Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Electrochemical pathways for sustainable energy storage and energy conversion
(USC Thesis Other)
Electrochemical pathways for sustainable energy storage and energy conversion
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
Electrochemical Pathways For Sustainable Energy Storage and Energy Conversion by Buddhinie Srimali Jayathilake A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA in Partial Fulfillment of the Requirement for the Degree DOCTOR OF PHILOSOPHY (CHEMISTRY) May 2019 ii iii Dedication I dedicate my thesis to my beloved family. My sweetheart, this thesis is the return for all your commitment and unbelievable moral support throughout a wonderful decade from our freshmen year to today for following my dreams. iv v Acknowledgements • Thank you, USC for accepting me for the Ph.D. program and the Loker Hydrocarbon Research Institute for allowing me to experience the cutting-edge research. • Dear Professor Sri Narayan, thank you very much for being my Ph.D. advisor, for your never-ending guidance, motivation and all the kinds of support for completing my studies. Our lab is like a small industrial plant with testing different battery technologies which begin from the fundamental electrochemical reactions. Thank you again, for identifying my capabilities and for giving me the opportunity to engage in multiple projects to expand my knowledge and understanding and shining skills in electrochemistry. • Professor Surya Prakash and Professor Hanna Reisler, you two were my first graduate advisors at USC and thank you very much for all the encouragement from the beginning in graduate school. • I thank all the members of my dissertation committee for all the advice given to me for a successful completion of my doctoral studies. • Thank you Dr. Nagarajan Vaidehi and Dr. Supriyo Bhattacharya for collaborating with us. It was exciting working with you! Thank you, Sayan Kar, for synthesizing chemicals for my research. • Thank you, all my past and present lab mentors and lab mates for all the support throughout the last few years! • My sincerest acknowledgments to Dr. Robert Aniszfeld, David Hunter, Jessie May and all the members at the Loker Hydrocarbon Research Institute, and Michele D, Magnolia Benitez and all the faculty and staff members in the Chemistry department, for all your kind support at different stages in the program. Donald Wiggins and Michael Cowan of the USC Dornsife Machine shop, thank you very much for support with the fabrication of electrodes and workshop activities. Thank you so much, Phillip Swag for building all the glassware for electrochemical measurements. Thank you very much Allan Kershaw, vi Dr. Frank Devlin and CEMMA staff for training and facilitating all the characterization instruments. • Dr. Rebecca Broyer, thank you very much! It was wonderful being a TA for you. Amanda Baxter, Dr. Courtney Downes and all the mentors and co-TAs at USC, it was nice working with you all and thank you for all the time we spent together at USC. • Graduate school was a challenge from the first day, but I had you, my best friends Parmeet Kaur and Wei Jiang, since we met in the very first day at USC in 2014 and all other friends at USC, thank you for all your help and cheers in graduate school. Special thanks to the Sri Lankan Crew at USC and Sri Lankan community in Los Angeles for all your caring like a family throughout the last several years. • My lovely, mom and dad without you as my heroes, without your unconditional love, caring and non-stop encouragement, fulfilling my doctoral studies will be a dream forever. My sisters, my love for you two for everything you do for me. Shamal, when I look back, memories in the last 11 years are admirable and thank you for everything. vii Table of Contents Dedication ........................................................................................................................... iii Acknowledgements ............................................................................................................. v List of Tables ....................................................................................................................... xiii List of Figures ...................................................................................................................... xv List of Abbreviation and Symbols ........................................................................................ xxii Abstract .............................................................................................................................. xxiv Introduction ................................................................................................. 1 1.1 Electrochemical Energy Storage Systems ........................................................... 1 1.2 Redox Flow Batteries ........................................................................................... 2 1.2.1 Screening of Redox Couples ............................................................................... 3 1.2.1.1 Aqueous Inorganic Redox Flow Cells .................................................................. 4 1.2.1.2 Aqueous All-Organic Redox Flow Cells .............................................................. 5 1.2.1.3 Aqueous Organic-Inorganic Mixed Redox Flow Cells ......................................... 6 1.2.1.4 All-Iron Redox Flow Cell ...................................................................................... 7 1.2.2 Electrode Materials .............................................................................................. 8 1.2.3 Charge Transfer Kinetics and Electrocatalysis ..................................................... 9 1.2.4 Mass Transport .................................................................................................... 11 1.2.5 Membrane Development .................................................................................... 12 1.2.6 Supporting Electrolytes ....................................................................................... 14 1.2.7 Efficiency .............................................................................................................. 14 1.2.8 Electrolyte Regeneration or Recombination ....................................................... 16 1.3 Lithium-Sulfur Batteries ....................................................................................... 17 1.4 Electrochemical Energy Conversion Pathways for Carbon Dioxide Reduction .. 18 1.4.1 Carbon Dioxide Emission and Environmental Impact ......................................... 18 1.4.2 Thermodynamics and Kinetics of Carbon Dioxide Reduction ............................. 19 1.4.3 Electrochemical Carbon Dioxide Reduction Using Metal Catalysts .................... 20 1.4.4 Electrochemical Carbon Dioxide Reduction Using Bio-Catalysts ....................... 22 1.4.5 Carbon Dioxide and Formate Interconversion Using Metal Independent FDH . 23 1.4.6 Molecular Level Understanding of Catalysis by FDH .......................................... 25 1.4.7 Challenges for Enzyme-Catalyzed Electrochemical Conversion of Carbon Dioxide to Formate ............................................................................................. 26 1.5 Chapter One References ..................................................................................... 28 Improvements to the Coulombic Efficiency of the Iron Electrode for ......... an All-Iron Redox Flow Battery .................................................................... 39 2.1 Background ......................................................................................................... 39 2.1.1 The Implications of Parasitic Hydrogen Evolution ............................................... 39 2.2 Experimental Methods ........................................................................................ 42 viii 2.2.1 Coulombic Efficiency of Iron Deposition and Dissolution ................................... 42 2.2.2 Investigations of the Hydrogen Evolution Reaction ............................................ 44 2.2.3 Correction of Electrode Potentials for Ohmic Drop ........................................... 45 2.2.4 Impedance Spectroscopy at Different Temperatures ......................................... 45 2.2.5 Near-surface pH and Bulk pH Measurement ...................................................... 45 2.3 Results and Discussion ......................................................................................... 46 2.3.1 The Effect of Electrolyte pH ................................................................................ 46 2.3.2 The Effect of Ascorbic Acid on the Kinetics of Hydrogen Evolution .................. 48 2.3.3 The Effect of Ascorbic Acid on the Kinetics of Electrodeposition of Iron .......... 50 2.3.4 Effect of Current Density on Coulombic Efficiency ............................................. 51 2.3.5 Effect of Electrolyte Additives ............................................................................. 54 2.3.5.1 Increasing the Solubility of the Electrolyte at Higher pH .................................... 54 2.3.5.2 Increasing the Solubility of Iron Hydroxide at the Electrode Surface ................. 56 2.3.5.3 Inhibition of kinetics of HER using a Second Metal ............................................. 58 2.3.6 Effect of Temperature ......................................................................................... 60 2.4 Chapter Two Conclusions .................................................................................... 63 2.5 Chapter Two References ..................................................................................... 66 Charge Transfer Kinetics of Iron Deposition and Hydrogen Evolution During Iron Plating― Insights Through the Electrolyte Properties and Iron(II) Complex Formation in Ammonium Chloride Solutions ..................... 69 3.1 Background ......................................................................................................... 69 3.1.1 Hydrogen Evolution Kinetics ............................................................................... 69 3.1.2 Iron Deposition Kinetics ...................................................................................... 71 3.2 Experimental Conditions ..................................................................................... 72 3.3 Results and Discussion ......................................................................................... 73 3.3.1 Electrolyte Properties .......................................................................................... 73 3.3.2 Dependence of Coulombic Efficiency of Iron Deposition on the Concentration of Iron (II) ............................................................................................................. 77 3.3.3 Effect of Adsorptive Corrosion-Inhibitors for the Overall Performance of the ... Iron Deposition Process ....................................................................................... 79 3.3.4 Kinetics of Iron Deposition and Hydrogen Evolution on Iron in Iron Chloride .. Solutions .............................................................................................................. 81 3.3.5 Impedance Spectroscopy and Mechanisms of Iron Deposition and Hydrogen Evolution on Iron in Iron Chloride Solutions ....................................................... 83 3.3.6 Evans Diagram Construction ............................................................................... 88 3.4 Chapter Three Conclusions ................................................................................. 89 3.5 Chapter Three References ................................................................................... 91 Critical Analysis of the Performance of Lithium-Sulfur Batteries with Mixed Conduction Membrane ..................................................................... 95 4.1 Background ......................................................................................................... 95 ix 4.1.1 Features of Discharging Curve ............................................................................ 95 4.1.2 Polysulfide Shuttling ............................................................................................ 97 4.1.3 Rate Dependent Li 2S Formation .......................................................................... 98 4.1.4 Characterization of Li-S System Using EIS ........................................................... 99 4.2 Experiments and Methods .................................................................................. 100 4.3 Results and discussion ......................................................................................... 102 4.3.1 Polysulfide Shuttle Hindrance and Charge Transfer via Lithium Ion Intercalation Through MCMs ............................................................................... 102 4.3.2 Improved Initial Cycling with Mixed Conduction Membrane ............................. 105 4.3.3 Analysis of Initial EIS of Li-S Cells ........................................................................ 106 4.3.4 Optimum Electrolyte for Better Ion Transport .................................................... 107 4.3.5 Different Types of Cell Constructions with MCM using KB, AB and Mixed AB . and KB as Carbon sources for Cathodes ............................................................. 110 4.3.6 Electrolyte Distribution ........................................................................................ 113 4.3.7 Material Utilization and Polysulfide Distribution ................................................. 114 4.3.8 Influence of Different Carbon Materials on Rate-Capability ............................... 115 4.3.9 Rate Capability Studies with Mixed Carbon Cathodes ....................................... 121 4.3.10 Summary of Critical Features in Cell Designing .................................................. 123 4.4 Chapter Four Conclusions ................................................................................... 124 4.5 Chapter Four References ..................................................................................... 125 Direct Electrochemical Detection and Quantification of NAD + and NADH .. Using an Unmodified Carbon-Fiber Microelectrode ..................................... 129 5.1 Background ......................................................................................................... 129 5.2 Methods ............................................................................................................... 132 5.2.1 Material and Reagents ......................................................................................... 132 5.2.2 Apparatus ............................................................................................................ 132 5.2.3 Preparation of Standard Solutions ...................................................................... 132 5.2.4 Electrochemical Generation of NADH ................................................................ 132 5.2.5 Electrochemical Detection .................................................................................. 133 5.2.6 Validation of the Electrochemical Detection ....................................................... 133 5.3 Results and Discussion ......................................................................................... 133 5.3.1 Cyclic Voltammetry Detection of NAD + and NADH ........................................... 133 5.3.2 Calibration of the Microelectrode and Quantification of NAD + and NADH ....... 134 5.3.3 Detection Limit Enhancement and Detection of Low Volume Samples ............. 136 5.3.4 Dependence of pH .............................................................................................. 138 5.3.5 Observing Enzymatic Reaction by the Measuring the Consumption and Generation of NAD Cofactors ............................................................................. 138 5.3.6 Detection of the Electrochemical Generation of NADH ..................................... 140 5.3.7 Detection of NAD + Analogue Molecules Using Microelectrode ........................ 142 5.4 Chapter Five Conclusions .................................................................................... 143 x 5.5 Chapter Five References ..................................................................................... 144 Efficient and Selective Electrochemically-Driven Enzyme-Catalyzed Reduction of Carbon Dioxide to Formate using Formate Dehydrogenase .. and an Artificial Cofactor ............................................................................. 147 6.1 Background ......................................................................................................... 147 6.1.1 Replacing NADH with an Artificial Redox Cofactor, Methyl Viologen ................ 148 6.2 Materials and Methods ........................................................................................ 150 6.2.1 Materials .............................................................................................................. 150 6.2.2 Cyclic Voltammetry of Cofactor Solutions ........................................................... 150 6.2.3 Bulk Electrolysis of Cofactor Solutions to Prepare MV •+ ..................................... 151 6.2.4 Carbon Dioxide Reduction Using FDH and the Reduced Form of Methyl Viologen ............................................................................................................... 151 6.2.5 Formate Oxidation to Carbon Dioxide Using FDH and MV 2+ ............................. 151 6.2.6 NMR Analysis ....................................................................................................... 152 6.2.7 Computational Studies ........................................................................................ 152 6.3 Results .................................................................................................................. 152 6.3.1 Electrochemical Reversibility of NAD + / NADH and MV 2+ / MV •+ ......................... 152 6.3.2 Carbon Dioxide Reduction Using FDH and the Reduced Form of Methyl Viologen ............................................................................................................... 154 6.3.3 Methyl Viologen is a Unidirectional Cofactor for FDH ........................................ 155 6.3.3.1 Experimental Evidences ...................................................................................... 155 6.3.3.2 Binding Free Energy Recapitulates the Thermodynamic Feasibility of the ........ Bicarbonate Conversion by Methyl Viologen and FDH ....................................... 157 6.3.3.3 Relative Stability and Flexibility of the Reactants and Products in FDH ............. 159 6.3.3.4 Involvement of A Second MV •+ in the Catalytic Reaction ................................... 160 6.3.4 Continuous Reduction of Carbon Dioxide to Formate in Two-Chamber and Three-Chamber Electrolyzer Configurations ....................................................... 162 6.3.4.1 Formate Accumulation under Continuous Steady-State Carbon Dioxide Reduction ............................................................................................................. 162 6.3.5 Formate Yield ...................................................................................................... 165 6.4 Chapter Six Conclusions ...................................................................................... 166 6.5 Chapter Six References ....................................................................................... 168 Performance Advancements for the Continuous Bio-Electrochemical Generation, Conservation and Separation of Formate from Carbon Dioxide using FDH and Artificial Cofactor Systems ................................................... 171 7.1 Background ......................................................................................................... 171 7.2 Thermodynamics and Kinetics of Bio-catalytic Carbon Dioxide Reduction Reaction ............................................................................................................... 172 7.2.1 Mass Transport in Bio-catalytic Carbon Dioxide Reduction Reaction ................. 173 7.2.1.1 Rate of Diffusion in Electrochemical Systems ...................................................... 173 7.2.1.2 Ion-Exchange Membranes ................................................................................... 174 xi 7.2.2 Coulombic Efficiency in Bio-catalytic Carbon Dioxide Reduction Reaction ........ 175 7.2.3 Focus of Chapter Seven ...................................................................................... 175 7.3 Experimental ........................................................................................................ 176 7.3.1 Materials .............................................................................................................. 176 7.3.2 Studying MV and Formate Diffusion Across PEMs ............................................. 176 7.3.3 Uptake of MV into PEMs ..................................................................................... 176 7.3.4 Bipolar Membrane Fabrication ............................................................................ 177 7.3.5 Carbon Dioxide Reduction using FDH and the Reduced form of MV ................ 177 7.3.6 Screening of the Artificial Cofactors ................................................................... 177 7.4 Results and Discussion ......................................................................................... 178 7.4.1 Replacing PEM and AEM Separators with a Bipolar Membrane. ....................... 178 7.4.2 pH Balancing During Bio-Electrocatalytic Carbon Dioxide Reduction ................ 179 7.4.3 In-situ Separation of Formate for Recycling the Reaction Mixture ..................... 180 7.4.4 Screening PEMs Based on Diffusion .................................................................... 184 7.4.4.1 Effective Concentration of MV ............................................................................ 184 7.1.1.1 Studying Formate Diffusion Across PEMs ........................................................... 186 7.4.4.2 Optimal Membranes and Configuration for Achieving CO 2 Reduction with ...... Advanced Performance ....................................................................................... 189 7.4.5 Approaches to Improving the Coulombic Efficiency ........................................... 190 7.4.6 Step-wise Reaction of MV •+ with FDH and Carbon Dioxide ............................... 