Citation
Theoretical and Experimental Investigation of Metal Oxides for Solar Thermochemical Fuel Production

Material Information

Title:
Theoretical and Experimental Investigation of Metal Oxides for Solar Thermochemical Fuel Production
Creator:
Carrillo II, Richard J
Publisher:
University of Florida
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Mechanical Engineering
Mechanical and Aerospace Engineering
Committee Chair:
Scheffe,Jonathan
Committee Co-Chair:
Hahn,David Worthington
Committee Members:
Moghaddam,Saeed
Hagelin Weaver,Helena Elisabeth
Graduation Date:
8/9/2019

Subjects

Subjects / Keywords:
fuel
hydrogen
metal-oxide
solar

Notes

General Note:
Solar-driven production of fuels, like H2, via a metal oxide redox cycle is a promising method of renewable energy conversion that enables relatively facile storage and transportation of intermittent solar energy. This cycle consists of (1) reduction of the metal oxide using concentrated solar energy and (2) re-oxidation of the reduced material by H2O or CO2 to produce H2 or CO, respectively. The thermodynamic and kinetic properties of the metal oxide greatly impact the efficiency with which solar fuels may be produced. In this work, a lab-scale reactor system and a novel experimental procedure were developed with which these properties may be extracted under well-controlled conditions. Within this experimental framework, the oxygen partial pressure was controlled by delivering a precise mixture of H2 and H2O and the redox reactions of interest were initiated via rapid changes in the delivered H2O. Residual gas analysis, comparing temporal changes in the measured H2 due to redox reactions to the steady-state baseline, was employed to measure the change in oxidation state of the metal oxide. Isothermal relaxation experiments with undoped ceria at 1173-1473 K were performed. The intrinsic kinetics of ceria were also examined; this was done by coupling a tanks-in-series mixing model, accounting for the effects of gas-phase dispersion, with a mechanistic kinetic model. The measured equilibrium data and extracted kinetic parameters were found to be in good agreement with the literature, providing validation for the experimental system and procedures. The experimental framework was employed for examination of a new class of materials, Sr- and Al-doped YMnO3, for which thermodynamic data was previously unknown. Lastly, a thermodynamic model was developed to assess the efficiencies of H2 production via isothermal or near-isothermal redox cycling of LaMnO3-based perovskites in comparison to those of ceria or Zr-doped ceria. For isothermal operation at 1473-1773 K with a reduction oxygen partial pressure of 10-6 atm and 95% effective gas-phase heat recovery, the perovskites examined were shown to outperform both ceria and Zr-doped ceria in terms of fuel yields and efficiency; in general, the highest efficiencies were observed for La0.6Sr0.4Mn0.6Al0.4O3-based redox cycles.

Record Information

Source Institution:
UFRGP
Rights Management:
All applicable rights reserved by the source institution and holding location.
Embargo Date:
2/29/2020

Downloads

This item has the following downloads:


Full Text

PAGE 1

THEORETICAL AND EXPERIMENTAL INVESTIGATION OF METAL OXIDES FOR SOLAR THERMOCHEMICAL FUEL PRODUCTION By RICHARD JOSEPH CARRILLO 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 FLO RIDA 2019

PAGE 2

2019 Richard Joseph Carrillo

PAGE 3

To my wife, parents, sister, and family

PAGE 4

4 ACKNOWLEDGMENTS This work was supported by the University of Florida Graduate School Fellowship Award from the Departme nt of Mechanical and Aerospace Engineeri ng I thank Dr. Jonatha n Scheffe for the opportunity to perform interesting and impactful research; I am genuine ly grateful for his mentors hip and his guidance in my work here at the University of Florida and my deve lopment as a researcher. I acknowledge and thank my colleagues in the Renewable Energy Conversion Laboratory for their contributions to my work Namely, Kent Warren for his support with thermogravimetric analysis and his contributions to the k inetic model presented in Chapter 2 Kangjae Lee for performing the scanning electron microscopy (SEM) and energy dispersive X ray spectroscopy (EDS) measurements presented in Appendix B and Caroline Hill for assistance with the th ermogravimetric experiments discussed in Chapter 3 I would also like to recognize Dr. David Hahn, Dr. Saeed Moghaddam, Dr. Wolfgang Sigmund and Dr. Helena Hagelin Weaver for serving on my supervisory committee. I would like to thank Dr. George Kame nov for performing inductively coupled pla sma mass spectrometry (ICP MS) measurements and Kristy Schepker for assistance with powder X ray diffraction (PXRD) scans both of which are shown in Chapter 3 I thank Cory Rogers for his help with the early stages of the development of the high temperatu re tubular reactor system described in Chapter 2 Additionally, I w ould like to express my gratitude for the help provided by Jacob Misura with performing experiments in the tubular reactor, maintenance of the reactor system, and development of the LabVIEW VI used to control the electrical instrumentation of the reactor system. Lastly, I am eternally grateful for the unwavering support provided by my wife, parents, sister, and family. I thank my wife for being a constant source of inspiration and positivity in my

PAGE 5

5 life. I thank my parents for rearing me into the person I am today, instilling in me a strong work ethic, and for their unco nditional love and support

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ........................... 8 LIST OF FIGURES ................................ ................................ ................................ ......................... 9 LIST OF ABBREVIATIONS ................................ ................................ ................................ ........ 14 ABSTRACT ................................ ................................ ................................ ................................ ... 20 CHAPTER 1 INTRODUCTION: ADVANCES AND TRENDS IN REDOX MATERIALS .................... 22 Background ................................ ................................ ................................ ............................. 25 Ceria and Doped Ceria ................................ ................................ ................................ ............ 26 Perovskites ................................ ................................ ................................ .............................. 34 Emerging Materials ................................ ................................ ................................ ................ 41 Thermodynamics of Solar Redox Material Performance ................................ ....................... 41 Additional Criteria for Redox Material Viability ................................ ................................ ... 49 Conclusion ................................ ................................ ................................ .............................. 50 2 E XPERIMENTAL FRAMEWORK FOR EVALUATION OF THE THERMODYNAMIC AND KINETIC PARAMETERS OF METAL OXIDES .................. 57 Experimental ................................ ................................ ................................ ........................... 60 High Temperature T ubular Reactor Setup ................................ ................................ ...... 60 Experimental Procedure and Evaluation of the Sample Averaged Nonstoichiometry ... 61 Results and Discussion ................................ ................................ ................................ ........... 64 Equilibrium and Transient Oxygen Nonstoichiometry Measurements of Undoped Ceria ................................ ................................ ................................ ............................. 64 Mechanistic Modeling of the Kinetics of Ceria ................................ .............................. 67 Conclusion ................................ ................................ ................................ .............................. 71 3 OXYGEN NONSTOICHIOMETRY AND DEFECT EQUILIBRIA OF YTTRIUM MANGANITE PEROVSKITES WITH STRONTIUM A SITE AND ALUMINUM B SITE DO PING ................................ ................................ ................................ ........................ 80 Experimental ................................ ................................ ................................ ........................... 82 Materials Synthesis and Characterization ................................ ................................ ....... 82 Therm ogravimetric Analysis ................................ ................................ ........................... 84 Isothermal Relaxation Experiments in H 2 /H 2 O ................................ ............................... 86 Results and Discussion ................................ ................................ ................................ ........... 88 Materials Characterization ................................ ................................ ............................... 88

PAGE 7

7 Oxygen Nonstoichiometry Measurements ................................ ................................ ...... 90 Defect Modeling ................................ ................................ ................................ .............. 92 Conclusion ................................ ................................ ................................ .............................. 98 4 THEORETICAL INVESTIGATION OF ISOTHERMAL AND NEAR ISOTHERMAL REDOX CYCLING OF NONSTOICHIOMETRIC OXIDES ................................ ............. 110 Thermodynamic Model for Determination of Solar to Fuel Efficiencies ............................ 113 Model Formulation ................................ ................................ ................................ ........ 113 Thermodyna mic Properties of Candidate Materials ................................ ...................... 124 Results and Discussion ................................ ................................ ................................ ......... 127 Isothermal Operation ................................ ................................ ................................ ..... 127 Temperature Swing Operation ................................ ................................ ...................... 131 Conclusion ................................ ................................ ................................ ............................ 133 5 SUMMARY AND CONCLUSIONS ................................ ................................ ................... 150 APPEND IX A SUPPLEMENT TO CHAPTER 2 ................................ ................................ ........................ 153 B SUPPLEMENT TO CHAPTER 3 ................................ ................................ ........................ 160 C SUPPLEMENT TO CHAPTER 4 ................................ ................................ ........................ 166 LIST OF REFERENCES ................................ ................................ ................................ ............. 175 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 189

PAGE 8

8 LIST OF TABLES Table page 2 1 Comparison of fitted kinetic and thermodynamic parameters with prior work. ................ 7 2 3 1 ICP MS analysis of YSMA8264, YSMA8246, and YSMA9164 shown in mol%. ........ 101

PAGE 9

9 LIST OF FIGURES Figure page 1 1 Gibbs free energy change versus temperature for the reduction and oxidation of ZnO, SnO 2 and Fe 3 O 4 redox pairs. ................................ ................................ ............................. 52 1 2 Partial molar enthalpy versus nonstoichiometry for the reduction of CeO 2 and Zr doped CeO 2 (circles), LSM based perovskites (diamonds), and LSMA6464 (crosses). ... 52 1 3 Partial molar entropy versus nonstoichiometry for the reduction of CeO 2 and Zr doped CeO 2 (circles), LSM based perovskites (diamonds), and LSMA6464 (crosses). ... 53 1 4 Gibbs free energy change versus temperature for the reduction of CeO 2 and Zr doped CeO 2 (circles), LSM based perovskites (diamonds), and LSMA6464 (crosses). .............. 53 1 5 Gibbs free energy change ver sus temperature for the oxidation of CeO 2 and Zr doped CeO 2 (circles), LSM based perovskites (diamonds), and LSMA6464 (crosses) with H 2 O. ................................ ................................ ................................ ................................ ... 54 1 6 Gibbs free energy change versus temperature for th e reduction and oxidation of CeO 2 and several fictitious metal oxides with varied partial m olar thermodynamic properties. ................................ ................................ ................................ ........................... 54 1 7 Predicted equilibrium H 2 yield for the WS reactions of CeO 2 and three fictitious metal oxide s versus oxidation temperature. ................................ ................................ ...... 55 1 8 T ) for numerous combinations h O s O ................................ ................................ ................................ .................. 56 2 1 Schematic of the high temperature tubular reactor sys tem. ................................ ............... 72 2 2 Exemplary experimental data obtained using an isotherma l experimental scheme in which the oxidation state of a ceria sample was cycled by shifting the p O 2 via step changes in the input H 2 O flow rate. ................................ ................................ ................... 73 2 3 (left axis, solid line) and p H 2 O/ p H 2 (right axis, dashed line) versus elapsed time during an isothermal relaxation experiment performed at 1373 K with undoped ceria. ... 74 2 4 Transient in undoped ceria measured via residual gas analysis using the isothermal experimental scheme de scribed herein. ................................ ................................ ............. 74 2 5 Iso therms of the equilibrium oxygen content (2 ) of undoped ceria versus the logarithm of p O 2 ................................ ................................ ................................ ............... 75 2 6 Reaction rate for reduction and oxidation reactions of undoped ceria performed in a H 2 /H 2 O environment. ................................ ................................ ................................ ......... 76

PAGE 10

10 2 7 Measured and CSTR corrected reaction rates during reduction of undoped c eria performed at 1473 K and p H 2 O: p H 2 of 2.24:1. ................................ ................................ 77 2 8 Comparison between the model predictions and CSTR corrected reaction rates during ceria reduction (top subplots) and oxidation (bottom subplots) in H 2 /H 2 O. .......... 78 2 9 Model predictions of ceria oxidation in a H 2 /H 2 O mixture and corresponding at 137 3 K and 1473 K. ................................ ................................ ................................ ........... 79 3 1 PXRD patterns of YSMA8264 after synthesis, reduction in an Ar atmosphere at 1173 K, and reduction in a H 2 /Ar atmosphere at 1173 K. ................................ ........................ 101 3 2 PXRD patterns of YSMA8246 after synth esis, reduction in an Ar atmosphere at 1173 K, and reduction in a H 2 /Ar atmosphere at 1173 K. ................................ ........................ 102 3 3 PXRD patterns of YSMA9164 after synthesis, reduction in an Ar atmosphere at 1173 K, and reduc tion in a H 2 /Ar atmosphere at 1173 K. ................................ ........................ 102 3 4 Percent mass change and temperature versus time for temperature programmed reduction experiments in H 2 /Ar ( p H 2 Y SMA8246, YSMA9164, an d ceria. ................................ ................................ ............... 103 3 5 Exemplary TGA results showing percent mass change and temperature versus time for nonisothermal experiments in O 2 /Ar ( p O 2 = 1.6110 4 atm) at 1173 1473 K performed with YSMA8264, YSMA8246, and YSMA 9164. ................................ ......... 104 3 6 Exemplary experimental data for isothermal relaxation experiments in H 2 /H 2 O with YSMA8 246. ................................ ................................ ................................ ..................... 105 3 7 Isotherms of the equilibrium oxygen content (3 ) of YSMA8264, YSMA8246, and YSMA9164 versus the logarithm of the p O 2 ................................ ................................ .. 106 3 8 Defect model fits for YSMA8264 at 973 1173 K in the low p O 2 region and 1173 1473 K in the high p O 2 region. ................................ ................................ ........................ 107 3 9 Defect model fits for YSMA8246 at 973 1173 K in the low p O 2 region and 1173 1473 K in the high p O 2 region. ................................ ................................ ........................ 108 3 10 Defect model fits for YSMA9164 at 9 73 1173 K in the low p O 2 region and 1173 1473 K in the high p O 2 region. ................................ ................................ ........................ 109 4 1 Process model schematic for the produc tion of H 2 via a two step metal oxide redox cycle. ................................ ................................ ................................ ................................ 135 4 2 E quilibrium p O 2 versus temperature for H 2 O and CO 2 res pectively. ............................. 136 4 3 Pumping efficiency (electric to pump) of a mechanical pump versus the logarithm of the inlet pressure to the pump ( i.e. the reduction p O 2 ). ................................ ................... 137

PAGE 11

11 4 4 Heat input per mole of O 2 pumped vers us the logarithm of the reduction p O 2 for a thermochemical oxygen pump driven by a Co 3 O 4 / CoO redox cycle. ............................. 138 4 5 H 2 = red ox ) per mole of metal oxide (MO) versus operating temperature for isothermal redox cycles with ea ch of the materials considered. ................................ ................................ ................................ ....................... 139 4 6 Equilibrium oxygen content (2 or 3 ) at 1773 K from 10 6 to 10 1 atm for e ach of the m aterials considered. ................................ ................................ ................................ .. 140 4 7 Inert gas requirement during reduction per mole of metal oxide (MO) versus operating temperature for isothermal redox cycles with each of the ma terials considered. ................................ ................................ ................................ ....................... 141 4 8 Oxidant delivery requirement per mole of metal oxide (MO) versus operating temperature for isothermal redox cycles with each of the materials consider ed. ............ 142 4 9 Energy p enalties due to heating of the oxidant ( Q sens,ox ), the reduction endotherm ( Q red ), heating of the inert gas ( Q sweep,ig ), heat equivalent of pumping work for a mechanical vacuum pump ( Q sweep,vp ), heat equivalent of pumping work for an electrochemical pump ( Q sweep,ep ), and the heat input required to drive a Co 3 O 4 /CoO thermochemical oxygen pump ( Q sweep,tp ) versus operating temperature for i sothermal redox cycles with ceria, CZO20, and LSM40. ................................ ................................ 143 4 10 Solar to fuel energy conversion efficiency for H 2 O splitting versus operating temperature for isothermal redox cycles with each of the mate rials considered. ............ 144 4 11 H 2 = red ox ) per mole of metal oxide (MO) versus temperature swing for nonisother mal redox cycles with ea ch of the materials considered. ................................ ................................ ................................ ....................... 145 4 12 Inert gas requirement during reduction per mole of metal oxide (MO) versus temperature swing for nonisothermal redox cycles with each of the ma terials considered. ................................ ................................ ................................ ....................... 146 4 13 Oxidant delivery requirement per mole of metal oxide (MO) versus temperature swing for nonisothermal redox cycles wi th each of the mate rials considered. ................ 147 4 14 Energy penalties due to heating of the oxidant ( Q sens,ox ), heating of the metal oxide from the oxidation temperature to the reduction temperature ( Q sens,s ), the reduction endotherm ( Q red ), h eating of the inert gas ( Q sweep,ig ), heat equivalent of pumping work for a mechanical vacuum pump ( Q sweep,vp ), heat equivalent of pumping work for an electrochemical pump ( Q sweep,ep ), and the heat input required to drive a Co 3 O 4 /CoO thermochemical oxygen p ump ( Q sweep,tp ) versus temperature swing for nonisothermal redox cycles wi th ceria, CZO20, and LSM40. ................................ ......... 148 4 15 Solar to fuel energy conversion efficiency for H 2 O splitting versus temperature swing for nonisothermal redox cycles with each of the m aterials considered. ................ 149

PAGE 12

12 A 1 Transient in undoped ceria measured via residual gas analysis using an isothermal relaxation scheme in H 2 /H 2 O. ................................ ................................ .......................... 153 A 2 Transient in undoped ceria measured via residual gas analysis using an isothermal relaxation scheme in H 2 /H 2 O. ................................ ................................ .......................... 154 A 3 Transient in undoped ceria measured via residual gas analysis using an i sothermal relaxation scheme in H 2 /H 2 O. ................................ ................................ .......................... 155 A 4 Reaction rate for reduction and oxidation reactions of undoped ceria performed in a H 2 /H 2 O environment. ................................ ................................ ................................ ....... 156 A 5 Reaction rate for reduction and oxidation reactions of undoped ceria performed in a H 2 /H 2 O environ ment. ................................ ................................ ................................ ....... 157 A 6 Reaction rate for reduction and oxidation reactions of undoped ceria performed in a H 2 /H 2 O environ ment. ................................ ................................ ................................ ....... 158 A 7 Exemplary temporal distributions of species concentrations during reduction (top subplots) and oxi dation (bottom subplots) at 1173 K and 1473 K. ................................ 159 B 1 PXRD patterns of YSMA8264, YSMA8246, YSMA9164, and YSMCo6464 (measured by Huang and Huang). ................................ ................................ ................... 160 B 2 PXRD patterns of YSM20 as synthesized and as measured by Huang and Huang. ....... 161 B 3 SEM image of YSMA8264 powder. ................................ ................................ ................ 162 B 4 SEM image of YSMA8246 powder. ................................ ................................ ................ 163 B 5 SEM image of YSMA9164 powder. ................................ ................................ ................ 164 B 6 EDS images showing the distributi ons of Y, Sr, Mn, and Al in YSMA8264, YSMA8246, and YSMA9164. ................................ ................................ ......................... 165 C 1 Partial molar enthalpy change per mole of monoatomi c oxygen versus nonstoichiometry for each of the materials considered. ................................ .................. 166 C 2 Equilibrium oxygen content of LSMA6464 versus p O 2 at 1473 1773 K. ....................... 167 C 3 Logarithm of the equilibrium constants of oxygen vacancy formation (K 1 ) and disproportionation (K 2 ) versus inver se temperature for LSM A6464. ............................. 168 C 4 Equilibrium oxygen content of LCM40 versus p O 2 at 1473 1 773 K. ............................. 169 C 5 Logarithm of the equilibrium constants of oxygen vac ancy formation (K 1 ) and disproportionation (K 2 ) versus inverse temperature for LCM40. ................................ .... 170 C 6 Equilibrium oxygen content of LCMA6464 versus p O 2 at 1473 1 773 K. ...................... 171

PAGE 13

13 C 7 Logarithm of the equilibrium constants of oxygen vacancy formation (K 1 ) and disproportionation (K 2 ) versus inverse tempe rature for LCMA6464. ............................. 172 C 8 Solar to fuel energy conversion efficiency for H 2 O splitting versus operating temperature for isothermal redox cy cles using ceria. ................................ ...................... 173 C 9 Solar to fuel energy conversion efficiency for H 2 O splitting ver sus temperature swing for nonisothermal redox cy cles using ceria. ................................ .......................... 174

PAGE 14

14 LIST OF ABBREVIATIONS BSCF B arium strontium coba lt iron BSCF25/75/8/2 Ba 0.25 Sr 0.75 Co 0.8 Fe 0.2 O 3 BSCF5582 Ba 0.5 Sr 0.5 Co 0.8 Fe 0.2 O 3 CA Citric acid CDS Carbon dioxide splitting CEM Controlled evaporator mixer CPC Compound parabolic concentrator CSTR Continuously stirred tank reactor CZO20 Ce 0.8 Zr 0.2 O 2 DI Deionized EDS Energy dispersive X ray spectroscopy EG E thylene glycol ESO Entropy stabilized oxide FT Fischer Tropsch HEX Heat exchanger ICP MS I nductively coupled plasma mass spectrometry ITCS I sothermal carbon dioxide splitting ITRC I sothe rmal redox cycling ITWS I sothermal water splitting LBM L anthanum barium manganese LCM L anthanum calcium manganese LCM40 L a 0.60 Ca 0.40 MnO 3 LCMA L anthanum calcium manganese aluminum LCMA6464 L a 0.6 Ca 0.4 Mn 0.6 Al 0.4 O 3

PAGE 15

15 LSCF L a 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 LSM L anth anum strontium manganese LSM10 L a 0.90 Sr 0.10 MnO 3 LSM20 L a 0.80 Sr 0.20 MnO 3 LSM30 La 0.70 Sr 0.30 MnO 3 LSM 35 La 0.65 Sr 0.35 MnO 3 LSM40 La 0.60 Sr 0.40 MnO 3 LSM50 La 0.50 Sr 0.50 MnO 3 LSMA L anthanum strontium manganese aluminum LSMA2882 La 0.2 Sr 0.8 Mn 0.8 Al 0.2 O 3 LSMA4664 La 0.4 Sr 0.6 Mn 0.6 Al 0.4 O 3 LSMA4682 La 0.4 Sr 0.6 Mn 0.8 Al 0.2 O 3 LSMA6446 La 0.6 Sr 0.4 Mn 0.4 Al 0.6 O 3 LSMA6464 La 0.6 Sr 0.4 Mn 0.6 Al 0.4 O 3 LSMA6482 La 0.6 Sr 0.4 Mn 0.8 Al 0.2 O 3 MO Metal oxide PCCD P aired charge compensating doped PCHIP P iecewise cubic Hermite interpolating polynomial PCO P oly cation oxide PV P hotovoltaics PXRD Power X ray diffraction RPC R eticulated porous ceramic RPM R evolutions per minute RTD R esidence time distribution SEM S canning electron microscopy

PAGE 16

16 SSA S pecific surface area STC S olar thermoche mical TGA T hermogravimetric analysis TPR T emperature programmed reduction TSRC T emperature swing redox cycling WS W ater splitting YCM50 Y 0.5 Ca 0.5 MnO 3 YSM50 Y 0.5 Sr 0.5 MnO 3 YSMA Y ttrium strontium manganese aluminum YSMA8246 Y 0.8 Sr 0.2 Mn 0.4 Al 0.6 O 3 YSMA 8264 Y 0.8 Sr 0.2 Mn 0.6 Al 0.4 O 3 YSMA9164 Y 0.9 Sr 0.1 Mn 0.6 Al 0.4 O 3 YSMCo Y ttrium strontium manganese cobalt YSZ Y ttria stabilized zirconia XPS X ray photoelectron spectroscopy NOMENCLATURE A P re exponential factor C G eometric concentration ratio c p S pecific heat capacity E a Activation energy E cell Electric potential E v V acancy formation energy F F araday constant F H 2 H 2 molar flow rate

PAGE 17

17 F H 2 O H 2 O molar flow rate G sc S olar constant h f F ormation enthalpy K E quilibrium constant K w E quilibrium constant f or the dissociation of H 2 O M Molar mass m Mass N T otal number of moles or number of continuously stirred tank reactors in series P Pressure p H 2 H 2 partial pressure p H 2 O H 2 O partial pressure p O 2 O 2 partial pressure Q H eat or heat equivalent energy R Ideal gas constant T Temperature t Time y Mole fraction GREEK Change G Gibbs free energy change for a reaction g H 2 O Gibbs free energy change for the formation of H 2 O g O P artial molar Gibbs free energy change per mole of monoatomic oxygen H E nthalpy change for a reaction h Sensible enthalpy change

PAGE 18

18 h O P a rtial molar enthalpy change per mole of monoatomic oxygen h T D efect formation enthalpy S E ntropy change for a reaction s O P artial molar entropy change per mole of monoatomic oxygen s T D efect formation entropy O xygen nonstoichiometry H eat re covery effectiveness Efficiency R eaction coordinate Stefan Boltzmann constant 2 Variance Tolerance factor or space time SUBSCRIPTS 0 Ambient aux Auxiliary df Defect formation ep Electrochemical Pump exp Experiment f Final gg Gas to g as i Initial ig Inert gas iso Isothermal ox Oxidation R Reaction

PAGE 19

19 red Reduction rerad Re radiation s Sample or solid sens Sensible ss Solid to solid sys System tot Total tp Thermochemical pump vp Vacuum pump

PAGE 20

20 Abstract of Dissertation Present ed to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THEORETICAL AND EXPERIMENTAL INVESTIGATION OF METAL OXIDES FOR SOLAR THERMOCHEMICAL FUEL PRODUCTION By Richard Joseph Carrillo August 2019 Chair: Jonathan R. Scheffe Major: Mechanical Engineering Sola r driven production of fuels, like H 2 via a metal oxide redox cycle is a promising metho d of renewable energy conversion that enables relatively facile s torage and transportation of intermittent solar energ y. This cycle consists of (1) reductio n of the metal oxide using concentrated solar energy and (2) re oxidation of t he reduced material by H 2 O or CO 2 to produce H 2 or CO, respectively. The thermodynamic and kinetic properties of the metal oxide greatly impact the efficiency with which solar fuels may b e produced. In this work, a lab scale reactor system and a novel experimental procedure were developed with which these properties may b e extracted under we ll controlled conditions Within this experimental framework, t he oxygen partial pressure was controlled by delivering a precise mixture of H 2 and H 2 O and the redox reactions of interest were initiated via rapid changes in the delivered H 2 O. Residual gas a nalysis, comparing temporal changes in the measured H 2 due to redox reactions to the steady state baseline, was employed to measure the change in oxidation state of the metal oxide I sothermal relaxation experiments with undoped ceria at 1173 1473 K and ox ygen partial pressures from 4.5410 18 1.0210 9 atm were performed. The intrinsic kinetics of ceria were also examined; this was done by coupling a tanks in series mixing model, accounting for the effects of gas phase dispersion, with a mechanistic kineti c model. The measured equilibrium data and extracted

PAGE 21

21 kinetic parameters were found to be in good agreement with the literature, providing validation for the experimental system and procedures. The experimental framework was employed for examination of a ne w class of materials, Sr and Al doped YMnO 3 for which thermodynamic data was previously unknown. Lastly, a thermodynamic model was developed to assess the efficiencies of H 2 production via isothermal or near isothermal redox cycling of LaMnO 3 based perov skites in com parison to those of ceria or Zr doped ceria. For isothermal operation at 1473 1773 K with a reduction oxygen partial pressure of 10 6 atm and 95% effective gas phase heat recovery, the perovskites examined were shown to outperform both ceria a nd Zr doped ceria in terms of fuel yields and efficiency; in general, the highest efficiencies were observed for La 0.6 Sr 0.4 Mn 0.6 Al 0.4 O 3 based redox cycle s.

