UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Investigation of a direct methanol redox fuel cell with design simplification Ilicic, Alan Bartol 2010

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2010_spring_ilicic_alan.pdf [ 5.36MB ]
Metadata
JSON: 24-1.0058853.json
JSON-LD: 24-1.0058853-ld.json
RDF/XML (Pretty): 24-1.0058853-rdf.xml
RDF/JSON: 24-1.0058853-rdf.json
Turtle: 24-1.0058853-turtle.txt
N-Triples: 24-1.0058853-rdf-ntriples.txt
Original Record: 24-1.0058853-source.json
Full Text
24-1.0058853-fulltext.txt
Citation
24-1.0058853.ris

Full Text

INVESTIGATION OF A DIRECT LIQUID REDOX FUEL CELL WITH DESIGN SIMPLIFICATION by ALAN BARTOL ILICIC B.A.Sc., The University of British Columbia, 2004 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF  THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (Chemical and Biological Engineering) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) April 2010 © Alan Bartol Ilicic, 2010 ii ABSTRACT A key objective of this work is to address a number of the central issues associated with the direct methanol fuel cell (DMFC) through the investigation of a redox flow battery (RFB) / DMFC hybrid fuel cell. The air cathode (Pt/carbon) of the DMFC is substituted by a Fe2+/Fe3+ redox couple cathode (carbon) with no platinum-group metal (PGM) catalyst. In this configuration, referred to as the direct liquid redox fuel cell (DLRFC), the Fe2+/Fe3+ redox couple cathode is selective to the redox couple reaction and fuel crossover does not cause cathode depolarization. A wide range of anolyte fuel concentrations were tested (2-24 M CH3OH) and the best DLRFC performance was obtained at 16.7 M CH3OH (equimolar CH3OH / H2O). A significant improvement in the DLRFC performance and catholyte charge density was obtained by switching from a sulfate-based iron salt to a perchlorate-based iron salt. This led to a greater than 150% increase in the solubility of the redox couple, greater than a 200 mV increase in the equilibrium half-cell potential of the redox couple and a greater than 200% improvement in the DLRFC peak power density (79 mW/cm² vs. 25 mW/cm²) relative to the sulfate-based system. The selective nature of the redox cathode enabled the demonstration and characterization of a novel mixed-reactant DLRFC (MR-DLRFC) where a mixed electrolyte containing methanol and the redox couple is fed to the cathode and fuel crossover is the mode of fuel supply. A non-optimized peak power density of 15 mW/cm² was obtained with this system. iii A novel in-situ redox couple regeneration approach was also demonstrated and characterized, which involved substituting the methanol anolyte by an air stream. This approach exploits the hybrid nature of the DLRFC and utilizes the PtRu catalyst at the DLRFC anode as an O2 reduction cathode during regeneration. Eliminating the use of PGM catalysts at the fuel cell cathode and enabling the use of high fuel concentrations are decisive advantages of DLRFC technology. Furthermore, the ability to extend DLRFC technology to a mixed- reactant architecture where fuel crossover is desirable paves new ground for the future of fuel cell research. iv TABLE OF CONTENTS Abstract ...............................................................................................................ii Table of Contents...............................................................................................iv List of Tables......................................................................................................vi List of Figures ...................................................................................................vii Nomenclature...................................................................................................xiv Acknowledgements ..........................................................................................xx Co-authorship Statement ................................................................................xxi 1. Introduction.....................................................................................................1 1.1 History and Overview of Fuel Cell Technology............................................1 1.2 Fundamental Principles of Fuel Cell Technology ........................................8 1.3 The Redox Flow Battery............................................................................38 1.4 The Direct Methanol Fuel Cell ...................................................................56 1.5 Mixed-Reactant Fuel Cells ........................................................................79 1.6 Research Objectives .................................................................................84 1.7 Thesis Layout ............................................................................................85 1.8 References................................................................................................87 2. High Fuel Concentration Direct Liquid Fuel Cell with a Redox Couple Cathode1 ............................................................................................................95 2.1 Introduction ...............................................................................................95 2.2 Experimental .............................................................................................99 2.3 Results and Discussion ...........................................................................103 2.4 Conclusions.............................................................................................115 2.5 Acknowledgements .................................................................................115 2.6 References..............................................................................................116 3. Improving the Performance of the Direct Liquid Redox Fuel Cell1.........119 3.1 Introduction .............................................................................................119 3.2 Experimental ...........................................................................................122 3.3 Results and Discussion ...........................................................................126 3.4 Conclusions.............................................................................................144 3.5 Acknowledgements .................................................................................145 3.6 References..............................................................................................146 4. Advancing Direct Liquid Redox Fuel Cells: Mixed Reactant and In-Situ Regeneration Opportunities1 .........................................................................149 v4.1 Introduction .............................................................................................149 4.2 Experimental ...........................................................................................153 4.3 Results and Discussion ...........................................................................159 4.4 Conclusions.............................................................................................176 4.5 Acknowledgements .................................................................................177 4.6 References..............................................................................................178 5. Conclusions ................................................................................................180 5.1 The Direct Liquid Redox Fuel Cell Approach ..........................................181 5.2 Direct Liquid Redox Fuel Cell Performance Improvements.....................188 5.3 The Mixed-Reactant DLRFC Architecture ...............................................193 5.4 Regeneration of the DLRFC Redox Couple ............................................197 5.5 Potential Applications of DLRFC Technologies.......................................201 5.6 Future Work and Recommendations.......................................................204 5.7 References..............................................................................................212 6. Appendices..................................................................................................216 Appendix A – Publications and Presentations...............................................216 Appendix B – Experimental Procedures........................................................218 Appendix C – Direct Liquid Redox Fuel Cell Thermodynamics .....................239 Appendix D – Competition Between Fe2+/Fe3+ Ions and Protons in Nafion® Membranes ...................................................................................................244 Appendix E – Repeatability and Reproducibility of the DLRFC .....................246 Appendix F – Advancements in the Direct Hydrogen Redox Fuel Cell1 ........248 vi LIST OF TABLES Table 1.1. Key properties of various fuel cells. .....................................................6 Table 1.2. Some electrochemical half-cell reactions [45]. ...................................10 Table 1.3. A list of reference electrodes [46].......................................................10 Table 1.4. Exchange current densities for various redox couples at 27°C [24]. ..26 Table 1.5. Short list of potential electrode materials and their electrical resistivities [49]. ...........................................................................................27 Table 1.6. Ionic molar conductivities at infinite dilution and 25°C for some anions and cations [50]. ..........................................................................................29 Table 1.7. List of various redox couples with their half-cell potentials at 25°C [45]. .....................................................................................................................43 Table 1.8. A list of some low molecular weight organic liquid fuels and various properties at 25°C........................................................................................62 Table 2.1. Summary of anodic and cathodic peak currents (ip) and peak potentials (Ep) for cyclic voltammograms shown in Figures 2.4a-c. ...........107 Table 3.1. List of ferric salts considered for use in the DMRFC catholyte. Half-cell potentials reported for standard conditions at 25°C. ..................................127 Table 3.2. Summary of anodic and cathodic peak current densities (ip), peak potentials (Ep) and apparent half-cell potentials (E) for cyclic voltammograms shown in Figure 3.4. All data were obtained at 90°C. ................................135 Table 4.1. Summary of anodic and cathodic peak current densities (ip) and peak potentials (Ep) for cyclic voltammograms shown in Figure 4.6...................162 Table 4.2. Summary of Nafion® 112 membrane conductivities at 25°C after membranes were exposed to various electrolytes. ....................................165 Table 6.1. Temperature coefficients and other thermodynamic properties for the DLRFC reactions at 25°C [45]. ..................................................................239 Table 6.2. Conductivity measurements of Nafion® 115 membrane before and after exposure to the iron based solution at 22°C. .....................................264 vii LIST OF FIGURES Figure 1.1. Schematic of Grove's "Gas Battery" powering an electrolyzer............2 Figure 1.2. Power output and applications of several fuel cells [44]. ....................7 Figure 1.3. An example of a fuel cell polarization curve which illustrates the various regions of a polarization curve.........................................................16 Figure 1.4. Conceptual graph illustrating activation energies of reaction intermediates during (a) standard conditions and (b) cathodic polarization. 21 Figure 1.5. A sample plot of the Butler-Erdey-Gruz-Volmer equation. ................25 Figure 1.6. Sketch of the potential distribution across the thickness of a 3-D cathode. Em and Es are the potentials of the ionic and electronic conductors, respectively..................................................................................................34 Figure 1.7. Sketch of the current density distribution across the thickness of a 3- D cathode. ...................................................................................................35 Figure 1.8. Simplified schematic of a redox flow battery. ....................................38 Figure 1.9. Images of various carbon substrates. Left to right: carbon fiber paper, carbon cloth, graphite felt [53]. Reproduced by permission of The Electrochemical Society...............................................................................40 Figure 1.10. Equilibrium cell potential of a Fe/Cr RFB vs. the state of charge at 25°C.............................................................................................................50 Figure 1.11. Schematic of a vanadium redox battery integrated with a renewable energy source. .............................................................................................55 Figure 1.12. Simplified schematic of a direct methanol fuel cell..........................58 Figure 1.13. Simplified schematic of an indirect methanol fuel cell. ....................59 Figure 1.14. Generalized reaction mechanism for methanol oxidation [76]. .......65 Figure 1.15. An illustration of a carbon supported PtRu catalyst particle. ...........66 Figure 1.16. Schematic of a cathode catalyst layer with a microporous layer.....69 Figure 1.17. The molecular structure of the Nafion® membrane. ........................70 Figure 1.18. The molecular structure of iron tetramethoxyphenylporphyrin. .......73 viii Figure 1.19. Effect of cathode humidifier temperature. (70°C cell temperature, 6 mL/min CH3OH flow rate, 2 M CH3OH, 600 sccm air flow rate, air pressure unavailable) [69]. Reprinted with permission from Elsevier..........................74 Figure 1.20. Effect of methanol flow rate. (70°C cell temperature, 1 M CH3OH, 1200 sccm air flow rate, 70°C cathode humidifier temperature, air pressure unavailable) [69]. Reprinted with permission from Elsevier..........................75 Figure 1.21. Effect of cell temperature. (3 M CH3OH, 600 sccm air flow rate, 4 mL/min methanol flow rate, 30°C cathode humidifier temperature, air pressure unavailable) [69]. Reprinted with permission from Elsevier...........77 Figure 1.22. Effect of methanol concentration. (70°C cell temperature, 6 mL/min CH3OH, 600 sccm air flow rate, 70°C air humidifier temperature, air pressure unavailable) [69]. Reprinted with permission from Elsevier..........................78 Figure 1.23. Schematic of a conventional fuel cell with two reactant streams. ...79 Figure 1.24. Schematic of a mixed-reactant fuel cell with transverse electrodes. .....................................................................................................................80 Figure 1.25. Schematic of a mixed-reactant fuel cell with co-planar electrodes..81 Figure 2.1. Schematic of the fuel cell test system.............................................102 Figure 2.2. Expanded view of the 4 cm² fuel cell. .............................................102 Figure 2.3. DSC results for mixed CH3OH/Fe 2+/Fe3+ redox with and without shavings of CFP. (50-75 mg sample mass, 1°C/min temperature ramp rate) ...................................................................................................................103 Figure 2.4a. Cyclic voltammograms of a Fe2+/Fe3+ redox electrolyte over Pt and GC. (70°C, 50 mV/s scan rate, IR-corrected).............................................105 Figure 2.5. (a) Conductivity of the Fe2+/Fe3+ redox electrolyte (0.81 M FeNH4(SO4)2, 0.09 M FeSO4, 0.5 M H2SO4). (b) Conductivity of Nafion ® 117 in deionized water. (c) Conductivity of Nafion® 117 in deionized water after exposure to the redox electrolyte...............................................................109 Figure 2.6. DLFC/DLRFC performance in the 4 cm² fuel cell for high and low fuel concentrations of (a) methanol and (b) formic acid. (Temperature, 70°C; reactant stoich, 4; air at 101.3 kPa abs. and not humidified). ....................111 ix Figure 2.7a. DLFC/DLRFC cell and electrode performance (IR-corrected) in the 4 cm² fuel cell for a 16.7 M methanol anolyte. (Temperature, 70°C; fuel flow rate, 5 mL/min; redox flow rate, 5 mL/min; air flow rate, 38 mL/min [@ 25°C and 101.3 kPa abs.]; air not humidified).....................................................113 Figure 3.1. Schematic of the direct liquid redox fuel cell test system. ...............125 Figure 3.2. Expanded view of the 4 cm² fuel cell. .............................................126 Figure 3.3. Electrolyte conductivities as a function of temperature for (a) 2.5 M Fe(ClO4)3; (b) 1 M FeNH4(SO4)2 / 0.5 M H2SO4; (c) 3 M Fe(NO3)3. ...........129 Figure 3.4a. Cyclic voltammograms over Pt and GC working electrodes for a 1 M Fe(ClO4)3 electrolyte. (90°C, 50 mV/s scan rate, IR-corrected) .................130 Figure 3.5. Cell and individual electrode polarization (IR-corrected) for a DMRFC supplied with (a) 2.5 M Fe(ClO4)3 catholyte and 1 M CH3OH, 0.2 M HClO4 anolyte (perchlorate system, Rcell= 0.03 ȍ) and (b) 1 M FeNH4(SO4)2 / 0.5 M H2SO4 catholyte and 1 M CH3OH, 0.1 M H2SO4 anolyte (sulfate system, Rcell= 0.12 ȍ·cm²). (90°C cell temperature, Nafion® 112 membrane, 1 mL/min anolyte flow rate, 2 mL/min catholyte flow rate). ........................................137 Figure 3.6. Power density curves calculated from the data shown in Figure 3.5. ...................................................................................................................138 Figure 3.7. Cell polarization curves (IR-corrected) for a DMRFC using anolytes with different methanol concentrations. (70°C cell temperature, Nafion® 112 membrane, X M CH3OH / 0.5 M H2SO4 anolyte, 5 mL/min anolyte flow rate, 0.81 M FeNH4(SO4)2 / 0.09 FeSO4 / 0.5 M H2SO4 catholyte, 5 mL/min catholyte flow rate).....................................................................................140 Figure 3.8. Cell polarization curves (IR-corrected) for a DMRFC at different temperatures. (Nafion® 112 membrane, 1 M CH3OH / 0.2 M HCl4 anolyte, 1 mL/min anolyte flow rate, 2.5 M Fe(ClO4)3 catholyte, 2 mL/min catholyte flow rate). ..........................................................................................................141 Figure 3.9a. Short-term durability testing and cell impedance measurements for a perchlorate-based DMRFC operating at 50°C and 50 mA/cm². (Nafion® 112 membrane, 1 mL/min anolyte flow rate, 2 mL/min catholyte flow rate). A xcurrent interruption for 10 seconds (0 mA/cm²) was introduced after 3.5 hrs. ...................................................................................................................142 Figure 4.1. Schematic of a mixed-reactant direct liquid redox fuel cell (MR- DLRFC)......................................................................................................151 Figure 4.2. Comparison of conventional vs. in-situ DLRFC regeneration. ........152 Figure 4.3. Schematic of the mixed reactant direct liquid fuel cell system. .......157 Figure 4.4. Expanded view of the 4 cm² fuel cell. .............................................157 Figure 4.5. Differential scanning calorimetry results for a 2 M methanol redox electrolyte (no acid) (a) with and (b) without carbon particles. ...................160 Figure 4.6. Comparison of cyclic voltammograms: (a) 0 M and (b) 2 M methanol redox electrolytes over GC (70°C, 50 mV/s scan rate, IR-corrected).........162 Figure 4.7. Conductivity of the methanol redox electrolyte as a function of methanol concentration and temperature. .................................................164 Figure 4.8. Cell polarization behaviour (IR-corrected) for a MR-DLRFC as a function of temperature and methanol concentration. Mixed electrolyte supplied to cathode at 5 mL/min and comprised of X M CH3OH, 0.81 M FeNH4(SO4)2, 0.09 M FeSO4, 0.5 M H2SO4. DLRFC reference data is also included for comparison (anolyte composition 2 M CH3OH, 0.5 M H2SO4; anolyte flow rate 5 mL/min; catholyte composition 0.81 M FeNH4(SO4)2, 0.09 M FeSO4, 0.5 M H2SO4; catholyte flow rate 5 mL/min). .............................168 Figure 4.9. Power density curves calculated from the data shown in Figure 4.8a. ...................................................................................................................171 Figure 4.10a. Galvanostatically measured polarization curves for in-situ redox couple regeneration in single-pass mode at 70 and 90°C. (anolyte composition 0.9 M FeSO4, 1 M H2SO4 anolyte flow rate 5 mL/min; cathode air flow rate 38 mL/min [@ 25°C and 101.3 kPa abs.]; air not humidified). 174 Figure 5.1. Schematic of a direct liquid redox fuel cell (DLRFC).......................183 Figure 5.2. Schematic of a mixed-reactant direct liquid redox fuel cell (MR- DLRFC)......................................................................................................194 Figure 5.3. Schematic of DLRFC redox couple regeneration using an external electrochemical regeneration cell. .............................................................198 xi Figure 5.4. Schematic of redox couple regeneration for a DLRFC using the novel in-situ regeneration approach. ...................................................................199 Figure 6.1. Photograph of a three-electrode cell used for cyclic voltammetry...219 Figure 6.2. Selecting a new instrument in Corrware. ........................................220 Figure 6.3. Setting the instrument convention in Corrware. ..............................221 Figure 6.4. The virtual instrument mapper in Corrware software. .....................222 Figure 6.5. Inserting a new experiment in Corrware. ........................................223 Figure 6.6. Configuring a cyclic voltammetry experiment in Corrware. .............224 Figure 6.7. Configuring the Zplot software. .......................................................226 Figure 6.8. Configuring the Zplot software for cyclic voltammetry impedance measurements. ..........................................................................................227 Figure 6.9. Example of an electrochemical impedance response. ....................227 Figure 6.10. Photograph of a conductivity cell. .................................................229 Figure 6.11. Configuring the Zplot software for membrane conductivity measurements. ..........................................................................................230 Figure 6.12. Photograph of the 4 cm² fuel cell disassembled. ..........................234 Figure 6.13. Photograph of the assembled 4 cm² fuel cell. ...............................235 Figure 6.14. Photograph of the fuel cell test system. ........................................235 Figure 6.15. Configuring a galvanostatic experiment in Corrware. ...................237 Figure 6.16. The standard potential of the DLRFC reactions as a function of temperature. ..............................................................................................240 Figure 6.17. Equilibrium cell potential of the DLRFC at 25°C as a function of the redox couple SOC. (All reactants/products apart from the redox couple are assumed to have an activity of 1) ..............................................................241 Figure 6.18.Equilibrium cell potential of the DLRFC at 70°C as a function of the redox couple SOC. (All reactants/products apart from the redox couple are assumed to have an activity of 1) ..............................................................242 Figure 6.19. Equilibrium cell potential of the DLRFC at 90°C as a function of the redox couple SOC. (All reactants/products apart from the redox couple are assumed to have an activity of 1) ..............................................................243 xii Figure 6.20. Study of competition between iron ions and protons in Nafion 117® membranes. Immersion of membrane samples in redox electrolyte [0.81 FeNH4(SO4)2, 0.09 M FeSO4 and X M H2SO4] performed at 70°C. Membrane conductivity and IEC measurements performed at 25°C. ..........................245 Figure 6.21. Repeatability tests (IR-corrected) conducted on a sulfate-based DLRFC operating at 70°C. (Nafion® 112 membrane, 2 mg/cm² 40% 1:1 mol/mol Pt:Ru on C anode catalyst, 2 M CH3OH / 0.5 M H2SO4 anolyte, 5 mL/min anolyte flow rate, TGP-H-120 (3x) cathode, 0.81 M FeNH4(SO4)2 / 0.09 FeSO4 / 0.5 M H2SO4 catholyte, 5 mL/min catholyte flow rate) ..........246 Figure 6.22. Replicate polarization curves (IR-corrected) for a perchlorate-based DLRFC at 90°C. (Nafion® 112 membrane, 2 mg/cm² 40% 1:1 mol/mol Pt:Ru on C anode catalyst, 1 M CH3OH / 0.2 M HClO4 anolyte, 1 mL/min anolyte flow rate, TGP-H-120 (3x) cathode, 2.5 M Fe(ClO4)3 catholyte, 2 mL/min catholyte flow rate).....................................................................................247 Figure 6.23. Schematic diagram of a direct hydrogen redox fuel cell showing it is a hybrid of a PEMFC and a redox flow battery. .........................................254 Figure 6.24. Cyclic voltammograms (70°C, scan rate 50 mV/s) for hydrogen and the Fe3+ / Fe2+ redox couple mixed on glassy carbon and for the redox couple alone on glassy carbon and platinum. ............................................256 Figure 6.25. Comparison of cyclic voltammograms (25°C, scan rate 100 mV/s) for the Fe3+/Fe2+ redox couple on glassy carbon and on Toray TGPH-090 carbon fiber paper in different solutions:  a) 5 mM K3Fe(CN)6 in 1 M KNO3; b) 0.81 M / 0.09 M Fe(NH4)(SO4)2. 12H2O / FeSO4. 7H2O in 0.5 M H2SO4. ...................................................................................................................258 Figure 6.26. Effect of compression on direct hydrogen redox fuel cell performance for carbon fiber paper and thin felt cathodes (40°C, catholyte flow rate 10 ml/min). ..................................................................................260 Figure 6.27. Comparison of polarization and power density curves for carbon fiber paper and thin felt cathodes in the direct hydrogen redox fuel cell (70°C, catholyte flow rate 10 ml/min). ........................................................261 xiii Figure 6.28. Direct hydrogen redox fuel cell polarization curves obtained at 40°C for different flow rates of the Fe3+ / Fe2+ solution. Toray TGPH-090 carbon fiber paper used for the cathode. ...............................................................261 Figure 6.29. Direct hydrogen redox fuel cell polarization and power density curves obtained in the temperature range of 25°C to 70°C for a constant Fe3+ / Fe2+ solution flow rate of 10 ml/min.  Toray TGPH-090 carbon fiber paper used for the cathode. .......................................................................263 xiv NOMENCLATURE Symbols A pre-exponential factor (m/s) Aactual actual electrode surface area (m²) Ac cathodic pre-exponential factor (m/s) Ageometric projected electrode surface area (m²) Aj cross-sectional area of component j (m) aproduct,j activity of the j'th product (-) areactant,j activity of the j'th reactant (-) AVol volume-specific area (m 2/m3) ba anodic tafel slope (V) bc cathodic tafel slope (V) C salt concentration (mol/m³) Cb,Ox bulk concentration of the oxidized species (mol/m³) Cb,Ox,j bulk concentration of the j'th oxidized species (mol/m³) Cb,Red bulk concentration of the reduced species (mol/m³) Cb,Red,j bulk concentration of the j'th reduced species (mol/m³) CD volumetric charge density (C/m³) Ci concentration of species j (mol/m³) COx concentration of the oxidized species (mol/m³) COx(0,t) concentration of the oxidized species at the electrode surface (x=0) as a function of time (t) (mol/m³) CRed(0,t) concentration of the reduced species at the electrode  surface (x=0) as a function of time (t) (mol/m³) Cs,Ox surface concentration of the oxidized species (mol/m³) Cs,Ox,j surface concentration of the j'th oxidized species (mol/m³) Cs,Red surface concentration of the reduced species (mol/m³) Cs,Red,j surface concentration of the j'th reduced species (mol/m³) DOx diffusion coefficient of the oxidized species (m²/s) e natural exponent (2.718) E0 potential at standard conditions and 25°C (V) E0a anode potential at standard conditions (V) E0c cathode potential at standard conditions (V) E0T potential at standard conditions and temperature T (V) xv Ea anode potential (V) Ec cathode potential (V) Ec,b cathode potential based on the bulk concentration(V) Ec,s cathode potential based on the surface concentration (V) Ecell cell potential (V) Ee equilibrium potential (V) Ee,a equilibrium anode potential (V) Ee,c equilibrium cathode potential (V) Ee,cell equilibrium cell potential (V) Ee,P1, Ee,P2 equilibrium potential at pressure 1 and pressure 2 (kPa) EOC open circuit voltage (V) F Faraday's number (96486 C/mol·e-) G Gibbs free energy (J) i superficial current density (A/m²) I current (A) iL mass-transfer limiting current density (A/m²) iL,Ox mass-transfer limiting current density of the oxidized species (A/m²) iL,Red mass-transfer limiting current density of the reduced species (A/m²) J flux (mol/m²·s) j actual current density (A/m²) j* local current denstiy (A/m²) j*L mass transport limited current denstiy (A/m²) j0 exchange current density (A/m²) ja anodic current density (A/m²) jc cathodic current density (A/m²) JOx flux of the oxidized species (mol/m²·s) K constant for calculating the molar conductivity (S m0.5/mol0.5) k0 standard heterogenous rate constant(m/s) kc reduction reaction rate constant (m/s) Km mass transport coefficient (m/s) Km,Ox mass transport coefficient of the oxidized species (m/s) Lj length of component j (m) mcell mass of the cell (g) n number of moles (mol) xvi nRDS number of moles transferred in the rate determining step (mol) P pressure (kPa abs.) P1, P2 pressure at state 1 and 2 (kPa abs.) q tortuosity factor (-) Q charge (C) R ideal gas constant (8314 kPa·m³ / mol·K) rc rate of reduction at the cathode (mol/m²·s) Rcell cell impedance (ȍ) Rj resistance of component j (ȍ) Rmem membrane impedance (ȍ) S entropy (J/mol K) sOx stoichiometric coefficient of the oxidized species (-) sOx,j stoichiometric coefficient of the j'th oxidized species (-) sRed,j stoichiometric coefficient of the j'th reduced species (-) T temperature (K) t time (s) V volume (m³) v- stoichiometric number of anions in a salt (-) v+ stoichiometric number of cations in a salt (-) Vcell volume of the cell (m³) Vj (dot) flow rate of reactant j (m 3/s) x length (m) X reaction conversion (-) zj charge number of ion j (-) Įa anodic transfer coefficient (mol) Įc cathodic transfer coefficient (mol) ȕ symmetry factor (-) į length of the diffusion boundary (m) ǻEIR ohmic voltage loss (V) ǻfG0Ox,j change in standard Gibbs free energy of formation of the j'th oxidized species (J) ǻfG0Red,j change in standard Gibbs free energy of formation of the j'th reduced species (J) ǻG change in Gibbs free energy (J) ǻG‡0 activation energy for the anodic and cathodic reaction intermediates at standard conditions (J) ǻG‡a activation energy of the anodic reaction intermediate (J) xvii ǻG‡c activation energy of the cathodic reaction intermediate (J) ǻG0 change in standard Gibbs free energy (J) ǻH change in enthalpy (J) ǻng change in number of moles of gas (mol) ǻV change in volume (m³) İ void fraction (-) Ș electrode overpotential (V) Șa anodic overpotential (V) Șc cathodic overpotential (V) Șcell cell overpotential (V) Șconc,a anodic concentration overpotential (V) Șconc,c cathodic concentration overpotential (V) Șs surface overpotential (V) Șs,a anodic surface overpotential (V) Șs,c cathodic surface overpotential (V) ț electronic conductivity of the solid phase (S/m) țe effective electronic conductivity of the solid phase (S/m) ȁ molar conductivity (S·m²/mol) ȁ0 molar conductivity at infinite dilution (S·m²/mol) Ȝ0- ionic molar conductivity of the anion at infinite dilution (S·m²/mol) Ȝ0+ ionic molar conductivity of the cation at infinite dilution (S·m²/mol) Ȝj ionic molar conductivity of ion j (S·m²/mol) Ȝs,j stoichiometric factor for reactant j (-) µj mobility of ion j (m²·mol/J·s) ȡj resistivity of component j (ȍ·m) ı ionic conductivity of the membrane/electrolyte (S/m) ıe effective ionic conductivity of the membrane/electrolyte (S/m) IJ maximum electro-active thickness of 3-D electrode (m) ĭ overall efficiency (-) ĭi current efficiency (-) ĭTh thermodynamic efficiency (-) ĭV voltage efficiency (-) xviii Acronyms 3-D three-dimensional abs. absolute AEM anion exchange membrane AFC alkaline fuel cell approx. approximately ATR auto-thermal reforming BET Brunauer-Emmett-Teller CC carbon cloth CD charge density CEM cation exchange membrane CFP carbon fiber paper CHBE Chemical and Biological Engineering CHP combined heat and power CV cyclic voltammetry DEFC direct ethanol fuel cell DI deionized DLFC direct liquid fuel cell DLRFC direct liquid redox fuel cell DMFC direct methanol fuel cell DMRFC direct methanol redox fuel cell DSC differential scanning calorimetry EDTA ethylenediaminetetraacetato FRA frequency response analyzer GC glassy carbon GDC gadolinium-doped cerium oxide GF graphite felt HEDTA ethylenediaminetetraacetic acid HHV higher heating value HRFC hydrogen redox fuel cell IEC ion exchange capacity IFCI Institute for Fuel Cell Innovation IMFC indirect methanol fuel cell LHV lower heating value LSM lanthanum-doped strontium manganese oxide MCFC molten carbonate fuel cell MEA membrane electrode assembly MPL micro-porous layer MR-DLRFC mixed-reactant direct liquid redox fuel cell MRFC mixed-reactant fuel cell xix NASA National Aeronautics Space Administration NRC National Research Council NSERC Natural Sciences and Engineering Research Council OCP open circuit potential OCV open circuit voltage ORP oxidation-reduction potential ORR oxygen reduction reaction Ox oxidized species PAFC phosphoric acid fuel cell PBI polybenzimidazole PEM proton exchange membrane PEMFC proton exchange membrane fuel cell PGM platinum group metal POX partial oxidation PTFE polytetrafluoroethylene Red reduced species RFB redox flow battery SHE standard hydrogen electrode SMR steam-methane reformation SOC state-of-charge SOFC solid oxide fuel cell SPEEK sulfonated poly ether ether ketone TBP triple-phase boundary TMPP tetramethoxyphenylporphyrin UBC University of British Columbia VRB vanadium redox battery YSZ yttria-stabilized zirconia xx ACKNOWLEDGEMENTS I extend my deepest gratitude towards my supervisor, Dr. David Wilkinson, for his mentorship and expertise. Working with him for the past 5 years has strengthened my ability to critically analyze and tackle complex problems – a skill valuable not only in scientific research but in many departments of life. I would also like to thank Dr. Khalid Fatih, my co-supervisor, and Franz Moraw who were involved in the research that provided a foundation for my project. Their training and guidance proved to be very valuable over the course of my graduate work. My appreciation also goes out to my committee members Dr. Khalid Fatih, Dr. Elod Gyenge and Colin Oloman for offering helpful feedback at critical times over the course of the project. I am also grateful towards members of the CHBE machine shop, NRC-IFCI machine shop and CHBE stores for providing logistical support for this project. Lastly, I acknowledge my peers Alfred Lam, Caroline Cloutier, Mauricio Blanco, Simon Fan, Omar Herrera, David Bruce and Mohammad Saad Dara for creating an enjoyable and supportive work environment. xxi CO-AUTHORSHIP STATEMENT In Chapter 3, the data presented in Figure 3.4a-c were produced by Mohammed Saad Dara. Processing and analysis of this data was performed by Alan B. Ilicic. In Appendix E, replicate 2 of Figure 6.22 was produced by Mohammad Saad Dara. In Appendix F, only the data presented in Figure 6.24 were produced by Alan B. Ilicic. All other data presented in Appendix F were produced by Franz Moraw. Khalid Fatih and David P. Wilkinson jointly wrote the manuscript included in Appendix F. Drafts of the published manuscripts included in this thesis were reviewed and fine-tuned by David P. Wilkinson as well as Khalid Fatih. This thesis was reviewed and fine-tuned by the Ph.D committee consisting of David P. Wilkinson, Khalid Fatih, Colin Oloman and Elod Gyenge. Individuals who contributed to the manuscripts presented in this thesis include David P. Wilkinson, Khalid Fatih, Franz Moraw, Francois Girard and Mohammad Saad Dara. Apart from the above exceptions, all experimentation, data processing/analysis and thesis/manuscript writing was done by Alan B. Ilicic.  1 1. INTRODUCTION 1.1 History and Overview of Fuel Cell Technology Fuel cell technology allows for the direct conversion of chemical energy into electrical energy without the use of combustion or turbine technology. The first discovery of the fuel cell phenomenon dates back to 1839 and the work of Dr. Christian F. Schönbein (1799-1868), a chemist at the University of Basel, Switzerland, and Sir William Robert Grove (1811-1896), a lawyer, judge and physical scientist [1, 2]. Grove later developed the “gas battery”, shown in Figure 1.1, which used hydrogen and oxygen gas in contact with platinum electrodes and a sulphuric acid electrolyte to power a water electrolyzer [3]. Grove also noted in the latter publication the importance of increasing the triple-phase boundary (TPB), which refers to the areas where the gas, the electrolyte and the solid electrode meet to accommodate the fuel cell reaction. The electrochemical reactions for Grove’s apparatus are as follows:  Fuel Cell Anode 2He2H2 ←+ −+  E0= 0 V vs. SHE (1.1) Cathode OH2e2ΗO2 1 22 →++ −+  E0= 1.23 V vs. SHE (1.2) Overall OHO2 1H 222 →+  E0= 1.23 V (1.3)     2 Electrolyzer Anode OHe2H2O2 1 22 ←++ −+  E0= 1.23 V vs. SHE (1.4) Cathode 2Η2e2Η →+ −+  E0= 0 V vs. SHE (1.5) Overall 222 O2 1HOH +→  E0= -1.23 V (1.6)  In 1889, Langer and Mond, in an effort to economize Grove’s device, fed the “gas chain” with highly impure industrial hydrogen [4]. Unfortunately, there were two main factors that restricted the fuel cell’s robustness: deactivation of the platinum catalyst by poisons such as carbon monoxide and the high cost of the platinum catalyst. Langer and Mond did, however, address the need to increase the TPB sites by employing a porous matrix to separate the gas/liquid phase boundary and by platinizing platinum to enhance the surface area.  Figure 1.1. Schematic of Grove's "Gas Battery" powering an electrolyzer.  3 A revolutionary era in fuel cell technology arose in the mid 20th century due in part to the work of Francis T. Bacon (1904-1992), who pioneered the high- pressure, high temperature alkaline H2/air fuel cell (AFC). By 1959, Bacon led a product development program at Cambridge University and fabricated a 5 kW alkaline fuel cell to power a forklift and other machines [5]. A major drawback of alkaline H2/air fuel cell technology was the need for electrolyte replacement after the inevitable degradation due to CO2 uptake from air. However, a niche market for this technology became apparent. In the late 1960’s and early 1970’s, Bacon’s alkaline fuel cell was used on the National Aeronautic Space Administration (NASA) Apollo space missions for on-board power [6]. Another milestone within this era was the birth of the proton exchange membrane (PEM), patented for use in an acidic H2/air fuel cell by William Grubb of General Electric in 1955 [7, 8]. These proton exchange membrane fuel cells (PEMFCs) were later used in the mid 1960s for the NASA Gemini space missions to generate on- board power in the space shuttle. Although PEMFC technology opened up new avenues to improve the number of TPB sites by combining membrane ionomer resin with the catalyst layer, new challenges were introduced regarding membrane stability, contamination, compatibility and hydration. In the 1960s, General Electric turned to DuPont to develop a novel membrane based on Teflon® [9]. From this joint effort emerged the perfluorosulfonic acid membrane Nafion®, which rapidly rose as the international standard membrane for PEMFCs. However, General Electric later lost faith in PEMFC technology during the late  4 70’s due to its poor power density, high catalyst loading requirement and the need for an expensive membrane. Advancements in PEMFC technology played a significant role in the development of direct methanol fuel cells (DMFCs). It is no coincidence that the initial major developments in DMFC technology began in the 1960s. Some investigators initially considered indirect methanol fuel cells, where the methanol is catalytically reformed in-situ to hydrogen, but this approach proved to be cumbersome over the decades. Researchers observed early on that direct methanol oxidation over Pt in an acidic environment poisoned the catalyst with CO. During the 1960s, Shell Company investigated many Pt alloys for use as a methanol oxidation catalyst and observed good results with only PtRu and PtRh, the former delivering the best performance [10]. Eventually, a 300 W unit which powered its own auxiliaries was fabricated but was unfortunately limited to specialized applications due to the high catalyst loading required at the anode (10 mg/cm² PtRu). Consequently, future research efforts by Shell were targeted towards reducing the cost by engineering the catalyst layer. Other notable DMFC research in the 1960s includes the work of M.W. Breiter, T. Biegler, D. Koch, V. S. Bagotzky and Y. Vassilyev covering the subjects of methanol oxidation mechanisms on Pt [11], adsorbates on Pt after methanol oxidation [12-14] and CO oxidation on Pt [15, 16]. In 1912, Nernst stated that mediating oxygen reduction with a redox couple may help overcome the irreversibility of oxygen reduction, since many redox couples exhibit high exchange current densities [17]. It is therefore not  5 surprising that one of the first applications considered for redox couples was redox-mediated coal oxidation to extract electrical energy from coal in the 1950’s and 1960’s [18, 19]. Indeed, this represented a novel approach to acquire electrical energy from coal, but coal combustion coupled with gas turbines prevailed due to its superior economics, performance and simplicity.  