191 7.4.7 Screening Redox-Active Cofactors ...................................................................... 191 7.5 Chapter Seven Conclusions ................................................................................. 197 7.6 Chapter Seven References .................................................................................. 198 Future Directions ......................................................................................... 201 8.1 All-Iron Redox flow Cell ....................................................................................... 201 8.1.1 Electrolyte Recombination .................................................................................. 201 8.1.2 Lowering HER to 0% ............................................................................................ 203 8.1.3 Other Directions .................................................................................................. 205 8.2 Lithium-Sulfur System .......................................................................................... 205 8.3 Bio-catalytic Carbon Dioxide Reduction ............................................................. 206 8.3.1 Evolving the Mechanism of FDH based CO 2 reduction ...................................... 206 8.3.1.1 NMR Studies ........................................................................................................ 207 8.3.1.2 Cyclic Voltammetry .............................................................................................. 207 8.3.1.3 EPR Studies .......................................................................................................... 207 8.3.1.4 Experimental Techniques for Mechanism Elucidation ........................................ 208 8.3.2 Ex-Situ Separation of Formate from the CO 2 Reduction Reaction Mixture ........ 209 8.3.3 Cofactor Immobilization ...................................................................................... 210 8.4 Chapter Eight References ................................................................................... 211 Publications and Presentations ........................................................................................... I Appendix ............................................................................................................................ III xii Appendix A………… .................................................................................................................... III A1. Calculation of the equilibrium potential shift at pH 3 due to iron (II) ascorbate complex formation…….. ............................................................................................................................. III A2. Calculation of the bulk and electrode surface pH ................................................................. IV A3. Calculation of the diffusion layer thickness (δ) ...................................................................... VI A4: Ascorbic acid dissociation with pH in 2 M Ammonium chloride medium (■:HL - , ▲H 2L) ...... VII A5: Tafel parameters for HER and iron deposition at different temperatures. ........................... VIII Appendix B…… ........................................................................................................................... IX B1: Activity of FeOH + , Fe 2+ , Cl - and Fe-Cl complex calculation ................................................... IX B2: Evan Diagram Construction ................................................................................................... XI B3: References used in Appendix B ............................................................................................. XII Appendix C….. ............................................................................................................................. XIII C1: Capacity fade with initial discharging .................................................................................... XIII C2: Significantly higher impedance with less electrolyte ............................................................. XIV C3: Discharging curve of a Li-S cell with a cathode of CNT ........................................................ XVI C4: Discharging capacity recovery by discharging at lower rates ............................................... XVII Appendix D……. .......................................................................................................................... XVIII D1: Thermodynamics of FDH catalyzed CO 2 reduction ............................................................... XVIII D2: Molecular dynamics studies ................................................................................................... XVIII D3: Electrochemical reduction of NAD + using carbon electrodes ............................................... XXIII D4: Detection of MV di-cation and reduced form during bulk electrolysis ................................. XXV D5: Steady-state reduction of CO 2 using an electrochemical cell assembled with an AEM ....... XXVI D6: Formate oxidation using FDH and MV di-cation ................................................................... XXVII D7: Steady-state reduction CO 2 reduction .................................................................................. XXVIII D8: Cross over of MV across proton exchange membrane ......................................................... XXIX D9: Rate constant (k cat) calculation ............................................................................................... XXX D10: References used in Appendix D .......................................................................................... XXXI Appendix E…… ............................................................................................................................ XXXIII E1: Tokuyama anion exchange membrane properties ................................................................ XXXIII E2: Different viologen molecules studied in photocatalytic CO 2 reduction using FDH…………XXXIV E3: 1 H NMR of 2,7-AQDS in D 2O and MES buffer ....................................................................... XXXV E4: References used in Appendix E……………………………………………………………………XXXVI xiii List of Tables Table 1-1: All-inorganic redox flow cells ...................................................................................................... 5 Table 1-2: All-organic redox flow cells ......................................................................................................... 6 Table 1-3: Inorganic-organic mixed redox flow cells ................................................................................... 7 Table 1-4: Standard reduction potentials of CO 2 reduction at pH 7 61, 66 .................................................. 20 Table 1-5: Metal catalysts used in carbon dioxide reduction .................................................................... 21 Table 1-6: Carbon dioxide reduction using metal-independent formate dehydrogenase enzyme from Candida boidinii ......................................................................................................................................... 25 Table 2-1: Predicted performance of the all-iron flow battery with coulombic efficiency losses due to HER. ............................................................................................................................................................ 41 Table 2-2: Electrode potential and HER overpotential at 20 mA/ cm 2 at different values of pH. ............. 47 Table 2-3: Effect of ammonium chloride additives for the solubility and dissociation .............................. 55 Table 2-4: Effect of temperature on electrochemical impedance parameters of iron electrode under potentiostatic conditions at -0.85 V vs. Ag/AgCl. ..................................................................................... 63 Table 3-1: HER overpotential and Tafel slopes of HER on different metals 2, 3, 7 ........................................ 71 Table 3-2: Experiment Conditions ............................................................................................................. 73 Table 3-3: Tafel parameters of iron deposition and hydrogen evolution reactions .................................. 83 Table 3-4: EIS data of iron deposition at -0.7 V vs. NHE at different ion (II) activities .............................. 85 Table 4-1: Single-layer mixed conduction membrane preparation ......................................................... 101 Table 4-2: Multi-layered mixed conduction membrane preparation ....................................................... 101 Table 4-3: Properties of Acetylene Black and Ketjen Black carbon particles .......................................... 110 Table 4-4: Cathode properties of different carbon supports .................................................................. 118 Table 4-5: Cathode properties of mixed carbon supports with MCM .................................................... 121 Table 4-6: Cell Parameters 2 ...................................................................................................................... 123 Table 5-1: Diffusion coefficients of NAD species in three different buffer solutions .............................. 136 Table 5-2: Rate of different dehydrogenase enzymatic reactions ........................................................... 139 Table 6-1: Electrochemical properties of MV in different buffer solutions .............................................. 153 Table 6-2: Yield of the formate in the three-compartment cell shown in Figure 6-6. ............................. 166 Table 7-1: CO 2 reduction using different membranes and configurations (40 mM, MV 2+ ) ..................... 179 Table 7-2: CO 2 reduction process combined with an in-situ formate separation ................................... 184 Table 7-3: Incorporation of MV di-cation into different PEMs ................................................................. 186 xiv Table 7-4: Formate cross over via different PEMs ................................................................................... 188 Table 7-5: Coulombic Efficiency of Methyl Viologen Reduction ............................................................. 190 Table 7-6: Electrochemical properties of candidate artificial cofactors .................................................. 196 xv List of Figures Figure 1-1: Basic construction of a redox flow cell unit ............................................................................... 2 Figure 1-2: Reduction potential stairway of a widely studied redox-active molecules ............................... 3 Figure 1-3: Schematic diagram of all-iron redox flow cell ........................................................................... 8 Figure 1-4: Tafel plot of iron deposition and dissolution .......................................................................... 11 Figure 1-5: Nernst diffusion model ............................................................................................................ 12 Figure 1-6: Pourbaix diagram of iron at ionic concentrations of 1.0 mM 51 (« represents the potential window over which the negative electrode operates.) .............................................................................. 15 Figure 1-7: Schematic of the electrochemical processes in a lithium-sulfur battery (C = charge, D = discharge) from reference. 54 ....................................................................................................................... 17 Figure 1-8: FDH catalyzed formate oxidation ............................................................................................ 23 Figure 1-9: Schematic representation of the hypothetical substrate binding of FDH 100 ........................... 26 Figure 2-1: Effects of the hydrogen evolution during the operation of all-iron redox flow battery ......... 40 Figure 2-2: Experimental set-up. ................................................................................................................ 43 Figure 2-3: (a) Effect of electrolyte pH on the coulombic efficiency of iron electrodeposition and electro dissolution. (b) Effect of electrolyte pH on electrode potential of iron electrode during electrodeposition; bath conditions (iron (II) chloride (3.0 M), ammonium chloride (2.0 M), ascorbic acid (0.3 M) at temperature = 25 ◦ C). ......................................................................................................... 46 Figure 2-4: (a) Potentiodynamic polarization studies of iron deposition from a solution of iron(II) chloride(3.0 M), ammonium chloride (2.0 M), T = 25 ◦ C in the presence and absence of ascorbic acid (0.3 M) at various values of pH as shown (AA = Ascorbic acid) and (b) Overpotential for HER on iron at pH 0 and 3 in the presence or absence of ascorbic acid (0.3 M) and ammonium chloride (2.0 M). ... 49 Figure 2-5: Experimental (♦,) and simulated (Ú,+) near-surface pH and experimental bulk pH (▲,¢) deviation during potentiodynamic polarization of metal electrodeposition using an electrolyte composed with iron (II) chloride (3.25 M) and ammonium chloride (2.0 M). ............................................. 53 Figure 2-6: Deviation of coulombic efficiency with stirring rate at constant current densities (¯−50 and ¢−100 mA/ cm 2 ) using an electrolyte composed with iron (II) chloride (3.0 M) and ammonium chloride (2.0 M) at pH = 2.00. ................................................................................................. 54 Figure 2-7: Apparent hydrolysis constant of iron (II) solutions ................................................................. 56 Figure 2-8: (a) Experimental near-surface pH (◊,) and bulk pH (●,▲) deviation during potentiodynamic polarization of metal electrodeposition using an electrolyte composed with iron (II) xvi chloride (3.25 M), ammonium chloride (2.0 M) and ascorbic acid (0.3 M). (b) Representation of the equilibrium reactions near the electrode surface during iron electrodeposition process (Ascorbic acid (H 2L) and hydrogen adsorption on iron and HER on iron are not illustrated). ........................................... 57 Figure 2-9: (a) Potentiodynamic studies of HER on iron(x) and cadmium(-). (b) Effect of cadmium chloride additives on the coulombic efficiency of iron electrodeposition and electro dissolution. Bath conditions (iron (II) chloride (3.0 M), ammonium chloride (2.0 M), ascorbic acid (0.3 M) and cadmium chloride (0.0 mM(■), 20 mM(▲) at pH = 3.0 and temperature = 25 ◦ C). ................................................... 59 Figure 2-10: Effect of temperature on coulombic efficiency of iron electrode- position and electro dissolution. Bath conditions (iron (II) chloride (3.0 M), ammonium chloride (2.0 M) and ascorbic acid (0.3 M) at pH = 3.0). Efficiency values indicated for each point. ............................................................... 61 Figure 2-11: Overlaid Nyquist diagrams at−0.85V vs. Ag/AgCl at 25-60 ◦ C at stirring (x) and no stirring (◊) conditions. Bath conditions (iron (II) chloride (3.0M), ammonium chloride (2.0 M) and ascorbic acid (0.3 M) at pH = 3.0). Temperatures indicated on the charts. (Rs -ohmic resistance and Rp -polarization resistance). .................................................................................................................................................. 62 Figure 3-1: Volcano plot of Hydrogen Evolution on metals 5 ..................................................................... 70 Figure 3-2:Electrolyte properties at different iron (II) concentrations A: Conductivity, B: pH and C: ionic strength (l -0 M and q- 2 M Ammonium Chloride electrolyte) ...................................................... 74 Figure 3-3: (A) Activity of different species in the electrolyte and (B) Activity coefficient of Fe 2+ in the electrolyte containing iron(II) chloride (0.6 to 3 M) and ammonium chloride (0 or 2 M);(n-FeOH + = H + , l-Fe 2+ ,ê-Cl- and t- Fe-Cl complex) .................................................................................................... 75 Figure 3-4: Relationship of coulombic efficiency with A: Current density and electrode potential and B: State of charge of the anolyte from 0 to 80% (in 3-0.6 M FeCl 2 and 2 M NH 4Cl). ................................ 77 Figure 3-5: Polarization curves of hydrogen evolution on iron at 10 mV/s scan rate (in 2M in ammonium chloride solution) ..................................................................................................................... 79 Figure 3-6: Iron deposition and dissolution with cysteine additives .......................................................... 80 Figure 3-7: Tafel lines of A: Iron deposition and B: Hydrogen evolution reactions on iron from 0.6 to 3 M iron (II) chloride (q-0.6, l-1, ¢-2, t-2.5 and ê-3M) and 2 M ammonium chloride. ......................... 82 Figure 3-8: A and B: EIS of iron deposition reaction in 0.6 and 3.0 M iron (II) chloride solution, C: Experimental (red) and fitted (green) EIS data with a proposed equivalence circuit (D) of iron deposition process in 0.6 M iron (II) chloride solution (Area of the electrode is 23 cm 2 ). ......................... 84 xvii Figure 3-9: Nature of the electrode surface and (A) non-catalytic and (B) catalytic mechanisms of iron deposition (d -diffusion layer thickness, q -coverage and In -Inhibitors) .................................................... 87 Figure 3-10: Representative Evan Diagrams of iron deposition and hydrogen evolution in iron (II) chloride solutions in higher concentrations from 1-3 M with A: HER at higher potentials and B: HER at higher potentials and near E corr (I HL=limiting current density of HER from proton) and C: Potentiodynamic curves of iron deposition using 0.6 to 3 M iron (II) chloride (p-0.6, l-1, ¢-2, t-2.5 and ê-3M) and 2 M ammonium chloride. ................................................................................................. 89 Figure 4-1: Lithium ion transport via (A) and (B): non-porous, (C): porous and (D): porous and non- porous layered MCMs ................................................................................................................................ 97 Figure 4-2: Effect of discharging rate for Li 2S deposition .......................................................................... 99 Figure 4-3: Electrical equivalent circuit for Li-S system ........................................................................... 100 Figure 4-4 (A): Discharging curves and (B): EIS of Li-S with KB cathodes using different layered membranes, Cell parameters: 15 µm thick cathode, 2 cm 2 of cell area, KB carbon with 63.3% S, 100 µL of electrolyte volume (extra 40 µL for ABA and BBB type MCM) and at C/64 rate. ................... 103 Figure 4-5: (A,C,F and H ) SEM images of pressed, (B,D,G and I) SEM images of not pressed and (E) Visible appearance of 15 µm thick single layer of MCM (A-E are thickness of 15 µm, slurry prepared using mechanical stirring and potassium hydroxide treated to remove the aluminum support and F-I are thickness of 75 µm, slurry prepared using ball milling and no potassium hydroxide treatment). ..... 104 Figure 4-6: Coulombic efficiency and specific discharge capacity of Li-S cells with MCM (●E101) and with no MCM (▼E157). Cell parameters: 15 µm thick cathode, 2 cm 2 of cell area, KB carbon with 63.3% S, 100 µL of electrolyte volume and at C/64 rate. ........................................................................ 105 Figure 4-7: Initial Specific Capacity vs. EIS (At the frequency of 0.1 Hz) of different Li-S cell configurations before cycling ................................................................................................................... 107 Figure 4-8: Effect of discharging capacity and features with lower volume of electrolyte; Cell parameters: 15 µm thick cathode, 2 cm 2 of cell area, KB carbon with 63.3% S, 100 µL of electrolyte volume and at C/64 rate. ......................................................................................................................... 109 Figure 4-9: EIS of different LI-S cell constructions with (A) Ketjen Black cathode, (B) Mixed carbon (Acetylene Black: Ketjen Black, 1:1) cathodes and (C) Acetylene Black cathodes. ................................. 111 Figure 4-10: 3-D structures of carbon supported -sulfur cathodes (A) KB, (B) AB and (C) mixed carbon KB:AB and 2D structure of a single pore created with carbon particles (D) KB, (E) AB and (F) mixed carbon KB:AB (C - ● and ❍, S -n, PVDF -g and Li + -n). ........................................................................ 112 xviii Figure 4-11: Discharging rate capability of (A and C) Ketjen Black cathodes and (B and D) Acetylene Black cathodes. ........................................................................................................................................ 118 Figure 4-12: Specific capacity at second plateau vs. Overpotential in Ketjen Black (p) Acetylene Black cathodes (¨). ............................................................................................................................................ 119 Figure 4-13: Analysis of kinetics of S 4 2- to Li 2S conversion and utilization in Ketjen Black (p) Acetylene Black cathodes (¨). ................................................................................................................................... 119 Figure 4-14: Discharging rate capability with different Li-S cell construction with layered MCM and (A) KB cathodes and (B) AB cathodes. ..................................................................................................... 120 Figure 4-15: Analysis of kinetics of S 4 2- to Li 2S conversion and utilization in Ketjen Black (p) and Acetylene Black cathodes (¨) with layered MCM. ................................................................................... 120 Figure 4-16: Discharging rate capability with different Li-S cell construction with mixed carbon cathodes with (A) layered MCM and (B) with single MCM ...................................................................... 122 Figure 4-17: Analysis of kinetics of S 4 2- to Li 2S conversion with MCM single layer (¢) and ABA layered (●). ............................................................................................................................................................ 123 Figure 5-1: Microelectrode detection of NAD + / NADH to investigate Bio-catalytic redox reactions (R=adenine dinucleotide) ......................................................................................................................... 