PAGE 22

22 CHAPTER 1 INTRODUCTION : ADVANCES AND TRENDS IN REDOX MATERIALS Solar energy, a ubiquitous and esse ntially unlimited resource, offers the ability to generate electricity via solar photovoltaics (PV) [1] supply the process heat needed to drive a heat engine ( e.g. Rankine cycle) [2] provide hot water for domestic heating applications [3] or any number of other thermally demanding applications. The major caveats with all of these storage technologies, e.g. batterie s [4] or thermal storage [5] if they are to b e used on demand. The conversion of incident sunlight directly to chemical fuels such as molecular hydrogen or fuel precursors like synthesis gas (syngas) offers the potential to efficiently store solar energy, transport it, and utilize it on demand [6] Given that the current transportation infrastructure is largely based on the use of liquid hydrocarbon fuels efficient conversion of concentrated solar energy into fungible fuels provides a pathway to begin the transition away from fossil fuels and towards renewable fuel sources. There are a variety of methods to convert sunlight to fuels through the dissociati on of H 2 O and/or CO 2 ; most of these are low temperature photon driven drive the process of interes t, the remainder of the spectrum is not capable of being u tilized [7, 8] However, thermochemical approaches that employ conce ntrated sunlight as a thermal input to drive the dissociation of H 2 O and CO 2 utilize the entire solar spectrum and, therefore, provide a thermodynamically attractive pathway to solar fuel production [9] A variety of mature three dimensional concentrating technologies such as parabolic dish es or heliostat fields [3, 9] may be e.g. 1000 to 10000 suns, where 1 sun is equivalent to 1 kW m 2 ) to drive a number of solar thermochemical (STC) processes at elevated temperatures, usually in the range of 873 to 2273 K [10 14]

PAGE 23

23 One promising method of STC fuel production is the metal oxide based thermochemical redox cycle. The cycle is characterized by two distinct steps, namely a high temperature reduction step and a lower temper ature oxidation step, during which the metal oxide serves as a reactive intermediate that facilitates oxygen exchange. The first step is defined by an endothermic reaction, shown in the equation below, during which the metal oxide (M x O y ) is reduced and gas eous O 2 is released upon heating with concentrated solar energy. ( 1 1 ) The deviation from stoichiometry or nonstoichiometry ( ) defines the maximum amount of oxygen exchange that may occur in the subsequent step (thereby setting an upper limit on the fuel production) and, thus, is a critical parameter in determining the effic acy of a particular material for use in solar fuel production [15] The second step typically occurs at a temperature lower than that of the prior step and is driven by the introduction of an oxidant such as H 2 O or CO 2 The respective H 2 O splitting (WS) and CO 2 splitting (CDS) reactions are shown in the equations below. ( 1 2 ) ( 1 3 ) Through a series of surface mediated reaction mechanisms, the reduced metal oxide is reacted with the delivered H 2 O or CO 2 to produce H 2 or CO, respectively, and all or a portion (as allowed by the thermodynamic favorability of the oxidation reaction under the co nditions employed) of the oxygen released by the metal oxide during the first step is replenished. The obtained during the reduction reaction and the change in induced by the oxidation reaction are dependent upon the respective temperatures and oxygen partial pressures ( p O 2 ) at which the two separate steps are performed [16] In general, increases with temperature and is inversely

PAGE 24

24 proportional to the p O 2 The net reactions for the WS and CDS cycles are simply H 2 2 + 0.5O 2 and CO 2 2 respectively, as the metal oxide is recycled. In contrast to direct H 2 O or CO 2 thermolysis, the two step cycle requires significantly lower temperatures to drive the reactions and produces gaseous O 2 and the desired fuel in separate steps, thus, bypassing the need for efficient high temperature gas separation (an energetically expensive process) [17, 18] The H 2 or CO produced may be used directly to generate power through combustion or electrochemical oxidation. Additionally, the mixture of the two, syngas, may be further processed to fungible li quid hydrocarbon fuels such as kerosene, diesel, and gasoline via catalytic processes such as Fischer Tropsch (FT) synthesis [19] Siegel et al. [20] suggested that, when evaluated in comparison to H 2 production via photovoltaics coupled with electrolysis, STC technologies need to achieve an annual average solar to fuel efficiency of at least 20% to be economically competitive. To date, the largest reported average solar to fuel energy conversion efficiency for STC CDS is 5.25% by Marxer et al. [21] using ceria (CeO 2 ) as the reactive intermediate. Currently, CeO 2 represents the pinnacle of performance because of its fast reaction rates, phys ico chemical stability, and the relative ease with which it is re oxidized [16, 20, 22, 23] The reaction kinetics and oxygen diffusion rates are rapid, enabling the use of a wide range of reactive structures ( e.g. reticulated porous ceramics) [14, 22, 24, 25] With regards to morphological stability, CeO 2 exhibits resistance to particle agglomeration and maintains its cubic fluorite structure over a wide variety of reductio n extents and operating conditions [16, 26, 27] However, high temperatures are required in order to achieve notable reduction extents and high efficiencies with CeO 2 because of its relatively high enthalpy change during oxygen exchange [23, 28] At such extreme temperatures, sintering and

PAGE 25

25 sublimation have been shown to negatively impact operability [19, 29] and energy losses due to thermal re radiation can be substantial. Background There exist three distinct classes of metal oxide redox pairs: (1) volatile stoichiometric, (2) nonvolatile stoichiometric, and (3) nonst oichiometric. In a volatile redox cycle, the temperature required for reduction is greater than the vaporization temperature of the metal oxide, thereby causing it to undergo a solid to gas phase transition (e.g. M x O y 0.5yO 2 ) [11] Some volatile redox pairs include ZnO/Zn [11] and SnO 2 /SnO [30] The practicality of large scale implementation of volatile redox cycles is largely limited by the need for rapid quenching of the gas phase products [31] In nonvolatile stoichiometric and nonstoichiometric cycles, the redox intermediat e remains in the solid phase throughout the reduction step [15] Nonvolatile stoichiometric reactions involve a change in crystal structure and the reduction of the cation to form solid compounds (e. g. M 3 O 4 2 ) [15, 17, 23] Examples of nonvolatile stoichiometric redox pairs include Fe 3 O 4 /FeO [17, 32, 33] ferrite systems in which transition metal cations (e .g. Co 3+/2+ and Ni 2+ ) are substituted for Fe 3+/2+ in Fe 3 O 4 [34 37] and hercynite systems wherein reduction of Fe 3 O 4 or ferrites is performed in the presence of Al 2 O 3 [13, 38] Stoichiometric reactions have a greater oxygen exchange capacity compared to that of nonst oichiometric reactions; however, as shown in iron oxide cycles, they typically exhibit poor stability ( i.e. the reduction temperatures required for thermodynamic favorability exceed the melting point of the reduction product, FeO) and slower reaction kinet ics [15, 23] ZrO 2 and yttria stabilized zirconia (YSZ) have been used as supports or thermally stable solvents to address problems regarding physical stability, but this implies an extra sensible heating energy penalty and the efficiency suffers as a result [34, 39 41] Fro m a thermodynamic perspective, large changes in entropy during oxygen exchange are desirable, as the entropy has a

PAGE 26

26 strong impact on the thermodynamic favorability and the temperature swing between reduction and oxidation. The solid to gas and crystallograp hic phase changes that occur during cyclic operation with the volatile and nonvolatile stoichiometric redox pairs, respectively, are associated with larger changes in entropy than those of nonstoichiometric oxygen exchange. Nonstoichiometric reactions with metal oxides such as CeO 2 and LaMnO 3 perovskites involve the partial reduction of the bulk metal oxide, while maintaining the crystallographic structure (e.g. MO 2 2 + 0.5 O 2 ) [15, 16] Similar to ferrite systems, doping schemes have been employed with CeO 2 and different perovskites to tune the thermodynamic parameters and redox performance. Ceria and Doped Ceria Reducti on of CeO 2 is typically performed at 1773 K and p O 2 between 10 6 and 10 3 atm, while the oxidation step is performed between 873 and 1273 K at p O 2 between 10 20 and 10 10 atm [29, 42] CeO 2 based t hermochemical cycles proceed according to the following reduction and oxidation reactions (here only WS is shown): ( 1 4 ) ( 1 5 ) where ox is the nonstoichiometry following oxidation and red is the nonstoichiometry after reduction. CeO 2 has notably high entropy change associated with oxygen exchange when compared to other nonstoichiometric redox materials [16, 43 45] leading to reduced temperature swings T ) between the reduction and oxidation steps [20] and more favorable oxidation thermodynamics than emerging metal oxides that have been recently investigated [46, 47]

PAGE 27

27 Chueh and Haile [12] first demonstrated the ability to perform simultaneous WS and CDS to generate syngas via the nonstoichiometric CeO 2 based redox cycle and showed stable H 2 and CO production over more than 50 cycles. They estimated efficiencies of 15% and 22.9% for cycling between 1073 and 1773 K with 0% and 50% heat recovery, respectively. The cyclability of WS and CDS using CeO 2 was further demonstrated within a directly irradiated 1.9 kW solar reactor that documented the first mea sured peak efficiencies of 0.7% and 0.8% for WS and CDS, respectively, and average efficiencies of 0.4% for both [42] At the time, this set the efficiency record for the conversion of CO 2 to CO using a solar driven device. 500 WS cycles were also performed in an infrared imaging furna ce, wherein CeO 2 was reduced at 1773 K with low p O 2 of 10 5 atm and oxidized at 1073 K. After the first 100 cycles, over which the reaction rates decreased due to sintering, the reduction and oxidation rates were relatively stable over the course of 400 ad ditional cycles. The H 2 and O 2 yields decreased correspondingly, dropping from approximately 8 and 4 mL g 1 respectively, to about 6 and 3 mL g 1 during the first 100 cycles [23] Furler et al. [29] demonstrated stable syngas production with tunable H 2 :CO molar ratios during ten consecutive simultaneous WS and CDS cycles using a porous CeO 2 felt in a 3.6 kW solar cavity receiver. The average solar to fuel ene rgy conversion efficiency was found to be 0.15%, lower than the 0.4% average efficiency shown by Chueh et al. [42] This lower efficiency was largely attributed to heat transfer limitations, as the CeO 2 lead to inhomogeneous temperature distributions and limited the majority of the reaction to the near surface layer exposed to the high flux radiant energy. A gradual decrease in O 2 evolution was observed during the reduction step; this was attributed to a decline in peak reactor temperature due to depositi on of sublimated CeO 2

PAGE 28

28 concentrator (CPC) and quartz window that limited the incident radiation transmission. This deposition was a direct repercussion of the extreme surface temperatures resulting from the poor heat tran sfer properties of the felt. To address this limitation, Furler et al. [25] subseq uently synthesized a CeO 2 based reticulated porous ceramic (RPC) foam with a macroporous structure (mm sized pores) to enable volumetric absorption of incident radiation. The enhanced heat transfer capabilities and greater bulk density of the RPC allowed f or a significant increase in the absolute rate of O 2 evolution and lower temperature gradients throughout the structure. As a result, the average efficiency increased by over a factor of 10; for a power input of 3.8 kW, the solar to fuel energy conversion efficiency was 1.73%. The oxidation kinetics of the RPC, however, were limited by its relatively low specific surface area (SSA), and the oxidation reaction took greater than four times longer than that of the felt to reach completion. To address this, Fur ler et al. [14] developed a series of CeO 2 based RPC structures with dual scale porosities. The dual scale RPC combined the improved heat transfer enabled by the millimeter sized pores of the prior RPC and improved oxidation kinetics resulting from the m icrometer sized pores within the struts. Thermogravimetric (TG) analysis showed no demonstrable difference in the rate or extent of reduction between RPC structures with porous and non porous struts, but a tenfold improvement in oxidation rates for samples with the highest strut porosity. Operation within a solar cavity receiver confirmed the improved oxidation behaviors observed during TG analysis; compared to the macroporous RPC [25] the dual scale porous RPC exhibited a threefold improvement in oxidation rates and a comparable efficiency of 1.72% (even with notably less CeO 2 mass). M arxer et al. [19] demonstrated simultaneous extended WS and CDS redox cycles using the CeO 2 based RPC with dual scale

PAGE 29

29 porosities in a cavity type solar reactor. Over the course of 291 redox cycles, a bout 700 standard liters of syngas were produced, compressed, stored, and transported to Shell Global Solutions in Amsterdam and then catalytically processed via Fischer Tropsch synthesis to produce the derived kerosene. Very recently, a solar to fuel energy conversion efficiency of 5.25% was achieved using the CeO 2 based dual scale porous RPC in a second generation cavity reactor that improved the heat and mass transfer characteristics of the system [21] The reaction kinetics and oxygen diffusion rates of CeO 2 are attractive relative to other solar redox materials, such as iron oxides, ferrites and zinc oxide [22, 23, 48 50] Knowledge of the ambipolar diffusion coefficient of CeO 2 suggests that reaction rates for STC fuel production are not likely to be limited by the time required for oxygen bulk diff usion [22, 23] Measurements of the oxygen surface exchange coefficient show that CeO 2 reduction operates within a surface controlled regime, even for dense samples 1 mm in thickness [51] The results of Ackermann et al. [22] suggest that small diffusion length scales in the micrometer and millimeter range allow for diffusion times on the order of milliseconds to seconds, re spectively. These results agree with experimentally observed reaction rates shown by Furler et al. [25] using dense and porous RPCs. Furthermore, both Chueh et al. [42] and Furler et al. [29] observed that heat tran sfer was always the reaction rate limiting factor during thermal reduction when using porous structures because of their rapid reaction kinetics. Additionally, Scheffe et al. [52] and Welte et al. [53] demon strated the ability to thermally reduce CeO 2 particles with median particle sizes in the micrometer range (12 70 m) in an aerosol reactor in which the residence times were less than 1 second. Heat transfer, not reaction kinetics, was found to be the limit ing factor in both cases.

PAGE 30

30 Oxidation kinetics of CeO 2 have been shown to be strongly dependent on the achieved after reduction and the manner in which this initial reduction extent is achieved ( i.e. thermally or chemically) [24] The CO production rate was shown to decrease with increasing initial and the initial reduction extent beyond which the rates begin to decrease rapidly was larger for higher oxidation temperatures. This abrupt transition in oxidation kinetics was attributed to a near order phase change resulting in a sudden change in ther modynamic and physical properties [54, 55] For thermally reduced CeO 2 (at 1773 K for varying reduction times), the oxidation rates were notably sl ower (about an order of magnitude) than those wherein the initial reduction extent was achieved via chemical reduction with H 2 This was attributed to the absence of micro cracks that were observed following chemical reduction. However, within this regime, the opposite trend was observed; the oxidation rates were shown to increase in a nearly linear manner with increasing initial This trend is believed to be due to even smaller surface cracks that were seen to increase with increasing reduction extent be cause of either chemical or thermal expansion [56] Due to the high reduction temperatures required to achieve adequate reduction of CeO 2 severa l doping schemes have been employed. By substituting transition metal and rare earth metal oxides into the crystal structure, the thermodynamic properties and redox performance of CeO 2 may be tuned. Recently considered dopants include: Li 2+ Sr 2+ Ca 2+ Mg 2+ Sc 3+ Dy 3+ La 3+ Sm 3+ Gd 3+ Y 3+ Pr 3+ Hf 4+ and Zr 4+ [28, 57 68] Substitution of a portion of the Ce 4 + with Zr 4+ lead to a notable improvement in reduction extent; however, Zr 4+ doped CeO 2 exhibited a decre ase in oxidation kinetics compared to pure CeO 2 [28, 61, 64] Gravimetric O 2 evolution increased with increasing Zr 4+ content up to dopant concentrations of about 25 mol% [62] Zr 4+ doped CeO 2 tested in a solar driven

PAGE 31

31 the rmogravimetric analyzer, had poor re oxidation yields due to its low thermodynamic favorability and oxidation kinetics [69] Unlike pure CeO 2 the chemical equilibrium of Zr 4+ doped CeO 2 may be accurately described by a single defect model, namely the development of doubly ionized oxygen vacancies (point defects) upon O 2 evolution [28, 70] Extracted partial molar thermodynamic properties for Zr 4+ doped CeO 2 indicated that both the partial molar enthalpies and entropies decrease in magnitude with increasing Zr 4+ content. The implications of Zr 4+ doping are improved reduction yields at the cost of diminished oxidation kinetics and T between reduction and oxidation. Incorporation of La 3+ Gd 3+ and Y 3+ doping in Zr 4+ doped CeO 2 lead to greatly improve d reduction extents and fuel yields in comparison to undoped CeO 2 Furthermore, La 3+ doping in the Zr 4+ doped CeO 2 improved the thermal stability and resistance to material deactivation during cyclic operation [60] Partial substitution of the trivalent lanthanides La 3+ Sm 3+ and Gd 3+ each lead to a reduction in redox performance compared to pure CeO 2 and O 2 and H 2 yields d ecreased as the dopant concentration increased [60] A nonmonotonic relationship between redox performance and Pr 3+ dopant incorporation was observed [65] With regards to both O 2 release during reduction and H 2 production during WS, Pr 3 + doping was shown to slightly improve yields at 5 and 10 mol% Pr 3+ ; however, the yields showed a sharp decline for dopant concentrations beyond 10 mol%. Extraction of the partial molar thermodynamic properties indicated that the improved reducibility of 5 and 10 mol% Pr 3+ doped CeO 2 was attributed to a reduced enthalpic energy penalty. Reduction extents were shown to increase for doping schemes which yield an effective cation radius smaller than that of pure CeO 2 ; thus, Mg 2+ Sc 3+ and Hf 4+ doped CeO 2 show ed greater reducibility than undoped CeO 2 [57] Of these three doping schemes, Hf 4+ doping

PAGE 32

32 exhibited the largest improvement in O 2 evolution; however, the ratio of H 2 produced during WS to O 2 released was lower than that of undoped CeO 2 Scheff e and Steinfeld [58] performed a thermodynamic analysis of Gd 3+ Y 3+ Sm 3+ Ca 2+ and Sr 2+ doped CeO 2 and found a negative correlation between dopant concentration and at high temperatures. At re duction temperatures greater than 1700 K and p O 2 of 10 5 atm, the reduction extents of pure, undoped CeO 2 were larger than those of all of the doped CeO 2 considered; however, at lower temperatures the opposite behavior was observed, wherein larger dopant c oncentrations led to an increase in the predicted O 2 release. For larger dopant concentrations, the change in Gibbs free energy for oxidation with H 2 O or CO 2 was less sensitive to changes in the reaction temperature (indicating a decrease in the magnitude of the entropy change). Because of this change in the thermodynamic properties, the oxidation of pure CeO 2 at temperatures below 1200 K was found to be more favorable than that of doped CeO 2 and results in greater equilibrium fuel yields. Commercial, large scale implementation of STC fuel production is hindered by low energy conversion efficiencies, a vital metric for quantitative assessment of the viability of a particular cycle. The efficiency of CeO 2 and doped CeO 2 redox cycling has been evaluated experi mentally [14, 19, 25, 29, 42, 71] and theoretically [20, 58, 72 78] Under idealized conditions, the incorporation of gas and solid phase heat recovery was shown to offer a substantial improvement in efficiency, which increased with increasing concentration ratio, when compared to operation with no heat recovery [72] The effects of p O 2 reduction schemes, both mechanical and chemical, on the ove rall efficiency have been analyzed [76, 78, 79] Vacuum pumping was shown to lessen the need for highly effective heat recovery and potentially yield higher efficiencies than p O 2 reduction via sweep gas flow. Moreo ver, the combination of

PAGE 33

33 chemical scavengers with gas sweeping or pumping may provide an avenue for enhanced efficiencies. Scheffe and Steinfeld [58] found an inverse relationship between dopant con centration (Gd 3+ Y 3+ Sm 3+ Ca 2+ and Sr 2+ ) and solar to fuel efficiency, with CeO 2 outperforming the doped variations over a wide range of oxidation temperatures. Bulfin et al. [77] determined that Zr 4+ doping schemes may allow for improved efficiencies due to their greater reducibility, but require large r temperature swings for favorable oxidation. T between the reduction and oxidation steps, particularly in the absence of solid phase heat recovery, negatively impact s the energy conversion efficiency and cycle operability. This is because solar energy must be used to reheat the solid back to t he reduction temperature, leading to an increase in energetic penalties, thermal stress on the reactor components, and cycle times. Isothermal operation, in which the reduction and oxidation steps are performed at the same temperature, allows these issues to be circumvented; however, these benefits come at the cost of a lower thermodynamic driving force for oxidation. Analyses of the thermodynamics and efficiencies of STC fuel production via isothermal redox cycling of CeO 2 have been performed [73, 74] Assuming a concentration ra tio of 3000 suns and an operating temperature of 1773 K, Bader et al. [73] determined that a gas phase heat recovery effectiveness of 95.5% would be required in order to achieve solar to fuel energy conversion efficiencies of 10% and 18% for CeO 2 based iso thermal WS (ITWS) and CDS (ITCS) cycles, respectively. Additionally, this efficiency may be augmented via an increase in the operating temperature (however, this would increase the likelihood of CeO 2 sublimation [25, 80] ), a reduction in the operating pressure of the T (e.g. 100 K) between the reduction and oxidation steps. Hao et al. [74] investigated the practicality of ITWS via reactor independent efficiency calculations, in which the respective energies requir ed for reduction of CeO 2 and vaporization and heating of the

PAGE 34

34 input H 2 O were considered relative to the energetic value of H 2 (taken as the higher heating value). The enthalpic contributions associated with the heating of inert gas were not considered; howe ver, Bader et al. [73] determined that this accounted for a large amount of the energetic penalties that would hinder the efficiency. Und er the abovementioned assumptions, Hao et al. [74] determined that the efficiency decreases sharply with increasing oxidant input in the absence of heat recovery. Ermanoski et al. [75] deemed ITWS unfeasible for efficient operation, citing high energy requirements for heating H 2 O as the major hindrance. Isothe rmal operation in the temperature range required for ample CeO 2 reduction requires large amounts of H 2 O (of which a majority is unreacted) to produce a relatively small amount of H 2 T > 0 was shown to be more efficient than ITWS, even when considering highly effective gas phase heat recovery and zero solid phase heat recovery. Experimentally, Hao et al. [74] performe d isothermal CeO 2 based WS cycles, showing the ability to produce H 2 despite the substantially decreased thermodynamic favorability of the hydrolysis reaction at such high temperatures (1673 to 1873 K). Hathaway et al. [71] performed 45 isothermal redox cycles with CeO 2 in a solar reactor designed by Bader et al. [81] wherein 95 and 93% of the sensible energies of the sweep gas and oxidizing gas, respectively, were recovered. The average solar to fuel energy c onversion efficiency for CDS, not considering the energy required to produce the sweep gas, was 1.64%; when considering this energetic penalty, the efficiency dropped to 0.72%. Perovskites Perovskites of the general form ABO 3 similar to those used in oxyg en permeable membranes [82] solid oxide fuel cells [83] an d metal air batteries [84] have been proposed as an alternative to CeO 2 for solar fuel production. When compared to CeO 2 perovskites show increased O 2 evolution during the high temperature reduction step and a notable decrease in

PAGE 35

35 operating temperatures [43 47, 85 92] The larger deviation from stoichiometry provides a higher upper limit for fuel production; however, the oxidation reaction must be sufficiently favorable such that the reduced perovskite can fully repleni sh the oxygen lost during reduction. This is an issue for many of the perovskites that have been analyzed thus far, unless excess oxidant is used to increase the thermodynamic driving force [43, 47, 85] However, c onsidering the large composition space of at least 27 suitable elements for the A site cations and 35 elements for the B site [93] perovskites offer the ability to tune the thermodynamic properties to a greater degree than with doped CeO 2 [94] Scheffe et al. [47] introduced lanthanum strontium manganese (LSM) perovskites as a new class of solar redox materials for use in two step STC WS and/or CDS. T he reduction extent of La 0.65 Sr 0.35 MnO 3 (LSM35) was about twice that of CeO 2 when reduced at 1773 K. Unlike CeO 2 however, LSM35 could not be completely re oxidized under any of the conditions investigated. Nonetheless, CO yields during CDS for LSM35 excee ded those of CeO 2 for the same operating conditions. Reduction extents of LSM based perovskites at a particular temperature and p O 2 have been shown to increase with increasing Sr 2+ content; however, this increase in reducibility was coupled with a decrease in WS and CDS favorability [45 47, 86] The introduction of a divalent dopant on the A site lead to a mixed Mn 3+ /Mn 4+ valence state, wherein higher Mn 4+ content (resulting from higher Sr 2+ dopant concentrations) f avored deeper reduction extents [46, 95] In terms of thermodynamics the partial molar enthalpy of reduction for LSM perovskites decreases monotonical ly with increasing Sr 2+ content, thereby decreasing the energetic penalties for reduction, while the entropy remains relatively constant [45] Such a change in thermodynamic properties favors a decrease in operating temperatures, but leads to T between redox steps. LSM based perovskites exhibit nota bly lower enthalpies and

PAGE 36

36 entropies than CeO 2 providing further explanation for the higher operating temperatures required for the latter. The effects of a variety of A site doping schemes for LaMnO 3 perovskites on redox performance have been evaluated. Th ese dopants include: Sr 2+ Ca 2+ Ba 2+ Y 3+ La 3+ Nd 3+ Sm 3+ Gd 3+ and Dy 3+ [45 47, 86 89] In general, La 1 x Ca x MnO 3 (LCM) perovskites have been shown to achieve deeper reduction extents than La 1 x Sr x MnO 3 (LSM) pe rovskites [44, 85, 89] Dey et al. [ 89] asserted that the improved redox performance of LCM based perovskites is a consequence of its orthorhombic crystal structure, as opposed to the rhombohedral structure of LSM perovskites. Analogous to the trends observed for LSM perovskites, O 2 evoluti on during reduction of LCM perovskites increases with increasing Ca 2+ dopant concentration. Furthermore, LCM perovskites exhibit ed diminished oxidation favorability when compared to CeO 2 La 1 x Ba x MnO 3 (LBM) perovskites showed lower O 2 evolution and fuel pr oduction than LSM perovskites, while similar trends regarding reduction and oxidation yields were observed [86, 87] Experimentally observed trends indicate that the size of the rare earth ion, the tolerance 2 ) are important in determining the viability of a material for STC WS and/or CDS [88] More specifically, O 2 evolution during reduction was shown to increase with decreasing rare 2 exhibit different redox performance, with the onset of the reduction reaction occurring at lower temperatures for the perovs 2 The largest reduction yields for these perovskites were observed for Y 0.5 Sr 0.5 MnO 3 (YSM50) and Y 0.5 Ca 0.5 MnO 3 (YCM50) [87, 88] YSM50 and YCM50 showed superior O 2 evolution and fuel yields when compared to those of CeO 2 ; however, neither showed completely reversible mass gain during oxidation. Demont et al observed traces of the hexagonal phase YMnO 3 in YSM50 after

PAGE 37

37 cycling; these changes were attributed to its low tolerance factor and, thus, greater distortion of the perovskite structure. When isothermally cycled, La 0.5 Sr 0.5 MnO 3 (LSM50) produced 3.5 and 2.8 times more CO than CeO 2 at 1773 and 1673 K, respectively [90] Total ITCS yields were bolstered further via the use of YSM50. The combination of A and B site doping creates a large composition space in which the redox performance of perovskites may be tuned. With regards to STC fuel applications, most experimental and theoretical efforts have involved the study of LaMnO 3 perovskites with Sr 2+ or Ca 2+ A site doping and Al 3+ B site doping. LSM and LCM based perovskites with Al 3+ B site doping (LSMA and LCMA) were found to release more O 2 during reduction than CeO 2 and Al 3+ free LSM and LCM based perovskites under comparable conditions [43, 44, 85, 92] For example, O 2 evolution during reduction of La 0.6 Sr 0.4 Mn 0.4 Al 0.6 O 3 (LSMA6446) at 1623 K was about 8 times greater than that of CeO 2 and the onset o f reduction began at a temperature 300 K lower [92] Furthermore, the reduction extent of La 0.6 Sr 0.4 Mn 0.6 Al 0.4 O 3 (LSMA6464) was 13 times larger than that of CeO 2 [85] and 150% greater than that of LSM40 when the respective meta l oxides were reduced under the same conditions [44] Reduction extents m ay be enhanced further when Ca 2+ rather than Sr 2+ is doped onto the A site, e.g. La 0.6 Ca 0.4 Mn 0.6 Al 0.4 O 3 (LCMA6464) achieved a deeper reduction than LSMA6464 [43, 44, 85] LSMA based compounds exhibited notable im provements in fuel yields compared to CeO 2 ; McDaniel et al. [92] observed 9 and 6 fold improvements in H 2 and CO yields, respectively. The oxidation kinetics have been shown to be similar to those of CeO 2 and cyclical stability with no degradation or significant changes in redox perform ance has been demonstrated for up to 80 redox cycles (wherein reduction and oxidation were performed at 1623 and 1273 K, respectively).

PAGE 38

38 The effects of varying A and B site doping schemes on the redox performance of perovskites have also been evaluated com putationally [43, 96, 97] Deml et al. [96] evaluated the vacancy formation energies (E v ) for LSMA based perovskites over a large composition space to elucidate their dependence on the respective A and B site dopant concentrations. Previous computationa l efforts by Kotomin et al. [98] and Merkle et al [99] for Ba 1 x Sr x Co 1 y Fe y O 3 (BS CF) perovskites showed a linear increase in E v with increasing Fe 3+ content for constant Ba 2+ and Sr 2+ dopant concentrations, where high E v indicates less energetically favorable reduction of the transition metal ion [100] Deml et al. [96] found that increasing Sr 2+ content at a constant Mn 2+/3+/4+ concentration yielded a decrease in E v allowing for more energetically favorable formation of oxygen vacancies. Increasing Mn 2+/3+/4+ content lead to either approxim ately no change or an increase in E v The optimal range of E v for STC redox cycling was found to be between 1.8 and 2.4 eV. Despite Al 3+ being redox inactive, its presence yields a notable effect on the redox performance of A and B site doped perovskites Ezbiri et al. [97] investigated the surface activity of LSMA and LCMA based perovskites and determined that Al 3+ doping on the B site leads to Mn 2+/3+/4+ enriched surfaces. X ray photoelectron spectroscopy (XPS) indicated that Mn is the only redox active metal on the surface, in agreement with XPS studies of LSM based perovskites [101] Enrichment of the perovskite surface with Mn wa s suggested to be the reason for the increased reduction extents of Al 3+ doped LSM and LCM based perovskites in comparison with their respective Al 3+ free counterparts. Moreover, Mn enrichment of the sur face of LSMA based perovskites wa s accompanied by Sr 2+ depletion; thus, these perovskites show ed a decreased tendency to form carbonates, wh ich has been observed experimentally by Cooper et al. [85] and Galvez et al. [102] For example, when oxidized at 923 K and 1123 K, the mass of

PAGE 39

39 LSM40 increased b eyond that expected of its fully oxidized state during CDS; this behavior was not seen during oxidation of LSMA6464, but was observed for LCMA6464 and LCM40 only during oxidation by CO 2 at 923 K. This irregular behavior was attributed to the formation of a carbonate phase, as the A site elements are known to form carbonates ( e.g. SrCO 3 or CaCO 3 ) and the excess mass was quickly removed during oxidation with air. The partial molar thermodynamic properties of LaMnO 3 perovskites with Sr 2+ or Ca 2+ A site doping and Al 3+ B site doping have been determined [43, 44, 85] In general, most LSM and LCM based perovskites with Al 3+ doping have partial molar enthalpies that decrease with increasing as opposed to Al 3+ free LSM and LCM perovskites whose enthalpies increase as the deviation from stoichiometry increases. Furthermore, the partial molar enthalpies and entropies of LSMA and LCMA perovskites are generally lower than those of CeO 2 The dim inished enthalpy values are responsible for the perovskites greater reduction extents at lower temperatures (due to a downward shift in the Gibbs free energy change). Ezbiri et al. [43] extracted thermodynamic properties for a wide variety of LSMA and LCMA based perovskites including: La 0.2 Sr 0.8 Mn 0.8 Al 0.2 O 3 (LSMA2882), La 0.4 Sr 0.6 Mn 0.6 Al 0.4 O 3 (LSMA4664), La 0.4 Sr 0.6 Mn 0.8 Al 0.2 O 3 (LSMA4682), LSMA6446, and La 0.6 Sr 0.4 Mn 0.8 Al 0.2 O 3 (LSMA6482). In terms of oxidation thermodynamics, it was shown that the re oxidation of CeO 2 is notably more favorable than that of the LSMA based perovskites (while having the least f avorable reduction thermodynamics ). Temperature shifts between thermodynamically favorable reduction and oxidation for CeO 2 were shown to be smaller than those of all of the perovskites examined thus far. Amongst the perovskites tested, LSMA6446 showed the most favorable oxidation thermodynamics; oxidation with CO 2 becomes thermodynamically favorable below 870 K. The least desirable oxidation thermodynamics were exhibited by LSMA2882, as favorable oxidation

PAGE 40

40 with CO 2 would require oxidation temperatures well below 400 K. Despite high during the reduction step, achieving high fuel yields with LSMA2882 necessitates the use of excess oxidant T between reduction and oxidation than for CeO 2 In accordance with observations made by Deml et al. [96] variation of the A site cation ratio while keeping the B site cation ratio constant played a larger role in tuning the redox behavior of LSMA based perovskites than variation of the B site cation ratio. Besides LaMnO 3 perovskites with Sr 2+ or Ca 2+ A site doping and/or Al 3+ B site doping, a number of other perov skites with varied compositions have been evaluated for STC redox cycling [46, 91, 94, 103] Jiang et al. [91] tested supported and unsupported LaFeO 3 perovskites with Sr 2+ or Ce 3+ A site doping and Co 3+ or Mn B site doping (La x A 1 x Fe y B 1 y O 3 ). In all cases, the unsupporte d La x A 1 x Fe y B 1 y O 3 compounds showed high O 2 evolution during reduction at 1573 K; however, the CO production during CDS at 1273 K was negligible. The poor fuel production of the unsupported A and B site doped LaFeO 3 perovskites was attributed to sintering ; thus, the perovskites were dispersed on ZrO 2 Al 2 O 3 and SiO 2 supports to improve the thermal stability at high temperatures. The use of SiO 2 supports was shown to greatly enhance the O 2 evolution rate and total CO production. Demont et al. [46] evaluated the redox performance of La x Sr 1 x CoO 3 La x Sr 1 x FeO 3 Ba x Sr 1 x Co y Fe 1 y O 3 LaSrCoO 4 and LaSrFeO 4 Despite having amongst the highest total O 2 productions during reduction, Ba 0.25 Sr 0.75 Co 0.8 Fe 0.2 O 3 (BSCF25/75/8/2), Ba 0.5 Sr 0.5 Co 0.8 Fe 0.2 O 3 (BSCF5582), and La 0.6 Sr 0.4 Co 0.2 Fe 0.8 O 3 (L SCF) performed very poorly in terms of their WS ability and overall fuel production. Bork et al. [94] measured the redox performances of La 0.6 Sr 0.4 Cr 1 x Co x O 3 perovskites. When reduced at 1473 K and oxidized at 1073 K with CO 2 La 0.6 Sr 0.4 Cr 0.8 Co 0.2 O 3 produced 25 times greater CO yields than CeO 2 The trends observed when varying the B site