In the 1980s, mediated H2/O2 fuel cells, also known as chemically regenerative fuel cells, became a subject of interest but never reached a commercial stage [20- 23]. It was in the 1970s that the most significant application for redox couple electrochemistry would be thoroughly investigated for power generation: redox flow batteries (RFBs) for electrical energy storage. The NASA Lewis Research Center rapidly emerged as a leader in this field through the reports and publications detailing the selection of suitable redox couples [24, 25], advancements in the Fe2+/Fe3+,Cr2+/Cr3+ RFB [26, 27],  system cost estimates [28], scale up considerations [29] and applications involving renewable energy [30]. The primary issues associated with the Fe/Cr RFB pioneered by NASA were: 1) crossover of the redox couple leading to permanent capacity loss; 2) the need to rebalance the state-of-charge (SOC) of the anolyte and catholyte due to hydrogen generation at the Cr2+/Cr3+ electrode and 3) high membrane resistivity (c.f., 2 Ω·cm²). Some of these key issues were addressed with the birth of a new type of RFB in the 1980s, which involved vanadium redox couples at the anode (V2+/V3+) and cathode (VO2+/VO2+) [31]. Much experimentation followed, which covered the areas of electrolyte composition [32, 33], electrode materials [34], membrane modification [35-38], electrochemical characterization [39-41] and  6 stack testing [42, 43]. The all-vanadium RFB eventually reached wide commercial deployment, particularly in backup power, renewable energy storage and load-levelling applications. Over the years, many other types of fuel cells emerged, including the phosphoric acid fuel cell (PAFC), the direct ethanol fuel cell (DEFC), the molten carbonate fuel cell (MCFC) and the solid oxide fuel cell (SOFC).  These fuel cell types are often distinguished by the fuel, the conducting ion and the range of operating temperatures. A summary of several types of fuel cells is shown in Table 1.1. Each fuel cell type is suitable for a specific set of applications and a limited range of power output. However, when several types of fuel cells are considered collectively, as shown in Figure 1.2, they cover a wide power output range (mW-MW) and many applications including portable, automotive and combined heat and power (CHP).  Table 1.1. Key properties of various fuel cells.    7  Figure 1.2. Power output and applications of several fuel cells [44].   8 1.2 Fundamental Principles of Fuel Cell Technology At the heart of fuel cell technology lies the basic elements of electrochemistry, the science where chemistry and electricity meet one another at the interface of an electrode and electrolyte. Electrochemical reactions always involve the transfer of an electron between an oxidized (Ox) and reduced (Red) species. In standard form, electrochemical reactions are written as a reduction reaction:  ∑∑ →+ − j jjdRe i iiOx dResneOxs ,,  (1.7) where sOx,i / sRed,j is the stoichiometric coefficient of an oxidized / reduced species and n is the number of electrons transferred.  1.2.1 Thermodynamics: Standard Potential The change in Gibbs free energy at standard conditions (∆G0, T= 298.15 K, P= 101.3 kPa absolute (abs.), activity of 1 for all products and reactants) for an electrochemical reaction can be written generally as a function of the Gibbs free energy of formation of the oxidized (∆fG0Ox) and reduced species (∆fG0Red):  ∑∑ ∆−∆=∆ j 0 jOxf i 0 idf 0 GGG ,,Re  (1.8) Since electrical energy is defined as the product of the electromotive force (E) and the charge transferred (Q), the standard Gibbs free energy of an electrochemical reaction can be expressed with the following equation:  000 nFEQEG −=−=∆  (1.9)  9 where n is the number of moles of electrons transferred in the electrochemical reaction, F is Faraday’s constant (96486 C/mol) and E0 is the standard potential at 25°C in V. The negative sign respects the convention for spontaneous reactions: a positive cell potential yields a negative Gibbs free energy. By definition, cathodic reactions involve reduction and anodic reactions involve oxidation. It follows that by combining equations 1.8 and 1.9, the standard cell potential (E0cell) can be expressed as a function of the standard anode (E0a) and cathode (E0c) potentials:  0a 0 c 0 cell EEE −=  (1.10) An electrochemical cell can be envisaged by combining two electrochemical half-cell reactions where one reaction would undergo reduction and the other oxidation. The spontaneity of this electrochemical cell at standard conditions will be dictated by the polarity of the standard cell potential. A list of some electrochemical half-cell reactions is given in Table 1.2. The phase of the reactant or product is included in Table 1.2 since the Gibbs free energy and consequently the half-cell potential is influenced by this property. As half-cell reactions represent only half of a complete electrochemical cell, the potential of a half-cell reaction must be reported against that of another half-cell reaction. In Table 1.2, potentials are given versus the standard hydrogen electrode (SHE) at 298 K, which is shown in Table 1.2 to have a potential of 0 V vs. SHE. Several types of reference electrodes such as the SHE are used for electrochemical measurements, each with their own reference potential. A list is given in Table 1.3.  10 Table 1.2. Some electrochemical half-cell reactions [45].    Table 1.3. A list of reference electrodes [46].     11 1.2.2 Thermodynamics: Effect of Temperature Thus far, only electrochemical reactions occurring at standard conditions have been discussed. Typically, the cell temperature, reactant/product pressure and reactant/product activity will not reflect that of standard conditions. In this case, corrections can be made to obtain the equilibrium cell potential, Ee, which reflects the current operating conditions. The dependency of the reaction potential on temperature can be expressed based on equation 1.9:  PP T G nF 1 T E ⎟⎠ ⎞⎜⎝ ⎛ ∂ ∆∂−=⎟⎠ ⎞⎜⎝ ⎛ ∂ ∂  (1.11) where T is the temperature in K and P is the pressure in kPa. From basic thermodynamics, the following expression for the Gibbs free energy for a reversible process can be written:  SdTVdPdG −=  (1.12) where V is the volume in m³ and S is the entropy in J/mol·K. By taking the partial derivative of equation 1.12 with respect to temperature at constant pressure, we obtain  ST G P −=⎟⎠ ⎞⎜⎝ ⎛ ∂ ∂  (1.13) where the same can be written for the difference between two states:  ST G P ∆−=⎟⎠ ⎞⎜⎝ ⎛ ∂ ∆∂  (1.14) Combining equations 1.11 and 1.14, a new relationship is realized:  12  nF S T E P 0 ∆=⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛ ∂ ∂  (1.15) Integrating equation 1.15 from 298.15 K to T, where ∆S does not change significantly over the temperature interval, a simple equation for temperature correction of a reaction potential is derived:  ( )15298TnF SEE 00T .−∆+=  (1.16)  1.2.3 Thermodynamics: Effect of Pressure In the case where reactants or products are gases which have a partial pressure other than 101.3 kPa abs., corrections to the reaction potential can be made to reflect the current operating conditions. The dependency of the reaction potential on pressure can be expressed based on equation 1.9:  TT P G nF 1 P E ⎟⎠ ⎞⎜⎝ ⎛ ∂ ∆∂−=⎟⎠ ⎞⎜⎝ ⎛ ∂ ∂  (1.17) From equation 1.12, the partial derivative with respect to pressure at constant temperature may be taken:  VP G T =⎟⎠ ⎞⎜⎝ ⎛ ∂ ∂  (1.18) where V is the volume of the gas in m³. The same may be stated for the difference between two states:  VP G T ∆=⎟⎠ ⎞⎜⎝ ⎛ ∂ ∆∂  (1.19)  13 where ∆V is the change in volume of gas in m³ for the electrochemical reaction referred to by ∆G. Combining equations 1.17 and 1.19, we obtain  nF V P E T ∆−=⎟⎠ ⎞⎜⎝ ⎛ ∂ ∂  (1.20) Assuming the gas phase is ideal,  P RTn V g ∆=∆  (1.21) where ∆ng is the change in number of moles of gas, R is the ideal gas constant in mol·K/kPa·m³, T is the temperature in K and P is the pressure in kPa. Inserting equation 1.21 into 1.20 and integrating, the following expression for the effect of pressure on the equilibrium potential is obtained:  g 12 n 1 20 PTPe P Pln nF RTEE ∆ ⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛−= ,,  (1.22)  1.2.4 Thermodynamics: Effect of Concentration The effect of reactant and product concentrations on the change in Gibbs free energy is given by the following fundamental thermodynamic relationship:  ⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛+∆=∆ ∏ ∏ jOx id s jttanreac s iproduct0 a a lnRTGG , ,Re , ,  (1.23) where aRed is the activity of the reduced species i, aOx is the activity of the oxidized species j and s is the stoichiometric coefficient. Combining equations 1.9 and 1.23, the Nernst equation which describes the effect of reactant and product concentrations on the potential is obtained:  14  ⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛−= ∏ ∏ jOx id s jttanreac s iproduct0 Te a a ln nF RTEE , ,Re , ,  (1.24) If the component is a solid or solvent, a=1. For ideal solutions, a=Ci/C0 where Ci is the concentration in mol/L and C0 is the reference concentration (1 mol/L); for non-ideal solutions, ai= γiCi/C0 where γ is the activity coefficient. For ideal gases, a=Pi/P0 where Pi is the partial pressure in kPa and P0 is the reference pressure (101.3 kPa abs.); for non-ideal gases, a= γiPi/P0  1.2.5 Thermodynamics: Efficiency The thermodynamic efficiency of a fuel cell, which represents the theoretical maximum, can be described with the following equation:  H G Th ∆ ∆=φ  (1.25) where ∆H is the enthalpy change in J of the electrochemical reaction. Utilizing equation 1.9 for non-standard conditions, a different expression for the thermodynamic efficiency can be written:  H nFEe Th ∆−=φ  (1.26) It is important to pay attention to the phase of the reactants and products of an electrochemical reaction when considering the thermodynamic efficiency as the phase influences both ∆G and ∆H. This is most often encountered with water and terms have been established to distinguish between the cases where the  15 product water is in the liquid phase (i.e., higher heating value, HHV) and the gaseous phase (i.e., lower heating value, LHV). Apart from the thermodynamic efficiency, current efficiency (Φi), the voltage efficiency (Φv) and reactant utilization are other parameters which affect the overall efficiency of a working fuel cell. Consequently, the overall energy efficiency (Φ) is a product of these efficiencies:  ∏ −λφφφ=φ j 1 jsiVTh ,  (1.27) where λs,j is the stoichiometric factor of reactant j (-) signifies the ratio of reactant j supplied to reactant j consumed, as defined below:  Is VnFC j jj js & =λ ,  (1.28) where sj is the stoichiometric coefficient (-), I is the current in A, Cj is the concentration of reactant j in mol/m3, Vj(dot) is the flow rate of reactant j in m3/s. The current efficiency represents the fraction of total current that is produced by the desired reactions. A low current efficiency indicates the strong presence of unwanted secondary reactions. The voltage efficiency reflects all combined voltage losses that reduce the operating voltage of the fuel cell below the theoretical equilibrium voltage of the fuel cell, Ee. The stoichiometric factors of the reactants simply account for the fraction of a reactant that leaves the fuel cell unreacted.   16 1.2.6 Cell Voltage and Overpotential The operating voltage of a fuel cell is dependent on many variables, such as catalysts, operating temperature, operating current density, degree of reactant crossover, reactant flow rate and membrane conductivity. The performance of a fuel cell is normally characterized by its polarization curve, which is a graph of the cell voltage vs. the current density. The polarization curve can be separated into three sections where different voltage loss mechanisms are dominant, as shown in Figure 1.3.  Figure 1.3. An example of a fuel cell polarization curve which illustrates the various regions of a polarization curve.  Often the operating voltage of a fuel cell at a current density of 0 mA/cm² (i.e., the open circuit voltage [OCV], EOC) does not correspond to the theoretical equilibrium cell voltage, Ee,cell, as one would anticipate. This can be due reactive  17 impurities present in the system or crossover of a reactant through the membrane to the opposite side of the fuel cell where it is able to react. Not only does reactant crossover waste the reactant, it also negatively affects the potential of the electrode it has crossed over to. It is this shift in potential that creates a discrepancy between the equilibrium cell voltage and the OCV. The magnitude of the difference indicates the degree of reactant crossover and/or impurities. The voltage loss of the cell, anode and cathode is normally referred to as the cell (ηcell), anode (ηa) and cathode (ηc) overpotential, respectively, as defined by the following equations:  cellecellcell EE ,−=η  (1.29)  aeaa EE ,−=η  (1.30)  cecc EE ,−=η  (1.31) The cell overpotential can be broken down into the kinetic/activation overpotential (ηs), mass transport/concentration overpotential (ηconc) and ohmic/IR losses (∆EIR). Consequently, the cell voltage can be expressed as a function of the various anodic and cathodic overpotentials:  c,conca,concIRc,sa,scell,ecell EEE η+η−∆−η+η−=  (1.32) where, by definition, anodic overpotentials are positive and cathodic overpotentials are negative for a fuel cell.   18 1.2.7 Superficial Current Density vs. Actual Current Density It is important to differentiate between the actual current density, j, and the superficial current density, i. The actual current density is defined as follows:  actualA Ij =  (1.33) where Aactual is the true electrochemically active surface area measured in m² using a sophisticated technique such as the Brunauer-Emmett-Teller (BET) method (see [47]). On the other hand, the superficial current density, i, is normalized by the geometric surface area:  geometricA Ii =  (1.34) where Ageometric is the projected surface in m² area of the electrode, which does not account for the roughness or porosity of the electrode. Often the superficial current density is reported in the literature rather than the actual current density due to the difficulty of measuring the actual surface area of the electrode. In this thesis, the theoretical section is developed for the most part using the actual electrode surface area whereas projected electrode surface area is used throughout the experimental sections when reporting values normalized by the electrode area (e.g., current density, power density).   19 1.2.8 Overpotential: Kinetics The details of electrode kinetics and activation overpotential can be understood by starting with some basic principles of chemistry and ending with a comprehensive equation known as the Butler-Volmer equation. The kinetic rate of a 1st order ideal reduction reaction involving only one oxidized and one reduced species can be expressed with the following equation:  ( )t0Ckr Oxcc ,=  (1.35) where rc is the reduction reaction rate in mol/m2·s, kc is the reduction reaction rate constant in m/s and COx(0,t) is the concentration of the oxidized species at x=0 (the surface of the electrode) as a function of time in mol/m³. The reaction rate can also be expressed as a function of the actual current density (j):  ( ) nF jr cc −=  (1.36) where jc is the reduction current density in A/m² and is negative in polarity by definition. Combining equations 1.35 and 1.36, an expression for the cathodic current density can be obtained:  ( )t0CnFkj Oxcc ,−=  (1.37) The temperature dependency of the reduction rate constant, kc, follows an Arrhenius type behaviour:  RT G cc c eAk m∆− =  (1.38) where Ac is the pre-exponential factor in m/s and ∆Gc‡ is the activation energy in J for the reduction reaction intermediate.  20 The activation energy of a reaction intermediate, ∆G‡, is strongly dependent on the potential of the electrode. This is best illustrated by graphing the Gibbs free energy vs. the reaction coordinate for a case where the system is at equilibrium and a case where the electrode potential is shifted from its equilibrium value, as shown in Figure 1.4. At equilibrium, the activation energy for the anodic (∆Ga‡) and cathodic (∆Gc‡) reaction intermediates is equal. as shown in Figure 1.4a. When the electrode is cathodically polarized, as shown in Figure 1.4b, the Gibbs free energy of the oxidized (Ox) and reduced (Red) species is displaced from one another by ∆G, according to equation 1.9:  ( )0RDS EEFnG −−=∆  (1.39) where nRDS is the number of electrons transferred in the rate determining step.  21                Figure 1.4. Conceptual graph illustrating activation energies of reaction intermediates during (a) standard conditions and (b) cathodic polarization. (b) (a)  22 What fraction of the total displacement is allocated to each species depends on a parameter known as the symmetry factor, β. Therefore,  ( ) ( )( ) ( )( )0RDS00RDS0a EE1FnGEEFn1GG −β−−∆=−−β−+∆=∆ mmm  (1.40) and  ( )0RDS0c EEFnGG −β+∆=∆ mm  (1.41)  where ∆G0‡ represents the activation energy for the anodic and cathodic reaction intermediates at standard conditions. Inserting equation 1.41 into equation 1.38, we obtain  ( ) RT EEFn RT G c 0 RDS0 eAek −β−∆−= m  (1.42) At this point, the standard heterogeneous rate constant, k0, may be defined:  RT G 0 0 Aek m∆− =  (1.43) The anodic and cathodic transfer coefficients, αa and αc, can also be defined:  ( )β−=α 1nRDSa  (1.44)  β=α RDSc n  (1.45) Returning to equation 1.37, equations 1.42, 1.43 and 1.45 can be utilized:  ( ) ( )RT EEFOx0c 0 c et0CnFkj −α− −= ,  (1.46) A similar expression may be written for the anodic current density:  ( ) ( )RT EEFd0a 0 a et0CnFkj −α = ,Re  (1.47)  23 where CRed(0,t) is the concentration of the reduced species at the electrode surface (x=0). The total current density, j, is simply a sum of the anodic and cathodic current density:  ( ) ( ) ( ) ( ) ⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛ −= −α−−α RT EEF Ox RT EEF d 0 0 c 0 a et0Cet0CnFkj ,,Re  (1.48) The equilibrium potential can be defined for this case:  ⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛−= bOx bd0 Te C C ln nF RTEE , ,Re  (1.49) where CRed,b and COx,b are the concentrations of the reduced and oxidized species in the bulk, respectively. It is also useful to define the exchange current density, j0:  ( ) ( ) ac bOxbd00 CCnFkj αα= ,,Re  (1.50) The electrode overpotential, η, is given by the following equation:  eEE −=η  (1.51) Combining equations 1.48, 1.49 and 1.50, the complete Butler-Volmer equation to describe the kinetics of a 1st order redox reaction is realized:  ( ) ( ) ηRTFα Oxb, Ox η RT Fα Redb, Red 0 ca e C t0,Ce C t0,C j j −−=  (1.52) where the electrode overpotential, η, is always positive for anodic polarization and always negative for cathodic polarization.   24 Under purely kinetic control with no mass transport limitations, the reactant concentration at the surface of the electrode will equal that of the bulk concentration and the total electrode overpotential, η, simplifies to the electrode surface overpotential, ηs. The Butler-Volmer equation then reduces to the Butler- Erdey-Gruz-Volmer (BEV) equation [48]:  s c s a RT F RT F 0 ee j j ηα−ηα −=  (1.53) An exemplary graph of the BEV equation is shown in Figure 1.5. When the electrode polarization is very small (less than 50 mV) both the anodic and cathodic contributions to the total current are significant and the j vs. η relationship approaches linearity. For this range of overpotential, one may apply the linear approximation of the BEV equation:  ( ) RT F j j s ca 0 ηα+α=  (1.54)   25  Figure 1.5. A sample plot of the Butler-Erdey-Gruz-Volmer equation.  If the electrode potential is sufficiently polarized in the cathodic direction, the anodic term in the BEV equation will become insignificant:  ηα−−= RT F 0 c e j j  (1.55) From here, the Tafel equation can be realized:  ⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛⋅=η 0 cs j jlnb  (1.56) and  RT Fb cc α−=  (1.57) j  26 Similarly for anodic polarization:  RT Fb aa α=  (1.58) The Tafel slope for a particular reaction depends on the electrode material, electrolyte composition and temperature whereas the exchange current density is a function of the electrode material, electrode state (roughness, presence of oxides, adsorbates), electrolyte composition and temperature. A list of exchange current densities for various redox couples at 27°C is shown in Table 1.4. For good fuel cell electrode performance, a high exchange current density and low Tafel slope are desirable for the fuel cell reactions of interest within this thesis.  Table 1.4. Exchange current densities for various redox couples at 27°C [24].   27 1.2.9 Ohmic Losses Revisiting equation 1.32 at this point, the ohmic/IR overpotential term, ∆EIR, can now be discussed. There are several components of a fuel cell other than the load that can contribute to the total resistance (ionic + electronic), including the membrane, electrolyte, flow field plates, current collectors and electrical connections. In general, the ohmic/IR overpotential can be expanded into all of its subcomponents:  ∑∑ ρ==∆ j j jj j jIR A L IRIE  (1.59) where I is the current in A, Rj is the electronic/ionic resistance of component j in Ω, ρ is the resistivity of component j in Ω ·m, A is the cross-sectional area of component j in m² and L is the thickness of component j in m. For the resistivity of electronic conductors, it is sufficient to consult a handbook. A short list of some common electrode materials is given in Table 1.5.  Table 1.5. Short list of potential electrode materials and their electrical resistivities [49].   If the electronically conductive material is porous, the conductivity may be corrected with the following equation [49]:  28  q e κε=κ  (1.60) where κe is the effective conductivity of the solid phase in S/m, κ is the conductivity of the pure solid phase in S/m, ε is the void fraction and q is the tortuosity factor (typically 1.5). Note,  1−κ=ρ  (1.61) An electrolyte exhibits conductive properties due to the mobility of charged ions in solution. A strong electrolyte is defined as one that undergoes complete dissociation:  BABA −+υυ υ+υ→−+  (1.62) where v+ and v- are the stoichiometric number of cations and anions in the salt, respectively. The ionic conductivity of a strong electrolyte at very low concentrations (below 5 mol/m³) can be expressed with the following equation:  ∑∑ λ=µ=σ j jj j jj 2 j 2 CCzF  (1.63) where σ is the electrolyte conductivity in S/m, zj is the charge number of ion j, µj is the mobility of ion j in m²·mol/J·s, Cj is the concentration of ion j in mol/m³ and λj is the ionic molar conductivity in S·m²/mol. The conductivity at very low concentrations may also be defined using the molar conductivity, Λ:  CΛ=σ  (1.64) where Λ is the molar conductivity in S·m²/mol and C is the concentration of the salt Av+Bv- in mol/m³. The molar conductivity can be quantified for low concentration electrolytes using the following empirical relationship:  29  CK0 −Λ=Λ  (1.65) where Λ0 is the molar conductivity at infinite dilution in S·m²/mol, K is a constant in S·m0.5/mol0.5 and C is the concentration of the salt in mol/m³. Furthermore,  000 ++−− λυ+λυ=Λ  (1.66) where λ-0 and λ+0 are the ionic molar conductivities of the anion and cation at infinite dilution and v+ and v- are the stoichiometric number of cations and anions in the salt, respectively. The ionic molar conductivities at infinite dilution for various ions can be found in handbooks. Table 1.6 includes these values for a handful of ions.  Table 1.6. Ionic molar conductivities at infinite dilution and 25°C for some anions and cations [50].   For electrolytes at elevated temperatures with concentrations greater than 5 mol/m³, one may use the modified Casteel-Amis equation [49]:  ( ) ( )'max' max ' max '' ' max ' max ' CCC xCCy x 2 e C C −−− ⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛σ=σ  (1.67)  30  T21 maxmax ' max σ+σ=σ ; TCCC 21 maxmax'max +=  (1.68) where σ is the electrolyte conductivity in S/m, C’ is the solution concentration in % weight, T is the temperature in °C, C’max1 and C’max2 are tabulated constants in % weight, σmax1 and σmax2 are tabulated constants in S/m and x and y are tabulated constants. If there are non-conductive phases present in the electrolyte, such as gas bubbles or non-conductive suspended solids, the conductivity may be corrected using the Maxwell equation, which is valid for ε<0.6 [49]:  21 1 0 ε+ ε−=σ σ  (1.69) where σ0 is the conductivity of the electrolyte without any non-conductive phases and ε is the volume fraction of the non-conducting phase. If the conductivity of an ion exchange membrane is of interest, it is often easiest to measure the conductivity of the membrane directly under the operating conditions of interest using electrochemical impedance spectroscopy. This experimental technique is described in detail in Appendix B.  1.2.10 Overpotential: Mass Transport At this point the various sources for ohmic losses (solid conductors, electrolytes and membranes) have been elaborated upon and it is possible to move forward to the final overpotential terms of equation 1.32, which are pertaining to concentration overpotential. These terms account for significant drops in the concentration of the reactants due to mass transport constraints at  31 high consumption rates (i.e., high current densities). The effect on the electrode potential is essentially Nernstian, according to equation 1.24. Consequently, the concentration overpotential is simply the difference between the cathodic equilibrium potential based on the surface concentration (Ec,s) and the bulk concentration (Ec,b):  bcsccconc EE ,,, −=η  (1.70) Which can be expanded to the general form: ⎥⎥⎦ ⎤ ⎢⎢⎣ ⎡ ⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛−− ⎥⎥⎦ ⎤ ⎢⎢⎣ ⎡ ⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛−=η ∑∑ j jOxb jdb jOx 0 j jOxs jds jd 0 cconc C C lns nF RTE C C lns nF RTE ,, ,Re, , ,, ,Re, ,Re,  (1.71) where surface concentrations are denoted with Cs and bulk concentrations are denoted with Cb. Assuming there is only one oxidized species experiencing starvation at the electrode surface, we may simplify the above equation and write  ⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛=η Oxb Oxs Oxcconc C C lns nF RT , , ,  (1.72) It then becomes necessary to define the surface concentration, Cs,Ox. Assuming one dimensional Fickian diffusion, the flux of the oxidized species towards the electrode surface can be expressed with the following equation:  ( ) 0 CC D dx dCDJ OxbOxsOxOxOxOx −δ −=−= ,,  (1.73) where JOx is the flux of the oxidized species in mol/m²·s, x is the length in m, DOx is the diffusion coefficient of the oxidized species in m²/s and δ is the length of the diffusion boundary layer in m.   32 The mass transport coefficient, Km (m/s), is defined as  δ= DKm  (1.74) Under pure mass transport control the current density can be estimated by that which is supplied to the electrode surface plane by diffusion:  J s nFi =  (1.75) Utilizing equations 1.73, 1.74 and 1.75, we may write  ( )OxbOxsOxm Ox CCK s nFi ,,, −=  (1.76) and  OxbOxm Ox OxL CKs nFi ,,, −=  (1.77) where iL is the mass transport limiting current density in A/m² pertaining to the oxidized species. It then becomes apparent that  OxLOxL OxL Oxb Oxs i i1 i ii C C ,, , , , −=−=  (1.78) And finally an estimate for the mass transport overpotential is derived by substituting equation 1.78 into equation 1.72:  ⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛ −=η OxL Oxcconc i i1lns nF RT , ,  (1.79)  Similarly, equation 1.79 can be written for the anodic concentration overpotential:  33  ⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛ −= RedL, Redaconc, i i1lns nF RT η  (1.80)  1.2.11 Three-Dimensional Electrode Theory Electrochemical cells invariably employ three-dimensional (3-D) electrodes. In some cases, thick porous 3-D electrodes are employed in electrochemical cells to increase the volume-specific electrode surface area. This approach can improve the space-time yield of the electrochemical cell but it can also lead to augmented non-uniformity in the current and potential distribution within the 3-D electrode. Assuming lateral uniformity, the potential distribution across the thickness of a porous electrode can be described with the following equation [51]:  ⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛ σ+κ= eeVol2 2 11jA dx Ed *  (1.81) where AVol is the real volume-specific electrode area in m2/m3, j* is the local current density in A/m², κe is the effective conductivity of the solid phase in S/m and σe is the effective conductivity of the liquid phase in S/m. Equation 1.81 is normally solved numerically using mathematical software.   34  Figure 1.6. Sketch of the potential distribution across the thickness of a 3-D cathode. Em and Es are the potentials of the ionic and electronic conductors, respectively.  35  Figure 1.7. Sketch of the current density distribution across the thickness of a 3- D cathode.  Under pure mass transfer control, a limiting case solution of equation 1.81 exists:  50 LVol e jA E2 . * ⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛ ∆κ=τ  (1.82) where τ is the maximum electro-active thickness of the 3-D electrode in m, ∆E is the voltage drop across τ in V and j*L is the mass transport limited current density in A/m².   36 1.2.12 Rechargeable Fuel Cells Some fuel cells, such as redox flow batteries (RFBs), can be regenerated after discharge. In this case, the total capacity, the charge density of the anolyte/catholyte and the state of charge (SOC) are important parameters. The total capacity of an electrolyte to pass electrons is a function of the concentration of the electroactive species, the number of electrons transferred, n, and the volume of the electrolyte:  nFCVQ =  (1.83) where Q is the electrolyte charge capacity in Coulombs, C is the concentration of the electroactive species in mol/m³ and V is the volume of the electrolyte in m³. When comparing the capacity of various electrolytes to one another, the volumetric charge density (CD) provides a robust comparison by normalizing the capacity with respect to the volume:  nFC V QCD ==  (1.84) where Q is the electrolyte charge capacity in Coulombs. The SOC indicates what fraction of an electrolyte’s maximum capacity is remaining in solution:  %100xC C SOC Total Charged=  (1.85) where the SOC is in %, CCharged is the concentration of the charged species and CTotal is the total concentration of the charged and discharged species combined.  37 1.2.13 Figures of Merit Important figures of merit for a fuel cell include, but are not limited to, the electrolyte charge density, the volumetric power density, the gravimetric power density, the cost and the energy efficiency. The polarization curve can also provide a wealth of information regarding losses, limitations and opportunities for performance enhancement. The power density of a fuel cell at a particular current density can be computed from a polarization curve:  celliEDensityPower =  (1.86) where the power density is in W/m², the current density, i, is in A/m² and the cell potential, Ecell, is in V. If the power density is graphed as a function of the current density over the complete range of current densities, a peak will become visible. The power density at this point is the peak power density, which is also a useful figure of merit. In order to account for the volume or mass of the fuel cell, it is common to normalize the power density by the volume (Vcell) or mass (mcell) of the cell in order to obtain the volumetric or gravimetric power density, respectively:  cell cell V iEDensityPowerVolumetric =  (1.87)  cell cell m iEDensityPowercGravimetri =  (1.88)     38 1.3 The Redox Flow Battery The redox flow battery (RFB) operates as a secondary battery in the way it is charged and discharged but resembles a fuel cell as it involves flowing electrolytes (electrochemical cell is not closed) and the electrodes do not participate in the electrochemical reactions apart from current collection and catalysis. A RFB in its simplest form is comprised of two electrolyte storage tanks, two electrolyte pumps and an electrochemical cell, as shown in Figure 1.8.   Figure 1.8. Simplified schematic of a redox flow battery.  1.3.1 Redox Flow Battery Components The storage tanks hold the electrochemically active redox couple in the anolyte and catholyte while the pumps circulate the anolyte and catholyte through the electrochemical cell during discharge or regeneration. The concentration of the redox couple in the electrolyte and the volume of the storage  39 tanks dictate the total charge/discharge capacity of the RFB. Since the volume of the storage tanks can be sized independently of the electrochemical cell, the power output can be chosen independent of the electrical energy storage capacity of the RFB. This is not the case for secondary batteries where the total active area and volume of active species are related. RFBs are unique relative to other fuel cells in that they operate as a closed system. For this reason, secondary side reactions, crossover of electroactive species or water crossover can lead to significant losses in system capacity or salt precipitation. It is therefore common to prevent or resolve these issues early in the design stage to ensure a long operating life [52]. The electrochemical cell is comprised of two electrodes with flow fields and a membrane or separator between the electrodes to provide ionic conduction, electronic insulation and reactant separation. A number of flow field designs are possible but the most common is an empty pocket packed with a porous material in order to extend the reaction zone and enhance the cell performance. This is an opportunity unique to redox couple reactions because many of them are exempt from TPB constraints as they do not involve a gaseous phase. Carbon fiber paper (CFP), carbon cloth (CC) and graphite felt (GF) are the most commonly used substrates as they yield good electrochemical activity for a variety of redox couples and offer chemical stability in both acidic and alkaline media. CFP, CC and GF each possesses its own unique characteristics, as shown in Figure 1.9. Many types of CFP, CC and GF are available from vendors,  40 each with their own unique properties. Nevertheless, some general physical properties for these carbon substrates can be identified: • GF generally exhibits the highest specific surface area (m²/g) • CFP and CC are the most durable (GF is easily destroyed by compression) • CFP is susceptible to cracking and GF and CC are susceptible to fraying • CFP and CC have a much lower porosity than GF which is important for flow considerations.   Figure 1.9. Images of various carbon substrates. Left to right: carbon fiber paper, carbon cloth, graphite felt [53]. Reproduced by permission of The Electrochemical Society.  The electronic insulator between the electrodes may possess selectivity towards cation transfer, anion transfer or neither. The ion selectivity of the membrane dictates which ions transfer charge during cell operation. A common approach is to employ a cation exchange membrane (CEM) and target highly mobile protons as the dominant charge transferring ion [42, 54-56]. However, since the metal ion redox couples are typically positively charged, this approach also facilitates crossover of the electroactive species. Where anion exchange  41 membranes (AEMs) are employed, crossover of positively charged electroactive species can be significantly reduced but a less mobile than proton anion such as Cl- or OH- must be the dominating charge transfer ion. Since many metal ion redox couples precipitate as hydroxides in alkaline environments, Cl- or SO42- are commonly targeted ions instead of OH- where an AEM is employed [36, 57]. The conductivity of the membrane is also an important consideration, particularly when the membrane is being contaminated by the redox couple. Membrane contamination by the redox couple inevitably occurs to some degree when the membrane conducts ions of the same polarity as the redox couple. An example of this could be occupation of the SO3- sites in a Nafion® cation exchange membrane by Ce2+/Ce3+ redox couple cations. Since the cerium ions are less mobile than the protons that normally occupy the SO3- sites, the conductivity of the membrane is negatively influenced by the Ce3+ cations. Porous separators such as glass filter paper offer the advantage of high conductivity (based on the supporting electrolyte) and low cost but permit relatively high crossover rates of the electroactive species. For this reason, porous separators are seldom found in RFBs.  1.3.2 Redox Couple Selection Selecting suitable redox couples for the anode and cathode of a RFB is a critical task. RFBs are normally defined by the redox couples they employ. Some criteria for redox couple selection include [25]:   42 • Good electrochemical activity and reversibility of the redox couple. • Sufficiently high aqueous solubility of the redox couple in both the reduced and oxidized state. • Electrochemical compatibility with the solvent, electrodes and membrane during discharge and regeneration. • A useful standard open circuit cell voltage after the selection of the fuel and oxidant redox couples • Avoid redox couple reactions which involve a high degree of oxygen transfer from water. These reactions cause large shifts in pH during discharge and regeneration. (e.g., Cr2O72- + 14H+ + 6e- → 2Cr3+ + 7H2O) • Avoid the use of complexing agents as they increase the complexity of the system and often lead to reduced solubility limits.  Although there are many redox couples to choose from, only a small fraction of them are suitable for use as a fuel or oxidant in a RFB. Some examples of redox couples that have been considered for use in RFBs are listed in Table 1.7 along with their respective half-cell potential.   43  Table 1.7. List of various redox couples with their half-cell potentials at 25°C [45].   Redox couples used as a fuel typically have standard half cell potentials within the vicinity of the standard potential for hydrogen production (0 V vs. SHE) whereas redox couples used as oxidants are selected to be closer to the standard potential for oxygen reduction (1.23 V vs. SHE). With this approach, one may obtain a reasonable open circuit voltage (OCV) while avoiding/minimizing hydrogen production or oxygen evolution during regeneration. It is important to note that the standard half-cell potential of a particular redox couple is dependent on more than just the metal ions participating in the electrochemical reaction. Neighbouring ions interact with the redox couple and can significantly influence the half-cell potential. For instance, the Ce3+/Ce4+ redox couple possesses a half-cell potential of 1.7 V vs. SHE in 1 M HClO4 but only 1.28 V vs. SHE in 1 M HCl. Complexing agents can also have a significant  44 impact on the half-cell potential of a redox couple. For example, the Fe3+/Fe2+ redox couple shifts to about 0.12 V vs. SHE in the presence of ethylenediaminetetraacetato (EDTA). Choosing an appropriate redox couple and supporting electrolyte is often a complex task and normally involves screening many redox couples through various levels of testing. It is common to test, at minimum, the redox couple solubility, electrochemical activity/reversibility using cyclic voltammetry and chemical/electrochemical stability with other fuel cell components or chemicals. A few notable publications on the selection of various redox couples were given in the 1970s [24, 25].  1.3.3 The Fe/Cr Redox Flow Battery The first serious attempt to find a suitable set of redox couples for use in a RFB resulted in the classic Fe/Cr system at the Lewis Research Center, USA [25]. It is worthwhile to elaborate on the details of this system as the challenges faced by this RFB led to several advancements in RFB technology which are of importance today. The Fe/Cr RFB is based on the following electrochemical reactions:  The Fe/Cr RFB during discharge Anode +−+ ←+ 23 CreCr  E0= -0.42 V vs. SHE (1.89) Cathode +−+ →+ 23 eFeeF  E0= 0.67 V vs. SHE (1.90) Overall ++++ +→+ 2332 eFrCeFrC  E0= 1.09 V (1.91)   45 The Fe/Cr RFB employs an AEM and depends on Cl- ions for charge transfer across the membrane during discharge and regeneration. The anolyte (fuel) is prepared with a chromium chloride salt whereas the catholyte (oxidant) is prepared with an iron chloride salt.  Both electrolytes contain a high concentration of hydrochloric acid to facilitate the transport of Cl- ions across the membrane. A carbon substrate such as graphite felt is used at the cathode since the Fe2+/Fe3+ redox couple exhibits good electrochemical activity over this substrate. Since the Cr2+/Cr3+ redox couple shows poor reversibility over carbon substrates, a layer of gold is deposited on a carbon substrate in order to obtain acceptable performance. The Cr2+/Cr3+ redox couple delivers a good standard OCV (1.09 V) in the Fe/Cr RFB despite the rather low standard half-cell potential of the Fe2+/Fe3+ redox couple (0.68 V vs. SHE). However, the standard half cell potential of the Cr2+/Cr3+ redox couple (-0.42 V vs. SHE) is significantly lower than the standard half-cell potential for hydrogen production (0 V vs. SHE), which creates a situation where regenerating the Cr2+/Cr3+ redox couple may cause significant amounts of undesired hydrogen evolution to occur. The issue of hydrogen evolution over the fuel electrode during regeneration is further exacerbated by the presence of gold since a relatively low kinetic overpotential is sufficient to generate hydrogen over gold. This creates a situation where the following two reactions are in competition at the fuel electrode during regeneration:  46 Competition at the fuel electrode of a Fe/Cr RFB during regeneration Primary +−+ ←+ 23 CreCr  E0= -0.42 V vs. SHE (1.92) Secondary 2H2 1eH →+ −+  E0= 0 V vs. SHE (1.93)  When undesirable secondary reactions compete with the primary reaction, the current efficiency drops below 100%, which directly affects the energy efficiency of the system. Undesirable secondary reactions also cause the anolyte and catholyte to become imbalanced and have asynchronous states of charge (SOCs). In the absence of secondary reactions in a RFB, the SOC of the anolyte and catholyte are equivalent throughout the discharge and regeneration cycles. When the SOCs of the electrolytes become asynchronous, the total capacity of the RFB to store and release electrical energy is compromised. Mitigating this issue involves both preventing the occurrence of secondary reactions and correcting the SOCs of the electrolytes. In the case of the Fe/Cr RFB, a novel approach was devised to significantly reduce the amount of hydrogen generated during regeneration [24]. The approach involved adding a small quantity of PbCl2 to the fuel electrolyte in order to invoke deposition of lead on to the gold-coated fuel electrode during regeneration according to the following reaction:  Deposition of Pb on the fuel electrode of a Fe/Cr RFB during regeneration Reduction Pbe2Pb2 →+ −+  E0= -0.12 V vs. SHE (1.94)  The purpose of this deposition process is to introduce an electrode surface during regeneration that inhibits the production of hydrogen. It is well  47 established in the electrochemical literature that hydrogen production over lead requires a high kinetic overpotential. Since the Cr2+/Cr3+ redox couple maintains a reasonable level of electrochemical activity over the lead electrode, a differential advantage of Cr3+ reduction over H2 production can be realized by the deposition of lead. The deposition of lead is able to occur due to the negatively shifted potential of the fuel electrode during regeneration. During discharge of the Fe/Cr RFB, the fuel electrode becomes positively polarized causing the lead on the fuel electrode to oxidize and dissolve back into solution.  