131 Figure 5-2: Cyclic voltammograms of equimolar (5 mM) NAD+ and NADH in 0.5 M phosphate buffer on A: Glassy carbon micro disk electrode and B Carbon fiber microelectrode (at scan rate of 25 mV/ s) .................................................................................................................................................................. 134 Figure 5-3: Calibration curve of NAD + and NADH detection on microelectrode (0.25 to 20 mM) ........ 135 Figure 5-4: Detection of A: NAD+ and B: NADH (5 mM) in MES, Phosphate buffer and Tris buffer at pH 7.00 ........................................................................................................................................................... 136 Figure 5-5: Relative current response variation with increasing the number of electrodes .................... 137 Figure 5-6: Detection of 7.5 mM NAD + in 0.5 M Phosphate Buffer ........................................................ 137 Figure 5-7: Current response of NAD + (5 mM) reduction at different pH in the region of 6.5-10.0 in 0.5 M phosphate buffer solution (pH adjusted using NaOH) .................................................................. 138 Figure 5-8: FDH enzyme catalyzed formate oxidation (100 mM Sodium formate and 5 mM NAD + ) ..... 139 Figure 5-9 A: Microelectrode Detection of NAD + electrolyzed sample B: H-NMR peaks of NADH from NAD + in the electrolyzed sample on platinum electrode and C: Correlation of H-NMR analysis and microelectrode analysis. ........................................................................................................................... 140 xix Figure 5-10 A: Microelectrode Detection of NAD + electrolyzed sample (electrolysis using graphite felt) and B: Cyclic voltammogram of NAD + and NADH (5 mM) using graphite rod in 0.1 M phosphate buffer at pH 7. .......................................................................................................................................... 142 Figure 5-11: Cyclic voltammogram of NAD + and NAM in phosphate buffer solution. ........................... 142 Figure 6-1: Enzyme-catalyzed CO 2 reduction using metal independent FDH and MV cofactor ............. 148 Figure 6-2: Cyclic Voltammograms of MV in A: Phosphate buffer (⎯), B: MES buffer (--) and C: Tris buffer (…) at pH 6.6 ................................................................................................................................. 153 Figure 6-3: Electrochemical formate generation: (A) Reactor configuration, (B) Experimental set-up and (C) Rate of formate generation for various concentrations of MV •+ , and the continuous current density for producing MV •+ at an applied potential of -0.44 V vs. NHE using 3-40 mM MV 2+ , 1.2 µM FDH, 15 mL buffer solution in each chamber and Tokuyama A901 AEM. .............................................. 155 Figure 6-4: Effect of 20 mM MV 2+ on the concentration of sodium formate (11 to 44 mM) in the presence of 4.3 µM FDH. ......................................................................................................................... 156 Figure 6-5: Orientation of bicarbonate ion in the representative conformation from the highest populated cluster in the MD ensemble in the presence of (A) hydronium ions and MV •+ (green), and (C) NADH (green); major active site residues shown in magenta; (B, D) probability density of bicarbonate position as function of distance with R258 and H311 as sampled during the MD simulations of (B) MV •+ bound FDH and (D) NADH bound FDH; for each residue, the minimum distance of the oxygen atoms in bicarbonate with the sidechain polar nitrogen atoms was measured; (E) Difference in binding free energy between the reactants and products of bicarbonate reduction in FDH in the presence of MV •+ and NADH. ............................................................................................ 158 Figure 6-6: Comparison of the reactants and products orientation in the FDH active site in MV facilitated bicarbonate reduction; (A) orientation of bicarbonate and hydronium ions in presence of MV •+ (green); residues making hydrogen bonds with the reactants (blue); (B) Orientation of the products formate and hydroxyl ions in presence of MV 2+ (green); residues that formed hydrogen bonds with the reactants in the MV •+ bound FDH (blue); these hydrogen bonds are disrupted upon product formation, as shown by the increased distances with the formate. ........................................... 160 Figure 6-7: (A) Cross section of FDH showing the binding site of two bound MV •+ molecules (green); the aromatic residues making contact with the second MV •+ in the secondary pocket (magenta); the major active site residues involved in catalysis (blue); reactants bicarbonate and hydronium (green); xx (B) Magnified view of the FDH cross section near the active and secondary sites and (C) Possible electron transfer pathway from the second MV •+ molecule. ................................................................... 161 Figure 6-8: Electrochemical formate generation: (A) Illustration of the two-chamber electrochemical reactor and (B) Results of continuous accumulation of formate from electrochemical reduction of carbon dioxide in the two-chamber reactor at an applied potential of -0.44 V vs. NHE using 40 mM MV 2+ , 2.6 µM FDH, 7.5 mL buffer solution in each chamber and Nafion® 117 PEM. ............................. 163 Figure 6-9: Electrochemical formate generation (A) Illustration and (B) Experimental setup showing diffusion of MV across PEM and (C) Steady-state CO 2 reduction in the novel three chamber reactor at an applied potential of -0.44 V vs. NHE using 40 mM MV 2+ , 2.4 µM FDH, 7.5 mL buffer solution in each chamber, Tokuyama A901 AEM and Nafion® 117 PEM. ................................................................ 164 Figure 7-1: Factors affecting in a bio-catalytic pathway of formate generation from carbon dioxide reduction .................................................................................................................................................. 171 Figure 7-2: Different types of ion-exchange membranes A: Proton-exchange membrane (PEM), B: Anion-exchange membrane (AEM) and C: Bipolar membrane (BPM) used in CO 2 reduction ................ 174 Figure 7-3: CO 2 reduction using a bipolar membrane (reactor configuration B) .................................... 178 Figure 7-4: Proposed CO 2 reduction scheme (configuration E) for achieving constant pH and formate separation (a modified design from reactor configuration A) ................................................................. 182 Figure 7-5: CO 2 reduction A: scheme and B: experimental set-up (configuration F) for achieving constant pH and formate separation (a modified design from reactor configuration B) ........................ 183 Figure 7-6: A: Bio-catalytic CO 2 reduction using in-situ formate separation using the reactor configuration G and B: Formate accumulation in the separation chamber with time ............................. 183 Figure 7-7: Methyl viologen diffusion across different PEMs .................................................................. 185 Figure 7-8: Methyl viologen concentration in the enzymatic chamber vs. electrolyzing time ................ 186 Figure 7-9: Formate Diffusion via different PEMs .................................................................................... 188 Figure 7-10: In-house created bipolar membrane by combining Tokuyama A-901 and Nafion Ò 325 membranes ............................................................................................................................................... 189 Figure 7-11: Schematic diagram of electrochemical reduction of MV using an anion exchange membrane and FDH catalyzed CO 2 reduction ........................................................................................ 191 Figure 7-12: Cyclic voltammogram of candidate artificial cofactors in phosphate buffer at pH 7 A: Different viologen molecules, B: Bio-compatible organic molecules and C: 2,7-AQDS ......................... 192 Figure 7-13: Cross-over testing of neutral red across an AEM ................................................................ 194 xxi Figure 8-1: A single chamber reactor for H 2 and iron(III) recombination A: schematic diagram with all iron redox flow cell, B: Experimental setup of the recombination reactor, C: Reaction mixture before recombination, D: Reaction mixture after recombination and E: Teflon treated, Platinum deposited Reticulated Vitreous Carbon Foam as the catalyst. ................................................................................. 202 Figure 8-2: A two-chamber reactor for H 2 and iron (III) recombination with all-iron redox flow cell A: schematic diagram and B: Experimental setup of the two-chamber reactor .......................................... 203 Figure 8-3: Proposed mechanisms of bicarbonate reduction into formate using Methyl viologen radical cations. A: Via a bicarbonate radical intermediate and B: Via no bicarbonate radical intermediates ........................................................................................................................................... 208 xxii List of Abbreviation and Symbols AA Ascorbic Acid AB Acetylene Black AEM Anion Exchange Membrane AEV Amino Ethyl Viologen AQDS Anthraquinone Disulfonic Acid BPM Bipolar Membrane CbFDH Candida boidinii Formate Dehydrogense CE Coulombic Efficiency CFME Carbon Fiber Microelectrode CNT Carbon Nano Tube CV Cyclic Voltammogram ECR Electrochemical Carbon Dioxide Reduction EDTA Ethylenediaminetetraacetic acid EES Electrochemical Energy Storage EIS Electrochemical Impedance Spectroscopy EPR Electron Paramagnetic Resonance FDH Formate Dehydrogenase GCMD Glassy Carbon Micro Disk HER Hydrogen Evolution Reaction KB Ketjen Black LCO Lithium Cobalt Oxide Li-S Lithium Sulfur MCM Mixed Conduction Membrane MES 2-(N-Morpholino) Ethane Sulfonic Acid MV Methyl Viologen NAD Nicotinamide Adenine Dinucleotide NADP Nicotinamide Adenine Dinucleotide Phosphate xxiii NAM Nicotinamide Mononucleotide NDL Nernst Diffusion Layer NHE Normal Hydrogen Electrode NMR Nuclear Magnetic Resonance OER Oxygen Evolution Reaction PEM Proton Exchange Membrane PS Poly Sulfide RFB Redox Flow Battery Tri-HCl Tris (Hydroxymethyl)Aminomethane Hydrochloride TsFDH Thiobacillus Formate Dehydrogenase UV/ Vis Ultra Violet/ Visible VE Voltage Efficiency xxiv Abstract This thesis focuses on two main topics in electrochemistry: (1) studying electrochemical energy storage systems and (2) developing electrochemical energy conversion pathways by the reduction of carbon dioxide to useful organic products. The first half of Chapter One provides an introduction to electrochemical energy storage technologies. Different categories of aqueous redox flow battery systems suitable for large-scale energy storage are discussed in detail. At the end of the first half of the introduction, a brief introduction to the lithium-sulfur battery is presented. The rest of Chapter One discusses the electrochemical carbon dioxide reduction. Among the different redox flow battery technologies, the all-iron redox flow battery is an attractive solution for large-scale energy storage because of the low-cost and eco-friendliness of iron-based materials. A major challenge for realizing a continuously operable all-iron redox flow battery is the parasitic evolution of hydrogen at the iron electrode during the charging step. Results and discussion presented in Chapter Two provide insights for minimizing hydrogen evolution in the all-iron battery system. At a given bulk concentration of iron (II), pH of the electrolyte, temperature, concentration of additives, and current density are recognized as key factors affecting the coulombic efficiency of the battery. Elevation of pH near the electrode surface during electrodeposition plays a significant role in hindering hydrogen evolution. Thus, electrolyte flow rates drastically influence the coulombic efficiency of the all-iron redox flow battery. By operating at 60 ◦ C and a pH of 3 with ascorbic acid and ammonium chloride, we could achieve a coulombic efficiency of 98%. This value of coulombic efficiency is among the highest values reported for the iron electrode of the all-iron flow battery. Chapter Three discusses the kinetics of electrodeposition of iron and the evolution of hydrogen during the charging of the all-iron redox flow cell. We could verify that the kinetics of iron deposition improved with the higher activity of iron (II), lowering the rate of the hydrogen evolution reaction (HER). Further, the increase of the pH near electrode surface results in increased coulombic efficiency at the higher charging current densities. We show that from 0 to 80% state-of-charge and charging at 30 and 40 mA/cm 2 , a steady coulombic efficiency of 98±1% xxv could be observed due to improved kinetics at the higher concentration of iron (II) or increased surface pH at raised current densities. Chapter Four delivers a critical analysis of studies of the lithium-sulfur battery. The lithium-sulfur battery is a promising technology that has the prospect of doubling the energy density of lithium-ion batteries due to the high specific capacity of the sulfur electrode. Further, sulfur is also an expensive material and has the potential to lead to a cost-effective battery. In the present study, we attempted to understand the use of the USC invented mixed-conduction membrane to address the issue of the polysulfide shuttle. Initial cycling data showed that about 20% improved coulombic efficiency and 11 to 16 % improved material utilization with this novel membrane. Further, by using different layers and different porosity of mixed conduction membranes, we attempted to validate the intercalation process of lithium ion to facilitate charge transfer kinetics of the lithium-sulfur cell. Further, the properties of Ketjen Black as a cathode matrix material for improving polysulfide retention and increasing material utilization are discussed. And the results presented in Chapter Four with a mixture of carbon materials point to approaches for achieving better utilization with improved rates of charge and discharge using cathodes based on Ketjen Black. Chapters Five, Six and Seven discuss the bio-catalytic pathways for the reduction of carbon dioxide to formate using the enzyme formate dehydrogenase and redox co-factors. Electrochemical analysis of NAD + /NADH and measurement of enzyme activity is challenging, and Chapter Five presents a simple electrochemical detection of NAD + / NADH using an unmodified carbon fiber microelectrode as an accurate, convenient and low-cost detection technique for studying enzymatic reactions. Diffusion-limited current measurements of reduction and oxidation of NAD + and NADH enables investigation of NAD-dependent enzyme activity and indirect analysis of concentrations of substances like formate and ethanol which are substrates for NAD-dependent enzymes. Further, this method was validated for the in-situ detection of electrochemically-generated NADH. This fast, in-situ determination of NAD + and NADH can be used in bio-electrochemical applications and be further developed to a biosensor. Chapter Six describes results on the bio-catalytic reduction of carbon dioxide using commercially- available formate dehydrogenase enzyme. We demonstrate an efficient and xxvi continuous conversion of carbon dioxide to formate using formate dehydrogenase enzyme derived from Candida boidinii yeast and an electrochemically-generated artificial co-factor, methyl viologen radical cation. Continuous regeneration of this artificial cofactor could be achieved at -0.44 V vs. the Normal Hydrogen Electrode (NHE) leading to a significantly lower overpotential for the reduction of carbon dioxide to formate. Thus, the electrochemically- regenerated methyl viologen radical cation works efficiently as a cofactor to support the proton- coupled electron transfer to the carbon dioxide molecule. With a special electrochemical reactor assembled with three chambers separated by anion and cation exchange membranes, we demonstrate the generation and accumulation of formate and the evolution of molecular oxygen. The proton-exchange membrane allows for the accumulation of formate by preventing loss to electrochemical re-oxidation at the oxygen evolution electrode while the anion- conducting membrane improves the utilization of the cofactor by suppressing its loss by cross- over of the oxygen electrode. This novel three-chamber reactor has been shown as a proof-of- concept for efficient and continuous conversion of carbon dioxide into formate. In Chapter Seven, we discuss various factors in improving the efficiency of the bio-catalytic carbon dioxide reduction including the role of thermodynamics, kinetics, mass transport and pH balance of the system, and in-situ formate separation on the coulombic efficiency. We also discuss the performance improvements resulting from various cell configurations. Chapter Eight summarizes the future directions for research on all of the projects discussed in the thesis. 1 Introduction 1.1 Electrochemical Energy Storage Systems The demand for energy is growing worldwide. The steady growth of the global population places an increasing demand for energy from fossil fuels resources. More than 80% of the world’s energy demand is supported by using carbon-based resources. 1 The burning of hydrocarbons results in greenhouse gas emissions with environmental consequences such as global warming and air pollution. For these reasons, renewable energy has become popular in recent decades. However, the availability of renewable energy is variable and even unpredictable. Therefore, reliable, low-cost, durable and eco-friendly methods of energy storage are essential to provide a continuous energy supply from solar photovoltaic arrays and wind turbines. A number of energy storage systems are available that use mechanical, chemical, electrical and electrochemical methods. Among these technologies, the rechargeable batteries present many advantages. In rechargeable batteries, reversible electrochemical reactions store and deliver energy. Lead-acid, nickel-cadmium, nickel-metal hydride and lithium-ion batteries are well-known rechargeable systems. Lithium-ion and nickel metal hydride batteries are used in all-electric and hybrid electric vehicles because of their high energy density. 2, 3 Lead-acid batteries are widely used due to they are low-cost. However, these inexpensive batteries low energy density and are a toxic hazard. Further, improvements in battery technologies are needed to store energy at a large-scale from renewable energy sources. A variety of other technologies including sodium-sulfur, metal-air, lithium-ion, lithium-sulfur, redox flow batteries have the potential to meet the demands of grid scale energy storage. 4 According to the U.S. Department of Energy, for energy storage systems to be economically viable, a target cost of US$100 per kWh must be achieved by 2030. 5 Of the various electrochemical energy storage technologies for large-scale applications, redox flow batteries are particularly promising because of their potentially low-cost and scalability. 6, 7 Thus, researchers worldwide are studying redox flow batteries based on metal ions and organic redox molecules, with aqueous and non- aqueous electrolytes, for grid-scale energy storage applications. 8-13 The next generation of 2 lithium batteries based on lithium and sulfur are also attractive due to the high specific capacity, high theoretical energy density, and low cost of sulfur. 1.2 Redox Flow Batteries Redox flow batteries are excellent candidates for large-scale energy storage because they are safe, scalable, and have low maintenance costs. A unique property of redox flow batteries is that the power output is not dependent on the amount of energy stored. Energy is stored in an electrolyte in external tanks and electrolyte, while power is delivered via a stack of cells. The size of the tanks and the stack can be independently varied. Therefore, to obtain more energy storage capability, the amount of redox-active electrolyte is increased. To deliver more power, the number of cells or the area of the cells in a stack is increased. A schematic diagram of the repeating unit of a redox flow cell is presented in Figure 1-1. Electrodes inside the cell housing and connected to current collectors. The cell is assembled with gaskets to prevent leakage of the fluids. In more recent designs redox flow cells, ion- exchange membranes serve as separators between the two electrodes preventing mixing of the negative and positive electrolytes. These membrane separators facilitate the transport of cations or anions during cell operation. Figure 1-1: Basic construction of a redox flow cell unit 3 Even though redox flow batteries have attractive features for the grid-scale application, improvements are required for commercial deployment. 14 In the next section, I explain some of the important requirements for redox flow cells needed for large-scale applications. 1.2.1 Screening of Redox Couples Figure 1-2: Reduction potential stairway of a widely studied redox-active molecules In principle, any reversible redox couple can be used to build a redox flow cell. An ideal candidate for a high-energy redox flow cell must have reversible redox reactions involving multiple electrons and high values of solubility. The larger the difference between the standard reduction potentials for the positive and negative electrode, the higher would be the cell voltage. Even though a large number of redox couples have been tested for redox flow cells in aqueous solutions, their practical use has been limited by their poor solubility and undesirable side reactions. Low solubility results in low power output and energy density, while side reactions lower the coulombic efficiency. Solubility can be increased by the addition of suitable functional groups. The standard reduction potentials and reactivity can also be altered by altering the functional groups. Thus, molecules may be engineered for use in redox flow 4 batteries. However, the values of standard reduction potentials for candidate redox couples for a flow battery operating in aqueous media are limited by the hydrogen and oxygen evolution reactions (Figure 1-2). Further, for large scale applications, the redox materials must be cost- effective. There is a numerous redox flow battery systems based on metal ions and organic redox molecules using both aqueous and non-aqueous electrolytes which are being studied widely for grid-scale energy storage applications. 