat e is decre ased by about 10% at 8 00 K in comparison to the first hour of production. The resulting one hour film thickness at 875 K is estimated at about 2.7 m (see Table 4 2 ). Thicknesses on the order of several microns are apparently enough to produce moderate r esistance to diffusion and thereby to affect the overall reaction rate due to 0.E+00 2.E-03 4.E-03 6.E-03 8.E-03 1.E-02 1.E-02 650 700 750 800 850 900 Hydrogen Production (mL/cm 2 min) Temperature (K) Hour 1 Hour 2 Hour 3
69 transport effects. The results express that diffusion becomes important for increased oxidation steps at high reactor temperatures, furthermore, factors attributed to material st ructure and behavior presented later in this chapter suggest that short oxidation steps are favored for long term species integrity. For this reason, only the first hour data is used to calculate the effective rate kinetic coefficients, thereby approximat ing the ideal conditions of a purely kinetic limited reaction attributed to the dissociative water adsorption step. The kinetic model for rate limiting conditions shown below has been applied for calculation of the pre exponential constant and activation energy associated with the effective Arrhenius form discussed in Chapter 1. Based on first order kinetic s the effective rate constant from E quation 1 13 can be determined Figure 4 3 is a plot of the production rates during the first hour with respect to an inverse temperature change. Figure 4 3 Arrhenius plot of first hour oxidation for hydrogen production -8.5 -8 -7.5 -7 -6.5 -6 -5.5 0.0011 0.00115 0.0012 0.00125 0.0013 0.00135 0.0014 0.00145 0.0015 Ln(k eff ) Inverse Temperature (1/K)
70 From the linear relation, the values of E a and A are readily determined from the slope and y intercept values, respectively. The same analysis can be done for h ours 2 and 3 and from Figure 4 4 you can see that the kinetic results are similar for the extent of the experiment. The values for activation energy and pre exponenti al factor are tabulated below in Table 4 2 for all three hours of oxidation. The effective activat ion energy only decreases by .6% and 2.8 % for the second and third hours, respectively. In contrast, the pre exponential term A increases by 1 8 % and 36 % ov er the second and third hours, reflecting the increasing role of diffusion and more importantly the effect of spallation This is given further attention to later in the chapter and is the basis for only hour 1 kinetic results being utilized Figure 4 4 Arrhenius form comparison of hydrogen production for extent of reaction hour to hour. -8.5 -8 -7.5 -7 -6.5 -6 -5.5 0.0011 0.0012 0.0013 0.0014 0.0015 Ln(k eff ) Inverse Temperature (1/K) Hour 1 Hour 2 Hour 3
71 Table 4 2 Kinetic constants calculated from the Arrhenius forms displayed in Figure 4 4 Hour 1 Hour 2 Hour 3 E eff (kJ/mol) 47.3 45.7 44.9 A (cm/s) 2.2 1.5 1.2 The activation energy of ~ 50 kJ/mol is considered to be consistent with the dissociation energy of water via chemisorption onto the iron oxide matrix following the initial transient induction period. This is nearly one order of magnitude less than the O H bond dissociation energy of a free water molecule; however, it is known that the formation of a transition state with the metal oxygen bond weakens one O H bond within the molecular water, greatly redu cing the dissociat ion energy [84 ]. In a study of wate r adsorption on epitaxial Fe 3 O 4 (111) by Joseph et al dissociative adsorption was reported with an isosteric heat of adsorption of 65 kJ/mol and a corresponding heat of desorption of 50 kJ/mol [85 ]. While these values do not directly corroborate the activ ation energy for the dissociative step as measured and proposed by Reaction (9), the current activation energy of 50 kJ/mol is very consistent with the associated energies reported for water chemisorption on magnetite. Neises and co workers, who examined hydrogen production in SiC honeycombs coated with zinc ferrite, reported an activation of 110 kJ/mol (10) for the peak reaction rate at the beginning of the water splitting step, as modeled using an Arrhenius approach  While a zinc ferrite system is considerably different than the current Fe Fe 3 O 4 oxidation, comparable activation energies (~50 vs ~100 kJ/mol ), together with the water adsorption studies of Joseph et al., do suggest the importance of the water dissociation step as the potentially rate limiting kinetic step when the oxide layers have not grown sufficiently thick for diffusion and transport rates to become important. This study is continued and duplicated later in
72 this dissertation for high temperature investigation for examination of kinetic limitations beyond thermodynamic favorability regimes. For the rate limiting condition of water chemisor ption, a first order approximation was expressed in relation to water concentration. The concentration dependence during the water dissociation step has been investigated previously by Nieses and coworkers to show zero orde r dependence . Joseph et al have also demonstrated this during the potentially rate limiting kinetic step when the oxide layers have not grown sufficiently thick for diffusion and transport rates to become important  Before obtaining kinetic constants based on this assumption, the water concentration effects were checked to ensure zero order dependency. The procedure for obtaining concentration dependency was to oxidize iron at a constant reactor temperature of 820 K for 1 hour while adjusting species concentrations every 5 mi nutes. The range of water concentration varied between 1% 15% with alternating patterns for each experiment. These tests were done 5 times in order to obtain sufficient data points at each concentration value with t he results plotted in Figure 4 4 Wha t the result s show quantitatively is that as concentratio n increases production rate does increase linearly until about 8% at which production levels off and suggests no dependence upon concentration change. Nieses et al first discussed the dependence of water concentration between 10% and 80% as being on the order of zero, which is consistent with our result for concentrations above 8% . However, for the production results presented in Figure 4 2, the concentration of water was 4% which is within th e linear region of Figure 4 5.
73 Figure 4 5 Hydrogen production rate with respect to water concentration at 800 K The slope of this linear region was used to calculate the order power from the relation below. ( 4 1) The plot of the relation shown in Figu re 4 6 displays an order of ~0.4 which promotes the concentration dependence of being 1/2 order fo r concentrations less than about 8%. Additional insight may be gained by considering the water impingement rates in combina tion with the surface consumption rate. For the average hydrogen production rates, as given by Table 4 1 on the order of 10 15 molecules/cm 2 s of H 2 O are consumed for hydrogen production. This may be compared with a water molecule impingement rate (at 8 00 K) of about 10 22 molecules/cm 2 s. Hence only about 1 in 10 7 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0 5 10 15 Hydrogen Production (mL/min) Water Concentration (%)
74 water molecules striking the surface are dissociatively adsorbed and likewise participate in the oxidation reaction. Such data again supports a kinetic limiting step of the water dissociation further suggesting a lack of dependence on water transport to the surface (i.e. relative insensitivity to the water concentration). Figure 4 6 Log log plot of hydrogen production rate with respect to water concentration Raman Spectroscopy species identification From the production rates of hydrogen, an oxide layer growth rate can be calculated from the overall global mechanism : xFe + y H 2 O y H 2 + Fe x O y (4 2 ) The specific oxide formed is reasonably assumed to be that of magnetite, which is thermodynamically favorable. However, to ensure that this is the true oxide state y = 0.4237x 1.0211 -1 -0.95 -0.9 -0.85 -0.8 -0.75 -0.7 -0.65 -0.6 -0.55 -0.5 0 0.2 0.4 0.6 0.8 1 1.2 1.4 Log(Hydrogen Production) Log ([H 2 O])
75 additional analytical measurements were recorded. Specifically, Raman spectroscopy measurem ents were recorded directly from the oxidized rods and compared to standards of magnetite (Fe 3 O 4 ) and hematite (Fe 2 O 3 ). All Raman spectra were recorded using 632 nm excitation with a dispersive micro Raman spectrometer (JY Horiba LabRam). Figure 4 7 show s Raman spectra of pressed pellets made from standards of these two reference iron oxides. Wstite (FeO) is another lower state iron oxide that can be present after the water splitting reaction; however, it was found that wstite is unstable at atmospheri c conditions below 840 K and r eadily oxidizes to magnetite [86 ]. As observed in Figure 4 7 Fe 2 O 3 produced a series of well defined Raman bands at 225, 293, 411, 613, 1315 cm 1 while Fe 3 O 4 is characterized by a single prominent peak at 664 cm 1 The stru cture in the Fe 3 O 4 spectrum below about 300 cm 1 is considered an artifact due to the high pass filter used to filter out elastic light in combination with the very low overall signal strength of magnetite. Figure 4 7 Raman signal responses for iron oxide reference standards Based on the reference measurements, Fe 3 O 4 has a much lower signal intensity (i.e. reduced signal to noise ratio) as compared to Fe 2 O 3 for similar sample bulk 0 20 40 60 80 100 120 140 160 0 1000 2000 Intensity (a.u.) Raman Shift( cm 1 ) Fe 3 O 4 0 1000 2000 3000 4000 5000 6000 7000 8000 0 1000 2000 Intensity (a.u.) Raman Shift ( cm 1 ) Fe 2 O 3
76 densities, as noted in the total signal comments below. Clearly Raman spectroscopy is more sensitive (i.e. greater scattering cross section) to hematite than magnetite by about two orders of magnitude. Raman spectra were recorded for the oxidized iron rods corresponding to each of the four investigated reactor temperatures, noting that the rods were stored at room temperature under inert gas prior to Raman analysis to avoid any post experiment oxidation. The resulting Raman spectra are reported in Figure 4 8 Figure 4 8 Resulting Rama n spectra for oxidized species after hydrogen production. The peak highlighted by the arrow corresponds to the principle Fe 3 O 4 band at 664 cm 1 which overlaps slightly with the 613 cm 1 Fe 2 O 3 band. The Raman spectra suggest that both magnetite and hematite are present after the oxidation reaction for all 2000 6000 10000 14000 18000 22000 26000 30000 34000 38000 0 500 1000 Intensity (a.u.) Raman Shift ( cm 1 ) 670 K 10000 12000 14000 16000 18000 20000 22000 24000 26000 28000 0 500 1000 Intensity (a.u.) Raman Shift ( cm 1 ) 735 K 2000 4000 6000 8000 10000 12000 14000 16000 18000 20000 0 500 1000 Intensity (a.u.) Raman Shift ( cm 1 ) 800 K 2000 4000 6000 8000 10000 12000 14000 16000 18000 0 500 1000 Intensity ( a.u .) Raman Shift ( cm 1 ) 875 K
77 reactor temperatures. However, the hematite Raman scattering response is about 100 fold greater than magnetite under the current excitation system, while the actual scattering intensities are comparable between the two oxides; hence it is evident that a significantly larger amount of magnetite is present. It is therefore concluded that the iron oxide levels produced by water splitting in the 600 to 875 K temperature range are primarily the oxide magnetite, Fe 3 O 4 Go et al. performed x ray diffraction analysis of the iron oxide state produced following re oxidation by steam for hydrogen production, and likewise reported magnetite as the predo minant oxide phase [87 ]. The current work is further corroborated by the EDS data presented below. Oxide layer development A selection of the oxidized rods were subjected to additional SEM and EDS analysis to further assess the chemical composition as well as the film morphology. For SEM/EDS analysis, the rods were sectioned with a diamond saw, embedded in an epoxy matrix, an d polished prior to imaging. Representative EDS mapping images are presented in Figure 4 9 which clearly show the presence of a distinct oxide layer above the ele mental iron bulk of the rods. To quantify the iron oxide states from the SEM images, EDS p oint scans were taken at three locations within the oxide films: the surface of the oxide layer, the middle of the oxide layer, and near the inner surface of the oxide layer. The oxygen content (O/Fe intensity ratio) at the inner iron iron oxide surface i nterface was at smaller oxygen to iron ratio than that recorded at the outer surface of the oxide layer, meaning there is a presence of a slight gradient in exact oxide state across the oxide layer.