This process exposes the underlying gold coating and allows for the fuel oxidation reaction (Cr2+→Cr3++e-) to proceed with a lower overpotential. The lead plating/deplating cycle occurs as the Fe/Cr RFB undergoes charge/discharge cycles and represents a novel way to utilize the high activity of the gold electrode during discharge while retaining the ability to impede the hydrogen production reaction over lead during regeneration. The amount of lead deposited and oxidized is extremely small (approx. 0.1 mg/cm²) and therefore does not affect the current efficiency at the fuel electrode to any measurable extent. Although the approach involving lead deposition and dissolution is very effective in reducing the amount of hydrogen production in the Fe/Cr RFB, it does not eliminate it completely. Consequently, the SOC of the anolyte will eventually deviate negatively from that of the catholyte over many charge/discharge cycles. At some point, rebalancing of the cell will need to be performed in order to recover the original full capacity of the RFB. One approach is to take the  48 hydrogen generated at the fuel electrode during regeneration and use it to electrochemically reduce the ferric ions in the catholyte using an external “rebalancing cell” [27]. In this way, the SOC of the Fe2+/Fe3+ catholyte can be corrected (i.e., lowered) to equal that of the Cr2+/Cr3+ anolyte according to the following electrochemical reactions:  The rebalancing cell within a Fe/Cr RFB system Anode 2H2 1eH ←+ −+  E0= 0 V vs. SHE (1.95) Cathode +−+ →+ 23 eFeeF  E0= 0.67 V vs. SHE (1.96) Overall +++ +→+ 232 eFHeFH21  E0= 0.67 V (1.97)  It would appear, in theory, that recycling the hydrogen in this manner would perfectly rebalance the SOC of the electrolytes but in practice a small amount of excess hydrogen (from an external source) is required due to loss of captured hydrogen and chemical oxidation of chromous at the anode due to air leaked into the system. The rebalancing process represents a performance loss but is a necessary measure in order to recover the maximum charge/discharge capacity of the Fe/Cr RFB. Overall, the rebalancing cell approach is rather elegant as only one rebalancing cell is required for a Fe/Cr RFB system with any size stack. Other approaches to rebalance the cell include deep-discharge and overcharging the cell. In order to implement basic control algorithms for a RFB system, it is necessary to monitor the SOC of the RFB in real time. Researchers pioneering the Fe/Cr RFB achieved this by supplying the Cr2+/Cr3+ anolyte and Fe2+/Fe3+  49 catholyte to an auxiliary electrochemical “blank” cell and measuring the OCV in real time [27]. Assuming that the anolyte and catholyte are balanced, it is possible to correlate the measured OCV from the “blank” cell to the SOC of the system using empirical data and/or the Nernst equation (1.24).  An example of a theoretical curve produced with the Nernst equation for the Fe/Cr RFB is shown in Figure 1.10. Since the cell potential varies with the SOC, it can be used as a calibration curve. The auxiliary cell can be very small with a low volume of flow as it does not draw current. Also, only one auxiliary cell is required for any size RFB stack since the anolyte and catholyte fed to each current-drawing cell are mixed in the anolyte and catholyte storage tanks, respectively. The dependence of the cell voltage on the SOC of a RFB creates a challenge with respect to power control. As the system is discharged and the SOC declines, so does the power output for a given set of operating conditions. This issue was solved early on for the Fe/Cr RFB by the use of “trim cells”, which are additional electrochemical cells installed on the end of a stack. When the SOCs of the electrolytes are near 100%, the additional trim cells are left unused with no reactants flowing through them. As the SOCs of the electrolytes declines as the reactants are consumed, the corresponding voltage drop is compensated for by activating the trim cells one by one as needed. Trim cells are activated by flowing reactants through them. With this approach, the voltage and power output of a stack may be regulated within a set of limits.  50 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30 0 20 40 60 80 100 State of Charge (%) Eq ui lib riu m  C el l P ot en tia l, E e  (V )  Figure 1.10. Equilibrium cell potential of a Fe/Cr RFB vs. the state of charge at 25°C.  Several key advancements in RFB technology were made during the development of the Fe/Cr RFB, including trim cells, rebalancing cells and SOC readout cells. These advancements eventually led to the development of a 1 kW pilot-scale Fe/Cr RFB [58]. However, some issues persisted and hindered the commercialization of the Fe/Cr RFB. Most significant was the high resistivity of the membrane (approx. 2 Ω·cm²), which arose due to the trade-off between membrane selectivity and membrane conductivity. Membrane selectivity was of paramount importance since the crossover of electroactive redox species represented a permanent loss in the charge/discharge capacity of the system.  51 Other major issues include the burden of the rebalancing cell, instability of the gold/lead catalyst and the need to improve the overall charge/discharge efficiency of the system. Furthermore, the need to improve the anode and cathode kinetics by increasing the temperature of the system compromises the selectivity of the membrane. Despite the progress of the Fe/Cr battery, the need for a new type of system free of these limitations became apparent.  1.3.4 The Vanadium Redox Flow Battery In the 1980s, researches at the University of New South Wales, Australia, recognized the limitations of the Fe/Cr RFB and developed a new type of RFB based on the four oxidation states of the vanadium ion [31]. This new RFB, termed the vanadium redox battery (VRB), was unique relative to all other types of RFBs since the electrochemical species at both the anode and the cathode were based on the same metal ion. This property alone has significant ramifications with respect to most of the issues persistent in the Fe/Cr RFB.  The electrochemical reactions for the VRB are as follows:  The vanadium redox battery during discharge Anode +−+ ←+ 23 VeV  E0= -0.26 V vs. SHE (1.98) Cathode OHVOe2HVO 2 2 2 +→++ +−++  E0= 1.00 V vs. SHE (1.99) Overall OHVOV2HVOV 2 23 2 2 ++→++ +++++  E0= 1.26 V (1.100)  The VRB contains carbon-based electrodes at both the anode and cathode as both vanadium redox couples exhibit sufficiently good  52 electrochemical activity over carbon. Hydrogen evolution at the fuel electrode during regeneration is not an issue for this RFB due to the high overpotential over carbon. By employing the same metal ion in the anolyte and the catholyte, crossover of the redox couple becomes much less of an issue. As a net migration of vanadium ions occurs from the anode to the cathode or vice versa over many charge/discharge cycles, one may simply mix the electrolytes in a discharged state, separate them and charge the resulting “over-discharged” electrolytes (i.e., mixtures of V3+/VO2+). This is a much more attractive solution compared to complete electrolyte replacement that would be required for a Fe/Cr RFB. Consequently, a membrane with lower selectivity and higher conductivity can be used in a VRB to improve the overall cell performance. In any case though, crossover of the redox couple will lead to some performance loss at each electrode and a reduction in the SOC of each electrolyte so caution must still be exercised to reduce crossover. The VRB showed promise to mitigate all of the major challenges associated with the Fe/Cr RFB. Without the issues of permanent capacity loss after redox ion crossover or hydrogen evolution during regeneration, higher charge/discharge cycle efficiencies and higher current densities became tangible. However, the VRB system brought new challenges such as water crossover, redox couple precipitation and membrane stability in the presence of the strong oxidant VO2+. Water transport across the membrane in a VRB is caused by the migration of solvated protons or vanadium ions through the membrane and also osmotic  53 pressure differences across the membrane [59]. As water accumulates in one electrolyte, it can cause the other to concentrate and precipitate. Generally speaking, water crossover due to proton migration is not a significant factor since an equal amount of water is transported in each direction within a full charge/discharge cycle. Also, the amount of water transport due to osmotic pressure differences can be kept low with proper pump and pressure control. The most significant long-term contributor to water crossover is net vanadium ion crossover. The V2+ and V3+ ions in the anolyte have a larger charge:diameter ratio than the VO2+ and VO2+ ions in the catholyte, allowing them to penetrate and cross the CEM more easily than the VO2+ and VO2+ ions. The V2+ and V3+ ions have relatively large solvation shells due to their relatively large charge to diameter ratio and thus carry a significant amount of water with them during crossover through the membrane. The degree of vanadium ion crossover is primarily a function of temperature, electrolyte composition, membrane thickness and type of membrane. Most efforts to minimize the degree of vanadium ion crossover have been targeted towards membrane development [35, 36, 38], which was also motivated by the separate objective of cost reduction. Salt precipitation can be an issue in the VRB under certain circumstances. The onset of salt precipitation is primarily a function of electrolyte composition, SOC, temperature and standing time. The solubility of V2+, V3+ and VO2+ ions behave similarly with respect to temperature and concentration of sulfuric acid: increased solubility with increasing temperature and decreasing sulfuric acid concentration [33]. The VO2+ ion, on the other hand, expresses the exact  54 opposite behaviour due to its ability to polymerize. High concentrations of sulfuric acid ensure that an excess of SO42- ions is present in the electrolyte, allowing for full coordination of the VO2+ ion which inhibits polymerization. As a result, a careful balance between the sulfuric acid concentration and temperature must be respected to ensure electrolyte stability. Kazacos et al. found that a catholyte with 2 M VO2+/VO2+ and 3-4 M H2SO4 delivered good stability over many charge/discharge cycles [32]. Reducing the length of time that the VRB stands with a high SOC also prolongs the stability of the catholyte. The strong oxidizing potential of the VO2+ ion limits the life span of the membrane. Not surprisingly, the SOC of the catholyte and cell temperature are key factors affecting the stability of the membrane. Lowering the total concentration of vanadium ions in the catholyte to reduce membrane degradation is not a logical option since maximizing the charge density of the catholyte through a high vanadium ion concentration is a primary objective for VRB development. Identifying membranes with a strong resistance against the VO2+ ion has been the main approach in addressing membrane degradation. Thus far, good stability has been observed for Nafion® 112, Selemion CMV (type 2), and crosslinked, sulfonated Daramic CEMs [38]. After many years of development, the VRB reached the commercial stage in the late 1990s. Various stationary applications such as backup power, load- levelling, peak-shaving, remote power and renewable energy storage have been identified for RFBs. VRBs have been installed all over the world, including Denmark, USA, Africa, Canada and Israel [60, 61]. Installations can fall within a  55 wide range of power to capacity ratios, depending on the application of the RFB. Examples of the extreme cases include a voltage sag protection project designed for 4 MW x 1.5 hr and university load levelling installation designed for 250 kW x 8 hr [62]. These projects demonstrate the power output and storage capacity flexibility of RFBs. Such flexibility is simply not possible with secondary battery technology. The ability for RFBs to store large quantities of electrical energy with high volume electrolyte storage tanks have made them particularly useful for renewable energy storage applications where other conventional approaches are not possible (e.g., absence of large body of water for pumped hydro). An example of how a VRB can be integrated with a renewable energy source to act as an electrical energy buffer is shown in Figure 1.11.  Figure 1.11. Schematic of a vanadium redox battery integrated with a renewable energy source.  56 1.4 The Direct Methanol Fuel Cell The direct methanol fuel cell (DMFC) has received rapidly increasing interest over the past few decades due to the growing demand for portable power and uncertainty whether secondary batteries will be able to meet the future demand for portable power. DMFCs are well suited for micro and portable applications due to the high volumetric charge/energy density of the methanol liquid fuel (3.95 kAh/L, 4.62 kWh/L) and the ease in which the methanol fuel can be handled, stored and transported. Relative to proton exchange membrane fuel cells (PEMFCs) powered with hydrogen, DMFCs have the advantage of requiring no auxiliary cooling or humidification equipment since the liquid anolyte addresses these issues autonomously. However, the power density achievable in DMFCs is several factors lower than that possible in PEMFCs. The DMFC utilizes methanol and oxygen from air to produce electrical energy according to the following electrochemical reactions:  The direct methanol fuel cell Anode OHOHCHe6H6CO 232 +←++ −+  E0= 0.06 V vs. SHE (1.101) Cathode OH3e66HO2 3 22 →++ −+  E0= 1.23 V vs. SHE (1.102) Overall OH2COO2 3OHCH 2223 +→+  E0= 1.17 V (1.103)  The above electrochemical reactions represent methanol oxidation and oxygen reduction in an acidic environment. Analogous reactions can be written  57 for alkaline conditions, which would then denote the direct methanol alkaline fuel cell (DMAFC). This thesis is concerned with only the conventional acidic DMFC.  1.4.1 Direct vs. Indirect Methanol Fuel Cells The word “direct” is an important term in “DMFC” in order to distinguish this system from the indirect methanol fuel cell (IMFC). The IMFC is actually a hydrogen PEMFC with a methanol reformer upstream to convert methanol into hydrogen. The indirect approach takes advantage of the facile storage, handling and distribution of the liquid methanol fuel as well as the high electrochemical activity of hydrogen. However, the IMFC is substantially complicated by the reformer, which operates near 20 bar and 250°C [63]. Two simple schematics outlining the basic components of the DMFC and IMFC are shown in Figure 1.12 and Figure 1.13, respectively. The methanol reformer in the IMFC promotes the following endothermic reaction:  Methanol reformation  22g2g3 H3COOHOHCH +→+ )()(  ∆H0= 49.2 kJ/mol (1.104) where ∆H0 is the enthalpy of the reaction at standard conditions.  The above reactions products can be separated with a hydrogen-selective PdAg membrane in order to obtain a nearly pure hydrogen feed for the fuel cell. However, not all of the hydrogen produced can be recovered through the PdAg membrane. Typically, the unrecoverable fraction of hydrogen in the H2/CO2 mixture is burned with air before discharge. Due to the complexity surrounding  58 the reformer in the IMFC, interest in the IMFC has declined significantly over the past 10 years. The majority of methanol fuel cell research is currently focused on the DMFC.  Figure 1.12. Simplified schematic of a direct methanol fuel cell.    59  Figure 1.13. Simplified schematic of an indirect methanol fuel cell.  1.4.2 Active vs. Passive DMFCs For some portable applications involving micro-electronics, low power outputs are acceptable and the capacity of the power supply paramount. In these cases, passive DMFCs are well suited for operation. Passive DMFCs are compact, simple and low-cost since the fuel and oxidant are supplied without the use of blowers or pumps and the fuel cell operates at ambient temperature. This normally implies that the air oxidant is supplied to an open cathode via diffusion and the fuel is either stored in the anode chamber or supplied through a wick [64, 65]. These conditions significantly simplify the operation of the fuel cell but restrict the power output substantially with current densities being limited to below 100 mA/cm². Passive DMFCs enable low-power devices to make use of  60 the high volumetric energy density of methanol without the need for auxiliary equipment such as heaters, pumps, blowers and flow controllers. Active DMFCs include all of the common auxiliary equipment needed to modify the operating conditions of the DMFC in order to obtain high power densities. Active DMFCs typically operate from 50 to 90°C and are able to reach current densities exceeding 500 mA/cm² [66-69]. At such elevated current densities, required reactant flow rates cannot be met by diffusion and natural convection alone. Active DMFCs must bear a parasitic load to power the auxiliary pumps and blowers with controllers. A variety of DMFC applications such as portable power and non-micro electronics are only accessible to active DMFCs. Based on the number of publications available (10,820 for active DMFCs and 974 for passive DMFCs [70]), it is apparent that active DMFCs have received significantly more attention than passive DMFCs.  1.4.3 Methanol Production The majority of methanol produced today is derived from methane obtained from natural gas. Unfortunately, it is not possible to partially oxidize methane directly to obtain methanol with high yield. First, methane must be converted to syn gas, a mixture of CO and H2, by reacting methane with steam over a Ni catalyst at 700-850°C:  Steam-methane reformation (SMR)  2g24 H3COOHCH +→+ )(  ∆H0= 206 kJ/mol (1.105)   61 Since the above reaction is highly endothermic, it is common to provide heat while producing syn gas through partial oxidation of methane in an oxygen deficient environment:  Partial oxidation (POX)  224 H2COO2 1CH +→+  ∆H0= -35 kJ/mol (1.106)  The combination of SMR and POX is known as auto-thermal reforming (ATR). Following conversion of methane into syn gas, methanol can be produced at 5- 10 MPa and 250°C over a catalyst mixture containing Cu, Zn, and alumina according to the following reaction:  Methanol production  OHCHH2CO 32 →+  ∆H0= -129 kJ/mol (1.107)  Lastly, the methanol must be separated from the mixture.  1.4.4 Methanol as a Fuel Within the class of low molecular weight organic liquid fuels, there are many candidates for fuel cell applications. Key parameters in determining the suitability of a particular fuel for use in a fuel cell include:  • The half-cell potential of the fuel • The number of electrons per mole of fuel  62 • The electrochemical activity of the fuel for a particular catalyst • The volumetric charge/energy density of the fuel • Fuel cell operation considerations • Cost, production and availability considerations  Table 1.8 is a list of some low molecular weight organic fuels available for use in fuel cells.  Table 1.8. A list of some low molecular weight organic liquid fuels and various properties at 25°C.          Formic acid has received much recognition as a viable fuel as it exhibits good electrochemical activity and has a lower crossover rate through Nafion® membranes than that of methanol [71]. However, its low volumetric charge and energy density represent a significant challenge. Ethanol and 2-propanol offer exceptional volumetric charge and energy densities but due to the C-C bond  63 present in the fuels, the electrochemical reaction kinetics are not favourable and complete oxidation is typically not possible. Incomplete oxidation directly affects the charge and energy density negatively, marginalizing the best feature these fuels have. The issue surrounding the C-C bond and incomplete oxidation can be addressed through the use of fuels with C-O-C bonds, such as dimethyl ether and dimethoxy methane. Dimethyl ether, for instance, exhibits good electrochemical activity, possesses a high volumetric charge and energy density, and does not oxidize at the oxygen cathode [72]. However, dimethyl ether is not produced in quantities comparable to methanol or ethanol and its cost may be too high for widespread use as a fuel. Dimethoxymethane unfortunately hydrolyzes to methanol during fuel cell operation and thus can not be used as a fuel for fuel cell operation [72]. The use of methanol as a fuel is not free of challenges. For instance its relatively high toxicity (mouse, oral LD50= 7300 mg/kg [73]) presents a challenge for widespread commercialization and a number of challenges related to fuel cell operation are present, as discussed below. However, it represents a fuel with a balanced set of features. The electrochemical activity of methanol compares well with the other organic fuels in addition to its high volumetric charge/energy density. It can undergo complete oxidation and since its global production levels are very high the cost is kept low. For these reasons, methanol has received the most attention out of all electroactive liquid organic fuels for use in a fuel cell. However, formic acid and ethanol are actively studied globally at this time for use as electroactive fuels.  64 1.4.5 The DMFC Anode Methanol oxidation requires the presence of a platinum group metal (PGM) catalyst for practical applications where reaction completion and electrochemical activity are paramount. Platinum alone provides good initial activity towards methanol oxidation but the electrochemical activity degrades over time as the CO coverage on the electrode surface increases [14]. A key mechanism for the removal of CO on Pt surfaces involves chemisorbed OH [15]. Consequently, metal catalysts which promote the adsorption of OH onto the electrode surface, such as Ru, Rh and Os, can be alloyed with Pt to enhance the electrochemical activity of methanol oxidation [74]. Perhaps the most significant milestone in the history of DMFC development to date occurred when the 1:1 mol/mol PtRu catalyst was identified as a bi-functional catalyst for methanol oxidation in the 1960’s [10]. To this day, the 1:1 mol/mol PtRu alloy stands as the international standard DMFC catalyst for methanol oxidation. It is typically applied to the anode at a loading of 2-8 mg/cm² [68, 75]. The reaction mechanism for methanol oxidation was generalized in the 1970’s by Bagotzky et al. according to Figure 1.14 [76]. The initial dehydrogenation of CH3OH to CH2OH is regarded as the rate determining step. Although a number of reaction pathways are possible, the detection of COH species on the Pt catalyst surface at lower oxidation potentials (below 0.7 V vs. SHE) suggests that the top-right outer path of Figure 1.14 is a key path for methanol oxidation over Pt.   65  Figure 1.14. Generalized reaction mechanism for methanol oxidation [76].  In later years, experimental techniques such as in-situ ellipsometry, in-situ Fourier transform infrared spectroscopy and x-ray absorption spectroscopy revealed the following methanol oxidation reaction mechanism for a 1:1 mol/mol PtRu catalyst [77, 78]: 1. −+ ++−→+ eHOHCHPtPtOHCH 23   (1.108) 2. −+ ++−→+− eHCHOHPtPtOHCHPt 2   (1.109) 3. −+ ++−→+− eHCHOPtPtCHOHPt   (1.110) 4a. −+ ++≡−→− eHOCPtCHOPt   (1.111) 4b. −+ ++=〉→+− eHOCPtCHOPt PtPt   (1.112) 5. −+ ++→+ eHRuOHOHRu 2   (1.113) 6. −+ ++++→−+ eHCOPtRuCOPtRuOH 2   (1.114)  The anode catalyst layer can be prepared by spraying, painting, printing or scraping a catalyst ink on to a catalyst support structure such as CFP or the membrane. The ink normally consists of water, alcohol, catalyst and a binder. The catalyst can be incorporated as a pure catalyst (e.g., PtRu black) or a supported catalyst. Supported catalysts are typically nano-particles of carbon  66 decorated with a metal catalyst, as shown in Figure 1.15. Supported catalysts are used to increase the utilization of PGM catalysts. Polytetrafluoroethylene (PTFE) or Nafion® ionomer can be used as a binder for the catalyst layer, depending on the properties desired for the final product.   Figure 1.15. An illustration of a carbon supported PtRu catalyst particle.  The removal of CO2 from the anode catalyst layer and diffusion layer is a critical function that plays a key role in the anode performance. Adjusting the hydrophobicity of the catalyst layer through the use of PTFE and Nafion® ionomer provides a tool to enhance CO2 gas removal from the catalyst layer [79]. However, a balance must be achieved as transport of the CH3OH/H2O anolyte is favoured by hydrophilic pores and transport of CO2 is favoured by hydrophobic pores. An ideal anode catalyst layer might be one which consists of many mutually exclusive hydrophilic and hydrophobic pores to accommodate simultaneous transport of the reactants to and products away from the electrode surface. The Nafion® content of the anode catalyst layer also affects the proton conductivity within the catalyst layer, although this effect is of little importance  67 when the proton conductivity is already established through the use of an acid in the liquid anolyte.  Other important aspects of the DMFC anode include control of the methanol concentration in the anolyte and CO2 separation from the anolyte. As shown in Figure 1.12 earlier, the CH3OH/H2O anolyte is recycled which requires that CH3OH and H2O be maintained at their desired concentrations while CO2 is purged via a gas/liquid separator. Recycling the anolyte permits high anolyte flow rates to enhance CO2 removal within the anode catalyst layer and diffusion layer without compromising fuel utilization. The methanol concentration in the anolyte recycle loop is maintained through the use of a methanol sensor, valve and controller. The fuel supply tank may contain a concentrated methanol anolyte or pure methanol, depending on how much water produced at the cathode is redirected to the anode during fuel cell operation.  1.4.6 The DMFC Cathode The oxygen reduction reaction requires the use of PGM catalysts to deliver high performance in acid conditions at practical current densities. The most commonly employed cathode catalyst is Pt black or carbon supported Pt. The design and fabrication of the cathode catalyst layer is complicated by the need for TPB sites. In addition, facilitating efficient product water removal is critical to obtain a high-performance cathode. According to equation 1.102, the oxygen reduction half-cell reaction, protons and oxygen must meet at the electron-conducting catalyst surface in  68 order for this reaction to proceed. The intersecting regions where these three phases meet make up the TPB sites. In order to establish the TPB, the porous catalyst layer must be integrated with a proton conducting medium, such as Nafion® ionomer. Similar to the anode catalyst preparation method described, the Nafion® ionomer is added to the catalyst ink prior to spraying, printing, painting or scraping it on to the catalyst substrate. The proton conductivity within the catalyst layer can be integrated with the proton conductivity within the membrane by hot- pressing the membrane electrode assembly (MEA) at 135°C [79]. This temperature exceeds the glass-transition temperature for the Nafion® ionomer and the Nafion® becomes more fluid-like. The weight percent of ionomer incorporated into the cathode catalyst ink is a critical parameter, often falling in the range of 10-30%. Water management at the cathode is essential for two reasons: (1) to ensure that reactant access to the cathode catalyst sites is maintained, and (2) to redirect the water to the anode where it is needed as a reactant. Without proper cathode water management, the cathode will be rapidly flooded, leading to high cathode overpotentials, and the product water will be unavailable to the anode, precluding the possibility of using a concentrated methanol feed at the anode. Cathode flooding is exacerbated by the electro-osmotic drag of water by protons from the anode to the cathode, typically in the amount of 2.5 H2O molecules per proton [80]. Cathode water management is particularly crucial at high current densities where the amount of water produced at the cathode and electro- osmotically dragged to cathode are high. A common approach to mitigate this  69 issue is to incorporate PTFE into the cathode catalyst layer to increase the hydrophobicity and promote water removal. Furthermore, a microporous layer (MPL) consisting of carbon and PTFE can be applied to the cathode, as shown in Figure 1.16, to increase the hydraulic pressure of water and promote convective transport of water through the membrane to the anode [53]. This approach effectively supplies the anode with reactant water, which is imperative when a concentrated methanol anolyte is supplied to the anode.        Figure 1.16. Schematic of a cathode catalyst layer with a microporous layer.  1.4.7 DMFC Membranes At this time, most DMFCs employ Nafion® 117 (175 µm thick) as the membrane due to its high proton conductivity (83 mS/cm [81]), durability and chemical compatibility. Nafion® is perfluourosulfonic acid membrane which incorporates a PTFE backbone, as shown in Figure 1.17.  A B C D A: membrane B: catalyst layer C: dense microporous layer D: catalyst substrate  70  Figure 1.17. The molecular structure of the Nafion® membrane.  However, many researchers have studied other membranes in order to circumvent the high cost and high methanol permeability of Nafion®. In particular, Carretta et al. reported that highly sulfonated polystyrene exhibited a conductivity equivalent to Nafion® with a 70% lower methanol crossover rate [82]. The BASF Company recently developed a polybenzimidazole based membrane named Celtec-V which reportedly outperforms Nafion® 117 at methanol concentrations exceeding 1 M [83]. Jörissen et al. showed good themal stability, chemical inertness, proton conductivity and fuel cell performance for a number of polymer blends involving sulfonated polyether-ether-ketone, poly(4-vinylpyridine), polybenzimidazole and polysulfones [84]. An excellent review on DMFC membranes is provided by Neburchilov et al. [85].  1.4.8 Methanol Crossover in DMFCs It has been observed that the most frequently employed membrane separator for DMFC’s, Nafion®, permits high rates of methanol diffusion [86]. For example, Lam et al. reported that the methanol crossover rate at 25°C through Nafion 117 was 1.4x10-3, 4.8x10-3 and 6.9x10-3 mol/m²·s at methanol  71 concentrations of 1.9, 6.2 and 7.8 M, respectively [87]. These methanol crossover rates can also be converted to short-circuit current densities of 81, 278 and 400 mA/cm², respectively. Methanol crossover is undesirable as it represents a fuel loss and can dramatically affect the system efficiency and performance. Once methanol reaches the cathode, it has the opportunity to be oxidized by oxygen over the platinum catalyst and create a mixed (reduced) potential at the cathode. Under certain conditions, the crossed-over methanol can consume a large fraction of the oxygen available at the cathode and cause oxygen starvation, which can have a profound negative impact on both the kinetic overpotential and mass-transport overpotential according to equations 1.52 and 1.72, respectively. The above phenomena can significantly reduce the power output and efficiency of the fuel cell. Consequently, suppressing methanol crossover is vital to the efficiency and performance of DMFC’s. The most common approaches to address methanol crossover include reducing the methanol concentration to around 1-2 M and utilizing a thicker membrane (e.g., Nafion® 117 instead of Nafion® 112) [88]. Since methanol undergoes Fickian diffusion through the membrane, these approaches are effective in reducing the methanol crossover rate. However, lowering the methanol concentration has a negative impact on the anode kinetics and complicates the supply of methanol from a concentrated methanol storage tank. Using a thicker membrane increases the resistance of the membrane, causing higher ohmic losses.  72 Some novel approaches to mitigating methanol crossover include the use of anode diffusion barriers and methanol-tolerant cathode catalysts. Lam and Wilkinson demonstrated the benefits of an extended 3D anode in a passive DMFC [89, 90]. The modified anode comprised multiple layers of CFP substrates loaded with PtRu catalyst in order to both hinder the diffusion of methanol and reduce the concentration of methanol substantially by the time it reaches the membrane. This approach significantly reduced methanol crossover, which was demonstrated through methanol crossover and fuel cell tests. A number of methanol-tolerant cathode catalysts have been identified as a suitable replacement for Pt. In particular, iron tetramethoxyphenylporphyrin (Fe- TMPP), shown in Figure 1.18, and RuSe based catalysts have shown promising selective oxygen reduction activity in the presence of methanol [91]. Gojkovic et al. studied the activity of Fe-TMPP and found comparable performance to that of carbon supported Pt [92]. Tributsch et al. found similar results for the RuSeO selective catalyst [93]. Although these catalysts are effective in eliminating cathode depolarization due to fuel crossover, they do not address fuel loss.  73         Figure 1.18. The molecular structure of iron tetramethoxyphenylporphyrin.  1.4.9 DMFC Operating Conditions The cell temperature, methanol anolyte flow rate, anolyte methanol concentration, anode pressure, air flow rate and cathode pressure significantly influence DMFC performance. Surampidi et al. investigated DMFC performance at temperatures of 30, 60 and 90°C and found a substantial improvement at higher temperatures [94]. The temperature of a DMFC influences a multitude of phenomena, including methanol crossover, reaction kinetics, oxygen partial pressure, water evaporation, membrane stability, cell resistance and mass transfer limiting current densities. Ge and Liu conducted an elaborate analysis on the effects of cell temperature, methanol concentration, methanol flow rate, air flow rate and humidifier temperature [69]. Nafion® 117 was used as the polymer electrolyte between serpentine flow fields and catalyst loadings were generous at 3 mg/cm²  74 for both Pt/Ru black and Pt black at the anode and cathode, respectively. It was previously mentioned that DMFC’s are advantageous in that they do not require external humidification. This is confirmed within Figure 1.19. The influence of cathode humidification on cell performance is small since membrane hydration is maintained through the use of a liquid aqueous anolyte. A small decrease in performance is observed at high humidifier temperatures due to the onset of cathode flooding.   Figure 1.19. Effect of cathode humidifier temperature. (70°C cell temperature, 6 mL/min CH3OH flow rate, 2 M CH3OH, 600 sccm air flow rate, air pressure unavailable) [69]. Reprinted with permission from Elsevier.  75 The effect of methanol flow rate is summarized in Figure 1.20. As expected, the mass transfer limiting current density increases with increasing anolyte flow rate and exhibits a diminishing gain with higher flow rates. The abnormal regression of the potentiostatically measured current density at low cell voltages and flow rates below 3 mL/min is likely due to reduced electrochemically active area from CO2 product accumulation at the anode.  Figure 1.20. Effect of methanol flow rate. (70°C cell temperature, 1 M CH3OH, 1200 sccm air flow rate, 70°C cathode humidifier temperature, air pressure unavailable) [69]. Reprinted with permission from Elsevier.  76 Figure 1.21 depicts the strong positive influence of temperature on the reaction kinetics. At higher temperatures, various negative phenomena become significant such as increased methanol crossover, decreased CO2 solubility and higher water vapour partial pressure in the cathode, causing the cell performance to plateau at 70°C. A decrease in CO2 solubility will cause a larger fraction of the CO2 evolved to exist in the gas phase, which can lead to a reduction in accessible active area on the anode and inhibit the methanol migration to the anode catalyst surface. For the system studied, the best cell temperature would be roughly 70°C. The effect of methanol concentration on cell performance is shown in Figure 1.22, which highlights the significance of methanol crossover. At 0.5 M, the cell quickly undergoes methanol starvation and exhibits a low mass transfer limiting current density. On the other hand, it is interesting to note that methanol concentrations above 3 M exhibit poorer performance than that at 0.5 M as a result of accelerated methanol crossover, despite the improvement in methanol oxidation kinetics. Thus, a balance between kinetics and methanol crossover must be respected. Additionally, the concentration of methanol in the fuel storage tank and the amount of product water that will be recycled from the cathode to the anode are important considerations when selecting the methanol concentration.  77  Figure 1.21. Effect of cell temperature. (3 M CH3OH, 600 sccm air flow rate, 4 mL/min methanol flow rate, 30°C cathode humidifier temperature, air pressure unavailable) [69]. Reprinted with permission from Elsevier.  78  Figure 1.22. Effect of methanol concentration. (70°C cell temperature, 6 mL/min CH3OH, 600 sccm air flow rate, 70°C air humidifier temperature, air pressure unavailable) [69]. Reprinted with permission from Elsevier.  79 1.5 Mixed-Reactant Fuel Cells Conventional fuel cells require the supply of two separate reactant streams, the fuel and the oxidant, as shown in Figure 1.23. The conventional architecture functions well but care must be taken to ensure that the fuel does not reach the cathode and the oxidant does not reach the anode. If this requirement is not met, issues similar to those present during methanol crossover such as electrode depolarization will manifest. It follows that stringent reactant sealing and a highly selective membrane are necessary to obtain good performance with conventional fuel cell architectures. However, the selectivity of a membrane is often far from perfect and significant performance losses occur due to reactant crossover. This issue represents one of the motivations behind mixed-reactant fuel cell (MRFC) architectures.   Figure 1.23. Schematic of a conventional fuel cell with two reactant streams.  All MRFC architectures generally fall within one of two categories defined by the orientation of the electrodes: (1) transverse and (2) co-planar. Both architectures require the presence of fully selective anode and cathode catalysts, which ensure that the fuel is inactive at the cathode and the oxidant is inactive at  80 the anode. It is important to note that the selectivity of an electrode can also be established physically rather than catalytically. Physical selectivity implies that an undesirable reactant (e.g., the oxidant at the anode) is denied physical access to the electrode. This approach is typically used when two phases are present and the hydrophobicity of an electrode surface can be modified to significantly limit the access of an undesirable reactant. If both electrodes can be designed to be fully selective, a mixed-reactant stream can be supplied to both electrodes without causing depolarization at either electrode. The use of a mixed-reactant supply introduces some attractive features, such as less stringent sealing, simpler manifolding, improved cell compactness and greater cell design flexibility. MRFCs based on the transverse electrode architecture are structurally similar to conventional fuel cells but differ in the reactant supply, as shown in Figure 1.24. In this mixed-reactant architecture, charge carrying ions migrate through the plane of the ionic conductor and the thickness of the ionic conductor has a significant influence on the ohmic resistance of the cell.    Figure 1.24. Schematic of a mixed-reactant fuel cell with transverse electrodes.   81 MRFCs incorporating the co-planar electrode architecture are rather unique in comparison to conventional fuel cell architectures, as shown in Figure 1.25. The anode and cathode reside on the same side of the ionic conductor and the fuel/oxidant mixture is supplied to only one side of the ionic conductor. Consequently, charge-carrying ions must migrate in the plane of the ionic conductor during fuel cell operation. The design of a co-planar electrode architecture is very different from that of conventional architectures since new design parameters such as the inter-electrode gap length and electrode width are present. These two parameters are of paramount importance due the direction in which the charge-carrying ions migrate and significantly influence the ohmic losses for this architecture. One advantage to the co-planar electrode architecture is that it lends itself well to printing and etching electrode fabrication techniques.   