PAGE 41

41 cation ratio included increased with increased Co 3+ content, reversible mass gain during CDS with Co 3+ doping up to 20%, and decreased oxidation favorability with Co 3+ dopant concentration of 50%. Emerging Materials Paired charge compensating doped (PCCD) CeO 2 [104, 105] and poly cation oxides (PCOs) [106] represent two potentially promising emerging classes of materials. Recently, Hoes et al. [104] and Muhich et al. [105] proposed PCCD CeO 2 formed by co doping pure CeO 2 with trivalent and pentavalent cations, as a new class of CeO 2 based metal oxides. The use of trivalent and pentavalent dopants in equimolar amounts was shown to y ield similar thermodynamic behavior to that of CeO 2 with a single tetravalent dopant ( e.g. Hf 4+ and Zr 4+ ), but with a more desirable balance between reducibility and oxidation favorability. It was found that co doping of Y 3+ with Nb 5+ at 5 mol% into CeO 2 l ead to a theoretical solar to fuel energy conversion efficiency of about 31%; in comparison, an efficiency of 26% was calculated for undoped CeO 2 Lastly, PCOs were discovered by Zhai et al. [106] PCOs, inspired by entropy stabilized oxides (ESOs) [107] are an oxide class containing three or more metal cations dispersed between two solid phases and were shown to thermochemically split H 2 O and produce more H 2 than CeO 2 when reduced at temperatu res as low as 1373 K. Thermodynamics of Solar Redox Material Performance Material thermodynamics are a paramount consideration when choosing or designing a solar redox material for two step WS and/or CDS. The thermodynamics of the reduction and oxidation r eactions will define the maximum possible extent to which they proceed, depending on the process temperature and environment (namely p O 2 ), and set an upper limit for solar to fuel energy conversion efficiency. Thermodynamics also indicates that the reducti on and oxidation temperatures cannot be equivalent when attempting to attain favorable conditions for

PAGE 42

42 both steps [108] A solar redox material with desi rable thermodynamic properties achieves an appreciable reduction extent at moderate temperatures (achievable with low enthalpy changes), while allowing for complete oxygen replenishment during the subsequent oxidation step. The difference between the requi red temperatures of the reduction and oxidation steps is of critical importance, as higher temperature swings imply greater energetic penalties associated with heating the metal oxide back to the reduction temperature and necessitate efficient solid to sol id heat recuperation [47] The temperatures which ultimately define the respective equilibria of the reactions of inter est are the reduction temperature ( T red ) and the oxidation temperature ( T ox ). T red and T ox may G ) for the respective reactions, i.e. G is zero [109] In practice, the respective reaction equilibria of both steps may be shifted by performing the reduction reaction at lower oxygen partial pressures, e.g. via the use of a chemica lly inert sweeping gas or operating under vacuum conditions, or by performing the oxidation step with excess oxidant [28] The Gibbs free energy change for a stoichiometric reaction is defined via the equation below [109] ( 1 6 ) Figure 1 1 G calculated for the reduction and WS reactions of stoichiometric redox pairs: ZnO/Zn, SnO 2 / SnO, and Fe 3 O 4 /FeO. Temperature dependent enthalpy and entropy values for stable phases of the involved chemical species between 400 and 2000 K were obtained from Fact Web [110] and values for formation enthalp y at standard conditions were taken from the NIST Chemistry WebBook [111] Figure 1 1 may be used to determine the temperatures at which the reduction and oxidation reactions reach their respective thermodynamic equilibria. For

PAGE 43

43 G are due to changes in phase or crystal structure which result in abrupt chan ges in the enthalpy and entropy. For nonstoichiometric redox pair s, the thermodynamic properties are defined by considering an infinitesimal change in the oxidation state of the metal oxide [16] Thus, the cha h O s O ), and partial molar Gibbs g O ) per mole of monoatomic oxygen may be defined in accordance with the equation below. ( 1 7 ) h O s O are known for a particular metal g O may be quantified via equation ( 1 8 ) These relationships are derived by assuming that the activity of the metal oxide is unity, the evolved O 2 h O s O are independent of the temperature and only dependent on [44] The Gibbs free energy change for nonstoichiometric reduction, g red is determined by evaluating equation ( 1 8 ) at the reduction extent of interest ( red ). ( 1 8 ) ( 1 9 ) Equation ( 1 8 ) h O s O may be calculated relatively easily. ( 1 10 ) Given empirical oxygen nonstoichiometry data as a function of temperature and p O 2 eq uation ( 1 10 ) may be used to determine h O s O From the form of eq uation ( 1 10 ) it is h O s O are found by evaluating the slope and intercept, respectively, of the

PAGE 44

44 logarithm o f p O 2 versus inverse temperature for a constant reduction extent. The partial molar enthalpies and entropies for several solar redox materials, obtained from literature, are presented in Figure 1 2 and Figure 1 3 The data in these figures was c ompiled fro m Yang et al. [45] and Takacs et al. [28, 44] h O for LSM based perovskites decreases with increasing Sr 2+ dopant h O s O for all of the perovskites shown are lower than those of pure CeO 2 s O for LSMA6464 are larger than those of the LSM based perovskites. As will be discussed in detail below, h O at higher reduction s O s O allows for smaller T h O enables re h O s O are of critical importance for determining the temperature domains in which equilibrium between the reactants and products is expected for the reactions of interest. This is evident in Figure 1 4 which g red versus temperature for the same metal oxides as shown in Figure 1 2 and Figure 1 3 Here, red has been chosen separately for each metal oxide such that the ratio of oxygen mass released to the mass of one mole of the metal oxide is equated. T his is done to ensure that comparison of the respective thermodynamic equilibria is done in a mathematically and physically unbiased manner. h O s O on the reduction thermodynamics are evident from the g red As v g red < 0 are desirable, thermodynamic favorability h O s O At the reduction extents presented in Figure 1 4 h O s O of CeO 2 h O being over 100 kJ mol 1 h O s O of LSM20, the smallest of the perovskites examined herein, was nearly half that of CeO 2 The

PAGE 45

45 poor reduction behavior of LSM20 under the examined conditions results from the combination s O h O compared to the other perovskites. LSMA6464, having s O h O about 150 kJ mol 1 lower than that of CeO 2 g red = 0 at about 1915 K. The chang e in Gibbs free energy for the reverse of equation ( 1 7 ) g ox ), the negative of g O (eq uation ( 1 8 ) ) at a given nonstoichiometry ( ox ) and temperature is shown below. ( 1 11 ) The Gibbs free energy change for oxidation with H 2 g ox,H 2 O ) may then be calculated g ox and the Gibbs free energy change for the formation of H 2 g H 2 O ). Temperature g H 2 O were obtained from the NIST JANAF thermochemical tables [111] Figure 1 5 g ox,H 2 O versus temperature for the abovementi oned metal oxides, wherein ox was selected for each in the same manner as red for Figure 1 4 ( 1 12 ) As was the case for the conditions shown in Figure 1 4 h O s O of CeO 2 are the largest amongst the metal oxides examined under oxidation conditions. In contrast to the low h O that are desirable for reduction, h O at lower allows for the oxidation reaction to be performed at higher temperatures. Thus, CeO 2 exhibits the most favorable oxidation thermodynamics. In contrast, most of the perovskites show reduced oxidation favorability largely as a result of s O h O does not trend in the same direction with as that of CeO 2 Figure 1 4 and Figure 1 5 serve as a qualitative confirmation of experimentally observed trends regarding the respective redox performances of perovs kites in comparison to CeO 2 In general, perovskites are more easily reduced, mostly

PAGE 46

46 resulting from their smaller enthalpies at higher nonstoichiometries; however, oxidation with H 2 O (or CO 2 ) requires larger temperature swings. The diminished entropy chang es of the perovskites, in comparison with those of pure and doped CeO 2 cause the Gibbs free energy to be less sensitive to changes in temperature (i.e. because the slope is smaller in magnitude) and necessitates operation of the oxidation step at notably lower temperatures or wit h high excess oxidant flows. To illustrate why this is the case, Figure 1 6 h O s O in g red g ox,H 2 O for CeO 2 and three fictitious metal oxides whose thermodynamic properties were defined by applying multiplying factors to those of CeO 2 are shown. The thermodynamic properties of the three fictitious metal h O h O,ceria s O s O,ceria h O h O,ceria s O s O,ceria h O h O,ceria s O s O,ceria h O g red downwards and g ox,H 2 O upwards, while the sensitivity (or slope) remains unchanged. The downwa rd shift of the reduction line enables the onset of the reduction reaction at a lower temperature; however, due to g H 2 O increas es monotonically with temperature, this change causes an increase in T required for thermodynamic ally favorable redox cycle operation. The effects s O s O (and, thus, the respective magnitudes of the slopes of the reduction and oxidation lines) allows for operation of the reduction step at a lower temperat T h O s O will lead to a further decrease in T red T will be larger than that of the case s O is increased. The trends observed for the first fictitiou s material are representative of those observed for LaMnO 3 perovskites with varying doping schemes;

PAGE 47

47 however, for the perovskites, the effects are exacerbated because their entropies are also lower than those of CeO 2 Neither the second nor the third fictit ious materials exist in the literature, as s O,ceria are larger than those of all other nonstoichiometric solar redox materials discovered thus far. h O s O quantitatively affect oxidation yields, equilibrium molar H 2 yields f or CeO 2 and the abovementioned fictitious metal oxides were calculated for a range of oxidation temperatures. WS performance w as calculated via equation ( 1 8 ) and the equilibrium constant for the dissociation of H 2 O ( K w ) [28] ( 1 13 ) ( 1 14 ) Here, n H 2 O,i is the initial H 2 O input, n H 2 is the equilibrium molar H 2 yield, and red is the reduction extent attained prior to the WS reaction. Utilizing an iterative root finding routine, e q uation ( 1 14 ) may be solved for n H 2 at any temperature. Figure 1 7 shows predicted equilibrium fuel yields for CeO 2 and the abovementioned fictitious redox m aterials for oxidation temperatures between 400 and 1500 K. In all cases, red = 0.05 and n H 2 O,i = red The reduction temperatures ( T red g red = 0, are shown for each material. In general what can be elucidated from this figure is that there is a clear connection between reduction temperature and oxidation temperature. For all material compositions, the oxidation temperature decreases as the reduction temp erature decreases. T o maximize the solar to fuel energy conversion efficiency one would like to maximize fuel yields

PAGE 48

48 while decreasing the temperature swing between reduction and oxidation steps ( i.e. the efficiency is proportional to the ratio of n H 2 and T ). h O s O affect the ratio of n H 2 to T this ratio has been plotted h O s O relative to the thermodynamic properties of CeO 2 in Figure 1 8. Here, the fuel yields are defined as the maximum fuel yields obtainable for oxidation temperatures between 400 and 1500 K (from Figure 1 7). The reduction and oxidation g red red = 0.05 and the temperature where n H 2 T i s defined as shown in the equation below ( 1 15 ) All combinations where s O s O,Ceria h O h O,Ceria have larger ratios than those of CeO 2 s O is desirable. Moreover, a redox s O s O,Ceria exhibits a decreased ratio for all combinations, indicating t hat these materials are not as desirable from a thermodynamic perspective. This trend is also observed for h O Considering the thermodynamic properties for candidate materials recently evaluated ( Figures 1 2 and 1 3 s O s O,Ceria h O h O,Ceria Thus, the current trends and thermodynamic limitations suggest that perovskites currently do not achieve thermodynamic properties within the desired entropic range (above that of CeO 2 ). For this, advanced co mputational efforts, such as those of Deml et al. [96] will be critical to determine the composition spaces in which the enthalpies and entropies of candidate redox materials lie within a more favorable regime. However, such an endeavor may prove to be difficult because it is known that the ran ge of entropy changes of metal oxides is limited [20]

PAGE 49

49 Additional Criteria for Redo x Material Viability Thermodynamic equilibrium is not an exclusive means of quantifying the viability of a candidate STC redox material, as it does not consider the temporal and spatial aspects of material performance under realistic operating conditions. While thermodynamics serve as a means of qualitatively and quantitatively assessing the upper limit of operational efficiency, the reaction kinetics must be sufficient such that they offer notable reducibility and fuel yields within a reasonable amount of time ( i.e. an infinitely small reaction rate will require an infinitely large reactor to produce useful amounts of fuel). Reduction and oxidation kinetics will determine the extent to which the respective reactions proceed in a given amount of time, thereb y affecting the practicality and efficiency of operation in a real reactor. For the likely case where the reduction and oxidation kinetics are not equivalent, this disparity may be addressed by decoupling the two steps such that their allotted reaction tim es may be controlled separately [18] As previously discussed, thermodynamics indicates that the WS or CDS reactions require te mperatures less than those of the prior solar driven reduction step, making it more likely that this step will be limited in terms of kinetics, rather than heat or mass transfer [20] In this case, increasing T ox at the expense of thermodynamic favorability, would increase the reaction rates. Additionally, effective surface area and mor phology have been shown to play a role in determining the reaction kinetics [14, 25, 112 114] Lastly, fast thermal and mass transport is required to limit temperature gradients and diffusion based kinetic limitati ons. In addition to thermodynamics and reaction kinetics, the physical and chemical attributes of the reactive intermediate must also be considered. As the processes of interest occur at elevated temperatures, the morphological stability of the redox mater ial must be known; phase changes, such as the solid to gas transitions seen during operation with volatile redox pairs, can complicate operation ( e.g. requiring rapid quenching of the produced species to prevent

PAGE 50

50 unwanted recombination) and sintering may ca use unwanted degradation and reduces the 2 exhibits noteworthy physico chemical stability during cyclical operation. During 1000 isothermal CDS cycles at 1773 K, fibrous CeO 2 maintained a porous microstructure and 89% of its peak fuel yields were retained after the first 300 cycles [115] However, in a reactor heated by concentrated radiation, where temperature gra dients are more severe, sintering and reactivity with insulation have been observed [29] In general, i t is known that sintering and material creep are relevant at temperatures exceeding 40% of the [108] These types of changes ne gatively impact the reaction kinetics and may necessitate implementation of microstructural design strategies. Furthermore, sublimation of the metal oxide has been shown to be noteworthy [16, 80, 116] For reducti on of CeO 2 to Ce 2 O 3 at 2273 K and total pressures between 100 and 200 mbar, Abanades and Flamant [80] observed CeO 2 vaporization of greater than 50%. At the r eduction temperatures typically employed, vaporization may cause significant losses of active material over a long period of time and should be considered when determining the viability of a material. Furthermore, as observed by Furler et al. [29] and Marxer et al. [19] sublimation of the metal oxide may negatively impact the achievable reactor temperatures if the sublimed material is deposited on the window or CPC of the reactor. Conclusion Development and commercialization of highly efficient thermochemical reactors capable of converting solar energ y into fungible fuels would allow for the use of renewable fuel sources within the currently adopted infrastructure. Compared to CeO 2 the current state of the art redox material, an ideal material should exhibit larger entropy changes during oxygen exchan ge. This results in reducing the temperature swing between reduction and oxidation. Furthermore, it should exhibit lower enthalpy changes at higher oxygen nonstoichiometries and relatively higher

PAGE 51

51 enthalpy changes at lower nonstoichiometries. This would ena ble oxygen release at relatively lower temperatures and oxidation at higher temperatures, compared to a material with constant enthalpy as a function of nonstoichiometry. As new candidate redox materials are considered, it is important to carefully conside r the oxidation conditions under which their performance is assessed. Experimental characterization should be performed under realistically relevant conditions (e.g. H 2 O:H 2 or CO 2 :CO < 10:1) that are representative of large scale operation and not with exc essive oxidant gases as is typical. Ultimately, achievement of economically viable STC fuel production is reliant upon increasing the attainable solar to fuel energy conversion efficiency. This is a stepwise process that should involve: (1) material disco very via data extraction from the literature and databases or high throughput computational screening, (2) experimental determination of thermodynamic properties, (3) synthesis and experimental characterization of redox performance ( e.g. cycling of enginee thermophysical properties ( e.g. kinetics and thermal conductivity), (4) process optimization via theoretical open system modeling, and (5) reactor design and development. Recent advancements point towards a promising future for renewable fuel production via high temperature thermochemical redox cycles. However, for this technology to become commercially viable, where prototype reactors demonstrate efficiencies of about 20% or higher, further advancements in material properti es and reactor design are of critical importance.

PAGE 52

52 Figure 1 1. Gibbs free energy change versus temperature for the reduction and oxidation of ZnO, SnO 2 and Fe 3 O 4 redox pairs. Figure 1 2. Partial molar enthalpy versus nonstoichiometry for the reduc tion of CeO 2 and Zr doped CeO 2 (circles), LSM based perovskites (diamonds), and LSMA6464 (crosses).

PAGE 53

53 Figure 1 3. Partial molar entropy versus nonstoichiometry for the reduction of CeO 2 and Zr doped CeO 2 (circles), LSM based perovskites (diamonds), and L SMA6464 (crosses). Figure 1 4. Gibbs free energy change versus temperature for the reduction of CeO 2 and Zr doped CeO 2 (circles), LSM based perovskites (diamonds), and LSMA 6464 (crosses).

PAGE 54

54 Figure 1 5. Gibbs free energy change versus temperature for the oxidation of CeO 2 and Zr doped CeO 2 (circles), LSM based perovskites (diamonds), and LSMA 6464 (crosses) with H 2 O. Figure 1 6. Gibbs free energy change versus temperature for the reduction and oxidation of CeO 2 and several fictitious metal oxides wi th varied partial molar thermodynamic g red g ox were calculated for a red of 0.05 and ox of 0.01,

PAGE 55

55 T ) required for operation of each metal oxide at ambient conditions within the thermodynamically fav orable ranges are shown. Figure 1 7. Predicted equilibrium H 2 yield for the WS reactions of CeO 2 and three fictitious metal oxides versus oxidation temperature. Initial H 2 O input is equated to the reduction extent prior to oxidation (i.e. n H 2 O,i = red ). Prior to the WS reactions shown, red is set to 0.05 for each metal oxide.

PAGE 56

56 Figure 1 8. T ) for numerous combinations h O s O Oxidation calculation s were performed at temperatures between 400 and 1500 K for red = 0.05 and n H 2 O,i = red

PAGE 57

57 CHAPTER 2 EXPERIMENTAL FRAMEWORK FOR EVALUATION OF THE THERMODYNAMIC AND KINETIC PAR AMETERS OF METAL OXIDES The reduction extents of perovskite materials like LSM40 and LSMA6464 can be substantially great er than those of ceria under comparable reduction conditions [44, 92] For example, at a reduction temperature of 1600 K and p O 2 of 10 3 bar, the mass specific oxygen evolution for LSM40 and LSMA6464 is about 0.047 and 0.113 mol kg 1 respectively; under these same conditions, ceria was shown to release 0.017 mol kg 1 [44] Furthermore, LSM40, LSMA6464, and other perovskites have been shown to outperform ceria in terms of total fuel productivity when the product gas was substantially diluted via the use of large oxidant inputs (thereby increasing the thermodynamic driving force for the oxidation reaction, but lowering the oxidant to fuel conversion) [47, 85] However, as a result of their less favorable thermodynamic (low p O 2 and high conversion) is expected to result in net fuel yields that are less than th os e of ceria under most conditions Cooper et al. calculated the equilibrium CO yields of ceria, LSMA6464, LCMA6464, and LCM40 during CDS after reduction at 1673 K and p O 2 of 10 5 bar [85] In spite of ach ieving up to five times that of ceria, the perovskites produced less fuel than ceria in cases of low excess oxidant delivery and moderate temperature swings between reduction and oxidation. Nevertheless, excess oxidant concentrations greater than those e xpected in solar reactors are usually used in experimental studies. For example, in WS and CDS experiments performed with LSMA based materials by McDaniel et al. [92] the H 2 O:H 2 or CO 2 :CO ratios were always greater than 100:1. In contrast, Furler et al. [14] have performed CDS with ce ria in a solar cavity receiver where the molar CO 2 :CO ratio at peak CO production was 6:1.

PAGE 58

58 Although it is known that the oxidation conditions strongly affect performance, in practice widely varying oxidant concentrations are used during experimental charac terization of candidate materials. As a result, comparison of material performance (namely, fuel yields and kinetics) can be difficult to infer from the literature. Often, oxidant concentrations (or p O 2 ) during oxidation are not reported, making comparison even more difficult. For example, Dey et al. [8 8] and Demont and Abanades [87] have measured the O 2 and CO production capabilities of Y 0.5 Sr 0.5 MnO 3 (YSM50) during red ox cycling. Both demonstrated high reduction extents at 1673 K, but the fuel yields were drastically different, presumably because of different CO 2 :CO ratios used during oxidation; however, this data was not provided. In this chapter, the development of a high temperature tubular reactor is thoroughly described in which the fundamental thermodynamic and kinetic behaviour of thermochemical materials can be readily assessed. This reactor system is capable of operating at temperatures up to 1873 K, total press ures ranging from vacuum to ambient and p O 2 as low as 10 29 atm. Compared to off the shelf systems like thermogravimetric analyzers (TGA) or indirect conductivity based measurement systems, this system has three inherent benefits: (1) the flexibility to c ontrol the sample morphology ( e.g. powder, packed bed, reticulated porous ceramic, or pellet), (2) the potential for a well developed and characterized flow, and (3) the ability to readily customize the system on demand ( e.g. easy integration with a steam generator to control and operate at very low p O 2 ). T he sample environment in this system is well characterized and precisely controlled, enabling the thermodynamic and kinetic parameters of candidate redox materials to be measured under appropriate ( i.e. r elatively low p O 2 during oxidation) and easily reproducible conditions. These parameters were measured using a novel experimental scheme wherein the p O 2 was

PAGE 59

59 controlled by subjecting the sample to a H 2 /H 2 O environment and concomitant changes in the oxidatio n state were measured in response to rapid shifts in the H 2 O delivery. In contrast to other experimental efforts in identifying the performance of redox materials, this procedure enables direct comparison of the viability of different materials and the use of low oxidant to fuel ratios ( e.g. H 2 O:H 2 of 10:1), a critical parameter in assessing the solar to fuel efficiency. The system enables high gas delivery, reducing the effect that external mass transfer limitations may have on the measured reaction rate. As a result of sufficient and uniform gas flow and the use of a controlled sample environment, both thermodynamic and kinetic insights into the performance of redox materials may be obtained in a single experiment, as opposed to evaluating fuel yields in a TGA and kinetics in a separate reactor with more desirable gas dynamics [106] The reactor system and experimental methods were validated by performing isothermal relaxation experiments with undoped ceria, wherein the sample environment was rapidly altered by stepwise changes in the delivered H 2 O vapor concentration, and co mparing measured oxygen nonstoichiometries with accepted data available in the literature. Data was measured at temperatures from 1173 1473 K and p O 2 from 4.5410 18 1.0210 9 atm. The measured equilibrium data displayed strong agreement with the literatur e and the expected trends were preserved. Kinetic data was extracted by first transforming reactant concentrations measured downstream of the reaction zone using a tanks in series mixing model to account for gas dispersion. Next, a mechanistic kinetic mode l distinguishing surface and bulk species concentrations was fit to the data to extract pertinent thermodynamic and kinetic parameters. The model assumed a two step reaction mechanism mediated by the formation of an intermediate hydroxyl species on the sur face. Activation energies and defect formation enthalpies and entropies for the forward and reverse reactions were found to be in good agreement with

PAGE 60

60 previous modelling efforts, providing further validation of the use of this system to explore thermodynami c and kinetic behaviour of emerging thermochemical materials. Experimental High Temperature Tubular Reactor Setup The experimental schemes to be described herein were performed in a custom high temperature tubular reactor system capable of precisely setting the sample environment via control of the reactor temperature ( T reactor ), total pressure ( p tot ), and the p O 2 Figure 2 1 shows a simplified schematic of the experimental system. A horizontally oriented high temperature tubular furnace (Carbolite STF 16/180), capable of achieving temperatures up to 1873 K, is used to control the temperature within an Al 2 O 3 work tube (inner diameter of 19.05 mm). An absolute pressure controller (MKS Type 640B) coupled with a rotary vane pump (Edwards E2M0.7) permits investigation of redox performance at p tot from 0.2 mbar to ambient pressure, and the p O 2 is controlled by utilizing H 2 H 2 O, and Ar mixt ures of varying gas and vapor concentrations. Gaseous H 2 O input is enabled via a vapor delivery system consisting of a mass flow controller (MKS GE50A) to supply Ar (Airgas, Ultra High Purity 5.0 Grade Argon) as a carrier gas, a liquid flow controller (Bro nkhorst LIQUI FLOW) providing low quantities of deionized H 2 O with allowable setpoints ranging from 0.18 9 g hr 1 and a Bronkhorst Controlled Evaporator Mixer (CEM). Gaseous Ar and liquid H 2 O enter the mixing chamber, forming an aerosol; the mixture then passes through a heat exchanger maintained at 473 K where total evaporation is achieved. Two additional MKS GE50A mass flow controllers are used to deliver Ar as an inert sweeping gas and a 5% H 2 mixture balanced in Ar (Airgas, Argon 95 Hydrogen 5). To pre vent condensation of the vapor mixture prior to entering and after exiting the heated work tube, all reactor plumbing downstream of the vapor delivery system has been affixed with flexible electric heating cable and basalt fabric insulation. The temperatur e of the stainless steel tubing is then

PAGE 61

61 maintained at 393 K using two Omega CN79000 series dual zone temperature controllers. Prior to being sampled by a residual gas analyzer (Stanford Research Systems QMS100), the reactor effluent passes through a vacuum trap submerged in an ice bath in order to remove the vast majority of unreacted H 2 O. The furnace temperature, input gas and vapor flowrates, and p tot are controlled via an in house developed LabVIEW VI and the product evolution, and reaction rates are evaluated using custom MATLAB scripts. Experimental Procedure and Evaluation of the Sample Averaged Nonstoichiometry Reaction rates and equilibrium of undoped ceria (2000 mg, Alfa Aesar REacton 99.9%, via isothermal relaxation experiments performed in a H 2 /H 2 O environment. The powdered sample was placed in an Al 2 O 3 combustion boat (Sigma Aldrich, Z561738, 70 mm 14 mm 10 mm, 5 mL capacity) and placed within the uniform temperature region of the work tube. A constant H 2 flowrate was utilized throughout the entire duration of an experiment to establish an equilibrium ( i.e. no net reaction) background signal. Once a desired temperature was achieved, the p O 2 was changed via rapid, stepwise shifts in the delivered H 2 O concentration and reaction yields and rates were measured by analysis of the subsequent relaxation of the H 2 production or consumption back to its baseline. The maximum delivered H 2 O:H 2 ratio during oxid ation used in these analyses was 36:1 a nd the minimum was 0.5:1. During heating and cooling, H 2 and H 2 O flowrates remained constant; consumption or production of H 2 due to re duction or oxidation was quantified by integration of the H 2 signal as it deviated from its baseline. The p O 2 was determi ned by assuming equilibrium of the H 2 O thermolysis reaction (H 2 O H 2 +0.5O 2 ), which was calculated via the following equation. ( 2 1 )

PAGE 62

62 p i s the standard state reference pressure (1 atm), K w is the equilibrium constant for temperature facilitated H 2 O dissociation (obtained from the NIST JANAF thermochemical tables [111] ), p H 2 O and p H 2 are the respective partial pressures of H 2 O and H 2 and F H 2 O,i and F H 2 ,i are the respective input molar flow rates of H 2 O and H 2 (standardized to 298 K and 1 atm). Other studies have us ed H 2 /H 2 O or CO/CO 2 mixtures to produce low p O 2 atmospheres [16, 66, 117] For example, Tuller and Nowick [117] utilized CO/CO 2 mixtures to control the p O 2 from 10 22 10 4 atm and confirmed agreement between the calculated p O 2 and that measured near their sample. Due to the high total flow rates used during experimentation (600 sccm or hig her) and the gas dynamics of the system, it was assumed that the loc al p O 2 seen by the sample was dictated by the delivered gases ( i.e. the product gases were swept away sufficiently fast such that they did not influence the sample environment). When the p O 2 is decreased by reducing the H 2 O input, the oxygen concentration gradient between the reactor environment and the sample drives the simultaneous reduction of the sample (thus, creating oxygen vacancies) and oxidation of H 2 to form H 2 O. This reaction proceeds according to the equation below : ( 2 2 ) where is the difference between the reduction extents after and before the reaction ( i.e. = f i ). By numerically integrating the difference between the transient outlet molar H 2 flow rate ( F H 2 ) and its baseline value prior to the reaction (defined here as the value of F H 2 at a reaction time of zero), the total changes in H 2 O and H 2 ( N H 2 O ,rxn and N H 2 ,rxn respectively) and, thus, the sample averaged change in were determined according to the equation below ( 2 3 )

PAGE 63

63 Here, N CeO 2 t rxn is the total tim e elapsed during the reaction being analyzed from its start ( t = 0) to the time at which equilibrium was achieved ( t = t end ), i.e. the relaxation time. Conversely, increasing the p O 2 by increasing the H 2 O input leads to concurrent oxidation of the sample and reduction of H 2 O to yield a sample with fewer oxygen vacancies and gaseous H 2 This results in an increase in the measured H 2 flow rate in the effluent. This reaction represents the reverse of that described in eq uation ( 2 2 ) ; however, to maintain consistency with eq uation ( 2 3 ) = i f (and, thus, the ca will be negative). Using a procedure analogous to the one described for the reduction reaction, the H 2 increase measured by the mass spectrometer is monitored and numerically integrated using eq uation ( 2 3 ) to quantify the total H 2 production during the oxidation reaction and, thus, the change in oxidation state. A detailed description of the experimental procedure is herein provided Duri ng all experiments, the total pressure was maintained at 1 atm. Each relaxation experiment consisted of five distinct stages: 1. Under ambient conditions, the H 2 and Ar flow rates were set and sufficient time was given (approximately 2 hr) to allow the mass s pectrometer measurements to stabilize. At this point, varying H 2 :Ar flow rate ratios were measured in order to produce a linear calibration by which the H 2 flow rate in the effluent may be quantified during the reduction and oxidation reactions (as Ar deli very is constant and known throughout the experiment). After calibration, the H 2 and Ar flow rates were kept constant for the remainder of the experiment. 2. Prior to heating to the temperature of interest for the particular experiment ( T exp ), the sample was heated to a lower temperature (773 K) that was sufficiently high to allow H 2 O vapor to be delivered through the work tube without condensing and low enough to prevent notable reduction of the sample, i.e. such that there was no reaction induced decrease in the H 2 measurement and the initial was approximately zero. 3. Once the system was given sufficient time to equilibrate, nonstoichiometric reduction of the sample was initiated by increasing T reactor to T exp in the presence of H 2 H 2 O, and Ar. 4. At T exp the reduction reaction was allowed to re ach completion and subsequent redox reactions were performed isothermally via randomized, stepwise shifts in the input H 2 O

PAGE 64

64 flow rate. The sample environments for the respective reduction and oxidation reactions were maintained until equilibrium was achieve d (indicated by a steady H 2 measurement). 5. Lastly, the ceria sample was re oxidized to its starting state by cooling the reactor while maintaining the gas and vapor flows. A similar procedure has also been successfully utilized wherein the reactions perform ed during the fourth segment were driven by changes in the reactor temperature. Here, the H 2 H 2 O, and Ar flow rates were kept constant throughout the entire experiment. These nonisothermal experiments are conceptually similar to the O 2 release and uptake experiments performed by McDaniel et al. during which samples were heated and cooled (leading to reduction and oxidation, respectively) in a constant flow of O 2 and He [92] Results and Discussion Equilibrium and Transient Oxygen Nonstoichiometry Measurements of Undoped Ceria The abovem entioned isothermal experimental procedure was employed in order to elucidate the reaction rates and equilibrium oxidation states of undoped ceria under conditions relevant to those under which oxidation may be performed in solar reactors ( i.e. high conver sion and low p O 2 ). Figure 2 2 shows exemplary measured data for an experiment performed at T exp = 1373 K. Figure 2 3 shows the concomitant changes in (left axis, solid line) and p H 2 O/ p H 2 (right axis, dashed line) over the course of the same experiment. Here, the sample was first reduced by heating from 773 to 1373 K at a rate of 20 K min 1 with p H 2 O/ p H 2 = 2.25. The H 2 signal noticeably decreased durin g heating, followed by an increase back to the baseline soon after reaching T exp = 1373 K. At this point, the p O 2 was estimated to be 4.8310 13 atm and f was 0.0679. Reduction and oxidation reactions were then initiated by varying the H 2 O input between 7 1 ( p H 2 O/ p H 2 from 1.12 to 6.75), as seen on the orange line (right axis) of panel (b) in Figure 2 2 from about 4750 10930 s. Upon doing so, subsequent increases or decreases in the H 2 signal (for oxidation or reduction, respectively), f ollowed by a slow decay