8-11, 15 In sections 1.1.2.1 to 1.1.2.3, three different categories of aqueous redox flow systems are discussed. 1.2.1.1 Aqueous Inorganic Redox Flow Cells Zinc/chlorine battery system is among the first redox flow batteries that were developed (1960). 4 Vanadium and Iron / titanium are also some of the initial developments of redox flow batteries. 7, 16 In addition, iron or vanadium based positive side redox materials are studied coupling with different redox chemistries such as chromium and bromine at the negative side for developing a variety of redox flow battery technologies. Table 1-1 provides a list of different redox couples used in inorganic flow battery systems. When the metallic species are deposited a solid form these systems are classified as ‘hybrid’ redox flow batteries. The all-iron, zinc- bromine, and soluble lead-based redox flow systems are examples of hybrid systems. Of the various kinds of inorganic redox flow batteries, the all-vanadium system is the only one that has been developed to a commercial scale. 16-18 The vanadium system has been feasible because of the kinetic reversibility of redox couples and high solubility of vanadium ions in aqueous electrolytes. 5 Table 1-1: All-inorganic redox flow cells Negative side Positive side Redox pairs Standard redox potential, V vs. NHE Redox pairs Standard redox potential, V vs. NHE V 3+ / V 2+19 −0.26 VO 2 + / VO 2+ 19 0.99 Cr 3+ / Cr 2+ −0.41 Ce 4+ / Ce 3+ 1.74 Zn 2+ / Zn 20 −0.76 Mn 3+ / Mn 2+ 1.54 Ti 3+ / TiO 2+21 −0.06 Br − / Br 2Cl 22 1.07 S 2 2- / S 4 2- 23 −0.45 Fe 3+ / Fe 2+24 0.77 Pb 2+ / Pb 25, 26 -0.13 PbO 2/ Pb 2+ 25, 26 1.49 Zn(OH) 4 2- / Zn −1.21 Cl - / Cl 2 20 1.36 Cr(EDTA) 2- / Cr(EDTA) - -0.96 Br - /Br 3-23 1.05 Sn 4+ /Sn 2+ 0.15 Fe (CN) 6 4- / Fe (CN) 6 3- 0.36 Fe 2+ / Fe (0)24 -0.44 NpO 2 2+ /NpO 2 + 27 1.14 Cr 2O 7 2−/ Cr 3+ 1.23 Ce 2O 6+ /Ce 3+ 1.87 Co 3+ /Co 2+ 1.95 Cu 2+ /Cu 0.34 1.2.1.2 Aqueous All-Organic Redox Flow Cells These flow battery systems use two different organic redox-active molecules for positive and negative electrolytes. This concept is relatively recent. There are a number of recent studies of aqueous all-organic redox flow cells in acidic, neutral and basic media. Organic molecules have been designed and synthesized for increasing solubility in water and increasing energy storage capability. Modifications of the structure have also improved kinetic reversibility. To achieving good solubility, the focus is on molecules with low molar mass. To this end quinones, stable nitroxide radicals and N, N’-disubstituted 4,4’-bipyridine (viologen) derivatives as well as polymers bearing these functionalities have been studied. These molecules have been used not 6 only in aqueous but also in non-aqueous media. Examples of different organic redox couples that are used in all-organic redox flow cells are listed in Table 1-2. Table 1-2: All-organic redox flow cells Negative side Positive side Anthraquinone 10, 28, 29 Benzaquinone 10, 28 Quinoxiline 30 TEMPO 31, 32 Methyl viologen 31, 33 Hydroxyl TEMPO 33 Polymer-based viologen 31 Phenazine 34 Fluorenone 34 1.2.1.3 Aqueous Organic-Inorganic Mixed Redox Flow Cells Also, there are flow cells that combine organic and inorganic redox couples. Among the first organic redox couple based batteries is the one reported by the Xu et al. in 2009 where benzoquinone disulfonic acid (Tiron) was used as the positive electrolyte and lead sulfate was used as a negative electrolyte, all in acidic media. 35 Examples of inorganic and organic hybrid flow cells are listed in Table 1-3. 7 Table 1-3: Inorganic-organic mixed redox flow cells Media Redox Flow System Negative Positive Acidic Cadmium-Chloranil 36 Cd 2+ Chloro benzoquinone Tiron 35 Lead Benzoquinone disulfonic acid Metal-free AQDS/Bromide 37 Anthraquinone Bromine/Bromide Neutral Poly(TEMPO)/Zinc Hybrid 38 Zinc Polymeric TEMPO Zinc-Benzoquinone 39 Zinc Benzoquinone Alkaline Alkaline quinone 40 Anthraquinone Fe(CN) 6 4- / Fe(CN) 6 3- Biomimetic 41, 42 Flavin mononucleotide Phenazine deivatives 42 Fe(CN) 6 4- / Fe(CN) 6 3- Ferrocene 42 Alloxazine 43 Alloxazine Fe(CN) 6 4- / Fe(CN) 6 3- 1.2.1.4 All-Iron Redox Flow Cell Even though the all-vanadium system is the most extensively studied and has been developed to a commercial scale, the widespread deployment of vanadium systems has been hindered by the cost of the vanadium-based electrolytes. 18, 44 . Additionally, vanadium solutions are hazardous when used in large quantities. Meeting the target of levelized cost of energy storage of US$0.025/kWh set by the US Department of Energy requires significantly less expensive systems than vanadium-based flow batteries. 5, 45 To this end, we believe that iron-based redox flow batteries are promising because iron is inexpensive and abundantly available. The all-iron redox flow battery uses Fe (III)/Fe (II) redox couple as the positive electrode and the Fe(II)/Fe(0) redox couple as the negative electrode (Figure 1-3 and Eqs. 1-1 and 1-2) yielding a cell voltage of 1.21 V. Fe 3+ + e − ⇆ Fe 2+ E 0 = 0.77 V Positive Electrode (1-1) Fe 2+ + 2e − ⇆ Fe 0 E 0 = −0.44 V Negative Electrode (1-2) 8 This battery system was introduced by Savinell and Hrushka in 1981. 46 During the charging of the all-iron battery, iron is electrodeposited from solution of iron (II) chloride at the negative electrode while iron (II) chloride is converted to iron (III) chloride at the positive electrode. During discharge, elemental iron at the negative electrode is returned to the solution as iron (II) chloride, while iron (III) chloride is reduced to iron (II) chloride at the positive electrode. The specific energy of a 3 M solution of iron (II) chloride is 257 Wh/kg. At a bulk price of iron (II) chloride of US$3/kg, the material cost of the iron chloride is estimated to be US$10/kWh. For an efficient cell operating at high current densities of 100-300 mA/cm2, a system cost of US$100/kWh is potentially reachable with the all-iron redox flow battery. Figure 1-3: Schematic diagram of an all-iron redox flow cell 1.2.2 Electrode Materials Electrical conductivity and chemical stability are important factors for selecting electrode materials for redox flow cells. Carbon materials are generally stable in acidic and basic media. Also, the kinetics of the redox reactions are facile, and the electrode cost is low. Carbon materials include graphite, graphite felts, thermal and acid treated graphite, carbon composite materials, carbon nano-tubes, reticulated-vitreous carbon, carbon foam, and carbon cloth. In addition, to carbon materials some metallic meshes, and foams are also used. 16 Porosity, wettability and permeability of electrode materials, are important for achieving a large active surface area of the electrode. Negative Electrolyte FeCl 2 Pump Positive Electrolyte FeCl 3 /FeCl 2 Fe 2+ Fe 2+ Fe 2+ Fe 2+ Cl - Cl - Anion Exchange Membrane Cl - Cl - Cl - Cl - Load or Power Source e - e - e - e - Fe Fe Fe Fe 3+ Fe 3+ Cl - Fe 2+ Fe 2+ Fe 2+ Cl - Cl - ▬ + 9 1.2.3 Charge Transfer Kinetics and Electrocatalysis Studying the kinetics of charge transfer in redox reactions is a fundamental aspect of electrochemistry. The Butler-Volmer equation (Eq. 1-3) relates the rate of electrochemical reaction at any particular electrode potential to the fundamental kinetic parameters. The equation sums up the contribution of both the oxidation and reduction reactions. The overpotential h is the difference between the electrode potential, E, and the equilibrium electrode potential, E eq (Eq. 1-4). The equilibrium potential is related to the bulk concentration of the oxidized and reduced species C* ox and C* red through the Nernst Equation (Eq. 1.5). The exchange current density (I 0) (Eq. 1-6) is a measure of the current corresponding to the rate of exchange of oxidized and reduced forms of the redox couple at the equilibrium potential. Thus, the exchange current density is proportional to the rate constant (k o) and bulk concentration (C*) at the equilibrium potential. The equation also includes the effect of surface concentration (C s ) and the bulk concentration on the rate. The effect of potential on the reaction rate is captured in the kinetic parameter, a, often referred to as the symmetry factor. Other constants are the R is the gas constant, n is the number of electrons transferred in the rate determining step, F is the Faraday constant and T is the temperature. # # $ =& ' $( ) ' $( ∗ +(, -. a 012 34 5 6−& ' 8+9 ) ' 8+9 ∗ +(, -. (;<a) 012 34 5 6 (1-3) η = E – Eeq (1-4) > +? => $ − 34 12 @1- ' 8+9 ' $( 5 (1-5) # $ =A $ 1 2 B ' $( ∗ (;. C) ' 8+9 ∗ (C) (1-6) When a redox reaction involves outer-electron transfer, the rate constants are large as the barriers to electron transfer are low. The ferrous/ferric couple is an example of this category of redox reactions. However, in the presence of chloride ions or cyanide ligands, iron (II) and iron (III) form coordination complexes. In some cases, the complexing ligand may not participate in the redox reaction. However, it can affect the electron transfer process due to changes in 10 solvation, structural relaxation, and mediated electron transfer. Further, some reactions like the breaking of the Br-Br bond and the oxygen transfer step in VO 2 + / VO 2+ redox reaction, involve different charge transfer pathways compared to Fe 3+ / Fe 2+ redox chemistry. 18 Therefore, the kinetics of the VO 2 + / VO 2+ couple is slow. However, as the kinetics determine the rate capability and voltage efficiency, electrode surface modifications could be used to enhance the kinetics on carbon electrodes. Chemical etching, thermal etching, chemical doping, carbon nano-tube addition, and the introduction of metallic catalysts are different kinds of surface modifications. 13, 16 Further, to achieve better kinetics of redox reactions, a layer of electrocatalysts may be introduced. Electrocatalysts serve to increase the value of the exchange current density and the symmetry factor. Increasing the temperature also increase the rate constant for electron- transfer. At large negative values of overpotential, the contribution of the oxidation reaction to the current is small enough to be neglected. Correspondingly at large positive values of overpotential, the contribution of the reduction reaction to the current may be neglected. The case of requiring large overpotentials is encountered with reactions that have a low exchange current density, where the large overpotentials enhance the rate of electron transfer. Under these conditions, we may simplify the Butler-Volmer relationship to the Tafel Equation (Eq. 1-7a or b) that neglects either the oxidation and reduction reaction currents. # # $ =& ' 8+9 ) ' 8+9 ∗ +(, -. (;<a) 012 34 5 6 (1-7a) or # # $ =& ' $( ) ' $( ∗ +(, -. a 012 34 5 6 (1-7b) When the surface concentration is not different from the bulk concentration, the Tafel equation is also often represented in the “log-linear” form as per Eq. 1-8. h = a + b log I (1-8) Where a = 2.303 (RT/αnF) log Io and b = 2.303 (RT/αnF), b is referred to as the Tafel Slope. 11 To illustratrating the use of the Tafel Equation, experimental data of overpotential and the current density obtained from our studies on electrodeposition and electro-dissolution of iron are shown in Figure 1-4. Figure 1-4: Tafel plot of iron deposition and dissolution 1.2.4 Mass Transport In addition to charge transfer, mass transport is also important for sustaining an electrochemical reaction. Mass transport rates determine the concentration of the redox active species at the surface for any particular bulk concentration. Mass transport to the electrodes occurs three different modes: diffusion, convection and migration. When the concentration of the reacting ions is much smaller than that of the supporting electrolyte, the transport of the electroactive ions by migration is likely to be a small contribution to the rate of mass transport, and this contribution is often neglected in the calculations. Convective mass transport is dependent on the rate of stirring of the flow rate to the electrode. Adjacent to the surface of the electrode is a static layer of the solution in which the only diffusion is the main mode of mass transport. Such a static layer may be termed a diffusion layer. The thickness of such a layer depends on the flow rate or rate of stirring. The Nernst diffusion layer (NDL) model provides a way to relate the concentration at the surface of the electrode to the bulk concentration for any particular value of the diffusion layer thickness. In Figure 1-4 we depict the essence of the NDL model. The surface concentration of an electroactive species at a steady-state current I can be related to -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 Overpotential, V Log current, (current in A) Fe 2+ (aq) + 2e - à Fe(s) Fe(s)à Fe 2+ (aq) +2e - 12 the function of the diffusion layer thickness, the diffusion coefficient, and the bulk concentration, as per Eq.1- 9. C s Fe(II) = C * Fe(II) - (I d / n F D) (1-9) Where C*= bulk concentration, I = current density, D = diffusion coefficient of iron (II), d = the diffusion layer thickness, n = number of electrons and F =faraday constant. Figure 1-5: Nernst diffusion model The mass transport rate may be increased by directing the flow towards the surface of the electrode at high flow rates. To this end, flow-field configurations are used in redox flow batteries that direct the flow in an optimum manner. The area of the electrode available for mass transport also determines the overall rate of reaction. Thus, the design of flow fields is a sub-area in redox flow cell engineering aimed at improving the utilization of the electrode surface for better mass transport. 7, 16 Among the flow field designs, are three common types: forced-flow, interdigitated, and unrestricted flow. Properties of these regions can be different, and these differences can affect the performance of the battery. 1.2.5 Membrane Development The membrane separator in a redox flow cell accomplishes two functions: a medium for ion transport and preventing the mixing of the positive and negative electrolytes. Thus, membrane development in redox flow cells is aimed at preventing the cross-over of redox-active materials and ensuring good ionic conductivity. The stability of the membrane in contact with the liquid C s C * ! Distance from the electrode Concentration 13 redox electrolyte is also a major concern for long-term applications. If cross-over of redox-active materials from one side to the other side happens the membrane separator, self-discharge occurs. Thus, the coulombic efficiency for the charge and discharge processes is reduced. Mixing of the positive and negative electrolyte results in the reduction of capacity often termed “capacity fade”. When redox-active materials are anionic, cation exchange membranes are used in redox flow cells to prevent the cross-over and self-discharging. Nafion® and Fumapem® proton-conducting membranes are widely used. When redox-active molecules are cationic, anion-exchange membranes are used. Tokuyama and Fumasep are anion-exchange membranes used in research and development of redox flow cells. Reliability of the membrane to prevent cross-over and curtail degradation is an important requirement for large-scale energy storage systems. Also, higher conductivity of the membrane is a reduces the internal resistance of the system to improve the energy efficiency. Therefore, designing membranes with the optimal properties at a low-cost is continued to be a challenge in the field of redox flow batteries. The earliest experiments on the all-iron flow battery system deployed iron chloride solutions and a porous membrane separator between the two electrodes. 46, 47 Inexpensive separator materials such as printer paper and Daramic membranes are also tested in all-iron redox flow systems. 48 Such arrangements resulted in unavoidable cross-diffusion of Fe(III) and Fe(II) between the positive to the negative sides of the cell and thus, reduced the coulombic efficiencies. In our group’s initial work on iron chloride flow cell, we introduced another cell configuration for the all-iron flow battery using an anion-exchange membrane (AEM) as a separator between the positive and negative electrodes. 49 We demonstrated that the AEM facilitates the transfer of chloride ions between positive and negative sides of the cell without mixing of the positive and negative electrolytes. Thus, self-discharge could be avoided (Figure 1-3). 14 1.2.6 Supporting Electrolytes Supporting electrolytes are ionic salts that are not redox active and are added to the battery electrolyte to improve the conductivity and thereby minimize the ohmic resistance losses. Also, in some electrochemical reactions, the reduction potentials can be shifted by the pH of the supporting electrolyte or by complexation with the constituents of the electrolyte. In acidic media, sulfuric acid is widely used as the supporting electrolyte. In alkaline media, sodium or potassium hydroxides are commonly used as the supporting electrolyte. In acidic systems, protons or the hydronium ion (H 3O + ) is the charge carrier while in alkaline solutions, hydroxide ions carry the charge. Further, supporting electrolyte is used to balance the osmotic pressure differences across the ion-exchange membrane caused by changes in the concentration of the redox active ionic species during charge and discharge. In all-iron redox flow cells, chloride ions are used as supporting electrolytes. In our group, ammonium chloride is used as supporting electrolyte. Energy Storage Systems (ESS) has used sodium chloride as an additive to the electrolyte and has operated the cell in a pH ranging from 1 to 4. 50 However, formation of iron-chloro-complexes or adsorption of chloride ions to the electrode surface can affect the performance of the system. Use of sulfate as a supporting electrolyte has been proposed to prevent iron-chloro complex formations. 48 In 2015, Weber et al. introduced a redox flow battery technology using iron sulfate as the redox electrolyte with sodium chloride or sodium sulfate as the supporting electrolyte. 48 1.2.7 Efficiency Performance of a redox flow cell is described by its efficiency values. Coulombic efficiency (CE, Eq.1-10), voltage efficiency (VE, Eq.1-11) and energy efficiency (EE, Eq.12) are three different efficiency metrics: CE=Q delivered/ Q input (1-10) VE=Vdischarging/ Vcharging (1-11) EE= E output / E input (1-12) Where Q is the charge, V is the cell voltage, and E is the energy. 15 Some redox flow systems show poor coulombic efficiency due to parasitic reactions during charging and/or discharging steps. Voltage efficiencies can be lower due to poor reaction kinetics, poor mass transport and also ohmic losses. To achieve high energy efficiency, both the coulombic efficiency and voltage efficiency of the system should be high. The energy efficiency for redox flow batteries is typically no more than 80%. Inefficiencies raise the burden on thermal management systems. Therefore, energy efficiency has a direct impact on the cost of the system. The major parasitic side reaction in the all-iron redox flow battery is the hydrogen evolution reaction which occurs at the potential of electrodeposition of iron at low pH as indicated in the Pourbaix diagram (Figure 1-6). Therefore, raising the pH would be considered a solution. However, the pH of the electrolyte has to be maintained below the value of 2 to avoid the precipitation of the iron hydroxides. Even if operating at a low pH, the hydrogen evolution occurring at the iron electrode causes an increase of electrolyte pH followed by precipitation of the iron hydroxides. Therefore, maintaining the pH of the electrolyte is a challenge. Figure 1-6: Pourbaix diagram of iron at ionic concentrations of 1.0 mM 51 (« represents the potential window over which the negative electrode operates.) Our group’s previous studies demonstrated some approaches to mitigate hydrogen evolution at the iron electrode by using complexing agents and additives. 49 Complexation of ferrous and 16 ferric ions with different ligands aimed at enhancing the solubility of iron species at elevated pH has been studied by Savinell et al. 52 However, the use of these organic ligands shifted the potential for the electrodeposition of iron to negative values, decreasing the coulombic efficiency. 47 Further, these researchers demonstrated in experiments using a rotating rod electrode that a combination of pH greater than 2 and supporting electrolyte with a high chloride concentration could increase the efficiency of electrodeposition to 97%, significantly reducing the rate of the hydrogen evolution reaction. 47 A technical white paper from ESS reported a cell with a porous separator that could operate at a round-trip Coulombic efficiency >97% and an energy efficiency >76% for over 10,000 cycles at over 80% depth of discharge with no performance degradation and minimal maintenance for iron chloride redox flow system. 50 To address the issue of hydrogen evolution during iron deposition, Yan et al. have developed an all-soluble, all-iron flow battery that did not involve the iron deposition reaction. They combined an iron-triethanolamine (TEOA) redox couple with an iron−cyanide redox couple and demonstrated stable coulombic efficiencies between 80 and 90% for 110 cycles. However, the performance of this battery was found to degrade due to cross-over of the TEOA ligand. 53 Thus, an all-iron flow battery continues to be a topic of research with the main outstanding issues being the suppression of hydrogen evolution and self-discharge. 1.2.8 Electrolyte Regeneration or Recombination Parasitic reaction not only lower the energy efficiency but also lead to capacity fade. Recombination methods are required for such systems to restore the composition of the solutions during long-term operation. However, recombination systems may require expensive catalysts and can increase the overall cost of the redox flow system. In an all-iron redox flow battery, as HER could not be completely prevented, an in-tank electrolyte rebalancing reactor has been studied by Wainright et al. This reactor recombines the evolved hydrogen with the excess of accumulated iron(III). 12 17 1.3 Lithium-Sulfur Batteries Figure 1-7: Schematic of the electrochemical processes in a lithium-sulfur battery (C = charge, D = discharge) from reference. 