78 Figure 4 9 EDS images that present the produced oxide layer over the range of temperatures. The EDS O/Fe ratios were calibrated by using scans of pure Fe 3 O 4 and Fe 2 O 3 pressed powders, which enabled semi quantitative determination of the oxide state throughout the oxide layer, as shown in Figure 4 10 Based on the calibration, the EDS data reveal an oxide layer that corresponds strongly to magnetite, with a slight decr ease in oxygen content when moving from the outer surface to the inner surface. The EDS results corroborate the Raman spectroscopy results, leading to the conclusion of the specific oxide magnetite as the predominant oxide species produced during water sp litting in these temperature ranges. The slight gradient suggests some minor diffusion effects that are in perfect agreement with the Cabrera Mott model, where the oxygen and iron ions diffuse toward the middle of the oxide layer and react. Hence one exp ects to see near stoichiometric oxide formation near the center for comparable
79 diffusion rates, with some excess iron toward the iron surface, and some excess oxygen toward the outer surface where the actual water dissociation occurs. Figure 4 10 Oxygen to iron ratios measured from EDS signals along the extent of the oxide layer. With the determination of magnetite as the predominant iron oxide chemical phase after water splitting, the reaction mechanism can be writt en in the following global form : 3Fe + 4H 2 O 4H 2 + Fe 3 O 4 (4 3 ) Because four mols of H 2 will be produced for every one mol of magnetite, an effective linear growth rate of the oxide layer can be calculated based on the hydrogen production rates of Table 4 1 Table 4 3 shows the magn etite iron oxide layer linear growth rate in nanometers per second at each reactor temperature. As observed in 1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 1 2 3 Oxygen to Iron Ratio Scan Location (1 Iron Surface) (2 Middle of Oxide Layer) (3 Oxide layer Surface) 600K 650K 700K 750 K Fe3O4 Fe2O3 Linear (Oxide Ratios)
80 Table 4 3 the growth rates range from about 0.1 to nearly 0.8 nm/s for the range of reactor temperatures studied, whic h correspond to values of 0.35 m/hr to 2.7 m/hr at temperatures of 670 and 875 K, respectively. Table 4 3 Calculated oxide growth rates from measured hydrogen production rates. Average Oxide Growth Rate (nm/sec) Hour 1 Hour 2 Hour 3 67 0 K 0.097 0.083 0.076 735 K 0.32 0.27 0.27 8 00 K 0.44 0.38 0.39 875 K 0.75 0.63 0.53 By definition, the iron oxide layer linear growth rates demonstrate the same trends as the hydrogen production rates, in that an increase in temperature shows an increase in growth rate, and as reaction time increases for the greater reactor temperatures, the linear growth rate decreases. While these values represent the effective amount of iron oxide layer produced, it is unclear from these calculations alone as to the degree of uniformity of the iron oxide layers. Along with the EDS images, high resolut ion SEM was done in order to investigate the homogeneity of the produced oxide layers. Representative SEM images are presented in Figure 4 11 for the four reactor operating temperatures. It can be observed from the SEM images that the oxide layer growth is not highly uniform, and that fracturing occurs along the length of the films near the interface as well as vertical fractures. Some of these fractures may originate during cutting and polishing, but the images presented above were selected as represent ative of the typical results. From the above calculations for the linear oxide growth rate, the total oxide layer thickness should increase with reactor temperature. However, such a trend was not always observed in the SEM images, where the higher temper ature
81 growth conditions often produced films of less than expected thickness along with noticeable fracturing and separation between the oxide layer and bulk iron interface. Figure 4 11 SEM images of developed oxide layer with the presence of spallation. The cause of such an effect may be attributed to spallation of the oxide layer at larger film thicknesses (i.e. larger growth rates). Spallation occurs due to shear stresses within the material resulting in the lattice mismatch as well as the modulus mismatch between the elemental iron and the oxide magnetite, causing parts of the material to separate. The occurrence of spallation may be dependent upon both the film thickness and the ra te of growth (i.e. strain rate), which is consistent with the fact
82 that the 670 K oxide layer consistently appears more uniform. The phenomenon of possible oxide layer spallation has significant implications in an actual water splitting reactor operating in a cyclic ma nner of oxidation and reduction. A n optimal growth condition (i.e. combination of rate and time) to mitigate the effects of oxide layer spallation during the hydrogen production process has been identified f or the specific reactor conditions used in the experimental procedures presented here. Based on a series of SEM images taken for multiple reaction times at the different oxidation temperatures, it was determined that the ideal initial oxidation step for o xide layer production was 800 K for 45 minutes. This would allow for necessary growth rate for significant oxide layer development without spallation occurring. High Temperature Hydrogen Production The topic of solar syngas production demonstrates an interesting technique where hydrogen and CO can be produced simultaneously and be combined for additional energy storage processes Based on the research done by collaborators the reduction regime for cyclic reactors is beyond 1000C (1273 K) This sugg ests that oxidation needs to also be within the same temperature for system compatibility and efficiency. Mainly, the temperature difference between each step must be appropriate so to not create significant lag time. Currently, the lower end of feasible reduction temperatures at decreased reactor pressures are in proximity of 1300C (1573 K) Iron oxidation under steam was repeated and compared to previous data to ensure the reactor functionality was adequate and to give insight into higher oxidation tem peratures. In the current research, e x periments were run from 700 10 (973 1273 K) for approximately one ho ur over a surface area of 50.65 cm 2 at three different flow rates of
83 steam. A new, un oxidized rod was inserted into the reactor between each exp eriment. The flow rates were adjusted throughout the hour of oxidation and the experimental results can be visualized in Figure 4 12 Each data point corresponds to steady state production over an average of about 5 minutes. Figure 4 12 Hydrogen produ ced from Iron oxidation under the influence of s team When combined with the previous data se t for temperatures between 400 (670 875 K) in the Arrhenius form an effect of temperature based on chemical thermodynamics can begin to be realized. In the previous Chapter it was stated that hydrogen prod uced from the reaction of Equation 3 1 is thermodynamically favorable based on chemical equilibrium up to temperatures around 1200 activation energy calculated from the slope of the solid line in Figure 4 13 is 47 kJ/mol and is consistent with the low temperature hydrogen production calculation. However, if 0 0.05 0.1 0.15 0.2 0.25 0.3 900 950 1000 1050 1100 1150 1200 1250 1300 Hydrogen Production (mL/cm 2 min) Temperature (K) 12.5 mL/min Water 25 mL/min Water 50 mL/min Water
84 an Arrhenius plot of only the data from 700 (973 1273 K) is made, an activation energy of 88 kJ/mol is found that is represented by the dashed line in Figure 4 13 Figure 4 13. Arrhenius c urves for the oxidation of i ron under the influence of s team This perceived increase in activation energy can be assoc i ated to either a thermodynamic e ffect or the presence of diffusion limitations. From Figure 3 9, thermodynamic favorability for the oxidation of iron under steam exists below temperatures of 950 K at equilibrium and that beyond this temperature, the redu ction of iron oxide by hydrogen becomes favorable. At higher temperatures, the reduction of iron oxide represented by the reverse reaction in Equation 3 1 will compete with the oxidation step causing a decrease in hydrogen production rates. From Figure 4 13, the change in the kinetic data occurs around 950 K which corresponds to the thermodynamic data but the effect on the kinetic results is not what would be anticipated from thermodynamic limitations. The production rate does not increase from -9 -8 -7 -6 -5 -4 -3 -2 0.0007 0.0009 0.0011 0.0013 0.0015 Ln(k eff ) Inverse Temperature (1/K) Arrhenius Curve 600-1275 K Arrhenius Curve 875-1275 K