Figure 1.25. Schematic of a mixed-reactant fuel cell with co-planar electrodes.  82 1.5.1 Mixed-Reactant Direct Liquid Fuel Cells Methanol crossover is a severe problem present in direct methanol fuel cells (DMFCs) as it leads to issues such as cathode depolarization and fuel loss. Selective anode and cathode catalysts can be employed in a DMFC to adopt a mixed-reactant architecture and eliminate the issues associated with methanol crossover. Since DMFCs are fed with a liquid CH3OH/H2O anolyte and air catholyte, the presence of two phases may be exploited to achieve physical selectivity at the anode or cathode. As many selective oxygen reduction catalysts are known to the scientific community, cathode selectivity in a DMFC is usually established catalytically while highly hydrophilic catalyst layers incorporating Nafion®  ionomer are used to achieve physical selectivity at the anode. Since the CH3OH/H2O DMFC anolyte is comprised of polar molecules, a hydrophilic anode surface would be strongly wetted by this electrolyte, inhibiting gaseous oxygen from reaching the anode. Barton et al. demonstrated the selectivity of various electrodes useful for mixed-reactant DMFCs using half-cell measurements at 80°C [95]. A physically selective hydrophilic PtRu/C anode incorporating Nafion® ionomer was tested with three anolytes: (a) CH3OH/H2O, (b) CH3OH/H2O + N2 and (c) CH3OH/H2O + air. Cases (b) and (c) produced essentially identical anode performance but showed poorer anode polarization behaviour than case (a) at higher current densities (above 100 mA/cm²). The shape of the anode polarization suggests that mass transport overpotential is responsible for the deviation. This work indicates that performance loss at higher current densities for a mixed-reactant  83 anode relative to that of a non-mixed-reactant anode are due to only reactant dilution and not anode depolarization. Similar tests were conducted for catalytically selective RuSeMo and Fe-TMPP cathodes. The cathode polarization behaviour in a half-cell configuration was measured for catholytes containing (1) air and (2) air + CH3OH/H2O. It was observed that the presence of the fuel had little (less than a 30 mV drop) or no effect on the cathode polarization behaviour, which demonstrates the selective nature of these oxygen reduction catalysts and suggests that dilution effects alone are responsible for the discrepancy in cathode performance. Later work by Barton et al. led to the same conclusions for Co-TMPP and a commercial metal chalcogenide cathode catalyst [96]. The performance of a mixed-reactant DMFC at 90°C employing various catalytically selective cathode catalysts (Co-TMPP, Fe-TMPP, FeCo-TMPP and RuSe/C) was established by Shukla et al. [97]. It was observed that the performance of the RuSe/C cathode catalyst was significantly higher than all of the TMPP-based cathode catalysts. A peak power density of 30 mW/cm² was achieved using air mixed with 1 M CH3OH. It is possible for a mixed-reactant fuel cell to employ only physically selective catalysts, as was demonstrated by Zeng et al. for an alkaline ethanol/air fuel cell [98]. A hydrophilic PtRu/C anode and hydrophobic Pt/C cathode were supplied with a 0.5 M ethanol / 2 M KOH / air mixed electrolyte and a peak power density of 31 mW/cm² was achieved.   84 1.6 Research Objectives The main objective of this research is to demonstrate and develop a direct liquid redox fuel cell (DLRFC) which utilizes a redox couple at the cathode and high energy density liquid methanol/water solution at the anode. The project research involves the following components:  1) Demonstrate a high fuel concentration DLRFC incorporating a conventional cell architecture. • Selection and electrochemical characterization of a suitable redox couple • Fuel cell testing of a high fuel concentration DLRFC  2) Characterize and improve the performance of the DLRFC. • Identify suitable operating conditions for the DLRFC through fuel cell tests • Identify and test opportunities to improve the charge density of the DLRFC catholyte  3) Demonstrate and electrochemically characterize a mixed-reactant DLRFC utilizing a novel cell architecture where methanol crossover is desirable.  4) Investigate the feasibility of electrochemical DLRFC catholyte regeneration. • Conduct fuel cell testing to establish the performance of electrochemical DLRFC catholyte regeneration using air as the oxidant.  85 1.7 Thesis Layout In Chapter 2, a direct liquid redox fuel cell (DLRFC) utilizing a CH3OH / H2O / H2SO4 electrolyte at the anode and a Fe2+/Fe3+ redox couple at the cathode is demonstrated. The absence of Pt catalyst at the carbon-based cathode and the ability to use high fuel concentrations (up to 16.7 M methanol) without introducing cathode depolarization is highlighted. The material in this chapter has been published: • A. B. Ilicic, D. P. Wilkinson, K. Fatih, F. Girard, “High Fuel Concentration Direct Liquid Redox Fuel Cell with Redox Couple Cathode”, J. Electrochem. Soc., 155, 12 (2008) B1322-B1327.  In Chapter 3, significant improvements for the DLRFC are obtained with respect to cell performance and catholyte charge density through the use of catholytes prepared with iron salts containing different anions (e.g., FeNH4(SO4)2, Fe(NO3)3, Fe(ClO4)3). The effect of the cell temperature and anolyte methanol concentration on the DLRFC performance is also studied. The material in this chapter has been submitted for publication: • A. B. Ilicic, M. S. Dara, D. P. Wilkinson, K. Fatih, “Improved Performance of the Direct Methanol Redox Fuel Cell”,  In Chapter 4, the concept of a mixed-reactant DLRFC is demonstrated using a novel cell architecture where methanol crossover is desirable as it represents the method of fuel supply. The mixed-reactant DLRFC performance is  86 characterized for different cell temperatures and methanol concentrations in the mixed electrolyte. In-situ DLRFC catholyte regeneration is also demonstrated using a novel cell architecture. The material in this chapter has been accepted for publication: • A. B. Ilicic, D. P. Wilkinson, K. Fatih, “Advancing Direct Liquid Redox Fuel Cells: Mixed-Reactant and In-situ Regeneration Opportunities”, J. Electrochem. Soc., 157, 4 (2010)  In Appendix F, a manuscript pertaining to the hydrogen redox fuel cell (HRFC), a system similar to the DLRFC, is included. The HRFC is powered with hydrogen fuel and the Fe2+/Fe3+ redox couple at the anode and cathode, respectively, and demonstrates some of the advantages present with the DLRFC. In this work, the Fe2+/Fe3+ redox couple is electrochemically characterized using cyclic voltammetry and the effect of cell temperature and cathode material on the performance of the HRFC is investigated. The material in this section has been published: • K. Fatih, D. P. Wilkinson, F. Moraw, A. Ilicic, F. Girard, “Advancements in the Direct Hydrogen Redox Fuel Cell”, Electrochem. Solid-State Lett., 11, 2 (2008) B11-B15.  87 1.8 References 1. W. R. Grove, Philos. Mag., 14, (1839) 127-130. 2. C. F. Shönbein, Philos. Mag., 14, (1839) 43-45. 3. W. R. Grove, Philos. Mag., 21, (1842) 417-420. 4. L. Mond, C. Langer, Proc. Roy. Soc. London, 46, (1889) 296-304. 5. Anon., Business Week, 33, (1959). 6. K. V. Kordesch, J. Electrochem. Soc., 125, 3 (1978) 77C-88C. 7. Grubb, W. T., US Pat. 2,913,511 (1959). 8. W. T. Grubb, L. W. Niedrach, J. Electrochem. Soc., 107, 2 (1960) 131-135. 9. L. E. Chapman, in Proc.7th Intersociety Energy conversion Engineering conference, PV, pp. 466-471 (1972). 10. M. R. Andrew, R. W. Glazebrook, An Introduction to Fuel Cells, Elsevier, (1966), p. 127. 11. V. S. Bagotzky, Y. Vassilyev, Electrochim. Acta, 12, 9 (1967) 1323-1343. 12. M. W. Breiter, J. Electroanal. Chem., 14, 4 (1967) 407-413. 13. M. W. Breiter, J. Electroanal. Chem., 15, (1967) 221-226. 14. T. Biegler, D. F. A. Koch, Electrochem. Soc., 114, 9 (1967) 904-909. 15. M. W. Breiter, J. Phys. Chem., 72, 4 (1968) 1305-1313. 16. M. W. Breiter, J. Electroanal. Chem., 19, 1-2 (1968) 131-136. 17. H. A. Liebhafsky, E. J. Cairns, Fuel Cells and Fuel Batteries, John Wiley and Sons, New York (1968), p. 27. 18. A. M. Posner, Fuel, 34, 3 (1955) 330-338.  88 19. E. C. Potter, Electrochemistry: Principles and Applications, Cleaver-Hume Press Ltd., (1956), p. 405. 20. B. Folkesson, J. App. Electrochem., 20, 6 (1990) 907-911. 21. R. Larsson, B. Folkesson, J. App. Electrochem., 20, 5 (1990) 737. 22. J. T. Kummer, D. G. Oei, J. App. Electrochem., 12, (1982) 87-100. 23. J. T. Kummer, D. G. Oei, J. App. Electrochem., 15, 4 (1985) 619-629. 24. J. Giner, L. Swette, K. Cahill, "Screening of Redox Couples and Electrode Materials" NASA-CR-134705, 1976. 25. L. H. Thaller, in , PV, pp. 924-928, 9th Intersociety Energy Conversion Engineering Conference Proceedings, San Francisco, CA, USA, (1974). 26. L. H. Thaller, in , PV I, pp. 715-721, Proceedings of the 14th Intersociety Energy Conversion Engineering Conference, Boston, MA, USA, (1979). 27. L. H. Thaller, "Redox flow cell energy storage systems" DOE NASA 1002-79 3; E-9996; NASA-TM-79143, 1979. 28. L. H. Thaller, "Redox Flow Cell Development and Demonstration Project" NASA-TM-79067, 1977. 29. M. Warshay, L. O. Wright, Systems - Computers - Controls (1976) 130-140. 30. P. O'Donnell, R. F. Gahn, W. Pfeiffer, in , PV, pp. 733-736, Twelfth IEEE Photovoltaic Specialists Conference 1976, Baton Rouge, LA, USA, (1976). 31. M. Kazacos, M. Rychcik, R. G. Robins, A. G. Fane, M. A. Green, J. Electrochem. Soc., 133, 5 (1986) 1057-1058. 32. M. Kazacos, M. Cheng, M. Skyllaskazacos, J. App. Electrochem., 20, 3 (1990) 463-467.  89 33. M. Skyllas-Kazacos, C. Peng, M. Cheng, Electrochem. Solid-State Lett., 2, 3 (1999) 121-122. 34. M. Rychcik, M. Skyllas-Kazacos, J. Power Sources, 19, 1 (1987) 45-54. 35. T. Mohammadi, M. Skyllas-Kazacos, J. Power Sources, 56, 1 (1995) 91-96. 36. T. Mohammadi, M. Skyllas-Kazacos, J. Membr. Sci., 98, 1-2 (1995) 77. 37. T. Mohammadi, M. Skyllas Kazacos, J. Power Sources, 63, 2 (1996) 179-186. 38. T. Mohammadi, M. Skyllas Kazacos, J. App. Electrochem., 27, 2 (1997) 153- 160. 39. E. Sum, M. Skyllas-Kazacos, J. Power Sources, 15, 2-3 (1985) 179-190. 40. E. Sum, M. Rychcik, M. Skyllas-Kazacos, J. Power Sources, 16, 2 (1985) 85- 95. 41. S. Zhong, M. Skyllas-Kazacos, J. Power Sources, 39, 1 (1992) 1-9. 42. M. Kazacos, D. Kasherman, D. R. Hong, M. Kazacos, J. Power Sources, 35, 4 (1991) 399-404. 43. R. L. Largent, M. Skylas-Kazacos, J. Chieng, in Proceedings of 23rd IEEE Photovoltaic Specialists Conference, PV, pp. 1119-1124, Conference Record of the Twenty Third IEEE Photovoltaic Specialists Conference - 1993 (Cat. No.93CH3283-9), Louisville, KY, USA, (1993). 44. J. Larminie, A. Dicks, Fuel Cell Systems Explained, John Wiley & Sons Ltd., (2003). 45. D. R. Lide, CRC Handbook of Chemistry and Physics, CRC Press LLC, Boca Raton (2005).  90 46. D. J. G. Ives, G. J. Janz, Reference Electrodes, Theory and Practice, New York Academic Press, New York (1961). 47. S. Brunauer, P. H. Emmett, E. Teller, J. Am. Chem. Soc., 60, (1938) 309. 48. T. Erdey-Gruz, M. Volmer, Z. Phys. Chem., 150, A (1930) 202-213. 49. E. Gyenge, CHBE 477: Fuel Cells and Electrochemical Systems Course Notes (2004). 50. M. R. Wright, An Introduction to Aqueous Electrolyte Solutions, Wiley, (2007). 51. A. Koshev, V. Varentsov, M. Chirkina, Protection of Metals and Physical Chemistry of Surfaces, 45, 4 (2009) 494-500. 52. K. E. Mosley, IEE Proceedings A (Physical Science, Measurement and Instrumentation, Management and Education, Reviews), 133, 6 (1986) 375- 386. 53. F. Q. Liu, G. Q. Lu, C. Y. Wang, J. Electrochem. Soc., 153, 3 (2006) A543- A553. 54. K. Fatih, D. P. Wilkinson, F. Moraw, A. Ilicic, F. Girard, Electrochem. Solid- State Lett., 11, 2 (2008) B11-B15. 55. N. Tokuda, M. Furuya, Y. Kikuoko, Y. Tsutui, T. Kumamoto, in , PV vol.3, pp. 1144-1149, Proceedings of the Power Conversion Conference-Osaka 2002 (Cat. No.02TH8579), Osaka, Japan, (2002). 56. M. Lopez-Atalaya, G. Codina, J. R. Perez, J. L. Vazquez, A. Aldaz, Journal of Power Sources, 39, 2 (1992) 147-154. 57. Y. Y. Wang, M. R. Lin, C. C. Wan, Journal of Power Sources, 13, 1 (1984) 65- 74.  91 58. M. A. Reid, L. H. Thaller, in , PV, pp. 1471-1476, "Energy to the 21st Century". Proceedings of the 15th Intersociety Energy Conversion Engineering Conference, Seattle, WA, USA, (1980). 59. T. Mohammadi, S. C. Chieng, M. S. Kazacos, J. Membr. Sci., 133, 2 (1997) 151-159. 60. Prudent Energy Inc., Projects and Installations, 1-1-2010, http://www.pdenergy.com/en/applications- solutions/projects_installations/by_vrb/by_vrb_1.html 61. Prudent Energy Inc., Prudent Energy: Capture the Wind, 1-1-2010, http://www.pdenergy.com/en/upload/pdf/brochure-wind.pdf 62. Prudent Energy Inc., Prudent Energy Provides Energy Storage and Power Quality Solutions, 1-1-2010, http://www.pdenergy.com/en/upload/pdf/brochure-company.pdf 63. G. P. Hyung, J. A. Malen, W. T. Piggott, J. D. Morse, R. Greif, J. Microelectromech. Syst., 15, 4 (2006) 976-985. 64. J. J. Martin, W. Qian, H. Wang, V. Neburchilov, D. P. Wilkinson, J. Power Sources, 164, 1 (2007) 287-292. 65. S. K. Kamarudin, W. R. W. Daud, S. L. Ho, U. A. Hasran, J. Power Sources, 163, 2 (2007) 743-754. 66. T. V. Reshetenko, H. T. Kim, U. Krewer, H. J. Kweon, Fuel Cells, 7, 3 (2007) 238-245. 67. V. G. Sister, V. N. Fateev, D. A. Bokach, Russ. J. Electrochem., 43, 9 (2007) 1097-1100.  92 68. S. Q. Song, Z. X. Liang, W. J. Zhou, G. Q. Sun, Q. Xin, J. Power Sources, 145, 2 (2005) 495-501. 69. J. B. Ge, H. T. Liu, J. Power Sources, 142, 1-2 (2005) 56-69. 70. ScienceDirect, Journal Search, 7-1-2010, http://www.sciencedirect.com 71. C. M. Miesse, J. Power Sources, 162, 1 (2006) 532-540. 72. J. T. Mueller, P. M. Urban, W. F. Hoelderich, K. M. Colbow, D. P. Wilkinson, J. Electrochem. Soc., 147, 11 (2000) 4058-4060. 73. Iowa State University, Methanol Material Safety Data Sheet, 7-1-2010, http://avogadro.chem.iastate.edu/MSDS/methanol.htm 74. N. A. Hampson, M. J. Willars, B. D. McNicol, J. Power Sources, 4, 3 (1979) 191-201. 75. J. F. Whitacre, T. Valdez, S. R. Narayanan, J. Electrochem. Soc., 152, 9 (2005) A1780-A1789. 76. V. S. Bagotzky, Y. Vassiliev, O. A. Khazova, J. Electroanal. Chem., 81, 2 (1977) 229-238. 77. A. S. Arico, S. Srinivasan, V. Antonucci, Fuel Cells, 1, 2 (2001) 133-161. 78. R. Dillon, S. Srinivasan, A. S. Arico, V. Antonucci, J. Power Sources, 127, 1-2 (2004) 112-126. 79. A. Lindermeir, G. Rosenthal, U. Kunz, U. Hoffmann, J. Power Sources, 129, 2 (2004) 180-187. 80. X. Ren, S. Gottesfeld, J. Electrochem. Soc., 148, 1 (2001) A87-A93. 81. Dupont, Nafion(R) PFSA Membranes Product Sheet,  (2002). 82. N. Carretta, V. Tricoli, F. Picchioni, J. Membr. Sci., 166, 2 (2000) 189-197.  93 83. L. Gubler, D. Kramer, J. Belack, O. Unsal, T. J. Schmidt, J. Electrochem. Soc., 154, 9 (2007) 981-987. 84. L. Jorissen, V. Gogel, J. Kerres, J. Garche, J. of Power Sources, 105, 2 (2002) 267-273. 85. V. Neburchilov, J. Martin, H. Wang, J. Zhang, J. Power Sources, 169, 2 (2007) 221-238. 86. P. Dimitrova, K. A. Friedrich, B. Vogt, U. Stimming, J. Electroanal. Chem., 532, 1-2 (2002) 75-83. 87. A. Lam, D. P. Wilkinson, Z. Jiujun, in Proton Exchange Membrane Fuel Cells, T. Fuller, C. Bock, C. Lamy, Eds., PV 1, pp. 273-281, The Electrochemical Society Proceedings Series, Pennington, NJ, (2005). 88. S. Hikita, K. Yamane, Y. Nakajima, JSAE Review, 22, 2 (2001) 151-156. 89. D. P. Wilkinson, A. Lam, in , PV MA 2005-02, p. 1827, Meeting Abstracts, Pennington, NJ, (2005). 90. A. Lam, D. P. Wilkinson, J. Zhang, Electrochimica Acta, 53, 23 (2008) 6890- 6898. 91. A. K. Shukla, R. K. Raman, Ann. Rev. Mater. Research, 33, (2003) 155-168. 92. S. Lj. Gojkovic, S. Gupta, R. F. Savinell, J. Electroanal. Chem., 462, 1 (1999) 63-72. 93. H. Tributsch, M. Bron, M. Hilgendorff, H. Schulenburg, I. Dorbandt, J. Appl. Electrochem., 31, 7 (2001) 739-748. 94. S. Surampudi, S. R. Narayanan, E. Vamos, H. Frank, G. Halpert, J. Power Sources, 47, 3 (1994) 377-385.  94 95. S. C. Barton, T. Patterson, E. Wang, T. F. Fuller, A. C. West, J. Power Sources, 96, 2 (2001) 329-336. 96. S. C. Barton, W. Deng, J. Gallaway, S. Levendovsky, T. S. Olson, in , PV 1, pp. 315-322, ECS Transactions, Los Angeles, CA, United states, (2005). 97. A. K. Shukla, R. K. Raman, K. Scott, Fuel Cells, 5, 4 (2005) 436-447. 98. R. Zeng, P. K. Shen, Journal of Power Sources, 170, 2 (2007) 286-290.    95 2. HIGH FUEL CONCENTRATION DIRECT LIQUID FUEL CELL WITH A REDOX COUPLE CATHODE1 2.1 Introduction Direct Liquid Fuel Cells (DLFCs) are presently targeted towards micro/portable applications due to compactness and facile storage/handling of the fuel. Relative to hydrogen fuel cell technology, DLFC technology possesses several important advantages, including high volumetric fuel energy density, no cooling requirements, no auxiliary humidification and ease of fuel storage/transportation. The major technological challenges faced by conventional DLFC technology include cathode flooding, fuel crossover, CO2 product removal at the anode and sluggish anode and cathode kinetics [1, 2]. Water dragged from the anode to the air cathode by the electro-osmotic effect and water produced by the cathode reaction significantly impacts the degree of cathode flooding. Currently, Nafion® is the most widely used membrane in DLFC systems due to its good proton conductivity, mechanical durability and chemical resistance. However, liquid fuels are known to permeate through Nafion® at rates where fuel loss and cathode depolarization become an issue, ultimately leading to reduced cell performance [3, 4]. Consequently, it is common to use low fuel concentrations relative to the reaction stoichiometry concentration limit to reduce fuel crossover in conventional DLFCs (e.g., 0.75-2 M [3, 5, 6] vs. approx. 17 M for CH3OH ; 5-6 M [7, 8] vs. approx. 26.5 M for HCOOH ). In conventional DLFCs, fuel crossover can cause cathode depolarization through two related but 1 A version of this chapter has been published. A. B. Ilicic, D. P. Wilkinson, K. Fatih, F. Girard. High Fuel Concentration Direct Liquid Redox Fuel Cell with Redox Couple Cathode. J. Electrochem. Soc., 155, 12 (2008) B1322-B1327. Reproduced by permission of The Electrochemical Society.  96 fundamentally different phenomena: 1) fuel oxidation at the cathode (Pt catalyst), creating a mixed potential, and 2) decreasing surface concentration of oxygen at the cathode due to the consumption of oxygen via direct fuel oxidation, leading to a Nernstian voltage loss and increased kinetic overpotential. A number of approaches have been investigated by various research groups to address methanol crossover, which include developing new membranes [9], modifying membranes [10], introducing fuel additives [11], optimizing operating conditions [12], designing novel electrodes [13] and employing methanol-tolerant cathode catalysts [14-16]. Transition-metal macrocycles such as iron tetramethoxyphenylporphoryn or RuSe-based metal catalysts are examples of methanol-tolerant cathode catalysts that have received much attention. Although methanol-tolerant catalysts typically exhibit inferior activity towards the oxygen reduction reaction relative to pure Pt, they represent a novel solution to cathode depolarization and allow for greater design flexibility due to their selective nature. For example, single-chamber, mixed-reactant direct methanol fuel cells based on methanol-tolerant cathode catalysts have been demonstrated [17-19]. It is well understood that poor cathode kinetics, stringent humidification requirements, cathode flooding and complex TPB requirements represent some of the key technological issues for the air cathode in Proton Exchange Membrane Fuel Cells (PEMFCs) and DLFCs [20]. Replacing the air cathode with a carbon- based cathode (no PGM catalyst) utilizing a redox couple has previously been demonstrated for a hydrogen-fuelled PEMFC by Fatih et al. [21]. Further advancements in the Direct Hydrogen Redox Fuel Cell were recently reported  97 which allow elimination of auxiliary humidification and cooling, improved cathode kinetics and enhanced design flexibility [22]. Overall performance obtained for this fuel cell system was promising (peak power densities exceeding 170 mW/cm² using either CH3OH or HCOOH as the fuel), despite various components such as the membrane not being optimized. In this paper we examine the use of a carbon-based cathode with a redox couple for DLFCs to address issues such as cathode depolarization, cathode flooding and poor cathode kinetics. A number of redox couples exhibit good electrochemical activity over carbon-based electrodes [23, 24]. This all-liquid fuel cell system is referred to as a Direct Liquid Redox Fuel Cell (DLRFC). Key advantages of the DLRFC over the conventional DLFC include no PGM catalyst at the cathode, the ability to utilize high fuel concentrations at the anode (e.g., 16.7 M CH3OH or 18 M HCOOH) without introducing significant cathode issues, no cathode flooding as it is an all-liquid system and enhanced design flexibility with respect to the cathode (i.e., no TPB constraints, use of 3-D electrode etc.). However, there are challenges associated with the DLRFC which include regeneration of the redox couple, membrane contamination by the redox couple and crossover of the redox couple. As the redox couple oxidant is reduced during cell discharge, it must be regenerated externally by chemical [25-28], electrochemical [29] or biological means [30], which will affect the system’s overall efficiency and simplicity. A DLRFC can be interpreted as a hybrid between a DLFC and a redox flow battery. It follows that many combinations are possible when selecting the  98 liquid fuel and the redox couple. Criteria for a suitable redox couple include reasonable solubility in the electrolyte, good electrochemical activity, chemical stability, a sufficiently high redox potential to provide a useful cell voltage and a sufficiently low redox potential to permit regeneration by oxygen when applicable. In this work, we consider the Fe2+/Fe3+ redox couple and either methanol or formic acid as the fuel. The standard exchange current density at 25°C of the Fe2+/Fe3+ redox couple on glassy carbon (approx. 10 mA/cm²) [31] is several orders of magnitude greater than oxygen on Pt (approx. 10-7 mA/cm²) [32]. The relevant electrochemical reactions for a methanol DLRFC employing the Fe2+/Fe3+ redox couple are:  Anode OHOHCHe6H6CO 232 +←++ −+  E0= 0.06 V vs. SHE (2.1) Cathode +−+ →+ 23 Fe6e6Fe6  E0= 0.77 V vs. SHE (2.2) Overall    +++ ++→++ 22323 Fe6H6COFe6OHOHCH  E0= 0.71 V (2.3)  Similarly, the individual reactions for a formic acid DLRFC are: Anode HCOOHe2H2CO2 ←++ −+  E0= -0.20 V vs. SHE (2.4) Cathode +−+ →+ 23 Fe6e6Fe6  E0= 0.77 V vs. SHE (2.5) Overall +++ ++→+ 223 Fe2H2COFe2HCOOH  E0= 0.97 V (2.6)  It is important to note that accelerated membrane degradation due to hydroxyl radical formation from peroxide in the presence of Fe2+/Fe3+ (i.e., Fenton’s reagent) is not expected to be an issue in the DLRFC due to the absence of metal catalyst at the cathode and the low concentration of dissolved oxygen at the cathode. In the event that peroxide is formed at the PtRu anode,  99 the peroxide or any hydroxyl radicals (generated from Fe2+/Fe3+ ions that have crossed over) would react with the fuel before reaching the membrane.  2.2 Experimental Four types of experiments were conducted to investigate the use of the Fe2+/Fe3+ redox couple in a Direct Liquid Redox Fuel Cell (DLRFC): Differential Scanning Calorimetry (DSC) of ferric/ferrous/methanol solutions, cyclic voltammetry of solutions containing iron ions and methanol or formic acid, membrane and electrolyte conductivity measurements and DLFC/DLRFC performance testing in a 4 cm² fuel cell fed with methanol or formic acid. All electrochemical measurements, unless otherwise stated, were performed with a Solartron 1470E Multistat. The acidic aqueous Fe2+/Fe3+ electrolyte, referred to as the “redox electrolyte”, contained  0.81 M Fe(III)NH4(SO4)2•12H2O (ACS, Fisher), 0.09 M Fe(II)SO4•7H2O (ACS, Fisher) and 0.5 M H2SO4 (ACS, Fisher). Methanol (electronic grade, Fisher) and formic acid (ACS, Sigma Aldrich) were used as fuels in the DLRFC. Deionized water (18 MΩ·cm) was used to prepare all electrolytes reported herein. Differential scanning calorimetry was conducted on two redox electrolytes with 2 M CH3OH. Pulverized Toray Carbon Fiber Paper (CFP) (TGP-H-060, E- TEK) with a particle diameter not exceeding 0.5 mm was uniformly dispersed in an amount not exceeding 2 wt% in one electrolyte but was absent in the other. Aluminum hermetic pans with an o-ring provided a high pressure seal for the  100 aqueous samples as they were heated from 25°C to 200°C at a rate of 1°C/min. Sample masses were in the range of 50-75 mg. Cyclic voltammetry experiments were conducted at 70°C using either a 5 mm Glassy Carbon (GC) (AFE1E050GC, Pine Instruments) or Pt (EDI101, Radiometer) working electrode. The counter electrode was a Pt flag (XM120, Radiometer) and the reference electrode was a Hg/Hg2SO4/saturated K2SO4 reference electrode (XR200, Radiometer). The temperature of the glass cell was maintained at 70°C by submerging the cell in a temperature bath (Neslab EX-7, Thermo Electron) and electrolytes were de-aerated by bubbling N2 for 30 minutes prior to testing under static conditions. The GC electrode was prepared by rinsing with alcohol, polishing with 0.05 µm alumina, rinsing with deionized water, sonicating in deionized water for two minutes and finally rinsing in deionized water. The Pt electrode was prepared by rinsing with alcohol, submerging in a hot solution of 1:1 concentrated sulfuric acid and 30% w/w hydrogen peroxide for 5 minutes followed by rinsing with deionized water. The electrochemical window tested was 0-1.2 V vs. SHE using a scan rate of 50 mV/s. A Solartron 1260 Frequency Response Analyzer (FRA) was used to obtain the internal cell resistance for IR correction of the cyclic voltammograms. Membrane conductivity experiments were performed on protonated Nafion® 117 (Ion Power) membrane samples in a custom-built 2-point conductivity cell containing Pt electrodes spaced 1 cm apart. A Solartron 1260 FRA was used to obtain the membrane impedance over a range of frequencies (105-107 Hz) for a 5 mV voltage perturbation. Membranes were cleaned and  101 protonated by soaking in deionized H2O for 2 h, boiling in 3 % H2O2 for 30 min, rinsing in deionized H2O, boiling in 0.5 M H2SO4 for 30 min, rinsing in deionized H2O and finally storing in deionized H2O for at least 24 h. The conductivity of the iron redox electrolyte was measured with a YSI 3200 conductivity meter. DLFC and DLRFC cell performance testing was conducted at 70°C in the single-pass fuel cell test system shown in Figure 2.1. An expanded view of the custom-built 4 cm² fuel cell is shown in Figure 2.2. The anode catalyst loading for both methanol and formic acid fed fuel cells was 2 mg/cm² (40% 1:1 a/o Pt/Ru on Vulcan XC-72, E-Tek) on TGP-H-060 CFP. Nafion® ionomer (5 % in H2O/alcohol, Aldrich) was added to the catalyst ink as a binding agent in the amount of 30% w/w (i.e., 30% Nafion® ionomer, not solution). The cathode was supplied with either air at 101.3 kPa abs. for DLFC systems or the Fe2+/Fe3+ redox electrolyte for DLRFC systems. DLFC cathodes contained 1 mg/cm² of Pt catalyst (40% Pt on Vulcan XC-72, E-Tek) sprayed on to TGP-H-060 CFP and also contained 30 % w/w Nafion® ionomer. DLRFC cathodes were simply layers of TGP-H-060 CFP with no PGM catalyst. Single-channel serpentine flow fields with a channel width and depth of 0.84 mm and 0.46 mm, respectively, were used for all reactants except for the Fe2+/Fe3+ redox cathode, in which case a 1 mm deep empty pocket was filled with six layers of 190 µm TGP-H-060 CFP. Nafion® 117 membranes were employed in all configurations and were cleaned and protonated by procedures outlined above. Nafion® membranes were not hot-pressed to the electrodes. In all configurations, the nitrogen bladder, depicted in Figure 2.2, was pressurized to approx. 200 kPa abs. to compress the fuel cell.  102   Figure 2.1. Schematic of the fuel cell test system.   Figure 2.2. Expanded view of the 4 cm² fuel cell.  103 2.3 Results and Discussion Differential scanning calorimetry was carried out on solutions containing both the Fe2+/Fe3+ redox couple and methanol with or without carbon. The objective was to determine at what temperature, if any, methanol present at the DLRFC cathode via crossover would react with Fe3+ ions and cause cathode issues. Since the cathode contains CFP, DSC was conducted with and without dispersed CFP shavings in the electrolyte to determine the effect of carbon on the stability of the fuel and oxidant. The absence of an exotherm in the results, shown in Figure 2.3, indicates that methanol and Fe3+ ions do not react with or without carbon in the temperature range of interest for the DLRFC. These results are consistent with Zawadzki et al. [33], who reported no reactivity of methanol over carbon up to 673 K. -5 -4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0 25 50 75 100 125 150 175 200 Temperature (°C) H ea t R at e (m W ) No carbon With carbon  Figure 2.3. DSC results for mixed CH3OH/Fe2+/Fe3+ redox with and without shavings of CFP. (50-75 mg sample mass, 1°C/min temperature ramp rate)  104 Cyclic voltammetry was used to compare the electrochemical activity of the Fe2+/Fe3+ redox couple over Pt and GC and to determine what effect the presence of methanol or formic acid has on the electrochemical activity of the redox couple over GC. In total, four configurations were investigated at 70°C: the redox electrolyte over a GC or Pt electrode; the redox electrolyte with 2 M methanol or 2 M formic acid over GC. All voltammograms were acquired at a scan rate of 50 mV/s over a potential window of 0-1.2 V vs. SHE. Stable voltammograms were obtained within 10 cycles and were subsequently IR corrected. Figure 2.4a, conveys comparable redox activity. Hence, there is an opportunity to use a carbon cathode for Fe3+ reduction in a fuel cell environment while obtaining performance similar to that over Pt, a key advantage with respect to cost and total PGM content. Figures 2.4b and 2.4c show the effect of 2 M methanol or formic acid, respectively, on the electrochemical activity of the redox couple over GC. It is important to note that a 2 M concentration is relatively high when simulating the cathode of a fuel cell after fuel crossover. The anodic and cathodic peak currents and peak potentials for Figures 2.4a-c are given in Table 2.1. Based on Figures 2.4b, 2.4c and Table 2.1, it is apparent that the presence of 2 M methanol or formic acid does not adversely suppress the electrochemical activity of the redox couple over GC. Physical adsorption of the fuel on the GC electrode may be responsible for the approx. 30% reduction in peak current density. Since practical concentrations of methanol or formic acid at a fuel cell cathode after fuel crossover would be significantly lower than 2 M, it may be  105 concluded that fuel crossover in a Fe2+/Fe3+ DLRFC at 70°C would have a negligible effect on the electrochemical performance of the cathode when methanol or formic acid is used as the fuel. This finding is consistent with the DSC results shown in Figure 2.3, which confirm the thermal stability of methanol and ferric ions on carbon over the temperature range of interest for the DLRFC.             Figure 2.4a. Cyclic voltammograms of a Fe2+/Fe3+ redox electrolyte over Pt and GC. (70°C, 50 mV/s scan rate, IR-corrected)  (a) -100 -80 -60 -40 -20 0 20 40 60 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Potential (V vs. SHE) C ur re nt  D en si ty  (m A /c m ²) Redox electrolyte over GC Redox Electrolyte 0.81 M FeNH4(SO4)2 0.09 M FeSO4 0.5 M H2SO4 Redox electrolyte over Pt  106           Figure 2.4b. Cyclic voltammograms of a Fe2+/Fe3+ redox electrolyte over GC with and without 2 M CH3OH. (70°C, 50 mV/s scan rate, IR-corrected)            Figure 2.4c. Cyclic voltammograms of a Fe2+/Fe3+ redox electrolyte over GC with and without 2 M HCOOH. (70°C, 50 mV/s scan rate, IR-corrected) -100 -80 -60 -40 -20 0 20 40 60 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Potential (V vs. SHE) C ur re nt  D en si ty  (m A /c m ²) Redox electrolyte over GC Redox Electrolyte 0.81 M FeNH4(SO4)2 0.09 M FeSO4 0.5 M H2SO4 Redox electrolyte + 2 M HCOOH over GC (c) -100 -80 -60 -40 -20 0 20 40 60 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Potential (V vs. SHE) C ur re nt  D en si ty  (m A /c m ²) Redox Electrolyte 0.81 M FeNH4(SO4)2 0.09 M FeSO4 0.5 M H2SO4 Redox electrolyte over GC Redox electrolyte + 2 M CH3OH over GC (b)  107 Table 2.1. Summary of anodic and cathodic peak currents (ip) and peak potentials (Ep) for cyclic voltammograms shown in Figures 2.4a-c. Voltammogram ipa (mA/cm²) ipc (mA/cm²) Epa (V vs. SHE) Epc (V vs. SHE) Redox electrolyte over GC 77 78 0.83 0.52 Redox electrolyte over Pt 100 101 0.76 0.61 Redox electrolyte + 2 M methanol over GC 53 52 0.87 0.43 Redox electrolyte + 2 M formic acid over GC 57 50 0.88 0.40  Membrane conductivity experiments were performed to observe the conductivity of Nafion® 117 membranes after exposure to the redox electrolyte between 25°C and 80°C. Membrane samples were submerged in the redox electrolyte at the specified temperature for at least 3 hours to permit ion exchange between the membrane and the electrolyte. The pores of the membranes were cleared of any residual redox electrolyte by first rinsing with deionized H2O, soaking in deionized H2O at room temperature for at least 12 hours, replacing the deionized H2O and allowing the membranes to sit for at least another 12 hours. During impedance measurements, the membranes were immersed in deionized water at the specified temperature. Membranes not exposed to the redox electrolyte were also tested for comparison. Membrane contamination by the redox couple is possible if the membrane conducts ions of the same polarity as the redox species, as would be the case for the Fe2+/Fe3+ redox couple and Nafion®. Membrane contamination could be addressed by ensuring the fixed charges in the membrane are the same polarity as the redox couple or by increasing the concentration ratio of charge-carrying ions (e.g.,  108 H3O+) to the redox couple, since the charge-carrying ions are in competition for the membrane’s ionic sites. Other possible approaches include the investigation of alternative membranes (e.g., sulfonated poly ether ether ketone (SPEEK), polybenzimidazole (PBI) [34]) or the addition of complexing agents (e.g., EDTA4-) to the catholyte to neutralize the electrostatic attraction between the redox couple and the membrane’s conducting sites and possibly provide steric transport resistance. The latter approach, however, will influence the redox couple’s standard half-cell potential and/or the kinetics. Crossover of the redox couple could result in a reduction of cell capacity and may cause anode depolarization as well. As shown in Figure 2.5, the membrane conductivity increases with temperature (curves (b) and (c)). However, 80-90% of the membrane’s ionic conductivity is lost after exposure to the redox electrolyte, which is due to substitution of the protons on the membrane’s sulfonic sites by the less mobile iron cations. The conductivity of the redox electrolyte, also shown in Figure 2.5, decreases with temperature (curve (a)).        109 0 20 40 60 80 100 120 140 160 180 20 30 40 50 60 70 80 Temperature (°C) C on du ct iv ity  (m S/ cm ) (a) (b) (c)  Figure 2.5. (a) Conductivity of the Fe2+/Fe3+ redox electrolyte (0.81 M FeNH4(SO4)2, 0.09 M FeSO4, 0.5 M H2SO4). (b) Conductivity of Nafion® 117 in deionized water. (c) Conductivity of Nafion® 117 in deionized water after exposure to the redox electrolyte.  Fuel cell tests were conducted in the fuel cell test system (Figure 2.1) with a 4 cm² fuel cell (Figure 2.2) to establish the effect of fuel concentration on the performance of DLRFCs and conventional DLFCs. Fuel anolytes were prepared at both “low” and “high” concentrations of either methanol or formic acid. For methanol this was 2 and 16.7 M, and for formic acid this was 2 and 18 M, both in 0.5 M H2SO4. High anolyte concentrations approximate equimolar water and fuel. For methanol fuel cells, equimolar water and fuel represents the stoichiometric concentration limit. Cathodes were supplied with either the Fe2+/Fe3+ redox  110 electrolyte or air at 101.3 kPa abs. for the DLRFC and DLFC systems, respectively. Reactant flow rates correspond to a reaction stoichiometry of 4 at each current density. Polarization curves for both methanol and formic acid variants, shown in Figure 2.6, were generated galvanostatically. The open circuit potential (OCP) can be used to detect cathode depolarization since activation, ohmic and mass transport losses are not present during zero-current conditions. It can be seen that in the case of the methanol DLFC (Figure 2.6a), the OCP diminished from 0.68 V to 0.31 V after increasing the methanol concentration from 2 M to 16.7 M, signifying the presence of a mixed potential at the cathode due to fuel crossover. A similar observation can be made for formic acid (Figure 2.6b). On the other hand, an increase in fuel cell performance is observed for both methanol and formic acid DLRFC systems at the high fuel concentration relative to the low fuel concentration, demonstrating the redox cathode’s insensitivity to fuel crossover.  111 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 25 50 75 100 125 150 Current Density (mA/cm²) C el l P ot en tia l ( V) ●      DLRFC, 2 M CH3OH ■    DLRFC, 16.7 M CH3OH ○    DLFC, 2 M CH3OH □    DLFC, 16.7 M CH3OH  0 0.1 0.2 0.3 0.4 0.5 0.6 0 25 50 75 100 125 150 Current Density (mA/cm²) C el l P ot en tia l ( V) ●      DLRFC, 2 M HCOOH ■    DLRFC, 18 M HCOOH ○    DLFC, 2 M HCOOH □    DLFC, 18 M HCOOH  Figure 2.6. DLFC/DLRFC performance in the 4 cm² fuel cell for high and low fuel concentrations of (a) methanol and (b) formic acid. (Temperature, 70°C; reactant stoich, 4; air at 101.3 kPa abs. and not humidified). (a) (b)  112 Additional fuel cell testing was performed on methanol and formic acid DLRFCs/DLFCs with a mercury sulfate reference electrode inserted upstream from the anode in order to monitor the individual anode and cathode polarization behavior. The fuel concentrations were 16.7 M methanol and 18 M formic acid, both in 0.5 M H2SO4. Reactant flow rates were held constant at 5 mL/min for liquid electrolytes and 38 mL/min for air. Galvanostatically generated polarization curves for methanol and formic acid fuel cells are shown in Figures 2.7a and 2.7b, respectively. Non-optimized peak power densities of 30 and 22 mW/cm2 were observed for the methanol and formic acid DLRFCs compared to 6 and 7 mW/cm2 for the methanol and formic acid DLFCs, respectively (not shown). The data in Figure 2.7 indicates that the anode polarization is essentially independent of the cathode oxidant (i.e., DLRFC vs. DLFC). It is also evident that the air cathode potentials at open circuit are significantly lower than that anticipated at 70°C (c.f. 0.5 V for methanol and 0.59 V for formic acid vs. Ee= 1.18 V), which can be attributed to cathode depolarization by fuel crossover. On the other hand, the redox cathode exhibits a potential at open circuit close to the anticipated value at 70°C (c.f. 0.71 V for methanol and 0.73 V for formic acid vs. Ee= 0.816 V; Ee based on E0= 0.67 V vs. SHE for 0.5 M H2SO4 [35]). The discrepancy in the cathode potential for the DLRFC is due to the sulfate anion, present from both sulfuric acid and the redox salt, which forms complexes with the iron and decreases its equilibrium potential. Based on Figure 2.6 and Figure 2.7, it is apparent that the Fe2+/Fe3+ redox cathode does not suffer from depolarization after fuel crossover, even at high fuel concentrations. The selective nature of the  113 redox cathode enables the DLRFC to deliver superior cell performance at high fuel concentration relative to a conventional DLFC. The DLRFC performance was found to be reproducible after 50 hours of accumulated intermittent experimentation. Additional long-term testing is underway.                 Figure 2.7a. DLFC/DLRFC cell and electrode performance (IR-corrected) in the 4 cm² fuel cell for a 16.7 M methanol anolyte. (Temperature, 70°C; fuel flow rate, 5 mL/min; redox flow rate, 5 mL/min; air flow rate, 38 mL/min [@ 25°C and 101.3 kPa abs.]; air not humidified).   0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 50 100 150 200 250 300 350 400 Current Density (mA/cm²) Po te nt ia l ( V)  ●   DLRFC, Cell  ▲  DLRFC, Anode  ■   DLRFC, Cathode  ○   DLFC, Cell  ∆   DLFC, Anode  □   DLFC, Cathode (a)  114             Figure 2.7b. DLFC/DLRFC cell and electrode performance (IR-corrected) in the 4 cm² fuel cell for a 18 M formic acid anolyte. (Temperature, 70°C; fuel flow rate, 5 mL/min; redox flow rate, 5 mL/min; air flow rate, 38 mL/min [@ 25°C and 101.3 kPa abs.]; air not humidified)  In summary, these data demonstrate that DLRFCs can accommodate high fuel concentrations to enhance overall cell performance and simplify fuel dispensing of a concentrated fuel without introducing issues at the cathode. For DLRFC systems, overall improved fuel cell performance at increased fuel concentrations is likely due to improved reaction kinetics, reduced anodic concentration overpotential and/or enhanced mass transport of the fuel. 0.000 0.100 0.200 0.300 0.400 0.500 0.600 0.700 0.800 0 50 100 150 200 250 300 350 400 Current Density (mA/cm²) Po te nt ia l ( V)  ●   DLRFC, Cell  ▲  DLRFC, Anode  ■   DLRFC, Cathode  ○   DLFC, Cell  ∆   DLFC, Anode  □   DLFC, Cathode (b)  115 2.4 Conclusions The concept of a Direct Liquid Redox Fuel Cell (DLRFC) incorporating the Fe2+/Fe3+ redox couple over a carbon-based cathode and either methanol or formic acid as a fuel has been demonstrated. This alternative type of direct liquid fuel cell system has significant benefits with respect to fuel concentration, performance, cost and design flexibility. Also, the issue of cathode flooding is eliminated as it is an all-liquid system. It has been shown that this fuel cell system can accommodate very high fuel concentrations (greater than 16 M) without introducing significant issues at the cathode due to increased fuel crossover. The cathode is fully selective to the cathode reduction reaction and cathode performance is therefore independent of the fuel concentration at the anode. In addition, the ability to use a three-dimensional carbon-based cathode without catalyst can result in a significant reduction in the platinum group metal loading for the fuel cell. Significant improvements in performance are expected with further improvements of electrolyte composition, operating conditions and materials, particularly the membrane.  