54 At the anode: Li + e - ⇋ Li (1-13) At the cathode: S 8 + 16e - ⇋ 8S 2- (1-14) Overall reaction: 16 Li + S 8 ⇋ 8Li 2S (1-15) A lithium-sulfur battery is composed of lithium metal anode and sulfur cathode (Figure 1-6). During discharge of this battery lithium metal and sulfur are converted to lithium sulfide (Eq. 1- 15). At the negative electrode lithium metal dissolution occurs (Eq 1-13), and at the positive electrode elemental sulfur is converted to sulfide ions (Eq. 1-14). With an average cell voltage of 2.2 V (Eq. 1-13) and specific capacities of 1672 mAh/ g for the sulfur electrode and 3860 mAh/ g of lithium electrode, the Li-S system has a theoretical specific energy of 2600 Wh/ kg. Thus, it is a promising rechargeable system to replace lithium-ion battery technology. Li-S system has the advantage of relying on abundantly-available and low-cost sulfur. Even though a half a century of time of research and development has been spent from the introduction of Li-S technology in 1960, commercialization of the technology is still limited by a number of intrinsic issues such as capacity fade, poor utilization of sulfur, low-coulombic efficiency, and low rate capability. 55 18 These issues arise from the complex electrochemistry of Li-S system. The multi-electron conversion between elemental sulfur and lithium sulfide is accompanied by phase changes at the electrodes, changes in the molar volume of the electroactive materials, and migration of dissolved species within the cell. To address these challenges, the focus of research has been in three broad areas namely materials studies, electrode design and cell engineering. 56 However, understanding of individual steps in the operation of Li-S system in greater detail is required to develop a reliable technology. Derek Moy and S. R. Narayanan at USC investigated a Li-S system with a mixed conduction membrane to improve capacity retention and material utilization. 54, 57, 58 Derek Moy’s Ph.D. dissertation provides an extensive introduction to the literature and experimental findings. 57 Continuing from Moy’s work, Chapter four explains in detail my understanding of the discharge process at the cathode and methods of improving the utilization and rate capability of Li-S system and further improvements to the novel membrane invented at USC. 1.4 Electrochemical Energy Conversion Pathways for Carbon Dioxide Reduction 1.4.1 Carbon Dioxide Emission and Environmental Impact Carbon dioxide is a very stable and inert molecule in the atmosphere that is a greenhouse gas. The rising level of carbon dioxide and other greenhouse gases is contributing to the increase in the average global temperature. Consequently, significant climate and environmental changes have already begun to occur. Carbon dioxide emissions arise from biological respiration, volcanic and geothermal activity, and anthropogenic activities. The rate of fossil fuel burning, deforestation, and industrial emissions has been increasing rapidly in the last 100 years. 59 It is found that the CO 2 level in the atmosphere has risen by 25% (100 ppm) in the last 40 years. The increase in the concentration has been associated with a rise of 1 degree Celsius in the global average temperature. 60 To avoid the rise in carbon dioxide concentration to catastrophic levels, we must not only curtail emissions and recycle the carbon dioxide to useful products. 19 1.4.2 Thermodynamics and Kinetics of Carbon Dioxide Reduction The complete oxidation of organic compounds is an exothermic process and results in the production of CO 2. The Gibbs free energy of formation of CO 2 is -394.4 kJ mol -1 at 298 K and thus reversing carbon dioxide to useful organic compounds requires the supply of energy and catalytic processes to make specific types of chemical bonds. 61 CO 2 recycling occurs in nature via photosynthesis. Man-made approaches for reducing the level of CO 2 in the atmosphere are classified as carbon capture, carbon sequestration, carbon utilization. The latter provides the added benefit of generating useful organic compounds. Over the last three decades there have been a large number of studies to convert CO 2 to useful organic products using catalytic, electrochemical and photochemical methods. 62-64 The energy requirement for these processes can be generated from solar and wind resources. The CO 2 molecule is linear with a bond length of 1.17 A˚. A decrease in the O-C-O angle is required to achieve CO 2 reduction. Bending the OCO bond enhances the electron-accepting ability of CO 2. 65 Products of CO 2 reduction can be either organic compounds or syngas (a mixture of carbon monoxide and hydrogen). Table 1-4 shows standard reduction potentials for the reduction of CO 2 to different organic molecules. The standard reduction potential can be used to determine the spontaneity of the carbon dioxide reduction reaction (Eq.1-16). CO 2 (g) + H 2 D Organic Compound + H 2 O (l) (1-16) The standard free energy change for this reaction can be calculated from the standard reduction potential and the number of electrons transferred in the reaction (Eq. 1-17). ΔG° = −n F E o (1-17) where n= number of electrons and F= Faraday constant For example, the calculated free energy for the synthesis of methane from carbon dioxide is shown below (Eq.1-19). CO 2 (g) + 8H + + 8e − → CH 4 (g) + 2H 2 O (l) E o =-0.24 (1-18) CO 2 (g) + 4H 2→ CH 4 (g) + 2H 2 O (l) ∆G o = -185 kJ/mol (1-19) 20 Table 1-4: Standard reduction potentials of CO 2 reduction at pH 7 61, 66 Half reactions of CO 2 reduction Standard reduction potential, V vs. Normal Hydrogen Electrode. CO 2 (g) + e − → *COO − −1.90 CO 2 (g) + 2H + + 2e − → HCOOH (l) −0.61 CO 2 (g) + H 2 O (l) + 2e − → HCOO − (aq) + OH − −0.43 CO 2 (g) + 2H + + 2e − → CO (g) + H 2 O (l) −0.53 CO 2 (g) + H 2 O (l) + 2e − → CO (g) + 2OH − −0.52 CO 2 (g) + 4H + + 2e − → HCHO (l) + H 2 O (l) −0.48 CO 2 (g) + 3H 2 O (l) + 4e − → HCHO (l) + 4OH − −0.89 CO 2 (g) + 6H + (l) + 6e − → CH 3 OH (l) + H 2 O (l) −0.38 CO 2 (g) + 5H 2 O (l) + 6e − → CH 3 OH (l) + 6OH − −0.81 CO 2 (g) + 8H + + 8e − → CH 4 (g) + 2H 2 O (l) −0.24 CO 2 (g) + 6H 2 O (l) + 8e − → CH 4 (g) + 8OH − −0.25 2CO 2 (g) + 12H + + 12e − → C 2 H 4 (g) + 4H 2 O (l) 0.06 2CO 2 (g) + 8H 2 O (l) + 12e − → C 2 H 4 (g) + 12OH − −0.34 2CO 2 (g) + 12H + + 12e − → CH 3 CH 2 OH (l) + 3H 2 O (l) 0.08 2CO 2 (g) + 9H 2 O (l) + 12e − → CH 3 CH 2 OH (l) + 12OH − (l) −0.33 1.4.3 Electrochemical Carbon Dioxide Reduction Using Metal Catalysts Electrocatalysts are typically screened for their selectivity, current efficiency, electron transfer rates, and energy efficiency. The overpotential (Eq. 1-4) and coulombic efficiency (Eq. 1-10) at a specific current density are often reported in CO 2 reduction literature as Figures of merit for various electrocatalyst. 61, 67 Usually, the overpotentials are high and the product selectivity is poor. Thus, the efficient electrochemical reduction of CO 2 into organic products has been a 21 challenge. Table 1-5 summarizes four categories of metal catalysts that have been studied for CO 2 conversion. Even though a variety of a metal-based electrocatalysts have been investigated, the product yield is low because of the parasitic hydrogen evolution. Formate formation from the carbon dioxide reduction is feasible compared to that of other organic products as formate generation does not require dissociation of strong C-O bonds, but a proton-coupled two-electron reduction step. Metal catalysts with higher overpotential for HER show better selectivity of formate. 68 CO formation happens via reductive disproportionation of weakly bound carbon dioxide on the metal surface. 69 Additionally, the poor adsorption of hydrogen and anions on those metal surfaces gives higher CO selectivity over HER. 68 Copper has been widely studied for methanol and methane production from CO 2. 64 Table 1-5: Metal catalysts used in carbon dioxide reduction Metal catalysts Cathodic reactions Pt, Ti, Ni, Fe HER is favored 68, 70 Sn, In, Pb, Hg, Bi, Cd, Tl CO 2 to formate 68, 71, 72 Au, Ag, Zn, Pd, Bi, Ga CO 2 to CO 68, 73 Cu CO 2 to Hydrocarbons, alcohols, formic acid and CO 64 Surface modified metals, metal alloys, metal oxides, metal chalcogenides, metal carbides, metal-organic complexes and metal-free carbon-based catalysts are studied for the electrochemical conversion of CO 2. 59, 66, 74 Metal alloys can provide a higher selectivity and better kinetics by adjusting the binding of the intermediates of carbon dioxide reduction. 66 Metal alloys with non-noble metals can reduce the cost of the catalysts as well. Metal oxides- based electrocatalysts require high temperature for the operation. 59 However, these systems could generate higher current densities for CO and H 2 generation. 59 Even though molecular catalysts provide indirect pathways of carbon dioxide reduction, they show higher product selectivity and lower overpotentials compared to metal catalysts. 59 Iron, selenium and molybdenum sulfide metal clusters have been studied as biomimetic CO 2 reduction pathways and low overpotentials for CO production have been reported. 66, 75 Further, different edge sites 22 of these chalcogen materials can lead different intermediates binding to achieve product selectivity in carbon dioxide conversion. 66 1.4.4 Electrochemical Carbon Dioxide Reduction Using Bio-Catalysts The most widely occurring process of CO 2 reduction is photosynthesis. Therefore, we can expect bio-catalytic pathways of CO 2 conversion to offer high product selectivity. Over the last 30 years, scientists have been investigating CO 2 reduction using different enzymes. The reaction studied in bio-catalytic CO 2 conversion is the generation of formate using the enzyme, formate dehydrogenase. Formate dehydrogenase is often used to convert formate to CO 2. However, a number of studies do report shifting the equilibrium toward formate generation by supplying carbon dioxide. 76-79 There are two different categories of FDH, namely, metal-dependent and metal-independent types. Metal-dependent FDH has metal containing active sites while metal- independent FDH does not have metal containing active sites. These metal components allow the electron transfer at the active site and faster turnover rates of enzymatic reactions are reported with metal-dependent FDH compared to that with metal-independent FDHs. All the metal-independent FDH uses cofactors to facilitate the electron transfer. Some metal- dependent FDH also uses cofactors for electron transfer. These cofactors act as terminal electron acceptors at the active site of FDH. Further, the FDH enzyme has been combined with a cascade of enzymes for generating higher carbon compounds, and also with carbonic anhydrase to enhance CO 2 dissolution. 80-82 Other than FDH, there are studies based on carbon dioxide reductase that catalyze CO 2 reduction to formate using molecular hydrogen. 83, 84 Carbon dioxide reductase enzyme extracted from Acetobacterium woodi has two active sites and they are connected to each other via eleven iron-sulfur (Fe-S) clusters distributed between all the subunits. 75, 83, 84 Carbon monoxide dehydrogenase whose typical function is to oxidize CO to CO 2 is another enzyme that has been studied for CO 2 reduction. Even though there are two types of carbon monoxide dehydrogenase enzymes with nickel-iron-sulfur (Ni-Fe-S) and molybdenum-copper-sulfur (Mo- 23 Cu-S) clusters, only the Ni-Fe-S containing enzyme was identified as being capable of CO 2 reduction. 85, 86 1.4.5 Carbon Dioxide and Formate Interconversion Using Metal Independent FDH Formic acid or formate salt is derived readily from carbon dioxide as it involves the formation of a just one carbon-hydrogen bond without requiring the dissociation of the strong carbon- oxygen bond in carbon dioxide. Further, formic acid is of considerable commercial value as it is useful as a chemical feedstock, a fuel in fuel cells, and an efficient carrier of hydrogen. 87, 88 Thus, it is not surprising that the electrochemical reduction of carbon dioxide to produce formate has been studied by others. For this process, enzymes offer a high level of selectivity. 89 In natural environments, FDH converts formate to carbon dioxide using the co-factor NAD + as per the reaction in Eq.1-15 (Figure 1-8). Peacock et al. reported that this reaction follows a sequential pathway where products are released only after both the substrate and the co-factor are bound to the enzyme. 90 Further, they described that formate binding happens after the binding of NAD + . 90 NAD + + HCOO - D CO 2 + NADH (1-17) Figure 1-8: FDH catalyzed formate oxidation Reported values of Michalis Menton constant (K m) for yeast FDH for formate oxidation are 32 µM and 1.7 mM for NAD + and formate respectively. 79 For plant FDH those values are 7.2 µM and 0.3 mM for NAD + and formate respectively. 90 Also, NADH is known as a non-competitive FDH catalyzed 24 inhibitor for formate oxidation. 91 In a dynamic chemical equilibrium, we can shift the reaction of formate oxidation (Eq. 1-17) towards the reactants by supplying a large excess of the products (NADH and carbon dioxide). In 1950, Martin et al. reported that there can be no significant direct reduction of CO 2 to formate in the presence of FDH and NADH. 92 Further, in 1951, Davison has also reported that attempts to reverse the formate oxidation reaction were not successful. 93 Martin et al. proposed that by coupling the FDH system with an energy-yielding reaction might generate formate from CO 2. 92 However, as the equilibrium constant for the reverse reaction of formate oxidation is 2300 times lower than the forward reaction, Peacock explained that achieving measurable steady-state kinetics of the reverse reaction of formate oxidation require higher concentrations of FDH (60 µM) and cofactors. 90 Studies of cofactor detection and interconversion of formate and carbon dioxide with NAD cofactors at relatively higher cofactor concentrations are discussed in Chapter five. Further inhibition of formate oxidation by bicarbonate (inhibition coefficient is 97 mM) suggests that reverse reaction can be achieved by increasing the bicarbonate concentration. 79, 90 However, as it is further reported that with FDH, NAD + and bicarbonate might act as a product inhibitors or dead-end inhibitors, a system with bicarbonate, FDH and NAD cofactor might not be a suitable approach for achieving the reverse reaction. 78 There are several recent reports on the generation of formate using FDH from the Candida boidinii with or without mediators and with natural cofactors (Table 1-6). 80, 82, 91, 94 Table 1-6 shows that the presence of Carbonic Anhydrase enzyme could generate higher formate concentration. 80 This variety of FDH is available commercially and has been studied as bound to electrode surfaces and as hybrid systems with microbial fuel cells. 94, 97 Kim et al. has also reported an electro-enzymatic cell of CO 2 conversion using FDH from Candida boidinii with [Cp*Rh(bpy)Cl] Cl complex and identified copper as the better metal catalyst for the electrochemical generation of NADH compared to that on gold, silver and indium. 91 25 Table 1-6: Carbon dioxide reduction using metal-independent formate dehydrogenase enzyme from Candida boidinii Electrolytes and Mediators Electrodes and applied potential, V vs. NHE Formate generation remarks Reference Phosphate buffer, pH 7.4, NADH, Neutral Red mediator Graphite rod at -0.6 to -0.8 413.77 mg/L or 9.36 mg/ 1 mole CO 2 and 15.48 µM enzyme 82 Phosphate buffer, pH 7.4, 100 mM, NADH, Neutral Red mediator, Carbonic Anhydrase Enzyme Cold rolled graphite– polytetrafluoroethylene (PTFE) composite layer on a stainless-steel mesh at -0.75 497-647 mg/ L 80 100 mM PBS, pH 6.0, 10 mM NaHCO 3, 1.0 mM NADH, Neutral Red mediator Graphite rod at -0.8 to -1 64.71 ± 3.38 mg /L /hr 94 1.4.6 Molecular Level Understanding of Catalysis by FDH FDH is a homodimer with two active sites. 98 Each monomer of FDH has two domains namely; “NAD binding” and “catalytic”. 99 Figure 1-9 shows a schematic representation of the postulated formate and NAD + binding in the Michaelis complex of FDH from Pseudomonas. 100 Arg284, Asn146, Ile122, and Hi332 are the main residues located near the active site of FDH. 100 Among them, Arg 284 and Asn 146 are the two critical amino acid residues for locating formate in the productive orientation for hydride transfer. 100, 101 Ambiguously, the strong interactions between the formate and Arg 284 and Asn 146 can lead FDH to a barrier for hydride transfer. 101 Thus, FDH activity can be relatively low compared to other NAD + dependent enzymes. 101 26 FDHs from Thiobacillus and Pseudomonas show higher CO 2 reducing activities than FDH from Candida boidinii. 67, 94, 100, 102, 103 Popover et al. explain that the additional residue between the third glycine residue in the conservative triad and the residue Asp 221 in FDH from yeast and fungi gives NAD + binding specificity. 104 Further, Dmitry et al. have postulated structural reasons for the changes in the free energy barrier of CO 2 reduction. 100 Figure 1-9: Schematic representation of the hypothetical substrate binding of FDH 100 1.4.7 Challenges for Enzyme-Catalyzed Electrochemical Conversion of Carbon Dioxide to Formate The equilibrium constant for the reaction as written in Eq. 1.15 is calculated to be 10 7 (see Appendix D). Thus, shifting the equilibrium towards formate will require at least two orders of higher concentration of NADH compared to the concentration of generated formate. Even if one is successful in producing formate by such an approach, the rapid rate of formate re- oxidation to carbon dioxide, supported by the normal role of FDH, will reduce the net yield of formate. If NADH can be electrochemically regenerated from NAD + (Eqs.1-16 and 1-18) we would be able to continuously supply NADH at high concentrations and thereby sustain the production of formate from carbon dioxide. However, we found that the electrochemical reduction of NAD + using carbon electrodes did not produce the enzyme-active isomer of NADH. 27 These observations were consistent with other reports in the literature as the continuous generation of NADH in an active form by the electro- reduction of NAD + is not promising because of the isomerization (Eq. 1-21) and dimerization (Eq. 1-19) of the reduced products. 82, 105 1 st electron reduction (fast step): NAD + + e - à NAD • (1-18) Dimerization (fast step): 2NAD • à (NAD)2 (enzyme inactive) (1-19) Protonation and 2 nd electron reduction (slow step): NAD • + H + + e - à 1,4- NADH (enzyme active form) (1-20) NAD • + H + + e - à 1,6- NADH (enzyme inactive form) (1-21) Even though, a successful electrochemical pathway for enzyme active NADH regeneration using platinum electrodes at -0.7 to -1.0 V vs. Ag/AgCl was previously reported by Yun et al., platinum- catalyzed electro-generation of NADH would be inefficient because of the slow kinetics requiring several hundreds of millivolts of overpotential for the electro-reduction process. 106 Thus, we found that the FDH-catalyzed conversion of carbon dioxide into formate using electrochemically-generated NADH is not promising. Further, we were not successful in our efforts to reverse the reaction in Eq. 1-17 even with a large excess of NADH. At concentration values higher than 0.3 mM, NADH is known to inhibit the activity of FDH for the production of formate (the reverse of the reaction shown in Eq. 1-17). 91 Consequently, we began to focus on a finding a suitable co-factor to replace NADH that can be active with FDH and also be regenerated electrochemically. Chapter Six and Seven discuss the replacement of NADH using potential artificial cofactors to achieve the reduction of CO 2 at high efficiency. 28 1.5 Chapter One References 1. Facts, U. S. E. The United States uses a mix of energy sources. https://www.eia.gov/energyexplained/?page=us_energy_home. 2. Blomgren, G. E., The Development and Future of Lithium Ion Batteries. Journal of The Electrochemical Society 2017, 164 (1), A5019-A5025. 3. Ovshinsky, S. R.; Fetcenko, M. A.; Ross, J., A Nickel Metal Hydride Battery for Electric Vehicles. Science 1993, 260 (5105), 176-181. 4. Dunn, B.; Kamath, H.; Tarascon, J.-M., Electrical Energy Storage for the Grid: A Battery of Choices. Science 2011, 334 (6058), 928. 5. Report, D. O. E., Levelized Cost and Levelized Avoided Cost of New Generation Resources in the Annual Energy Outlook 2017, U.S. Energy Information Administration: 2017 April. 6. Pan, F.; Wang, Q., Redox Species of Redox Flow Batteries: A Review. Molecules 2015, 20 (11), 19711. 7. Alotto, P.; Guarnieri, M.; Moro, F., Redox flow batteries for the storage of renewable energy: A review. Renewable and Sustainable Energy Reviews 2014, 29, 325-335. 8. Leung, P. K.; Mohamed, M. R.; Shah, A. A.; Xu, Q.; Conde-Duran, M. B., A mixed acid based vanadium-cerium redox flow battery with a zero-gap serpentine architecture. J. Power Sources 2015, 274, 651-658. 9. Lim, H. S.; Lackner, A. M.; Knechtli, R. C., Zinc-Bromine Secondary Battery. J. Electrochem. Soc. 1977, 124 (8), 1154-1157. 10. Yang, B.; Hoober-Burkhardt, L.; Krishnamoorthy, S.; Murali, A.; Prakash, G. K. S.; Narayanan, S. R., High-Performance Aqueous Organic Flow Battery with Quinone-Based Redox Couples at Both Electrodes. J. Electrochem. Soc. 2016, 163 (7), A1442-A1449. 11. Rajarathnam, G. P.; Easton, M. E.; Schneider, M.; Masters, A. F.; Maschmeyer, T.; Vassallo, A. M., The influence of ionic liquid additives on zinc half-cell electrochemical performance in zinc/bromine flow batteries. RSC Advances 2016, 6 (33), 27788-27797. 29 12. Selverston, S.; Savinell, R.; Wainright, J., In-tank hydrogen-ferric ion recombination. J. Power Sources 2016, 324, 674-678. 13. Leung, P.; Shah, A. A.; Sanz, L.; Flox, C.; Morante, J. R.; Xu, Q.; Mohamed, M. R.; Ponce de León, C.; Walsh, F. C., Recent developments in organic redox flow batteries: A critical review. J. Power Sources 2017, 360, 243-283. 14. Bartolozzi, M., Development of redox flow batteries. A historical bibliography. Journal of Power Sources 1989, 27 (3), 219-234. 15. Selverston, S.; Savinell, R. F.; Wainright, J. S., In-tank hydrogen-ferric ion recombination. Journal of Power Sources 2016, 324, 674-678. 16. Weber, A. Z.; Mench, M. M.; Meyers, J. P.; Ross, P. N.; Gostick, J. T.; Liu, Q., Redox flow batteries: a review. Journal of Applied Electrochemistry 2011, 41 (10), 1137. 17. Lin, W.; Stocker, K. M.; Schatz, G. C., Mechanisms of Hydrogen-Assisted CO 2 Reduction on Nickel. Journal of the American Chemical Society 2017, 139 (13), 4663-4666. 18. Kear, G.; Shah, A. A.; Walsh, F. C., Development of the all-vanadium redox flow battery for energy storage: a review of technological, financial and policy aspects. Intl. J. Energy Res. 2012, 36 (11), 1105-1120. 19. Skyllas-Kazacos, M.; Grossmith, F., Efficient vanadium redox flow cell. Journal of the Electrochemical Society 1987, 134 (12), 2950-2953. 20. Jorné, J.; Kim, J.; Kralik, D., The zinc-chlorine battery: half-cell overpotential measurements. Journal of Applied Electrochemistry 1979, 9 (5), 573-579. 