85 875 975 K but then it rapidly increases for temperatures of 1075 and 1175 K. This is illustrated in Figure 4 13 by the increase in activation energy which more closely corresponds to a transition between a reaction rate limited regime to a diffusion limited regime More data points would need to be gathered to have a better understanding in what is actually occurring but what can be concluded is that the oxidation of iron from steam should not be done at temperatures that greatly This conclusion is made due to the presence of both diffusion limitations along with competing reactions due to thermodynamic constraints. The impact this may have on syngas production is worth noting due to the depen dence of iron oxidation from CO 2 to produce CO. Based up on the same thermodynamic analysis presented in the previous Chapter for chemical equilibrium of Equation 3 2, CO production is favored at elevated temperatures well beyond 900C. The kinetic results provided later in this Chapter along with the expected reduction temperatures around 1300C suggest that syngas production would occur at temperatures beyond the thermodynamic favorability of hydrogen production. In order to optimize the oxidation step for both hydrogen and CO production, an understanding in combined process efficiency is needed. High Temperature Syngas Production How the competing reactions of Equations 3 1 and 3 2 would interact with one another during an oxidation step is quite interesting, not only in how they would coexist simultaneously but also the possible side reactions that may occur. Before moving into syngas production from a supply of both steam and CO 2 the independent reactions should be compared. An in depth analysis into the fundamental kinetics of the steam iron process has already been presented in the previous section but, CO 2 splitting
86 needs to be examined as well. The following section demonstrates an initial study into CO production from CO 2 splitting based on the same procedures for water splitting. CO 2 Production Iron was oxidized under the influence of CO 2 under the same criteria as the high temperature hydrogen producti on experiments. Oxidation tests commenced for 1 hour for three different flow concentrations of CO 2 Figure 4 14 shows the CO 2 splitting process that is monitored by the mass spectrometer interface. The CO 2 flow rate was 12.5 mL/min for the first 30 min as seen in the figure. The flow was then increased to 25 mL/min and 50 mL/min at 5 minute intervals then brought back down to 12. 5 ml/min for the conclusion of the one hour experiment. Figure 4 14. Species trends from real time mass spectromete r sampling of He, Ar, CO and CO 2 Flow rates of CO 2 are labeled in mL/min. 1.00E-09 1.00E-08 1.00E-07 1.00E-06 1.00E-05 0 10 20 30 40 50 60 Signal Response (Torr) Time (min) Argon Helium CO CO2 25 50 25 12.5
87 The oxidation process resembles the previously depic ted hydrogen production reaction with steady state CO production rates occurring throughout the reaction with a small downward trend during the extent of reaction due to diffusion limitations. CO production rates were quantified using the previously descr ibed process and assumptions. Oxidation results from CO 2 splitting are shown below in Table 4 4 for each of the three flow rates of CO 2 and graphically represented in Figure 4 15 for temperatures of 700 (973 1373 K) with each point corresponding to an average production rate over 5 minutes The analysis for CO 2 was done based on the model of kinetic limited reactions presented for hydrogen production. The production rates for CO were quite low and steady over the 1 hour oxidation and correspond to slow oxide growth rates. Even at these elevated temperatures the oxide layer thickness over the one hour experiment should only represent a negligible barrier to diffusion of reactive ions. Table 4 4 Produced CO volume tric flow rates for each of the 3 flow rates of CO 2 during a reaction time of 1hr. Average Production Rate (mL/min cm 2 ) (RSD) CO 2 Flow Rate 12.5 mL/min 25 mL/min 50 mL/min 973 K 0.0053 (14 %) 0.0084 (3 %) 0.0112 (14 %) 1073 K 0.0077 (22 %) 0.0097 (11 %) 0.0145 (8 %) 1173 K 0.0 106 (13 %) 0.0138 (12 %) 0.0206 (17 %) 1273 K 0.0097 (11 %) 0.0130 (11 %) 0.0202 (14 %) 13 73 K 0.0 209 (3 %) 0.0322 (5 %) 0.0552 (7 %) Volumetric flow defined at 298 K and 1 atm.
88 Figure 4 15 CO Produced from Iron Oxidatio n Under the Influence of CO 2 Again, kinetic data was calculated based on rate limiting conditions for obtaining activation energies and pre exponential constants. Figure 4 16 is an Arrhenius plot that corresponds to CO production from a CO 2 supplied flow rate 12.5 m L/min. The activation energy and pre exponential factor was estimated from a linear relation that fit well to the lower temperatures of 700 900 (973 1173 K) but at the higher temperatures, the same pattern occurs that was presented in the high temperatu re hydrogen production where a decrease occurs in the production rate followed by a rapid increase in the production r ate. This occurrence is further discussed later in this chapter. Figure 4 17 shows the three different flow regimes represent a similar Arrhenius relation with the calculated a ctivation energy being between 32 3 8 kJ/mol. The values for activation energy and pre exponential factor are tabulated in Table 4 5 for each of the flow rates. 0 0.01 0.02 0.03 0.04 0.05 0.06 900 1000 1100 1200 1300 1400 CO 2 Produced (mL/min cm 2 ) Temperature (K) 12.5 mL/min CO2 25 mL/min CO2 50 mL/min CO2
89 Figure 4 16. Arrhenius curve for the oxidation of ir on under the influence of CO 2 supplied at 12.5 mL/min Figure 4 17 Arrhenius form comparison of CO productio n at different flow rates of CO 2 -6.5 -6.3 -6.1 -5.9 -5.7 -5.5 -5.3 -5.1 -4.9 -4.7 -4.5 0.0007 0.0008 0.0009 0.001 0.0011 Ln(keff) Inverse Temperature (1/K) -7 -6.5 -6 -5.5 -5 -4.5 -4 0.0007 0.0008 0.0009 0.001 Ln(keff) Inverse Temperature (1/K) 12.5 mL/min CO2 25 mL/min CO2 50 mL/min CO2
90 Table 4 5 Kinetic constants calculated from the Arrhenius forms displayed in Figure 4 17. 12.5 25 50 E eff (kJ/mol) 32.8 31.8 37.9 A (cm/s) 2.1 2.5 2.1 The previous value for activation energy of ~50 kJ/mol for water splitting was associated to the water disassociation step as the rate limiting step due to the relation of the activation energy to the reported energies for water chemisorption on magnetite. The activation energy reported here of ~35 kJ/mol for the CO 2 splitting process can likewise be identified as the disassociation of CO 2 into CO and O The following mechanism represents a Cabrera Mott reaction pathway for the oxidation of iron by CO 2 CO 2 (ads ) + V CO (g) + O ( 4 4 ) 4O + 8e 4O 2 Fe (bulk) +2 + 2e 2 Fe (bulk) 2 Fe +3 + 6 e 4O 2 + 2Fe +3 + Fe +2 3 O 4 ( 4 5 ) ( 4 6 ) ( 4 7 ) (4 8) Reaction 4 hea et al, who demonstrated the activation of CO 2 on cobalt surface s [ 88 ] This study was duplicated by Glezakou and Dang for activation of CO 2 on iron surfaces and the decomposition energy barrier of CO 2 was estimated to be 5.0 kcal/mol [ 89 ] This energy barrier of about ~25 kJ/mol reported by Glezakou and Dang again shows the importance of the species dissociation step as the rate limiting step. A first order approximation was made for the rate limiting condition of CO 2 chemisorption. This approximation was inspired by the water concentration results
91 done for the water splitting process in which it was determined there was an insensitivity to the water concentration. A rate limiting step in both the water and CO 2 dissociation suggests a lack of dependence on species transport to the surface. However the results did show an order of about ~1/2 for water concentration below 8%. Duplicating the analysis for the CO 2 concentration an order of ~1/2 was again found but the first order approximation was maintained. Though first order dependence on concent ration was used for both the water splitting and CO 2 splitting kinetic analysis, the approximate order dependence that was determined duri ng both experimental cases further support s the proposed Cabrera Mott mechanisms for species chemisorption. An inte rpretation of dissociative adsorption by Langmuir kinetics dictates that a species can dissociate into two atoms upon adsorption. This understanding of a single gas phase molecule producing two adsorbed species results in a half order dependence for the a dsorbed species as derived from Langmuir kinetics. In corroboration with the experimental data for species concentration dependence, this analysis can be viewed as additional support for the proposed reaction mechanism. As with the higher temperature wa ter splitting experiments, a kink in the Arrhenius curve was identified that demonstrated an apparent increase in activation energy. Figure 4 18 shows one Arrhenius curve that correspond s to the four temperatures discussed earlier for CO 2 splitting along with a second curve for 1000 and 1100 separately. The change in the kinetic data occurs around 1000 so two additional points at 1050 and 1150 were experimentally measured and plotted with the second curve. The trend with both species splitting proce sses is that once a certain temperature is achieved, production rates begin to change much more rapidly as
92 temperature increases resulting in much higher activation energies. This result appears to reflect a transition between a kinetic limited regime and a diffusion limited regime The activation differed from 50kJ/mol to 88kJ/mol for water splitting and 35 kJ/mol to 130 kJ/mol for CO 2 splitting. It is desirable to add additional data points in this region with any future work. In an attempt to assign an uncertainly to the 130 kJ/mol activation energy, the curve was refit using various subsets of the four data points. Based on the standard deviation of the resulting slopes, an uncertainty of 41 kJ/mol is representative. These values are consistent wit h published kinetic constants that were estimated by fitting diffusion models to experimental data. Mehdizadeh et al found that for their iron silica porous structure used for water splitting hydrogen production, an activation energy of 88 kJ/mol can be e stimated when fitting a first order model that is dependent upon a hybrid of diffusion and boundary controlled mechanisms [90 ] Figure 4 18. Comparison of Arrhenius curve s for low temperature (700 ) and high temperature (1000 1150 ) CO 2 splitting -7 -6.5 -6 -5.5 -5 -4.5 -4 0.0006 0.0007 0.0008 0.0009 0.001 0.0011 Ln(keff) Inverse Temperature (1/K) Activation Energy of 33 kJ/mol Activation Energy of 130 kJ/mol