PAGE 65

65 back to the baseline, can also be seen. For this experiment a maximum of 0.0908 ( p O 2 = 1.2110 13 ) and a minimum of 0.0334 ( p O 2 = 4.3510 12 ) were measured. In the last segment, the sample was completely re oxidized by cooling t o 773 K under the same flow conditions as utilized during the initial nonisothermal reduction reaction. This can be observed by the increase in the H 2 signal as the sample was cooled. In total, isothermal relaxation experiments were performed at temperatur es from 1173 1473 K and p O 2 from 4.5410 18 1.0210 9 atm. Figure 2 4 shows the calculated of undoped ceria versus time, all normalized to a starting time of zero, during isothermal reduction and oxidation reactions performed at 1473 K. The results clearly show that the final achieved is dictated only by the reaction temperature and p O 2, and not the initial This is exemplified by considering the blue (open circles) solid and dashed lines in panel (a) of Figure 2 4 ; while the initial for the respective reactions are 0.106 and 0.062, respectively, the achieved after the reactions are com parable. Further evidence of this trend may be obtained by considering the orange (diamonds) solid and dashed lines. Comparing the blue and orange lines, it is clear that the final achieved in the case of the former is greater than that of the latter as a result of the lower p O 2 employed. Repeatability was also observed; approximately equivalent transient profiles were seen for reactions in which the initial and sample environment were the same. Comparable conclusions can be observed from the oxidatio n reactions shown in panel (b) of Figure 2 4. Figures A 1 to A 3 of Appendix A show versus time for isothermal reactions performed between 1173 and 1373 K, and the same trends are observed. We note that Zhai et al. [106] have also evaluated H 2 production during the oxidation step in the presence of a H 2 and H 2 O mixture dur ing experiments performed with PCOs in a TGA. However, the ratios of

PAGE 66

66 delivered H 2 O to H 2 used in their studies are orders of magnitude larger than those employed herein. Equilibrium values of the aforementioned data are shown in Figure 2 5 where calculated oxygen content of undoped ceria (2 ) is plotted as a function of temperature and p O 2 This data is compared to thermogravimetric measurements made by Panlener et al. and defect model calculations by Tuller and Nowick [16, 117] A piecewise cubic Hermite interpolating polynomial (PCHIP in MATLAB) was used to interpolate the measurements from Panlener et al. and directly compare to the experimental results. The average percent differences for experimental data compared to Panlener et al at 1173, 12 73, 1373, and 1473 K were 7.1, 7.8, 9.1, and 4.3%, respectively; thus, good quantitative agreement with accepted literature was attained. C ontributions to the error in the calculations (namely and p O 2 ) include: uncertainties in the delivered gas and vapor flow rates, the assumptions regarding thermal equilibration of the sample and delivered gases and thermodynamic equilibrium of the H 2 O thermolysis reaction, errors inherent to the mass spectr ometer or due to signal drift over the course of an experiment, and errors in analysis induced by the use of a numerical integration routine. Panels (a) and (b) of Figure 2 6 show d /d t versus for reduction and oxidation reactions, respectively, at 1473 K. Reaction rates were obtained by numerically differentiating the transient measurements with respect to reaction duration using a second order three point differentiation routine [118] In all cases, the rat es were characterized by a rapid initial change in d /d t corresponding to the time and initial at which the p O 2 was shifted by altering the H 2 O input. Following this initial spike, the rate approaches zero as approaches the equilibrium values corresponding to the temperature and p O 2 under which the reac tion was performed. As would be expected, the rates are dependent upon the temperature and p O 2 During reduction of

PAGE 67

67 the sample, the reaction rate at a constant was inversely proportional to the p O 2 In contrast, the rates during oxidation of the sample b y H 2 O were hindered by decreasing the p O 2 Importantly, equivalent ( or near equivalent) d /d t at a fixed composition were measured for reactions performed under the same temperature and p O 2 irrespective of the value of prior to the reaction giving conf idence in the experimental approach. This can be observed by considering the blue (open circles) solid and dashed lines in panel (a) of Figure 2 6 ; here, despite having significantly different initial d /d t for both reactions were equivalent when the in stantaneous values of were the same. Figures A 4 to A 6 of Appendix A show d /d t versus for the isothermal reactions performed at temperatures between 1173 and 1373 K. Mechanistic Modeling of the Kinetics of Ceria To elicit kinetic information with mec hanistic interpretation, the redox behavior of ceria during H 2 oxidation or H 2 O splitting reactions was analyzed assuming a reversible two step, surface mediated reaction mechanism that accounts for adsorption and dissociation of reactants, surface charge transfer, bulk to surface oxygen transport, and association and desorption of products. The reaction mechanism, written in Krger Vink notation, is shown below: ( 2 4 ) ( 2 5 ) where , , and are all surface defect species and refer to an oxygen vacancy, oxygen anion, hydroxyl group incorporated on an oxygen vacancy, Ce 3+ cation, and Ce 4+ cation, respectively. Model formulation and accompanying assumptions are thoroughly discussed in the kinetic analysis of chemical looping WS over ceria performed by Zhao et al. [119] The equations used to describe surface and bulk defect concentrations and the objective function used in the minimization routine were kept consistent to enforce similarity between

PAGE 68

68 model implementations. However, unlike Zhao et al. who assumed that the local sample environment was identical to downstream mass spectrometric measurements [119] in this analysis the measured rate was examined after removing the effects of gas phase mixing and disp ersion, an artefact of the experimental setup [35, 92, 120] Furthermore, local reactant partial pressures are assumed to be determined by the delivered gases, a valid assumption under the conditions of an infinite ly thin sample and sufficiently high sweep gas flow rates. To simulate the rate wit hout dispersion, a tanks in series mixing model was implemented consisting of N ideal continuously stirred tank reactors (CSTR) in series. Step tracer experiments were perfo rmed with H 2 to obtain the residence time distribution (RTD) function ( E ) for the reactor; this was accomplished by comparing the mole fraction of the tracer in the effluent ( y tracer ) with that in the feed ( y 0 ) during the step. With knowledge of the RTD, t he mean residence time 2 ), and number of CSTRs ( N ) needed to approximate the RTD were calculated as shown below [118, 121] ( 2 6 ) ( 2 7 ) ( 2 8 ) ( 2 9 ) The rate without dispersion was obtained by working iteratively backwards through the series of CSTRs using the equation below : ( 2 10 )

PAGE 69

69 where y H 2 i 1 and y H 2 i are the mole fractions of H 2 at the inlet and outlet, respectively, of the i th CSTR reactor volume. Figure 2 7 shows th e difference between the measured (downstream of the reactor) and simulated and d /d t (corrected for dispersion) for an isothermal reduction reaction performed at 1473 K and p H 2 O: p H 2 of 2.24:1. The CSTR parameters used in this calculation are shown in th e figure caption. Here, it can be seen that the maximum of the CSTR corrected rate (red squares) is both greater in magnitude and achieved sooner than that of the measured rate (blue circles). The total area under the respective curves is conserved, as evi denced by the measured (dashed line) and CSTR corrected (dot dashed line) profiles; thus, while the profiles themselves differ (a repercussion of the dispersion free rate), the final achieved is equivalent for both cases. The forward and reverse reaction rates of the postulated two step mechanism were assumed to follow an Ar rhenius dependence, thus each rate constant contains a unique pre exponential factor A and activation energy E a As a result, in addition to the defect formation h T s T ) used to describe bulk to surface transport equilibrium, a total of 10 fitted parameters (thermodynamic and kinetic) were extracted from the minimization routine. Several redox tests, performed at different p O 2 initial and temperatures, were categorized by reaction ( i.e. H 2 oxidation versus H 2 O splitting) and modeled simultaneously to ensure that the fitted parameters are globally descriptive of the wide range of experimental conditions considered. The intrinsic reaction rates (during ceria reduction and oxidation) and corresponding model predictions are displ ayed in Figure 2 8 Good agreement is observed in all comparisons, regardless of the employed experimental conditions. Exemplary distributions of defect species can be seen in Figure A 7 of Appendix A and Table 2 1 summarizes the thermodynamic and kinetic parameters used to generate the aforementioned results. Subscript

PAGE 70

70 notation within Table 1 is as follows: f and r refer to the forward and rev erse of reactions 1 and 2 (equations ( 2 4 ) and ( 2 5 ) respectively). When compared to the prior work, quantitative agreement is shown, as the majority of fitted parameters fa ll within the reported uncertainty. Disc repancies may be ascribed to the post process efforts to remove dispersion and differing experimental systems and procedures. Furthermore, Zhao et al. [119] obtained parameters from results at very high concentrations of 14% H 2 and 26% H 2 O (and, therefore, much different p O 2 ) during ceria reduction and oxidation respectively. In contrast, the p O 2 during reduction and oxidation used in this work were more comparable to each other, as a fixed p H 2 O: p H 2 ratio was delivered to initiate the respective reactions. With knowledge of the fitted model parameters, evaluatin g redox kinetics of ceria under a wider range of conditions becomes feasible. Thus, using the modeled rate of ceria oxidation at 1473 K as a reference (see the bottom right panel of Figure 2 8 ), other rates were examined at conditions difficult or impossib le to replicate in the current experimental setup ( i.e. very high oxidant to fuel ratios). The results are shown in Figure 2 9 which displays the predicted rate of ceria oxidation via H 2 O splitting and corresponding profile at equivalent i for two temperatures ( i.e. 1373 and 1473 K) and two different p O 2 For p O 2 = 5.6010 11 atm ( Figure 2 9 ), in ac cordance with an experimentally observed temperature dependence, the predicted peak rate decreases with decreasing te mperature. Conversely, as temperature increases, the change in predicted at equilibrium decreases, in agreement with prior literature (see Figure 2 5 ) [16, 117] These trends are also observed for a tenfold incr ease in delivered oxidant ( i.e. p O 2 = 5.6010 9 atm), shown in panel (b) of Figure 2 9 Notably, at constant temperature, as p H 2 O: p H 2 (and thus, p O 2 ) increases from panel (a) to panel (b) of Figure 2 9 the predicted peak rate increases with an accompanie d decrease in the predicted at equilibrium (indicating greater fuel yields and re

PAGE 71

71 oxidation of the metal oxide). Although these predictions were made for fixed p H 2 O: p H 2, the benefit of such a model is that it could be implemented in systems with spatially and temporally varying com positions, e.g. in large scale solar reactors, in order to predict and optimize performance without exhaustive experimentation. Conclusion A n experimental framework was described by which the thermodynamic and intrinsic kinetic parameters of candidate ther mochemical materials may be measured. The experimental procedures employed herein enable the study and direct comparison of material performance under controlled and industrially relevant p H 2 O: p H 2 ratios, a parameter which has been shown to have a signific ant impact on the measured fuel yields and efficiency of solar to fuel energy conversion. T he current state of the art redox material, undoped ceria, was used as a platform to demonstrate the ability to produce accurate measurements of at various tempera tures and p O 2 Isothermal relaxation experiments with a constant H 2 background, during which the oxidation state of the sample was shifted via rapid changes in the delivered H 2 O, were performed at temperatures from 1173 1473 K and p O 2 from 4.5410 18 1.02 10 9 atm and measurements obtained via residual gas analysis were found to be in good agreement with accepted data from the literature. The temporal effect s of gas phase dispersion in the reactor system on the mass spectrometer measurements were accounted for via a tanks in series mixing model. This mixing model was coupled with a mechanistic kinetic model to show the ability to extract the intrinsic kinetic behavior

PAGE 72

72 Table 2 1. Comparison of fitted kinetic and thermodynamic parameters with prior work. Fitted Parameters This Work Zhao et al. Red. Ox. A 1,f (s 1 ) 2.9010 2 1.1010 4 1.3010 2 E a1,f (kJ mol 1 ) 21 22 7 7 A 1,b (s 1 ) 6.9010 14 6.1010 14 8.2010 14 E a1,b (kJ mol 1 ) 209 232 210 50 A 2,f (s 1 ) 3.5010 14 1.8010 14 1.5010 14 E a2,f (kJ mol 1 ) 181 176 190 50 A 2,b (s 1 ) 8.1010 4 5.1010 4 4.4010 4 E a2,b (kJ mol 1 ) 64 89 97 5 h T (kJ mol 1 ) 69.5 149.8 107.6 16.8 s T (J mol 1 K 1 ) 20.3 11.7 54 11.9 Figure 2 1. Schematic of the high temperature tubular reactor system. Dry gases (pure Ar and 5% H 2 balanced in Ar) are delivered via three mass flow controllers and deionized liquid H 2 O is delivered using a highly accurate liquid flow controller. The liquid H 2 O is mixed with Ar (which functions as a carrier gas), aerosolized, an d vaporized in the CEM. All gas and vapor flows are directed through the work tube during experimentation. Most of the unreacted H 2 O is removed from the reactor effluent before being sampled by the mass spectrometer, wherein the product species are

PAGE 73

73 measure d and quantified, using a condenser consisting of a vacuum trap and an ice bath. p tot is controlled using a pressure controller coupled with a vacuum pump. Figure 2 2. Exemplary experimental data obtained using an isothermal experimental scheme in whic h the oxidation state of a ceria sample was cycled by shifting the p O 2 via step changes in the input H 2 O flow rate. 6.86 mol s 1 of H 2 and 402.67 mol s 1 of Ar were delivered at all times, and H 2 O flow rates were varied from 7.72 46.30 mol s 1 (a) Reac tor temperature versus time. (b) Input H 2 molar flow rate (left vertical axis) and input H 2 O molar flow rate (right vertical axis) versus time. (c) H 2 molar flow rate in the reactor effluent, as measured by the mass spectrometer, versus time.

PAGE 74

74 Figure 2 3 (left axis, solid line) and p H 2 O/ p H 2 (right axis, dashed line) versus elapsed time during an isothermal relaxation experiment performed at 1373 K with undoped ceria. In accordance with the experimental procedure, the first and last reactions shown were i nitiated via changes in the reactor temperature while keeping p H 2 O/ p H 2 constant. Figure 2 4. Transient in undoped ceria measured via residual gas a nalysis using the isothermal experimental scheme described herein Separate reactions for which the p O 2 are equivalent are distinguished via solid, dashed, and dot dashed lines. The p O 2 in atm,

PAGE 75

75 for each reaction is given in the legend. (a) versus elapsed time for isothermal reduction reactions performed at 1473 K and p O 2 between 2.2210 12 and 4.4910 10 atm. (b) versus elapsed time for isothermal oxidation reactions performed at 1473 K and p O 2 between 3.5710 11 and 1.0210 9 atm. Figure 2 5. Isotherms of the equilibrium oxygen content (2 ) of undoped ceria versus the logarithm of p O 2 Measured equilibrium values (scattered data points) were obtained at T reactor from 1173 1473 K and p O 2 between 4.5410 18 1.0210 9 atm via the isothermal and nonisothermal experimental schemes described herein and are compared to thermogravimetric measurements by Panlener et al. (solid lines) and model predictions by Tuller and Nowick (dashed lines)

PAGE 76

76 Figure 2 6. Reaction rate for reduction and oxidation reactions of undoped ceria performed in a H 2 /H 2 O environment. Separate reactions for which the p O 2 are equiv alent are distinguished via solid, dashed, and dot dashed lines. The p O 2 in atm, for each reaction is given in the legend. (a) d /d t versus for isothermal reduction reactions, proceeding from the left of the figure to the right, performed at 1473 K and p O 2 between 2.2210 12 and 4.4910 10 atm. (b) d /d t versus for isothermal oxidation reactions, proceeding from the right of the figure to the left, performed at 1473 K and p O 2 between 3.5710 11 and 1.0210 9 atm.

PAGE 77

77 Figure 2 7 Measured and CSTR cor rected reaction rates during reduction of undoped ceria performed at 1473 K and p H 2 O: p H 2 of 2.24:1. Based on the tracer experiments, = 31.7 s and 2 = 557.6 s 2 Two CSTRs in series were used to model the RTD of the reactor system. The corresponding measu red and CSTR corrected (dashed and dot dashed lines, respectively) are plotted with respect to the right ordinate.

PAGE 78

78 Figure 2 8 Comparison between the model predictions and CSTR corrected reaction rates during ceria reduction (top subplots) and oxidat ion (bottom subplots) in H 2 /H 2 O. The corresponding CSTR corrected of ceria (dashed lines) refers to the right ordinate.

PAGE 79

79 Figure 2 9 Model predictions of ceria oxidation in a H 2 /H 2 O mixture and corresponding at 1373 K and 1473 K. Subplots (a) and (b ) examine predicted rates at equivalent i and log( p O 2 ) = 10.25 and 8.25, respectively. Black circles represent the CSTR corrected reaction rate at the given conditions, and the predicted (dashed lines) refers to the right ordinate.

PAGE 80

80 CHAPTER 3 OXYGEN N ONSTOICHIOMETRY AND DEFECT EQUILIBRIA OF YTTRIUM MANGANITE PEROVSKITES WITH STRONTIUM A SITE AND ALUMINUM B SITE DOPING Ceria is considered to be the state of the art redox material; it has been used to demonstrate the highest efficiency for solar driven C O production of 5.25% [21] may be re oxidized w ith low oxidant to fuel ratios [85, 122] and exhibits rapid reaction rates [22] and high phas e stability [16, 23] Due to its high partial molar entropy change duri ng oxygen exchange, ceria is uniquely able to drive the reduction of H 2 O/ CO 2 to H 2 /CO with small temperature swings However, its high partial molar enthalpy changes necessitate high temperatures ( e.g. greater than 1773 K) and low gas phase oxygen chemical potential to achieve modest reduction extents ( ) and high solar to fuel efficiencies [20, 73, 75] The thermodynamic properties may be tuned by introducing dopants into the crystal structure; however, many doping strategies did not improve the redox performance beyond that of pure ceria [57, 59 61, 123 125] In contrast, tetravalent dopants ( e.g. Zr 4+ and Hf 4+ ) were shown to increase when subjected to comparable temperat ures and oxygen partial pressures [28, 57, 61, 68, 70, 116, 126] Beyond ceria and doped ceria, perovskite oxides have received significant attention as H 2 O/CO 2 splitting materials due to their lower enthalpies, wh ich enable the reduction reaction to be performed at lower temperatures. LaMnO 3 perovskites with Sr 2+ or Ca 2+ doping on the A site and Al 3+ doping on the B site have been the focus of many of these experimental efforts [44, 47, 85, 88, 92] While this class of materials has been shown to achieve notably greater than ceria under comparable conditions, their lower entropies lead to less favorable oxidation thermodynamics a nd the amount of H 2 /CO produced is highly se nsitive to the oxidant concentration [85]

PAGE 81

81 Besides LaMnO 3 based perovskite oxides, YMnO 3 doped with Sr 2+ (YSM ) namely Y 0.5 Sr 0.5 MnO 3 (YSM50), has also been examined [87, 88, 90, 127, 128] YSM50 release s notably more O 2 during therm al reduction than ceria and LaMnO 3 based perovskites. Dey et al. [88] attributed this increase in O 2 evolution to the smaller ionic radius of Y 3+ compared to that of La 3+ which results in a lower tolerance factor and an increase in distortion of the lattice [129, 130] While experimental examinations of this class of materials agree unanimously in regard to their improved reduction performance, contradicting oxidation results have been reported. Demont and Abanades [87] Nair and Abanades [127] Dey et al. [88] and Dey and Rao [90] computed oxidation yields for YSM50 via thermogravimetric analysis and attributed mass changes in their samples to splitting of CO 2 to produce CO. For reduction in Ar at 1673 K and oxidation by CO 2 at 1323 K, the results of Demont and Abanades and Nair and Abanades indicated poor CO production; in contrast, Dey et al. reported significantly higher oxidation yields for reduction at 1473 1673 K and oxidation at 1173 or 1373 K. In a tubular reactor Dey et al. measured H 2 O s plitting yields via gas chromatography (GC) and observed H 2 production during oxidation at 1373 K. During isothermal operation at 1573 and 1673 K, Dey and Rao calculated CO production approaching complete re oxidation. In the same study, oxidation measurem ents for Sr 2+ doped LaMnO 3 were supplemented by GC and found to be within 10 15% of the ir thermogravimetric results ; however, these measurements were not shown for YSM50 Rodenbough and Chan [128] also examined YSM50 and used an IR electro chemical sensor to measure CO production directly but did not observe CO when cycling between 1173 and 1473 K Quantitative c omparison of the abovementioned experimental works is difficult as the oxidation conditions are either not sufficiently defined or controlled, i.e. CO 2 :CO or H 2 O:H 2 are

PAGE 82

82 not provided or vary as the reaction progresses or the employed conditions are simply different In this chapter, the redox performance of YSM based perovskite oxides was investigated using the experimental framework described in Chapter 2. In regard to thermochemical fuel production only 50 mol% Sr 2+ d oping on the A site has been examined and B site doping in YMnO 3 has not been considered. Herein, Al 3+ doping in this class of materials was considered for the first t ime. YMnO 3 perovskites with 10 or 20 mol% Sr 2+ doping and 40 or 60 mol% Al 3+ doping were synthesized via a modif ied Pechini method. The crystal structures were examined before and after thermochemical treatment via power X ray diffraction (PXRD), elemental compositions were measured via inductively coupled plasma mass spectrometry (ICP MS), and the sample surfaces were observed via scanning electron microscopy (SEM) and energy dispersive X ray spectroscopy (EDS). Equilibrium maps of as a function of tempe rature and oxygen partial pressure ( p O 2 ) were produced for each material. Redox experiments were performed at 973 1173 K and p O 2 from 1.2410 20 2.2610 13 atm in the high temperature tubular reactor described in Chapter 2 wherein the sample environment w as controlled by delivering H 2 and H 2 O. Additionally, thermogravimetric analysis was employed to quantify at 1173 1473 K and p O 2 from 1.6110 4 3.2310 2 atm ; here, p O 2 was controlled via an O 2 /Ar atmosphere. Lastly, a chemical defect model considering the formation of doubly ionized oxygen vacancies coupled with reduction of Mn 4+ to Mn 3+ and/or Mn 2+ as well as the formation of extended defects was produced. Experimental Materials Synthesis and Characterization A modified Pechini meth od [131] was utilized to synthesize perovskite oxide samples of the form Y 1 x Sr x Mn 1 y Al y O 3 (YSMA). Within this composition space, the A and B site dopant concentrations were varied to assess their effects on redox performance. The following

PAGE 83

83 compounds were formed and will be abbreviated herein as follows: Y 0.8 Sr 0.2 Mn 0.6 Al 0.4 O 3 (YSMA8264), Y 0.8 Sr 0.2 Mn 0.4 Al 0.6 O 3 (YSMA8246), and Y 0.9 Sr 0.1 Mn 0.6 Al 0.4 O 3 (YSMA9164). Stoichiometric amounts, according to the desired cation ratios, of the nitrate precursors Y(NO 3 ) 3 6H 2 O (Sigma A ldrich, 99.8% trace metals basis), Sr(NO 3 ) 2 (Sigma Aldrich, ACS 3 ) 2 (Sigma Aldrich, 45 50 wt.% in dilute HNO 3 ), and Al(NO 3 ) 3 9H 2 O (Sigma 6 H 8 O 7 (CA, Sigma Aldrich, ACS reagent, 2 O. In accordance with Scheffe et al. [61] and Cooper et al. [85] the molar ratio of CA to total metal cations was set to 3:2. To fo rm a solution, the mixture was stirred with a magnetic stirrer for 2 hr. Following stirring, C 2 H 6 O 2 (EG, Sigma Aldrich, 99.8%) was added at a ratio of 2 mol EG per mol CA to promote homogeneity [132] ; the solution was stirred for an additional 10 min. While stirring, the solution was gradually heated to 363 K; the rotation speed was progressively increased from 300 to 800 RPM during the temperature ramp The solution was maintained at this temperature for about 1.5 hr until a gel was formed. After gelification was complete, the gel was allowed to cool and was subsequently transferred to two Al 2 O 3 boat crucibles (to prevent sample loss, as the gels expand dramatically upon heating). The gel was dried at 573 K [57] for 3 hr in a Carbolite RHF 16/8 box furnace; during heating, the ramp rate was set to 2 K min 1 The resulting powder was then pulverized with a mortar and pestle upon cooling to for m a fine powder. Lastly, the powder was calcined in air at 1473 K for 24 hr, again with a ramp rate of 2 K min 1 PXRD was performed at room temperature over 2 from 20 80 with a step size of rt Powder Diffractometer with Cu HighScore Plus v. 3.0e software was used to perform background detection and subtraction to

PAGE 84

84 account for the effects at low angles ( e.g. 2 f rom 0 20) of the glass slide on which the samples YSMA based materials that were synthesized. Thus, the PXRD patterns obtained for YSMA8264, YSMA8246, and YSMA9164 were compared to those for Y 1 x Sr x Mn 1 y Co y O 3 (YSMCo) with x from 0.2 0.4 and y from 0 0.4 as reported by Huang and Hua ng [133] Additionally, potential crystal structure changes due to deviation from stoichiometry were assessed via simple reduction experiments in a horizontally oriented thermogravimetric analyzer (Mettle r Toledo HT TGA/DSC 2 ) ; each of the samples were reduced at 1173 K in Ar and H 2 /Ar atmospheres, then were immediately cooled to ambient temperature to prevent re oxidation. The diffraction patterns of the reduced samples were obtained and compared to the as synthesized samples The respective elemental compositions of the synthesized samples ( i.e. the concentrations of Y, Sr, Mn, and Al ) w ere evaluated via ICP MS on an Element 2 ICP MS (Thermo Scientific). Prior to ICP MS, the samples were dissolved in aqua regia (HNO 3 and HCl) SEM and EDS were employed using an FEI Nova NanoSEM 430 with 15 kV accelerating voltage to assess the surfaces of the samples and the homogeneity of the dopant elements. Thermogravimetric Analysis Initial assessment s of the oxygen exchange capacities o f the YSMA based perovskites were performed via temperature programmed reduction (TPR) experi ments in the th ermogravimetric analyzer Ceria (Alfa Aesar REacton 99.9%, 5 m powder ) was also tested using the sam e procedure to serve as a benchmark for compariso n. Sample loading was approximately 14 mg in all cases, and the samples were prepared in a single layer in a platinum dish crucible (Alfa Aesar, 46686). During the TPR experiments, the samples were initially heated to 573 K at 20 K min 1 1 Ar (Airgas, Ultra High Purity 5.0 Grade Argon). Note, gas flow rates were standardized to 298 K and 1 atm. Reduction was initiated by

PAGE 85

85 heating from 573 to 1173 K at 10 K min 1 1 1 H 2 (Airgas, Argon 95 Hydrogen 5) ( p H 2 0.01 atm). Upon reaching 1173 K, a short (1 min) purge 1 1 O 2 1 Ar for 5 min ( p O 2 0.15 atm). Th e O 2 delivery was maintained during cooling back to 573 K; the procedure was then repeated for a total of four cycles. Each O 2 min 1 Ar to remove any residual O 2 from the balance chamber of the TGA. Two e xperiments (thus, eight total cycles) were performed per material to assess repeatability. Immediately following each experiment using a reactive sample mass (henceforth referred to as the sample run), the procedure was repeated with an approximately equiv alent mass of Al 2 O 3 powder (Sigma Aldrich, 98% Al 2 O 3 basis ); this was done to correct for buoyancy effects and will be referred to herein as the blank run. Additional experiments were performed with the YSMA based perovskites to quantify the redox behavior at higher temperatures and p O 2 from 11 73 1473 K and 1.6110 4 3.2310 2 atm, respectively. The samples were initially heated to 1173 K at a rate of 20 K min 1 in Ar. Based on the results of the TPR experiments, a p O 2 of approximately 0.15 atm was used to ensure complete re oxidation of the per ovskites at this temperature prior to and after each experiment. The sample environment was controlled by delivering an O 2 /Ar mixture; p O 2 from 1.6110 4 3.2110 3 atm were achieved by mixing 0.5% O 2 (Airgas, 0.5% Oxygen Balance Argon) and Ultra High Purit y Ar, and the highest p O 2 of 3.2310 2 and 0.15 atm were obtained by mixing Ultra High Purity O 2 and Ar. After the initial oxidation step at 1173 K, the p O 2 was rapidly decreased to isothermally reduce the samples. The samples were held at 1173 K under the se conditions for an extended amount of time (3 hr for YSMA8264 and 2 hr for YSMA8246

PAGE 86

86 and YSMA9164) to allow the initial reduction reaction to reach equilibrium. Next, the samples were reduced further by heating to 1273 K at a rate of 5 K min 1 This tempe rature was used as a reference point for all subsequent reactions; after reduction at a higher temperature, the sample was cooled back to 1273 K before the next measurement. For all isothermal segments hereafter, the dwell times were either 20 min for YSMA 8264 or 45 min for both YSMA8246 and YSMA9164. Reduction extents were then measured at 1373 and 1473 K. Following these measurements, the samples were re oxidized at 1173 K under the same conditions as used at the beginning of the experiment. Due to the sm aller reduction extents exhibited by YSMA8246 and YSMA9164 under the tested conditions compared to those of YSMA8264, sample loading was higher for these materials (100 and 250 mg, respectively) than for the latter (25 mg) to promote larger total changes i n mass. All samples were arranged as loosely Al 2 O 3 crucibles (Mettler Toledo, 00024124 or 30077260). As done for the TPR experiments, a blank run was performed following each sample run. The transient sample averaged of the r espective perovskites was calculated via the equation below. ( 3 1 ) Here m s is the buoyancy corrected sample mass and M YSMA and M O are the molar masses of the YSMA sample and monoatomic oxygen, respectively. Isothermal Relaxation Experiments in H 2 /H 2 O An isothermal experimental scheme was employed within a tubular reactor to obtain thermodynamic insight into the performance of the YSMA based perovskites at low temperature s and p O 2 typifying the oxidation step of a temperature swing redox cycle During these experiments, the p O 2 was controlled by delivering a precise mixture of H 2 and H 2 O. The

PAGE 87

87 reactor system and e xperimental procedures were d escribed thoroughly in Chapter 2. A schematic of the experimental system is shown in Figure 2 1. The transient and equilibrium redox behavior of the YSMA based materials was evaluated isothe rmally in a controlled H 2 /H 2 O sample environment at 973 1173 K and p O 2 f rom 1.2410 20 2.2610 13 atm The experiments were performed with a constant H 2 background, and changes in the oxygen content of the samples were evaluated via mass spectrometry by an alyzing deviations from this background signal initiated by the redox reactions. At the temperatures of interest, reduction and oxidation reactions were initiated via rapid changes in the delivered H 2 O (and, thus, the p O 2 ). The p O 2 was assumed to be govern ed by the H 2 O thermolysis reaction (H 2 O H 2 +0.5O 2 ) an d was calculated as shown in equation ( 2 1 ) H 2 O delivery during the isothermal relaxation reactions was v aried from 7.72 123.46 1 and the H 2 O:H 2 ratio ranged from 1.51 72.54; low H 2 O:H 2 ratios ( i.e. less than 100) were employed to emulate high conversion conditions. Note, the maximum H 2 O:H 2 ratio employed herein is lower than the minimum values emplo yed by Zhai et al. [106] and Barcellos et al. [134] where the ratios ranged from 104 2.110 4 and 285 1333, respectively. Regarding the experimental procedure, it is also important to note that t he analysis is simplified by the use of a constant H 2 background throughout the measurement segments. In cases where the H 2 yields were sufficiently low, Barcellos et al. observed that the background H 2 signal upon simultaneous introduction of H 2 and H 2 O w as larger than that associated with the production of H 2 ; thus, their analysis of the fuel production in these cases was reliant upon measurement of the O 2 evolution in the subsequent reduction step. Decreasing the H 2 O delivery (thus, decreasing the p O 2 ), drives the reduction of the sample and consumption of H 2 to produce H 2 O in accordance with the reaction shown below.