21. Savinell, R.; Liu, C.; Galasco, R.; Chiang, S.; Coetzee, J., Discharge Characteristics of a Soluble Iron-Titanium Battery System. Journal of the Electrochemical Society 1979, 126 (3), 357- 360. 22. Skyllas-Kazacos, M., Novel vanadium chloride/polyhalide redox flow battery. Journal of Power Sources 2003, 124 (1), 299-302. 23. Remick, R. J.; Ang, P. G., Electrically rechargeable anionically active reduction-oxidation electrical storage-supply system. Google Patents: 1984. 30 24. Jayathilake, B. S.; Plichta, E. J.; Hendrickson, M. A.; Narayanan, S. R., Improvements to the Coulombic Efficiency of the Iron Electrode for an All-Iron Redox-Flow Battery. Journal of The Electrochemical Society 2018, 165 (9), A1630-A1638. 25. Linden, D.; Reddy, T. B., Handbook of Batteries. 3rd. McGraw-Hill 2002. 26. Collins, J.; Kear, G.; Li, X.; Low, C. T. J.; Pletcher, D.; Tangirala, R.; Stratton-Campbell, D.; Walsh, F. C.; Zhang, C., A novel flow battery: A lead acid battery based on an electrolyte with soluble lead(II) Part VIII. The cycling of a 10cm×10cm flow cell. Journal of Power Sources 2010, 195 (6), 1731-1738. 27. Yamamura, T.; Watanabe, N.; Yano, T.; Shiokawa, Y., Electron-Transfer Kinetics of Np3 + ∕ Np4 + , NpO2 + ∕ NpO2 2 + , V2 + ∕ V3 + , and VO2 + ∕ VO2 + at Carbon Electrodes. Journal of The Electrochemical Society 2005, 152 (4), A830-A836. 28. Hoober-Burkhardt, L.; Krishnamoorthy, S.; Yang, B.; Murali, A.; Nirmalchandar, A.; Prakash, G. S.; Narayanan, S., A new Michael-reaction-resistant benzoquinone for aqueous organic redox flow batteries. Journal of The Electrochemical Society 2017, 164 (4), A600-A607. 29. Lin, K.; Chen, Q.; Gerhardt, M. R.; Tong, L.; Kim, S. B.; Eisenach, L.; Valle, A. W.; Hardee, D.; Gordon, R. G.; Aziz, M. J., Alkaline quinone flow battery. Science 2015, 349 (6255), 1529-1532. 30. Brushett, F. R.; Jansen, A. N.; Vaughey, J. T.; Su, L.; Milshtein, J. D., Materials for use with aqueous redox flow batteries and related methods and systems. Google Patents: 2017. 31. Hagemann, T.; Winsberg, J.; Grube, M.; Nischang, I.; Janoschka, T.; Martin, N.; Hager, M. D.; Schubert, U. S., An aqueous all-organic redox-flow battery employing a (2, 2, 6, 6- tetramethylpiperidin-1-yl) oxyl-containing polymer as catholyte and dimethyl viologen dichloride as anolyte. Journal of Power Sources 2018, 378, 546-554. 32. Janoschka, T.; Martin, N.; Hager, M. D.; Schubert, U. S., An Aqueous Redox-Flow Battery with High Capacity and Power: The TEMPTMA/MV System. Angewandte Chemie International Edition 2016, 55 (46), 14427-14430. 33. Liu, T.; Wei, X.; Nie, Z.; Sprenkle, V.; Wang, W., A total organic aqueous redox flow battery employing a low cost and sustainable methyl viologen anolyte and 4-HO-TEMPO catholyte. Advanced Energy Materials 2016, 6 (3). 31 34. Kwon, G.; Lee, S.; Hwang, J.; Shim, H.-S.; Lee, B.; Lee, M. H.; Ko, Y.; Jung, S.-K.; Ku, K.; Hong, J., Multi-redox Molecule for High-Energy Redox Flow Batteries. Joule 2018. 35. Xu, Y.; Wen, Y.-H.; Cheng, J.; Cao, G.-P.; Yang, Y.-S., A study of tiron in aqueous solutions for redox flow battery application. Electrochimica Acta 2010, 55 (3), 715-720. 36. Xu, Y.; Wen, Y.; Cheng, J.; Cao, G.; Yang, Y., Study on a single flow acid Cd–chloranil battery. Electrochemistry Communications 2009, 11 (7), 1422-1424. 37. Huskinson, B.; Marshak, M. P.; Suh, C.; Er, S.; Gerhardt, M. R.; Galvin, C. J.; Chen, X.; Aspuru-Guzik, A.; Gordon, R. G.; Aziz, M. J., A metal-free organic–inorganic aqueous flow battery. Nature 2014, 505 (7482), 195. 38. Winsberg, J.; Janoschka, T.; Morgenstern, S.; Hagemann, T.; Muench, S.; Hauffman, G.; Gohy, J.-F.; Hager, M. D.; Schubert, U. S., Poly(TEMPO)/Zinc Hybrid-Flow Battery: A Novel, “Green,” High Voltage, and Safe Energy Storage System. Advanced Materials 2016, 28 (11), 2238-2243. 39. Leung, P.; Martin, T.; Shah, A.; Anderson, M.; Palma, J., Membrane-less organic– inorganic aqueous flow batteries with improved cell potential. Chemical Communications 2016, 52 (99), 14270-14273. 40. Kwabi, D. G.; Lin, K.; Ji, Y.; Kerr, E. F.; Goulet, M.-A.; De Porcellinis, D.; Tabor, D. P.; Pollack, D. A.; Aspuru-Guzik, A.; Gordon, R. G., Alkaline Quinone Flow Battery with Long Lifetime at pH 12. Joule 2018. 41. Orita, A.; Verde, M. G.; Sakai, M.; Meng, Y. S., A biomimetic redox flow battery based on flavin mononucleotide. Nature communications 2016, 7, 13230. 42. Hollas, A.; Wei, X.; Murugesan, V.; Nie, Z.; Li, B.; Reed, D.; Liu, J.; Sprenkle, V.; Wang, W., A biomimetic high-capacity phenazine-based anolyte for aqueous organic redox flow batteries. Nature Energy 2018, 3 (6), 508-514. 43. Lin, K.; Gómez-Bombarelli, R.; Beh, E. S.; Tong, L.; Chen, Q.; Valle, A.; Aspuru-Guzik, A.; Aziz, M. J.; Gordon, R. G., A redox-flow battery with an alloxazine-based organic electrolyte. Nature Energy 2016, 1, 16102. 44. Wei, L.; Zhao, T. S.; Xu, Q.; Zhou, X. L.; Zhang, Z. H., In-situ investigation of hydrogen evolution behavior in vanadium redox flow batteries. Appl. Energy 2017, 190, 1112-1118. 32 45. Eyer, J.; Corey, G. Energy Storage for the Electricity Grid: Benefits and Market Potential Assessment Guide; Sandia National Laboratories: 2010 February. , 2010 February 46. Hruska, L.; Savinell, R., Investigation of Factors Affecting Performance of the Iron-Redox Battery. J. Electrochem. Soc. 1981, 128 (1), 18-25. 47. Hawthorne, K. L.; Petek, T. J.; Miller, M. A.; Wainright, J. S.; Savinell, R. F., An Investigation into Factors Affecting the Iron Plating Reaction for an All-Iron Flow Battery. J. Electrochem. Soc. 2015, 162 (1), A108-A113. 48. Tucker, M. C.; Phillips, A.; Weber, A. Z., All-Iron Redox Flow Battery Tailored for Off- Grid Portable Applications. ChemSusChem 2015, 8 (23), 3996-4004. 49. Manohar, A. K.; Kim, K. M.; Plichta, E.; Hendrickson, M.; Rawlings, S.; Narayanan, S. R., A High Efficiency Iron-Chloride Redox Flow Battery for Large-Scale Energy Storage. J. Electrochem. Soc. 2016, 163 (1), A5118-A5125. 50. Inc, E. S. S., Technical White Paper, All-Iron Flow Battery – Overview: 2017 June. 51. http://chem.libretexts.org. 52. Hawthorne, K. L.; Wainright, J. S.; Savinell, R. F., Studies of iron-ligand complexes for an all-iron flow battery application. J. Electrochem. Soc. 2014, 161 (10), A1662-A1671. 53. Gong, K.; Xu, F.; Grunewald, J. B.; Ma, X.; Zhao, Y.; Gu, S.; Yan, Y., All-Soluble All- Iron Aqueous Redox-Flow Battery. ACS Energy Lett. 2016, 1 (1), 89-93. 54. Moy, D.; Narayanan, S., Mixed conduction membranes suppress the polysulfide shuttle in lithium-sulfur batteries. Journal of The Electrochemical Society 2017, 164 (4), A560-A566. 55. Herbert, D.; Ulam, J., US Patent 3,043,896 (1962). There is no corresponding record for this reference 1962. 56. Fang, R.; Zhao, S.; Sun, Z.; Wang, D.-W.; Cheng, H.-M.; Li, F., More Reliable Lithium- Sulfur Batteries: Status, Solutions and Prospects. Advanced Materials 2017, 29 (48), 1606823. 57. Moy, D. Advancing High Energy Lithium-Sulfur Batteries – Understanding and Solving Key Degradation Issues Affecting Cycle Life. University of Sourthern California, 2017. 58. Moy, D.; Manivannan, A.; Narayanan, S., Direct measurement of polysulfide shuttle current: A window into understanding the performance of lithium-sulfur cells. Journal of the electrochemical society 2015, 162 (1), A1-A7. 33 59. Jones, J. P.; Prakash, G.; Olah, G. A., Electrochemical CO 2 reduction: recent advances and current trends. Israel Journal of Chemistry 2014, 54 (10), 1451-1466. 60. Agency, U. S. E. P., 2018. 61. White, J. L.; Baruch, M. F.; Pander, J. E.; Hu, Y.; Fortmeyer, I. C.; Park, J. E.; Zhang, T.; Liao, K.; Gu, J.; Yan, Y.; Shaw, T. W.; Abelev, E.; Bocarsly, A. B., Light-Driven Heterogeneous Reduction of Carbon Dioxide: Photocatalysts and Photoelectrodes. Chemical Reviews 2015, 115 (23), 12888-12935. 62. Hwang, H.; Yeon, Y. J.; Lee, S.; Choe, H.; Jang, M. G.; Cho, D. H.; Park, S.; Kim, Y. H., Electro-biocatalytic production of formate from carbon dioxide using an oxygen-stable whole cell biocatalyst. Bioresource technology 2015, 185, 35-39. 63. Jitaru, M., Electrochemical carbon dioxide reduction-fundamental and applied topics. Journal of the University of chemical Technology and Metallurgy 2007, 42 (4), 333-344. 64. Kuhl, K. P.; Cave, E. R.; Abram, D. N.; Jaramillo, T. F., New insights into the electrochemical reduction of carbon dioxide on metallic copper surfaces. Energy & Environmental Science 2012, 5 (5), 7050-7059. 65. Mondal, B.; Song, J.; Neese, F.; Ye, S., Bio-inspired mechanistic insights into CO 2 reduction. Current opinion in chemical biology 2015, 25, 103-109. 66. Zhang, W.; Hu, Y.; Ma, L.; Zhu, G.; Wang, Y.; Xue, X.; Chen, R.; Yang, S.; Jin, Z., Progress and perspective of electrocatalytic CO 2 reduction for renewable carbonaceous fuels and chemicals. Advanced Science 2018, 5 (1), 1700275. 67. Rusching, U.; Müller, U.; Willnow, P.; Höpner, T., CO 2 reduction to formate by NADH catalysed by formate dehydrogenase from Pseudomonas oxalaticus. The FEBS Journal 1976, 70 (2), 325-330. 68. Marken, F.; Fermin, D., Electrochemical Reduction of Carbon Dioxide: Overcoming the Limitations of Photosynthesis. Royal Society of Chemistry: 2018; Vol. 21. 69. Sullivan, B. P.; Krist, K.; Guard, H., Electrochemical and electrocatalytic reactions of carbon dioxide. Elsevier: 2012. 70. Ting, L. R. L.; Yeo, B. S., Recent Advances in Understanding Mechanisms for the Electrochemical Reduction of Carbon Dioxide. Current Opinion in Electrochemistry 2018. 34 71. Baruch, M. F.; Pander III, J. E.; White, J. L.; Bocarsly, A. B., Mechanistic insights into the reduction of CO 2 on tin electrodes using in situ ATR-IR spectroscopy. ACS Catalysis 2015, 5 (5), 3148-3156. 72. Dutta, A.; Kuzume, A.; Rahaman, M.; Vesztergom, S.; Broekmann, P., Monitoring the chemical state of catalysts for CO 2 electroreduction: an in operando study. ACS Catalysis 2015, 5 (12), 7498-7502. 73. Back, S.; Yeom, M. S.; Jung, Y., Active sites of Au and Ag nanoparticle catalysts for CO 2 electroreduction to CO. Acs Catalysis 2015, 5 (9), 5089-5096. 74. Apaydin, D. H.; Schlager, S.; Portenkirchner, E.; Sariciftci, N. S., Organic, Organometallic and Bioorganic Catalysts for Electrochemical Reduction of CO 2. ChemPhysChem, n/a-n/a. 75. Roldan, A.; Hollingsworth, N.; Roffey, A.; Islam, H.-U.; Goodall, J.; Catlow, C.; Darr, J.; Bras, W.; Sankar, G.; Holt, K., Bio-inspired CO 2 conversion by iron sulfide catalysts under sustainable conditions. Chemical Communications 2015, 51 (35), 7501-7504. 76. Amao, Y.; Watanabe, T., Photochemical and enzymatic methanol synthesis from HCO 3 − by dehydrogenases using water-soluble zinc porphyrin in aqueous media. Applied Catalysis B: Environmental 2009, 86 (3–4), 109-113. 77. Amao, Y.; Shuto, N.; Furuno, K.; Obata, A.; Fuchino, Y.; Uemura, K.; Kajino, T.; Sekito, T.; Iwai, S.; Miyamoto, Y.; Matsuda, M., Artificial leaf device for solar fuel production. Faraday Discussions 2012, 155 (0), 289-296. 78. Amao, Y.; Ikeyama, S., Discovery of the Reduced Form of Methylviologen Activating Formate Dehydrogenase in the Catalytic Conversion of Carbon Dioxide to Formic Acid. Chemistry Letters 2015, 44 (9), 1182-1184. 79. Blanchard, J. S.; Cleland, W., Kinetic and chemical mechanisms of yeast formate dehydrogenase. Biochemistry 1980, 19 (15), 3543-3550. 80. Srikanth, S.; Alvarez Gallego, Y.; Vanbroekhoven, K.; Pant, D., Enzymatic Electrosynthesis of Formic Acid through CO 2 Reduction in Bioelectrochemical System (BES): Effect of Immobilization and Carbonic Anhydrase Addition. ChemPhysChem 2017. 35 81. Kuwabata, S.; Tsuda, R.; Yoneyama, H., Electrochemical conversion of carbon dioxide to methanol with the assistance of formate dehydrogenase and methanol dehydrogenase as biocatalysts. Journal of the American Chemical Society 1994, 116 (12), 5437-5443. 82. Srikanth, S.; Maesen, M.; Dominguez-Benetton, X.; Vanbroekhoven, K.; Pant, D., Enzymatic electrosynthesis of formate through CO 2 sequestration/reduction in a bioelectrochemical system (BES). Bioresource technology 2014, 165, 350-354. 83. Ceccaldi, P.; Schuchmann, K.; Müller, V.; Elliott, S. J., The hydrogen dependent CO 2 reductase: the first completely CO tolerant FeFe-hydrogenase. Energy & Environmental Science 2017. 84. Schuchmann, K.; Müller, V., Direct and Reversible Hydrogenation of CO to Formate by a Bacterial Carbon Dioxide Reductase. Science 2013, 342 (6164), 1382-1385. 85. Drennan, C. L.; Heo, J.; Sintchak, M. D.; Schreiter, E.; Ludden, P. W., Life on carbon monoxide: X-ray structure of Rhodospirillum rubrum Ni-Fe-S carbon monoxide dehydrogenase. Proceedings of the National Academy of Sciences 2001, 98 (21), 11973-11978. 86. Dobbek, H.; Svetlitchnyi, V.; Gremer, L.; Huber, R.; Meyer, O., Crystal structure of a carbon monoxide dehydrogenase reveals a [Ni-4Fe-5S] cluster. Science 2001, 293 (5533), 1281- 1285. 87. Olah, G. A.; Goeppert, A.; Prakash, G. S., Beyond oil and gas: the methanol economy. John Wiley & Sons: 2011. 88. Narayanan, S.; Haines, B.; Soler, J.; Valdez, T., Electrochemical conversion of carbon dioxide to formate in alkaline polymer electrolyte membrane cells. Journal of The Electrochemical Society 2011, 158 (2), A167-A173. 89. Blumenfeld, L. A.; Tikhonov, A. N., Principles of Enzyme Catalysis. In Biophysical Thermodynamics of Intracellular Processes, Springer: 1994; pp 86-111. 90. Peacock, D.; Boulter, D., Kinetic studies of formate dehydrogenase. Biochemical Journal 1970, 120 (4), 763-769. 91. Kim, S.; Kim, M. K.; Lee, S. H.; Yoon, S.; Jung, K.-D., Conversion of CO 2 to formate in an electroenzymatic cell using Candida boidinii formate dehydrogenase. Journal of Molecular Catalysis B: Enzymatic 2014, 102, 9-15. 36 92. Mathews, M. B.; Vennesland, B., Enzymic oxidation of formic acid. Journal of Biological Chemistry 1950, 186 (2), 667-682. 93. Davison, D. C., Studies on plant formic dehydrogenase. Biochemical Journal 1951, 49 (4), 520. 94. Zhang, L.; Ong, J.; Liu, J.; Li, S. F. Y., Enzymatic electrosynthesis of formate from CO 2 reduction in a hybrid biofuel cell system. Renewable Energy 2017, 108, 581-588. 95. Reda, T.; Plugge, C. M.; Abram, N. J.; Hirst, J., Reversible interconversion of carbon dioxide and formate by an electroactive enzyme. Proceedings of the National Academy of Sciences 2008, 105 (31), 10654-10658. 96. Thauer, R. K.; Fuchs, G.; Jungermann, K., Role of iron-sulfur proteins in formate metabolism. Iron-sulfur proteins 1977, 3, 121-156. 97. Lee, S. Y.; Lim, S. Y.; Seo, D.; Lee, J. Y.; Chung, T. D., Light-Driven Highly Selective Conversion of CO 2 to Formate by Electrosynthesized Enzyme/Cofactor Thin Film Electrode. Advanced Energy Materials 2016, 6 (11). 98. Guo, Q.; Gakhar, L.; Wickersham, K.; Francis, K.; Vardi-Kilshtain, A.; Major, D. T.; Cheatum, C. M.; Kohen, A., Structural and Kinetic Studies of Formate Dehydrogenase from Candida boidinii. Biochemistry 2016, 55 (19), 2760-2771. 99. Schirwitz, K.; Schmidt, A.; Lamzin, V. S., High-resolution structures of formate dehydrogenase from Candida boidinii. Protein Science 2007, 16 (6), 1146-1156. 100. Nilov, D. K.; Shabalin, I. G.; Popov, V. O.; Švedas, V. K., Molecular modeling of formate dehydrogenase: the formation of the Michaelis complex. Journal of Biomolecular Structure and Dynamics 2012, 30 (2), 170-179. 101. Schiøtt, B.; Zheng, Y.-J.; Bruice, T. C., Theoretical investigation of the hydride transfer from formate to NAD + and the implications for the catalytic mechanism of formate dehydrogenase. Journal of the American Chemical Society 1998, 120 (29), 7192-7200. 102. Choe, H.; Ha, J. M.; Joo, J. C.; Kim, H.; Yoon, H.-J.; Kim, S.; Son, S. H.; Gengan, R. M.; Jeon, S. T.; Chang, R., Structural insights into the efficient CO 2-reducing activity of an NAD- dependent formate dehydrogenase from Thiobacillus sp. KNK65MA. Acta Crystallographica Section D: Biological Crystallography 2015, 71 (2), 313-323. 37 103. Ihara, M.; Kawano, Y.; Urano, M.; Okabe, A., Light driven CO 2 fixation by using cyanobacterial photosystem I and NADPH-dependent formate dehydrogenase. PloS one 2013, 8 (8), e71581. 104. Lamzin, V. S.; Dauter, Z.; Popov, V. O.; Harutyunyan, E. H.; Wilson, K. S., High Resolution Structures of Holo and Apo Formate Dehydrogenase. Journal of Molecular Biology 1994, 236 (3), 759-785. 105. Jensen, M. A.; Elving, P. J., Nicotinamide adenine dinucleotide (NAD + ). Formal potential of the NAD+/NAD· couple and NAD · dimerization rate. Biochimica et Biophysica Acta (BBA)- Bioenergetics 1984, 764 (3), 310-315. 106. Yun, S.-E.; Taya, M.; Tone, S., Direct reduction of NAD + by electrochemical procedure and application of the regenerated NADH to enzyme reaction. Biotechnology Letters 1994, 16 (10), 1053-1058. 38 39 Improvements to the Coulombic Efficiency of the Iron Electrode for an All-Iron Redox Flow Battery 2.1 Background In this Chapter, we provide further discussion of our publication 1 and our presentation at the 231 st Electrochemical Society Meeting in New Orleans, Louisiana. 2 Reliable, low-cost, durable and eco-friendly methods of storing electrical energy are essential for utilizing the energy generated by large-scale solar photovoltaic arrays and wind turbines. We discussed in Chapter One that among the various electrochemical redox flow battery technologies the all iron-redox flow battery is particularly promising because of its low-cost, environmental safety, high energy content and scalability. 3, 4 The major problem of all-iron redox flow cell is the parasitic hydrogen evolution which results in significant limitations for the long- term operation of the system. This Chapter describes the impact of the parasitic hydrogen evolution and discusses several approaches to mitigate this problem. 2.1.1 The Implications of Parasitic Hydrogen Evolution The reactions at the negative electrode of the all-iron redox battery is as follows (Eq. 2-1). The hydrogen evolution reaction (Eq.2-2) also occurs in the same range of potentials. Fe 2+ + 2e − ⇆ Fe 0 E 0 = −0.44 V (2-1) H 3O + + e - à ½ H 2 + H 2O E o = 0 V (2-2) As per the reversible electrode potential values, hydrogen will be evolved not only during charging but also during open-circuit stand leading to the corrosion of iron from a charged negative electrode. 5 Most importantly, hydrogen evolution affects the repeated cycling of the redox flow cell by rapidly changing the composition of the electrolyte in the system (Figure 2- 1). 6 As protons are converted to hydrogen the pH of the solution increases. The result is an 40 increase in hydroxide ion concentration. The imbalance of hydroxide ions between the positive and negative side results in diffusion of the hydroxide ions through the anion exchange membrane from the negative to the positive side of the cell (in Figure 2-1, Step 1C). Thus, as hydrogen evolution occurs the pH of the negative and positive side will continue to increase (Figure 2-1 step 2A), finally resulting in the precipitation of iron hydroxides (Figure 2-1). Further, a charging process with the low faradaic efficiencies at the negative electrode leads to a continuous buildup of iron (III) ions in the positive electrolyte leading to a capacity imbalance between the positive and negative sides of the cell (steps 2B and 3B). Figure 2-1: Effects of the hydrogen evolution during the operation of all-iron redox flow battery In Table 2-1 we project the impact of hydrogen evolution on the rechargeability of the all-iron redox flow battery for various assumed values of coulombic efficiencies. For this calculation, we start with a 2 M solution of iron (II) chloride and simulate the effect of charge and discharge at 41 various values of coulombic efficiency. Then we apply the criteria of capacity imbalance of more than 20% or pH > 3 as the point at which the battery becomes inoperable as a rechargeable system; at pH values greater than 3, precipitation of the hydroxides will start to occur. These simulations were conducted using the relationships in Eqs. 2-3 and 2-4. Number of cycles limited by the pH increase = [(10 -initial pH -10 -pH of precipitation )]/[0.8 *C t*3600*(1-α)/F] (2-3) Number of cycles limited by the capacity fade = [0.25/(1 − α)] (2-4) Where α = coulombic efficiency, F = faraday constant and Ct = total capacity Table 2-1: Predicted performance of the all-iron flow battery with coulombic efficiency losses due to HER. Calculations based on Eq. 2-3 and 2-3 indicate that even when the coulombic efficiency values are as high as 95%, just in a few cycles, the cell operation goes out of capacity balance or leads to precipitation of the hydroxides (Table 2-1). Thus, for the continuous operation of this battery Assumed Coulombic efficiency % Number of Achievable Charge/ Discharge Cycles Criteria 1: Capacity imbalance >20% Criteria 2: pH increased to 3 95.00 5 2 96.00 6 3 97.00 8 4 98.00 13 6 99.00 25 12 99.99 2500 1238 Total theoretical capacity for 2 M FeCl 2 in positive electrolyte=53.6 Ah L -1 ; total volume in both chambers=2 L; starting pH of the electrolyte =1; pH at which precipitation is projected to occur=3. 42 for thousands of cycles without onerous capacity re-balancing schemes and re-adjustment of pH, we must suppress hydrogen evolution completely at the negative electrode. The rate of hydrogen evolution is governed by pH and electrolyte composition (in the bulk and near the surface of the electrode), adsorption of substances, temperature, electrode potential and the specific mechanism of hydrogen evolution operative at the iron electrode. 7 Investigation of the kinetics and mechanism of hydrogen evolution from practical solutions used in the all-iron flow battery is challenging because the effect of the foregoing factors is difficult to separate. 5 In studies from our group, Manohar et al. had demonstrated the benefit of raising the bulk pH and using ascorbic acid as a complexing additive to prevent precipitation of iron hydroxides. 8 In this chapter, we report new findings relating to: (i) the specific effect of ascorbic acid on the overpotential for hydrogen evolution, (ii) the effect of pH near the surface of the electrode on the coulombic efficiency, (iii) the role of ascorbic acid in buffering the pH near the surface of the electrode, (iv) the effect of flow rate on the coulombic efficiency, and (v) the relative effect of temperature on the kinetics of the iron deposition reaction and the hydrogen evolution reaction. 