93 The transition between the two regimes is also interesting in that an apparent nega tive slope (i.e. negative activation energy) is present, which could indicate possible surface morphological changes within this temperature range. There could also be some internal chemistry occurring due to trace amounts of species within the bulk material, such as carbon. This occurrence could be further investigated in order to identify more conclusively what is taking place by evaluating data at additional temperatures or by cycling the temperature for a given experiment Oxide Layer Morphology Raman spectrum were taken of two samples after oxidation experiments run at 19. Represented by the red arrow, the magnetite peak found at a Raman shift of 664 cm 1shows an increased intensity value when compared to the magnetite standard response in Figure 4 7. This again demonstrates that magnetite is the abundant oxide that forms from during CO2 splitting even thou gh hematite peaks are present. Figure 4 19. Resulting Raman spectr a for oxidized species after CO 2 splitting. 0 2000 4000 6000 8000 10000 12000 14000 16000 0 500 1000 Intensity (a.u.) Raman Shift (cm 1 ) 900 0 1000 2000 3000 4000 5000 6000 7000 0 500 1000 Intensity (a.u.) Raman Shift (cm 1 ) 1100
94 As explained in the low temperature hydrogen production discussion, Raman spectroscopy appears to be sensitive to the presence of hematite with the 613 cm 1 peak being an order of magnitude higher than the magnetite peak at 664 cm 1. Because those two peaks show similar intensities when analyzed from the oxidized species, magnetite is the resultant oxide during formation. Figures 4 20 and 4 21 show EDS images for oxidized species from CO 2 splitting with a corresponding SEM image form temperatures of 900 and 1000 respectively. Figure 4 20. SEM image and EDS point scan mapping for a from CO 2 splitting. The SEM images show a well developed oxide layer on the order of ~15 microns, which is expected for oxide development at these temperatures for one hour experiments. The EDS image maps presented below each of the SEM images represent oxygen re sponse in the left image and iron response on the right. Where the
95 oxide appears to be in the SEM image, oxygen shows a strong response detailing the significant oxide layer presence. The right image depicts the bulk iron with a slightly softer response t o iron in the oxide layer as expected. The oxide in Figure 4 21 appears to be of the same thickness even though oxidized at a higher temperature resulting in more production of CO. This re illustrates the dependence of oxide layer integrity to extent of reaction or oxide growth. The SEM image shows a much more aggressive topography for the oxide layer resembling spallation occurrence. As with water splitting, it is recommended that oxidation steps for CO 2 splitting remain short to limit oxide thickness growth and preserve material integrity. The point scan results in Figure 4 21. SEM image and EDS point scan mapping for a 1 from CO 2 splitting. Figure 4 22 shows the EDS O/Fe ratios normalized to the average oxygen to iron ratio at the center of the oxide layer with a slight gradient as oxygen response
96 decreases toward the bulk iron. This reestablishes the Cabrera Mott model as the source reacti on mechanism where the oxygen and iron ions diffuse toward the middle of the oxide layer and react. Figure 4 22. Normalized o xygen to iron ratios measured from EDS signals along the extent of the oxide layer following CO 2 splitting Conclusions of Iro n Oxidation from Species Splitting The objective of the study into monolith iron species ox idation by flow of steam and CO 2 was to identify the fundamental kinetic parameters and reaction mechanisms that pertain to the independent species splitting processes. The results presented in the previous sections detailed activation energies and pre exponential coefficient constants that can be used to model these reactions in larger scaled reactors with either porous or particle based iron species. An understanding of the possible reaction pathways and kinetic dependencies allows for accurate reactor optimization and system efficien cy. The study did find reaction limitations that pertained to extent of reaction and reaction temper ature. As reaction rate increases due to temperature and oxide layer growth becomes more pronounced, production limitations are presented due to a loss in 0.6 0.8 1 1.2 1.4 1 2 3 Normalized Oxygen to Iron Ratio Scan Location (1 Iron Surface) (2 Middle of Oxide Layer) (3 Oxide layer Surface)
97 material integrity along with an increase in diffusion limitations. The study into iron oxide morphology showed that at increased thick ness of magnetite oxide layers species separation or spallation is more prevalent, which can result in the loss of act ive material. Likewise, as diffusion e ffects become the limiting factor on reaction rate, competing kinetic parameters cause increases in reaction rate along with an increase in CO2 splitting, the previously presented results show large increases in reaction rates as temperature changes along with a significant increase in activation energies. The influence this may have on large scale reactor design and system parameters is apparent in that for certain temperatures production rates may vary based on reaction limitations. For a cyclic process where both reduction and oxidation of iron will occur, the temperature of the reduction process is limited based on effective oxygen release rates. The practical low end of reduction temperatures has been previously discussed reduction steps and o xidation steps to be maintained, the difference in tempera ture between each step would need to be minimized. This would more than likely require the oxidation steps for both species splitting processes to be well within the diffusion limited kinetic range even when short oxidation steps are considered for limited oxide layer development. Another factor on temperature which has been discussed throughout this dissertation is the effect thermodynamics could play on process efficiency. Because CO 2 dynamics will be actively favorable for CO generation and the chemical reduction step of CO 2 generation should not play a competing role during the process. However, as
98 temperature rises, water splitting rates will begin to compete with chemical reduction of iron by H 2 It is also possible for thermal reduction to inhibit both oxidation steps at temperatures above 1100 causing a decrease in peak species production. Scheffe et al identified this e ffect in their study on cobalt fer rite zirconia oxidation by splitting water at temperatures above 1100 their obser ved p e a k hydrogen production begins to decrease and is steady [ 91 ] They accredit this to oxygen release occurring throughout the extent of the oxidation steps and identify a secondary issue of the recombination of H 2 and O 2 into steam. Based on the analysis presented here it is recommended to maintain reactor temperatures for the oxidation of iron species to below 1100
99 CHAPTER 5 CONCLUS ION OF STUDY AND CONCENTRATED SOLAR REACTOR IMPLEMENT ATION The implementation of a concentrated solar furnace to drive chemical reactions for fuel generation would have a quite different conception to the reactor designs presented in this dissertation The reactive material would be a porous or powder based species that would have a large reactive surface area. Also due to sintering affects at high temperatures, supporting substrates would be used such as silica, zirconia or YSZ. This would help prevent significant surface area decrease and preserve activit y of the species. Mixed metal species would also be implemented due to the melting temperatures of iron and FeO being within the temperature range for thermal reduction. Mixed metals or iron ferrites that contain other transition metals like cobalt or ni ckel would promote the thermal properties of the reactive material allowing for increased reactor temperatures and species integrity. The reactor concept would also be of interest in regards to the intended products from reaction. The work in the previous section pertained to independent species production and did not discuss the influence these independent reactions would have on one another if they were combined for simultaneous production. This chapter introduces a discussion on combined fuel productio n of H 2 and CO in a simultaneous oxidation step where production can occur in separate parallel reactors within a single furnace or together in the same reactor. A study into the impact of oxide layer development is als o introduced to show the influence a n iron ferrite may have on reaction rates in comparison to pure iron oxidation. This chapter and dissertation will conclude with an analysis on realistic solar reactor implementation for fuel production that will focus on the cost of fuel produced and how it compares with the competitive technology.