PAGE 88

88 ( 3 2 ) is defined as the difference between the reduction extents following and prior to the reaction. Contrarily, increasing the H 2 O delivery leads to an increase in the p O 2 and the initiates the reverse of the reaction shown in equation ( 3 2 ) The reaction induced changes in the sample averaged were evaluated via numerical integration of the transient H 2 consumption or production using the equation below : ( 3 3 ) where N YSMA is the total moles of YSMA sample being tested, t start and t end are the respective starting and ending times for the reaction of interest, F H 2 is the measured H 2 molar flow rate, and N H 2 O and N H 2 are the molar changes in H 2 O and H 2 respectively, due to the reaction. As the redox performance of the YSMA based perovskites was not well understood prior to experimentation, slight modifications were made to the experimental procedure (fr om that presented in Chapter 2) to ensure a constant initial Rather than cooling in a H 2 /H 2 O environment, as done previously for the experiments performed with ceria in the same reactor system, air (Airgas, Ultra Zero Grade Air) was used to ensure compl ete re oxidation (as confirmed during investigations in the TGA). H 2 and H 2 O delivery was ceased after completion of the isothermal p O 2 shift reactions and air was delivered for 1 hr at the temperature of the experiment and during cooling to room temperatu re. Results and Discussion Materials Characterization The PXRD patterns for YSMA8264, YSMA8246, and YSMA9164 after synthesis are shown in Figure B 1 of Appendix B in comparison to PXRD patter ns extracted from Huang and

PAGE 89

89 Huang [133] for Y 0.6 Sr 0.4 Mn 0.6 Co 0.4 O 3 (YSMCo 6464 ) In general, the diffraction patterns for the YSMA materials are comparable to that of YSMCo6464. For YSMA9164, however, three low intensity peaks ca n be seen at approxi mately 29, 30, and 31.67 ; these peaks correspond to the YMnO 3 hexagonal phase [87] and are not present in YSMA8264 and YSMA8246 after synthesis These observations are in agreement with those o f Huang and Huang who s howed that crystallization to the orthorhombic perovskite phase wa s promoted by the presence of Mn 4+ (the amount of which increases with Sr 2+ doping). Figures 3 1, 3 2, and 3 3 compare the as synthesized PXRD patterns for the YSMA materials to those after reduction in either Ar (moderate ) or H 2 /Ar (high ). for Ar and H 2 reduction were approximately 0.0521 and 0.3925, respectively, for YSMA8264; 0.0429 and 0.3319 for YSMA8246; and 0.0287 and 0.3388 for YSMA9164. For YSMA8264, changes in the PXRD patterns were observed after both Ar a nd H 2 reduction. After reduction in Ar at 1173 K, new peaks e xist at 28.87, 29.94, 31.19, 51.4, and 56.74 ; each of these peaks are present in Y 0.8 Sr 0.2 MnO 3 (PXRD patterns for which are shown in Figure B 2 of Appendix B), which crystallizes to a hexagonal non perovskite structure [133, 135] After reduction in H 2 the low intensity peaks at 28.87, 29.94, and 31.19 are no longer visible and a relatively high intensity peak at 28.99 can be seen. Additionally, notable separation of the main Bragg peak was observed for the H 2 reduced YSMA8264 sample. For YSMA9164, there were no notable changes in the PXRD pattern after reduction in an Ar atmosphere ; however, similar to YSMA8264, a new peak at approximately 29.01 was observed after reduc tion in H 2 In contrast to YSMA8264 and YSMA9164, the PXRD patterns for YSMA8246 were qualitatively unchanged even after reduction to high in an H 2 atmosphere. A very low intensity peak, almost indistinguishable from the signal noise, can be seen at abou t 29

PAGE 90

90 ICP MS analysis of the elemental compositions of the YSMA based perovskites showed that the actual dopant concentrations were all within 1 mol% of their respective targe t values; thus, the previously defined notations for the nominal compositions a re maintained The measured concentrations of Y, Sr, Mn, and Al in comparison to the targeted concentrations for YSMA8264, YSMA8246, and YSMA9164 are shown in Table 3 1. SEM images were taken after synthesis (prior to redox analysis). These images for all ma terials are shown in Figures B 3 through B 5 in Appendix B. EDS maps of the sample surfaces of each material were prod uced and are shown in Figure B 6 In general, these images show near homogeneous distribution of the respective cations. Oxygen Nonstoic hiometry Measurements TPR experiments were performed in H 2 /Ar to assess the oxygen exchange capacities of the YSMA perovskites. Figure 3 4 shows the mass profiles for the YSMA perovskites and ceria during heating from 573 to 1173 K for the final cycle of t he two experimental runs performed per material. Compared to ceria, the onset of reduction for the YSMA perovskites occurs at significantly lower temperatures and reduction extents are greater. YSMA8264 achieved the deepest reduction with a perc ent mass de crease of 2.55 % upon reaching 1173 K, comp ared to 2.357 %, 1.868 % and 1.257 % fo r YSMA9164 YSMA8246 and ceria, respectively. Repeatability between separate runs was achieved, as can be seen by the similarity in the mass profiles and extents of reduction b etween run 1 (solid lines) and run 2 (dashed lines). of the YSMA materials at 1173 1473 K and p O 2 from 1.6110 4 3.2310 2 atm were measured via thermogravi m e tric analysis. Mass and chang es due to the redox reactions were calculated relative to the st eady state mass of the sample at 1173 K and p O 2 = 0.15 atm, the 0 during the TPR experiments. Figure 3 5 shows exemplary TGA results for YSMA8264, YSMA8246, and

PAGE 91

91 YSMA 9164 during exp eriment s performed at 1173 1473 K with a constant p O 2 of 1.6110 4 atm during temperature swings As noted previously, the sample loading was greater for YSMA8246 and YSMA9164 than for YSMA8264 to improve analysis of the mass changes. T his change necessita ted adjustments to the isothermal dwell times to allow the sample masses to reach equilibrium T o promote visual comparison and enable plotting on the same time scale, some of the isothermal segments were artificially extended after a steady state mass was achieved. After stabilization of the masses at the temperature and p O 2 for which 0, the p O 2 was decreased by switching the method gas from Ultra High Purity O 2 to 0.5% O 2 balanced in Ar. This shift in p O 2 at constant temperature occurred at about 163. 8 min and resu lted in a decrease in mass and a concomitant increase in Figure 3 5 clearly shows that YSMA8264 reduces significantly more than YSMA8246 and YSMA9164 under these conditions ; t he mass decreases for YSMA 8264, YSMA8246, and YSMA9164 were 0.21 48 0.0346, and 0.01096 %, respectively, which correspond to of 0.02423 0.00378 and 0.001237 Once equilibrium was established after the p O 2 swing reaction additional were measured by changing the temperature while keeping p O 2 constant; increasing th e temperature resulted in reduction of the sample, while oxidation was initiated via a decrease in temperature As expected, the largest mass change during this experiment was observed at 1473 K where the mass changes for the three materials were 0.5065, 0 .1727, and 0.04254%, respectively, and the corresponding were 0.05711, 0.01887 and 0.0048 These experiments were repeated with different p O 2 to map the dependence of on temperature and p O 2 To measure under controlled conditions representative of oxidation with high conversion of H 2 O to H 2 ( e.g. H 2 O:H 2 100:1), experiments were performed in a high temperature tubular reactor in a H 2 /H 2 O atmosphere at temperatures from at 973 1173 K and p O 2

PAGE 92

92 from 1.2410 20 2.2610 13 atm Figure 3 6 shows representative experimental data for an experiment performed at 10 73 K and p O 2 from 3.4510 17 2.2110 15 atm with YSMA8246 ; the top, middle, and bottom subplots show the reactor temperature, p O 2 that the sample wa s exposed to, and the molar H 2 flow rate at the outlet of the reactor (measured via mass spectrometry), resp ectively. The sample was heated from 473 to 1073 K while maintaining constant H 2 and H 2 O delivery ( p H 2 = 0.00481 atm and p H 2 O = 0.03633 atm ) ; the increase in temperature lead to reduction of the sample, indicated by the deviation in the H 2 flow rate at the outlet of the reactor from the baseline flow rate (established by the delivered H 2 ). At 1073 K, the equilibrium p O 2 was approximately 3.4510 17 atm and was 0.09048. The remaining reactions were performed isothermally via rapid shifts in the delivered H 2 O concentration. For example, the subsequent reaction was performed by increasing the H 2 O flow rate ( p H 2 O = 0.1846 atm); the rise in p H 2 O increased the p O 2 to 1.2410 15 atm resulting in H 2 production and re oxidation of the sample to = 0.07636. Figure 3 7 shows equilibrium for YSMA8264, YSMA8246, and YSMA9164 obtained during the experiments performed in the TGA and the tubular reactor. An inset is provided within each of the respective figures to highlight the values measured at higher p O 2 Defect Modeling A defect model was developed to describe the chemical equilibria of the YSMA perovskites. As relatively low p O 2 were employed during the experimental investigation in the tubular reactor, the model considers both point defects and defect clusters. In YMnO 3 the Mn cation is trivalent; however, upon substitution of a portion of the trivalent Y with divalent Sr, the valency of an equivalent portion of Mn changes from trivalent to tetravalent to maintain charge neutrality This charge compensation is analogous to that which occurs upon replacement of La 3+ with Sr 2+ in La 1 x Sr x MnO 3 [44, 47, 95] In Krger Vink notation [136] the chemical formula fo r

PAGE 93

93 Sr doped YMnO 3 (Y 1 x Sr x MnO 3 ) is where , , a nd represent Y 3+ Sr 2+ Mn 3+ Mn 4+ and O 2 respectively. It is assumed that the ad dition of trivalent Al on the B site will not impact the molar amount of Mn 4+ ; in this case, the presence of Al 3+ affects only the molar amount of Mn 3+ (in accordance with the crystal site balance for Mn) The chemical formula for Sr and Al doped YMnO 3 ( Y 1 x Sr x Mn 1 y Al y O 3 ) becomes where represents Al 3+ T he relationships which define electroneutrality and the site balances for Mn and O for stoichiometric Y 1 x Sr x Mn 1 y Al y O 3 are shown below, where the brac keted terms denote the respective species concentrations in reference to the number of moles of perovskite ( 3 4 ) ( 3 5 ) ( 3 6 ) During reduction of the metal oxide, oxygen vacancies (point defects) are created; this model assumes that these oxygen vacancies are doubly ionized, meaning two electrons are left behind for each oxygen vacancy created [137] This defect reaction is shown below. ( 3 7 ) and e represent doubly ionized oxygen vacancies and ele ctrons, respectively. It has been proposed that these electrons will interact with Mn 4+ as shown below. ( 3 8 )

PAGE 94

94 Combining the reactions shown in equations ( 3 7 ) and ( 3 8 ) gives the equation for the first defect mechanism considered by the model; namely, the creation of doubly ionized oxygen vacancies accompanied by reduction of Mn 4+ to Mn 3+ [137] This reaction and the equation for the corresponding equili brium constant ( K r ) are shown below. ( 3 9 ) ( 3 10 ) Considerin g this mechanism alone, is equivalent to and the maximum is equal to half of the divalent dopant concentration; at this point, all of the Mn 4+ has been consumed by the defect formation reaction and, thus, the reaction can no longer proceed. As some of the measured at low p O 2 were greater than this maximum value this mechanism alone cannot describe the equilibrium of the YSMA materials in this pressure range. Disproportionation of Mn 3+ into Mn 4+ and Mn 2+ was also considered [44, 95] This reaction is defined below, where represents Mn 2+ ( 3 11 ) This reaction mechanism may be re written as shown below by combining equations ( 3 9 ) and ( 3 11 ) ( 3 12 ) T his reaction represents the formation of doubly ionized oxygen vacancies coupled with double reduction of Mn 4+ to Mn 2+ As the c oncentration of oxygen vacancies increases, it is possible for these point defects to interact and for m extended defects called defect clusters [47, 137] It wa s assumed that all of the Mn 2+ produced by the abovementioned d isproportionation reaction

PAGE 95

95 formed associations with nearby oxygen vacancies [47] The defect cluster formation reaction is shown below. ( 3 13 ) Combining equations ( 3 9 ) ( 3 11 ) and ( 3 13 ) gives the overall reaction f or the formation of defect clusters This reaction and the equilibrium constant ( K c ) are shown in equations ( 3 14 ) and ( 3 15 ) below. ( 3 14 ) ( 3 15 ) T he reactions defined in equations ( 3 9 ) ( 3 12 ) and ( 3 14 ) were used to model the equilibrium of the YSMA perovskites. Considering these reactions, the charge neutrality and site conservation equations (shown previously for the stoichiometric metal oxide) may be updated based on the imposed changes in the defect concentrations. These equations, a long with an updated definition of that accounts for the presence of defect clusters, are shown below. ( 3 16 ) ( 3 17 ) ( 3 18 ) ( 3 19 ) and are known for each of the YSMA materials based on the ICP MS measurements; thus, they are def ined according to the general formula for YSMA ( i.e. = x

PAGE 96

96 and = y). Due to the assumption that all Mn 2+ forms clusters with nearby oxygen vacancies, may be eliminated from equa tions ( 3 16 ) and ( 3 17 ) , and may be written as functions of and alone; additionally, equations ( 3 10 ) and ( 3 15 ) may be also be re wri tten in terms of these two values ( 3 20 ) ( 3 21 ) Fits to the experimental measurements were obtained using fmincon, an optimization algorithm inherent to MATLAB. This algorithm was also used within each iteration of the minimization routine to simultaneously solve equations ( 3 20 ) and ( 3 21 ) for and The solution routine was subjected to the linear inequali ty constraints shown below to ensure that only a single solution was obtained. ( 3 22 ) ( 3 23 ) Separately, fits were obtained via simultaneous consideration of the reactions defined in equations ( 3 12 ) and ( 3 13 ) ; in this case, Mn 4+ is reduced to Mn 2+ by the two electrons left behind upon production of an oxygen vacancy and it is not assumed that all of the Mn 2+ created forms defect clusters ( i.e. 0). Considering the relevant charge neutrality and site balance equations, the equations for the equilibrium constants of the double reduction and defect cluster reactions were written in terms of and as shown below.

PAGE 97

97 ( 3 24 ) ( 3 25 ) In a similar manner as done with the previous model, the linear inequality constraints shown below were applied to the solution routine ( 3 26 ) ( 3 27 ) Figures 3 8, 3 9, and 3 10 show the fits to the equilibrium measurements for YSMA8264, YSMA8246, and YSMA9164, respectively, for both the Mn 4+ /Mn 3+ (equations ( 3 20 ) and ( 3 21 ) ) and Mn 4+ /Mn 2+ (equations ( 3 24 ) and ( 3 25 ) ) defect models In general, neither the Mn 4+ /Mn 3+ nor the Mn 4+ /Mn 2+ model fit the data for YSMA8264 or YSMA9164 well in the low p O 2 regions shown in subplot (a) of Figures 3 8 and 3 10. For YSMA9164, the fits produced with the Mn 4+ /Mn 3+ model are notably poor; the measured data exhibits low sensitivity to changes in p O 2 while the mod el predicts a significantly greater slope in the equilibrium isotherm. The fits from the Mn 4+ /Mn 2+ model for YSMA9164 at 1073 and 1173 K and YSMA8264 at 1173 K have relatively low error. However, this is likely due to the fact that the measured values ar e near the maximum value that can be predicted by this model ( i.e. max = x ). The lack of agreement between the proposed defect models and the measured data is likely a repercussion of the changes in crystal structure observed in both YSMA8264 and YSMA91 64 via the PXRD scans after reduction (discussed above) For YSMA8246, fits obtained via the Mn 4+ /Mn 2+ model in the low p O 2 region are poor, as the slope s of the predicted isotherms at 973 1173 K are notably greater than those of the measured data. However in contrast to YSMA8264

PAGE 98

98 and YSMA9164, the Mn 4+ /Mn 3+ model predictions for YSMA8246 agree with the measured well (as represented by the solid lines in subplot (a) of Figure 3 9 ). It is important to note that the PXRD patterns of YSMA8246 were largely unchanged even when reduced to high Subplot (b) of Figures 3 8 through 3 10 shows the defect model fits in t he high p O 2 region at 1173 1473 K for each material. For YSMA8264, model predictions via both the Mn 4+ /Mn 3+ and Mn 4+ /Mn 2+ models agreed with the measured data; however, the sum of squared residuals was lower for the former. For YSMA8246 and YSMA9164 where the measured were notably less than those observed for YSMA8264 in the high p O 2 region, lower error fits were obtained via the Mn 4+ /Mn 2+ mo del. While both models underpredict at the highest p O 2 this fact is more apparent with the Mn 4+ /Mn 3+ model. This likely indic ates that the assumption that all Mn 2+ forms defect clusters with nearby oxygen vacancies, as done in the case of the Mn 4+ /Mn 3+ model, is not accurate at very small or high p O 2 Conclusion YMnO 3 perovskites with Sr 2+ A site doping and Al 3+ B site doping namely Y 0.8 Sr 0.2 Mn 0.6 Al 0.4 O 3 Y 0.8 Sr 0.2 Mn 0.4 Al 0.6 O 3 and Y 0.9 Sr 0.1 Mn 0.6 Al 0.4 O 3 were synthesized and examined for thermochemical H 2 production for the first time. ICP MS results revealed that the intended compositions were achieved within 1 mol% of the re spective targets. Examination of the PXRD patterns after sy nthesis and after thermochemical treatment in either Ar or H 2 /Ar revealed phase instability for YSMA8264 and YSMA9164 while only minimal changes were observed for YSMA8246 TPR experiments reveale d a notable increa se in reduction extent for the YSMA based materials compared to ceria. Thermogravimetric analysis was employed to measure the oxygen content of the materials at 1173 1473 K and p O 2 from 1.6110 4 3.2310 2 atm controlled via delivery of O 2 /Ar. The novel experimental framework described and validated in Chapter 2 was successfully implemented with a new class of materials with

PAGE 99

99 previously unknown thermodynamic properties. of the materials were examined at 973 1173 K at p O 2 from 1.2410 20 2.2610 13 atm ( H 2 O:H 2 from 1.51 72.54 ) to emulate high conversion conditions. In contrast to previous studies examining this class of materials without Al 3+ doping, the oxidation perfor man ce was examined while controlling the o xygen activity during the reaction ; additionally, t he use of a constant H 2 baseline throughout the experimental procedure enabled quantification of s mall changes in oxidation state without having to infer th e react ion extent via analysis of subsequent reactions D e fect models were produced to describe the oxygen deficiency in the materials versus p O 2 and to obtain insight into the reaction mechanisms Two models were formulated which each considered (1) creation of oxygen vacancies accompanied by a change in the Mn 4+ concentration and (2) formation of extended defects (defect clusters) between Mn 2+ and nearby oxygen vacancies. In the first model, the creation of oxygen vacancies lead to the reduction of Mn 4+ to Mn 3+ and it was assumed that all of the Mn 2+ produced by disproportionation was consumed by the cluster formation reaction In contrast, the second model explicitly considered double reduction of Mn 4+ to Mn 2+ during oxygen vacancy formation and no assumptions were made regarding the Mn 2+ concentration. For YSMA8264 and YSMA9164 neither of the proposed models could accurately describe the oxygen content at 973 1173 K in the low p O 2 region likely due to the observed phase impurities. In contrast, the Mn 4+ /Mn 3+ model agreed well with the measured data for YSMA8246 at all temperatures in this region. In the higher temperature and p O 2 r egion the data for YSMA8264 was best describe d by the Mn 4+ /Mn 3+ model For YSMA8246 and YSMA9164 better fits to the data in this region were obtained by assuming that formation of oxygen vacancies was accompanied by double reduction of Mn 4+ to Mn 2+

PAGE 100

100 The work her ein suggests that these materials are not viable for efficient solar thermochemical fuel production due to the presence of phase impurities ( e.g. the hexagonal YMnO 3 phase) and the lack of sensitivity in their oxygen content to changes in p O 2 particularly for low H 2 O:H 2 ( i.e. high conversion). For example, at 1173 K the equilibrium p O 2 established during oxidation where the ratio of H 2 O to H 2 at equilibrium is 100:1 is approximately 6.310 13 atm. Considering the Mn 4+ /Mn 3+ model fit produced for YSMA8246 under this condition is 0.0728 ; in contrast, of La 0.8 Sr 0.2 MnO 3 (LSM20) would be about 0.0114 [138] The reduction extent s at high temperature and p O 2 must exceed these value s for fuel production to be possible. The flat isotherms for the YSMA based perovskites under the desirable, high conversion oxidation conditions indicate that a large excess of oxidant ( i.e. H 2 O:H 2 ) is required to achieve high oxidation extents and fuel yields.

PAGE 101

101 Table 3 1. ICP MS analysis of YSMA8264, YSMA8246, and YSMA9164 shown in mol%. Material Y Sr Mn Al Targeted Measured Targeted Measured Targeted Measured Targeted Measured YSMA8264 80 79.82 20 20.18 60 59.91 40 40.09 YSMA8246 80 79.55 20 20.45 40 40.55 60 59.45 YSMA9164 90 89.76 10 10.24 60 60.56 40 39.44 Figure 3 1. P XRD patterns of YSMA8264 after synthesis, reduction in an Ar atmosphere at 1173 K, and reduction in a H 2 /Ar atmosphere at 1173 K

PAGE 102

102 Figure 3 2. PXRD patterns of Y SMA8246 after synthesis, reduction in an Ar atmosphere at 1173 K, and reduction in a H 2 /Ar atmosphere at 1173 K. Figure 3 3. PXRD patterns of YSM A9164 after synthesis, reduction in an Ar atmosphere at 1173 K, and reduction in a H 2 /Ar atmosphere at 117 3 K.

PAGE 103

103 Figure 3 4 P ercent mass change and temperature versus time for temperature programmed reduction experiments in H 2 /Ar ( p H 2 0.01 atm) performed with YSMA8264, YSMA8246, YSMA9164, and ceria. The solid and dashed lines show two separate experiments performed to assess the repeatability of the reduction curves.

PAGE 104

104 Figure 3 5 Exemplary TGA results showing percent mass change and temperature versus time for nonisothermal experiments in O 2 /Ar ( p O 2 = 1.6110 4 atm ) at 1173 1473 K performed with YSMA8264, YSMA8246, and YSMA9164. An inset was included in the upper left of the figure to emphasize the mass change for YSMA9164 during the first reduction at 1473 K. As the total masses used during the experiments were different for each of the materials ( to improve analysis of the mass changes ) the temperature program for YSMA8264 was different than that used for the experiments with YS MA8246 and YSMA9164 Th us, some of the isothermal segments were artificially extended after equilibrium was achieved to allow the experiments to be shown on a single figure.

PAGE 105

105 Figure 3 6 Exemplary experimental data for isothermal relaxation experiments in H 2 /H 2 O with YSMA8246. The top, middle, and bottom subplots show the temperature, p O 2 and H 2 molar flow rate in the reactor effluent, respectively.

PAGE 106

106 Figure 3 7 Isotherms of the equilibrium oxygen content (3 ) of YSMA8264, YSMA8246, and YSMA9164 versus the logarithm of the p O 2 Data at temperatures from 973 1173 K and p O 2 from 1.2410 20 2.2610 13 atm were measured during the isothermal H 2 /H 2 O experiments in the tubular reactor. Insets are included in the low er right of the respective subplots to emphasize the 3 values at temperatures from 1173 1473 K and p O 2 from 1.6110 4 3.2310 2 atm obtained during the nonisothermal O 2 /Ar experiments in the TGA.

PAGE 107

107 Figure 3 8 Defect model fits for YSMA8264 at (a) 973 1173 K in the low p O 2 region and (b) 1173 1473 K in the high p O 2 region Markers represent experimentally measured data. Solid lines represent model fits considering (1) creation of doubly ionized oxygen vacancies coupled with reduction of Mn 4+ to Mn 3+ and (2) formation of defect clusters with the assumption that all Mn 2+ produced by disproportionation form s associations with nearby oxygen vacancies. Dashed lines show the model fits produced when accounting for (1) formation of doubly ionized oxygen vacanci es accompanied by double reduction of Mn 4+ to Mn 2+ and (2) formation of defect clusters with no assumption made regarding the Mn 2+ concentration

PAGE 108

108 Figure 3 9 Defect model fits for YSMA82 4 6 at (a) 973 1173 K in the low p O 2 region and (b) 1173 1473 K in the high p O 2 region. Markers represent experimentally measured data. Solid lines represent model fits considering (1) creation of doubly ionized oxygen vacancies coupled with reduction of Mn 4+ to Mn 3+ and (2) formation of defect clusters with the assumpti on that all Mn 2+ produced by disproportionation forms associations with nearby oxygen vacancies. Dashed lines show the model fits produced when accounting for (1) formation of doubly ionized oxygen vacancies accompanied by double reduction of Mn 4+ to Mn 2+ and (2) formation of defect clusters with no assumption made regarding the Mn 2+ concentration. The black solid and dashed lines represent the model fits when fitting the high and low p O 2 regions at 1173 K simultaneously.

PAGE 109

109 Figure 3 10 Defect model fits for YSMA91 64 at (a) 973 1173 K in the low p O 2 region and (b) 1173 1473 K in the high p O 2 region. Markers represent experimentally measured data. Solid lines represent model fits considering (1) creation of doubly ionized oxygen vacancies coupled with redu ction of Mn 4+ to Mn 3+ and (2) formation of defect clusters with the assumption that all Mn 2+ produced by disproportionation forms associations with nearby oxygen vacancies. Dashed lines show the model fits produced when accounting for (1) formation of doub ly ionized oxygen vacancies accompanied by double reduction of Mn 4+ to Mn 2+ and (2) formation of defect clusters with no assumption made regarding the Mn 2+ concentration.

PAGE 110

110 CHAPTER 4 THEORETICAL INVESTIGATION OF ISOTHERMAL AND NEAR ISOTHERMAL REDOX CYCL ING OF NONSTOICHIOMETRIC OXIDES Perovskite oxides have garnered significant attention as thermochemical H 2 O and CO 2 splitting materials as a result of their enhanced oxygen exchange capacities when compared to pure and doped ceria. Many research efforts ha ve focused on assessing the viability of LaMnO 3 perovskites with various doping schemes, e.g. Sr 2+ or Ca 2+ on the A site and Al 3+ on the B site, for use in temperature swing redox cycles [43 45, 47, 85, 86, 88, 89, 92] La 1 x Sr x MnO 3 (LSM) and La 1 x Ca x MnO 3 (LCM) have been shown to achieve deeper reduction extents ( ) than ceria and produce more fuel when subjected to excess H 2 O/CO 2 [44, 85] The addition of Al 3+ on the B site, i.e. La 1 x Sr x Mn 1 y Al y O 3 (LSMA) and La 1 x Ca x Mn 1 y Al y O 3 (LCMA), increased further [43, 44, 85, 92, 97] The partial molar enth alpy and entropy changes of doped LaMnO 3 perovskites during oxygen exchange are lower than those of ceria. The onset of the reduction reaction for the perovskites at lower temperatures is a repercussion of the lower enthalpy values; this shifts the partial molar Gibbs free energy change downward, lowering the temperature at which the reaction becomes exergonic. The entropy affects the sensitivity of the Gibbs free energy to changes in temperature. Thus, temperature swing redox cycles (TSRC) are more effecti vely driven by materials which exhibit large changes in entropy, like ceria, as they enable operation with a smaller temperature swing. As a result, temperature swing cycles are usually predicted to be less efficient with perovskites when compared to ceria In contrast, isothermal cycles have attracted a large amount of attention in recent years. H 2 and CO production via isothermal redox cycling (ITRC) has been demonstrated experimentally for ceria and hercynite based processes [13, 71, 74, 139] Hao et al. [74] demonstrated isothermal H 2 O splitting with ceria at 1773 K; while the total fuel production was

PAGE 111

111 less than when using a temperature swing prior to oxid ation (as would be expected), time averaged fuel production rates were comparable. The first demonstration of isothermal redox cycling in a solar reactor was performed by Hathaway et al. [71] where continuous CO production was demonstrated at 1750 K. During operation, 9 5 and 93% of the sensible heat of the N 2 (used as an inert sweeping gas) and CO 2 respectively, were recovered; despite the high heat recuperation, the measured efficiency was only 0.72% when considering the energy required to produce N 2 Additionally, Tou et al. [139] have demonstrated isothermal single step CO 2 splitting using a novel ceria membrane reactor. Using the hercynite cycle, in which CoFe 2 O 4 reacts with Al 2 O 3 during reduction and the resulting aluminates are reacted with H 2 O to reform the original reactants and produce H 2 Muhich et al. [13] observed a twelve fold improvement in H 2 yields for isothermal operation at 1623 K in comparison to non isotherm al operation of ceria with reduction at 1623 K and oxidation at 1273 K. Little theoretical or experimental work has been focused on utilizing perovskites isothermally, even though isothermal cycles should be independent of partial molar entropy changes. D ey and Rao [90] have experimentally demonstrated improved CO production compared to the current state of the art ceria during isothermal CO 2 splitting using La 1 x Sr x MnO 3 with x from 0.3 to 0.5 (LSM30, LSM40, and LSM50, respectively) and Y 0.5 Sr 0.5 MnO 3 (YSM50). During their experimental assessments, reduction was performed at an oxygen partial pressure ( p O 2 ) of 10 5 atm and the pressure swing for isothermal oxidation was initiated by deliv ering 100% CO 2 In a recent work, Carrillo et al. [140] evaluated the effects of Cr doping in LSM based perovskites during isothermal and near isothermal CO 2 splitting; while decreased with increasing C r substitution, the reaction rates were shown to increase.