2.5 Acknowledgements The authors acknowledge the National Research Council, Institute for Fuel Cell Innovation (NRC-IFCI) and the Natural Sciences and Engineering Research Council of Canada (NSERC) for their financial support.   116 2.6 References 1. H. Yang, T. S. Zhao, Electrochim. Acta, 50, 16-17 (2005) 3243-3252. 2. W. Qian, D. P. Wilkinson, J. Shen, H. Wang, J. Zhang, J. Power Sources, 154, 1 (2006) 202-213. 3. J. B. Ge, H. T. Liu, J. Power Sources, 142, 1-2 (2005) 56-69. 4. J. G. Liu, T. S. Zhao, R. Chen, C. W. Wong, Electrochem. Commun., 7, 3 (2005) 288-294. 5. V. G. Sister, V. N. Fateev, D. A. Bokach, Russ. J. Electrochem., 43, 9 (2007) 1097-1100. 6. A. Lindermeir, G. Rosenthal, U. Kunz, U. Hoffmann, J. Power Sources, 129, 2 (2004) 180-187. 7. K. L. Chu, M. A. Shannon, R. I. Masel, J. Electrochem. Soc., 153, 8 (2006) A1562-A1567. 8. C. M. Miesse, W. S. Jung, K. J. Jeong, J. K. Lee, J. Lee, J. Power Sources, 162, 1 (2006) 532-540. 9. N. Carretta, V. Tricoli, F. Picchioni, J. Membr. Sci., 166, 2 (2000) 189-197. 10. A. Sungpet, J. Membr. Sci., 226, 1-2 (2003) 131-134. 11. A. Lam, D. P. Wilkinson, Z. Jiujun, in Proton Exchange Membrane Fuel Cells, T. Fuller, C. Bock, C. Lamy, Eds., PV 1, pp. 273-281, The Electrochemical Society Proceedings Series, Pennington, NJ, (2005). 12. S. Hikita, K. Yamane, Y. Nakajima, JSAE Review, 22, 2 (2001) 151-156. 13. D.P.Wilkinson, M. C. Johnson K. M. Colbow S. A. Campbell, US Patent No 5,672,439 (1997).  117 14. A. K. Shukla, R. K. Raman, Ann. Rev. Mater. Research, 33, (2003) 155-168. 15. S. Lj. Gojkovic, S. Gupta, R. F. Savinell, J. Electroanal. Chem., 462, 1 (1999) 63-72. 16. H. Tributsch, M. Bron, M. Hilgendorff, H. Schulenburg, I. Dorbandt, J. Appl. Electrochem., 31, 7 (2001) 739-748. 17. M. A. Priestnall, V. P. Kotzeva, D. J. Fish, E. M. Nilsson, J. Power Sources, 106, 1-2 (2002) 21-30. 18. A. K. Shukla, C. L. Jackson, K. Scott, G. Murgia, J. Power Sources, 111, 1 (2002) 43-51. 19. S. C. Barton, T. Patterson, E. Wang, T. F. Fuller, A. C. West, J. Power Sources, 96, 2 (2001) 329-336. 20. T. F. Fuller, M. L. Perry, Journal of the Electrochemical Society, 149, 7 (2002) 59-67. 21. K. Fatih, D. P. Wilkinson, F. Moraw, F. Girard, in Electrocatalysis, R. Adzic, V. Birss, G. M. Brisard, A. Wieckowski, Eds., PV 11, pp. 341-350, The Electrochemical Society Proceedings Series, Pennington, NJ, (2005). 22. K. Fatih, D. P. Wilkinson, F. Moraw, A. Ilicic, F. Girard, Electrochem. Solid- State Lett., 11, 2 (2008) B11-B15. 23. E. Sum, M. Skyllas-Kazacos, J. Power Sources, 15, 2-3 (1985) 179-190. 24. L. H. Thaller, in , PV, pp. 924-928, 9th Intersociety Energy Conversion Engineering Conference Proceedings, San Francisco, CA, USA, (1974). 25. M. R. Ronnholm, J. Warna, T. Salmi, I. Turunen, M. Luoma, Chemical Engineering Science, 54, 19 (1999) 4223-4232.  118 26. M. R. Ronnholm, J. Warna, D. Valtakari, T. Salmi, E. Laine, in , PV 66, pp. 447-452, Catalysis Today, Napoles, Italy, (2001). 27. J. T. Kummer, D. G. Oei, J. App. Electrochem., 12, (1982) 87-100. 28. J. T. Kummer, D. G. Oei, J. App. Electrochem., 15, 4 (1985) 619-629. 29. G. B. Adams, R. P. Hollandsworth, D. N. Bennion, J. Electrochem. Soc., 122, 8 (1975) 1043-1048. 30. A. Mazuelos, F. Carranza, I. Palencia, R. Romero, Hydrometall., 58, 3 (2000) 269-275. 31. F. C. Anson, Anal. Chem., 33, (1961) 939. 32. A. Damjanovic, J. O. Bockris, Electrochim. Acta, 11, 3 (1966) 376-378. 33. J. Zawadzki, B. Azambre, O. Heintz, A. Krzton, J. Weber, Carbon, 38, 4 (2000) 509-515. 34. V. Neburchilov, J. Martin, H. Wang, J. Zhang, J. Power Sources, 169, 2 (2007) 221-238. 35. J. G. Speigt, Lange's Handbook of Chemistry and Physics, 16th Ed., p. 1.385, McGraw-Hill, NY (2005).    119 3. IMPROVING THE PERFORMANCE OF THE DIRECT LIQUID REDOX FUEL CELL1 3.1 Introduction In an era of ever increasing demand for portable power, promising technologies such as direct methanol fuel cells (DMFCs) are being investigated by a large number of researchers across the globe [1-3]. The DMFC is targeted towards micro and portable applications due to the facile storage/transportation of the fuel and high volumetric energy/charge density of the fuel (4.62 kWh/L / 3.95 kAh/L at 25°C). These specific markets are generally not considered for the H2 proton exchange membrane fuel cell (PEMFC) due to the low volumetric energy/charge density of the fuel (approx. 0.9 kWh/L / 0.7 kAh/L at 34.5 MPa [5000 psi] and 25°C). Other DMFC advantages over the PEMFC include no need for auxiliary humidification or cooling equipment [4, 5]. There are a number of limitations to the DMFC, however, such as cathode flooding, fuel cross-over, CO2 product removal at the anode, and sluggish anode and cathode kinetics. Water dragged from the anode to the air cathode by the electro-osmotic effect and water produced by the cathode reaction significantly impact the degree of cathode flooding [6]. Currently, Nafion® is the most widely used membrane in DMFC systems due to its good proton conductivity, mechanical durability and chemical resistance. However, liquid fuels are known to permeate through Nafion® at rates where fuel loss and cathode depolarization 1 A version of this chapter has been submitted for publication.  A. B. Ilicic, M. S. Dara, D. P. Wilkinson, K. Fatih. Improved Performance of the Direct Liquid Redox Fuel Cell (2010).  120 become an issue, ultimately leading to reduced cell performance [7, 8]. Consequently, it is common to use low fuel concentrations relative to the reaction stoichiometry concentration to reduce fuel crossover in conventional DMFCs. For example, methanol concentrations in the range of 0.75-2 M are typically used [7, 9, 10] which are significantly lower than the reaction stoichiometry concentration of approximately 17 M CH3OH.  In conventional DMFCs, fuel crossover can cause cathode depolarization through two related but fundamentally different phenomena: 1) fuel oxidation at the cathode (Pt catalyst), creating a mixed potential, and 2) decreasing surface concentration of oxygen at the cathode due to the consumption of oxygen via direct fuel oxidation, leading to a Nernstian voltage loss and increased kinetic overpotential. A number of approaches have been investigated by various research groups to address methanol crossover, which include developing new membranes [11], modifying membranes [12], introducing fuel additives [13], optimizing operating conditions [14], designing novel electrodes [15] and employing methanol-tolerant cathode catalysts [16-18]. Our work involves an alternative approach to the classical DMFC architecture in which the air cathode containing Pt is substituted by a metal-ion redox couple (e.g., Fe2+/Fe3+) over a carbon cathode containing no platinum group metal (PGM) catalyst. This system is referred to herein as the direct methanol redox fuel cell (DMRFC). Initial pioneering work on this type of hybrid system was conducted by Fatih et al., but with a hydrogen redox fuel cell [19, 20]. The relevant electrochemical equations for the DMRFC using methanol fuel at the anode and the Fe2+/Fe3+ redox couple at the cathode are given below:  121  Anode OHOHCHe6H6CO 232 +←++ −+  E0= 0.06 V vs. SHE (3.1) Cathode +−+ →+ 23 Fe6e6Fe6  E0= 0.77 V vs. SHE (3.2) Overall    +++ ++→++ 22323 Fe6H6COFe6OHOHCH  E0= 0.71 V (3.3)  The DMRFC offers some significant advantages over the conventional DMFC including no PGM catalyst at the cathode, the ability to utilize high fuel concentrations at the anode (e.g., equimolar CH3OH/H2O, approximately 17 M CH3OH) without introducing significant issues at the selective cathode, no cathode flooding as it is an all-liquid system and enhanced design flexibility with respect to the cathode (i.e., no triple phase boundary constraints, use of 3-D electrode, etc.). However, there are some challenges associated with the DMRFC which include regeneration of the redox couple, membrane contamination by the redox couple, crossover of the redox couple and mismatched charge density of the anolyte and catholyte. After the redox couple oxidant is reduced during cell discharge, it must be regenerated externally by chemical, electrochemical [21] or biological means [22], which will affect the system’s overall efficiency and simplicity. The volumetric charge density of the anolyte and catholyte governs the volume of the electrolyte reservoirs required for a desired energy output. In our previous work on the DMRFC [23, 24], FeNH4(SO4)2 and FeSO4 salts were used to prepare the catholyte, in which case the aqueous solubility limit of the iron salt is less than 1 M at 25°C. As each methanol molecule carries six times the  122 number of electrons as a ferric ion, there is an issue of mismatched anolyte/catholyte charge densities, particularly at high methanol concentrations. A primary objective of this work was to increase the DMRFC catholyte charge density and improve the performance of the DMRFC by investigating catholytes prepared with irons salts apart from FeNH4(SO4)2. Membrane compatibility, electrolyte conductivity, cyclic voltammetry and fuel cell tests were used to evaluate and screen the selected iron salts. Additional studies were performed to investigate the effect of the anolyte methanol concentration and the cell temperature on the DMRFC performance. A series of short DMRFC durability tests are also included.  3.2 Experimental All electrolytes were prepared with deionized water (18 MΩ•cm) in glassware cleaned with equal parts of concentrated HNO3 (ACS, Fisher Scientific) and concentrated H2SO4 (ACS, Fisher Scientific). Electrolytes containing the Fe2+/Fe3+ redox couple were based on one of three anions: ClO4-, NO3- or SO42-. Iron perchlorate electrolytes were prepared with Fe(ClO4)3•6H2O (non-yellow, GFS Chemicals), no acid; iron nitrate electrolytes were prepared with Fe(NO3)3•9H2O (ACS, Fisher Scientific), no acid; iron sulfate electrolytes were prepared with FeNH4(SO4)2•12H2O (ACS, Fisher), FeSO4•12H2O (ACS, Fisher) and H2SO4 (ACS, Fisher Scientific). No acid was added to the iron perchlorate or iron nitrate solutions as these salts already contain a sufficient amount of residual perchloric or nitric acid, respectively, to produce a strongly  123 acidic solution. For instance, the pH of the 2.5 M Fe(ClO4)3 and 1 M Fe(NO3)3 electrolytes were approximately -0.25 and 0.5, respectively. An Oaklon 110 series pH meter was used for the pH measurements. All Nafion® 112 membranes used in the various experiments were cleaned and protonated by soaking in deionized H2O for 2 h, boiling in 3 % H2O2 for 30 min, rinsing in deionized H2O, boiling in 0.5 M H2SO4 for 30 min, rinsing in deionized H2O and finally storing in deionized H2O for at least 24 h. Chemical compatibility of the perchlorate and nitrate electrolytes with the Nafion® 112 membrane was determined by measuring any difference in dry weight of the membrane before and after exposure to the electrolytes. Nafion® membranes were initially dried in vacuum at 60°C overnight, weighed and then hydrated. The hydrated membranes were then immersed in a saturated or nearly saturated solution of ferric perchlorate (approx. 3 M) or ferric nitrate (approx. 3.9 M) at 90°C and left overnight. The membranes were finally rinsed with deionized water, soaked in deionized water, dried by the method above and weighed. The ionic conductivity of various redox electrolytes was measured with a YSI 3200 conductivity meter. Electrolytes were kept in a sealed glass cell to avoid evaporation and were immersed in a constant temperature bath (Neslab EX-7, Thermo Electron) to achieve the desired temperature. Conductivity measurements were taken once the electrolyte reached the desired temperature. Cyclic voltammetry experiments were conducted at 90°C using either a 5 mm diameter glassy carbon (GC) (AFE1E050GC, Pine Instruments) or 5 mm diameter Pt (EDI101, Radiometer) working electrode. The counter electrode was  124 a Pt flag (XM120, Radiometer) and the reference electrode was an Ag/AgCl reference electrode (XR300, Radiometer). The temperature of the glass cell was maintained at 90°C by submerging the cell in a constant temperature bath (Neslab EX-7, Thermo Electron) and electrolytes were de-aerated by bubbling N2 for 30 minutes at room temperature prior to testing under static conditions. The GC electrode was prepared by rinsing with alcohol, polishing with 0.05 µm alumina, rinsing with deionized water, sonicating in deionized water for two minutes and finally rinsing in deionized water. The Pt electrode was prepared by rinsing with alcohol, submerging in a hot solution of 1:1 concentrated sulfuric acid (ACS, Fisher) and 30% w/w hydrogen peroxide (ACS, Fisher) for 5 minutes followed by rinsing with deionized water. A Solartron 1260 Frequency Response Analyzer (FRA) was used to obtain the internal cell resistance for IR correction of the cyclic voltammograms. Stable voltammograms were acquired after 10 cycles. DMRFC performance testing was performed in the single-pass fuel cell test system shown in Figure 3.1. An expanded view of the custom-built 4 cm² fuel cell is shown in Figure 3.2. The anode catalyst loading for all fuel cell testing was 2 mg/cm² (40% 1:1 a/o Pt/Ru on Vulcan XC-72, E-Tek) on TGP-H-060 carbon fiber paper (CFP). Nafion® ionomer (5 % in H2O/alcohol, Aldrich) was added to the catalyst ink as a binding agent in the amount of 30% w/w (i.e., 30% Nafion® ionomer, not solution). Anolytes for fuel cell testing were prepared with methanol (electronic grade, Fisher) and either HClO4 or H2SO4, depending on whether the catholyte contained perchlorate or sulfate anions, respectively. A serpentine flow field with a channel width and depth of 0.84 mm and 0.46 mm, respectively, was  125 used at the anode. The cathode consisted of 3 layers of 370 µm thick TGP-H-120 CFP with no PGM catalyst packed into an empty 1 mm deep pocket. The cathode was supplied with a Fe2+/Fe3+ redox electrolyte incorporating either SO42- or ClO4- anions. A Nafion® 112 membrane was employed for all DMRFC tests. Nafion® membranes were not hot-pressed to the electrodes. In all configurations, the nitrogen bladder, depicted in Figure 3.2, was pressurized to approx. 200 kPa abs. to compress the fuel cell. The cell impedance used for IR- correction of polarization curves was obtained by taking the real component of the cell impedance measured at 20 kHz using a GwINSTEK LCR-821 unit. All fuel cell testing was performed after at least 30 min of potentiostatic cell conditioning at 0 V.   Figure 3.1. Schematic of the direct liquid redox fuel cell test system.  126  Figure 3.2. Expanded view of the 4 cm² fuel cell.  3.3 Results and Discussion In determining which non-sulfate iron salts would be suitable candidates for the DMRFC catholyte, the cost, chemical compatibility, half-cell potential, catalyst poisoning effects, purity, availability and solubility (when known) were considered. Iron salts with acetate, acetylacetonate, ammonium citrate, carbonate, chloride, citrate, hypophosphite, nitrate, oxalate, perchlorate, phosphate or tetrafluoroborate anions were initially screened. A list of all iron salts considered including reasons for excluding various salts is shown in Table 3.1. For some anions, such as citrate and oxalate, ligand effects contributed to the decision to exclude these salts from the study as they negatively shift the half cell potential of the Fe2+/Fe3+ redox couple. Upon completing the initial screening process, it was determined that nitrate and perchlorate iron salts should be further investigated through electrolyte conductivity, cyclic voltammetry and fuel cell tests.  127 Table 3.1. List of ferric salts considered for use in the DMRFC catholyte. Half-cell potentials reported for standard conditions at 25°C.   The aqueous solubility of Fe(ClO4)3•6H2O and Fe(NO3)3•9H2O at 25°C was experimentally determined to be greater than 2.5 and 3 M, respectively, which compares favourably with the aqueous solubility of FeNH4(SO4)2 (approx. 1 M) [25]. The Fe(NO3)3•9H2O salt melts at 47°C and thus preparation of a saturated ferric nitrate solution simply involves heating the salt. Positive results were obtained from the chemical compatibility test for Nafion® 112 membranes in saturated or nearly saturated solutions of ferric perchlorate and ferric nitrate at 90°C. No significant change in the dry membrane weight was observed after overnight exposure to the electrolytes indicating no rapid membrane degradation due to oxidation. It should also be noted that previous differential scanning calorimetry tests [23] showed chemical compatibility of methanol and the Fe2+/Fe3+ redox couple up to 200°C in the presence and absence of carbon, which confirms the selectivity of the Fe2+/Fe3+ redox cathode.  128 The electrolyte conductivity of 2.5 M Fe(ClO3)3, 3 M Fe(NO3)3 and  1 M FeNH4(SO4)2 / 0.5 M H2SO4 was measured at 50, 70 and 90°C. High concentrations for each type of iron salt were selected to determine the conductivity for fuel cell conditions. The conductivity measurements, shown in Figure 3.3, indicate that the non-sulfate electrolytes exhibited higher conductivity than the sulfate containing electrolyte, particularly at higher temperatures. It is interesting to note that the conductivity of the sulfate containing electrolyte decreases with temperature whereas the conductivity of the other electrolytes increases. This trend can be attributed to the formation of complexes such as FeH(SO4)2 and HSO4- at elevated temperatures [26]. Based on the conductivity data acquired here, it is apparent that the conductivities of the ferric nitrate and ferric perchlorate electrolytes are sufficiently high for employment in the DMRFC given the good performance obtained with ferric sulfate which has a lower conductivity.   129 0 50 100 150 200 250 300 40 50 60 70 80 90 100 Temperature (˚C) C on du ct iv ity  (m S/ cm ) (a) (b) (c)  Figure 3.3. Electrolyte conductivities as a function of temperature for (a) 2.5 M Fe(ClO4)3; (b) 1 M FeNH4(SO4)2 / 0.5 M H2SO4; (c) 3 M Fe(NO3)3.  Cyclic voltammetry experiments were conducted at 90°C to quantify the electrochemical activity and reversibility over Pt and GC of electrolytes prepared with the three iron salts: (a) 1 M Fe(ClO4)3, (b) 1 M FeNH4(SO4)2 / 0.5 M H2SO4 and (c) 0.05 M and 1 M Fe(NO3)3. The electrochemical activity over Pt provides a meaningful baseline for comparison of the activity over GC, which is chemically similar to the CFP used in the DMRFC cathode. The voltammograms for the iron electrolytes, shown in Figure 3.4, indicate comparable electrochemical activity between Pt and GC for the perchlorate (Figure 3.4a) and sulfate (Figure 3.4b)  130 electrolytes. For the nitrate electrolyte (Figure 3.4c), the electrochemical response was largely dependent on the concentration of the iron salt.  -200 -150 -100 -50 0 50 100 150 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Potential vs. SHE (V) C ur re nt  D en si ty  (m A /c m ²) Pt GC  Figure 3.4a. Cyclic voltammograms over Pt and GC working electrodes for a 1 M Fe(ClO4)3 electrolyte. (90°C, 50 mV/s scan rate, IR-corrected)   (a)  131 -200 -150 -100 -50 0 50 100 150 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Potential vs. SHE (V) C ur re nt  D en si ty  (m A /c m ²) GC Pt  Figure 3.4b. Cyclic voltammograms over Pt and GC working electrodes for a 1 M FeNH4(SO4)2 / 0.5 M H2SO4 electrolyte. (90°C, 50 mV/s scan rate, IR-corrected) (b)  132  Figure 3.4c. Cyclic voltammograms over Pt and GC working electrodes for (I) Pt, 0.05 M Fe(NO3)3, (II) GC, 0.05 M Fe(NO3)3, (III) Pt, 1 M Fe(NO3)3, and (IV) GC, 1 M Fe(NO3)3 electrolytes. (90°C, 50 mV/s scan rate, IR-corrected)  Initially, the 1 M Fe(NO3)3 electrolyte was tested and irreversible electrochemical behaviour with a significant degree of noise was observed (not shown). Consequently, a lower concentration was tested and it was observed that at 0.05 M Fe(NO3)3, voltammograms with little noise and reversible behaviour could be generated on Pt and GC. The irreversible electrochemical response of 1 M Fe(NO3)3 is probably related to the instability of the NO3- ions and may be due NO3- intermediates (e.g., NO, NO2) adsorbing onto the electrode surface or forming complexes with iron ions. Since one of the primary objectives of this work was to improve the charge density of the redox electrolyte, (c)  133 it would not be logical to pursue the ferric nitrate electrolyte if a reversible electrochemical response could not be generated at a concentration of 1 M Fe(NO3)3. The anodic and cathodic peak current densities (ip), peak potentials (Ep) and apparent half cell potential (E) for the perchlorate, nitrate and sulfate voltammograms from Figure 3.4 are shown in Table 3.2. The data for the 0.05 M Fe(NO3)3 system were included in Table 3.2 for completeness but since this electrolyte is not acceptable for use as a catholyte in a DMRFC, no further discussion on the nitrate voltammogram will be given. From Table 3.2, it is apparent that the peak current density over GC is reduced by roughly 15% and 20% relative to that over Pt for the perchlorate and sulfate electrolytes, respectively. The anodic and cathodic peak current density for the perchlorate system over GC is 83% and 54% greater than that observed for the sulfate system, respectively. In addition, the electrochemical reversibility of the perchlorate system, represented by the difference in peak potentials, is superior to that of the sulfate system over both Pt and GC. The advantages of the perchlorate system with respect to the electrochemical activity and reversibility indicate that this system is a better candidate for use as a DMRFC catholyte than the sulfate system.  The apparent Fe2+/Fe3+ half-cell potential, taken as the midpoint of the peak potentials, is significantly higher for the perchlorate electrolyte (0.83 V vs. SHE) than that observed for the sulfate electrolyte (0.64 V vs. SHE). The difference is due to ligand effects of the sulfate and perchlorate anions on the Fe2+ and Fe3+ cations. In general, all ligands have a certain  134 tendency to form complexes with a particular metal ion, which is expressed by the formation constant of the complex. In the case of a redox couple, if the ligand has a stronger preference to form complexes with one metal ion over the other, the half-cell potential can shift due to changes in the relative activity of the metal ions. The Fe2+/Fe3+/C3H5O(COO)33− (citrate) system demonstrates this phenomenon quite well. The formation constant for the Fe(II)(cit)- complex is 3.08 while the formation constant for the Fe(III)(cit) complex is 12.5 [25]. Consequently, a larger fraction of the Fe3+ ions are complexed than Fe2+ ions and the Fe3+/Fe2+ activity ratio is reduced. This has a direct Nernstian impact on the half-cell potential of the redox couple, which explains the negative shift in the standard half-cell potential of the Fe2+/Fe3+ redox couple from 0.77 V vs. SHE [27] to 0.03 V vs. SHE after complexation with citrate [28]. Overall, the perchlorate electrolyte is a strong candidate as a DMRFC catholyte because it exhibits significantly greater electrochemical activity, electrochemical reversibility, half-cell potential and solubility relative to the sulfate electrolyte.   135 Table 3.2. Summary of anodic and cathodic peak current densities (ip), peak potentials (Ep) and apparent half-cell potentials (E) for cyclic voltammograms shown in Figure 3.4. All data were obtained at 90°C.        DMRFC testing was performed to identify operating conditions that yield improved fuel cell performance and to observe the short-term performance stability of the DMRFC. The DMRFC tests can be divided into four groups which investigate: 1. The effect of the anion in the iron salt (incorporates the iron concentration effect) 2. The effect of the methanol concentration in the anolyte 3. The effect of the cell temperature 4. The short-term durability of the DMRFC  DMRFC tests observing the effect of the anion in the iron salt were performed at 90°C using catholytes prepared with either 2.5 M Fe(ClO4)3 or 1 M FeNH4(SO4)2 / 0.5 M H2SO4 (a DMRFC test using a 1 M Fe(NO3)3 catholyte produced no significant amount of power and had extremely poor stability). The iron concentration in the catholyte for both cases was in the vicinity of the solubility limit for that particular salt. The anolyte consisted of 1 M CH3OH and  136 either 0.2 M HClO4 or 0.1 M H2SO4, depending on the iron salt being used in the catholyte. The anolyte included some acid in order to establish ionic conductivity between the anode and the reference electrode, as shown in Figure 3.1. The cell and individual electrode polarization behaviour for DMRFC tests observing anion effects are shown in Figure 3.5. The results indicate that the perchlorate system exhibits a significantly higher open-circuit potential (OCP), lower ohmic losses and significantly improved overall performance. Furthermore the corresponding power densities for Figure 3.5, shown in Figure 3.6, indicate a greater than 3-fold increase in the peak power density of the perchlorate system (79 mW/cm²) relative to the sulfate system (25 mW/cm²). The individual anode and cathode polarizations shown in Figure 3.5 allow for a deeper understanding of the influence of the anion. The cathode polarization behaviour indicates a greater than 200 mV positive shift in the cathode OCP when switching from a sulfate system to a perchlorate system. The primary reason for this shift in half cell potential is the difference in the interaction of the iron ions with the sulfate and perchlorate ligands. If a ligand preferentially binds to one redox couple ion over the other, the activity of that preferred ion will be reduced and will affect the potential of the redox couple in a Nernstian fashion. Since perchlorate anions are known to exhibit weak ligand effects [29], it is likely that the oxidizing ability of the ferric ion is greater in the perchlorate system. More detailed analyses would be required to confirm this. The anode polarizations show higher overpotential for the perchlorate system. Two factors in the perchlorate system that may be responsible include: (1) perchlorate anions  137 influencing the polarization behaviour of methanol oxidation, and (2) increased anode depolarization due to the higher crossover rate of ferric ions from the cathode to the anode. The latter phenomenon is almost certainly a primary factor as the authors have previously observed the effect of anode depolarization due to crossover of ferric ions and the ferric ion concentration is significantly higher in the perchlorate system (2.5 M vs. 1 M in the sulfate system). 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0 100 200 300 400 500 Current Density (mA/cm²) Po te nt ia l ( V) Perchlorate- Cell Perchlorate- Anode Perchlorate- Cathode Sulfate- Cell Sulfate- Anode Sulfate- Cathode  Figure 3.5. Cell and individual electrode polarization (IR-corrected) for a DMRFC supplied with (a) 2.5 M Fe(ClO4)3 catholyte and 1 M CH3OH, 0.2 M HClO4 anolyte (perchlorate system, Rcell= 0.03 Ω) and (b) 1 M FeNH4(SO4)2 / 0.5 M H2SO4 catholyte and 1 M CH3OH, 0.1 M H2SO4 anolyte (sulfate system, Rcell= 0.12 Ω·cm²). (90°C cell temperature, Nafion® 112 membrane, 1 mL/min anolyte flow rate, 2 mL/min catholyte flow rate).  138 0 10 20 30 40 50 60 70 80 90 0 100 200 300 400 500 Current Density (mA/cm²) Po w er  D en si ty  (m W /c m ²) Perchlorate Sulfate  Figure 3.6. Power density curves calculated from the data shown in Figure 3.5.  These non-optimized fuel cell results clearly demonstrate the advantages of using an iron perchlorate salt in the catholyte of a DMRFC rather than a sulfate-based iron salt. The benefits arise primarily due to the increased solubility limit of the Fe2+/Fe3+ redox couple (more than 2.5-fold) and the increased half cell potential of the Fe2+/Fe3+ redox couple (more than 200 mV increase). The higher catholyte charge density and better overall cell performance of the iron perchlorate system indicate it is a better choice for the methanol DMRFC. A wide range of anolyte fuel concentrations can be tested in the DMRFC since the selective nature of the redox couple cathode eliminates the issue of  139 cathode depolarization after fuel crossover. The DMRFC polarization behaviour for methanol concentrations in the range of 2-24 M is shown in Figure 3.7. For these experiments, the anolyte included 0.5 M H2SO4 and the catholyte was comprised of 0.81 M FeNH4(SO4)2, 0.09 FeSO4 and 0.5 M H2SO4. The fuel concentration that delivers the best performance is 16.7 M CH3OH, which agrees with stoichiometry as this concentration corresponds to an equimolar mixture of methanol and water (in 0.5 M H2SO4). It is likely that at low and high fuel concentrations the anode reaction kinetics are negatively affected by the reduced fuel and water concentration, respectively. It is surprising that the DMRFC can operate at 24 M CH3OH, which represents pure methanol with 0.5 M H2SO4. In this case, at least one of the following must be true: a) the methanol is only partially oxidized (i.e., to CH2O, CO); b) the water required for the full methanol oxidation reaction is supplied via water crossover from the cathode. As a point of reference, Jiang et al. reported the water crossover rate through Nafion 117 at 62°C to be 1.10 x10-6 mol/cm²·s, which can theoretically accommodate 636 mA/cm² of complete methanol oxidation [6]. For Nafion 112, which is 71% thinner than Nafion 117, the water crossover rate would naturally be significantly greater.   140 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0 50 100 150 200 250 Current Density (mA/cm²) C el l V ol ta ge  (V ) 2 M methanol 8 M methanol 16.7 M methanol 20.2 M methanol 24 M methanol  Figure 3.7. Cell polarization curves (IR-corrected) for a DMRFC using anolytes with different methanol concentrations. (70°C cell temperature, Nafion® 112 membrane, X M CH3OH / 0.5 M H2SO4 anolyte, 5 mL/min anolyte flow rate, 0.81 M FeNH4(SO4)2 / 0.09 FeSO4 / 0.5 M H2SO4 catholyte, 5 mL/min catholyte flow rate).  The DMRFC performance at 50, 70 and 90°C is shown in Figure 3.8. In this case the anolyte consisted of 1 M CH3OH and 0.2 M HClO4 whereas the catholyte was 2.5 M Fe(ClO4)3. The results clearly show that the DMRFC performs best at 90°C. Cell temperatures exceeding 90°C were not tested to avoid electrolyte boiling. The performance improvements observed at elevated cell temperatures are likely due to enhanced reaction kinetics, diffusion rates and/or membrane/electrolyte conductivity.  141 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 100 200 300 Current Density (mA/cm²) Po te nt ia l ( V) 90°C 70°C 50°C  Figure 3.8. Cell polarization curves (IR-corrected) for a DMRFC at different temperatures. (Nafion® 112 membrane, 1 M CH3OH / 0.2 M HCl4 anolyte, 1 mL/min anolyte flow rate, 2.5 M Fe(ClO4)3 catholyte, 2 mL/min catholyte flow rate).  Short-term DMRFC durability testing was performed galvanostatically at 50 mA/cm² for three different configurations: (1) A perchlorate-based DMRFC at 50°C, (2) A perchlorate-based DMRFC at 90°C and (3) A sulfate-based DMRFC at 90°C. The results are shown in Figures 3.9a-c. Linear approximations of the voltage vs. time curves show that the DMRFC degrades at approximately 2.3, 2.5 and 3.2 mV/hr for cases (1), (2) and (3), respectively, over this time period. Similarly, the cell impedance increased at a rate of approximately 4.6, 9.3 and 0.9 mΩ/hr for cases (1), (2) and (3), respectively. It is interesting to note that the  142 sulfate-based DMRFC at 90°C showed the greatest voltage degradation rate but the lowest rate of increase for the cell impedance. Loss mechanisms unrelated to the cell impedance (e.g., accumulation of adsorbates on electrodes) are evidently significant factors in the voltage degradation characteristics for the DMRFC. The current-interrupt (0 mA/cm² for 10 sec) after 3.5 hours of operation did not have any notable effect on the DMRFC voltage degradation. Overall, the short-term durability tests demonstrate that the DMRFC performance is stable over the 4 hour time period.   Figure 3.9a. Short-term durability testing and cell impedance measurements for a perchlorate-based DMRFC operating at 50°C and 50 mA/cm². (Nafion® 112 membrane, 1 mL/min anolyte flow rate, 2 mL/min catholyte flow rate). A current interruption for 10 seconds (0 mA/cm²) was introduced after 3.5 hrs.  143  Figure 3.9b. Short-term durability testing and cell impedance measurements for a perchlorate-based DMRFC operating at 90°C and 50 mA/cm². (Nafion® 112 membrane, 1 mL/min anolyte flow rate, 2 mL/min catholyte flow rate). A current interruption for 10 seconds (0 mA/cm²) was introduced after 3.5 hrs.   144  Figure 3.9c. Short-term durability testing and cell impedance measurements for a sulfate-based DMRFC operating at 90°C and 50 mA/cm². (Nafion® 112 membrane, 1 mL/min anolyte flow rate, 2 mL/min catholyte flow rate). A current interruption for 10 seconds (0 mA/cm²) was introduced after 3.5 hrs.  3.4 Conclusions Electrolyte composition and cell temperature studies were performed to identify opportunities to improve the performance of the DMRFC. In one set of experiments, redox electrolytes prepared with Fe(ClO4)3 or Fe(NO3)3 were considered for DMRFC employment as a substitute for the FeNH4(SO4)2 catholyte in order to improve the catholyte charge density (i.e., solubility) and electrochemical performance. Electrolyte conductivity measurements, cyclic  145 voltammetry and DMRFC testing were used to evaluate the electrolytes. The Fe(ClO4)3 catholyte was determined to be a more suitable candidate than the FeNH4(SO4)2 catholyte due to its significantly higher solubility (more than 2.5 M vs. approx. 1 M), increased electrochemical activity, superior electrochemical reversibility and higher observed half-cell potential (0.83 V vs. SHE as opposed to 0.64 V vs. SHE at 90°C). The advantageous properties of the Fe(ClO4)3 catholyte became apparent through the increased DMRFC peak power density (79 mW/cm²) relative to that observed for the FeNH4(SO4)2 catholyte (25 mW/cm²). Anolyte composition studies demonstrated the best fuel cell performance at 16.7 M CH3OH (equimolar H2O/CH3OH) for the configuration studied. DMRFC temperature sensitivity tests indicated that maximizing the cell temperature (while respecting the anolyte and catholyte boiling points) delivers the best fuel cell performance. Short-term durability tests confirmed stable DMRFC performance over a 4 hr test period for a range of operating conditions. Further advancements in DMRFC technology are anticipated with future experiments targeted towards increasing the membrane conductivity, reducing Fe2+/Fe3+ crossover and exploring novel cell and electrode designs.  3.5 Acknowledgements The authors acknowledge the National Research Council, Institute for Fuel Cell Innovation (NRC-IFCI) and the Natural Sciences and Engineering Research Council of Canada (NSERC) for their financial support. The NRC-IFCI assisted in meeting the publication costs of this paper.  146 3.6 References 1. R. Dillon, S. Srinivasan, A. S. Arico, V. Antonucci, J. Power Sources, 127, 1-2 (2004) 112-126. 2. S. K. Kamarudin, W. R. W. Daud, S. L. Ho, U. A. Hasran, J. Power Sources, 163, 2 (2007) 743-754. 3. B. D. McNicol, D. A. J. Rand, K. R. Williams, J. Power Sources, 83, 1-2 (1999) 15-31. 4. S. Litster, C. R. Buie, T. Fabian, J. K. Eaton, J. G. Santiago, J. Electrochem. Soc., 154, 10 (2007) 1049-1058. 5. K. P. Adzakpa et al., J. Power Sources, 179, 1 (2008) 164-176. 6. R. Jiang, D. Chu, J. Electrochem. Soc., 155, 8 (2008) B798-B803. 7. J. B. Ge, H. T. Liu, J. Power Sources, 142, 1-2 (2005) 56-69. 8. J. G. Liu, T. S. Zhao, R. Chen, C. W. Wong, Electrochem. Commun., 7, 3 (2005) 288-294. 9. V. G. Sister, V. N. Fateev, D. A. Bokach, Russ. J. Electrochem., 43, 9 (2007) 1097-1100. 10. A. Lindermeir, G. Rosenthal, U. Kunz, U. Hoffmann, J. Power Sources, 129, 2 (2004) 180-187. 11. N. Carretta, V. Tricoli, F. Picchioni, J. Membr. Sci., 166, 2 (2000) 189-197. 12. A. Sungpet, J. Membr. Sci., 226, 1-2 (2003) 131-134. 13. A. Lam, D. P. Wilkinson, Z. Jiujun, in Proton Exchange Membrane Fuel Cells, T. Fuller, C. Bock, C. Lamy, Eds., PV 1, pp. 273-281, The Electrochemical Society Proceedings Series, Pennington, NJ, (2005).  147 14. S. Hikita, K. Yamane, Y. Nakajima, JSAE Review, 22, 2 (2001) 151-156. 15. D.P.Wilkinson, M. C. Johnson K. M. Colbow S. A. Campbell, US Patent No 5,672,439 (1997). 16. A. K. Shukla, R. K. Raman, Ann. Rev. Mater. Research, 33, (2003) 155-168. 17. S. Lj. Gojkovic, S. Gupta, R. F. Savinell, J. Electroanal. Chem., 462, 1 (1999) 63-72. 18. H. Tributsch, M. Bron, M. Hilgendorff, H. Schulenburg, I. Dorbandt, J. Appl. Electrochem., 31, 7 (2001) 739-748. 19. K. Fatih, D. P. Wilkinson, F. Moraw, F. Girard, in Electrocatalysis, R. Adzic, V. Birss, G. M. Brisard, A. Wieckowski, Eds., PV 11, pp. 341-350, The Electrochemical Society Proceedings Series, Pennington, NJ, (2005). 20. K. Fatih, D. P. Wilkinson, F. Moraw, A. Ilicic, F. Girard, Electrochem. Solid- State Lett., 11, 2 (2008) B11-B15. 21. G. B. Adams, R. P. Hollandsworth, D. N. Bennion, J. Electrochem. Soc., 122, 8 (1975) 1043-1048. 22. A. Mazuelos, F. Carranza, I. Palencia, R. Romero, Hydrometall., 58, 3 (2000) 269-275. 23. A. B. Ilicic, D. P. Wilkinson, K. Fatih, F. Girard, J. Electrochem. Soc., 155, 12 (2008) B1322-B1327. 24. A. B. Ilicic, D. P. Wilkinson, K. Fatih, F. Girard, in ECS Transactions, PV 16, pp. 1549-1560 (2008). 25. J. G. Speigt, Lange's Handbook of Chemistry, 16th Ed., McGraw-Hill, New York (2005).  148 26. J. M. Casas, G. Crisostomo, L. Cifuentes, Hydrometall., 80, 4 (2005) 254-264. 27. D. R. Lide, CRC Handbook of Chemistry and Physics, CRC Press LLC, Boca Raton (2005). 28. Y. H. Wen et al., J. Electrochem. Soc., 153, 5 (2006) 929-934. 29. A. Seidel, Kirk-Othmer Encyclopedia of Chemical Technology, John Wiley and Sons, Inc., (2001).    149 4. ADVANCING DIRECT LIQUID REDOX FUEL CELLS: MIXED REACTANT AND IN-SITU REGENERATION OPPORTUNITIES1 4.1 Introduction In recent years, interest in direct liquid fuel cells (DLFCs) has grown substantially with many commercial prototypes emerging [1, 2]. Core markets for DLFC technology include portable and micro applications, which stems from their compact nature and facile storage/handling of the fuel. The major technological challenges faced by conventional DLFC technology include cathode flooding, fuel crossover, CO2 product removal at the anode and sluggish anode and cathode kinetics [3, 4]. Water dragged from the anode to the air cathode by the electro- osmotic effect and water produced by the cathode reaction significantly impacts the degree of cathode flooding. Currently, Nafion® is the most widely used membrane in DLFC systems due to its good proton conductivity, mechanical durability and chemical resistance [5]. However, liquid fuels are known to permeate through Nafion® at rates where fuel loss and cathode depolarization become an issue, ultimately leading to reduced cell performance [6, 7]. In previous work, we emphasized the ability to address a number of the above DLFC issues by substituting the air cathode with a metal-ion redox couple over a carbon cathode [8, 9]. This alternative architecture, referred to as a direct liquid redox fuel cell (DLRFC), offers the advantages of cathode selectivity, no 1 A version of this chapter has been published.  A. B. Ilicic,  D. P. Wilkinson, K. Fatih. Advancing Direct Liquid Redox Fuel Cells: Mixed-Reactant and In-situ Regeneration Opportunities. J. Electrochem. Soc. 157, 4 (2010) B529-B535. Reproduced by permission of The Electrochemical Society.  150 platinum group metals (PGMs) at the cathode, no flooding issues, enhanced design flexibility at the cathode and the ability to use high fuel concentrations (e.g., 16.7 M CH3OH) without introducing cathode depolarization. Although many fuel/redox couple combinations are possible, our current work focuses on methanol and the Fe2+/Fe3+ redox couple. Our previous work has shown that the Fe2+/Fe3+ redox couple is suitable for use in a fuel cell [10, 11]. The relevant electrochemical reactions are:  Anode OHOHCHe6H6CO 232 +←++ −+  E0= 0.06 V vs. SHE (4.1) Cathode +−+ →+ 23 Fe6e6Fe6  E0= 0.77 V vs. SHE (4.2) Overall    +++ ++→++ 22323 Fe6H6COFe6OHOHCH  E0= 0.71 V (4.3)  Due to the cathode selectivity and the high crossover rate of methanol through the Nafion® membrane, a unique opportunity exists for a novel mixed- reactant DLRFC architecture (MR-DLRFC) where methanol crossover through the membrane is desirable! This approach, shown in Figure 4.1, involves feeding a mixed-reactant solution containing methanol, water, the redox couple and acid to the cathode while supplying nothing to the anode. The methanol permeates from the cathode to the anode where it is oxidized to produce CO2 and water. This new architecture allows for simpler manifolding, fewer auxiliary units, reduced cost and potentially increased system volumetric and gravimetric power density of the cell. In addition, the exiting gas at the anode is rich in CO2, which provides a significant cost advantage where CO2 separation and sequestration is sought.  151   Figure 4.1. Schematic of a mixed-reactant direct liquid redox fuel cell (MR- DLRFC).  Challenges associated with the DLRFC include regeneration of the redox couple, low charge density of the redox couple, membrane contamination by the redox couple and crossover of the redox couple. As the redox couple oxidant is reduced during cell discharge, it must be regenerated by chemical [12, 13], electrochemical [14] or biological means [15], which will affect the system’s overall efficiency and simplicity. In this work we present an in-situ electrochemical redox couple regeneration approach where no separate regeneration cell is required for the regeneration reactions (e.g., Fe2+ oxidation / O2 reduction).  