100 Combined CO H 2 Production Hydrogen production has been a consistent theme presented in this dissertation as a means for energy storage as a readily available fuel source. Hydrogen can be implemented as a combustion source to drive turbines for electrical power plants and internal combustion engine vehicles. Alternatively, hydrogen also is commonly used in fuel cell applications that p ertain to electric automobile technology. The apparent negative for hyd rogen as a fuel source is the lack of infrastructure that can support a hydrogen economy. The infrastructure of our society has been formulated for hydro carbon based fuels and a solar storage pathway that produces hydrocarbon fuels would be favorable in that regard. As independent steps, water splitting and CO 2 splitting produce their respective products of H 2 and CO which could then be combined or reformed to produce a range of hydrocarbons. As discussed previously, the Fischer Tropsch process is a pop ular pathway for reforming synthesis gas (H 2 and CO) into liquid hydrocarbons. The ratio of hydrogen to CO is important for reforming the produced species into the necessary fuel type, so the amount of mols for both species will determine the hydrocarbon produced. Equation 5 1 gives the Fisher Tropsch reaction step with n being the number of mols of CO. (5 1) The hydrogen to CO ratio that corresponds to the common hydrocarbons produced from this process range between 1.7 and 2.1. Conceptually, as the products of CO and H 2 are formed in separate reactors and are separated from their respective product streams, the species would then converge into a post reactor to form a liquid
101 hydrocarbon based on their concentrations Table 5 1 displays the production rates of hydrogen and CO from 700 1100 with the ratio of production being 8:1 10:1 in favor of hydrogen when temperature and flow conditions are equivalent Table 5 1 Production rates of hydrogen and CO at subseque nt temperature and flow rates. H 2 Production from Water Splitting (mL/min cm 2 ) CO P roduction from CO 2 Splitting (mL/min cm 2 ) Species Flow Rate 12.5 mL/min 25 mL/min 50 mL/min 12.5 mL/min 25 mL/min 50 mL/min 700 0. 0138 0. 0178 0.0219 0. 0053 0. 0084 0. 0112 800 0. 0150 0.0225 0.0334 0. 0077 0. 0097 0. 0145 900 0.0637 0.0892 0.1435 0. 0106 0. 0138 0.0206 1000 0.1336 0.1701 0.2179 0. 0097 0. 0130 0.0202 1100 0.0209 0.0322 0.0552 Volumetric flow defined at 298 K and 1 atm. In a solar furnace design that houses numerous reactors that produce H 2 or CO independently, the furnace can be optimized to produce a hydrogen to CO ratio ~2 by having a larger amount of CO 2 splitting reactors within the solar cavity, increase the flow rate of suppl ied CO 2 to the CO 2 splitting processes to increase local CO production rates or a combination thereof. Regardless, this issue can be easily addressed through furnace implementation and design. The problem that is presented has to do with sacrificing the amount of hydrogen produced to coincide with CO production. The benefit of combined fuel production has been stated that liquid hydrocarbon storage having favor over hydrogen storage but this benefit could be lessoned due to limiting total fuel productio n from the slower kinetics of CO 2 splitting. Instead of having the simultaneous fuel production occur ring in separate reactors, the oxidation steps could commence within one reactor concurrently to maximize fuel output There are several implementation limitations that exist resulting in this simultaneous oxidation reactor design from being favorable such as post reaction
102 species separation and competitive reaction pathways. Since the CO 2 splitting and water splitting would be competing for the same act ive sites throughout the reactive material bead, the individual production steps would be impacted negatively. Post reaction, the separation of CO and H 2 from CO 2 and H 2 O would now have to occur together which would be more difficult however the presence of these four species together represents the interesting side reaction that will take place to influence product gas ratio control. After the formations of CO and H 2 a third reaction is possible in the form of the water gas shift reaction. Since H 2 O, CO 2 H 2 and CO will coexist, the following reaction is likely to occur. H 2 O (g) +CO (g) H 2(g) + CO 2(g) ( 5 2 ) Figure 5 1 represents the Gibbs free energy for the water gas shift reaction with respect to temperature. Coincidentally, the chemical equilibrium point for the reaction is 1100 K, which is where the water oxidation and CO 2 oxidation processes intersect one another. Figure 5 1 also gives the species content for the syngas species after the water gas shift reaction with respect to temperature. As temperatures exceed 1100 K 5 2 favors the reverse reaction and it could be expected that for temperatures above 1100 K, CO quantities in the product stream may be higher and H 2 quantities lower The ther modynamic analysis of Reaction 5 2 was done independent of Reactions 3 1 a nd 3 2. It is important to note that, the thermod ynamic analysis for Reactions 3 1, 3 2, and 5 2 were done based on c hemical equilibrium which is a condition the experimental process may not be able to
103 achieve. Due to the continuous flow of input species, reactant and product gases will constantly be swe pt out of the reaction zone causing the forward reactions to become more spontaneous at elevated temperatures. The chemical kinetics of each reaction will be influenced by higher reactor temperatures so the thermodynamic results for Gibbs free energy mini mization can only be seen as an acknowledgement of how thermodynamic equilibrium may impact reaction rates Figure 5 1 Gibbs free energy approximation for chemical equilibrium water gas shift. Because of the impact of temperature on oxidation under steam and the production under the co mbined presence of steam and CO 2 for simultaneous species splitting Water and CO 2 were supplied at equal volumetric flow rates throughout the experiments with the flow rates adjusted to three different values during the experiment as done in the previous independent reaction experiments. The production rate results are shown in Table 5 2 at each of the three flow rates with hydrogen production on the -30 -25 -20 -15 -10 -5 0 5 10 15 400 600 800 1000 1200 1400 Gibbs Free Energy (kJ/mol) Temperature (Kelvin) Water Gas Shift Chemical Equalibrium 0 0.2 0.4 0.6 0.8 1 1.2 400 600 800 1000 1200 1400 Species Content (mols) Temperatue (Kelvin) Hydrogen Carbon Monoxide
104 right side of the table and CO production on the left. with the values from Table 5 1 for hydrogen production with a slight decrease in prod uction rate which can be attributed to the presence of a competing reaction The same cannot be said about the CO production however There is significant increase in CO production during the higher flow rates of CO 2 especially during the 900 C reactions Table 5 2. Production rates of hydrogen and CO during simultaneous reaction within the same reactor. H 2 From Syngas Production (mL/min cm 2 ) CO From Syngas Production (mL/min cm 2 ) Species Flow Rate 12.5 mL/min 25 mL/min 50 mL/min 12.5 mL/min 25 mL/min 50 mL/min 900 0.0343 0.0626 0.1178 0.0073 0.0742 0.0381 1000 0.0999 0.1228 0.1727 0.0067 0.0124 0.0241 1100 0.3063 0.5660 1.0345 0.0371 0.0754 0.1155 Volumetric flow defined at 298 K and 1 atm. The greater impact is on the ratio between hydrogen and CO at this temperature where the ratio is near 1:1 at 25mL/min of steam and CO 2 and 3:1 for 50mL/min. Because hydrogen and CO 2 are readily available post reaction, it is expected for the water gas shift to occur at these elevat ed temperatures. However, it is evident that temperature alone does not fully drive the reaction, and the concentration of CO 2 has a considerable effect on the impact of the water gas shift reaction. As interesting as this appears to be, issues were prese nted during experiments that suggest difficulty in optimizing for accurate product species ratio. Though the input species of steam and CO 2 were supplied at the same rates, the response of steam is quite sensitive and allowed for the CO 2 concentration to be increased or decreased for short periods of time. When the CO 2 concentration was dominate, CO production increased to a much larger steady state value. Because of frequent steam flow
105 oscillations and post reaction species separation design constraints it is recommended to keep these oxidation steps in independent reactors during simultaneous syngas production. Based on the results from water splitting and CO 2 splitting as separate independent reactions, the production of CO and H 2 during the combined oxidation step should be predicted quite accurately and allow for solar furnace design to be implemented for the appropriate product gas ratios. Iron Ferrite Oxidation To date, all oxidation results discussed in this disse rtation wer e the product of interaction with initial pure iron The reduction step that will ultimately drive the thermal process to temperatures that exceed 1300C will limit the effectiveness of iron as a useful material. The higher temperatures will prevent oxid es from reducing by causing them to melt while this could also lead to deactivation of the pure iron core for further oxidation steps. Cobalt oxides have been found to be actively reduced thermally at temperatures as low as 1100C with similar results fo r nickel at slightly higher temperatures. If these alloys can be combined with iron to strengthen the thermal properties of the material while promoting the thermal reduction tendencies, a major hurdle in utilizing solar driven fuel production would be ov ercome. Obtaining iron monoliths with a significant loading of cobalt, nickel or a mixture of both and investigating fundamental oxidation kinetics of those ferrites would be of great interests for the advancement of this technology. How the ferrite oxid ation characteristics compare to pure iron sources would be a major influence when associated to thermal reduction improvement. Because iron would still be the motivating factor in the oxidation process, it is assumed the kinetics would be quite similar ; however, depending
106 on how a material like cobalt interacts with the iron and mediates oxygen transfer; there could be significant effects in production behavior. Surepure monoliths were obtained that are an alloy of iron (55%), nickel (28%) and cobalt (17 species, EDS was done to ensure the alloy loading. These species were prepared and oxidized in the same manner as the previous iron species for CO 2 splitting for temperatures of 800 110 0C. The production rates of CO are displayed in Table 5 3 and illustrated in Figure 5 2. At the lower temperatures, when the kinetics are much slower, the rate comparison to the pure iron CO 2 splitting results from Table 5 1 have only a minor difference that ranges from 68 86% during the 12.5 mL/min supply flow of CO 2 The higher flow rates have an increased relation up to about 96% of the pure iron flow rate. This in not consistent with the 55% loading of iron for the same available surface area and s uggests the other components of the alloy may also be oxidizing to form nickel and or cobalt oxides. Another consideration for this result may be that due to the slow kinetics at lower temperatures, the smaller loading does not have a strong effect on rea ction. Table 5 3. Produced CO volumetric flow rates for each of the 3 flow rates of CO 2 during a reaction time of 1hr. Average Production Rate (mL/min cm 2 ) (percent ratio of alloy production rates to pure iron sample rates) CO 2 Flow Rate 12.5 mL/min 25mL/min 50 mL/min 1073 K 0.0070 (76%) 0.0091 (85%) 0.0147 (95%) 1173 K 0.0079 (68%) 0.0092 (65%) 0.0141 (69%) 1273 K 0.0089 (86%) 0.0109 (92%) 0.0171 (93%) 1373 K 0.0134 (63%) 0.0164 (52%) 0.0253 (47%) Volumetric flow defined at 298 K and 1 atm.