PAGE 112

112 La 0.6 Sr 0.4 Mn 0.15 Cr 0.85 O 3 was shown to exhibit greater CO production rates and yields than ceria during isothermal operation at 1673 K. Most efficiency models consider ceria or doped ceria in their analyses [58, 72, 73, 75, 77, 78, 141, 142] In general, when the effectiveness of gas phase heat recovery is less than unity, isothermal efficiencies for ceria based cycles increase with operating temperature and s olar concentration ratio [73, 74] Efficient use of the inert gas was shown to be important; counter flow models predicted higher efficiencies and lower inert sweep gas requirements than mixed flow models [72, 73, 142] Additionally, highly effective gas phase heat recovery is paramount due to the large amount of inert and oxidant gases required [73, 74, 76, 143] Several methods of lowering the p O 2 during reduction have been considered during efficiency analyses such as inert gas sweeping (ideal mixing [72] or countercurrent [73, 76, 141, 142] ), vacuum pumping [75, 76, 78, 144, 145] electrochemical oxygen pumping [145] or chemical scavengers in combination with inert gas or vacuum pumping [76] Bork et al. [146] calculated efficiencies for La 0.8 Sr 0.2 MnO 3 (LSM20), LSM30, and LSM40; interestingly, considering reduction at 1773 K with a reduction p O 2 of 10 6 atm and an initial H 2 O concentration a factor of ten greater than the initial reduction extent, the authors observed greater efficiencies for isothermal cycling than for operation with a temperature swing. As noted by the authors, these calculations were performed without consideration of the sensible heating duty for excess steam. Muhich et al. [147] compared the efficiencies of ceria, Zr doped ceria, and several LaMnO 3 perovskites. For isothermal operation at 1673 K, a reduction p O 2 of 10 4 bar, and 50% effective gas phase heat recovery, the calculated efficiencies of all of the materials were very low (approximately 0.3%). At this temperature, oxidation in a pure stream of H 2 O would lead to an effective p O 2 of about

PAGE 113

113 3.8710 4 bar; thus, the isothermal efficiencies were a repercussion of the low changes in due to the small pressure swing. During isothermal operation, it is desirable to have a steep slope in the equilibrium isotherm with respect to p O 2 s uch that small pressure swings result in large changes in Ceria has a steep slope only at very low p O 2 typical of oxidation conditions ( e.g. 10 21 10 18 atm at 1073 K) during temperature swing operation. However, in the higher pressure range ( e.g. 10 7 10 3 atm) where isothermal or near isothermal operation is viable, of ceria is less sensitive to changes in p O 2 and is, thus, not ideal [16] Many perovskites, on the other hand, are well known to reduce mo re easily than ceria because of their low oxygen vacancy formation enthalpies [43, 44, 47, 85, 92] resulting in a characteristic curve that is less steep in the lower p O 2 range, but steeper in the higher p O 2 range of isothermal operation. This behavior should result in more efficient isothermal operation; to exp lore this hypothesis, a process model is presented in this chapter usi ng a thermodynamic equilibrium approach to predict solar to fuel efficiencies of LSM20, LSM40, La 0.6 Sr 0.4 Mn 0.6 Al 0.4 O 3 (LSMA6464), La 0.6 Ca 0.4 MnO 3 (LCM40), and La 0.6 Ca 0.4 Mn 0.6 Al 0.4 O 3 (LCMA6464) in comparison to those of ceria and Ce 0.8 Zr 0.2 O 2 (CZO20). Four d ifferent methods of lowe ring the p O 2 during the reduction step, a component of the solar fuel production process known to incur large efficiency hindering energy penalties, were considered. These include: inert gas sweeping, mechanical vacuum pumping, elec trochemical oxygen pumping, and thermochemical oxygen pumping. Thermodynamic Model for Determination of Solar to Fuel Efficiencies Model Formulation A thermodynamic model was developed in order to determine theoretical energy conversion efficiencies for the candidate materials LSM20, LSM40, LSMA6464, LCM40, LCMA6464, ceria and CZO20. A schematic of the model is shown in Figure 4 1. Thin, solid

PAGE 114

114 lines with arrows signify mass flows, the larger arrows indicate the heat transfer terms considered in the model, and the dashed line surrounding the process components defines the system boundary. The reduction and oxidation reactions, shown b elow, occur in separate reaction chambers, which are maintained uniformly at a reduction temperature ( T red ) and an oxidation temperature ( T ox ), respectively. ( 4 1 ) ( 4 2 ) Here, M x O y is the metal oxide of interest and red and ox are the reduction and oxidation nonstoichiometries, respectively, which are defined by the temperature and p O 2 at which the redox steps are performed. In the case of inert gas sweeping, gases enter the reduction chamber at state R1 and exit at state R2. A gas to gas heat exchanger (HEX gg,red ) is considered to enable pre heating of the inert gas. Other methods of lowering the p O 2 during the reduction step include mechanical vacuum pumping, electrochemical pumping, and thermochemical oxygen pumpi ng. The metal oxide is shuttled from states R3 to R4 during reduction and, subsequently, from states O3 to O4 during the oxidation step (in a counter flow arrangement with the delivered gases). In the case of redox cycling with a temperature swing, a solid to solid heat exchanger (HEX ss ) is employed. The oxidant, H 2 O in the representative case described here, enters the oxidation chamber at state O1 after pre heating in an additional gas phase heat exchanger (HEX gg,ox ) and exits at state O2. Defining the so lar to fuel efficiency. The high temperature process heat required to drive the fuel production cycle is provided by concentrated solar energy ( Q solar ) and is defined as the sum of the heat terms for each component of the overall process as shown below in equation ( 4 3 )

PAGE 115

115 ( 4 3 ) Here, Q aux,red and Q aux,ox are the heat or heat equivalent requirements for auxiliary processes associated with the reduction and oxidation steps, respectively, such as he ating of the delivered gases lower ing the p O 2 during reduction, and gas separation [141] ; Q sens,s represents the energy required to heat the reactive solid from T ox to T red ; Q red and Q ox are the thermal energies required to facilitate the reduction and oxidation reactions respectively; and Q rerad is the thermal energy loss due to reradiation. Q rerad is defined as a fraction of Q solar and is related to T red and the geometric concentration ratio ( C ) through equation ( 4 4 ) assuming blackbody emissivity of unity ( 4 4 ) Here, is the Stefan Boltzmann constant (5.670510 8 W m 2 K 4 ) and G sc is the solar irradiance constant (1.367 kW m 2 ). To maintain consistency with previous analyses [73, 141, 148] C was chosen to be 3000 in all cases herein. In defining Q rerad Q solar may be re written as shown below. ( 4 5 ) The process model derivation and descriptions to follow assume a H 2 O splitting redox cycle wherein H 2 is the produced fuel. All calculations described herein were performed using an in house developed MATLAB code, which was built with sufficient gen erality to quantify either H 2 O or CO 2 splitting efficiencies for any material for which equilibrium and thermodynamic data are known. The solar to fuel energy conversion efficiency ( solar to fuel ) i s defined as the ratio of the calorific value of the fuel equivalent to the product of the molar amount of fuel produced ( n H 2 ) and the higher heating value of that fuel (HHV H 2 ) to Q solar

PAGE 116

116 ( 4 6 ) Model assumptions and quantification of The model formulation relies on a number of assumption s, most importantly : (1) all processes operate at steady st ate; (2) all gases are treated as ideal ; (3) a blackbody cavity receiver absorbs the incident solar radiation; (4 ) the reduction and oxidat ion chambers are spatially isothermal and maintained at T red and T ox respectively; (5 ) the metal oxides and the gases with which they interact are at chemical equilibrium at all points ; and (6 ) no side reactions occur Additionally, reactant and product g ases travel through the reaction chambers in plug flow; that is, they are assumed to be perfectly mixed radially, with axial changes in gas composition occurring only due to the intended redox reactions with the metal oxide [73] The metal oxide is reduced to red in the reduction cham ber; red is calculated knowing T red and the reduction p O 2 ( p O 2,red ) both of which are inputs to the model via interpolation of equilibrium data extracted from literature or by using a defect model (described in further detail below). Interpolation is performed using a piecewise cubic Hermite interpolating polynomial (PCHIP in MATLAB). ox is achieved at the gas inlet to the oxidati on chamber, the point at which the metal oxide exits As the H 2 O entering and the metal oxide leaving the chamber are assumed to be in equilibrium, the p O 2 to which the metal oxide is subjected at this point is defined by the equil ibrium p O 2 due to H 2 O thermolysis at T ox ( i.e. p O 2,O1 = p O 2,O4 ). The equation governing the p O 2 at state s O1 and O4 is defined below: ( 4 7 )

PAGE 117

117 where n H 2 O is the molar amount of oxidant and is the reaction coordinate (equivalent to the molar amount of H 2 due to thermolysis). When considering a pure stream of H 2 O, p O 2 is unaffec ted by the value of n H 2 O and is only a function of temperature; for simplicity, calculation of the p O 2 at the oxidation reactor inlet was performed by set ting n H 2 O to unity (note, n H 2 O is calculated explicitly below prior to calculating the oxidant heating requirements) Temperature dependent values for K w were obtained from the NIST JANAF thermochemical tables. With knowledge of K w and an arbitrary value selected for n H 2 O equation ( 4 7 ) may be solved at any temperature to obtain p O 2,O1 and p O 2,O4 may then be calculated using the equation below. ( 4 8 ) Figure 4 2 shows the equilibrium p O 2 for pure streams of H 2 O and CO 2 respectively, at temperatures from 1473 1773 K. ox is calculated based on the equilibrium thermodynamics of the metal oxide at T ox and p O 2,O1 in the same manner as described above for the reduction reaction The difference between red and ox ( i.e. ) defines both the O 2 evolution during the reduction step ( n O 2 ) and the H 2 production during oxidation ( n H 2 ) T he calculations for the molar and heat terms in this chapter are performed per unit molar flow rate of metal oxide ( ); thu s, the y are not written in rate form. n O 2 and n H 2 are calculated according to equations ( 4 9 ) and ( 4 10 ) below. ( 4 9 ) ( 4 10 )

PAGE 118

118 Lowering the p O 2 during reduction. The method by which the p O 2 in the reduction c hamber is achieved directly impacts Q aux,red and solar to fuel In regard to the auxiliary heating requirements for the reduction step, this model considers only the heating or heat equivalent requirements for the p O 2 reduction method of interest; thus, Q aux,red will be referred to as Q sweep hereafter ( with the appropriate subscript indicating the method). The e nergetic penal ties associated with inert gas sweeping ( Q sweep,ig ) mechanical vacuum pumping ( Q sweep,vp ), electrochemical pumping ( Q sweep,ep ), and thermochemical oxygen pumping ( Q sweep,tp ) were co nsidered. Within the model formulation, Q sweep was defined as shown below: ( 4 11 ) where the terms are binary variables taking the form of either 0 or 1 to indicate the p O 2 reduction method being analyzed. For inert gas sweeping, the model assumes that the inert gas (Ar) is recycled and is in counter flow to the metal oxide [73, 141] The amount of Ar require d during reduction is determined via a mole balance on the reduction chamber where the molar change in O 2 across the reduction chamber is equivale nt to the amount released by the metal oxide ( 4 12 ) Dividing both side s of equation ( 4 12 ) by n inert the inert gas requirement per mole of metal oxide, and utilizing the mathematical definition of the partial pressure gives the following equation: ( 4 13 )

PAGE 119

119 where P sys is the total pressure in the red uction chamber (1 atm) ; p O 2,R1 is the p O 2 at the gas inlet, an input to the model that determines red at T red ; and p O 2,R2 is the p O 2 at the gas outlet and is defined by the p O 2 above the metal oxide at T red and ox The inert gas is pre heated in a gas to gas heat exchanger (HEX gg,red ) by the effluent of the reduction chamber ( i.e. Ar and the produced O 2 ) prior to heating by solar energy. An energy balance, shown in equation ( 4 14 ) below, was used to determine the temperature of the delivered Ar after passing through the heat exchanger ( T HEX,red ) for a given gas to gas heat recovery effectiveness ( gg ). ( 4 14 ) and are the respective molar enthalpy changes for Ar and O 2 from ambient temperature ( T 0 = 298 K) to T Enthalpy data for Ar and O 2 at the temperatures of interest were obtained from NIST JANAF. Note, w he n gg = 0, = 0 and T HEX,red = T 0 The energy penalties associated with inert gas sweeping may then be determined using equation ( 4 15 ) Bader et al. [73] found that the theoretical minimum work required to recycle the sweep gas was low compared to the total solar input; thus, this energy penalty was not considered. ( 4 15 ) The energy required to control the p O 2 via mechanical vacuum pumping wa s ca lculated via the equation below: ( 4 16 ) where R is the ideal gas c onstant (8.314 J mol 1 K 1 ); P 0 is ambient pressure (1 atm); heat to elec is the heat to electricity conversion efficiency, assumed to be 40% [144, 149] ; and pump is the

PAGE 120

120 pumping efficiency. pump is strongly affected by the desired partial pressure; the envelope function derive d by Brendelberger et al. [149] was used to calculate pump based on the input for p O 2,red (see Figure 4 3). O 2 removal via an electrochemical oxygen pump wherein the transport of oxygen ions through a membrane is promoted via an applied elect ric potential [145, 150] was also considered T he en ergy required to drive such a process was calculated as shown in the equation below. ( 4 17 ) F is the Faraday constant (96.485 kJ V 1 mol 1 ) E cell is the electric potential, and heat to work is the heat to work efficiency. For simplicity, E cell was assumed to be 1.32 V [145, 151] heat to work was defined as 40%. The final method considered to lower p O 2 was a thermochemical oxygen pump [149, 152, 153] This effectively couples the redox cycle of interest with a much lower temperature, second ary redox cycle that serves as the O 2 pump. Herein, a pumping material ( e.g. CoO, SrFeO 3 or SrMnO 3 ) is reduced at high temperature ( e.g. 1213 K) and releases O 2 at ambient pressure; the pumping material is re oxidized by absorbing the O 2 released during r eduction of the fuel producing metal oxide. The concept was proposed as a means of p O 2 reduction for use in conjunction with H 2 O or CO 2 splitting cycles by Ezbiri et al. [152] and Brendelberger et al. [149] and has, since then, been demonstrated experimentally [154] and analyzed theoretically [155] Due to the availability of thermochemical data and to enable comparison with the literature, the model considers a Co 3 O 4 /CoO redox cycle f or use as the thermochemical oxygen pump. However, in practical application, other materials such as SrFeO 3 or SrMnO 3 may be more desirable due to potential kinetic limitations associated with the phase transition in the

PAGE 121

121 Co 3 O 4 /CoO system [155, 156] The energy required to drive this process is shown in the equation below: ( 4 18 ) where is the energy requirement for the reduction reaction, is the change in enthalpy of Co 3 O 4 from T 0 to T and T red,tp and T ox,tp are the respective reduction and oxidation temperatures for the pumping material The temperature dependent enthalpy changes for Co 3 O 4 were calculated via integration of the specific heat capacity, obtained from the NIST Chemistry WebBook. Equilibrium of the redox reaction for the pumping material mandates the equality below [149] ( 4 19 ) and were obtained from FactWeb and were used to determine p O 2 for the thermochemical oxygen pump as a function of temperature. T red,tp was selected such that the p O 2 was 1 atm and T ox,tp was determined based on the desired reduction p O 2 with lower oxidation temperatures enabling pumping to lower p O 2 With T red,tp and T ox,tp known, equation ( 4 18 ) was used to calculate Q sweep,tp for the thermochemical oxygen pump. Figure 4 4 shows the calculations for the heat required to drive the thermochemical oxygen pump per mole of O 2 pumped versus p O 2 For comparison, calculations by Brendelberger et al. [149] are also shown. Oxidant and h eating requirements for oxidation. n H 2 O is defined such that the gas stream leaving the oxidation chamber is in equilibrium with the metal oxide entering ( i.e. p O 2,O2 = p O 2,O3 ) ; p O 2,O3 the p O 2 above the metal oxide is determined at T ox and red At state O2, the oxidant gas stream contains a mixture of H 2 O, H 2 and O 2 The molar amount of H 2 in the gas

PAGE 122

122 stream at this state is equal to the sum of the H 2 produced by the redox reaction and that associated with the temperature induced dissociati on of H 2 O A system of equations describing the equilibrium at state O2 and by which the oxidant requirement may be determined is shown below. ( 4 20 ) ( 4 21 ) Simultaneously solving equations ( 4 20 ) and ( 4 21 ) gives the reaction coordinate and, more importantly, the H 2 O delivery which maximizes the H 2 yield in the oxidation chamber. With knowledge of the H 2 O input and the H 2 production the auxiliary heating requirements for the oxidation step may be determined. The model considers only the energy required to heat liquid H 2 O at ambient temperature ( T 0 ) to steam at T ox ; thus, Q aux,ox is referred to as Q sens,ox hereafter. Prior to heating with solar energy, the oxidant gas is pr e heated in a heat exchanger (HEX gg,ox ) by the effluent from the oxidation chamber ( i.e. the unreacted H 2 O and produced H 2 ) The energy balance below was used to determine the temperature of the H 2 O stream after passing through HEX gg,ox ( T HEX,ox ) for a giv en gg ( 4 22 ) Here, and are the molar enthalpy changes for H 2 O and H 2 respectively, from T 0 to T Temperature dependent enthalpy data for H 2 O and H 2 were obtained from NIST JANAF. For the case where gg = 0, = 0 and T HEX,ox = T 0 After calculat ing T HEX,ox the heating duty for the oxidant gas may be determined as shown in equation ( 4 23 ) ( 4 23 )

PAGE 123

123 Thermal energy to facilitate reduction and oxidation. The energy required to facilitate thermal reduction of the metal oxide is determined via integration of the partial molar enthalpy change ( ) from ox to red as shown below. ( 4 24 ) Calculation of the thermal energy required for the oxidation reaction with H 2 O i s calculated in a similar manner, but must also account for the enthalpy change required to drive H 2 O splitti ng ( ). ( 4 25 ) ( 4 26 ) is the enthalpy of formation of H 2 O and was obtained from the NIST Chemistry WebBook [111] For high temperature isothermal operation, the oxidation exotherm could be used to assist with preheat ing of the inlet gases [148] ; however, the current model formulation assumes that the entirety of this energy is rejected to the ambient. S olid phase s ensible heating d uty. For cases where T red and T ox are not equal, thermal energy is required to heat the metal oxide from T ox to T red prior to reduction; this heating duty ( Q sens,s ) is defined by the specific heat capacity of the metal oxide ( c p,MO ) and the temperature swing between steps. When considering a solid to solid heat exchanger (HEX ss ) with an arbitrary heat recovery effectiveness ( ss ) Q sens,s may be calculated as shown below. ( 4 27 ) The formulation in equation ( 4 27 ) assumes that c p,MO is not a function of the temperature nor the reduction extent. A s done by Muhich et al. [147] the specific heat capacities of ceria and CZO20

PAGE 124

124 were assumed to be 80 J mol 1 K 1 [157] and those of all of the perovskites were taken to be 140 J mol 1 K 1 [158] Thermodynamic Properties of Candidate Materials Implementation of the process model to predict solar to fuel requires knowledge of the thermodynamic properties of the metal oxides of intere st i.e. over a wide range of temperatures and p O 2 and from the minimum ox to the maximum red Da ta was extracted from the followi ng sources: Panlener et al. [16] for ceria; Hao et al. [66] for CZO20; Bork et al. [146] for LSM20 and LSM40; and Takacs et al. [44] for LSMA6464, LCM40, and LCMA6464. For ceri a, LSM20, and LSM40 both the equilibrium and enthalpy data were directly extracted with no nee d for extrapolation. For CZO20, LSMA6464, LCM40, and LCMA6464 defect models were produced to model the equilibrium data as needed and the enthalpy data was extracted and extrapolated to the maximum red for each respective material. versus for each of these materials is shown in figure C 1 of Appendix C. Defect model for CZO20. T he defect equilibrium of CZO20 was modeled via consideration of point defects alone [28, 70] The reduction of CZO20 was described using the equation below: ( 4 28 ) where , and represent Ce 4+ ions (cerium on a cerium lattice site) oxygen on an o xygen lattice site, Ce 3+ ions (localized electron on a ceri um lattice site) and doubly ionized oxygen vacancies, respectively. Loss of oxygen leads to the creation of oxygen vacancies, which is accompanied by the reduction of Ce 4+ ions to Ce 3+ ions. As the cerium cation is tetravalent in ceria, Zr 4+ doping does n ot lead to the creation of intrinsic oxygen vacancies, and simply lowers the amount of Ce 4+ by the cor responding dopant concentration, e.g. When

PAGE 125

125 considering only point defects, the concentration of oxygen vacancies is equal to and charge neutrality mandates that the concentration of Ce 3+ must be twice that of oxygen vacancies. The respective concentrations of the defe ct species are defined below. ( 4 29 ) ( 4 30 ) ( 4 31 ) Using the species concentrations i n equations ( 4 29 ) ( 4 30 ) and ( 4 31 ) the defect equilibrium for this reaction may be defined in terms of and p O 2 as shown below. ( 4 32 ) K df represents the temperature dependent equilibrium constant for the defect reaction defined in equation ( 4 28 ) A nonlinear least squares fitting routine was used to determine K df at 1473, 1573, 1673, and 1763 K using data extracted from Hao et al. [66] ; a linea r fit for the logarithm of K df versus inverse temperature was produced and used to extrapolate to 1773 K. Defect models for LSMA 6464, LCM40, and LCMA6464. The defect equilibria of LSMA6464, LCM40, and LCMA6464 were described by considering (1) formation of doubly ionized oxygen vacancies accompanied by reduction of Mn 4+ to Mn 3+ and (2) disproportionation of Mn 3+ to Mn 4+ and Mn 2+ T he first of these defect reactions is defined by the equation ( 4 33 ) and the associated equilibrium con stant ( K 1 ) is defined in equation ( 4 34 ) ( 4 33 ) ( 4 34 )

PAGE 126

126 Here, Mn 4+ ions ( ), inherently present in LaMnO 3 upon introduction of a divalent dopant, are reduced to Mn 3+ ions ( ) by the electrons left behind upon the creation of d oubly ionized oxygen vacancies. Mn 3+ may disproportionate into Mn 4+ and Mn 2+ ( i.e. where represents Mn 2+ ); combining the disproportionation reaction with that shown in equation ( 4 33 ) yields reaction defined below. ( 4 35 ) Equation ( 4 35 ) represents the formation of doubly ionized oxygen vaca ncies accompanied by the reduction of Mn 4+ to Mn 2+ The equilibrium constant ( K 2 ) describing the double red uction rea ction is shown below. ( 4 36 ) Considering the reactions shown in equatio ns ( 4 33 ) and ( 4 35 ) is defined by Site balances for manganese and oxygen and charge neutrality yield the following equations: ( 4 37 ) ( 4 38 ) ( 4 39 ) where represents the concent ration of the divalent A site dopant ( e.g. Sr 2+ or Ca 2+ ) and is the concentration of the trivalent B site dopant Al 3+

PAGE 127

127 Equations ( 4 34 ) and ( 4 36 ) through ( 4 39 ) were combined into a single l inear equation describing K 1 and K 2 in terms of only and p O 2 as done by Takacs et al. [44] This linearized equation was used with the extracted data to determine K 1 and K 2 at 1573, 1673, and 1773 K for LSMA6464, LCM40, and LCMA6464 Linear fits for the logarithm of K 1 and K 2 respectively, versus inverse temperature were produced to extrapolate and determine the equilibrium constants at 1473 K. With the equilibrium constants known, the abovementioned equations describing the defect equilibria were used to predict the e quilibrium at 1473 K and extrapolate to lower p O 2 than tested experimentally in the literature. Extrapolations of the equilibrium for LSMA6464, LCM40, and LCMA6464 at 1473 1773 K and the derived equilibrium constants are shown in figures C 2 through C 7 of Append ix C. Results and Discussion I sothermal Operation Total fuel yields, gas delivery requirements (inert and oxidant), energy penalties, and solar to fuel were comput ed for ITRC performed at 1473 1773 K with ceria, CZO20, LSM20, LSM40, LSMA6464, LCM40, and L CMA6464. As a representative example the calculations to be described below were performed considering H 2 O as the oxidant and a reduction p O 2 ( p O 2,R4 ) of 10 6 atm Each of the p O 2 reduction methods described herein were considered, separately, when calcul ating the heat terms and solar to fuel ; for these calculations, the gas phase heat recovery effectiveness was the same for HEX gg,red (where applicable) and HEX gg,ox and assumed to be 0.95. Figure 4 5 shows the H 2 production per mole of metal oxide for each material versus isother mal operating temperature ( T iso ) for p O 2,R4 = 10 6 atm Here, it is evident that the fuel yields using ceria are lower than those of all other materials considered at each of the operating temperatures Fuel yields at 1773 K are ordered as follows: LCM40 > LSMA6464 > LCMA6464

PAGE 128

128 > LSM40 > CZO20 > LSM20 > Ceria. Fuel yields per mole of metal oxide for LCM40 were 0.1601 mol es H 2 while those of ceria were 0.0711 mol es H 2 As expected, the fuel production for each material decreases monotonically as T iso decrease s. To promote visualization of the thermodynamics that dictate the isothermal fuel yields, Figure 4 6 shows the equilibrium oxygen content (2 for ceria and CZO20 on the left axis or 3 for the perovskites on the right axis) at 1773 K. The black dashed v ertical line shows the p O 2 at which the oxidation reaction with H 2 O would occur when operating isothermally. Using this equilibrium map, computation of and, thus, the fuel yields per mole of metal oxide may be performed by determining the vertical dista nce between the oxygen content at the reduction p O 2 (the leftmost portion of the figure) and that at the oxidation p O 2 As the domain across which the calculations are performed remains constant for each of the materials, it is clear that a higher slope ( i .e. larger changes in per unit change in p O 2 ) results in greater fuel yields when operating isothermally. Thus a material that reduces more easily elicits a larger and is m ore desirable for ITRC. The amount of inert gas per mole of metal oxide required during the reducti on reaction is lowest for ceria at all T iso examined as shown in Figure 4 7 At 1773 K, the inert gas requirements are highest for LCM40, while they are highest for LCMA6464 at 1473 K. Since T red = T ox for ITRC the p O 2 at the gas outlet of the reduction reactor ( p O 2,R2 = p O 2,R3 ) is equivalent to the p O 2 at the gas inlet of the oxidation reactor ( p O 2,O1 = p O 2,O4 ), which is only a function of temperature. Therefore p O 2,R2 is equivalent for all materials during ITRC and, according to equation ( 4 13 ) the differences between the inert gas requirement s for each of the materials depends only on the amount of O 2 released At constant T iso the material that releases the most O 2 during reduction (and, thus, produces the most fuel) requires more inert gas to control the p O 2 in the reduction chamber For most of the materials examined, t he inert gas

PAGE 129

129 requirements increase with decreasing T iso In contrast, ceria exhibits non monotonic behavior wherein the inert gas requirements slightly increase from 1773 to 1673 K, but decrease with further decreases in T iso For LSM20, the inert gas requirement decreases from 1573 to 1473 K. The H 2 O requirements for each of the materia ls are shown in Figure 4 8. In genera l, the H 2 O requirements are directly proportional to the fuel yields and the p O 2 at the gas outlet of the oxidation chamber ( p O 2,O2 = p O 2,O3 ) and are inversely proportional to T ox For all materials, this value increase s as T iso decreas es; the highest H 2 O requirements per mole of metal oxide at the extrema of T iso are 3.931 moles H 2 O for LCM40 (24.549 moles H 2 O per mole H 2 produced) at 1773 K and 55.987 moles H 2 O for LCMA6464 (746.195 moles H 2 O per mole H 2 produced) at 1 473 K. The trends in the heat or heat equivalent terms are mostly independent of composition for the perovskites; therefore, to simplify the discussion we will focus only on ceria, CZO20, and LSM40 here. The heating requirements for these materials, which largely dictate solar to fuel are shown in Figure 4 9. The blue solid, dashed, dot dashed, and dotted lines show the energy penalties for lowering the p O 2 during reduction via inert gas sweeping, vacuum pumping, electrochemical pumping, and thermochemica l oxygen pumping, respectively. Note, only one of these terms is considered at a time when calculating solar to fuel ; they are shown on the same plot for the sake of comparison. For ceria, Q sens,ox varies only slightly with T iso ; this is because the oxid ant requirements do not significantly change under the examined conditions. H owever, this value increases notably with decreasing T iso for both CZO20 and LSM40. Q sens,ox for LSM40 is greater than those of ceria and CZO20 at all T iso At 1473 K Q sens,ox fo r LSM40 is 13.167 and 2.862 times greater than those of ceria and CZO20, respectively. Q red is depende nt on the dependent partial

PAGE 130

130 molar enthalpy. For all considered herein, the partial molar enthalpies are ordered as follows: Ceria > CZO20 > LSM40. At 1773 K, Q red for ceria exceeds those of CZO20 and LSM40, but Q red for L SM40 is the largest at all T iso below abo ut 1749 K Lastly, Q sweep,ig Q sweep,vp Q sweep,ep and Q sweep,tp for LSM40 are greater than those of ceria and CZO20 at all T iso T he differences in the energy penalties for lowering the p O 2 during reduction via any of the methods discussed herein are dictated solely by the material specific reduction thermodynamics ( i.e. the molar amount of O 2 released) With knowledge of the fuel yields and the energy penalties for each process in the theoretical solar fuel production plant, solar to fuel may be calculated. The predicted solar to fuel for ITRC at T iso from 1473 1773 K for each of the materials are shown in Figure 4 10; subplots (a) (d) show the calculations for each of the p O 2 redu ction methods. Across all p O 2 reduction meth ods and T iso considered, solar to fuel for ceria are the lowest of all of the materials. Additionally, solar to fuel increases monotonically with increasing T iso The peak efficiencies for operation with a thermochemical pump are the highest amongst the p O 2 reduction methods and vacuum pumping leads to the lowest peak efficiencies, largely due to the low pump of 1.181 % (see Figure 4 3) at the chosen reduction p O 2 of 10 6 atm From Figure 4 4, it is clear that the energy required to drive the thermochemic al pump does not drastically increase when pumping to lower p O 2 providing the potential for even further increases in solar to fuel As shown in Figure 4 10, solar to fuel for ceria during ITRC at 1773 K with a reduction p O 2 of 10 6 atm is 28.259 %; perf orming the reduction reaction at 10 7 atm would increase solar to fuel to 31.314 %. For all p O 2 reduction methods, solar to fuel for the respective materials during ITRC at 1773 K are ordered as follows: LSMA6464 > LSM40 > LCMA6464 > LSM20 > LCM40 > CZO20 > Ceria. The maximum solar to fuel achieved by LSMA6464 at 1773 K were 35.17% for

PAGE 131

131 thermochemical pumping, 22.98% for electrochemical pumping, 17.3 % for inert gas sweeping, and 6.081 % for vacuum pumping. at 1773 K, solar to fuel for LCM40 are lower than those of the other perovskites examined. The lower solar to fuel for LCM40 at 1773 K is a repercussion of a notably larger Q solar requirement to drive the cycle; Q sens,ox Q red and Q sweep (the material dependent terms whi ch make up the denominator of the solar to fuel equation) are larger for this material than for all other materials. Q sens,ox is greater due to a larger oxidant requirement; the increased Q red and higher partial molar enthalpies than the other perovskites; lastly, the Q sweep terms are highest The enhanced solar to fuel for ITRC with LSMA6464 is largely a result of its relatively low partial molar enthalpy, which decr eases with increasing For example, even when for LSMA6464 is large r than that of LCMA6464 Q red for LSMA6464 is lower. Thus, from a thermodynamic perspective, LSMA6464 is the most advantageous material of those evaluated herein for use in ITRC. Temp erature Swing Operation The abovementioned analysis was also perfo rmed for TSRC Temperature swings between the reduction and oxidation steps ( T ) up to 300 K were considered. Calculations for the reduction reaction were performed at 1773 K with a reductio n p O 2 of 10 6 atm. Gas phase heat recovery of 95% was assumed for both HEX gg,red and HEX gg,ox but solid phase heat recovery was not considered. Figures 4 11 through 4 13 show the fuel yields, inert gas requirements, and oxidant requirements for each of t he materials versus T As T increases (and T ox decreases), the fuel yields increase and the inert gas requirements decrease for all materials. During ITRC, p O 2,R2 = p O 2,O1 as a result of the constant operating temperature and these values are not influenced by the thermodyn amics of the material; however, this is not the case during TSRC. When T red and T ox

PAGE 132

132 are not equivalent, t he p O 2 above the material at T red and ox will be greater than in the isothermal case; this is due to the fact that ox decreases with T ox Mathematica lly speaking, this causes the denominator of equation ( 4 13 ) to increase as T ox decreases and T increases, which results in a lower inert gas requirement. Fo r each of the materials, the inert gas requirement approaches zero as T increases; this trend is consistent with the work of Ehrhart et al. [144] For LSMA6464, LCMA6464, an d LSM40, the oxidant requirements increase with increasing T while they decrease for all other materials. p O 2,O2 = p O 2,R1 for ITRC but this value is influenced by material specific thermodynamics when a T is introduced and decreases as T increases. Co mparing ceria and LSMA6464, it can be seen that p O 2,O2 for ceria decreases more per unit decrease in T than does that of LSMA6464. This fact, in combination with the greater increases in fuel yields with changing T for LSMA6464, is responsible for the di ssimilar trends regarding the oxidant requirements. The energy requirements for TSRC with ceria, CZO20, and LSM40 are shown in Figure 4 14. Q sens,s for LSM40 are the largest at all T ox due to its higher specific heat capacity. In contrast to ITRC, Q red for ceria are the largest of the three materials until T of about 291 K. As a result of the inert gas requirements approaching zero with increasing T Q sweep,ig also approaches zero. Q sweep,vp Q sweep,ep and Q sweep,tp all increase with T as the molar O 2 evolutio n increases; these values are largest for LSM 40 at all T Considering the fuel yields and heat terms for TSRC, solar to fuel were calculated at T from 0 300 K for each of the materials and are shown in Figure 4 15. Interestingly, the trends are notably different than those of ITR C; solar to fuel for inert gas sweeping exhibit non monotonic behavior and those for vacuum pumping, electrochemical pumping, and thermochemical T For inert gas sweeping, solar to fuel for ceria increases from

PAGE 133

133 15.42% for ITRC at 1773 K T T solar to fuel for ceria decreases as the enhancements in fuel production which incre ase the numerator of equation ( 4 6 ) are not substantial enough to overcome the rising values of Q red and Q sens,s The maximum solar to fuel T of 200 K. It is also noteworthy that CZO20 outperforms all other materials except for LCM40 at this T Under the conditions examined, solar to fuel for the other p O 2 reduction methods decrease T increases. This trend persists at lower gg for all materials except for ceria, CZO20, and LCM40, where solar to fuel at lower gg T or exhibit non monotonic behavior in the temperature range examined. Figure s C 8 and C 9 in Appendix C shows the effect of varying gg on solar to fuel for ceria. Based on the abovementioned obs ervations, inert gas sweeping is the most efficient method of p O 2 reduction (of those examined herein) T Regarding the material choice, ceria is most desirable T low ( i.e. less than 100 K but greater than about 15 K ) and LCM40 enables higher solar to fuel T Conclusion A thermodynamic model of a solar fuel production plant was developed and described. This model was utilized to perform a theoretical investigation of the solar to fuel ener gy conversion efficiencies of ceria, CZO20, LSM20, LSM40, LSMA6464, LCM40, and LCMA6464 with an emphasis on isothermal and near isothermal redox cycling. The fuel productivity of the respective materials were calculated using thermodynamic data taken from the literature and defect models (possessing excellent agreement with the extracted experimental data) were produced as necessary to extrapolate to the desired oxygen partial pressure and temperature ranges. The effects of four p O 2 reduction methods were c onsidered, namely: inert gas sweeping, vacuum pumping, electrochemical pumping, and thermochemical pumping. The

PAGE 134

134 relative performance of the examined materials were dependent upon the operating scheme ( i.e. isothermal or temperature swing), the process para meters for the solar plant, and the conditions under which the respective redox steps were performed. Under most conditions tested, ceria produces the lowest amount of fuel and requires notably lower inert gas and oxidant delivery. For isothermal redox cyc ling, LSMA6464 was shown to give the highest efficiency of 35.17 % at 1773 K for p O 2 reduction v ia thermochemical pumping In contrast, the best performance during cycling with a temperature swing was 43.27 % for LCM40 with inert gas sweeping when the reduct ion reaction was performed at 1773 K and oxidation occurred at 1573 K. However, ceria is the most advantageous material in this case when a low temperature swing is desired (despite its relatively low fuel yields) giving an efficiency of 37.39 % with a tem perature swing of only 100 K. The work herein shows the possibility of enhanced solar to fuel efficiencies during isothermal or temperature swing redox cycling with the use of perovskite oxides or varying p O 2 reduction methods. Commercialization of this fu el production technology should consider both the material and reactor dependent parameters as well as the cycling scheme during the design process in order to achieve the highest possible efficiencies.