When air is used to regenerate a redox couple, an auxiliary regeneration cell containing a suitable oxygen reduction catalyst is typically required. However, in the case of the DLRFC, a unique opportunity arises due to the hybrid architecture of the DLRFC in which noble metal catalyst (e.g., Pt/Ru) is available at the anode. Consequently, spontaneous DLRFC redox couple regeneration can be achieved by simply substituting the methanol anolyte with an air stream on the anode side.  152 The methanol anode then becomes an air cathode which reverses the direction of electron flow and regenerates the redox couple. A direct comparison of a conventional DLRFC regeneration system against the in-situ DLRFC approach is given in Figure 4.2, which shows that in-situ regeneration presents an opportunity to significantly reduce the overall system cost, size and complexity. One disadvantage for this type of regeneration configuration is that the DLRFC must be operated in either discharge or regeneration mode as it is not possible to perform both simultaneously. The electrochemical reactions for in-situ DLRFC regeneration are shown below:  Anode +−+ ←+ 23 Fe2e2Fe2  E0= 0.77 V vs. SHE (4.4) Cathode OHe2H2O2 1 22 →++ −+  E0= 1.23 V vs. SHE (4.5) Overall     OHFe2H2O2 1Fe2 2 3 2 2 +→++ +++  E0= 0.46 V (4.6)   Figure 4.2. Comparison of conventional vs. in-situ DLRFC regeneration.   153 4.2 Experimental A variety of experiments were conducted to assess the feasibility and performance of a MR-DLRFC: differential scanning calorimetry (DSC), cyclic voltammetry (CV), electrolyte conductivity measurements and fuel cell tests. In addition, a separate set of fuel cell tests was performed to investigate in-situ regeneration of the Fe2+/Fe3+ redox couple. The mixed electrolyte that was used for the MR-DLRFC tests is referred to as the “methanol redox electrolyte”. The electrolyte consisted of 0.81 M Fe(III)NH4(SO4)2•12H2O (ACS, Fisher), 0.09 M Fe(II)SO4•7H2O (ACS, Fisher), 0.5 M H2SO4 (ACS, Fisher) and some concentration of methanol (electronic grade, Fisher). This electrolyte represents the mixed electrolyte that was used for the MR-DLRFC tests as well cyclic voltammetry (CV) and electrolyte conductivity tests. If no methanol is present in an electrolyte, it is simply referred to as a “redox electrolyte”. Deionized water (18 MΩ·cm) was used to prepare all electrolytes reported herein. Differential scanning calorimetry (DSC) was conducted on 2 M methanol redox electrolytes. Shavings of Toray Carbon Fiber Paper (TGP-H-060 CFP, E- TEK) were dispersed in one electrolyte but were absent in the other for comparison. Aluminum hermetic pans with an o-ring provided a high pressure seal for the aqueous samples as they were heated from 25°C to 200°C at a rate of 1°C/min. Sample masses were in the range of 50-75 mg. Cyclic voltammetry (CV) tests were conducted at 70°C using a 5 mm Glassy Carbon (GC) working electrode (AFE1E050GC, Pine Instruments). The  154 counter electrode was a Pt flag (XM120, Radiometer) and the reference electrode was a Hg/Hg2SO4/saturated K2SO4 reference electrode (XR200, Radiometer). The temperature of the glass cell was maintained at 70°C by submerging the cell in a constant temperature bath (Neslab EX-7, Thermo Electron) and electrolytes were de-aerated by bubbling N2 for 30 minutes prior to testing under static conditions. Electrolytes tested included a redox electrolyte and a 2 M methanol redox electrolyte. The GC electrode was prepared by rinsing with alcohol, polishing with 0.05 µm alumina, rinsing with deionized water, sonicating in deionized water for two minutes and finally rinsing in deionized water. A Solartron 1260 frequency response analyzer was used to obtain the internal cell resistance for IR correction of the cyclic voltammograms. The conductivity of 1, 2, and 4 M methanol redox electrolytes were measured using a YSI 3200 Conductivity Instrument equipped with a YSI 3252 conductivity probe. Conductivities were measured as a function of temperature by immersing an enclosed glass cell into a constant-temperature water bath with the probe inserted into the glass cell. Electrolyte conductivity measurements were taken at 25, 50, 75 and 90°C. Care was taken to maintain a sealed glass cell to avoid concentration changes in the electrolyte due to methanol evaporation. The conductivity of Nafion® 112 membranes was measured at 25°C after exposure to various electrolytes at 90°C. Strips of 1 cm x 2 cm protonated membranes were exposed to the electrolyte for 4 hours, rinsed and subsequently immersed in deionized water for 2 hours to clear the pores of residual ions. A  155 Solartron 1260 frequency response analyzer was used to measure the impedance of the membranes from 1 kHz to 10 MHz (10 mV perturbation) in a custom-built conductivity cell equipped with Pt electrodes. The resulting impedance data were regressed against a suitable equivalent circuit model to extract the resistance (Rmem) of the membrane. The conductivity (σ) was then calculated using the following equation:  AR L mem =σ  (4.7) where L is the distance between the Pt electrodes and A is the cross-sectional area of the membrane. MR-DLRFC performance testing was conducted in a single-pass fuel cell test system, shown in Figure 4.3. An expanded view of the custom-built 4 cm² fuel cell is shown in Figure 4.4. The anode catalyst loading was 2 mg/cm² (40% 1:1 a/o Pt/Ru on Vulcan XC-72, E-Tek) on TGP-H-060 CFP. Nafion® ionomer (5% in H2O/alcohol, Sigma Aldrich) was added to the anode catalyst ink as a binding agent in the amount of 30% w/w (i.e., 30% Nafion® ionomer, not solution). No solution was fed to the anode, which contained a serpentine flow field with a channel width and depth of 0.84 mm and 0.46 mm, respectively. The cathode consisted of 3 layers of 370 µm thick TGP-H-120 CFP (no PGM catalyst) packed into a 1 mm deep empty pocket. A 1, 2, or 4 M methanol redox electrolyte was supplied at 5 mL/min to the cathode, which fueled both the anode and the cathode. Nafion® 112 membranes were employed in all configurations and were cleaned and protonated by soaking in deionized H2O for 2 h, boiling in 3 % H2O2 for 30 min, rinsing in deionized H2O, boiling in 0.5 M H2SO4 for 30 min,  156 rinsing in deionized H2O and finally storing in deionized H2O for at least 24 h. Nafion® membranes were not hot-pressed to the electrodes. In all configurations, the nitrogen bladder, shown in Figure 4.4, was pressurized to approx. 200 kPa abs. to compress the fuel cell. The fuel cell was oriented vertically with the fuel cell inlets located at the bottom and the exits at the top. A mercury sulfate reference electrode was used to measure the anode and cathode potentials during galvanostatic fuel cell operation. As shown in Figure 4.3, the mercury sulfate reference electrode was ionically connected to the unused anode inlet, which was filled with stagnant 0.5 M H2SO4. A wetted ground glass plug interface was used to ensure that no flow of the stagnant electrolyte occurred during fuel cell operation. The cell impedance used for IR-correction of polarization curves was obtained after cell testing by taking the real component of the cell impedance measured at 20 kHz using a GwINSTEK LCR-821 unit.  157  Figure 4.3. Schematic of the mixed reactant direct liquid fuel cell system.   Figure 4.4. Expanded view of the 4 cm² fuel cell.   158 Regular DLRFC tests (non-mixed reactant) were also conducted to provide a comparison against the MR-DLRFC data. In this case, all operating parameters and components were identical to the MR-DLRFC case except for the electrolytes: the cathode was supplied with a redox electrolyte containing no methanol whereas the anode was supplied with an aqueous electrolyte containing 2 M CH3OH and 0.5 M H2SO4. In-situ electrochemical DLRFC regeneration performance testing was conducted at 70°C and 90°C in a single-pass 4 cm² fuel cell using galvanostatic measurements. The anode (cathode in discharge mode) was supplied with 1 mL/min of 0.9 M FeSO4 and 1 M H2SO4 in deionized water, which represents a redox electrolyte with a 0% state-of-charge (SOC). The H2SO4 concentration here (vs. 0.5 M for the DLRFC tests) reflects the accumulation of charge-carrying protons in the DLRFC catholyte as the cell is discharged. Three layers of 370 µm thick TGP-H-120 CFP were used as the electrode at the anode (no catalyst) which employed a 1 mm deep empty pocket flow field. The cathode (anode in discharge mode) was supplied with 30 mL/min of dry air at 101.3 kPa abs. through a single-channel serpentine flow field plate with a channel width and depth of 0.84 mm and 0.46 mm, respectively. The cathode catalyst loading was 2 mg/cm² (40% 1:1 a/o Pt/Ru on Vulcan XC-72, E-Tek) on TGP-H-060 CFP. Nafion® ionomer (5% in H2O/alcohol, Aldrich) was added to the cathode catalyst ink as a binding agent in the amount of 30% w/w (i.e., 30% Nafion® ionomer, not solution). Nafion® 112 membranes were employed for both temperatures and were cleaned and protonated by the procedures given previously. Nafion®  159 membranes were sandwiched between the electrodes but not hot pressed. Cells were conditioned for at least 30 min prior to testing to allow the membrane to approach an equilibrium state. In addition, batch redox electrolyte in-situ regeneration with a recycled 100 mL redox electrolyte was performed at 70°C in a short-circuited cell. All of the cell configuration parameters were identical to the above, apart from the fact that the redox electrolyte was recycled.  4.3 Results and Discussion In any mixed-reactant fuel cell it is imperative to confirm chemical compatibility between the fuel and oxidant. The DSC and CV experiments outlined below were selected such that confidence in the chemical compatibility could be established for a relevant temperature range (25-90°C) and potential window (0-1.2 V vs. SHE) in the presence or absence of carbon. DSC was carried out on 2 M methanol redox electrolytes in the presence or absence of fine shavings of TGP-H-060 CFP. The absence of an exotherm in the results, shown in Figure 4.5, indicates that methanol and Fe3+ ions do not react with or without carbon in the temperature range of interest for the DLRFC (below 100°C). These results are consistent with Zawadzki et al. [16], who reported no reactivity of methanol over carbon up to 673 K. It follows that no direct redox reaction would take place if a mixed electrolyte consisting of both methanol and Fe3+ ions were supplied to a DLRFC carbon cathode, which is the case for a MR-DLRFC.  160 -5 -4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0 25 50 75 100 125 150 175 200 Temperature (°C) H ea t R at e (m W ) (b) (a)  Figure 4.5. Differential scanning calorimetry results for a 2 M methanol redox electrolyte (no acid) (a) with and (b) without carbon particles.  Cyclic voltammograms were acquired at 70°C at a scan rate of 50 mV/s in the potential window of 0-1.2 V vs. SHE. Stable voltammograms were obtained within 10 cycles and were subsequently IR corrected. Voltammograms of the Fe2+/Fe3+ redox couple, shown in Figure 4.6, convey similar redox activity in the presence and absence of 2 M methanol in the redox electrolyte. Table 4.1 summarizes the peak currents, peak potentials and peak separations of the voltammograms presented in Figure 4.6. These data show that the presence of 2 M methanol affects the electrochemical activity and reversibility of the Fe2+/Fe3+ redox couple by reducing the peak current density by approximately 32% and  161 increasing the peak separation by approximately 42%. However, this effect is not so adverse as to signify a direct redox reaction between methanol and Fe3+ ions. It appears that physical adsorption of the methanol on the GC electrode may be responsible for the reduced electrochemical activity and reversibility. Some researchers have used methanol as a refrigerant over carbon substrates [17] or as an inhibitor for carbon nanotube growth [18] due to methanol’s adsorption properties on various carbon substrates. Nevertheless, these data indicate that the cathode of a MR-DLRFC can be fed with 2 M methanol without overly suppressing the electrochemical activity of the redox couple as does occur on Pt. This finding is consistent with the DSC results shown in Figure 4.5, which confirm the thermal stability of methanol and ferric ions on carbon over the temperature range of interest for the MR-DLRFC.  162 -100 -80 -60 -40 -20 0 20 40 60 80 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Potential (V vs. SHE) C ur re nt  D en si ty  (m A /c m ²) (a) (b)  Figure 4.6. Comparison of cyclic voltammograms: (a) 0 M and (b) 2 M methanol redox electrolytes over GC (70°C, 50 mV/s scan rate, IR-corrected).   Table 4.1. Summary of anodic and cathodic peak current densities (ip) and peak potentials (Ep) for cyclic voltammograms shown in Figure 4.6.   163 The conductivities of the Fe2+/Fe3+ redox electrolyte and various methanol redox electrolytes (1, 2 and 4 M CH3OH) were measured from 25 to 90°C. Reported conductivities are accurate within ± 15 mS/cm. Interestingly, the results, shown in Figure 4.7, indicate that the presence of methanol reverses the effect of temperature on conductivity. Where no methanol is present, the conductivity decreases with temperature from 142 to 107 mS/cm. This trend is attributed to the formation of complexes such as FeH(SO4)2 and HSO4- at elevated temperatures [19]. Conversely, in the 1, 2, and 4 M methanol redox electrolytes, the conductivity increases with temperature leading to conductivities in the range of 177 to 263 mS/cm at higher temperatures (70-90°C). The inversion of the conductivity temperature dependence upon the addition of methanol is most likely due to methanol inhibiting the formation of complexes (e.g., FeH(SO4)2 and HSO4-). It was also observed that the conductivity decreases with increasing methanol concentration, which may be due to an impeding effect of methanol on the Grotthus mechanism for proton transport. Furthermore, the substitution of water molecules for methanol in the proton solvation shell size may provide additional drag due to increased diameter, as implied by the work of Schaffer et al. [20]. In any case, the methanol redox electrolytes all exhibit a sufficiently high ionic conductivity for the catholyte in a MR-DLRFC, particularly at elevated temperatures.   164 0 50 100 150 200 250 300 20 30 40 50 60 70 80 90 100 Temperature (°C) C on du ct iv ity  (m S/ cm ) 0 M MeOH 1 M MeOH 2 M MeOH 4 M MeOH  Figure 4.7. Conductivity of the methanol redox electrolyte as a function of methanol concentration and temperature.  The conductivities of protonated Nafion® 112 membranes were measured after exposure to one of the following electrolytes: 1) deionized water (control); 2) the redox electrolyte; 3) a 4 M CH3OH redox electrolyte. The measured conductivities, shown in Table 4.2, indicate a reduction in membrane conductivity of about 88% upon exposure to the redox electrolyte, regardless of the presence of methanol. This is due to the substitution of mobile protons on the sulfonic sites within the Nafion® membrane for larger, less mobile ferric and ferrous cations.  165 Table 4.2. Summary of Nafion® 112 membrane conductivities at 25°C after membranes were exposed to various electrolytes.  aMembranes soaked in electrolyte for 4 hrs at 90°C, rinsed with de-ionized water and soaked in de-ionized water for 2 hrs prior measuring the conductivity at 25°C.  Fuel cell testing for the mixed reactant architecture was performed at temperatures of 70 and 90°C and methanol concentrations of 1, 2, and 4 M in the methanol redox electrolyte. Higher concentrations of methanol in the mixed electrolyte were not tested due to the negative effect of methanol on the activity of the redox couple and the observed long term instability of the mixed electrolyte at high methanol concentrations (e.g., 8 M CH3OH). The reason for salt precipitation at high methanol concentrations is likely related to the lower solubility of the inorganic iron salt in methanol than water. Cell voltage-current data (i.e., polarization data) were acquired as well as individual anode and cathode potentials through the use of a reference electrode ionically connected to the unused anode inlet of the MR-DLRFC. The ionic connection between the reference electrode and anode, shown in Figure 4.3, was established by filling the interconnecting tubing with stagnant 0.5 M H2SO4. The anode was not flooded due to the presence of this electrolyte since the electrolyte is stagnant and the anode chamber was purged with product CO2 at a high current density prior to testing. The polarization curves as a function of temperature and methanol concentration for the cell, anode and cathode are shown in Figures  166 4.8a, 4.8b and 4.8c, respectively. The cell polarization data indicates a significant increase in overall performance with increasing temperature. Peak power densities increased from 13-16 mW/cm² at 70°C to 22-26 mW/cm² at 90°C. No significant improvement in cell performance was observed when testing a higher mixed electrolyte flow rate (10 mL/min vs. 5 mL/min, not shown), suggesting the absence of significant concentration overpotential. A minor improvement in cell performance with decreasing methanol concentration is observed at both temperatures, which is better understood through the individual electrode polarization curves. In comparing the non-mixed DLRFC cell polarization data to that of the MR-DLRFC at 2 M CH3OH and 70°C in Figure 4.8a, it is apparent that DLRFC yields significantly higher performance than the MR-DLRFC at 70°C. The power density curves for the cell polarization data in Figure 4.8a are shown in Figure 4.9. From this graph it is apparent that the peak power density of the non- mixed DLRFC with 2 M CH3OH at 70°C is 24 mW/cm² while that of an equivalent MR-DLRFC is 14 mW/cm². The dominant loss mechanism is most likely reduced electrochemical activity of the redox couple in the MR-DLRFC due to the presence of methanol at the cathode. The anodic polarization behaviour, shown in Figure 4.8b, conveys an interesting point: the anode polarization behaviour is not significantly influenced by the methanol concentration despite it being the fuel supply driving force. This implies that even a 1 M methanol redox electrolyte provides a sufficient methanol crossover rate to the anode without introducing significant concentration overpotential at high current densities. The fuel cell performance data is  167 consistent with reported methanol crossover rates through Nafion 112 (hot- pressed into an MEA) at 90³C of 3.33x10-7, 6.67x10-7 and 9.33x10-7 mol/cm²·s for 1.2, 2.5 and 2.7 M CH3OH, resprectively [21]. At these crossover rates, complete methanol oxidation can proceed at current densities of 193, 386 and 540 mA/cm², respectively. Lastly, strong kinetic improvements at the anode were observed when increasing the temperature from 70 to 90°C. The polarization behaviour at the cathode, shown in Figure 4.8c, provides an explanation for the minor increase in overall cell performance with decreasing methanol concentration. Increasing the methanol concentration appears to have a minor negative impact on the activity of the redox cathode, which is consistent with the CV results. This phenomenon is most likely due to physical adsorption of methanol on the carbon cathode, as stated previously. The relatively high overpotential observed with the cathode data after IR-correction may also be due to methanol adsorption on the carbon electrode. This effect needs further investigation since we expect the redox reaction to be kinetically fast based on the CV data. The absence of significant cathode depolarization (normally apparent through a negatively shifted OCP at the cathode), even though a mixed reactant electrolyte is being supplied to the cathode, is characteristic of the selective carbon cathode. Lastly, increasing the temperature from 70 to 90°C significantly reduces the cathodic overpotential.   168 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0 50 100 150 200 250 300 Current Density (mA/cm²) C el l P ot en tia l ( V) 70°C, 1 M MeOH 70°C, 2 M MeOH 70°C, 4 M MeOH DLRFC, 70°C, 2 M MeOH 90°C, 1 M MeOH 90°C, 2 M MeOH 90°C, 4 M MeOH  Figure 4.8. Cell polarization behaviour (IR-corrected) for a MR-DLRFC as a function of temperature and methanol concentration. Mixed electrolyte supplied to cathode at 5 mL/min and comprised of X M CH3OH, 0.81 M FeNH4(SO4)2, 0.09 M FeSO4, 0.5 M H2SO4. DLRFC reference data is also included for comparison (anolyte composition 2 M CH3OH, 0.5 M H2SO4; anolyte flow rate 5 mL/min; catholyte composition 0.81 M FeNH4(SO4)2, 0.09 M FeSO4, 0.5 M H2SO4; catholyte flow rate 5 mL/min).    (a)  169 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0 50 100 150 200 250 300 Current Density (mA/cm²) A no de  P ot en tia l ( V) 70°C, 1 M MeOH 70°C, 2 M MeOH 70°C, 4 M MeOH 90°C, 1 M MeOH 90°C, 2 M MeOH 90°C, 4 M MeOH  Figure 4.8b. Anode polarization behaviour (IR-corrected) for a MR-DLRFC as a function of temperature and methanol concentration. Mixed electrolyte supplied to cathode at 5 mL/min and comprised of X M CH3OH, 0.81 M FeNH4(SO4)2, 0.09 M FeSO4, 0.5 M H2SO4. (anolyte composition 2 M CH3OH, 0.5 M H2SO4; anolyte flow rate 5 mL/min; catholyte composition 0.81 M FeNH4(SO4)2, 0.09 M FeSO4, 0.5 M H2SO4; catholyte flow rate 5 mL/min).  (b)  170 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 0 50 100 150 200 250 300 Current Density (mA/cm²) C at ho de  P ot en tia l ( V) 70°C, 1 M MeOH 70°C, 2 M MeOH 70°C, 4 M MeOH 90°C, 1 M MeOH 90°C, 2 M MeOH 90°C, 4 M MeOH  Figure 4.8c. Cathode polarization behaviour (IR-corrected) for a MR-DLRFC as a function of temperature and methanol concentration. Mixed electrolyte supplied to cathode at 5 mL/min and comprised of X M CH3OH, 0.81 M FeNH4(SO4)2, 0.09 M FeSO4, 0.5 M H2SO4. (anolyte composition 2 M CH3OH, 0.5 M H2SO4; anolyte flow rate 5 mL/min; catholyte composition 0.81 M FeNH4(SO4)2, 0.09 M FeSO4, 0.5 M H2SO4; catholyte flow rate 5 mL/min). (c)  171 0 5 10 15 20 25 30 35 40 0 50 100 150 200 250 300 Current Density (mA/cm²) Po w er  D en si ty  (m W /c m ²) 70°C, 1 M MeOH 70°C, 2 M MeOH 70°C, 4 M MeOH DLRFC, 70°C, 2 M MeOH 90°C, 1 M MeOH 90°C, 2 M MeOH 90°C, 4 M MeOH  Figure 4.9. Power density curves calculated from the data shown in Figure 4.8a.  In summary, the concept of a MR-DLRFC was successfully demonstrated, highlighting in particular the effects of temperature and methanol concentration in the methanol redox electrolyte. It was shown that the high crossover rate of methanol through Nafion® can be exploited to realize the MR-DLRFC architecture. The unique mixed-reactant configuration creates an opportunity to reduce the cost and enhance the volumetric and gravimetric power density of the DLRFC by eliminating the need for a fuel supply line and reducing the complexity and size of the anode compartment.  172 In-situ electrochemical redox couple regeneration in a DLRFC was studied using two experimental approaches: 1) generation of galvanostatic polarization curves for a single pass fuel cell and 2) determination of the current density and SOC over time during short-circuit regeneration of a recycled 100 mL redox electrolyte (initially at 0% SOC). In both cases, power is produced during regeneration as opposed to conventional regeneration approaches where power is required. Single pass fuel cell experiments were conducted using a 0% SOC redox electrolyte at cell temperatures of 70 and 90°C. The resulting polarization curves are shown in Figure 4.10a. The overall performance is nearly identical at both temperatures with an open circuit potential (OCP) around 0.18 V. The discrepancy between the theoretical OCP for these operating conditions (0.41 V for 90°C, Fe2+:Fe3+=100:1, 21% O2) and the actual OCP (0.18) is understandable since air cathodes typically exhibit an OCP much lower than the theoretical value (i.e., in the vicinity of 0.95 V vs. SHE). Taking this into consideration, the observed OCP of the Fe2+/air electrochemical cell is a reasonable value. At both temperatures the cell voltage drops off rapidly with increasing current density yielding a maximum current density of approximately 20 mA/cm2. The relatively low current density achieved with this system is due to the low theoretical OCP and high oxygen reduction overpotential. The batch regeneration experiment of a 100 mL recycled redox electrolyte at 70°C provides an alternative perspective on the performance of in-situ DLRFC redox couple regeneration. In this case, both current density and total charge  173 passed were recorded as the regeneration cell was potentiostatically at 0 V. Based on the volume (100 mL) and concentration of the redox couple (0.9 M Fe2+), the total number of coulombs available to be passed was determined (8684 C). Thus the fraction of 8684 C passed represents the current SOC in the cell. The current density and SOC as a function of time for this batch regeneration experiment during short-circuit conditions is shown in Figure 4.10b. The initial current density (22 mA/cm2) agrees with that observed in the single- pass experiment. Over the 80 hr testing period, the current density decays exponentially to approximately 0 mA/cm2 as the Fe2+ reactant is oxidized to Fe3+. Similarly, the SOC asymptotically approaches the maximum achievable SOC for this experiment of 87% over the 80 hr testing period. Factors influencing the maximum achievable SOC include: 1) The diminishing driving force for regeneration as the redox potential of the Fe2+/Fe3+ redox couple increases during regeneration; 2) The increased crossover rate of Fe3+ ions and consequently air cathode depolarization as the Fe3+ concentration at the redox anode increases during regeneration; 3) Cell design parameters influencing the rate of Fe3+ ion crossover. Practical DLRFC discharge and in-situ regeneration cycling times would be determined as a function of the regeneration rate and the SOC that the cell can produce useful power.   174 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.20 0 5 10 15 20 25 Current Density (mA/cm²) C el l P ot en tia l ( V) 70°C 90°C  Figure 4.10a. Galvanostatically measured polarization curves for in-situ redox couple regeneration in single-pass mode at 70 and 90°C. (anolyte composition 0.9 M FeSO4, 1 M H2SO4 anolyte flow rate 5 mL/min; cathode air flow rate 38 mL/min [@ 25°C and 101.3 kPa abs.]; air not humidified).  (a)  175 0 10 20 30 40 50 60 70 80 90 100 0 20 40 60 80 Time (h) %   S ta te  o f C ha rg e 0 5 10 15 20 25 C ur re nt  D en si ty  (m A /c m ²)  Figure 4.10b. SOC and current density as a function of time for a 100 mL batch in-situ redox regeneration test measured potentiostatically at 0 V and 70°C. (anolyte composition 0.9 M FeSO4, 1 M H2SO4 [at t=0]; anolyte flow rate 5 mL/min; cathode air flow rate 38 mL/min [@ 25°C and 101.3 kPa abs.]; air not humidified).  In-situ redox couple regeneration in a DLRFC has been successfully demonstrated with data generated for both single-pass and batch regeneration modes. It has been shown that the maximum current density and %SOC for this regeneration approach at 70°C is 22 mA/cm2 and 87%, respectively. This regeneration architecture provides an avenue to reduce the number of auxiliary units, size and cost of a DLRFC system and can potentially provide useful power. (b)  176 4.4 Conclusions A novel mixed reactant approach for the direct liquid redox fuel cell (DLRFC) which makes use of fuel crossover was successfully demonstrated for the Fe2+/Fe3+ methanol system. The cell architecture for this approach is particularly advantageous in allowing for simpler manifolding, fewer auxiliary units, reduced cost and potentially increased system volumetric and gravimetric power density. Differential scanning calorimetry, cyclic voltammetry and conductivity tests were collectively used to show adequate the thermal and electrochemical stability, adequate electrochemical activity and adequate electrolyte conductivity of the methanol redox electrolyte. Peak fuel cell performance (26 mW/cm²) was observed for highest temperature (90°C) and the lowest methanol concentration (1 M CH3OH) tested. Further performance improvements are anticipated over this non-optimized performance with advancements in the electrolyte composition and membrane. In-situ DLRFC regeneration of Fe3+ ions up to an 87% state of charge was accomplished by substituting the methanol fuel supply with air, thus spontaneously reversing the direction of electron flow. This regeneration approach is unique in that regeneration is spontaneous and produces power in contrast to conventional regeneration approaches which require power. Also, the in-situ architecture creates an opportunity to reduce the overall system cost, size and complexity as the need for an auxiliary regeneration unit is eliminated. Further improvements in the regeneration rate and maximum achievable SOC are expected by optimizing the cathode catalyst for oxygen reduction, improving  177 the membrane conductivity and reducing the crossover rate of Fe3+ ions through the membrane.  4.5 Acknowledgements The authors acknowledge the National Research Council, Institute for Fuel Cell Innovation (NRC-IFCI) and the Natural Sciences and Engineering Research Council of Canada (NSERC) for their financial support. The NRC-IFCI assisted in meeting the publication costs of this paper.    178 4.6 References 1. R. Dillon, S. Srinivasan, A. S. Arico, V. Antonucci, J. Power Sources, 127, 1-2 (2004) 112-126. 2. C. M. Miesse, J. Power Sources, 162, 1 (2006) 532-540. 3. H. Yang, T. S. Zhao, Electrochim. Acta, 50, 16-17 (2005) 3243-3252. 4. W. Qian, D. P. Wilkinson, J. Shen, H. Wang, J. Zhang, J. Power Sources, 154, 1 (2006) 202-213. 5. V. Neburchilov, J. Martin, H. Wang, J. Zhang, J. Power Sources, 169, 2 (2007) 221-238. 6. J. B. Ge, H. T. Liu, J. Power Sources, 142, 1-2 (2005) 56-69. 7. J. G. Liu, T. S. Zhao, R. Chen, C. W. Wong, Electrochem. Commun., 7, 3 (2005) 288-294. 8. A. B. Ilicic, D. P. Wilkinson, K. Fatih, F. Girard, J. Electrochem. Soc., 155, 12 (2008) B1322-B1327. 9. A. B. Ilicic, D. P. Wilkinson, K. Fatih, F. Girard, in ECS Transactions, PV 16, pp. 1549-1560 (2008). 10. K. Fatih, D. P. Wilkinson, F. Moraw, F. Girard, in Electrocatalysis, R. Adzic, V. Birss, G. M. Brisard, A. Wieckowski, Eds., PV 11, pp. 341-350, The Electrochemical Society Proceedings Series, Pennington, NJ, (2005). 11. K. Fatih, D. P. Wilkinson, F. Moraw, A. Ilicic, F. Girard, Electrochem. Solid- State Lett., 11, 2 (2008) B11-B15. 12. R. Pattabiraman, V. K. Venkatesan, H. V. K. Udupa, Pak. J. Sci. Ind. Res., 40, 7 (1981) 432-447.  179 13. J. T. Kummer, D. G. Oei, J. App. Electrochem., 12, (1982) 87-100. 14. G. B. Adams, R. P. Hollandsworth, D. N. Bennion, J. Electrochem. Soc., 122, 8 (1975) 1043-1048. 15. A. Mazuelos, F. Carranza, I. Palencia, R. Romero, Hydrometall., 58, 3 (2000) 269-275. 16. J. Zawadzki, B. Azambre, O. Heintz, A. Krzton, J. Weber, Carbon, 38, 4 (2000) 509-515. 17. I. I. El-Sharkawy, M. Hassan, B. B. Saha, S. Koyama, M. M. Nasr, Int. J. Refrig., 32, 7 (2009) 1579-1586. 18. Z. P. Wu, J. N. Wang, J. Ma, Carbon, 47, 1 (2009) 324-327. 19. J. M. Casas, G. Crisostomo, L. Cifuentes, Hydrometall., 80, 4 (2005) 254-264. 20. T. Schaffer, T. Tschinder, V. Hacker, J. O. Besenhard, J. Power Sources, 153, 2 (2006) 210-216. 21. S. Hikita, K. Yamane, Y. Nakajima, JSAE Review, 22, 2 (2001) 151-156.    180 5. CONCLUSIONS The conventional direct methanol fuel cell (DMFC) has gained wide recognition as an energy conversion device suitable for portable and micro applications due to the high volumetric energy density of the fuel and facile storage and handling of the fuel. The DMFC also offers some key advantages over H2 proton exchange membrane fuel cells (PEMFCs) such as no need for external humidification or cooling equipment. However, the commercial viability of DMFCs is currently hindered by various key issues including methanol crossover, poor anode/cathode kinetics, cathode flooding, and cost. The central objective of this work was to address a number of the key issues present in DMFCs by investigating an alternative cell configuration involving elements of both the DMFC and the redox flow battery (RFB). This hybrid configuration, referred to as the direct liquid redox fuel cell (DLRFC), was shown to have a number of significant advantages over the conventional DMFC. The unique configuration of the DLRFC also allowed for a novel mixed-reactant cell architecture to be studied, which represents a significant portion of this work. The research presented in this thesis focused on the development of the DLRFC, a novel variant to the conventional DMFC. Within this framework, the following objectives were identified: (1) demonstration of the concept and advantages of a DLRFC, (2) improvement of the performance of the DLRFC, (3) demonstration of a novel mixed-reactant cell architecture for the DLRFC and (4) investigation of the electrochemical regeneration of the redox couple.   181 In Chapter 2 of this thesis, the concept and key advantages of a DLRFC were demonstrated and performance of the DLRFC was characterized for various operating conditions. In Chapter 3, it was shown that significant improvements in the DLRFC performance and volumetric charge density of the redox electrolyte can be obtained through the use of a perchlorate-based iron salt in the redox electrolyte. In Chapter 4, a non-conventional mixed-reactant architecture was demonstrated for the DLRFC where methanol crossover is the mode of fuel supply to the anode. In addition, some initial experiments performed on a novel approach to in-situ DLRFC redox couple regeneration were included in Chapter 4. Collectively, the research reported in this thesis serves as the groundwork for an alternative fuel cell system and provides a preliminary investigation for a non-conventional mixed-reactant fuel cell architecture and a novel in-situ approach to redox couple regeneration. The following sections are intended to briefly summarize the work presented, highlight the significance and ultimate impact of the research conducted in light of knowledge within the scientific community, discuss the potential applications of the systems studied, and provide recommendations for future experimental work on the DLRFC.  5.1 The Direct Liquid Redox Fuel Cell Approach 5.1.1 Summary For DMFC systems, methanol crossover is a primary challenge since it leads to multiple problems, including cathode depolarization, fuel loss and  182 potentially oxidant starvation [1-5]. Methanol crossover from the anode to the cathode arises due to the concentration gradient present across the membrane and the high methanol permeability of the (Nafion®) membrane. Once the methanol fuel migrates to the cathode, it is lost and compromises the efficiency of the system. Since oxygen and Pt catalyst are present at the cathode, crossed over methanol has the opportunity to react with oxygen and compromise the cathode performance. Under certain circumstances, the rate of oxygen consumption due to methanol crossover can deplete the cathode of oxygen, leading to severe performance losses. For these reasons, reducing or mitigating methanol crossover in a DMFC is of paramount importance. Although the DMFC does not require complex humidification equipment, water management at the cathode represents a challenge. As the cell undergoes discharge, charge-balancing protons migrate from the anode to the cathode and carry a significant amount of water with them. Furthermore, water is produced at the cathode during oxygen reduction. These factors can lead to cathode flooding at higher current densities, which inhibits oxygen access to the cathode [6-8]. The reaction kinetics for the DMFC anode and cathode are relatively sluggish and require the application of high platinum group metal (PGM) catalyst loadings (more than 1 mg/cm²) to obtain reasonable performance [9-12]. Consequently, the DMFC catalyst accounts for a substantial portion of the total system cost, which negatively influences the commercial viability of this system.  183 All of the above DMFC issues can be addressed wholly or partially through the use of a novel configuration involving the substitution of the DMFC air cathode with a redox couple cathode, as shown in Figure 5.1.   Figure 5.1. Schematic of a direct liquid redox fuel cell (DLRFC).  In this configuration, referred to as the direct liquid redox fuel cell (DLRFC), the redox couple exhibits good electrochemical activity over carbon [13] and thus no PGM catalyst is required at the cathode. The absence of PGM catalyst on the DLRFC cathode mitigates the issues of methanol crossover and catalyst cost. Furthermore, cathode flooding is not an issue for the DLRFC since the redox couple is supplied to the cathode as a liquid electrolyte (i.e., no  184 gaseous reactants). The following items are the key research outcomes of the DLRFC tests discussed in Chapter 2:  • Demonstration of the DLRFC – The air cathode over Pt catalyst in the DMFC was replaced with the Fe2+/Fe3+ redox couple over carbon and fuel cell testing was performed. It was possible to employ a three-dimensional carbon cathode since the redox electrolyte can conduct protons and a TPB is not required.  • Observed fuel flexibility for the DLRFC – DLRFC operation was demonstrated using both methanol and formic acid as the fuel. Both fuels delivered performance similar to or surpassing that observed with a conventional fuel cell incorporating an air cathode. Other low-molecular weight organic liquid fuels such as ethanol or dimethyl ether could also be used in the DLRFC.  • No PGM catalyst required at the DLRFC cathode – Cyclic voltammetry results confirmed comparable electrochemical activity and reversibility of the Fe2+/Fe3+ redox couple over glassy carbon relative to that observed on Pt. Consequently, it is possible to employ a PGM-free carbon cathode in the DLRFC while retaining electrochemical activity for the redox couple similar to that attainable on Pt.   185 • Cathode selectivity achieved – The DLRFC carbon cathode is not active towards methanol oxidation. After methanol from the anode undergoes crossover to the cathode, it is not able to react there and no cathode depolarization or starvation can occur. Cyclic voltammetry, differential scanning calorimetry and fuel cell tests confirmed that the electrochemical activity of the redox couple is not significantly changed by the presence of methanol. This behaviour is representative of a selective electrode.  • High fuel concentrations possible – Because the DLRFC cathode is selective towards the cathode reaction, high fuel concentrations at the anode can be utilized without concern for performance losses due to fuel crossover. Fuel concentrations of 16.7 M and 18 M for methanol and formic acid, respectively, were tested in a fuel cell environment. Improved performance at these extremely high fuel concentrations was observed relative to lower fuel concentrations, in contrast to severe performance losses observed for conventional fuel cells involving an air cathode.  • Cathode flooding not an issue – Since there are no gas reactants at the DLRFC cathode and thus no TPB requirements, cathode flooding is no longer a concern. Indeed, the redox couple cathode is constantly flooded with the redox electrolyte.   186 Potential opportunities to strengthen this portion of the research include (a) identifying alternative membranes or different operating conditions that allow the conductivity of the membrane to be preserved in the presence of the Fe2+ and Fe3+ redox cations and (b) testing other redox couples that are electrochemically active on carbon but offer improved electrochemical activity, a smaller negative effect on the separator conductivity and/or a higher standard half-cell potential (e.g., the VO2+/VO2+ redox couple [1.0 V vs. SHE]).  5.1.2 Research Significance and Ultimate Impact The DLRFC configuration with a redox couple cathode represents a novel approach to mitigate a number of issues present with the conventional DMFC. Many researchers have directed their efforts towards mitigating the primary issues surrounding the DMFC utilizing various approaches. A standard approach to address the sluggish anode and cathode kinetics of the conventional DMFC is to apply high PGM catalyst loadings (i.e.,  greater than 1 mg/cm²) [14-19]. It is not uncommon to see anode PtRu catalyst loadings in the range of 2-8 mg/cm² and cathode Pt catalyst loadings in the range of 1-4 mg/cm². Such high PGM catalyst loadings significantly contribute to the overall system cost, hindering the commercial viability of the DMFC. Methanol crossover in the DMFC is typically addressed by using thicker membranes [7], reducing the methanol concentration [6, 20], engineering the anode [21, 22] or utilizing methanol-tolerant cathode catalysts [16, 23, 24]. Thick membranes cause increased ohmic losses while low methanol concentrations have a negative effect on anode kinetics. Methanol  187 tolerant cathode catalysts exhibit significantly lower activity towards oxygen reduction than Pt catalysts. The issue of cathode flooding is normally addressed by increasing the hydrophobicity of the cathode through the application of a PTFE sub-layer and/or supplying air at a high flow rate to facilitate water removal [25, 26]. The DLRFC configuration is unique to all of the above mentioned approaches targeted towards addressing core DMFC issues. The following points represent the significance and ultimate impact of DLRFC technology:  • Novel cell configuration – The DLRFC offers a number of advantages over the DMFC and represents a novel approach to address several core DMFC issues.  • Configuration independent – Any low molecular weight liquid fuel can be used to at the anode of the DLRFC. In addition, many redox couples are available for use in the DLRFC. This creates many options and creates potential for significant performance improvements for this relatively new architecture.  • Significant cell cost reduction – The high PGM metal loadings utilized in DMFCs account for a substantial portion of the DMFC cost [27]. Eliminating the PGM catalyst at the cathode, as is possible for the DLRFC, has the potential to reduce the total PGM cost by 30-50%, depending on the anode loading.  