107 Figure 5 2. CO Produced from Iron Alloy Oxid ation Under the Influence of CO 2 At the higher temperature of 1100C, there is a decrease in CO production for the oxidation of the alloy of 50 60%, which is consistent to the loading of iron in the alloy and signifies the oxidation of iron as the dominate mechanism during reaction. This would favor that at lower temperatures, the production rates would be less dependent upon loading. The high activity of the surface of the iron allo ws the production rates to be maintained when kinetics are slower while promoting the thermal properties of the material for increased reduction temperatures. Figure 5 3 show s an SEM image of the oxidized alloy at 1000C on the left with a clear oxide layer developed on the surface of the bulk iron. The image on the left is an EDS species signal map where the highlighted points represent oxygen response. As presented, oxygen response occurs strongly within the oxide and shows almost no response within the bulk material, which is consistent with the previously shown EDS data. 0 0.005 0.01 0.015 0.02 0.025 0.03 1000 1100 1200 1300 1400 CO Produced (mL/min cm 2 ) Tmerperature (Kelvin) 12.5 mL/min CO2 25 mL/min CO2 50 mL/min CO2
108 Figure 5 3. oxidized iron alloy rod from C O 2 splitting. Figure 5 4 displays the EDS signal response map for the 3 species of the alloy with iron being the top image, cobalt the bottom left image and nickel the bottom right. Figure 5 4. EDS mapping of Fe (top), Co (Bt. Right) and Ni (Bt. Left) to illustrate the oxide layer being primarily iron oxide. What the images show is iron being abundantly present throughout the bulk material and oxide layer in the top image but the other two species components only
109 showing strong signals withi n the bulk material from their respective images. From EDS on the bulk alloy, the percentage of species was found to be that of the reported loading but within the oxide only iron was found to be significant. This suggests that the cobalt and nickel do n ot influence or participate in the oxide layer development during the oxidation step and that mixed metal ferrite material used in a solar furnace for species splitting fuel production will predominantly be iron oxides. These iron oxides will continue to be that of magnetite as presented in the Raman spectrum below in Figure 5 5 for an alloy species oxidized at 1100C. Figure 5 5 Resulting Raman spectra for an oxid ized iron alloy specie after CO 2 splitting Concentrated Solar Fuel Production Economic Analysis When addressing the concept of fuel pro duction economic analysis, it is convenient to focus on the cost of hydrogen production. Syngas production is highly dependent upon hydrogen yield and the total cost to run a solar furnac e should have an 0 500 1000 1500 2000 2500 3000 0 200 400 600 800 1000 1200 Intensity (a.u.) Raman Shift (cm 1 ) Fe 3 O 4
110 influence if only hydrogen were produced. The consideration of the following analysis is on technology comparison and how the non commercial, yet emerging technology of concentrated solar hydrogen production competes with the current comm ercial units of solar hydrogen technology. The main pathway for producing hydrogen today is to strip it from natural gas, primarily through methane reformation. This accounts for 48% of hydrogen produced worldwide. Electrolysis accounts for the majority of the remanding hydrogen production and is dependent upon a source of electricity from the burning of hydrocarbons, the utilization of solar, wind and hydro technologies, and nuclear processes. When identifying the competition for concentrated solar hyd rogen production, solar PV electrolysis appears to be the initial opponent that this emerging technology must improve beyond. Though hydrocarbon reformation and electrolysis accounts for the vast majority of hydrogen production and is cost effective, wind and solar electrolysis show s potential to produce larger yields of hydrogen that could compete with global energy demand. With the expectation for solar PV electrolysis to continue to progress in the hydrogen fuel market, concentrated solar hydrogen prod uction must show competiveness with solar PV. Numerous technical reports have been published by National Renewable Energy Laboratory (NREL) that pertain to hydrogen production by electrolysis detailing how the price of electricity far out ways operational and capital costs [92, 93 ] A study done by Levene et al along with a report by Kroposki, Lebene and Harrison projected hydrogen cost from a large scale electrolysis system. These studies investigated common electrolysis devices that require 53 kW h /kg H 2 which are 75% efficient. This is the common commercial electrolyzer proficiency which can produce up to 380,000 kg H 2 /yr  Using the
111 Hydrogen Analysis Tool (H2A), they estimated the cost of hydrogen to be $4.09/kg and $5.40/kg for industrial and co mmercial application respectively. The electricity cost represented 60% of the cost at $0.045/kWh for industrial systems and 68% of cost at $0.069/kWh for commercial. Capital cost, material and operation cost were also calculated but it is clear that el ectricity cost drives the potential of electrolysis units from becoming economically productive. The electricity cost provided was based on average electricity rates from all sources of power not just from solar PV. If solar PV electricity cost were cons idered solely, the price of electricity would be much greater and estimated to be $0.10 $0.15/kWh. The interest in this price of hydrogen originates from the U.S Department of Energy setting a cost target of $2.00 $3.00/kg H 2 for effectiveness in the tran sportation market. For hydrogen cost to be below $3.00/kg, electricity cost would have to be below $.03/kWh. This is not a reasonable target that Solar PV electricity will be able to obtain without much improvement in material and capital cost for curren t PV electricity facilities. NREL has shown through their analysis of effective wind generation though reports that optimize wind electrolysis such as Saur and Ramsdens investigation that using current wind turbine technology $4/kg H 2 can be achieved [95 ] This promotes electrolysis to progress in the direction of wind applicatio ns and away from Solar PV in regards to renewable hydrogen production. Of course capital costs and operation costs could decrease, but i mprovements in that regard do not seem to show much progression in the near future. Noting that solar PV electrolysis suffers a significant challenge for economic efficiency, solar concentrated hydrogen production may present an opportunity for decreased hydrogen cost while maintaining species output. A study done by Graf et al
112 demonstrated that when compared to electrolysis, hydrogen production from thermochemical processes has the potential to produce a greater amount of hydrogen but at increased costs [96 ] This result was based on the inc reased cost of metal oxide replenishing under operation costs. Based on the current investigation presented in this dissertation, current and anticipated advancements in material integrity will allow for increased life time for the reactive material used. When considering th ese two techniques, the thermochemical process possesses an opportunity to greatly increase t o much higher process efficiencies tha n are currently present. Efforts currently being done look to reasonably increase total system efficiency to ~20%. This however cannot be assumed for electrolysis, which has produced ~70% efficiencies for an extended period of time. PV efficiencies also seem to have met a commercial limit around 20% with most market solar cells ranging from 8 15 %. By implying the an anticipated near futur e cycle efficiency of 20% for a concentrated solar furnace design while maintaining current PV and electrolysis system characteristics, the following analysis hopes to show the prom ise solar thermal chemical fue l production presents. The previous studies mentioned have analyzed electrolysis systems that produce hydrogen on the order of 1000 kg/day with a system efficiency of 75% 1000 kg/day would require a power requirement of 2300 kW, however the interest her e is in smaller s olar systems that run on a peak power supply of 660 kW. For a 75% efficient electrolyzer and a 15% efficient solar PV array this would represent a hydrogen production rate of 45 kg/day, if the system ran for 24 hours per day as described in the above mentioned studies. If electricity from PV was being used to operate the process and produce hydrogen on site, the PV array would only be able to operate at peak
113 operation for about 8 hours per day. This would suggest a production rate of 15 kg/day. For a concentrated solar furnace, the input energy would be 66 0 kW of solar energy with a syste m efficiency estimated at 20% which would correspo nd to 26.4 kg/day of hydrogen. When comparing capital cost, operation and maintenance, material and o ther variable costs the two systems would actually be quite comparable having similar price. The difference is strictly the electricity cost for the electrolysis process or the cost of operating a solar PV array. The before mentioned H2A analysis on curr ent electrolysis technology used for hydrogen production estimated an industrial price of $4.04/kg H 2 This analysis projected electricity costs from all sources of energy and not just solar PV. If solar PV electricity was the sole contributor, electricity costs would more than double Doubling the electricity costs would increase the price of hydrogen to $6.45/kg H 2 resulting in a $96.75/day cost for the 15 kg/day system. In regards to the solar furnace, i f 60% of the cost was removed from the H2A analysi s based on current industrial electricity fees, then the price of the hydrogen from the solar furnac e would be $1.64/kg H 2 Because the solar finance would operate at a power over 6 times greater than the electrolyzer system, it was assumed the capital and variable costs were should be scaled up based on a factor of 3.3. The price of hydrogen from the solar furnace was estimated to be $4.53/kg with a daily operation cost of $120/day In conclusion, when compared to an electrolysis system, a ther mochemical redox system under similar operating conditions would produce 76% more hydrogen with only a 24% increase in cost. This result is based on the assumption that the system components would be of similar operation and lifetime along with the solar concentrating dish /dishes being on the same order of cost as an electrolyzer (~$25,000 /dish ).
114 Summary Thoughts This investigation into iron oxidation by species splitting for solar fuel production has hoped to present a reasonable process to produc e energy sources that can compete in the current market of renewable energy with an ultimate goal of relieving world consumption of fossil fuels. An efficiency of 20% for this thermochemical process has yet to be achieved and can only be obtained if the n ecessary reactor implementation can be realized by continued research into ferrite materials that allow for controlled reduction steps and species longevity. This study into oxidation of elemental iron and iron alloys concludes but should help with the c ontinued effort into redox system characterization and thermal reduction mechanism identification. Ultimately for this technology to mature, intermediary materials of iron or other highly active transition metals need to show an ability to remain active f or increased cycle times while surviving physically at reactor temperatures beyond 1300C. The efficiency of a solar furnace also must improve beyond the current design value of 3% in order for it to adequately compete with current and emerging technologi es. System design and optimization is the dependent factor when addressing efficiency and must be a focus to obtain reactor efficiencies near 20%. Solar fuel and hydrogen production represents a unique opportunity to utilize a number of energy system pat hways to compensate for increased global energy demand. Storing fuel from solar and wind sources through the combined process of thermochemical and electrochemical de vices while simultaneously producing electricity by PV and wind turbine power plants poss ess the potential to produce energy on the scale of fossil fuels. However, only by continued research into engineering development and implementation can this be realized by showing system reliability and economic benefit.