2.2 Experimental Methods 2.2.1 Coulombic Efficiency of Iron Deposition and Dissolution To study the coulombic efficiency of iron deposition and dissolution under various operating conditions of pH, current density, temperature and electrolyte composition, we used a three- electrode half-cell (Figure 2-2). The working electrode (area = 25 cm 2 ) was made of impervious graphite (Graphitestore, catalog number GT001672) to mimic an electrode substrate with high hydrogen overpotential that will be used in an all-iron flow battery. We used a mild steel mesh as the counter electrode and a silver/silver chloride reference electrode. The use of a mild-steel counter electrode ensured that the composition of the solution and pH stays relatively constant because mild steel either underwent dissolution to iron (II) under anodic conditions, or the electrodeposition of iron occurred on mild steel during cathodic conditions. We avoided the use of a platinum counter electrode as we generate Fe (III) during anodic conditions and evolve hydrogen during cathodic conditions changing the pH. Thus, using a platinum counter electrode 43 results in the change of bath composition. The electrolyte solution consisted of iron (II) chloride (3.25 M) with ammonium chloride (2 M) and ascorbic acid (0.3 M). Ammonium hydroxide was used to adjust the pH of the electrolyte. Figure 2-2: Experimental set-up. The coulombic efficiency of electrodeposition of iron on a graphite-working electrode was determined over the current density range of 4–100 mA/ cm 2 . Iron was electrodeposited by passing a fixed amount of charge of 60 Coulombs at various current densities of interest. The actual amount of iron that was electrodeposited was determined by anodic stripping at 20 mA/ cm 2 . The coulombic efficiency was calculated, from the ratio of the charge delivered during the anodic stripping step, Q delivered, to the charge uptake during electrodeposition of iron, Q loaded, as per Eq. 2-5. % Coulombic efficiency = (Q delivered / Q loaded) *100 (2-5) The effect of flow rate (or stirring rate) of the electrolyte on coulombic efficiency was studied for an electrolyte solution that did not contain ascorbic acid. The half-cell experiments were performed using a potentiostat (VMC-4, Princeton Applied Research). During preparation and testing, the electrolyte was de-aerated using argon gas to prevent the air oxidation of Fe (II). Solution of iron (II) chloride and additives 44 The electrolyte was stirred using a magnetic stirrer bar during the electrodeposition of iron, to avoid the limitations from the mass transport of iron (II). 2.2.2 Investigations of the Hydrogen Evolution Reaction To investigate the rate of hydrogen evolution in different electrolytes we conducted polarization studies on an electrodeposited iron surface that mimicked the negative electrode in an all-iron flow battery. To study the effect of cadmium on the surface of the iron electrode, we used a cadmium electrode prepared by electrodeposition from an electrolyte containing cadmium (II) chloride. Other additives of interest were dissolved in the electrolyte, as required. Potentiodynamic polarization studies at very slow scan rates (5 mV/s) were carried out in a three- electrode cell configuration. The working electrode for this experiment was the relevant metal- covered graphite electrode as described in the foregoing. To study an iron-covered electrode, iron was electrodeposited with 120 Coulombs on a graphite sheet at 60 ◦ C using the same electrolyte described in the section 2.2.1 (Coulombic efficiency of Iron deposition and Dissolution) at a pH value of 3.0. Similarly, for studying a cadmium surface, we electrodeposited cadmium on a graphite sheet at a current density of 40 mA cm -2 by passing 120 Coulombs from a 0.1 M solution of cadmium chloride in 1 N hydrochloric acid. These electrodes appeared to have a smooth surface after electrodeposition. The electrodes were rinsed thoroughly with de- ionized water prior to the hydrogen evolution studies. For the hydrogen evolution studies, we used a platinum wire counter electrode and a silver/silver chloride reference electrode. In these studies, the electrolyte solution was free of the iron or cadmium ions to avoid electrodeposition of the metal; the solution consisted of just ammonium chloride (2.0 M) and ascorbic acid (0.3 M) at pH = 3. We used a pure iron rod (Alfa Aesar, catalog number 40500) as a working electrode to study the effect of pH on HER in the presence of ascorbic acid (0.3 M) in ammonium chloride (2 M) electrolyte. 45 2.2.3 Correction of Electrode Potentials for Ohmic Drop All the electrode potentials reported here have been corrected for the potential drop across the ohmic resistance between the working electrode and the reference electrode. The ohmic resistance value used for the correction was determined from the real component of the electrochemical impedance at 100 kHz, obtained from an impedance spectroscopy experiment. 2.2.4 Impedance Spectroscopy at Different Temperatures Electrochemical impedance under potentiostatic conditions was measured over the frequency range of 100 kHz to 5 Hz for a pure iron rod working electrode at −0.85 V vs. Ag/AgCl for different temperatures in the range of 25 ◦ −60 ◦ C using the electrolyte described in the section 2.2.1 (Coulombic Efficiency of Iron Deposition and Dissolution) at pH = 3. 2.2.5 Near-surface pH and Bulk pH Measurement The pH close to the electrode surface and the pH of the bulk electrolyte during charging, discharging and potentiodynamic polarization studies were measured using two pH-sensing glass electrodes (Thermo Scientific Orion- ROSS Ultra pH/ATC Triode). We have referred to the pH close to the electrode surface as “near-surface pH” in the rest of the manuscript. 46 2.3 Results and Discussion 2.3.1 The Effect of Electrolyte pH Figure 2-3: (a) Effect of electrolyte pH on the coulombic efficiency of iron electrodeposition and electro dissolution. (b) Effect of electrolyte pH on electrode potential of iron electrode during electrodeposition; bath conditions (iron (II) chloride (3.0 M), ammonium chloride (2.0 M), ascorbic acid (0.3 M) at temperature = 25 ◦ C). The coulombic efficiency during electrodeposition of iron increased with increasing pH at all values of current density (Figure 2-3a). Based on the increase of coulombic efficiency from 85% to 93.6% at 20 mA/ cm 2 when the electrolyte pH is raised from 0 to 3.15, we calculated a -850 -750 -650 -550 1 10 100 Potential, V vs. NHE Charging Current Density, mA/ cm 2 (b) pH =3.15 pH =3.0 pH =0.0 70 75 80 85 90 95 100 0 5 10 15 20 25 % Coulombic Efficiency Charging Current density, mA/cm 2 pH =3.15 pH =3.0 pH =0.0 (a) 47 decrease of 55% in the rate of hydrogen evolution corresponding to this increase of coulombic efficiency. These results are consistent with previous reports of high coulombic efficiency at pH values greater than 2. 9, 10 Upon changing the pH value from zero to 3.15 the equilibrium potential for the hydrogen evolution reaction (as calculated from the Nernst Equation) shifts from 0.0 V to −0.185 V vs. NHE. Consequently, at any operating electrode potential for iron deposition, the overpotential applied to the hydrogen evolution reaction would also be reduced by 0.185 V when the pH at the electrode/solution interface is 3.15 compared to a pH value of 0. Based on the electrode potential values observed during iron deposition (Figure 2-3b), we have calculated the overpotential applied to the hydrogen evolution reaction at various values of pH (Table 2-2). Table 2-2: Electrode potential and HER overpotential at 20 mA/ cm 2 at different values of pH. pH of the electrolyte Electrode potential, V vs. NHE Equilibrium potential for hydrogen evolution, V vs. NHE Calculated overpotential for hydrogen evolution, V 0.00 -0.771 0.000 -0.771 3.00 -0.611 -0.177 -0.434 3.15 -0.618 -0.186 -0.432 From the polarization data presented in Table 2-2 we can conclude that the overpotential applied to the hydrogen evolution decreased by 0.337 V by changing the pH from 0 to 3. Further, based on the value of Tafel slope for hydrogen evolution on an iron surface of 0.180 V/decade (Figure 2-4b) we could expect this decrease of overpotential of 0.337 V to result in a reduction in the value of hydrogen current by a factor of 10 (0.337/Tafel slope) , which is approximately a seventy-five-fold reduction in HER current. It is reasonable to assume that any change in roughness of the surface between pH = 0 and pH = 3 affects the hydrogen evolution and iron deposition reaction rates in the same proportion. However, we only obtain a factor of 2 reduction in current and an increase of coulombic efficiency of just 5.9% from pH = 0 to pH = 3 at 20 mA/ cm 2 . While this may seem surprising, this apparent inconsistency could be attributed 48 to the effect of adsorption of the electrolyte additives, particularly ascorbic acid on the kinetics of hydrogen evolution at the iron electrode. We could expect the effect of adsorption of the electrolyte additives to overshadow the benefit from the shift of equilibrium potentials. We discuss this effect in detail in the following. 2.3.2 The Effect of Ascorbic Acid on the Kinetics of Hydrogen Evolution Ascorbic acid has been used as an electrolyte additive in iron plating because ascorbic acid is a strong two-electron reducing agent. Thus, the presence of ascorbic acid served to minimize the air oxidation of iron (II) to iron (III). The formation of iron (III) in the negative electrolyte must be avoided as it will reduce the coulombic efficiency by electrochemical conversion to iron (II) and thus competing with the electrodeposition of iron. 11 Spinelli et al. studied ascorbic acid as a mixed type corrosion inhibitor on mild steel in solutions of pH in the range of 2 to 4. 12 Ascorbic acid at pH 3 exists as a mixture of the protonated and de- protonated forms and the relative composition of these forms will vary with pH. 11, 12 Both ascorbic acid and its oxidized form, dehydroascorbic acid, have been shown to adsorb on the iron surface facilitated by the presence of polar groups and pi- electrons. 12 The formation of adsorbed hydrogen, a required intermediate step for hydrogen evolution, will now have to compete with the adsorption of ascorbic acid. Consequently, we can expect the kinetics of hydrogen evolution to be hindered in the presence of ascorbic acid. The degree of inhibition will depend on the relative coverage of the surface by ascorbic acid and hydrogen atoms. When the pH is increased from 0 to 3.15, we could expect a decreased adsorption coverage because ascorbic acid would be ionized into a negatively charged species. Such species will be repelled from the surface of the electrode. Thus, ascorbic acid is not likely to have the same strong inhibitory effect at pH = 3.15 as it would at pH = 0. A decrease of electrode potential during iron deposition at 10 mA/ cm 2 with increasing pH from 0 to 3.15 in the presence of ascorbic acid is consistent with the reduced adsorption of ascorbic acid (Figure 2-3b). Further, when we compare the overpotentials for hydrogen evolution at pH = 0 and pH = 3 in the range of 49 potentials for iron deposition (Figure 2-4b), we find a significantly higher overpotential for hydrogen evolution at pH = 0 compared to pH = 3. Figure 2-4: (a) Potentiodynamic polarization studies of iron deposition from a solution of iron(II) chloride(3.0 M), ammonium chloride (2.0 M), T = 25 ◦ C in the presence and absence of ascorbic acid (0.3 M) at various values of pH as shown (AA = Ascorbic acid) and (b) Overpotential for HER on iron at pH 0 and 3 in the presence or absence of ascorbic acid (0.3 M) and ammonium chloride (2.0 M). -0.80 -0.70 -0.60 -0.50 -0.40 -5 -4 -3 -2 -1 Potential, V vs. NHE log (Current Density), current density in A/ cm 2 (a) pH =3.0 with AA pH =0.0 with AA pH =3.0 pH =1.66 -1.00 -0.80 -0.60 -0.40 -0.20 0.00 -5 -4 -3 -2 -1 0 Overpotential, V log (Current denisty) current density in A/ cm 2 pH =3.0 with AA pH =0.0 with AA pH =0.0 pH =3.0 (b) iron deposition range at pH =3.0 iron deposition range at pH =0.0 50 These observations suggest that in spite of the shift of the equilibrium potential for hydrogen evolution when we change the pH from 0 to 3, the benefit from a reduction of the applied overpotential is not realized because of the relatively low baseline resulting from the inhibition of hydrogen evolution by the adsorption of ascorbic acid at pH = 0. Consequently, the net effect observed by changing the pH from 0 to 3 was reduced to just a two-fold reduction in hydrogen evolution current at a cathodic current density of 20 mA/cm 2 . 2.3.3 The Effect of Ascorbic Acid on the Kinetics of Electrodeposition of Iron Ascorbic acid forms an intensely green complex with iron (II) that is quite stable even on exposure to oxygen. 13 The color of this complex changes from green to violet and then to black upon increase of pH. 14 Since the complexation of ascorbic acid with iron is facilitated at pH = 3 by the formation of the mono-ascorbate anion, the overpotential for metal electrodeposition will be increased. Thus, the increase of pH in the presence of ascorbic acid can be expected to have a significant effect on the kinetics of electrodeposition of iron. In the absence of ascorbic acid in the electrolyte, iron deposition kinetics is not affected significantly by the change of pH from 0 to 3 (Figure 2-4a). The polarization curves at various pH values fall right on top each other. This observation is not surprising as the equilibrium potential of the Fe(II)/ Fe(0) couple is not expected to change with pH until precipitation of the hydroxides begins to occur. When 0.3 M ascorbic acid is added to the iron chloride electrolyte, the electrode potentials during iron deposition shifted to more negative values. At pH = 3, at a current density of 10 mA/ cm 2 , the shift in electrode potential was 41 mV. However, at pH = 0 upon addition of ascorbic acid, the shift increased further to 95 mV. Thus, it was clear that ascorbic acid impacted the kinetics of the iron deposition reaction to different extents depending on the pH value of the electrolyte. Thus, the coulombic efficiency at pH = 0 is dominated by the effect of adsorption of ascorbic acid on the kinetics of hydrogen evolution and iron deposition. However, at pH = 3 51 and 3.15, we expect complexation reactions of ascorbate with iron to play a significant role (described here below). Ascorbic acid can interfere with the electrodeposition process by complexing with iron (II) and through the formation of adsorbed layers. The effect of complexation is two-fold: (i) shifting the equilibrium potential to more negative values and (ii) hindering of the charge transfer process by imposing an additional kinetic barrier of “de-complexation” before ion to atom transformation occurs during electrodeposition. Based on the formation constants of iron (II) ascorbate complexes, we expect the shift of equilibrium potential at pH 3 to be about 35 mV (see Appendix A1). 11 The observed shift in electrode potential at pH = 3 at 10 mA/cm 2 is consistent with this shift in equilibrium potential. At pH = 0, ascorbic acid being unionized (pKa = 4.17), complex formation is not significant, and we do not expect a shift in the equilibrium potential for the iron deposition reaction. However, at pH = 0 unionized ascorbic acid, as mentioned earlier, is reported to form adsorbed layers on the iron electrode. 12, 15 Such adsorption will reduce the exchange current density and lead to polarization losses (Figure 2- 4b). Consequently, we attributed the shift of electrode potential by 95 mV during electrodeposition of iron at pH 0 and 10 mA/cm 2 to the hindered kinetics resulting from the adsorption of ascorbic acid on the iron surface. 2.3.4 Effect of Current Density on Coulombic Efficiency We observed that increasing the charging current density also improved the coulombic efficiency (Figure 2-3a). However, this effect was more significant at pH = 3 than at pH = 0. During the electrodeposition of iron, the concentration of iron (II) in the diffusion layer adjacent to the surface of the electrode is lower than in the bulk simply because of the concentration gradient required for mass transport. This lowered surface concentration of iron (II), resulted in a lower extent of hydrolysis of iron (II) as per Eq. 2-6. The extent of hydrolysis of iron (II) can be calculated using the activity of each of the species (a x, x = Fe(OH) + , H 3O + and Fe 2+ ) using Eq. 2- 7. 52 Fe 2+ + H 2O ⇆ Fe(OH) + + H + (2-6) K h= aFe(OH)+ * aH3O+/ aFe2+ (2-7) Consequently, because of lower extent of hydrolysis, the pH at the surface of the electrode during electrodeposition will be higher than the bulk value. Such an increase in pH will result in the reduction in the hydrogen evolution current at any given overpotential and thus an improved coulombic efficiency. Further, at a moderately higher value of pH, both Fe 2+ and Fe(OH) + species would be present in significant concentrations. Dahms et al. have reported that the local pH increase at the electrode surface can influence the charge transfer process by involving Fe(OH) + in the rate determining step. 16 We have verified experimentally that an increase in near-surface pH occurs with increasing charging current densities (Figure 2-5). The near-surface pH can rise to as high as 6 at 100 mA/cm 2 while the bulk pH is still lower than 4. We can expect this rise in near-surface pH with current density to be affected by the diffusion layer thickness which in turn is dependent on the stirring or agitation rate. We have estimated the change in pH in the diffusion layer at any current density using the Nernst diffusion layer model and the hydrolysis constant (Eq.2-8) (see Appendix A2 for the derivation). Near-surface pH = - (1/2) log[(C * Fe(II) - (Iα d / n F D)) K h ] (2-8) where Kh = hydrolysis constant, δ = diffusion layer thickness, D = diffusion coefficient of iron(II), I = current density, α = coulombic efficiency for iron deposition, F = Faraday constant and C ∗ Fe(II) = bulk formal concentration of Fe(II) at a given pH. 53 Figure 2-5: Experimental (♦,) and simulated (Ú,+) near-surface pH and experimental bulk pH (▲,¢) deviation during potentiodynamic polarization of metal electrodeposition using an electrolyte composed with iron (II) chloride (3.25 M) and ammonium chloride (2.0 M). Using Eq. 2-8 and assuming a diffusion layer thickness of 0.62 mm, we can reproduce the change in near-surface pH with current density (Figure 2-5). This assumed value of diffusion layer thickness is reasonable to expect under the stirring conditions of the deposition experiment. Since the diffusion layer thickness decreases with increasing flow velocity, we can expect that the extent of stirring of the electrolyte to be an important factor in determining the observed charging efficiencies (see Appendix A3) for estimate of diffusion layer thickness as a function of flow velocity. In the experiments where we changed the stirring rate, we did observe for charging cur- rent densities of 50 and 100 mA/cm 2 that the coulombic efficiency of iron deposition decreased with increasing stirring rate (Figure 2-6). This observation also verified that the coulombic efficiency at a given current density can be a lower value at higher stirring rates due to the decreased diffusion layer thickness. Based on these observations and the simulation we can attribute with confidence the improvements in coulombic efficiency with charging current density to the increase of near-surface pH. Thus, flow velocity or stirring rate is an important factor in determining the coulombic efficiency during battery charging. 0 2 4 6 0 20 40 60 pH Current Density, mA/ cm 2 Bulk pH Near-surface pH(◆, ) Simulated near-surface pH ( *, +) 54 Figure 2-6: Deviation of coulombic efficiency with stirring rate at constant current densities (¯−50 and ¢−100 mA/ cm 2 ) using an electrolyte composed with iron (II) chloride (3.0 M) and ammonium chloride (2.0 M) at pH = 2.00. While the interaction of charging current density with flow rate and near-surface pH can be exploited to achieve the highest coulombic efficiencies, precipitation of insoluble hydroxides can occur if the optimal current density is exceeded or if the flow rate falls below the set value. The maximum value of pH that we can achieve in the solution of 3.25 M iron chloride without causing formation of the hydroxo-complexes of iron is pH = 3.15. Any attempt to raise the pH of the bulk solution where the solution concentration of iron is 3.25 M does not yield a clear solution. For this reason, although bulk values of pH = 3 and 3.15 are not too different, the increase in pH occurring near the electrode surface can be markedly different. In the following, we describe results with ascorbic acid containing electrolyte that can achieve relatively stable efficiencies over a wide range of current densities. 2.3.5 Effect of Electrolyte Additives 2.3.5.1 Increasing the Solubility of the Electrolyte at Higher pH We have studied the electrodeposition of iron from an electrolyte consisting of a mixture of ammonium chloride and ascorbic acid. Similar compositions were shown in previous studies to 96.0 98.0 100.0 0 250 500 750 1000 1250 1500 % Coulombic Efficiency Stirring rate, rpm 100 mA/cm 2 50 mA/cm 2 55 increase the stability of iron(II) in air, reduce corrosion of iron, and decrease the rate of hydrogen evolution during the electrodeposition and dissolution of iron. 10, 17 We have observed that iron (II) chloride solutions exhibit a slightly higher pH in the presence of 2.0 M ammonium chloride (Table 2-3). This observation is consistent with the effect of increased ionic strength in reducing the inter-ionic interactions and the tendency to form hydrolyzed products. Table 2-3: Effect of ammonium chloride additives for the solubility and dissociation Electrolyte composition pH Apparent Hydrolysis constant * of iron(II) x 10 6 , Ascorbic acid ionization constant * x10 -4 , Iron (II) chloride, M Ammonium chloride (2.0 M) Ascorbic acid (0.3 M) 0.50 - - 3.07 1.45 - 1.00 - - 2.89 1.66 - 1.00 √ - 2.96 1.20 2.00 - - 2.21 19.1 - 3.25 - - 1.50 311 - 3.25 √ - 1.97 35.4 - 3.25 √ √ 0.71 - - - √ √ 2.15 - 1.71 Check marks (√ ) indicate the presence of materials and Asterisk marks (*) represent calculated values. 56 Figure 2-7: Apparent hydrolysis constant of iron (II) solutions Based on the measured pH shift with ammonium chloride we can estimate the change in the activity of the iron (II), using Eq. 2-9 (readily obtained from the equilibrium constant expressions). Kh = (10 -(2*pH) )/ ( aFe2+) (2-9) We have estimated that the apparent hydrolysis constant in 3.25 M iron (II) chloride is 8.8 times lower with 2.0 M ammonium chloride (Figure 2-7). The lower apparent hydrolysis constant implies that a higher pH will have to be reached before precipitation occurs in ammonium chloride containing solutions. Thus, the addition of ammonium chloride will extend the pH range of operation of the all-iron flow battery. 2.3.5.2 Increasing the Solubility of Iron Hydroxide at the Electrode Surface We reported in our previous study that ascorbic acid also enhanced the solubility of iron (II) by the formation of metal-ligand complexes allowing the electrolyte pH to be as high as 3.15. 