188  • Greater design flexibility – The DLRFC cathode exhibits reaction selectivity and does not require TPB sites as all reactants are transported in the liquid phase. This relaxes the typical cathode design constraints, allows for the use of a three-dimensional extended cathode and creates new possibilities for novel cell architectures.  5.2 Direct Liquid Redox Fuel Cell Performance Improvements 5.2.1 Summary One of the most significant differences between the conventional direct methanol fuel cell (DMFC) and the direct liquid redox fuel cell (DLRFC) is that the oxidant (i.e., redox couple) is not freely available in the DLRFC. The redox couple oxidant within the DLRFC is circulated in a closed loop through the cathode and has a fixed capacity to store and release charge. The capacity of the redox electrolyte is quantified through its volumetric charge density, which accounts for the concentration of the redox couple and the number of electrons per redox ion. Initial work on the DLRFC, summarized in Chapter 5.1, involved the use of a 0.9 M total Fe2+/Fe3+ ion concentration, which corresponds to a volumetric catholyte charge density of 24 Ah/L. If the methanol concentration is between 2 M and 16.7 M, the volumetric anolyte charge density would be in the range of 322-2685 Ah/L. Based on these values, the catholyte volume would need to be 13 to 111 times larger than the anolyte volume in order for the anolyte and catholyte to have the same capacity for charge transfer. Such  189 disproportionality in volume of electrolytes is a burden to the DLRFC system design and inhibits the commercial viability of the DLRFC. There is also room for performance improvements in the DLRFC as it is a non-optimized system. A number of potential avenues exist to enhance the DLRFC performance, such as increasing the membrane conductivity, adjusting the operating conditions, increasing the redox couple half-cell potential or improving the redox cathode polarization behaviour. In the work reported in Chapter 3, various approaches to improve the performance of the DLRFC are pursued. A major component of this work was the investigation of irons salts apart from those based on sulfate anions. Of all the iron salts considered, Fe(ClO4)3 demonstrated the most promising properties. Additional minor components of this work included cell temperature sensitivity tests, anolyte methanol concentration sensitivity tests and short-term fuel cell durability tests. The main research outcomes of the work discussed in Chapter 3 are the following:  • Significantly improved volumetric catholyte charge density – A 2.5 M Fe(ClO4)3 catholyte, which exhibited good storage stability (months), was successfully tested in a fuel cell. This concentration corresponds to a volumetric catholyte charge density of 67 Ah/L, which implies the catholyte should be 5-40 times larger in volume than the anolyte based on methanol concentrations in the range of 2-16.7 M. This is a significant improvement  190 relative to catholyte excess volume factors of 13-111 for a 0.9 M total Fe2+/Fe3+ catholyte with sulfate anions.  • Increased redox half-cell potential – It was observed through cyclic voltammetry experiments that the Fe2+/Fe3+ redox couple exhibited apparent half-cell potentials of 0.64 and 0.83 V vs. SHE for 1 M FeNH4(SO4)2 and Fe(ClO4)3 electrolytes, respectively, at 90°C. The increased half-cell potential for the perchlorate salt is attributed to ligand effects governing the activity of each iron ion.  • Enhanced electrochemical activity of the redox couple – Cyclic voltammetry experiments showed a 64% increase in the cathodic peak current density and a 54% reduction in peak potential separation over a glassy carbon electrode at 90°C for the 1 M Fe(ClO4)3 electrolyte relative to the 1 M FeNH4(SO4)2 electrolyte. DLRFC tests showed a 216% improvement in the peak power density (79 mW/cm² vs. 25 mW/cm²) for a 2.5 M Fe(ClO4)3 catholyte relative to that observed for a 1 M FeNH4(SO4)2 catholyte. This demonstrates that the Fe(ClO4)3 electrolyte has superior electrochemical activity and reversibility than the FeNH4(SO4)2 electrolyte.  • Increased knowledge of suitable DLRFC operating conditions – The cell temperature and anolyte methanol concentration sensitivity studies included temperatures between 50-90°C and methanol concentrations  191 between 2-24 M. The univariate experiments showed that a cell temperature of 90°C and methanol concentration 16.7 M (i.e., equimolar CH3OH/H2O) delivered the best cell performance.  • Demonstrated stable short-term fuel cell durability – The performance of the DLRFC at 90°C was shown to be relatively stable over a 4 hour period. This was demonstrated for DLRFC catholytes based on sulfate and perchlorate anions.  Aspects of this research requiring more work include testing a wide range of operating conditions for the DLRFC supplied with a Fe(ClO4)3 electrolyte, investigating system interactions more closely, investigating the reasons for the observed poor electrochemical activity/reversibility of the 1 M Fe(NO3)3 redox electrolyte and comparing the electrochemical performance of different salts for redox couples other than the Fe2+/Fe3+ redox couple. Longer DLRFC durability testing is also required.  5.2.2 Research Significance and Ultimate Impact Several improvements to the DLRFC performance have been realized. The approaches that were taken to accomplish this are similar to those found in the literature. Kazacos considered various vanadium salts containing different anions in order to improve the charge density of an electrolyte used in a vanadium chloride/polyhalide RFB [28]. Recognizing that the solubility of a metal  192 salt is dependent on the anion of the salt can be used to enhance the solubility of a redox couple. It is possible to adjust the half-cell potential of a redox couple by changing the ligands (or complexing agents) present in the system, as indicated by the electrochemical section of the CRC Handbook of Chemistry and Physics [29]. For instance, the half-cell potential of the Fe2+/Fe3+ redox can be shifted from 0.77 V vs. SHE to 0.12 V vs. SHE by the addition of ethylenediaminetetraacetic acid (HEDTA) [30]. This particular system (Fe2+/Fe3+/edta4-) has been electrochemically studied by a number of researchers as redox couple fuel (rather than an oxidant) [31, 32]. The approaches taken to advance DLRFC technology are standard methods which have shown considerable progress. The significance of the research discussed in Chapter 3 can be itemized as follows:  • Improved overall DLRFC performance – The potential commercial viability of the DLRFC technology has progressed substantially as a result of the advancements made with respect to catholyte charge density, redox couple half-cell potential and cathode polarization behaviour.  • Competitive non-optimized DLRFC performance – It is important to emphasize that DLRFC technology is still in its infancy. However, the performance is reaching levels competitive with developed conventional DMFCs. For example, a DMFC operating at 60°C supplied with 2 M CH3OH at the anode and air (101.3 kPa abs.) at the cathode delivered a  193 peak power density of 100 mW/cm² [33] whereas 79 mW/cm² at 90°C can be obtained with the DLRFC at this time.  5.3 The Mixed-Reactant DLRFC Architecture 5.3.1 Summary One of the most significant features of the direct liquid redox fuel cell (DLRFC) is the cathode selectivity. It eliminates the possibility for cathode depolarization by methanol crossover and allows for the use of high fuel concentrations at the anode.  The presence of cathode selectivity and methanol crossover also creates an opportunity to devise a mixed-reactant DLRFC (MR- DLRFC) architecture where the methanol is mixed with the redox catholyte and is supplied to the anode via fuel crossover, as shown in Figure 5.2. The methanol in the catholyte will not depolarize the cathode since the cathode is selective, as was demonstrated in Chapter 2. Since both of the anode reactants (CH3OH + H2O) are able to penetrate the Nafion® membrane, it is possible to supply the anode reactants from the catholyte to the anode through the Nafion® membrane. In this configuration, there would be no need for an inlet at the anode. Also, the anode flow field would be responsible for CO2 product removal only. The mixed-reactant architecture described above is fundamentally different from the mixed-reactant architectures discussed in Chapter 1.1. The MR-DLRFC requires selectivity at the cathode only and utilizes a unique method for fuel supply through the membrane. One of the most interesting properties of  194 the MR-DLRFC is that methanol crossover is desirable.  However, Fe2+/Fe3+ crossover is not desirable.   Figure 5.2. Schematic of a mixed-reactant direct liquid redox fuel cell (MR- DLRFC)  The following points summarize the research findings from the MR- DLRFC work discussed in Chapter 4:  • Successful demonstration of the MR-DLRFC concept – Fuel cell testing was performed for methanol concentrations of 1, 2 and 4 M in the mixed electrolyte at both 70 and 90°C. The peak power density of the MR-  195 DLRFC was shown to be 40% lower than that obtained with an equivalent non-mixed DLRFC (15 mW/cm² vs. 25 mW/cm²).  • Low effect of methanol concentration on fuel cell performance in the range of 1-4 M – The MR-DLRFC cathode performance showed little sensitivity towards the concentration of methanol in the catholyte (i.e., mixed-electrolyte), which is representative of the cathode selectivity. Also, the anode performance was not influenced by the methanol concentration in the catholyte over the tested range, indicating that rate of methanol crossover through the membrane is not a limiting factor yet.  • Improved MR-DLRFC performance at elevated cell temperatures – A substantial improvement in cell performance was observed when increasing the cell temperature from 70 to 90°C. A cell temperature of 90°C is near the maximum operating temperature possible for the MR- DLRFC since it operates at ambient pressure and boiling of the electrolyte must be avoided.  Potential opportunities to strengthen this research include: (1) Conducting an in-depth investigation around the mechanisms by which methanol affects the redox cathode (various works in the scientific literature suggest that methanol adsorption on the carbon-based cathode may significantly reduce the electrochemically active area at the cathode [34, 35]), (2) replacing the  196 membrane with a microporous separator or eliminating it completely to reduce the ohmic losses, and (3) utilizing a perchlorate-based iron redox salt in the mixed electrolyte to increase the performance of the cathode, according to the research reported in Chapter 3.  5.3.2 Research Significance and Ultimate Impact During the period where the MR-DLRFC technology was demonstrated and developed, no publications in the scientific literature or patent literature disclosing the concept of a MR-DLRFC architecture were identified. The concept of a MR-DLRFC appeared to be new and an application for the intellectual property was submitted.  This application was preceded by some interfering intellectual property by the ACAL Energy company in the UK by only a few months. However, the work presented in Chapter 4 represents the first publication in the scientific literature for the novel MR-DLRFC architecture. The significance of this research is outlined below:  • Demonstration of a MR-DLRFC where methanol crossover is desirable – The MR-DLRFC concept redefines the meaning of fuel crossover. This approach is fundamentally unique to other mixed-reactant architectures and opens the avenue for a new family of fuel cell types.  • Simpler manifolding and system control – The mixed-reactant architecture greatly reduces the complexity of the DLRFC. With only a  197 single electrolyte present, reactant manifolding becomes trivial and the number of auxiliary temperature, flow and concentration control units diminishes substantially. These properties have favourable implications for the cost of the MR-DLRFC.  • Opportunity to significantly reduce cell thickness – The absence of an anolyte in the MR-DLRFC significantly simplifies the flow field requirements at the anode. Since only CO2 removal is required for the MR-DLRFC anodic flow field, something as simple as a wire mesh could be used to accommodate CO2 removal. A wire mesh could be much thinner than the currently employed flow field, creating an opportunity to significantly reduce the thickness of the anode compartment. For example, if the thickness of the flow field plates, catalyst support layers and membrane in a fuel cell stack are 1.5, 0.2 and 0.05 mm, respectively, a 75% reduction in the thickness of the anode flow field plate will result in a 33% reduction in the cell thickness.  5.4 Regeneration of the DLRFC Redox Couple 5.4.1 Summary An important aspect of DLRFC operation is regeneration of the redox couple after the catholyte is partially discharged. Regeneration of the DLRFC redox couple can be achieved by oxidizing it with air/O2 in an external electrochemical regeneration cell, as shown in Figure 5.3.  198  Figure 5.3. Schematic of DLRFC redox couple regeneration using an external electrochemical regeneration cell.  However, the above approach requires a separate regeneration cell, which will significantly add to the total weight and volume of the DLRFC system. Due to the unique hybrid architecture of the DLRFC involving elements of both the RFB and DMFC, an opportunity exists to regenerate the redox couple in-situ without the use of an external electrochemical regeneration cell. By substituting the methanol anolyte with an air stream, the PtRu catalyst at the DLRFC anode becomes the regeneration cathode and is used for O2 reduction. This substitution effectively reverses the flow of electrons and represents an in-situ regeneration approach for the DLRFC, as shown in Figure 5.4.   199  Figure 5.4. Schematic of redox couple regeneration for a DLRFC using the novel in-situ regeneration approach.  In Chapter 4, results for both single-pass and batch/recycle DLRFC redox couple regeneration tests are reported. The research findings are outlined below:  • Successful demonstration of the novel in-situ DLRFC regeneration concept – In-situ DLRFC regeneration of a 0% SOC redox electrolyte was characterized at 70°C and 90°C using a single-pass configuration. Nearly identical performance was observed for both temperatures using air as the oxidant.  • Characterized in-situ DLRFC regeneration for a batch/recycle configuration – 100 mL of a 0.9 M FeSO4 / 1 M H2SO4 (0% SOC) redox electrolyte was recycled through the 4 cm² fuel cell at 70°C. It was observed that a maximum SOC of 85% was possible for this configuration. After 40 hr of regeneration, a SOC of 75% is attainable. Advancements in the regeneration performance are required to increase the space-time yield and improve the viability of this regeneration approach.  200  • Established baseline performance for in-situ DLRFC redox couple regeneration – Using the current non-optimized PtRu catalyst and a 0% SOC redox electrolyte, a maximum regeneration current density of 20 mA/cm² was observed.  Potential opportunities to strengthen this research include: (1) Engineering the PtRu electrode to exhibit higher activity towards oxygen reduction. This may involve adjusting the hydrophobicity of the electrode structure and/or modifying the catalyst composition. (2) Characterize DLRFC regeneration using a higher concentration redox electrolyte prepared with a perchlorate-based iron salt.  5.4.2 Research Significance and Ultimate Impact At this time, no publications describing an in-situ redox couple regeneration approach have been identified in the scientific literature. The concept of in-situ redox couple regeneration is novel and an application for intellectual property has been filed. The significance of the research presented in Chapter 4 is outlined below:  • Novel in-situ DLRFC regeneration approach successfully demonstrated – This regeneration architecture represents a novel approach to redox couple regeneration. It is not limited to DLRFCs and may be useful for other electrochemical applications.  201 5.5 Potential Applications of DLRFC Technologies The DLRFC and MR-DLRFC technologies are presently at an early stage. As a result, the applications accessible for these technologies is a strong function of future developments.  At this time, targeted applications for the DLRFC and MR-DLRFC technologies include portable electronic device power (several W) and remote power generation (several kW).  5.5.1 Portable Electronic Device Power Power sources for cameras, cell phones and laptops are bound by stringent demands for compactness. It would be possible for the DLRFC or MR- DLRFC systems to meet this constraint through the use of single use recyclable fuel cartridges. In the DLRFC the reactant cartridge would need two chambers for the methanol/water anolyte and the redox catholyte whereas the MR-DLRFC would require just one compartment for the mixed-electrolyte. Upon complete discharge of the system, the reactant cartridge would need to be returned to the retailer and subsequently recycled at a central regeneration facility where depleted cartridges could be replenished and returned to the market. The MR-DLRFC may be more suitable than the DLRFC for this particular application due to the simplicity of a controlling and distributing a single mixed electrolyte. Fewer auxiliary units are required to control the reactant temperature, flow and concentration. A quick capacity comparison can be made for a typical Li-ion camera battery and a MR-DLRFC system. A Li-ion battery with dimensions of 0.75 x 3.5 x 4 cm (10.5 cm³) will contain approximately 750 mAh of charge. A  202 10.5 cm³ mixed-reactant electrolyte with 2.5 M Fe2+/Fe3+ as the limiting reactant will contain approximately 700 mAh of charge. It is apparent that the MR-DLRFC technology must undergo significant further development in order to compete with the charge capacity of Li-ion battery technology as the latter figure for the MR-DLRFC does not account for the volume of the cell and auxiliary units.  5.5.2 Remote Power Generation This market sector is currently dominated by diesel generators and is not bound by the stringent volume constraints present in small electronic devices. If the DLRFC or MR-DLRFC technologies are targeted towards this application, it would make sense to incorporate the redox regeneration component into the system to achieve a stand-alone unit requiring only refuelling of methanol and water. The redox couple can be regenerated electrochemically or chemically using air as the oxidant. At this time it is difficult to gauge the cost and size of a complete DLRFC or MR-DLRFC with a redox regeneration unit. However, some rough figures can be given for the stack of a DLRFC operating at the highest peak power density documented in this thesis (80 mW/cm²). Assuming a 25 cell stack where each individual cell is 0.4 cm thick, the area and volume of a 1 kW DLRFC stack (cells only) would be 500 cm² and 5000 cm³, respectively. As the DLRFC and MR-DLRFC technologies are currently non-optimized systems, future improvements in the redox electrolyte charge density, cell design and cell performance are expected to significantly advance the commercial viability of these technologies.  203 5.5.3 Scale-Up Considerations Increasing the scale of an electrochemical cell involves the consideration of non-linear variables and effects that may have not been important at the lab- scale. Such variables include the pressure drop, parasitic load, reactant conversion per pass, reactant distribution, continuity of electronic and ionic conductivity, cell temperature uniformity, electrolyte heat management and start up/shut down logistics. The pressure drop across an electrochemical cell/stack depends on the reactant flow rate, channel geometry, electrode porosity (for 3-D electrodes) and solution viscosity. For large-scale systems, large pumps are required to maintain the desired flow rate and parasitic losses due to pumping must be considered. Commercial scale testing of a VRB showed that a 1.5 M VO2(SO4)2 / 2.6 M H2SO4 electrolyte pumped a flow rate of 6 L/min through a 10 cell stack containing carbon felt (0.3 mm thick, 0.15 m² area) yields a pressure drop of 80 kPa [36]. This energy required to pump the electrolytes in this case was reported to be 2-3% of the electrical energy produced by the VRB. The flow channels of commercial-scale electrochemical cells need to be carefully designed to avoid reactant distribution issues. The inefficient use of the cell electrode area will lead to a highly non-uniform current and potential distribution across the area of the electrode and ultimately significant performance losses. A related subject is the reactant conversion per pass through the electrochemical cell as it can lead to non-uniform reactant distributions. The reactant conversion (X) is defined by the following equation:  204  jsjj j 1 VnFC Is X ,λ == &  (5.1) Where Cj is the concentration of reactant j in mol/m3, Vj (dot) is the flow rate of reactant j in m3/s, sj is the stoichiometric coefficient of reactant j, I is the current in A and λs,j is the stoichiometric coefficient of reactant j. As implied by equation 5.1, if a low stoichiometric factor is selected for fuel cell operation (e.g., 2), a significant fraction of the reactant will be consumed each pass and a significantly non-uniform reactant distribution will result. Heat management and maintaining cell temperature uniformity present a challenge in large scale electrochemical cells. If portions of an electrochemical cell are significantly offset from the set point temperature, reduced performance, side reactions or even damage to the cell can result. It may be necessary to segment the electrochemical cell/stack into independent thermal control loops in some cases. For large scale electrochemical cells operating at elevated temperatures, reactant loops would need to be well insulated to reduce heat loss and a heater may be required in the reactant storage tank for system start-up.  5.6 Future Work and Recommendations The original goal of this work was to demonstrate and develop a methanol-fed fuel cell which could take advantage of the high energy density of the methanol fuel but be relieved of many challenges present in conventional direct methanol fuel cell (DMFC) technology. The concept of a direct liquid redox fuel cell (DLRFC) offers numerous advantages over the conventional DMFC,  205 which have been demonstrated and discussed in previous chapters. However, the DLRFC presents new challenges related to the redox couple and is currently a non-optimized system with many areas available for improvement. The aim of this recommendation section is to shed light on opportunities available to future researchers to further advance DLRFC technology.  5.6.1 DLRFC Membrane/Separator Studies In Chapter 2, it was shown that the conductivity of the Nafion® membrane in the DLRFC is reduced by 80-90% after it is exposed to the Fe2+/Fe3+ redox electrolyte. The Fe2+/Fe3+ cations are able to penetrate the Nafion® cation exchange membrane and displace the highly mobile protons which reside on the SO3- sites within the Nafion® membrane. Consequently, there is an opportunity to improve the performance of the DLRFC by increasing the membrane/separator conductivity. Some potential strategies to address this issue are given below:  • Investigate alternative membranes – It is possible that membranes other than Nafion® are less susceptible to contamination by the Fe2+/Fe3+ redox couple. A variety of membranes suitable for use in DMFCs [37] can also be studied ex-situ in the presence of the Fe2+/Fe3+ redox couple to determine their potential as a membrane for the DLRFC.  • Add a thin film of anionic membrane on to the cathode side of the membrane – If the penetration of Fe2+/Fe3+ into the cation exchange  206 membrane is prevented, the conductivity of the membrane would not degrade. A thin film of anionic membrane (applied as an ionomer resin or sandwiched thin film) containing positively charged immobilized ions would provide a strong electrostatic repulsion against the multivalent Fe2+/Fe3+ cations and potentially inhibit or prevent membrane contamination.  • Replace the membrane with a microporous separator – At the cost of higher methanol and redox couple crossover, a porous separator may be employed to utilize the high conductivity of the electrolyte and eliminate the issue of membrane contamination (i.e., no ionic sites present for the redox couple to occupy).  In the case where pure or nearly pure methanol is supplied as the DLRFC anolyte, water crossover from the cathode to the anode acts a significant water source for the anode reaction. Thus, it would be relevant to investigate methods to increase the rate of water crossover through the membrane/separator under these conditions.  5.6.2 Methanol Adsorption on Carbon In the main body chapters, it was hypothesized that methanol adsorption on carbon is likely responsible for the observed reduced electrochemical performance of the redox couple on a carbon electrode in the presence of  207 methanol. This hypothesis stemmed from the work of other researchers who utilized the adsorption properties of methanol on carbon for refrigeration [34] and nano-tube growth control [35]. It would be beneficial to investigate this phenomenon on a variety of carbon substrates (i.e., CFP, GF and CC) to confirm its occurrence and, if necessary, identify strategies to reduce its impact. One approach is to study the adsorption of methanol on carbon and its effect on the Fe2+/Fe3+ redox couple using cyclic voltammetry and other electrochemical techniques. Another approach would be to use temperature programmed desorption methods.  5.6.3 Opportunities for the MR-DLRFC In Chapter 3, the solubility of the redox couple was increased by a factor of 2.5 using a perchlorate-based iron salt and the DLRFC cathode performance was substantially improved. In Chapter 4, a novel mixed-reactant architecture was demonstrated for the DLRFC. Further experimental work needs be conducted to combine these research outcomes. It is likely that the cathode performance enhancements reported in Chapter 3 can be extended to the mixed- reactant DLRFC (MR-DLRFC) through the use of a perchlorate-based iron salt. Fuel cell testing utilizing a reference electrode to observe individual electrode polarization behaviour would be a suitable experimental approach. A separate recommendation is to investigate how uptake of the redox couple in the membrane affects the methanol crossover characteristics. Since  208 fuel crossover serves as the mode of fuel supply in the MR-DLRFC, this is an area of interest.  5.6.4 Alternative Redox Couples for the DLRFC The Fe2+/Fe3+ redox couple was originally employed for use in the DLRFC based on the electrochemical data available on this redox couple in the literature and the existence of a preceding project based on a hybrid hydrogen redox fuel cell utilizing the Fe2+/Fe3+ redox couple [13, 38]. A significant portion of time has been dedicated to the identification of other redox couples suitable for employment in the DLRFC. The VO2+/VO2+ redox couple (E0= 1.0 V vs. SHE at 25°C) is of particular interest due to its higher theoretical half-cell potential relative to the Fe2+/Fe3+ redox couple and its demonstrated viability as a suitable redox couple for redox flow battery applications [36, 39-43]. Some initial work on the DLRFC focused on the VO2+/VO2+ redox couple. However, issues surrounding electrolyte preparation and electrolyte stability obstructed the use of this redox couple for initial DLRFC demonstration tests. Future researchers working on DLRFC technology may wish to revisit the VO2+/VO2+ redox couple. Kazacos et al. prepare the VO2+/VO2+ redox electrolyte by electrochemically oxidizing the VO2+ from VO(SO4)2 to VO2+ in an electrochemical cell [44] and by electrochemically dissolving V2O5 in an acidic electrolyte [45]. Also, some success in stabilizing the VO2+/VO2+ redox electrolyte has been demonstrated through the use of temperature control and precipitation inhibitors [46, 47].  209 5.6.5 Redox Couple Regeneration Some preliminary work on redox couple regeneration involving a novel in- situ configuration for the DLRFC was reported in Chapter 4. This regeneration approach provides a convenient solution to redox couple regeneration in the DLRFC without introducing additional auxiliary hardware. However, the observed in-situ regeneration rate is non-optimized and there is much room for improvement. Potential avenues for increasing the in-situ regeneration rate may include:  • Engineering the PtRu electrode, which is used for both methanol oxidation (during discharge) and oxygen reduction (during regeneration), to be more active towards oxygen reduction. This may involve optimizing the hydrophobicity of the electrode for efficient water removal or adjusting the composition or loading of the catalyst.  • Operating the in-situ DLRFC regeneration unit at elevated temperatures (above 100°C) and elevated pressures (above 2 bar abs.). Such operating conditions will improve the oxygen reduction reaction kinetics and the partial pressure of oxygen in the cathode air supply.  • Conventional approaches to redox couple regeneration such as chemical or electrochemical regeneration in an auxiliary unit can also be investigated. Rönnholm et al. performed homogenous chemical oxidation  210 of Fe2+ ions by pure O2 in a stirred-tank semi-batch reactor with finely dispersed particles of activated carbon [48, 49]. However, high temperatures (60-130°C) and high pressures (4-10 bar) were required to achieve practical reaction rates. Future work on ex-situ chemical or electrochemical redox couple regeneration for the DLRFC could involve the design and construction of an auxiliary regeneration unit operating at elevated temperatures and pressures.  • Expanding on the last point, it may be possible to regenerate the redox couple in the fuel cell during discharge by injecting air or oxygen into the catholyte stream. Further experimentation should be performed to confirm the viability of this approach.  5.6.6 Long-term Durability Testing To date, short-term 4 hr DLRFC durability tests using 100% SOC electrolytes have been performed with positive results. Longer testing periods (e.g., 24 hrs or 1 week) are needed to better understand the durability of the DLRFC. Full charge/discharge cycling of the redox couple should also be performed once the redox regeneration approach is more developed. Eventually, long-term durability testing involving many (i.e., hundreds) of charge/discharge cycles should be performed. Long-term durability tests are also required for the MR-DLRFC.   211 5.6.7 System Testing and Stack Development The research performed for this thesis serves as an initial demonstration and characterization of the DLRFC, MR-DLRFC and an in-situ regeneration approach. Once these technologies have undergone sufficient testing such that the system components, design and operating conditions deliver commercially competitive performance, detailed system testing and stack development should be performed as well as an analysis of scale-up considerations. This could involve measurement of the overall system energy efficiency, the parasitic load, the gravimetric system peak power density, the volumetric peak power density and the durability of the stack. Cost estimates would also be relevant at this stage.  212 5.7 References 1. S. Hikita, K. Yamane, Y. Nakajima, JSAE Review, 22, 2 (2001) 151-156. 2. E. Kjeang, J. Goldak, M. R. Golriz, J. Gu, D. James, Journal of Power Sources, 153, 1 (2006) 89-99. 3. A. Lam, D. P. Wilkinson, Z. Jiujun, in Proton Exchange Membrane Fuel Cells, T. Fuller, C. Bock, C. Lamy, Eds., PV 1, pp. 273-281, The Electrochemical Society Proceedings Series, Pennington, NJ, (2005). 4. J. H. Yang, Y. C. Bae, J. Electrochem. Soc., 155, 2 (2008) 194-199. 5. P. Mukoma, B. R. Jooste, H. C. M. Vosloo, Journal of Membrane Science, 243, 1-2 (2004) 293-299. 6. J. B. Ge, H. T. Liu, J. Power Sources, 142, 1-2 (2005) 56-69. 7. P. Dimitrova, K. A. Friedrich, B. Vogt, U. Stimming, J. Electroanal. Chem., 532, 1-2 (2002) 75-83. 8. X. Ren, S. Gottesfeld, J. Electrochem. Soc., 148, 1 (2001) A87-A93. 9. K. Scott, W. M. Taama, S. Kramer, P. Argyropoulos, K. Sundmacher, Electrochimica Acta, 45, 6 (1999) 945-957. 10. H. Liu, C. Song, L. Zhang, H. Wang, D. P. Wilkinson, Journal of Power Sources, 155, 2 (2006) 95-110. 11. A. Casalegno, R. Marchesi, Journal of Power Sources, 175, 1 (2008) 372- 382. 12. H. A. Gasteiger, S. S. Kocha, B. Sompalli, F. T. Wagner, Applied Catalysis B: Environmental, 56, 1-2 SPEC ISS (2005) 9-35.  213 13. K. Fatih, D. P. Wilkinson, F. Moraw, F. Girard, in Electrocatalysis, R. Adzic, V. Birss, G. M. Brisard, A. Wieckowski, Eds., PV 11, pp. 341-350, The Electrochemical Society Proceedings Series, Pennington, NJ, (2005). 14. K. V. Kordesch, J. Electrochem. Soc., 125, 3 (1978) 77C-88C. 15. S. Surampudi, S. R. Narayanan, E. Vamos, H. Frank, G. Halpert, J. Power Sources, 47, 3 (1994) 377-385. 16. A. K. Shukla, R. K. Raman, Ann. Rev. Mater. Research, 33, (2003) 155-168. 17. T. V. Reshetenko, H. T. Kim, U. Krewer, H. J. Kweon, Fuel Cells, 7, 3 (2007) 238-245. 18. V. G. Sister, V. N. Fateev, D. A. Bokach, Russ. J. Electrochem., 43, 9 (2007) 1097-1100. 19. Y. C. Park, D. H. Peck, D. H. Jung, S. K. Kim, S. Y. Lim, in , PV 16, pp. 801- 805, ECS Transactions, Honolulu, HI, United states, (2008). 20. J. G. Liu, T. S. Zhao, R. Chen, C. W. Wong, Electrochem. Commun., 7, 3 (2005) 288-294. 21. F. Q. Liu, G. Q. Lu, C. Y. Wang, J. Electrochem. Soc., 153, 3 (2006) A543- A553. 22. D. P. Wilkinson, A. Lam, in , PV MA 2005-02, p. 1827, Meeting Abstracts, Pennington, NJ, (2005). 23. S. Lj. Gojkovic, S. Gupta, R. F. Savinell, J. Electroanal. Chem., 462, 1 (1999) 63-72. 24. K. Scott, A. K. Shukla, C. L. Jackson, W. R. A. Meuleman, Journal of Power Sources, 126, 1-2 (2004) 67-75.  214 25. A. Oedegaard, C. Hebling, A. Schmitz, S. Moller-Holst, R. Tunold, J. Power Sources, 127, 1-2 (2004) 187-196. 26. A. Lindermeir, G. Rosenthal, U. Kunz, U. Hoffmann, J. Power Sources, 129, 2 (2004) 180-187. 27. R. Dillon, S. Srinivasan, A. S. Arico, V. Antonucci, J. Power Sources, 127, 1-2 (2004) 112-126. 28. M. Kazacos, Journal of Power Sources, 124, 1 (2003) 299-302. 29. D. R. Lide, CRC Handbook of Chemistry and Physics, CRC Press LLC, Boca Raton (2005). 30. J. G. Speigt, Lange's Handbook of Chemistry, 16th Ed., McGraw-Hill, New York (2005). 31. E. Juzeliunas, K. Juettner, Electrochimica Acta, 43, 12-13 (1998) 1691-1696. 32. H. J. Wubs, A. A. C. M. Beenackers, Industrial & Engineering Chemistry Research, 32, 11 (1993) 2580-2594. 33. G. Q. Lu, C. Y. Wang, Journal of Power Sources, 144, 1 (2005) 141-145. 34. I. I. El-Sharkawy, M. Hassan, B. B. Saha, S. Koyama, M. M. Nasr, Int. J. Refrig., 32, 7 (2009) 1579-1586. 35. Z. P. Wu, J. N. Wang, J. Ma, Carbon, 47, 1 (2009) 324-327. 36. M. Kazacos, D. Kasherman, D. R. Hong, M. Kazacos, J. Power Sources, 35, 4 (1991) 399-404. 37. V. Neburchilov, J. Martin, H. Wang, J. Zhang, J. Power Sources, 169, 2 (2007) 221-238.  215 38. K. Fatih, D. P. Wilkinson, F. Moraw, A. Ilicic, F. Girard, Electrochem. Solid- State Lett., 11, 2 (2008) B11-B15. 39. E. Sum, M. Skyllas-Kazacos, J. Power Sources, 15, 2-3 (1985) 179-190. 40. M. Kazacos, M. Rychcik, R. G. Robins, A. G. Fane, M. A. Green, J. Electrochem. Soc., 133, 5 (1986) 1057-1058. 41. M. Skyllas-Kazacos, F. Grossmith, Journal of the Electrochemical Society, 134, 12 (1987) 2950-2953. 42. M. Rychcik, M. Skyllas-Kazacos, Journal of Power Sources, 22, 1 (1988) 59- 67. 43. C. Menictas, D. R. Hong, Z. H. Yan, Y. J. Wilson, M. Kazacos, in , PV 1, pp. 299-303, National Conference Publication - Institution of Engineers, Australia, Sydney, USA, (1994). 44. S. Zhong, M. Skyllas-Kazacos, J. Power Sources, 39, 1 (1992) 1-9. 45. E. Sum, M. Rychcik, M. Skyllas-Kazacos, J. Power Sources, 16, 2 (1985) 85- 95. 46. M. Kazacos, M. Cheng, M. Skyllaskazacos, J. App. Electrochem., 20, 3 (1990) 463-467. 47. M. Skyllas-Kazacos, C. Peng, M. Cheng, Electrochem. Solid-State Lett., 2, 3 (1999) 121-122. 48. M. R. Ronnholm, J. Warna, T. Salmi, I. Turunen, M. Luoma, Chemical Engineering Science, 54, 19 (1999) 4223-4232. 49. M. R. Ronnholm, J. Warna, D. Valtakari, T. Salmi, E. Laine, in , PV 66, pp. 447-452, Catalysis Today, Napoles, Italy, (2001).  216 6. APPENDICES Appendix A – Publications and Presentations Publications • A. B. Ilicic, D. P. Wilkinson, K. Fatih, “Advancing Direct Liquid Redox Fuel Cells: Mixed Reactant and In-situ Regeneration Opportunities” J. Electrochem. Soc., 157, 4 (2010) (accepted). • D. P. Wilkinson, K. Fatih, F. Moraw, A. Ilicic, F. Girard, “Advancements in the Direct Hydrogen Redox Fuel Cell” Electrochem. Solid-State Lett., 11, 2 (2008) B11-B15. • A. B. Ilicic, D. P. Wilkinson, K. Fatih, F.Girard, “High Fuel Concentration Direct Liquid Fuel Cell with Redox Couple Cathode” J. Electrochem. Soc., 155, 12 (2008) B1322-B1327. • A. B. Ilicic, D. P. Wilkinson, K. Fatih, F. Girard, “High Fuel Concentration Direct Liquid Fuel Cell with Redox Couple Cathode” ECS Transactions, PV 16, 2 (2008) 1549-1560. • A. B. Ilicic, M. S. Dara, D. P. Wilkinson, K. Fatih, “Improved Performance of the Direct Methanol Redox Fuel Cell” (submitted). • A. B. Ilicic, V. Gogel, L. Joerissen, D. P. Wilkinson, “Effect of Electrode Structure and Anions on CO2 Tolerant Alkaline DMFCs” (in preparation).   217 Presentations1 • A. B. Ilicic, D. P. Wilkinson, W. Merida, K. Fatih; “The Future of Redox Systems in the Hydrogen Economy”; 6th Symposium for New Materials for Electrochemical Systems, Montreal QC, July 9-12 2006. • A. B. Ilicic, D. P. Wilkinson, K. Fatih, F. Girard, “High Fuel Concentration DMFC with Redox Couple Cathode”; 58th Meeting of the International Society of Electrochemistry, Banff AB, Sept. 9-14, 2007. • A. B. Ilicic, D. P. Wilkinson, K. Fatih, F. Girard, “High Fuel Concentration Direct Liquid Fuel Cell with Redox Couple Cathode”; 214th Meeting of the Electrochemical Society, Honolulu HI, Oct. 12-17, 2008. • D. P. Wilkinson, A. B. Ilicic, K. Fatih, “Advancements in Hydrogen and Liquid Fuel Redox Fuel Cells”; 216th Meeting of the Electrochemical Society, Vienna AT, Oct 4-9, 2009.         1Presenter underlined  218 Appendix B – Experimental Procedures B.1 Glass Cleaning Prior to use, glass cells, beakers and volumetric flasks should be cleaned with a 1:1 vol/vol concentrated H2SO4 / concentration HNO3 solution to remove residuals. A small amount of the concentrated cleaning solution (10-20 mL) should be swirled in the glassware ensuring the entire surface is coated. This must be done in a fumehood to avoid the corrosive HNO3 fumes. Clean glassware should be triply rinsed with deionized water and used immediately.  B.2 Redox Electrolyte Preparation The Fe2+/Fe3+ redox electrolyte has been prepared with both Fe(ClO3)3 and FeNH4(SO4)2 / FeSO4 salts. For the sulfate system, the dissolution time is long. After weighing the desired mass of salt into a glass beaker of appropriate volume, add 90% of the total water required in addition to a stir stick and thermometer. Place the beaker on a hot/stir plate and turn on both the heat and stirring power. The dissolution of the sulfate-based salts is significantly accelerated at 40-50°C. Allow the prepared redox electrolyte to stand overnight for the solution equilibria to be established (i.e. hydrolysis).   219 B.3 Cyclic Voltammetry A photograph of the cyclic voltammetry setup used for the work reported in this thesis is shown in Figure 6.1. Cyclic voltammetry measurements are extremely sensitive to impurities. Caution must be exercised to reduce contamination from the glass cell walls, the electrolyte and all of the electrodes.   Figure 6.1. Photograph of a three-electrode cell used for cyclic voltammetry.  Prior to setting up a cyclic voltammetry experiment, ensure that the electrolyte is prepared and the data acquisition software/hardware is functional  220 and properly configured. To set up a Solartron 1470E Multistat using Corrware software, following these instructions: 1. Turn on the Solartron 1470E and load the Corrware software. 2. Press “New Instrument” and a dialogue similar to Figure 6.2 will pop up. 3. Select an instrument that is currently not in use and press modify instrument.   Figure 6.2. Selecting a new instrument in Corrware.  4. Configure the instrument as shown in Figure 6.3. These parameters control the conventions used by the multistat. Press OK.   221  Figure 6.3. Setting the instrument convention in Corrware.   222 5. Ensure that the floating window which maps virtual instruments to hardware channels is set correctly. Figure 6.4 shows an example of how one would map (virtual) Instrument #1 to hardware channel #1.   Figure 6.4. The virtual instrument mapper in Corrware software.  6. Press “Insert New Experiment”. A dialogue similar to Figure 6.5 will appear. Select “Cyclic Voltammetry” and press OK.  223  Figure 6.5. Inserting a new experiment in Corrware.  7. A new dialogue similar to Figure 6.6 will appear. Choose a file name and location for the data file at the top of the window. Select 0 V vs. open circuit for the start and end points. Input the appropriate vertex potentials vs. the reference. Press OK. 8. You will now see the experiment listed and ready for execution. Once the cell is ready for testing, select the experiment and press “Measure Selected”.   224  Figure 6.6. Configuring a cyclic voltammetry experiment in Corrware.  To clean a Pt working electrode or counter electrode, immerse it in a fresh solution of 1:1 vol/vol concentrated H2SO4 and 30% H2O2 for 5 minutes. Oxygen evolution should occur evenly over the electrode surface. To clean a glassy carbon electrode, rinse it with methanol and polish it on soft fibrous pad with 0.05 µm alumina for 5-10 minutes. Rinse the polished electrode with deionized water and then sonicate the electrode for 10 minutes in deionized water to remove and residual alumina. For cyclic voltammetry experiments performed at elevated temperatures, insert a thermometer in the glass cell and immerse the cell in a constant  225 temperature bath at the desired temperature. If the solution must be dearated with N2, do this prior to heating the cell to minimize solvent evaporation.  