115 LIST OF REFERENCES  Styring S Artificial photosynthesis of solar f uels Faraday Discuss 2012; 155: 357 76  Cook T, Dogutan D, Reece S, Surendranath Y, Teets T, Nocera D. Solar energy s upply and storage for the legacy and nonlegacy w orlds Chem Rev 2010;110 6474 502  Barber J. Biological solar e nergy Phil Trans R Soc A 2007; 365: 1007 23  Li F, Kim H, Sridhar D, Wang F Zeng L, Chen J, Fan S. Syngas chemical looping g asification process: Oxygen carrier particle selection and p erformance Energy Fuels 2009;23: 4182 9  Allendorf M, Diver R, Siegel N, Miller J. Two step water splitting using mixed metal ferrites: Thermodynamic analysis and characterization of synthesized m aterials Energy Fuels 2008 22: 4115 24  Svoboda K, Slowinski G, Rogut J, Baxter D. Thermod ynamic p ossibilities and constraints for pure hydrogen production by iron based chemical looping process at lower temperatures. Energy Convers Manag 2007;48: 3063 373  Svoboda K, Siewiorek A, Baxter D, Rogut J, Puncochar M. Thermodynamic possibilities and constraints of pure hydrogen production by a chromium, nickel and manganese based chemical looping process at lower t emperatures Chem Pap 2007;61(2): 110 120  Steinfeld A, Sanders S, Palumbo R. Design aspects of solar thermochemical engineering a case study: Two step water splitting cycle using the Fe3O4/FeO redox system. Sol Energy 1999; 65(1): 43 53  Otsuka K, Mito A, Takenaka S, Yamanaka I. Production of hydrogen f rom methane without CO2 emissions mediated by indium oxide and iron oxide. Int J Hydrogen Energy 2001;26: 191 4  Gupta P, Velazquez Vargas LG, Fan LS. Syngas redox (SGR) process to produce hydrogen from coal derived syngas. Energy Fuels 2007;21: 2900 0 8  Stamatiou A, Loutz enhiser PG, Steinfeld A. Solar syngas p roduction via H2O/CO2 splitting thermochemical c ycles with Zn/ZnO and FeO/Fe3O4 redox r eactions. Chem Mater 2010;22: 851 9  Neises M, Roeb M, Schmcker M, Sattler C, Pitz Paal R. Kinet ic investigations of the hydrogen step of a thermochemical cycle using mixed iron oxides coated on ceramic substrates Int J Energy Research 2010;34: 651 61
116  Abanades S, Charvin P, Flamant G, Neveu P. Screening of water splitting t h ermochemical cycles potentially attractive for hydrogen production by concentrated solar e nergy. Energy 2006;31: 2805 22  Steinfeld A. Solar hydrogen production via a two step water splitting thermochemical cycle based on Zn/ZnO redox r eactions Int. J Hydrogen Energy 2 002;27: 611 9  Svoboda K, Slowinski G, Rogut J, Baxter D Pohorely M Thermodynamic p ossibilities and constraints for pure hydrogen production by a nickel and cobalt based chemical looping process at lower temperatures. Energy Conve rs Manag 2008; 49 : 221 31  Miser DE, Shin EJ, Hajaligol MR, Rasouli F. HRTEM characterization of phase changes and the occurrence of maghemite during catalysis by an iron oxide Appl Catal A Gen 2004;258: 7 16 [17 ] Bleeker M, Veringa H, Kersten S. Deactivation of iron oxide used in the steam iron proc ess to produce hydrogen. Appl Catal A Gen 2009;357: 5 17 [18 ] Hiroshi K, Hirofumi A, Hideyuki I, Yoichiro T, Noriko H, Akio S, Yutaka T. Conversion of concentrated solar energy to c hemical e nergy ( 54) two step water splitting cycle with Ni ferrite/zirconia s ystem Nip pon Kagakkai Koen Yokoshu 2004;84(1): 361 [19 ] Ehrensberger K, Frei A, Kuhn P, Oswald H, Hug P. Comparative experimental investigations of the w ater s plitting reaction with i r on o xide Fe1Yo and iron manganese o xides (Fe1 Xmnx)(1 Y)O Solid State Ionics 1995;78: 151 60 [20 ] Kaneko H, Kodama T, Gokon N, Tamaura Y, Lovegrove K, Luzzi A. Decomposition of Zn ferrite for O2 generation by concentrated s olar radiation. Sol Energy 2 004;76: 317 2 2 [21 ] Kodama T, Gokon N. Thermochemical cycles for high temperature solar hydrogen p roduction Chem Rev 2007;107: 4048 77 [22 ] Kodama T, Kondoh Y, Yamamoto R, Andou H, Satou N. Thermochemical hydrogen production by a redox s ystem of ZrO2 supported Co(II) f errite Sol Energy 2005;78: 623 31  Nakamura T. Hydrogen production from water utilizing solar heat at high temperatures. Sol Energy 1976;19: 467 75 [24 ] Charvin P, Abanades S, Flamant G, Lemort F. Two step water splitting thermochemical cycl e based on iron oxide redox pair for solar hydrogen production. Energy 2007;32: 1124 33
117 [25 ] Funk JE. Thermochemical hydrogen production: past and present Int J Hydrogen Energy 2001;26: 185 90 [26 ] Steinfeld A. Solar thermochemical production of hydrog en a review. Sol Energy 2005;78: 603 15 [27 ] Gokon N, Murayama H, Umeda J, Hatamachi T, Kodama T. Monoclinic zirconia supported Fe2O3 for the two step water splitting thermochemical cycle at high thermal reduction temperatures of 1400 Int J Hydrog en Energy 2009;34: 1208 17 [28 ] Hacker V, Frankhauser R, Faleschini G, Fuchs H, Friedrich K, Muhr M, et al. Hydrogen production by steam iron process J Power Sources 2000;86: 531 5 [29 ] Lorente E, Pena JA, Herguido J. Separation and storage of hydrogen by steam iron process: Effect of added metals upon hydrogen release and solid stability J Power Sources 2009;192: 224 9 [30 ] Lorente E, Cai Q, Pena JA, Herguido J, Brandon NP. Conceptual design and modeling of the Steam Iron process and fuel cell integrated system Int J Hydrogen Energy 2009;34: 5554 62 [31 ] Vannice MA. An analysis of the Mars van Krevelen rate expression. Catal Today 2007;123: 18 22 [32 ] Routr ay K, Reddy KRSK, Deo G. Oxidative dehydrogenation of propane on V2O5/Al2O3 and V2O5/TiO2 catalysts: understanding the effect of support by parameter estimation Appl Catal A Gen 2004;265: 103 13 [33 ] Arnold E, Sundaresan S. The role of lattice oxygen in the dynamic b ehavior of oxide c a talysts. Chem. Eng. Comm. 1987;58: 213 30 [34 ] Routray K, Deo G. Kinetic parameter estimation for a multiresponse nonlinear reaction model. AIChE J 2005;51: 1733 46 [35 ] Shee D, Rao M, Deo, G. Kinetic parameter estimation for supported vanadium oxide catalysts for propane ODH reaction: Effect o f loading and support. Catal Today 2006;118: 288 97 [36 ] Mars P, van Krevelen DW. Oxidations carried out by means of vanadium oxide catalysts. Chem E ng Sci 1954;3(1): 41 59 [37 ] Roberts M, Wood P. The mechanism of the oxidation and passivation of iron by water vapour an electron spectroscopic study. J Electron Spectrosc Relat Phenom 1977;11: 431 7
118 [38 ] Johnson WA, Mehl RF. Reaction kinetics in processes of nucleation and growth Trans Inst Min Metall 1939;135: 416 42 [39 ] Avrami M. Kinetics of phase change. I general theory. J Chem Phys 1939;7: 1103 12 [40 ] Avrami M. Kinetics of phase change. II t ransformation time relations for random dist ribution of nuclei. J Chem Phys 1940; 8: 212 24 [41 ] Avrami M. Granulation, Phase change, and microstructure kinetics of phase c hange. III J Chem Phys 1941;9: 177 84 [42 ] Lorente E, Pena JA, Herguido J. Kinetic Study of the redox process for separating and storing hydrogen: Oxidation stag e and ageing of the solid Int J Hydrogen Energy 2008;33: 615 26 [43 ] Roberts MW. Mechanism of the oxidation of iron films at temperatures f rom to 99 109 [44 ] Mott N. The theory of the formation of protective oxide films on metals, II. Trans Faraday Soc 1940;39: 472 83 [45 ] Mott N. Oxidation of metals and the formation of protective films. Nature 1940;145: 966 1000 [46 ] Mott N. The theory of the formation of protective ox ide films on metals, III. Ibid 1947;43: 429 34 [47 ] Cabrera N, Mott NF. Theory of the oxidation of m etals. Rep Prog Phys 1948;12: 163 84  Stehle R, Bobek M, Hooper R, Hahn D. Oxidation reaction k inetics for the steam iron process in support of hyd rogen production. Int J Hydrogen Energy 2011;36: 15125 35 [49 ] Steinfeld JI, F rancisco JS, Hase WL. Chemical kinetics and d ynamics. 2nd ed. New Jersey: Prentice Hall; 1999 [50 ] Roeb M, Gathmann N, Neis es M, Sattler C, Pitz Paal R. Thermodynamic analysis of two step solar water splitting with mixed iron oxides Int J Energy Research 2009;33: 893 902 [51 ] Abanades S, Le Gal A. CO2 splitting by thermo chemical looping based on ZrxCe1 xO2 oxygen carriers for synthet ic fuel generation. Fuel 2012;102: 180 6