PAGE 135

135 Figure 4 1. Process model schematic for the p roduction of H 2 via a two step metal oxide redox cycle. The reduction and oxidation reactions occur in two separate chambers. Ar is used as an inert sweeping gas during reduction and oxidation of the metal oxides is performed in the subsequent step in the presence of H 2 O. Gas phase heat exchangers pre heat the inert and oxidant gases, respectively and solid phase heat recovery is considered during temperature swing cycles. Arrows traveling between the reduction and oxidation chambers (R4 to O3 and O4 to R3 respectively) signify the path of the metal oxide, and the bold arrows denote the energy terms considered when defining the process efficiency. The dashed line denotes the system boundary.

PAGE 136

136 Figure 4 2. Equilibrium p O 2 versus temperature for H 2 O and C O 2 respectively. The requisite thermodynamic data was obtained from the NIST JANAF thermochemical tables. Calculations were performed by assuming equilibrium of the respective thermolysis reactions.

PAGE 137

137 Figure 4 3 Pumpi ng efficiency (electric to pump ) of a mechanical pump versus the logarithm of the inlet pressure to the pump ( i.e. the reduction p O 2 ) This efficiency was calculated using the envelope function derived by Brendelberger et al. When calculating the heat equivalent of pump work, a heat to el ectricity conversion efficiency of 40% was assumed.

PAGE 138

138 Figure 4 4 Heat input per mole of O 2 pumped versus the logarithm of the reduction p O 2 for a thermochemical oxygen pump driven by a Co 3 O 4 /CoO redox cycle. Reduction of Co 3 O 4 occurred at approximately 1212 K such that the p O 2 was 1 atm; the oxidation temperature was set such that the desired reduction p O 2 for the fuel producing redox cycle was achieved. The solid line shows calculations performed herein using thermodynamic data from FactWeb and the NIS T Chemistry WebBook. For comparison, the markers show calculations by Brendelberger et al.

PAGE 139

139 Figure 4 5. H 2 production (equivalent to = red ox ) per mole of metal oxide (MO) versus operating temperature for isothermal redox cycles with each of the materials considered red was calculated at T iso and p O 2 = 10 6 atm while ox was calculated at T iso and the equilibrium p O 2 defined b y H 2 O thermolysis at T iso

PAGE 140

140 Figure 4 6. Equilibrium oxygen content (2 or 3 ) at 1773 K from 10 6 to 10 1 atm for each of the materials considered. The black dashed vertical line, located at p O 2 = 7.5610 4 atm, shows the p O 2 at which oxidation by H 2 O would occur during isothermal operation. Extracted data shown for ceria and CZO20 refer to the left ordinate (2 ), while all other data refer to the right ordinate (3 ). The axis scales were kept consistent to enable comparison.

PAGE 141

141 Figure 4 7 Inert g as requirement during reduction per mole of metal oxide (MO) versus operating temperature for isothermal redox cycles with each of the materials considered. The initial p O 2 of the gas stream was based on the sweep gas purity, which was set to 10 6 atm and the total pressure of the reduction chamber was set to 1 atm. The calculations assume that the inert gas is recycled and is in counter flow to the metal oxide. The p O 2 at the gas outlet of the reduction chamber is assumed to be equivalent to the p O 2 above the metal oxide at T iso and ox

PAGE 142

142 Figure 4 8 Oxidant delivery requirement per mole of metal oxide (MO) versus operating temperature for isothermal redox cycles with each of the materials considered. Th e calculations were performed such that the p O 2 a t the gas outlet of the oxidation chamber was equivalent to that above the metal oxide at T iso and red ; red was determined at T iso and a p O 2 of 10 6 atm.

PAGE 143

143 Figure 4 9 Energy penalties due to heating of the oxidant ( Q sens,ox ), the reduction endotherm ( Q red ), heating of the inert gas ( Q sweep,ig ), heat equivalent of pumping work for a mechanical vacuum pump ( Q sweep,vp ), heat equivalent of pumping wor k for an electrochemical pump ( Q sweep,ep ), and the heat input required to drive a Co 3 O 4 /CoO thermochemical oxygen pump ( Q sweep,tp ) versus operating temperature for isothermal redox cycles with (a) ceria, (b) CZO20, and (c) LSM40 The calculations were performed considering a reduction p O 2 of 10 6 atm and a gas to gas heat recovery effectiveness of 95%.

PAGE 144

144 Figu re 4 10 Solar to fuel energy conversion efficiency for H 2 O splitting versus operating temperature for isothermal redox cycles with each of the materials considered. The reduction p O 2 was 10 6 atm, the gas to gas heat recovery effectiveness was 95%, and t he geometric concentration ratio was 3000. The reduction p O 2 was controlled via (a) inert gas sweeping, (b) a mechanical vacuum pump, (c) an electrochemical pump, or (d) a thermochemical oxygen pump.

PAGE 145

145 Figure 4 11 H 2 production (equivalent to = red ox ) per mole of me tal oxide (MO) versus temperature swing for non isothermal redox cycles with each of the materials considered. red was calculated at T red = 1773 K and p O 2 = 10 6 atm, while ox was calculated at T ox = T red T and the equili brium p O 2 defined by H 2 O thermolysis at T o x

PAGE 146

146 Figure 4 12 Inert gas requirement during reduction per mole of metal oxide (MO) versus temperature swing for nonisothermal redox cycles with each of the materials considered. The initial p O 2 of the gas str eam was based on the sweep gas purity, which was set to 10 6 atm, and the total pressure of the reduction chamber was set to 1 atm. The calculations assume that the inert gas is recycled and is in counter flow to the metal oxide. The p O 2 at the gas outlet of the reduction chamber is assumed to be equivalent to the p O 2 above the metal oxide at T red and ox

PAGE 147

147 Figure 4 13 Oxidant delivery requirement per mole of metal oxide (MO) versus temperature swing for nonisothermal redox cycles with each of the mater ials considered. The calculations were performed such that the p O 2 at the gas outlet of the oxidation chamber was equivalent to that above the metal oxide at T ox and red ; red was determined at T red = 1773 K and a p O 2 of 10 6 atm.

PAGE 148

148 Figure 4 14 Energy penalties due to heating of the oxidant ( Q sens,ox ), heating of the metal oxide from the oxidation temperature to the reduction temperature ( Q sens,s ), the reduction endotherm ( Q red ), heating of the inert gas ( Q sweep,ig ), heat equivalent of pumping work for a mechanical vacuum pump ( Q sweep,vp ), heat equivalent of pumping work for an electrochemical pump ( Q sweep,ep ), and the heat input required to drive a Co 3 O 4 /CoO thermochemical oxygen pump ( Q sweep,tp ) versus temperature swing for nonisothermal redox cycles w ith (a) ceria, (b) CZO20, and (c) LSM40. The calculations were performed considering a reduction p O 2 of 10 6 atm, a gas to gas heat recovery effectiveness of 95%, and zero solid to solid heat recovery.

PAGE 149

149 Figure 4 15 Solar to fuel energy conversion effic iency for H 2 O splitting versus temperature swing for nonisothermal redox cycles with each of the materials considered. The reduction p O 2 was 10 6 atm, the gas to gas heat recovery effectiveness was 95%, zero solid to solid heat recovery was considered, and the geometric concentration ratio was 3000. The reduction p O 2 was controlled via (a) inert gas sweeping, (b) a mechanical vacuum pump, (c) an electrochemical pump, or (d) a thermochemical oxygen pump.

PAGE 150

150 CHAPTER 5 SUMMARY AND CONCLUSIONS Solar thermochem ical redox cycling is a promising means of renewable fuel production. Improving the energy conversion efficiencies such that commercial realization of this technology becomes viable requires (1) discovery of candidate materials, (2) experimental examinat ion of the thermodynamics and kinetics of these materials, (3) theoretical investigations into the performance, and (4) development of prototype reactors and practical demonstr ation of efficiencies. The major focuses of this work have been experimental cha racterization of redox materials under carefully controlled operating conditions and theoretical investigation of the material dependent parameters which govern how efficiently solar energy may be converted in to chemical fuels. A high temperature tubular r eactor and novel experimental framework were developed with which the thermodynamic and kinetic param e ters of redox materials may be assessed The reactor system enables precise control of the sample environment, e.g. the temperature, total press ure, and p O 2 Here, t he p O 2 was controlled by delivering a mixture of H 2 and H 2 O and r edox reactions were initiated isothermally via rapid shifts in the H 2 O concentration Quantitative analysis of the reaction products in the reactor effluent was performed using mas s spectrometry by evaluating reaction induced deviations in the measured H 2 from the steady state baseline. The experimental system and procedure were validated by measuring the equilibrium of ceria at 1173 147 3 K and p O 2 from 4.5410 18 1.0210 9 atm; t he experimentally measured results showed excellent agreement with accepted data from the literature with percent differences being less than 10% across all temperatures The utility of the system for extracting kinetic pa rameters was also demonstrated. T o simulate the measured reaction rates without reactor inherent experimental artefacts, a tanks in series mixing model was developed that accounted for

PAGE 151

151 the effects of gas dispersion on the mass spectrometer measurements. The intrinsic kinetic behavior of c eria was extracted by coupling the mixing model with a mechanistic kinetic model; this model assumed a reversible two step reaction mechanism mediated by the formation of an intermediate hydrox y l species and considered the concentrations of defect species in both the bulk and on the sample surf ace. A minimization routine was used to extract the defect formation enthalpy and entropy and the pre exponential factors and activation energies of the forward and reverse reactions. In general, the extracted paramet ers displayed good agreement with the literature ; differences in the derived values were attributed to differences in the experimental setups and procedures and efforts to remove the effects of dispersion. Following validation, the experimental system and methods were used in characterizing the thermodynamics of Sr and Al doped YMnO 3 a new class of materials for which neither physico chemical nor thermodynamic data exists in literature. The crystal structures, elemental compositions, and sample surfaces were evalua ted via PXRD (performed both before and after redox cycling), ICP MS, SEM, and EDS Isothermal relaxation experiments were performed in the high temperature tubular reactor at 973 117 3 K and p O 2 from 1.2410 20 2.2610 13 atm. Thermogravimetric analysis was also performed at 1173 147 3 K and p O 2 from 1.6110 4 3.2310 2 atm. PXRD scans of as synthesized and reduced Y 0.8 Sr 0.2 Mn 0.6 Al 0.4 O 3 and Y 0.9 Sr 0.1 Mn 0.6 Al 0.4 O 3 revealed the presence of phase impurities upon reduction, while the diffraction patter ns of Y 0.8 Sr 0.2 Mn 0.4 Al 0.6 O 3 were largely unchanged even for large deviations from stoichiometry. D efect model s were developed considering (1) oxygen vacancy formation coupled with reduction of Mn 4+ to Mn 3+ and/or Mn 2+ and (2 ) formation of defect clusters b etween Mn 2+ and nearby oxygen vacancies. The model s were used to produce fits to the thermogravimetric data for Y 0.8 Sr 0.2 Mn 0.6 Al 0.4 O 3 and Y 0.9 Sr 0.1 Mn 0.6 Al 0.4 O 3 ; fits to the lower

PAGE 152

152 tempera ture and p O 2 data obtained in the tubular reactor were poor as a resul t of the lack of phase purity observed via PXRD For Y 0.8 Sr 0.2 Mn 0.4 Al 0.6 O 3 the Mn 4+ /Mn 3+ model was found to fit the data well in the lower temperature and p O 2 region, while lower error fits were obtained in the high p O 2 region via the Mn 4+ /Mn 2+ model. Las tly, a theoretical investigation into the solar to fuel energy conversion efficien cies achievable with perovskite oxides was performed. Data from the experimental investigations herein and those presented in literature suggest s that the thermodynamics of m any perovskites are not amenable to nonisothermal operation with large temperature swings. Instead, the focus of t he theoretical assessment wa s to evaluate their performance during isothermal or near isothermal cycling in comparison to that of the current state of the art, ceria. Efficiencies were calculated for seven candidate materials during isothermal redox cycling at 1473 1773 K with a p O 2 during reduction of 10 6 atm, 95% gas phase heat recovery, and a geometric concentration ratio of 3000. Additional ly, the impacts of four p O 2 reduction methods were analyzed. It was found that all material s examined outperformed ceria during isothermal operation, largely due to their increased oxygen exchange capacities. The efficiency of LSMA6464 at 1773 K was ap prox imately 35.17% when the p O 2 during reduction was controlled using a thermochemical oxygen pump, while that of ceria under identical conditions was 28.26%. During near isothermal operation, LCM40 achieved the highest efficiency of 43.27% with inert gas swee ping when the reduction reaction was performed at 1773 K and oxidation occurred at 1573 K.

PAGE 153

153 APPENDIX A SUPPLEMENT TO CHAPTER 2 Figure A 1. Transient in undoped ceria measured via residual gas analysis using an isothermal relaxation scheme in H 2 /H 2 O Separate reactions for which the p O 2 are equivalent are distinguished via solid, dashed, and dot dashed lines. The p O 2 in atm, for each reaction is given in the legend. (a) versus elapsed time for isothermal reduction reactions performed at 1173 K and p O 2 between 4.5410 18 and 3.1410 16 atm. (b) versus elapsed time for isothermal oxidation reactions performed at 1173 K and p O 2 between 1.2610 17 and 7.0910 16 atm.

PAGE 154

154 Figure A 2. Transient in undoped ceria measured via residual gas analysis usin g an isothermal relaxation scheme in H 2 /H 2 O. Separate reactions for which the p O 2 are equivalent are distinguished via solid, dashed, and dot dashed lines. The p O 2 in atm, for each reaction is given in the legend. (a) versus elapsed time for isothermal reduction reactions performed at 1273 K and p O 2 between 1.0010 15 and 1.4310 13 atm. (b) versus elapsed time for isothermal oxidation reactions performed at 1273 K and p O 2 between 1.6010 14 and 1.4310 13 atm.

PAGE 155

155 Figure A 3. Transient in undoped ce ria measured via residual gas analysis using an isothermal relaxation scheme in H 2 /H 2 O. Separate reactions for which the p O 2 are equivalent are distinguished via solid, dashed, and dot dashed lines. The p O 2 in atm, for each reaction is given in the legend (a) versus elapsed time for isothermal reduction reactions performed at 1373 K and p O 2 between 1.1710 13 and 5.3510 11 atm. (b) versus elapsed time for isothermal oxidation reactions performed at 1373 K and p O 2 between 1.8810 12 and 9.5310 11 atm.

PAGE 156

156 F igure A 4. Reaction rate for reduction and oxidation reactions of undoped ceria performed in a H 2 /H 2 O environment. Separate reactions for which the p O 2 are equivalent are distinguished via solid, dashed, and dot dashed lines. The p O 2 in atm, for each rea ction is given in the legend. (a) d /d t versus for isothermal reduction reactions, proceeding from the left of the figure to the right, performed at 1173 K and p O 2 between 4.5410 18 and 3.1410 16 atm. (b) d /d t versus for isothermal oxidation reactions, proceeding from the right of th e figure to the left, performed at 1173 K and p O 2 between 1.2610 17 and 7.0910 16 atm.

PAGE 157

157 Figure A 5. Reaction rate for reduction and oxidation reactions of undoped ceria performed in a H 2 /H 2 O environment. Separate reactions for which the p O 2 are equiva lent are distinguished via solid, dashed, and dot dashed lines. The p O 2 in atm, for each reaction is given in the legend. (a) d /d t versus for isothermal reduction reactions, proceeding from the left of the figure to the right, performed at 1273 K and p O 2 between 1.0010 15 and 1.4310 13 atm. (b) d /d t versus for isothermal oxidation reactions, proceeding from the right of th e figure to the left, performed at 1273 K and p O 2 between 1.6010 14 and 1.4310 13 atm.

PAGE 158

158 Figure A 6. Reaction rate for reduction and oxidation reactions of undoped ceria performed in a H 2 /H 2 O environment. Separate reactions for which the p O 2 are equiva lent are distinguished via solid, dashed, and dot dashed lines. The p O 2 in atm, for each reaction is given in the legend. (a) d /d t versus for isothermal reduction reactions, proceeding from the left of the figure to the right, performed at 1373 K and p O 2 between 1.1710 13 and 5.3510 11 atm. (b) d /d t versus for isothermal oxidation reactions, proceeding from the right of th e figure to the left, performed at 1373 K and p O 2 between 1.8810 12 and 9.5310 11 atm.

PAGE 159

159 Figure A 7. Exemplary temporal distributions of species concentrations during reduction (top subplots) and oxidation (bottom subplots) at 1173 K and 1473 K. For ea ch calculation, the input p H 2 O: p H 2 ratio can be determined from the corresponding conditions shown on Figure 2 8 ( e.g. p H 2 O: p H 2 = 0.56:1 for reduction at 1173 K). Surface (solid lines) and bulk (dashed lines) species concentrations are defined as the mola r ratio of defect species to ceria in the surface and bulk, respectively.

PAGE 160

160 APPENDIX B SUPPLEMENT TO CHAPTER 3 Figure B 1. PXRD patterns of YSMA8264, YSMA8246, YSMA9164, and YSMCo6464 (measured by Huang and Huang )

PAGE 161

161 Figure B 2. PXRD patterns of Y SM20 as synthesized and as me asured by Huang and Huang

PAGE 162

162 Figure B 3 SEM image of YSMA8264 powder.

PAGE 163

163 Fi gure B 4 SEM image of YSMA8246 powder.

PAGE 164

164 Fi gure B 5 SEM image of YSMA9164 powder.

PAGE 165

165 Fi gure B 6 EDS images showing the distributions of Y, S r, Mn, and Al in ( a) YSMA8264, ( b) YSMA8246, and ( c) YSMA9164

PAGE 166

166 APPENDIX C SUPPLEMENT TO CHAPTER 4 Figure C 1. Partial molar enthalpy change per mole of monoatomic oxygen versus nonstoichiometry for each of the materials considered.

PAGE 167

167 Figure C 2 Equilibrium oxygen content of LSMA6464 versus p O 2 at 1473 1773 K. The marker s indicate measured data extracted from Takacs et al. The solid lines represent defect model fits considering (1) production of doubly ionized oxygen vacancies and reduction of M n 4+ to Mn 3+ and (2) disproportionation of Mn 3+ to Mn 4+ and Mn 2+

PAGE 168

168 Figure C 3 Logarithm of the equilibrium constants of oxygen vacancy formation (K 1 ) and disproportionation (K 2 ) versus inverse temperature for LSMA6464. Markers represent the extracted va lue at a single temperature and the solid lines represent linear fits.

PAGE 169

169 Figure C 4 Equilibrium oxygen content of LCM40 versus p O 2 at 1473 1773 K. The markers indicate measured data extracted from Takacs et al. The solid lines represent defect model fi ts considering (1) production of doubly ionized oxygen vacancies and reduction of Mn 4+ to Mn 3+ and (2) disproportionation of Mn 3+ to Mn 4+ and Mn 2+

PAGE 170

170 Figure C 5 Logarithm of the equilibrium constants of oxygen vacancy formation (K 1 ) and disproportionati on (K 2 ) versus inverse temperature for LCM40. Markers represent the extracted value at a single temperature and the solid lines represent linear fits.

PAGE 171

171 Figure C 6 Equilibrium oxygen content of LCMA6464 versus p O 2 at 1473 1773 K. The markers indicate me asured data extracted from Takacs et al. The solid lines represent defect model fits considering (1) production of doubly ionized oxygen vacancies and reduction of Mn 4+ to Mn 3+ and (2) disproportionation of Mn 3+ to Mn 4+ and Mn 2+

PAGE 172

172 Figure C 7 Logarithm of the equilibrium constants of oxygen vacancy formation (K 1 ) and disproportionation (K 2 ) versus inverse temperature for LCMA6464. Markers represent the extracted value at a single temperature and the solid lines represent linear fits.

PAGE 173

173 Figure C 8. Sola r to fuel energy conversion efficiency for H 2 O splitting versus operating temperature for isothermal redox cycles using ceria. The reduction p O 2 was 10 6 atm and the geometric concentration ratio was 3000. The gas to gas heat recovery effectiveness was var ied from 0.80 to 0.95. The reduction p O 2 was controlled via (a) inert gas sweeping, (b) a mechanical vacuum pump, (c) an electrochemical pump, or (d) a thermochemical oxygen pump.

PAGE 174

174 Figure C 9. Solar to fuel energy conversion efficiency for H 2 O splittin g versus temperature swing for nonisothermal redox cycles using ceria. The reduction p O 2 was 10 6 atm, zero solid to solid heat recovery was considered, and the geometric concentration ratio was 3000. The gas to gas heat recovery effectiveness was varied f rom 0.80 to 0.95. The reduction p O 2 was controlled via (a) inert gas sweeping, (b) a mechanical vacuum pump, (c) an electrochemical pump, or (d) a thermochemical oxygen pump.

PAGE 175

175 LIST OF REFERENCES [1] Parida, B., Iniyan, S., and Goic, R ., 2011, "A review of solar photovoltaic technologies," Renewable and sustainable energy reviews, 15(3), pp. 1625 1636. [2] Steinfeld, A., 2005, "Solar thermochemical production of hydrogen a review," Solar energy, 78(5), pp. 603 615. [3] Duffie, J. A., and Beckman, W. A., 1980, "Solar engineering of thermal processes." [4] Cook, T. R., Dogutan, D. K., Reece, S. Y., Surendranath, Y., Teets, T. S., and Nocera, D. G., 2010, "Solar energy supply and storage for the legacy and nonlegacy worlds," Chemical re views, 110(11), pp. 6474 6502. [5] Kreith, F., and Kreider, J. F., 1978, "Principles of solar engineering." [6] Graves, C., Ebbesen, S. D., Mogensen, M., and Lackner, K. S., 2011, "Sustainable hydrocarbon fuels by recycling CO 2 and H 2 O with renewable or nuclear energy," Renewable and Sustainable Energy Reviews, 15(1), pp. 1 23. [7] Lewis, N. S., 2016, "Research opportunities to advance solar energy utilization," Science, 351(6271), p. aad1920. [8] McKone, J. R., Lewis, N. S., and Gray, H. B., 2013, Will solar driven water splitting devices see the light of day?," Chemistry of Materials, 26(1), pp. 407 414. [9] Steinfeld, A., and Palumbo, R., 2001, "Solar thermochemical process technology," Encyclopedia of physical science and technology, 15(1), pp. 237 256. [10] Tamaura, Y., Steinfeld, A., Kuhn, P., and Ehrensberger, K., 1995, "Production of solar hydrogen by a novel, 2 step, water splitting thermochemical cycle," Energy, 20(4), pp. 325 330. [11] Steinfeld, A., 2002, "Solar hydrogen production via a two step water splitting thermochemical cycle based on Zn/ZnO redox reactions," International Journal of Hydrogen Energy, 27(6), pp. 611 619. [12] Chueh, W. C., and Haile, S. M., 2009, "Ceria as a thermochemical reaction medium for selectively generatin g syngas or methane from H2O and CO2," ChemSusChem, 2(8), pp. 735 739. [13] Muhich, C. L., Evanko, B. W., Weston, K. C., Lichty, P., Liang, X., Martinek, J., Musgrave, C. B., and Weimer, A. W., 2013, "Efficient generation of H2 by splitting water with an isothermal redox cycle," Science, 341(6145), pp. 540 542. [14] Furler, P., Scheffe, J., Marxer, D., Gorbar, M., Bonk, A., Vogt, U., and Steinfeld, A., 2014, "Thermochemical CO 2 splitting via redox cycling of ceria reticulated foam structures with dual sc ale porosities," Physical Chemistry Chemical Physics, 16(22), pp. 10503 10511.

PAGE 176

176 [15] Scheffe, J. R., and Steinfeld, A., 2014, "Oxygen exchange materials for solar thermochemical splitting of H 2 O and CO 2: a review," Materials Today, 17(7), pp. 341 348. [ 16] Panlener, R., Blumenthal, R., and Garnier, J., 1975, "A thermodynamic study of nonstoichiometric cerium dioxide," Journal of Physics and Chemistry of Solids, 36(11), pp. 1213 1222. [17] Nakamura, T., 1977, "Hydrogen production from water utilizing sol ar heat at high temperatures," Solar energy, 19(5), pp. 467 475. [18] Miller, J. E., McDaniel, A. H., and Allendorf, M. D., 2014, "Considerations in the Design of Materials for Solar Driven Fuel Production Using Metal Oxide Thermochemical Cycles," Advanced Energy Materials, 4(2). [19] Marxer, D., Furler, P., Scheffe, J., Geerlings, H., Falter, C., Batteiger, V., Sizmann, A., and Steinfeld, A., 2015, "Demonstration of the entire produ ction chain to renewable kerosene via solar thermochemical splitting of H2O and CO2," Energy & Fuels, 29(5), pp. 3241 3250. [20] Siegel, N. P., Miller, J. E., Ermanoski, I., Diver, R. B., and Stechel, E. B., 2013, "Factors affecting the efficiency of sola r driven metal oxide thermochemical cycles," Industrial & Engineering Chemistry Research, 52(9), pp. 3276 3286. [21] Marxer, D., Furler, P., Takacs, M., and Steinfeld, A., 2017, "Solar thermochemical splitting of CO2 into separate streams of CO and O2 wit h high selectivity, stability, conversion, and efficiency," Energy & Environmental Science. [22] Ackermann, S., Scheffe, J. R., and Steinfeld, A., 2014, "Diffusion of oxygen in ceria at elevated temperatures and its application to H2O/CO2 splitting thermo chemical redox cycles," The Journal of Physical Chemistry C, 118(10), pp. 5216 5225. [23] Chueh, W. C., and Haile, S. M., 2010, "A thermochemical study of ceria: exploiting an old material for new modes of energy conversion and CO2 mitigation," Philosophi cal Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 368(1923), pp. 3269 3294. [24] Ackermann, S., Sauvin, L., Castiglioni, R., Rupp, J. L., Scheffe, J. R., and Steinfeld, A., 2015, "Kinetics of CO2 reduction over nonstoichiometric ceria," The Journal of Physical Chemistry C, 119(29), pp. 16452 16461. [25] Furler, P., Scheffe, J., Gorbar, M., Moes, L., Vogt, U., and Steinfeld, A., 2012, "Solar thermochemical CO2 splitting utilizing a reticulated porous ceria redox system," Energy & Fuels, 26(11), pp. 7051 7059. [26] Mogensen, M., Sammes, N. M., and Tompsett, G. A., 2000, "Physical, chemical and electrochemical properties of pure and doped ceria," Solid State Ionics, 129(1), pp. 63 94.

PAGE 177

177 [27] Sun, C., Li, H., an d Chen, L., 2012, "Nanostructured ceria based materials: synthesis, properties, and applications," Energy & Environmental Science, 5(9), pp. 8475 8505. [28] Takacs, M., Scheffe, J., and Steinfeld, A., 2015, "Oxygen nonstoichiometry and thermodynamic chara cterization of Zr doped ceria in the 1573 1773 K temperature range," Physical Chemistry Chemical Physics, 17(12), pp. 7813 7822. [29] Furler, P., Scheffe, J. R., and Steinfeld, A., 2012, "Syngas production by simultaneous splitting of H2O and CO 2 via cer ia redox reactions in a high temperature solar reactor," Energy & Environmental Science, 5(3), pp. 6098 6103. [30] Abanades, S., Charvin, P., Lemont, F., and Flamant, G., 2008, "Novel two step SnO 2/SnO water splitting cycle for solar thermochemical produ ction of hydrogen," International Journal of Hydrogen Energy, 33(21), pp. 6021 6030. [31] Alxneit, I., 2008, "Assessing the feasibility of separating a stoichiometric mixture of zinc vapor and oxygen by a fast quench Model calculations," Solar Energy, 82( 11), pp. 959 964. [32] Charvin, P., Abanades, S., Flamant, G., and Lemort, F., 2007, "Two step water splitting thermochemical cycle based on iron oxide redox pair for solar hydrogen production," Energy, 32(7), pp. 1124 1133. [33] Gokon, N., Mizuno, T., N akamuro, Y., and Kodama, T., 2008, "Iron containing yttria stabilized zirconia system for two step thermochemical water splitting," Journal of Solar Energy Engineering, 130(1), p. 011018. [34] Kodama, T., Kondoh, Y., Yamamoto, R., Andou, H., and Satou, N. 2005, "Thermochemical hydrogen production by a redox system of ZrO 2 supported Co (II) ferrite," Solar Energy, 78(5), pp. 623 631. [35] Scheffe, J. R., McDaniel, A. H., Allendorf, M. D., and Weimer, A. W., 2013, "Kinetics and mechanism of solar thermoch emical H 2 production by oxidation of a cobalt ferrite zirconia composite," Energy & Environmental Science, 6(3), pp. 963 973. [36] Allendorf, M. D., Diver, R. B., Siegel, N. P., and Miller, J. E., 2008, "Two step water splitting using mixed metal ferrite s: thermodynamic analysis and characterization of synthesized materials," Energy & Fuels, 22(6), pp. 4115 4124. [37] Scheffe, J. R., Allendorf, M. D., Coker, E. N., Jacobs, B. W., McDaniel, A. H., and Weimer, A. W., 2011, "Hydrogen production via chemical looping redox cycles using atomic layer deposition synthesized iron oxide and cobalt ferrites," Chemistry of Materials, 23(8), pp. 2030 2038. [38] Scheffe, J. R., Li, J., and Weimer, A. W., 2010, "A spinel ferrite/hercynite water splitting redox cycle," International Journal of Hydrogen Energy, 35(8), pp. 3333 3340.