B.4 Cyclic Voltammetry: IR correction To remove ohmic effects from the measured voltammogram, one must measure the cell impedance and compute the ohmic losses. The cell impedance can be measured before or after acquiring the voltammogram. Using a 1260 Solartron frequency response analyzer (FRA) connected to a 1470E Multistat, one may use the following instructions to measure the cell impedance and perform IR correction:  1. Power on the 1260 Solartron FRA and load the Zplot software. 2. Run the Zplot setup. Configure the setup as shown in Figure 6.7 so that it is configured to operate through the Solartron 1470E Multistat. 3. Set up an experiment under the “Ctrl E: Sweep Freq” tab. Figure 6.8 gives an exemplary set of parameters that can be used in this window. 4. Press “Measure Sweep” to perform the impedance measurement. 5. An impedance response with at least a portion of a semicircle should result. An example is shown in Figure 6.9. 6. The cell impedance is taken as the x-axis intercept after extrapolating the left side of the semi-circle. 7. Using the obtained cell impedance value, the voltage points of the cyclic voltammogram can be IR-corrected by applying the following formula:  226  celliAREE +=′  (6.1) Where E’ is the IR-corrected potential in V, E is the measured potential in V, i is the measured current density in A/cm² and Rcell is the measured cell impedance in Ω.   Figure 6.7. Configuring the Zplot software.  227  Figure 6.8. Configuring the Zplot software for cyclic voltammetry impedance measurements.  Nyquist Plot 0 50 100 150 200 250 300 0 200 400 Real Impedance (ohms) Im ag in ar y Im pe da nc e  (o hm s) Rs Rp Im ag in ar y Im pe da nc e  (o hm s)  Figure 6.9. Example of an electrochemical impedance response.   228 B.5 Nafion® Membrane Preparation New Nafion® membranes must be hydrated, cleaned and protonated according to the following procedure:  1. Soak membrane sample in Millipore water for approx. 2 hours 2. Boil membrane sample in 3 vol% hydrogen peroxide for 30 minutes 3. Rinse membrane sample with Millipore water 4. Boil in Millipore water for 30 minutes 5. Rinse membrane sample with Millipore water 6. Boil in 0.5M sulfuric acid for 30 minutes 7. Rinse membrane sample with Millipore water 8. Inspect membrane to ensure clarity 9. If membrane is clear, store in Millipore water 10. If membrane is not clear: a. Boil in stronger peroxide, 10 vol% b. Rinse with Millipore water c. Boil in 2M nitric acid d. Repeat steps 7-10    229 B.6 Membrane Conductivity Measurements The conductivity of a membrane is normally measured by mounting a membrane sample in a membrane conductivity cell and using a frequency response analyzer to measure the impedance. An example of a membrane conductivity cell is shown in Figure 6.10. The cell is shown here disassembled but must be closed and screwed shut after mounting a membrane sample to maintain good contact between the Pt electrodes and the membrane.  L Pt  electrodes  Figure 6.10. Photograph of a conductivity cell.  The conductivity of a membrane can be obtained with a Solartron 1470E Multistat, 1260 FRA using the following instructions:  1. Power on the 1260 Solartron FRA and load the Zplot software. 2. Run the Zplot setup and configure it as shown in Figure 6.7 so that it operates through the Solartron 1470E Multistat.  230 3. Set up an experiment under the “Ctrl E: Sweep Freq” tab. Figure 6.11 gives an exemplary set of parameters that can be used in this window. Notice that the frequency range is different than the cell impedance measurement configuration used for cyclic voltammetry.   Figure 6.11. Configuring the Zplot software for membrane conductivity measurements.  4. Press “Measure Sweep” to perform the impedance measurement. 5. An impedance response with at least a portion of a semicircle should result. An example is shown in Figure 6.9. 6. The membrane impedance is taken as the diameter of the semicircle. The membrane conductivity can be computed using the following equation:  231  AR L mem =σ  (6.2) Where σ is the membrane conductivity in S/m, L is the inter-electrode gap in m, Rmem is the membrane impedance in Ω and A is the cross-sectional area of the membrane in m².  B.7 Catalyst Spraying for Fuel Cell Electrodes Fuel cell electrodes can be fabricated by spraying a catalyst layer on to carbon fiber paper (e.g., Toray TGP-H-060). After proper training, the following instructions can be used to prepare a fuel cell electrode: 1. Choose an appropriate catalyst loading (e.g., 2 mg/cm2). 2. Calculate the mass of Pt/C powder mixture that should be used:  Notes for calculation -Catalyst loading symbolized by σCatalyst (mg per geometric cm²; actual can be calculated later) -The catalyst is purchased as a dispersed deposit on Vulcan XC carbon powder. Some common compositions of this powder based on mass are given: (20% Pt, 80% C; etek-inc.com ~$1500/100g), (40% Pt, 60% C), (20% Pt, 20% Ru, 60% C) and (10% Pt, 10% Ru, 80% C). The following example uses the last composition. Let fCatalyst denote the mass fraction of catalyst on the carbon -To account for losses (i.e. off-edge spraying, air-borne spray) while spraying the ink, the premixed Pt/C is added with an excess factor of 3 for hand-spraying or 2 for the automatic sprayer, which may vary with preference, technique or procedure. -The active area for the automatic sprayer is 11cm x 11cm, which equates to 121 cm2.  Catalyst GeomCatalyst VulcanXCCatalyst f A m σ2=+  (6.3)  ( ) ( ) ( ) gmg g cm cm mg m VulcanXCCatalyst 42.21000 1 2.0 12122 22 =⎟⎟⎠ ⎞ ⎜⎜⎝ ⎛⎟⎠ ⎞⎜⎝ ⎛ =+  (6.4)   232 3. Choose an appropriate Nafion® loading (~30% w/w is common). Sprayed Nafion® provides bonding when pressing the MEA and increases the TPB length. 4. Calculate the mass of Nafion® solution required: Notes for calculation -Nafion® loading is symbolized as fNafion (fraction Nafion in catalyst + Vulcan XC + Nafion) -Nafion® is available as a 5% w/w solution. Mass of solution expressed as mNaf_Sol  ⎟⎠ ⎞⎜⎝ ⎛ −= + 05.0 1 1_ Nafion VulcanXCCatalystNafion SolNaf f mf m  (6.5) The derivation can be realized by starting with:  ( )NafionVulcanXCCatalystNafionNafion mmfm += +  (6.6)  5. Clean an appropriate size beaker with isopropanol. 6. Weigh the calculated amount of premixed catalyst/C powder in the beaker. 7. Add water (before adding isopropanol) to the weighed powder. The carbon powder will agglomerate over the water surface. Submerge as much of the carbon powder as possible by stirring. 8. Add isopropanol (after adding water) to completely solvate the carbon powder. Try to obtain reasonable dispersion by stirring. One can constantly adjust the amount of water and isopropanol added to find what works best. Generally, much less isopropanol is added than water. If the isopropanol is added before water, the carbon powder and isopropanol will combust together.  233 9. Add the calculated mass of Nafion® solution (after adding water) with a micropipette. Stir. 10. Cover the mixing beaker with parafilm and sonicate the mixture for 30- 60 min. Inspect and stir the mixture every 10-15 minutes. A homogenous dispersion is desired. 11. Thoroughly clean the spray gun with isopropanol. 12. Ensure that the purple nozzle cover is horizontal. 13. Partially pressing the spray gun trigger will permit air to pass through the nozzle, but not ink. While shooting air, adjust the air pressure to read 15 psi. The pressure should be greater than 15 psi after releasing the trigger. 14. Select your substrate and record the initial weight. The difference between this weight and the weight after spraying indicates the catalyst loading. 15. Set the hot plate under the spray plate to 3 (~80°C). 16. Pour approximately 2/3 of the prepared ink into the spray gun. Spray all of it, acknowledging that it won’t be enough to overspray. Spraying an electrode for use in a fuel cell should be done only after practicing proper spraying techniques. 17. Add the last third of the ink to the gun and weigh the sprayed substrate periodically to achieve your desired catalyst loading. 18. Place the sprayed substrate in the oven at 80°C for 30 minutes to remove the water and isopropanol.  234 B.8 Fuel Cell Test System Start-Up The fuel cell should be stored disassembled. A photograph of the disassembled 4 cm² fuel cell is shown in Figure 6.12.   Figure 6.12. Photograph of the 4 cm² fuel cell disassembled.  The first step is to insert the membrane and electrodes into the active areas of the flow field plates and reassemble the fuel cell by stacking the components in the following order: 1) Cathode insulating support (brown plastic), 2) Cathode current collector, 3) Cathode flow field, 4) Anode flow field, 5) Anode current collector, 6) Anode insulating support, 7) Blue top plate. Finally, insert the support pins into the slots on the top of the fuel cell. A photograph of the assembled 4 cm² fuel cell is shown in Figure 6.13. This fuel cell is integrated into the fuel cell test system shown in Figure 6.14.  235  Figure 6.13. Photograph of the assembled 4 cm² fuel cell.   Figure 6.14. Photograph of the fuel cell test system.  236 The following series of steps outline the process of starting up the fuel cell test system:  1. Turn on the constant temperature bath and set it to the desired temperature. Top up the water bath with distilled water if needed. 2. Connect the N2 line to the lower blue stainless steel plate and pressurize/compress the cell to the desired pressure (usually 30 psi). 3. Insert the color coded banana wires into the fuel cell current collectors. 4. Insert the anode and cathode thermocouples into the holes in the flow fields. 5. Remove the caps from the fuel cell outlet lines. 6. Power on the anolyte and catholyte peristaltic pumps. 7. Insert the peristaltic pump inlet lines into the desired anolyte and catholyte storage bottles. 8. Set the flow rate of the peristaltic pumps and press “Start” on the pumps to purge the lines with the desired electrolytes. 9. Turn on the fuel cell temperature controllers and set the desired cell temperature. 10. Start the Solartron 1470E Multistat and load the Corrware software. 11. Follow instructions 1-4 from the Cyclic Voltammetry section in Appendix B to perform initial configuration of Corrware software. 12. Insert a new galvanostatic experiment, double click it and configure the measurement as needed. Insert a file name, location and comments if  237 useful as well as the total current, duration and data sampling rate. This process may be repeated if a series of galvanostatic experiments will be run in succession. An example of the galvanostatic experiment configuration window is shown in Figure 6.15.   Figure 6.15. Configuring a galvanostatic experiment in Corrware.  13. Select all of the experiments that will be executed and press “Measure Selected”    238 B.9 Fuel Cell Test Sytem Shut-Down 1. Turn off the current by pressing “Stop”. 2. Stop the anolyte and catholyte flow by pressing “Start” on the peristaltic pumps. 3. Shut off the temperature controllers. 4. Remove the peristaltic pump inlet lines from the electrolytes and place them in a beaker of deionized water. 5. Purge the lines with deionized water. 6. Place caps on the ends of the outlet lines of the fuel cell. 7. Depressurize the fuel cell by loosening the nut on the bottom plate of the fuel cell. 8. Disassemble the fuel cell.     239 Appendix C – Direct Liquid Redox Fuel Cell Thermodynamics The reaction temperature can have a significant impact on the potential of an electrode, according to equation 1.16Error! Reference source not found.. The temperature coefficient of a particular reaction, expressed in mV/°C, can be determined using equation 1.15 and the change in entropy. Table 6.1 below summarizes the temperature coefficients for the DLRFC reactions.  Table 6.1. Temperature coefficients and other thermodynamic properties for the DLRFC reactions at 25°C [1].   The temperature coefficients can be used to determine the standard potentials of the DLRFC over a range of temperatures, as shown in Figure 6.16. It is apparent that the potential of the DLRFC cathode is significantly influenced by the temperature whereas the DLRFC anode potential is not.  240 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 20 40 60 80 100 Temperature (°C) St an da rd  P ot en tia l, E 0 T  (V ) Anode Cathode Overall  Figure 6.16. The standard potential of the DLRFC reactions as a function of temperature.   The activity of the reaction products and reactants also have an effect on the equilibrium cell potential, according to equation 1.24. Assuming that the activity of the redox couple ions can be represented by their respective concentrations, a rough estimate for the equilibrium potential of the DLRFC reactions can be determined as a function of the redox couple SOC. The equilibrium potential of the DLRFC cathode and cell reactions as a function of the redox couple SOC is shown in Figure 6.17, Figure 6.18 and Figure 6.19 for temperatures of 25, 70 and 90°C, respectively.  241 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0 20 40 60 80 100 Fe3+/Fe2+ Redox Couple SOC (%) Eq ui lib riu m  P ot en tia l a t 2 5° C  (V ) Cathode Cell  Figure 6.17. Equilibrium cell potential of the DLRFC at 25°C as a function of the redox couple SOC. (All reactants/products apart from the redox couple are assumed to have an activity of 1)  242 0.40 0.50 0.60 0.70 0.80 0.90 1.00 0 20 40 60 80 100 Fe3+/Fe2+ Redox Couple SOC (%) Eq ui lib riu m  P ot en tia l a t 7 0° C  (V ) Cathode Cell  Figure 6.18.Equilibrium cell potential of the DLRFC at 70°C as a function of the redox couple SOC. (All reactants/products apart from the redox couple are assumed to have an activity of 1)  243 0.40 0.50 0.60 0.70 0.80 0.90 1.00 0 20 40 60 80 100 Fe3+/Fe2+ Redox Couple SOC (%) Eq ui lib riu m  P ot en tia l a t 9 0° C  (V ) Cathode Cell  Figure 6.19. Equilibrium cell potential of the DLRFC at 90°C as a function of the redox couple SOC. (All reactants/products apart from the redox couple are assumed to have an activity of 1)     244 Appendix D – Competition Between Fe2+/Fe3+ Ions and Protons in Nafion® Membranes In Chapter 2, Figure 2.5 highlights the negative impact of the Fe2+/Fe3+ redox couple on the conductivity of protonated Nafion® 117 membranes samples. A loss of 80-88% in the membrane conductivity was observed in the temperature range of 25-80°C after exposure to the redox electrolyte [0.81 FeNH4(SO4)2, 0.09 M FeSO4, 0.5 M H2SO4] relative to that measured prior to contamination. It was hypothesized that the detrimental effect of the iron ions on the Nafion® membrane conductivity might be reduced by increasing the proton concentration in the redox electrolyte, based on the assumption that the proton and iron ions are in active competition with one another for the SO3- sites within the Nafion® membrane. The effect of the sulfuric acid concentration in the redox electrolyte on the membrane was investigated through membrane ion exchange capacity (IEC) and membrane conductivity measurements. In these experiments membrane contamination was performed at 70°C by immersing membrane samples the redox electrolyte. Membrane conductivity measurements and IEC measurements were performed at 25°C. The redox electrolyte was comprised of 0.81 FeNH4(SO4)2, 0.09 M FeSO4 and X M H2SO4. The results of this study, shown in Figure 6.20, indicate that increasing the sulfuric acid concentration in the redox electrolyte from 0 M to 3.5 M reduces the negative impact of the redox ions on the membrane conductivity and IEC from 87% to 69% and 41% to 19%, respectively. It is interesting to note that despite the significant improvement in the IEC, the membrane conductivity does not  245 follow in a proportional manner. In summary, increasing the sulfuric acid concentration in the redox electrolyte above 0.5 M does not provide a substantial benefit towards the membrane conductivity that would warrant the increased electrolyte preparation cost and viscosity (i.e., pumping requirements).  0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 0 0.5 1 1.5 2 2.5 3 3.5 4 [H2SO4] (M) Io n Ex ch an ge  C ap ac ity  (m Eq /d ry  g ) 0 10 20 30 40 50 60 70 80 90 100 C on du ct iv ity  (m S/ cm ) Pure H2O Dupont's quoted values IEC: 0.83 mEq/dry g Conductivity: 89 mS/cm Pure H2O After exposure to redox electrolyte After exposure to redox electrolyte  Figure 6.20. Study of competition between iron ions and protons in Nafion 117® membranes. Immersion of membrane samples in redox electrolyte [0.81 FeNH4(SO4)2, 0.09 M FeSO4 and X M H2SO4] performed at 70°C. Membrane conductivity and IEC measurements performed at 25°C.   246 Appendix E – Repeatability and Reproducibility of the DLRFC E.1 DLRFC Repeatability A small series of repeatability tests were conducted for the DLRFC at 70°C. The results, shown in Figure 6.21, show that all galvanostatic measurements (with the exception of one) are within 20 mV of the others at the same current density.  0.0 0.1 0.2 0.3 0.4 0.5 0.6 0 20 40 60 80 100 120 140 160 Current Density (mA/cm²) C el l P ot en tia l ( V) Repetition 1 Repetition 2 Repetition 3  Figure 6.21. Repeatability tests (IR-corrected) conducted on a sulfate-based DLRFC operating at 70°C. (Nafion® 112 membrane, 2 mg/cm² 40% 1:1 mol/mol Pt:Ru on C anode catalyst, 2 M CH3OH / 0.5 M H2SO4 anolyte, 5 mL/min anolyte flow rate, TGP-H-120 (3x) cathode, 0.81 M FeNH4(SO4)2 / 0.09 FeSO4 / 0.5 M H2SO4 catholyte, 5 mL/min catholyte flow rate)  247 E.2 DLRFC Reproducibility The following replicate measurements of a perchlorate-based DLRFC at 90°C, shown in Figure 6.22, were performed on separate dates using different sets of electrolytes and electrodes (with identical composition). The data indicates a high degree of reproducibility for the perchlorate-based DLRFC.  0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0 100 200 300 400 500 Current Density (mA/cm²) Po te nt ia l ( V) Replicate 1 Replicate 2  Figure 6.22. Replicate polarization curves (IR-corrected) for a perchlorate-based DLRFC at 90°C. (Nafion® 112 membrane, 2 mg/cm² 40% 1:1 mol/mol Pt:Ru on C anode catalyst, 1 M CH3OH / 0.2 M HClO4 anolyte, 1 mL/min anolyte flow rate, TGP-H-120 (3x) cathode, 2.5 M Fe(ClO4)3 catholyte, 2 mL/min catholyte flow rate)   248 Appendix F – Advancements in the Direct Hydrogen Redox Fuel Cell1 F.1 Introduction The proton exchange membrane fuel cell (PEMFC) is considered to be one of the most promising power devices with a broad range of use including transportation, stationary, portable, and micro-applications [2].  However, there are still a number of challenges especially concerning performance, cost, and reliability of these fuel cell systems.  A large effort is directed toward the development of improved proton-exchange membranes and cathode materials, which are largely responsible for the loss of efficiency and the high cost [3].  The oxygen reduction reaction (ORR) at the cathode-PEM interface is fundamentally the main limiting factor for PEMFC performance and efficiency [4].  This cathode- PEM interface is complex and limiting because of the slow kinetics for the ORR, even with noble metal catalysts, and the triple-phase boundary constraints requiring effective electronic, protonic and reactant contact with the catalyst. This TPB requires complex catalyst layer structures in order to control reactant distribution and diffusion, and the complicated water management. Replacing the oxygen cathode in a conventional PEMFC by a redox species cathode could be a promising approach to resolve many of the issues with the ORR.  When oxygen is replaced by a redox oxidant the rate of the cathode reaction even without noble based catalysts is significantly higher than that observed for the ORR. The exchange current density of the Fe3+/Fe2+ redox 1 A version of this manuscript has been published.  K. Fatih, D. P. Wilkinson, F. Moraw, A. Ilicic, F. Girard. Advancements in the Direct Hydrogen Redox Fuel Cell. Electrochem. Solid- State Lett., 11, 2 (2008) B11-B15. Reproduced by permission of The Electrochemical Society.  249 couple is usually several orders of magnitude higher (approx. 10-2 A/cm2 [5]) than for oxygen (approx. 10-10 A/cm2 [6]).  Also, the rate of mass transfer of the oxidant to the electrode surface is increased by several orders of magnitude compared to oxygen due to the much higher aqueous solubility of the redox oxidant (1 M or higher) compared to oxygen, 2.67x10-4 M at 25ºC and 1.73x10-4 at 70ºC [7].  Although many benefits are possible with a redox oxidant there are challenges with respect to regeneration, which need to be considered in an overall system analysis.  Regeneration can be carried out by a variety of processes including, chemical methods [7], electrochemical methods (electrically rechargeable redox-flow cell) [8], biological methods [9, 10] and possibly radiochemical methods [11].  The biological regeneration of Fe3+ appears to be a promising method based on the oxidation of concentrated Fe2+ solution by autotrophic microorganisms, which are able to withstand harsh conditions such as a pH of 1 and a temperature of 40°C [9].  In this paper we will only discuss the fuel cell energy conversion aspect of the redox oxidant. Yeager et al. [11] in 1962 were first to report the use of  H2 with a redox couple (Fe3+/Fe2+) based on a radiochemical regeneration fuel cell scheme first proposed in 1957 by J.A. Ghomley at the Parma Research Laboratory at Union Carbide.  They were interested in the radioactive regeneration of Fe3+ according to the reaction  2Fe2+ + 2H+ → H2 + 2Fe3+ (6.7)  The unit cell was based on a concentric design that used a microporous polyvinyl chloride separator material (PorvicTM), high loaded platinum catalyzed  250 wet-proofed carbon electrodes, and an electrolyte consisting of equimolar FeSO4 and Fe2(SO4)3 with 1 M H2SO4.  Performance results were very poor typically in the voltage range of 0.35 to 0.50V at 20 mA/cm2 with very low overall gravimetric and volumetric power density. Surprisingly, the hydrogen redox fuel cell approach appears not to have been revisited in detail since Yeager’s proceeding article in 1962.  The terms redox flow cell, redox battery and redox fuel cell have been used synonymously in the literature to refer to redox couples on both the anode and cathode, and the term redox- hybrid in the case when metallic species are deposited in one half- cell [8].  However, direct hydrogen / oxygen fuel cell technology has significantly advanced with respect to materials, design and performance over the last twenty years.  In particular, the low-temperature PEMFC is now at a field trial level / early commercialization stage but a number of technology gaps remain. Therefore, it is useful to revisit the redox approach to look for opportunities to advance conventional fuel cell technology.  F.2 Experimental The electrochemical evaluation of the redox couple Fe3+/Fe2+ was investigated using cyclic voltammetry (CV) with a potentiostat-galvanostat (Solartron 1480, multi-channels).  In addition electrochemical impedance spectroscopy (EIS) with a frequency response analyzer (Solartron 1255B) was used for determining resistance for iR compensation.  Working electrode materials consisted of platinum, Glassy Carbon (GC) and three-dimensional  251 porous carbon fiber paper, TGP-H090 (Toray, E-Tek, thickness 0.28 mm) and thin carbon felt (Electrolytica, thickness 0.3 mm).  All electrochemical measurements were conducted in an oxygen-free single compartment three- electrode glass cell with a Pt gauze counter electrode and a Hg/Hg2SO4/K2SO4 (saturated) reference electrode (unless otherwise specified).  The carbon electrodes were prepared by connecting one edge to a lead wire with silver epoxy (Trancene, Inc) and sealed in a glass tube with an epoxy patch (Hysol-C1, Loctite).  The electrolytic solutions were prepared using DI water, Fe(NH4)(SO4)2. 12H2O, FeSO4. 7H2O and H2SO4 as a supporting electrolyte mixed to the desired electrolyte composition.  The GC electrodes were polished with a 1 µm diamond suspension (Buehler) then with a 0.05 µm Alumina suspension (Buehler), sonicated in DI water and then stored in concentrated sulfuric acid.  The GC electrode was cleaned with DI water prior to use and the porous electrodes were cleaned with isopropanol followed by DI water.  Porous electrodes were stored in concentrated sulfuric acid and sonicated in DI water prior to use.  The exposed superficial area of the porous electrode was 0.3 cm2 for the TGP-H090.  CV measurements were carried out in a static electrolyte solution and were repeated at least three times to ensure reproducibility The hydrogen anode was fabricated by first spraying a sub-layer on the wet-proofed gas diffusion layer (e.g., Toray TGPH-090 contained 10wt% PTFE) consisting of a mixture of 80 wt % carbon (Vulcan XC-72) and 20 wt % PTFE (DuPont) in isopropanol.  The catalyst ink consisting of a mixed catalyst powder (20 wt % Pt on Vulcan XC-72, E-Tek), Nafion® solution (5 wt % in alcohols/water,  252 Aldrich) and isopropanol was then spray deposited on the previously prepared sub-layer.  Both layers were dried at room temperature and then at 120°C for 30 min.  This procedure resulted in a carbon loading in the sub-layer of approx. 0.2 mg/cm2 (80% C and 20% PTFE w/w) and a Pt loading in the catalyst layer of approx. 0.2 mg/cm2 while the Nafion® composition in the catalyst layer was 30 wt %.  Nafion®115 membrane square pieces were pre-treated for 30 min in a 3 wt % H2O2 solution at 90°C then boiled in 0.5 M H2SO4 for 30 min and stored in DI water after rinsing.  A 4 cm2 square piece of the anode as prepared above, was hot-pressed to Nafion® 115 at 80 kg/cm2 and 145°C for 240 s.  The cathode carbon electrodes were first pre-treated with 1M HNO3 and thoroughly rinsed with DI water prior to being cut, and several pieces (4 for Toray TGPH-090 and 7 for the thin felt) were fitted into a square pocket (4 cm2 area with a depth of 0.1 cm) in the cathode carbon current collector plate.  The catholyte used for fuel cell testing was 0.9M in total iron with a Fe3+/Fe2+ ratio of 9 (to mimic a 90% oxidant regeneration process) at a pH of 1 and was preheated prior to entry in the fuel cell.  This solution was prepared using DI water (Millipore water, 18.2 MΩ cm), ferric ammonium sulfate Fe(NH4)(SO4)2. 12H2O (ACS. Fisher), ferrous sulfate FeSO4. 7H2O (ACS. Fisher), and adjusting the pH with sulfuric acid (ACS. Anachemia).  A single fuel cell (active area approx. 4 cm2) with a single pass serpentine anode and a pocket cathode was assembled, compressed with a nitrogen bladder, and electrically connected to a potentiostat / galvanostat (Solartron 1480, multi-channels).  After the fuel cell temperature had stabilized with the flowing redox solution, the open circuit voltage (OCV) was recorded for 5  253 min followed by a potentiodynamic polarization curve (scan rate of 0.2 mV/s) and then a galvanostatic polarization curve.  All fuel cell testing was carried out at least three times to ensure reproducibility.  F.3 Results and Discussion Figure 6.23 shows that the direct hydrogen redox fuel cell is a hybrid between the H2 / air PEMFC and a redox fuel battery.  The liquid redox cathode enables the use of a three dimensional electrode without the complex catalyst microstructure that is required for the ORR.  Also, as a result of the liquid cathode no separate or external humidification or cooling is required and there are no water management issues for the cathode.  Variation of the hydrogen humidification over a wide range (0 to 100% RH) had no effect on cell performance. In the research reported here the Fe3+ / Fe2+ redox couple is used as the model cathode in the direct hydrogen redox fuel cell with the following electrochemical reactions:  Anode 2H+ + 2e- ← H2 E0 = 0 V vs. SHE (6.8) Cathode 2Fe3+(aq) + 2e- → 2Fe2+(aq) E0 = 0.77 V vs. SHE (6.9) Overall 2Fe3+(aq) + H2 → 2Fe2+(aq) + 2H+ E0 = 0.77 V (6.10)  This redox cathode has a standard potential high enough to get useful cell voltage and at the same time low enough for ferric species to be regenerated by oxygen.  Also, this redox couple is readily available and inexpensive.  This redox  254 couple was extensively studied in the second half of the last century in the Fe-Cr redox flow battery [8, 12].   Figure 6.23. Schematic diagram of a direct hydrogen redox fuel cell showing it is a hybrid of a PEMFC and a redox flow battery.  It has been reported in the literature that the mechanism of the charge transfer of the aqueous ferric/ferrous complexes in reaction (2) is not an outer- sphere process but rather an inner-sphere process [13].  The kinetics of the ferric/ferrous couple on glassy carbon was reported to be surface and oxide sensitive [14, 15].  For inner-sphere charge transfer as expected with the Fe3+/Fe2+ redox couple on carbon-based electrodes, the kinetics of the charge transfer strongly depends on the nature of the iron complexes and their electrochemical behavior.  A recent study of the kinetics of the Fe3+/Fe2+ redox  255 couple showed similar kinetic data for platinum, glassy carbon and Toray TGPH - 090 (carbon fiber paper), and the complex association of ferric ions with sulphate [16].  This makes the electrolyte composition and its speciation an important consideration for replacement of the challenging oxygen cathode with the Fe3+/Fe2+ redox couple.  In fact, in a complex medium such as sulfuric acid based solutions, ferrous and ferric species may exist as free ions and /or complex compounds.  Therefore, using a redox cathode presents opportunities as well as challenges to tailor the electrolyte composition to provide the desired properties for better charge transfer kinetics at given operating conditions. Key to the use of the liquid redox system with hydrogen is a tolerance to cross-over and minimal depolarization at the active electrode surfaces.  Figure 6.24 shows a comparison of cyclic voltammograms for hydrogen and the liquid redox system mixed on glassy carbon, and for the liquid redox system alone on glassy carbon and platinum.  These tests were done at 70°C, the maximum temperature used in the study where activity would be greatest.  These results show similar activity for the redox system on both glassy carbon and platinum. Also, there is little to no activity of hydrogen on glassy carbon indicating hydrogen permeation to the cathode is not a problem as a result of reaction selectivity.  However, since hydrogen and the redox couple are both active on platinum redox species cross-over to the anode is considered to be a problem. Improved membrane selection would be expected to mitigate this issue. Although the redox couple has shown higher activity in diluted redox solution on platinum compared to glassy carbon, this activity was observed to be similar in  256 concentrated redox solution [16], and highlights the potential for significant platinum group metal reduction in the direct hydrogen redox fuel cell.   Figure 6.24. Cyclic voltammograms (70°C, scan rate 50 mV/s) for hydrogen and the Fe3+ / Fe2+ redox couple mixed on glassy carbon and for the redox couple alone on glassy carbon and platinum.  Also key to the use of the liquid redox cathode system is the ability to use a three dimensional electrode to extend the reaction zone and increase active surface area.  This approach has been used predominantly with liquid fuel feeds such as in the direct methanol fuel cell to increase fuel utilization, and performance, and decrease fuel crossover [17, 18].  It has generally involved distributing catalyst throughout the electrode structure.  In the case of the liquid  257 redox cathode system the carbon diffusion layer itself is active and no further catalyst application is required, thus greatly simplifying the electrode structure and processing.  Figure 6.25 shows a comparison of cyclic voltammograms for the redox system in two different solutions on glassy carbon and on Toray TGP- H090 carbon fiber paper based on the electrode superficial area.  Similar potentials are observed for the redox couple indicating similar activation on the two different carbons.  The real area accessed for the carbon fiber paper (or any other three dimensional electrode) can be estimated from the ratio of peak current based on the Randles - Sevcik equation [19],  ( ) 2121235p CvDAn106862i ×= .  (6.11) where ip is the peak current (A), n is the number of electron involved in the electrode process, A is the electrode area (cm2), D is the diffusion coefficient (cm2 s-1), C is the concentration of the electroactive species (mol cm-3), and v is the scan rate (V s-1). The concentration of the electroactive species used in equation 6.11 was taken as the total iron in the solution.  Based on this analysis there is an approximate 4 times increase in accessible active area for the three- dimensional carbon fiber paper over that of the glassy carbon electrode.  The total potential area of the carbon fiber paper is substantially larger but accessibility is limited by cell design, electrolyte composition and operating conditions.  258   Figure 6.25. Comparison of cyclic voltammograms (25°C, scan rate 100 mV/s) for the Fe3+/Fe2+ redox couple on glassy carbon and on Toray TGPH-090 carbon fiber paper in different solutions:  a) 5 mM K3Fe(CN)6 in 1 M KNO3;  b) 0.81 M / 0.09 M Fe(NH4)(SO4)2. 12H2O / FeSO4. 7H2O in 0.5 M H2SO4.  259 Fuel cell performance for the direct hydrogen redox fuel cell was evaluated for different electrode materials and operating conditions.  For all experiments the total iron concentration was kept at 0.9M with the ratio of the ferric / ferrous species kept at 9 to simulate a redox regeneration process with an efficiency of 90%, i.e., a typical feed composition in a redox system.  Figure 6.26 shows the effect of cell compression on performance with respect to a carbon fiber and a thin felt three-dimensional cathode.  The best performance is obtained at a similar compression pressure of 1.97-2.11 kg/cm2 indicating the importance of balancing contact resistance and electrode porosity.  Comparison of polarization performance and associated power density for the two different porous cathode materials is shown in Figure 6.27.  The TGPH-090 had superior performance and was therefore used for evaluation of different operating conditions. Figure 6.28 shows the effect of different fixed iron solution flow rates on polarization behavior.  Adequate stoichiometry is indicated for the high current density points but it is clear that a minimum threshold level of flow is required to achieve the highest current density.  This is likely related to the utilization of the 3D cathode with respect to the present cell design for reactant distribution. Similarly, if the stoichiometry is set high enough similar performance is achieved compared to the fixed flow polarization.  These results show that the nature of the three dimensional cathode, the cathode flow field design, and the cathode flow regime can have a significant impact on the performance of the redox system.   260  Figure 6.26. Effect of compression on direct hydrogen redox fuel cell performance for carbon fiber paper and thin felt cathodes (40°C, catholyte flow rate 10 ml/min).  261  Figure 6.27. Comparison of polarization and power density curves for carbon fiber paper and thin felt cathodes in the direct hydrogen redox fuel cell (70°C, catholyte flow rate 10 ml/min).  Figure 6.28. Direct hydrogen redox fuel cell polarization curves obtained at 40°C for different flow rates of the Fe3+ / Fe2+ solution. Toray TGPH-090 carbon fiber paper used for the cathode.  262 Figure 6.29 shows the polarization performance of the direct hydrogen redox fuel cell at different temperatures and the associated power density curves. Very stable and reproducible open circuit voltages of around 0.775 V were obtained that were practically independent of the temperature.  The calculated Nernst values of the open circuit voltage (OCV) using redox species concentrations at the cathode, and a pH of 1 and a pressure of 2 atm at the anode are 0.895 V and 0.913 V at 25oC and 70oC, respectively.  The observed difference in the OCV compared to thermodynamic values is mainly due to the activity of the redox species at the cathode.  In fact, the redox potential of the Fe3+/Fe2+ couple measured at 25oC using a commercial ORP (oxidation reduction potential probe) electrode was around 0.700 V vs NHE.  This results in a calculated OCV of 0.768 V, which compares well with the measured value of 0.775V.  Figure 6.29 also shows that there is a strong dependence of the cell performance on temperature with an adequate catholyte flow rate.  The linearity of the polarization plots at higher current densities and the strong dependence on temperature are indicative of ohmic control.  An approximate theoretical calculation assuming all of the electrode surface area is used and the mass transfer coefficient is about 10-5 m/s (common for turbulent liquid flow) reveals that the limiting currents should be greater than 1 A/cm2, well beyond what is observed experimentally.  263  Figure 6.29. Direct hydrogen redox fuel cell polarization and power density curves obtained in the temperature range of 25°C to 70°C for a constant Fe3+ / Fe2+ solution flow rate of 10 ml/min.  Toray TGPH-090 carbon fiber paper used for the cathode.  Further clarification of the ohmic issue was determined by taking conductivity measurements of the Nafion® 115 membrane before and after exposure to the iron based solution. The conductivity of Nafion® 115 after overnight exposure to DI water, sulfuric acid solution (pH = 1), and a working iron solution (pH = 1) is shown in Table 6.2. It is clear that as a result of the uptake of Fe3+ and Fe2+ the membrane conductivity is significantly reduced by around 86% from its baseline value. Similar conductivity decrease (approx. 80%) with Nafion® 117 uptake of Cr3+ was reported by Shores and Deluga [20].  However, after exposing contaminated membrane to sulfuric acid solution (pH = 1) for one hour, the conductivity partially  264 recovers to about 40% of the baseline value.  Under other conditions such as higher temperature and more concentrated acid solution it is likely possible to recover more of the baseline membrane conductivity. These results clearly indicate the importance of improving the membrane/separator for the direct hydrogen redox fuel cell in order to reduce contamination and cross-over by the redox species.  Table 6.2. Conductivity measurements of Nafion® 115 membrane before and after exposure to the iron based solution at 22°C.    amembrane exposed overnight in solution at 22°C bmembrane soaked for 1 hour in H2SO4 (pH = 1) solution at 22°C after being exposed to the iron solution  265 F.4 Conclusions In this study we have shown that the direct hydrogen redox fuel cell has significant merit as an alternative type of fuel cell system.  Using a redox couple to replace the oxygen reduction reaction at the cathode has a number of potential benefits with respect to performance and design.  The ability to use three dimensional carbon based cathodes without catalyst allows a reduction in total platinum group metal loading in the range of 65 to 80% over the state-of-art conventional air/hydrogen PEMFC and much lower platinum loadings could be used on the anode.  Although a cell performance of up to 0.17 W/cm2 (or 0.85 W/mg Pt) was achieved it was severely limited by ohmic control.  Significant improvements in performance are expected with further improvement of electrolyte composition, operating conditions and materials, particularly the membrane.  F.5 Acknowledgements This research was supported by the National Research Council of Canada Institute for Fuel Cell Innovation (NRC-IFCI) and by the Natural Sciences and Engineering Research Council of Canada (NSERC). The authors thank Robert Chow and Titichai Navessin for their contribution in the MEA fabrication.   266 F.6 References 1. D. R. Lide, CRC Handbook of Chemistry and Physics, CRC Press LLC, Boca Raton (2005). 2. D.P. Wilkinson, The Electrochem. Soc. Interface, 10-1, 22 (spring 2001). 3. E. J. Carlson, P. Kopf, J. Sinha, S. Sriramulu, and Y. Yang, Nat. Renew. Energ. Lab / TIAX LLC (Eds.), NREL, Report n° NREL/SR-560-39104 (2005). 4. D. Thompsett, in Handbook of Fuel Cells, Fundamentals, Technology and Applications, W. Vielstich, H. A. Gasteiger, and A. Lamm , Editors, vol. 3, Ch. 37, p. 467, John Wiley and Sons Ltd (2003). 5. F. C. Anson, Anal. Chem., 33, 939 (1961). 6. A. Damjanovic, and J. O. Bockris, Electrochim. Acta., 11 , 376 (1966). 7. R. Battino, T. R. Rettich, and T. Tominaga, J. Phys. Chem. Ref. Data, 12, 163 (1983). 8. M. Bartolozzi, J. Power Sources, 27, 219 (1989). 9. D. Karamanev, PCT n° PCT/CA04/00943 (2004), US Patent Application n° 20060251959 (2006). 10.T. HIDEO, Japan Patent n° 62256382 (1987). 11.J. F. Yeager, R. J. Bennett, and D. R. Allenson, Proc. Ann. Power Sources Conf., 16, 39 (1962). 12.K. Sawai, I. Tari, T. Ohzuku, and T. Hirai, Mem. Fac. Eng. , Osaka City Univ., 29, 139 (1988). 13.P. Chen, and R. L. McCreery, Anal. Chem., 68, 3958 (1996). 14.P. Chen, M. A. Fryling, and R. L. McCreery, Anal. Chem., 67, 3115 (1995).  267 15.I. Yagi, H. Notsu, T. Kondo, D. A. Tryk, and A. Fujishima, J.  Electroanal. Chem., 473, 173 (1999). 16.K. Fatih, D. P. Wilkinson, F. Moraw, and F. Girard, in Electrocatalysis Proceeding, Adzic, Birss, Brisard, and Wieckowski, Editors, PV 2005-11, p. 341, The Electrochemical Society Proceedings Series, Pennington, NJ (2005). 17.D. P. Wilkinson, M. C. Johnson, K. M. Colbow, and S. A. Campbell, US Patent n° 5,672,439  (1997). 18.D. P. Wilkinson, M. C. Johnson, K. M. Colbow, and S. A. Campbell, US Patent n° 5,874,182 (1999). 19.A. J. Bard, and L. R. Faulkner, Electrochemical Methods: Fundamentals and Applications.,  2nd Ed. ed., John Wiley & Sons Inc., 2001. 20.D. A. Shores, and G. A. Deluga, in Handbook of Fuel Cells, Fundamentals, Technology and Applications, W. Vielstich, H. A. Gasteiger, and A. Lamm , Editors, vol. 3, Ch. 23, p. 273, John Wiley and Sons Ltd (2003).

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.24.1-0058853/manifest

Comment

Related Items