119 [52 ] Collins A, Trapnell B. CO2 Chemisorption on evaporated metal films. Trans Faraday Soc 1957;53: 1476 82 [5 3 ] Copperthwaite R, Davies P, Morris M, Roberts M, Ryder R. The reactive chemisorption of carbon dioxide at magnesium and copper surfaces at lo w temperature. Catal Lett 1988;1: 11 20 [54 ] Kim S, Lee S, Hong S. A study on the reaction characteristics of CO 2 decomposition using iron oxides. J Industrial Eng Chem. 2012;18: 860 4 [55 ] Tamaura Y, Tabata M. Complete dioxide to carbon using cation excess magnetite. Nature 1990;346: 255 6 [56 ] Akanuma K, Nishizawa K, Kodama T, Tabata M, Mimori K, Yoshida T, Tsuj i M, Tamaura Y. Carbon dioxide decomposition into carbon with the rhodium bearing magnetite activated by H2 reduction. J Materi als Sci 1993;28: 860 4 [57 ] Ma L, Chen L, Chen S. Study of the CO2 decompositi on over doped Ni ferrites. J Phys Chem Solids 2007 ;68: 1330 5 [58 ] Arifin D, aston V, Liang X, McDaniel A, Weimer A. CoFe2O4 on a porous Al2O3 nanostructure for the solar thermochemical CO2 splitting. Energy Environ Sci 2012;10: 1039 44 [59 ] Giesen A, Herzler J, Roth P. High temperature oxidation of iro n atoms by CO2 Phys Chem Chem Phys 2002;4: 3665 8 [60 ] Menecier S, Valette S, Denoirjean P, Lefort P. Invar oxidation in CO2 k inetics and mechanism of formation of a wustite layer. J Therm Anal Calorim 2012;107: 607 16 [61 ] Galvez M, Loutzenhiser P, Hischier L, Steinfeld A. CO2 splitting via two step solar thermochemical cycles with Zn/ZnO and FeO/Fe3O4 redox reactions: t hermodynamic analysis. Energy Fuels 2008;22: 3544 50 [62 ] Loutzenhiser P, Galvez M, Hischier L, Steinfeld A. CO2 splitting via two step solar thermochemical cycles with Zn/ZnO and FeO/Fe3O4 redox reactions II: kinetic analysis. E nergy Fuels 2009;23: 2832 9 [63 ] Venstrom L, Davidson J. Splitting water and carbon dioxide via the heterogeneous oxidation of zinc vapor: thermodynamic cons iderations. J Sol Energy Eng 2011;133. [64 ] Venstrom L, Petkovich N, Rudisill S, Stein A, Davidson J. The effects of morphology on the oxidation of ceria by water and carbon dioxide. J Sol Energy Eng. 2012; 134.
120 [65 ] Coker E, Ambrosini A, Rodriguez M, Mi ller J. Ferrite YSZ composites for solar thermochemical production of synthetic fuels: in operando characterization of CO2 reduction. J Materials Chem. 2001;21: 10767 76 [66 ] Coker E, Rodriguez M, Ambrosini A, Miller J, Stechel E. Using in situ techniques to probe high temperature reactions: thermochemical cycles for the production of synthetic fuels from CO2 and water. Powder Diffraction 2012;27(2): 117 25 [67 ] Smestad G, Steinfeld A. Review: Photochemical and thermochemical production of solar fuels fro m H2O and CO2 us i ng metal oxide catalysts. Ind Eng Chem Res 2012;51: 11828 40 [68 ] Loutzenhiser P, Meier A, Steinfeld A. Review of the two step H2O/CO2 splitting solar thermochemical cycle based on Zn/ZnO r edox reactions. Materials 2010;3: 4922 38 [69 ] Li H, Zhang J, Young D. Oxidation of Fe Si, Fe Al and Fe Si Al alloys in CO2 H2O gas at 800C. Corrosion Sci 2012;54: 127 38 [70 ] Kim J, Johnson T, Miller J, Steechel E, Maravelias T. Fuel production from CO2 using solar thermal energy: system level anal ysis. Energy Environ Sci 2012;5: 8417 29 [71 ] Loutzenhiser P, Steinfeld A. Solar syngas production from CO2 and H2O in a two step thermochemical cycle via Zn/ZnO redox reactions: thermody namic cycle analysis. Int J Hy drogen Energy 2011;36: 12141 7 [72 ] Stamatiou A, Loutzenhiser P, Steinfeld. Solar Syngas production from H2O and CO2 via two step thermochemical cycles based on Zn/ZnO and FeO/Fe3O4 redox reactions: kinet ic analysis. Energy Fuels 2010,24: 2716 22 [73 ] Loutzenhiser P, Stamatiou A, Steinfeld A. Concentraed solar energy for thermochemical producing liquid fuels from CO2 and H2 O. Energy Conversion 2011;63(1): 32 4 [74 ] Thompson JJ. Rays of positive electricity and their application to chemical analysis. Monographs on Phys 1913 [75 ] Aston F. M ass spectra and isotopes. Nobel Lecture 1922 [76 ] Borman S. A brief history of mass spect rometry instrumentation. Chem Eng News 1998 [77 ] Kiser R. Introduction to mass spectrometry and its application. Prentice Hall 1968 [78 ] Yost R. Mass spectrometry principles & instrumentation, l ectures 2 4. Retrieved from http://lss.at.ufl.edu/ 2010
121 [79 ] Yost R. Ionization techniques, l ectures 5 8. Retrieved from http://lss.at.ufl.edu/ 2010 [80 ] Yost R. Mass analyzers, l ectures 9 12. Retrieved from http://lss.at.ufl.edu/ 2010 [81 ] Hahn DW. Raman scattering t heory. Retrieved from http://plaza.ufl.edu/dwhahn/Raman%20Scattering%20Theory.pdf 2007 [82 ] Go KS, Son SR, Kim SD. Reaction kinetics of reduction and oxidation of metal oxides for hydrogen producti on. Int J Hydrogen Energy 2008;33: 5986 95 [83 ] Weidenkaff A, Nuesch P, Wokaun A, Reller A. Mechanistic studies of the water splitting reaction for producing solar hydrogen. Solid States Ionics 1997; 101 103:915 22 [84 ] Joseph Y, Kuhrs C, Ranke W, Ritter M, Weiss W Absorption of FeO(111) and Fe3 O4(111): identification of active sites for disassociations. Chem Phys Lett 1999;314: 195 202 [85 ] Joseph Y, Ranke W, Weiss W. Water on FeO(111) and Fe3O4(111): Adsorption behavior on different surface t erminations. J Phys Chem B 2000;104: 3224 36 [86 ] Pineau A, Kanari N, Gabalah I. Kinetics of reduction of iron oxides by H2. Part 1. Low temperature reduction of hematite Thermochem Acta 2006;447: 89 100 [87 ] Go KS, Son SR, Kim SD, Kang KS, Park CS Hydrogen production from two step steam methane refor ming in a fluidized bed reactor Int J Hydrogen Energy 2009;34: 1301 09  Evidence for spontaneous CO2 activation on cobalt surfaces. Chem Phys Lett 2008;454: 262 8  Glezakou V, Dang L. Spon taneous a ctivation of CO2 and possible corrosion pathways on the low index iron s urfa ce Fe(100). J Phys Chem C 2009;113: 3691 6  Mehdizadeh A, Klausner J, Barde A, Rahmation N, Mei R. Investigation of hydrogen production reaction kinetics for an iron silica magnetically stabilized porous structur e. Int J Hydrogen Energy 2012 ;37: 13263 71 [91 ] Scheffe J, McDaniel A, Allendorf M, Weimer A. Kinetics and mechanism of solar thermochemical H2 production by oxidation of a cobalt ferrite zirconia compos ite. Energy Environ Sci 2013;6: 963 73  Levene J, Mann M, Margolis R, Milbrandt A. An analysis of hydrogen production from renewable electricity sources. Sol Energy 2007;81:773 80
122  Kroposki B, Levene J, Harrison K, Sen P, Novachek F. Electrolysis: Inf ormation and opportunities for power utilities. NREL/TP 581 40605 September 2006  Technology brief: Analysis of current day commercial electrolyzers. NREL/FS 560 36705 September 2004.  Saur G, Ramsden T. Wind electrolysis: Hydrogen cost optim ization. NREL/TP 5600 50408 May 2011.  Graf D, Monnerie N, Roeb M, Schmitz M, Sattler C. Economic comparison of solar hydrogen generation by means of thermochemical cycles and electrolysis. Int J Hydrogen Energy 2008;33:4511 9.
123 BIOGRAPHICAL SKETCH Richard Craig Stehle was born in Miami, FL to James Austin Stehle Jr. and Debra Baker Chamberlain. His lone sibling is his older brother James Ryan Stehle who is a proud veteran of the Iraq War. He graduated from Eau Gallie High School in Melbourne, FL in 2005 and enrolled at The University of Miami where he received his Bachelor of Science in mechanical e ngineering in 2009. Despite being an avid Miami Hurricane sport enthusiast, he enrolled at T he University of Florida to begin his graduate studies and received his Master of Science (non thesis) in mechanical e ngineering in 2011. Richard married Yijing Yin Stehle on November 2 2011 who he cherishes more than anything else. The work presented in this dissertation is the culmination of the research carried out for his Doctorate of Philosophy in m echanical e ngineering f rom the University of Florida.