PAGE 178

178 [39] Coker, E. N., Ambrosini, A., Rodriguez, M. A., and Miller, J. E., 2011, "Ferrite YSZ composites for solar thermochemical production of synthetic fuels: in operando characterization of CO2 reduction," Journal of Materials Chemistry, 21(29), pp. 10767 10776. [40] Kodama, T., Nakamuro, Y., and Mizuno, T., 2006, "A two step thermochemical water splitting by iron oxide on stabilized zirconia," Journal of Solar Energy Engineering, 128(1), pp. 3 7. [41] Gokon, N., Murayama, H., Umeda, J., Hatamachi, T., and Kodama, T., 2009, "Monoclinic zirconia supported Fe 3 O 4 for the two step water splitting thermochemical cycle at high thermal reduction temperatures of 1400 1600 C," international journal o f hydrogen energy, 34(3), pp. 1208 1217. [42] Chueh, W. C., Falter, C., Abbott, M., Scipio, D., Furler, P., Haile, S. M., and Steinfeld, A., 2010, "High flux solar driven thermochemical dissociation of CO2 and H2O using nonstoichiometric ceria," Science, 330(6012), pp. 1797 1801. [43] Ezbiri, M ., Takacs, M., Theiler, D., Michalsky, R., and Steinfeld, A., 2017, "Tunable thermodynamic activity of La x Sr 1 x Mn y Al 1 y O 3 solar thermochemical fuel synthesis," Journal of Materials Chemistry A. [44] Takacs, M. Hoes, M., Caduff, M., Cooper, T., Scheffe, J., and Steinfeld, A., 2016, "Oxygen nonstoichiometry, defect equilibria, and thermodynamic characterization of LaMnO 3 perovskites with Ca/Sr A site and Al B site doping," Acta Materialia, 103, pp. 700 710. [4 5] Yang, C. K., Yamazaki, Y., Aydin, A., and Haile, S. M., 2014, "Thermodynamic and kinetic assessments of strontium doped lanthanum manganite perovskites for two step thermochemical water splitting," Journal of Materials Chemistry A, 2(33), pp. 13612 1362 3. [46] Demont, A., Abanades, S. p., and Beche, E., 2014, "Investigation of perovskite structures as oxygen exchange redox materials for hydrogen production from thermochemical two step water splitting cycles," The Journal of Physical Chemistry C, 118(24) pp. 12682 12692. [47] Scheffe, J. R., Weibel, D., and Steinfeld, A., 2013, "Lanthanum strontium manganese perovskites as redox materials for solar thermochemical splitting of H2O and CO2," Energy & Fuels, 27(8), pp. 4250 4257. [48] Dieckmann, R., and S chmalzried, H., 1977, "Defects and cation diffusion in magnetite (I)," Berichte der Bunsengesellschaft fr physikalische Chemie, 81(3), pp. 344 347. [49] Loutzenhiser, P., Glvez, M., Hischier, I., Stamatiou, A., Frei, A., and Steinfeld, A., 2009, "CO2 sp litting via two step solar thermochemical cycles with Zn/ZnO and FeO/Fe3O4 redox reactions II: Kinetic analysis," Energy & Fuels, 23(5), pp. 2832 2839.

PAGE 179

179 [50] VanHandel, G., and Blumenthal, R., 1974, "The Temperature and Oxygen Pressure Dependence of the Io Electrochemical Society, 121(9), pp. 1198 1202. [51] Knoblauch, N., Drrer, L., Fielitz, P., Schmcker, M., and Borchardt, G., 2015, "Surface controlled reduction kinetics of nominally undoped polycrystalline CeO 2," Physical Chemistry Chemical Physics, 17(8), pp. 5849 5860. [52] Scheffe, J. R., Welte, M., and Steinfeld, A., 2014, "Thermal reduction of ceria within an aerosol reactor for H2O and CO2 splitting," Industrial & Engineering Chemistry Research, 53(6), pp. 2175 2182. [53] Welte, M., Barhoumi, R., Zbinden, A., Scheffe, J. R., and Steinfeld, A., 2016, "Experimental demonstration of the thermochemical reduction of ceria in a solar aerosol reactor," Industrial & Engineering Chemis try Research, 55(40), pp. 10618 10625. [54] Ko, R., Ricken, M., No, J., and Riess, I., 1989, "Phase transformations in reduced ceria: Determination by thermal expansion measurements," Journal of Solid State Chemistry, 78(1), pp. 136 147. [55] Srensen, O T., 1976, "Thermodynamic studies of the phase relationships of nonstoichiometric cerium oxides at higher temperatures," Journal of Solid State Chemistry, 18(3), pp. 217 233. [56] Bishop, S., Duncan, K., and Wachsman, E., 2009, "Defect equilibria and che mical expansion in non stoichiometric undoped and gadolinium doped cerium oxide," Electrochimica Acta, 54(5), pp. 1436 1443. [57] Meng, Q. L., Lee, C. i., Ishihara, T., Kaneko, H., and Tamaura, Y., 2011, "Reactivity of CeO 2 based ceramics for solar hydro gen production via a two step water splitting cycle with concentrated solar energy," international journal of hydrogen energy, 36(21), pp. 13435 13441. [58] Scheffe, J. R., and Steinfeld, A., 2012, "Thermodynamic analysis of cerium based oxides for solar thermochemical fuel production," Energy & Fuels, 26(3), pp. 1928 1936. [59] Meng, Q. L., Lee, C. i., Shigeta, S., Kaneko, H., and Tamaura, Y., 2012, "Solar hydrogen step water splitting cycle," Journal of Solid State Chemistry, 194, pp. 343 351. [60] Le Gal, A., and Abanades, S. p., 2012, "Dopant incorporation in ceria for enhanced water splitting activity during solar thermochemical hydrogen generation," The Journal of Physica l Chemistry C, 116(25), pp. 13516 13523. [61] Scheffe, J. R., Jacot, R., Patzke, G. R., and Steinfeld, A., 2013, "Synthesis, characterization, and thermochemical redox performance of Hf4+, Zr4+, and Sc3+ doped ceria for splitting CO2," The Journal of Phys ical Chemistry C, 117(46), pp. 24104 24114.

PAGE 180

180 doped ceria based mixed oxides for solar thermochemical hydrogen generation via two step water splitting cycles," Energy & fuels, 27(10), pp. 6068 6078. [63] Meng, Q. L., and Tamaura, Y., 2014, "Enhanced hydrogen production by doping Pr into Ce 0.9 Hf 0.1 O 2 for thermochemical two step water splitting cycle," Journal of Physics and Chemistry of Solids, 75(3), pp. 328 333. [64] Le Gal, A., Abanades, S., and Fl amant, G., 2011, "CO2 and H2O splitting for thermochemical production of solar fuels using nonstoichiometric ceria and ceria/zirconia solid solutions," Energy & Fuels, 25(10), pp. 4836 4845. [65] Meng, Q. L., Lee, C. i., Kaneko, H., and Tamaura, Y., 2012, "Solar thermochemical process for hydrogen production via two reaction," Thermochimica Acta, 532, pp. 134 138. [66] Hao, Y., Yang, C. K., and Haile, S. M., 2014, "Ceria Zirconia Solid Solution s (Ce1 x Zr x of Materials, 26(20), pp. 6073 6082. Paal, R., 2015, "Ceria doped with zirconium and la nthanide oxides to enhance solar thermochemical production of fuels," The Journal of Physical Chemistry C, 119(13), pp. 6929 6938. [68] Bulfin, B., Call, F., Vieten, J., Roeb, M., Sattler, C., and Shvets, I., 2016, "Oxidation and reduction reaction kineti cs of mixed cerium zirconium oxides," The Journal of Physical Chemistry C, 120(4), pp. 2027 2035. [69] Takacs, M., Ackermann, S., Bonk, A., Neises von Puttkamer, M., Haueter, P., Scheffe, J., Vogt, U., and Steinfeld, A., 2016, "Splitting CO2 with a ceria based redox cycle in a solar driven thermogravimetric analyzer," AIChE Journal. [70] Bulfin, B., Hoffmann, L., de Oliveira, L., Knoblauch, N., Call, F., Roeb, M., Sattler, C., and Schmuecker, M., 2016, "Statistical thermodynamics of non stoichiometric cer ia and ceria zirconia solid solutions," Physical Chemistry Chemical Physics, 18(33), pp. 23147 23154. [71] Hathaway, B. J., Bala Chandran, R., Gladen, A. C., Chase, T. R., and Davidson, J. H., 2016, "Demonstration of a solar reactor for carbon dioxide spl itting via the isothermal ceria redox cycle and practical implications," Energy & Fuels, 30(8), pp. 6654 6661. step solar thermochemical non stoichiometric redox cycles with heat r ecovery," Energy, 37(1), pp. 591 600.

PAGE 181

181 analysis of isothermal redox cycling of ceria for solar fuel production," Energy & Fuels, 27(9), pp. 5533 5544. [74] Hao, Y., Yang, C. K., and Haile, S. M., 2013, "High temperature isothermal chemical cycling for solar driven fuel production," Physical Chemistry Chemical Physics, 15(40), pp. 17084 17092. [75] Ermanoski, I., Miller, J., and Allendorf, M., 2014, "Efficiency maximi zation in solar thermochemical fuel production: challenging the concept of isothermal water splitting," Physical Chemistry Chemical Physics, 16(18), pp. 8418 8427. [76] Lin, M., and Haussener, S., 2015, "Solar fuel processing efficiency for ceria redox cy cling using alternative oxygen partial pressure reduction methods," Energy, 88, pp. 667 679. [77] Bulfin, B., Lange, M., de Oliveira, L., Roeb, M., and Sattler, C., 2016, "Solar thermochemical hydrogen production using ceria zirconia solid solutions: Effi ciency analysis," International Journal of Hydrogen Energy, 41(42), pp. 19320 19328. [78] Bulfin, B., Call, F., Lange, M., Lubben, O., Sattler, C., Pitz Paal, R., and Shvets, I., 2015, "Thermodynamics of CeO2 thermochemical fuel production," Energy & Fuels 29(2), pp. 1001 1009. [79] Brendelberger, S., Roeb, M., Lange, M., and Sattler, C., 2015, "Counter flow sweep gas demand for the ceria redox cycle," Solar Energy, 122, pp. 1011 1022. [80] Abanades, S., and Flamant, G., 2006, "Thermochemical hydrogen prod uction from a two step solar driven water splitting cycle based on cerium oxides," Solar Energy, 80(12), pp. 1611 1623. [81] Bader, R., Chandran, R. B., Venstrom, L. J., Sedler, S. J., Krenzke, P. T., De Smith, R. M., Banerjee, A., Chase, T. R., Davidson, to Split CO2 Via Isothermal Redox Cycling of Ceria," Journal of Solar Energy Engineering, 137(3), p. 031007. [82] Hendriksen, P. V., Larsen, P. H., Mogensen, M., Poulsen, F. W., and Wiik, K., 2000, "Prospects and problems of dense oxygen permeable membranes," Catalysis Today, 56(1), pp. 283 295. [83] Muecke, U. P., Beckel, D., Bernard, A., Bieberle Htter, A., Graf, S., Infortuna, A., Mller, P., Rupp, J. L., Schneider, J., and Gauckler, L. J., 2008, "Micro solid oxide fuel cells on glass ceramic substrates," Advanced Functional Materials, 18(20), pp. 3158 3168. [84] Suntivich, J., Gasteiger, H. A., Yabuuchi, N., Nakanishi, H., Goodenough, J. B., and Shao Horn, Y., 2011, "Design principles for oxygen reduction activity on perovskite oxide catalysts for fuel cells and metal air batteries," Nature chemistry, 3(7), pp. 546 550.

PAGE 182

182 [85] Cooper, T., Scheffe, J. R., Galvez, M. E., Jacot, R., Patzke, G., and Steinfeld, A., 2015, "Lanthanum Manganite Perovskites with Ca/Sr A site and Al B site Doping as Effective Oxygen Exchange Materials for Solar Thermochemical Fuel Production," Energy Technology, 3(11), pp. 1130 1142. [86] Demont, A., and Abanades, S., 2014, "High redox activity of Sr substituted lanthanum mang anite perovskites for two step thermochemical dissociation of CO 2," RSC Advances, 4(97), pp. 54885 54891. [87] Demont, A., and Abanades, S., 2015, "Solar thermochemical conversion of CO 2 into fuel via two step redox cycling of non stoichiometric Mn conta ining perovskite oxides," Journal of Materials Chemistry A, 3(7), pp. 3536 3546. [88] Dey, S., Naidu, B., and Rao, C., 2015, "Ln0. 5A0. 5MnO3 (Ln= lanthanide, A= Ca, Sr) perovskites exhibiting remarkable performance in the thermochemical generation of CO a nd H2 from CO2 and H2O," Chemistry A European Journal, 21(19), pp. 7077 7081. Ca x MnO 3 perovskites in generating H 2 and CO by the thermochemical splitting of H 2 O and CO 2," Physical Chemistry Chemical Physics, 17(1), pp. 122 125. [90] Dey, S., and Rao, C. N. R., 2016, "Splitting of CO2 by Manganite Perovskites to Generate CO by Solar Isothermal Redox Cycling," ACS Energy Letters, 1(1), pp. 237 243. [91] Jiang Q., Tong, J., Zhou, G., Jiang, Z., Li, Z., and Li, C., 2014, "Thermochemical CO 2 x, y 1) perovskite oxides," Solar Energy, 103, pp. 425 437. [92] McDaniel, A. H., Miller, E. C., Arifin, D., Ambrosini, A., Coker, E. N., O'Hayre, R., Chueh, W. C., and Tong, J., 2013, "Sr and Mn production," Energy & Environmental Science, 6(8), pp. 2424 2428. [93] Ishihara, T., 2009, Perov skite oxide for solid oxide fuel cells, Springer Science & Business Media. thermochemical fuel production: strategies to lo wer the operation temperature," Journal of Materials Chemistry A, 3(30), pp. 15546 15557. [95] Nowotny, J., and Rekas, M., 1998, "Defect chemistry of (La, Sr) MnO3," Journal of the American Ceramic Society, 81(1), pp. 67 80. B., 2014, "Tunable Oxygen Vacancy Formation Energetics in the Complex Perovskite Oxide Sr x La1 x Mn y Al1 y O3," Chemistry of Materials, 26(22), pp. 6595 6602.

PAGE 183

183 [97] Ezbiri, M., Becattini, V., H oes, M., Michalsky, R., and Steinfeld, A., 2017, "High redox capacity of Al doped La1 xSrxMnO3 enriched surfaces," ChemSusChem. [98] Kotomin, E., Mastrikov, Y. A., Kuklja, M., Merkle, R., Roytburd, A. and Maier, J., 2011, "First principles calculations of oxygen vacancy formation and migration in mixed conducting 5. [99] Merkle, R., Mastrikov, Y. A., Kotomin, E. A., Kukl ja, M. M., and Maier, J., 2011, "First B226. [100] Mastrikov, Y. A., Merkle, R., Kotomin, E. A., Kuklja, M. M., and Maier, J., 2013, Physical chemistry chemical ph ysics, 15(3), pp. 911 918. [101] Wu, Q. H., Liu, M., and Jaegermann, W., 2005, "X ray photoelectron spectroscopy of La 0.5 Sr 0.5 MnO 3," Materials Letters, 59(16), pp. 1980 1983. [102] Galvez, M., Jacot, R., Scheffe, J., Cooper, T., Patzke, G., and Steinf eld, A., 2015, "Physico chemical changes in Ca, Sr and Al doped La Mn O perovskites upon thermochemical splitting of CO 2 via redox cycling," Physical Chemistry Chemical Physics, 17(9), pp. 6629 6634. [103] Babiniec, S. M., Coker, E. N., Miller, J. E., and Ambrosini, A., 2015, "Investigation of storage media," Solar Energy, 118, pp. 451 459. [104] Hoes, M., Muhich, C. L., Jacot, R., Patzke, G., and Steinfeld, A., 2017, "Ther modynamics of paired charge compensating doped ceria with superior redox performance for solar thermochemical splitting of H2O and CO2," Journal of Materials Chemistry A. [105] Muhich, C., Hoes, M., and Steinfeld, A., 2018, "Mimicking tetravalent dopant be havior using paired charge compensating dopants to improve the redox performance of ceria for thermochemically splitting H2O and CO2," Acta Materialia, 144, pp. 728 737. [106] Zhai, S., Rojas, J., Ahlborg, N., Lim, K., Toney, M. F., Jin, H., Chueh, W. C., and Majumdar, A., 2018, "The use of poly cation oxides to lower the temperature of two step thermochemical water splitting," Energy & Environmental Science. [107] Rost, C. M., Sachet, E., Borman, T., Moballegh, A., Dickey, E. C., Hou, D., Jones, J. L., Cur tarolo, S., and Maria, J. P., 2015, "Entropy stabilized oxides," Nature communications, 6, p. 8485.

PAGE 184

184 [108] Miller, J. E., Ambrosini, A., Coker, E. N., Allendorf, M., and McDaniel, A., 2014, "Advancing oxide materials for thermochemical production of solar f uels," Energy Procedia, 49, pp. 2019 2026. [109] Smith, J. M., 1975, "Introduction to chemical engineering thermodynamics," Rensselaer Polytechnic Institute. [110] Bale, C. W., Blisle, E., Chartrand, P., Decterov, S. A., Eriksson, G., Gheribi, A. E., Hack K., Jung, I. H., Kang, Y. B., Melanon, J., Pelton, A. D., Petersen, S., Robelin, C., Sangster, J., and Van Ende, M. A., 2016, "FactSage Thermochemical Software and Databases, 2010 2016," Calphad, pp. 35 53. [111] Linstrom, P. J., and Mallard, W., 2001, "NIST Chemistry webbook; NIST standard reference database No. 69." [112] Venstrom, L. J., Petkovich, N., Rudisill, S., Stein, A., and Davidson, J. H., 2012, "The effects of morphology on the oxidation of ceria by water and carbon dioxide," Journal of solar energy engineering, 134(1), p. 011005. [113] Rudisill, S. G., Venstrom, L. J., Petkovich, N. D., Quan, T., Hein, N., Boman, D. B., Davidson, J. H., and Stein, A., 2013, "Enhanced oxidation kinetics in thermochemical cycling of CeO2 through templated poros ity," The Journal of Physical Chemistry C, 117(4), pp. 1692 1700. [114] Petkovich, N. D., Rudisill, S. G., Venstrom, L. J., Boman, D. B., Davidson, J. H., and Stein, A., 2011, "Control of heterogeneity in nanostructured ce1 x zr x o2 binary oxides for enha nced thermal stability and water splitting activity," The Journal of Physical Chemistry C, 115(43), pp. 21022 21033. [115] Gladen, A. C., and Davidson, J. H., 2016, "The morphological stability and fuel production of commercial fibrous ceria particles for solar thermochemical redox cycling," Solar Energy, 139, pp. 524 532. [116] Abanades, S., Legal, A., Cordier, A., Peraudeau, G., Flamant, G., and Julbe, A., 2010, "Investigation of reactive cerium based oxides for H2 production by thermochemical two step wa ter splitting," Journal of Materials Science, 45(15), pp. 4163 4173. [117] Tuller, H., and Nowick, A., 1979, "Defect structure and electrical properties of nonstoichiometric CeO2 single crystals," Journal of The Electrochemical Society, 126(2), pp. 209 217 [118] Fogler, S. H., 1999, Elements of Chemical Reaction Engineering, Prentice Hall, Inc., Upper Saddle River, New Jersey 07458. [119] Zhao, Z., Uddi, M., Tsvetkov, N., Yildiz, B., and Ghoniem, A. F., 2016, "Redox Kinetics Study of Fuel Reduced Ceria for Chemical Looping Water Splitting," The Journal of Physical Chemistry C, 120(30), pp. 16271 16289.

PAGE 185

185 [120] Arifin, D., and Weimer, A. W., 2018, "Kinetics and mechanism of solar thermochemical H 2 and CO production by oxidation of reduced CeO 2," Solar Energy 160, pp. 178 185. [121] Levenspiel, O., 1999, Chemical Reaction Engineering ," Wiley, New York. [122] McDaniel, A. H., 2017, "Renewable energy carriers derived from concentrating solar power and nonstoichiometric oxides," Current Opinion in Green and Sus tainable Chemistry, 4, pp. 37 43. [123] Jiang, Q., Zhou, G., Jiang, Z., and Li, C., 2014, "Thermochemical CO2 splitting reaction Solar Energy, 99, pp. 55 66. [124] Ramos F ernandez, E. V., Shiju, N. R., and Rothenberg, G., 2014, "Understanding the solar driven reduction of CO 2 on doped ceria," RSC Advances, 4(32), pp. 16456 16463. [125] Kaneko, H., Miura, T., Ishihara, H., Taku, S., Yokoyama, T., Nakajima, H., and Tamaura, Y., 2007, "Reactive ceramics of CeO2 MOx (M= Mn, Fe, Ni, Cu) for H2 generation by two step water splitting using concentrated solar thermal energy," Energy, 32(5), pp. 656 663. [126] Kuhn, M., Bishop, S., Rupp, J., and Tuller, H., 2013, "Structural charact erization and oxygen nonstoichiometry of ceria 61(11), pp. 4277 4288. [127] Nair, M. M., and Abanades, S., 2016, "Insights into the Redox Performance of Non stoichiometric Lanthanum Manganite P erovskites for Solar Thermochemical CO2 Splitting," ChemistrySelect, 1(15), pp. 4449 4457. [128] Rodenbough, P. P., and Chan, S. W., 2018, "Thermal oxygen exchange cycles in mixed manganese perovskites," Ceramics International, 44(2), pp. 1343 1347. [129] Arulraj, A., Santhosh, P., Gopalan, R. S., Guha, A., Raychaudhuri, A., Kumar, N., and Rao, C., 1998, "Charge ordering in the rare earth manganates: the origin of the extraordinary sensitivity to the average radius of the A site cations," Journal of Physics : Condensed Matter, 10(38), p. 8497. [130] Woodward, P., Vogt, T., Cox, D., Arulraj, A., Rao, C., Karen, P., and Cheetham, A., 1998, "Influence of cation size on the structural features of Ln1/2A1/2MnO3 perovskites at room temperature," Chemistry of Materi als, 10(11), pp. 3652 3665. [131] Pechini, M. P., 1967, "Method of preparing lead and alkaline earth titanates and niobates and coating method using the same to form a capacitor," Google Patents. [132] Han, Y. S., and Kim, H. G., 2000, "Synthesis of LiMn2O 4 by modified Pechini method and characterization as a cathode for rechargeable Li/LiMn2O4 cells," Journal of power sources, 88(2), pp. 161 168.

PAGE 186

186 [133] Huang, C. Y., and Huang, T. MnO 3 properties f or SOFC cathode material," Journal of materials science, 37(21), pp. 4581 4587. a promising perovskite type oxide for solar thermochemica l hydrogen production," Energy & Environmental Science. [135] Thakur, R. K., Thakur, R., Shanmukharao Samatham, S., Kaurav, N., Ganesan, V., Gaur, xSrxMnO3 (x= 0.1 and 0.2) ," Journal of Applied Physics, 112(10), p. 104115. [136] Krger, F., and Vink, H., 1956, "Relations between the concentrations of imperfections in crystalline solids," Solid state physics, Elsevier, pp. 307 435. [137] Van Roosmalen, J., and Cordfunke, E., 1991, "A new defect model to describe the oxygen deficiency in perovskite type oxides," Journal of Solid State Chemistry, 93(1), pp. 212 219. [138] Mizusaki, J., Mori, N., Takai, H., Yonemura, Y., Minamiue, H., Tagawa, H., Dokiya, M., Inaba, H., Naraya, K. and Sasamoto, T., 2000, "Oxygen nonstoichiometry and defect equilibrium in the perovskite 163 177. [139] Tou, M., Michalsky, R., and Steinfeld, A., 2017, "Solar driven thermochemical spl itting of CO2 and in situ separation of CO and O2 across a ceria redox membrane reactor," Joule, 1(1), pp. 146 154. [140] Carrillo, A. J., Bork, A. H., Moser, T., Sediva, E., Hood, Z. D., and Rupp, J. L., 2019, "Modifying La0. 6Sr0. 4MnO3 Perovskites with Cr Incorporation for Fast Isothermal CO2 Splitting Kinetics in Solar Driven Thermochemical Cycles," Advanced Energy Materials, p. 1803886. [141] Ehrhart, B. D., Muhich, C. L., Al Shankiti, I., and Weimer, A. W., 2016, "System efficiency for two step metal oxide solar thermochemical hydrogen production Part 1: Thermodynamic model and impact of oxidation kinetics," International Journal of Hydrogen Energy, 41(44), pp. 19881 19893. [142] Krenzke, P. T., and Davidson, J. H., 2015, "On the efficiency of solar H2 and CO production via the thermochemical cerium oxide redox cycle: the option of inert swept reduction," Energy & Fuels, 29(2), pp. 1045 1054. [143] Al Shankiti, I., Ehrhart, B. D., and Weimer, A. W., 2017, "Isothermal redox for H2O and CO2 splitting A re view and perspective," Solar Energy, 156, pp. 21 29. [144] Ehrhart, B. D., Muhich, C. L., Al Shankiti, I., and Weimer, A. W., 2016, "System efficiency for two step metal oxide solar thermochemical hydrogen production Part 3: Various

PAGE 187

187 methods for achieving l ow oxygen partial pressures in the reduction reaction," International Journal of Hydrogen Energy, 41(44), pp. 19904 19914. [145] Hao, Y., Jin, J., and Jin, H., 2019, "Thermodynamic Analysis of Isothermal CO2 Splitting and CO2 H2O Co Splitting for Solar Fue l Production," Applied Thermal Engineering. [146] Bork, A. H., Povoden Karadeniz, E., and Rupp, J. L., 2017, "Modeling Thermochemical Solar to Fuel Conversion: CALPHAD for Thermodynamic Assessment Studies of Perovskites, Exemplified for (La, Sr) MnO3," Adv anced Energy Materials, 7(1). [147] Muhich, C. L., Blaser, S., Hoes, M. C., and Steinfeld, A., 2018, "Comparing the solar to fuel energy conversion efficiency of ceria and perovskite based thermochemical redox cycles for splitting H2O and CO2," internation al journal of hydrogen energy, 43(41), pp. 18814 18831. [148] Warren, K. J., Reim, J., Randhir, K., Greek, B., Carrillo, R., Hahn, D. W., and Scheffe, J. R., 2017, "Theoretical and experimental investigation of solar methane reforming through the nonstoich iometric ceria redox cycle," Energy Technology, 5(11), pp. 2138 2149. [149] Brendelberger, S., von Storch, H., Bulfin, B., and Sattler, C., 2017, "Vacuum pumping options for application in solar thermochemical redox cycles assessment of mechanical jet an d thermochemical pumping systems," Solar Energy, 141, pp. 91 102. [150] Zhou, W., Shao, Z., Ran, R., Chen, Z., Zeng, P., Gu, H., Jin, W., and Xu, N., 2007, "High performance electrode for electrochemical oxygen generator cell based on solid electrolyte ion transport membrane," Electrochimica acta, 52(22), pp. 6297 6303. [151] Narasimhaiah, G., and Janardhanan, V. M., 2013, "Modeling CO 2 electrolysis in solid oxide electrolysis cell," Journal of Solid State Electrochemistry, 17(8), pp. 2361 2370. [152] Ezbi ri, M., Allen, K. M., Glvez, M. E., Michalsky, R., and Steinfeld, A., 2015, "Design principles of perovskites for thermochemical oxygen separation," ChemSusChem, 8(11), pp. 1966 1971. [153] Kawabe, M., Ono, H., Sano, T., Tsuji, M., and Tamaura, Y., 1997, "Thermochemical oxygen pump with praseodymium oxides using a temperature swing at 403 873 K," Energy, 22(11), pp. 1041 1049. [154] Brendelberger, S., Vieten, J., Vidyasagar, M. J., Roeb, M., and Sattler, C., 2018, "Demonstration of thermochemical oxygen pu mping for atmosphere control in reduction reactions," Solar Energy, 170, pp. 273 279. [155] Brendelberger, S., Vieten, J., Roeb, M., and Sattler, C., 2019, "Thermochemical oxygen pumping for improved hydrogen production in solar redox cycles," Internationa l Journal of Hydrogen Energy.

PAGE 188

188 [156] Block, T., and Schmcker, M., 2016, "Metal oxides for thermochemical energy storage: A comparison of several metal oxide systems," Solar Energy, 126, pp. 195 207. [157] Ricken, M., Nlting, J., and Riess, I., 1984, "Spec ific heat and phase diagram of 99. [158] Szewczyk, A., Gutowska, M., and Dabrowski, B., 2005, "Specific heat and phase diagram x 1.0)," Physical Review B, 72(22), p. 224429.

PAGE 189

189 BIOGRAPHICAL SKETCH Richard Joseph Carrillo was born in Cape Canaveral, Florida, and was reared in Titusville, Florida. He attended Titusville High School, where he graduated first in his class and earned varsity letters in basketball, cross country, and track After graduation from high school, he enrolled at the University of Central Flor ida to major in mechanical e ngineering. During his undergraduate studies, he had the opportunity to do research in the lab of Dr. Dawn Tilbury at the University of Michigan. Here, he worked on the development and application of an algorithm for semi autonomous object retrieval with a five degree of freedom robotic arm. Afterwards, he joined the Hybrid Sustainable Energy Sy stems Laboratory to work as an undergraduate research assistant under the advisement of Dr. Tuhin Das. He designed experiments and fabricated an experimental apparatus by which the phenomenon of autorotation could be studied. In May 2015, he graduated magn a cum laude with a B.S. in mechanical e ngineering and a minor in m athematics. Later that year, in August 2015, he enrolled in the Ph.D. program i n the Department of Mechanical and Aerospace Engineering at the University of Florida He joined the Renewable En ergy Conversion Laboratory to perform graduate research under the advisement of Dr. Jonathan Scheffe. Here, he studied solar thermochemical production of fuels facilitated by a two step metal oxide redox cycle. He developed a lab scale tubular reactor for use in studying the thermodynamic and kinetic properties of metal oxides and performed experimental and theoretical investigations into the fuel productivity of candidate materials. He earned a M.S. in mechanical e ngineering in May 2017 and comp leted his P h.D. in August 2019.