8, 11 Since the pH near the electrode is even more critical than the bulk pH, we have investigated the effect of ascorbic acid on maintaining the near-surface pH. y = 2E-07e 1.5888x R² = 0.97 0.E+00 2.E-05 4.E-05 6.E-05 8.E-05 0 1 2 3 4 Calculated Apperent K h , M FeCl 2 Concentration in 2 M NH 4 Cl, M 57 Figure 2-8: (a) Experimental near-surface pH (◊,) and bulk pH (●,▲) deviation during potentiodynamic polarization of metal electrodeposition using an electrolyte composed with iron (II) chloride (3.25 M), ammonium chloride (2.0 M) and ascorbic acid (0.3 M). (b) Representation of the equilibrium reactions near the electrode surface during iron electrodeposition process (Ascorbic acid (H 2L) and hydrogen adsorption on iron and HER on iron are not illustrated). In an electrolyte of bulk pH = 0, in the presence of ascorbic acid, the near-surface pH of the iron electrode during electrodeposition rose rapidly with current density (Figure 2-8a). However, when the bulk pH was 3, the near-surface pH remained close to that of the bulk pH value. This difference is attributed to the buffering action of ascorbic acid. With a pKa of 4.17, ascorbic acid 0.00 2.00 4.00 0 20 40 60 80 pH Current density, mA/ cm 2 (a) Bulk pH Near-surface pH 58 will dissociate in the diffusion layer maintaining a relatively stable pH close to the pKa value. 14 Further, the effective pKa of the ascorbic acid in ammonium chloride medium is 2.5 times higher compared to that of the standard value (see Table 2-3). 11 Thus, the buffering action of ascorbic acid in 2.0 M ammonium chloride shifts to even lower pH values than expected from the standard pKa value of 4.17 (see Appendix A4 ). This buffering action at even a lower pH value is essential for operating at higher current densities as the rapid rise of near-surface pH can cause the precipitation of the iron (II) hydroxo complexes. The near-surface pH in the presence of ascorbic acid at various current densities of operation can be determined by applying the Henderson-Hasselbalch equation. Experimental measurements confirm that in the presence of ascorbic acid, the near-surface pH does not deviate significantly from the bulk value (Figure 2- 8a). As mentioned earlier, in addition to the buffering action of ascorbic acid, the chelation of iron (II) by ascorbate also avoids the precipitation of the insoluble hydroxides (Figure 2-8b). 2.3.5.3 Inhibition of kinetics of HER using a Second Metal The composition of the electrode during electrodeposition of iron is expected to affect the rate of hydrogen evolution. In particular, metals such as platinum and nickel facilitate hydrogen evolution because of the enhanced chemisorption of hydrogen resulting from the partially- occupied d-bands of the metal. 7 On the contrary, elements with filled d-bands like cadmium, zinc, lead, bismuth and mercury do not support hydrogen evolution readily. Specifically the overpotential for hydrogen evolution at a current density of 10 mA/cm 2 in 1 N hydrochloric acid is 1.2 V on cadmium. 18 Drazic et al. explained the inhibitory effect of cadmium (II) in the corrosion of iron in sulfuric medium by the increase of the Tafel slope from 130 to 230 mV per decade. 19 Thus, if a second metal such as cadmium with high overpotential for hydrogen evolution can be presented to the surface during the electrodeposition of iron, then hydrogen evolution would be suppressed during charging. For such a metal to be always present at the surface, it must be co-deposited with iron and also be immiscible with iron; the immiscibility will ensure that the second metal remains on the surface at all times and does not diffuse into the bulk of the iron 59 electrode. By staying on the surface, the second metal will continue to provide the suppression of hydrogen evolution during the entire course of electrodeposition. Figure 2-9: (a) Potentiodynamic studies of HER on iron(x) and cadmium(-). (b) Effect of cadmium chloride additives on the coulombic efficiency of iron electrodeposition and electro dissolution. Bath conditions (iron (II) chloride (3.0 M), ammonium chloride (2.0 M), ascorbic acid (0.3 M) and cadmium chloride (0.0 mM(■), 20 mM(▲) at pH = 3.0 and temperature = 25 ◦ C). Cadmium is one such metal and thus a monolayer of cadmium on the surface of iron can be expected to hinder the kinetics of the hydrogen evolution reaction. Our previous results with small amounts of indium chloride in the bath suggested that the benefit of reduced hydrogen evolution can be realized with the co-deposition of metals with a high overpotential for -3.0 -2.5 -2.0 -1.5 -1.0 0.00 0.40 0.80 1.20 log (Current Density), current density in A/cm 2 Overpotential, V Iron, Tafel slope =280mV/ decade Cadmium, Tafel slope =353mV/ decade (a) 65 70 75 80 85 90 95 100 0 10 20 30 40 % Coulombic Efficiency Current Density, mA/ cm 2 with 0 mM cadmium additive with 20 mM cadmium additive (b) 60 hydrogen evolution, indium being one such. 8 However, a quantitative understanding of this benefit of co-deposition was not presented. We have therefore explored this effect further with cadmium addition, a metal similar to indium with respect to the high overpotential for hydrogen evolution. We have observed the doubling of the overpotential for HER at 10 mA/cm 2 with cadmium metal (Figure 2-9a). Further, the Tafel slope for hydrogen evolution on cadmium was higher than that on iron by 73 mV/ decade. These findings implied that cadmium could be exploited to inhibit hydrogen evolution from the negative electrolyte and increase the coulombic efficiency. We expected this effect to be more significant at higher charging current densities. When we electrodeposited iron from solutions containing 20 mM cadmium chloride, we found that the coulombic efficiency was consistently higher than for the electrolyte without cadmium when operating higher than 20 mA/cm 2 (Figure 2-9b). The improvement in coulombic efficiency from 90 to 93% resulting from the addition of cadmium, despite seeming small, translates to a 30% reduction in the amount of hydrogen evolved during charging. Thus, even a 3% improvement in coulombic efficiency is significant for increasing the cycle life of the all-iron flow battery. While this comparison proves the viability and basis of this approach of using a second metal to raise coulombic efficiency, the use of cadmium chloride is not desirable for developing an environmentally-friendly battery. A similar strategy relying on bismuth, tin and other non-toxic metals needs to be investigated. 20 Inhibition by organic molecules that can adsorb on the iron electrode also needs to be explored. 21, 22 2.3.6 Effect of Temperature We found that the coulombic efficiency of iron deposition and dissolution increased by 0.17% per degree Celsius rise of bath temperature in the range of 25–60 ◦ C at a charging current of 20 mA/ cm 2 (Figure 2-10). Hruska et al. demonstrated a coulombic efficiency of 90% using ferrous ammonium chloride electrolyte bath operated at 60 ◦ C however, their focus was not to study the effect of temperature on the coulombic efficiency. 9 We observed a coulombic efficiency of 61 97.9% at 60 ◦ C compared to 91.8% at 25 ◦ C. The increase of temperature would increase the solubility of the iron chloride and iron hydroxides, increase the ionic conductivity of the solution, increase the diffusion coefficient and increase the rate constant for charge transfer. 23 However, these various effects are not easily separated. Figure 2-10: Effect of temperature on coulombic efficiency of iron electrode- position and electro dissolution. Bath conditions (iron (II) chloride (3.0 M), ammonium chloride (2.0 M) and ascorbic acid (0.3 M) at pH = 3.0). Efficiency values indicated for each point. The improved kinetics of iron deposition is supported by the decrease of electrochemical impedance with temperature over the entire range of frequencies (Figure 2-11). We ascribed the real component of the high frequency impedance to the ohmic component arising from the electrolyte and electrode and designated it as the series resistance Rs. Using a Randles’ type model, the real component of the impedance at low frequencies (5–10 Hz range) was ascribed to a polarization resistance associated with the charge-transfer process and denoted as Rp (Table 2-4). We find that Rp also decreased with increasing temperature. Absence of no significant difference in the impedance data with and without stirring suggested that the mass transport of the electroactive species is not a limiting process. This finding is consistent with reports in the scientific literature that electrodeposition of iron proceeds via an adsorbed intermediate involved in the rate determining step. 24-31 Thus, we could attribute the improved 91.8 92.8 94.3 97.9 90 95 100 0 10 20 30 40 50 60 70 % Coulombic Efficiency Temperature, o C 62 kinetics to the desorption and surface diffusion of adsorbed intermediates resulting from an increase of temperature. In addition, we find that upon increasing the temperature from 25 to 60 ◦ C the Tafel slope for electrodeposition of iron decreases by about 30 mV/ decade (Appendix A5). Figure 2-11: Overlaid Nyquist diagrams at−0.85V vs. Ag/AgCl at 25-60 ◦ C at stirring (x) and no stirring (◊) conditions. Bath conditions (iron (II) chloride (3.0M), ammonium chloride (2.0 M) and ascorbic acid (0.3 M) at pH = 3.0). Temperatures indicated on the charts. (Rs -ohmic resistance and Rp -polarization resistance). Just as the iron deposition process proceeds through adsorbed reaction intermediates, the hydrogen evolution process also involves formation of adsorbed hydrogen. 24-31 Thus, the deposition of iron and the evolution of hydrogen can be expected to compete for the surface sites. Further, since the electrolyte contains ascorbic acid and ammonium chloride, equilibrium constants of iron hydrolysis and the extent of adsorption of hydrogen and ascorbic acid on the iron electrode surface would also be changed with increasing the temperature. The coulombic efficiency values suggest that the surface energetics of the iron deposition reaction is favored at higher temperatures over that of the hydrogen evolution reaction. 0 1 2 0 2 4 -Z imaginary , Ω cm 2 Z real , Ω cm 2 25 o C 0 1 2 0 2 4 -Z imaginary , Ω cm 2 Z real , Ω cm 2 30 o C R s R p 0 1 2 0 2 4 -Z imaginary , Ω cm 2 Z real , Ω cm 2 60 o C 0 1 2 0 2 4 -Z imaginary , Ω cm 2 Z real , Ω cm 2 40 o C 63 Table 2-4: Effect of temperature on electrochemical impedance parameters of iron electrode under potentiostatic conditions at -0.85 V vs. Ag/AgCl. Temperature, o C With no stirring With stirring R S, Ω cm 2 R p, Ω cm 2 R S, Ω cm 2 R p, Ω cm 2 25 1.223 2.074 1.218 1.998 30 1.133 1.496 1.117 1.487 40 1.009 1.009 0.991 1.023 60 0.782 0.466 0.773 0.407 Another benefit of operation at a higher temperature is that the internal stress of the electrodeposit can be relieved. A stress-free deposit will avoid the formation of powdery particles and metal detachment. 31 Electroplaters of iron have known these beneficial effects of temperature for some time and the electroplating of iron is always practiced at temperatures greater than 60 ◦ C, and even at as high as 80 ◦ C. 31 Although the overall beneficial effect of temperature on the coulombic efficiency has been measured for the electrolyte containing ascorbic acid and ammonium chloride (Figure 2-10), additional resolution of the contribution of the other factors affected by increase of temperature requires further investigation. 2.4 Chapter Two Conclusions The all-iron flow battery using iron chloride electrolyte is an attractive solution for large-scale energy storage if the hydrogen evolution that occurs during charging can be suppressed. Our study provides insight into several promising approaches to improving the coulombic efficiency of the iron electrode. We demonstrate a coulombic efficiency of 97.9% using a combination of Previous summary: Improvements to the Charging Efficiency 4 70 80 90 100 85.9 91.8 92.6 94.2 97.9 %Charging Efficiency Electrolyte 3.25 M FeCl 2 2 M NH 4 Cl 0.3 M Ascorbic Acid pH 0; Room Temp; Charging at 20 mA cm -2 pH 3; Room Temp; Charging at 20 mA cm -2 pH 3; Room Temp; Charging at 20 mA cm -2 ; 20 mM CdCl 2 additive; pH 3; Room Temp; Flow cell Experiment; Charging at 80 mA cm -2 pH 3; Temp 60ºC; Charging at 20 mA cm -2 64 functional electrolyte additives, pH and elevated temperature. To our knowledge this value of coulombic efficiency is among the highest reported for the “round-trip” coulombic efficiency of charge and discharge of the iron electrode. The coulombic efficiency during electrodeposition of iron was found to improve with increasing pH at all values of current density. We have found that ascorbic acid has an important role in determining the coulombic efficiency. The adsorption of ascorbic acid on iron inhibits hydrogen evolution over the pH range of 0 to 3. We have found that beyond the pH of the bulk electrolyte, the pH at the surface of the electrode is even more crucial in achieving a high coulombic efficiency; maintaining a high near-surface pH without precipitation is important for increasing the coulombic efficiency. Thus, the coulombic efficiency was also found to be sensitive to the rate of stirring of electrolyte. A key finding is that the near-surface pH can be regulated close to an optimal value of 3 by exploiting the acid-base buffering properties of ascorbic acid. Such an arrangement using ascorbic acid allows the iron electrode to operate at high current densities without the danger of precipitation of the insoluble hydroxides. Thus, ascorbic acid is again shown to be critical in achieving high coulombic efficiency. We have also demonstrated that a metal such as cadmium that has a high overpotential for hydrogen evolution and that is also immiscible with iron can provide a significant improvement in coulombic efficiency during iron electrodeposition. By the co-deposition of cadmium with iron, we found that the rate of hydrogen evolution can be reduced. We also conclude that increasing the operating temperature to 60 ◦ C has a strong beneficial impact on the coulombic efficiency. We have found that the charge-transfer kinetics of iron deposition was improved relative to that of the hydrogen evolution reaction by increasing the temperature, leading to a higher coulombic efficiency. With the fundamental insights gained in this study and the improvements in coulombic efficiency demonstrated we believe that the all-iron redox flow battery based on iron chloride will continue 65 to present an attractive pathway for large-scale electrical energy storage and the authors believe that the future will benefit from research on approaches to achieving 100% coulombic efficiency. 66 2.5 Chapter Two References 1. Jayathilake, B. S.; Plichta, E. J.; Hendrickson, M. A.; Narayanan, S. R., Improvements to the Coulombic Efficiency of the Iron Electrode for an All-Iron Redox-Flow Battery. Journal of The Electrochemical Society 2018, 165 (9), A1630-A1638. 2. Jayathilake, B. S.; Manohar, A.; Plichta, E. J.; Hendrickson, M. A.; Narayanan, S.R., Improvements to the Charging Efficiency of an All-Iron Redox-Flow Cell, Meeting Abstracts, The Electrochemical Society: 2017; pp 177-177. 3. Pan, F.; Wang, Q., Redox Species of Redox Flow Batteries: A Review. Molecules 2015, 20 (11), 19711. 4. Alotto, P.; Guarnieri, M.; Moro, F., Redox flow batteries for the storage of renewable energy: A review. Renewable and Sustainable Energy Reviews 2014, 29, 325-335. 5. Hibert, F.; Miyoshi, Y.; Eichkorn, G.; Lorenz, W., Correlations between the Kinetics of Electrolytic Dissolution and Deposition of Iron II. The Cathodic Deposition of Iron. J. Electrochem. Soc. 1971, 118 (12), 1927-1935. 6. Selverston, S.; Savinell, R. F.; Wainright, J. S., In-tank hydrogen-ferric ion recombination. Journal of Power Sources 2016, 324, 674-678. 7. Gabe, D. R., The role of hydrogen in metal electrodeposition processes. J. Appl. Electrochem. 1997, 27 (8), 908-915. 8. Manohar, A. K.; Kim, K. M.; Plichta, E.; Hendrickson, M.; Rawlings, S.; Narayanan, S. R., A High Efficiency Iron-Chloride Redox Flow Battery for Large-Scale Energy Storage. J. Electrochem. Soc. 2016, 163 (1), A5118-A5125. 9. Hruska, L.; Savinell, R., Investigation of Factors Affecting Performance of the Iron-Redox Battery. J. Electrochem. Soc. 1981, 128 (1), 18-25. 10. Hawthorne, K. L.; Petek, T. J.; Miller, M. A.; Wainright, J. S.; Savinell, R. F., An Investigation into Factors Affecting the Iron Plating Reaction for an All-Iron Flow Battery. J. Electrochem. Soc. 2015, 162 (1), A108-A113. 11. Martell, A. E., Chelates of Ascorbic Acid. In Ascorbic Acid: Chemistry, Metabolism, and Uses, Am. Chem. Soc.: 1982; Vol. 200, pp 153-178. 67 12. Ferreira, E. S.; Giacomelli, C.; Giacomelli, F. C.; Spinelli, A., Evaluation of the inhibitor effect of l-ascorbic acid on the corrosion of mild steel. Mater. Chem. Phy. 2004, 83 (1), 129-134. 13. Freedman, L.; Sack, A., Color Reactions and Stability of Iron Ascorbate**From the Research Laboratories of U. S. Vitamin Corporation, New York N. Y. J. Am. Pharm. Association 1944, 33 (9), 316-318. 14. Plug, C. M.; Dekker, D.; Bult, A., Complex stability of ferrous ascorbate in aqueous solution and its significance for iron absorption. Pharmaceutisch Weekblad 1984, 6 (6), 245-248. 15. Sekine, I.; Nakahata, Y.; Tanabe, H., The corrosion inhibition of mild steel by ascorbic and folic acids. Corrosion Science 1988, 28 (10), 987-1001. 16. Dahms, H., The influence of hydrolysis on the deposition and CO-DE-position of iron- group metals (Fe, Co, Ni) at the dropping mercury electrode. J. Electroanal. Chem. 1964, 8 (1), 5-12. 17. Darwish, N. A.; Hilbert, F.; Lorenz, W. J.; Rosswag, H., The influence of chloride ions on the kinetics of iron dissolution. Electrochim. Acta 1973, 18 (6), 421-425. 18. Gaida, B., Electroplating Science: Fundamental chemistry, Electrochemistry, Physics and Electricity with particular reference to the needs of electroplating students. Draper: Teddington, 1970. 19. Dražić, D.; Vorkapić, L., Inhibitory effects of manganeous, cadmium and zinc ions on hydrogen evolution reaction and corrosion of iron in sulphuric acid solutions. Corrosion Science 1978, 18 (10), 907-910. 20. Manohar, A. K.; Yang, C.; Malkhandi, S.; Prakash, G. S.; Narayanan, S., Enhancing the performance of the rechargeable iron electrode in alkaline batteries with bismuth oxide and iron sulfide additives. J. Electrochem. Soc. 2013, 160 (11), A2078-A2084. 21. Malkhandi, S.; Yang, B.; Manohar, A. K.; Prakash, G. S.; Narayanan, S., Self-assembled monolayers of n-alkanethiols suppress hydrogen evolution and increase the efficiency of rechargeable iron battery electrodes. J. Am. Chem. Soc. 2012, 135 (1), 347-353. 22. Yang, B.; Malkhandi, S.; Manohar, A. K.; Surya Prakash, G. K.; Narayanan, S. R., Organo- sulfur molecules enable iron-based battery electrodes to meet the challenges of large-scale electrical energy storage. Energy Environ. Sci. 2014, 7 (8), 2753-2763. 68 23. Gamburg, Y. D.; Zangari, G., Theory and practice of metal electrodeposition. Springer Science & Business Media: 2011. 24. Hilbert, F.; Miyoshi, Y.; Eichkorn, G.; Lorenz, W. J., Correlations between the Kinetics of Electrolytic Dissolution and Deposition of Iron: I . The Anodic Dissolution of Iron. J. Electrochem. Soc. 1971, 118 (12), 1919-1926. 25. Baker, B. C.; West, A. C., Electrochemical Impedance Spectroscopy Study of Nickel-Iron Deposition: II. Theoretical Interpretation. J. Electrochem. Soc. 1997, 144 (1), 169-175. 26. Matlosz, M., Competitive Adsorption Effects in the Electrodeposition of Iron-Nickel Alloys. J. Electrochem. Soc. 1993, 140 (8), 2272-2279. 27. Keddam, M.; Mattos, O. R.; Takenouti, H., Reaction model for iron dissolution studied by electrode impedance II. Determination of the reaction model. Journal of The Electrochemical Society 1981, 128 (2), 266-274. 28. Keddam, M.; Mottos, O. R.; Takenouti, H., Reaction model for iron dissolution studied by electrode impedance I. Experimental results and reaction model. J. Electrochem. Soc. 1981, 128 (2), 257-266. 29. Bockris, J. M.; Drazic, D.; Despic, A., The electrode kinetics of the deposition and dissolution of iron. Electrochim. Acta 1961, 4 (2-4), 325-361. 30. Bockris, J. M.; Kita, H., Analysis of galvanostatic transients and application to the iron electrode reaction. Journal of the Electrochemical Society 1961, 108 (7), 676-685. 31. Schlesinger, M.; Paunovic, M., Modern Electroplating. John Wiley & Sons: New Jersey, 2011. 69 Charge Transfer Kinetics of Iron Deposition and Hydrogen Evolution During Iron Plating― Insights Through the Electrolyte Properties and Iron(II) Complex Formation in Ammonium Chloride Solutions 3.1 Background This Chapter is an extended discussion from the conference proceeding publication 1 and the presentation delivered at the 48 th Power Sources Conference at Denver, Colorado, June 2018. 3.1.1 Hydrogen Evolution Kinetics Studies of charge transfer kinetics of the hydrogen evolution reaction (HER) can enhance our understanding of the processes occurring during the electroplating of iron that is relevant to the all-iron flow battery and electrowinning of iron. Facilitating the deposition of iron in preference to HER is important to achieve higher coulombic efficiencies during the operation of the all-iron flow battery and electroplating. Electrochemical production of hydrogen gas from aqueous electrolytes can occur in two ways― by hydronium ion reduction (Eq 3-1) or by water reduction (Eq 3-2). In both acidic and alkaline media, three well-known steps of hydrogen evolution reactions namely; Volmer (3-3 a&b), Heyrovsky (3-4 a&b) and Tafel (3-5) are identified. 2H 3O + + 2e − à H 2 + 2H 2O (3-1) 2H 2O + 2e − à H 2 + 2OH − (3-2) Volmer : H 3O + + e − + Mà M-H + H 2O (3-3a) Volmer : H 2O + e − + Mà M-H + OH − (3-3b) Heyrovsky : M-H + H 3O + + e − à H 2 + H 2O+M (3-4a) Heyrovsky : M-H + H 2O + e − à H 2 + OH − +M (3-4b) Tafel : 2M-H à H 2 + 2M (3-5) 70 Overpotential is an important parameter for determining the rate of the hydrogen evolution reaction. Bockris reported that the HER overpotential at a given current density is a periodic function of the atomic number. 2 Heavy metals like lead, mercury, and cadmium give higher overpotentials for hydrogen evolution reaction. 3 On the other hand, metals like platinum give lower overpotentials for HER as M-H bonding is favorable on them (Table 3-1). 2 As described in Chapter One, the exchange current density and Tafel slope are the principal parameters that are used to characterize the charge transfer kinetics and mechanism. Transition metals generally have high exchange current densities for hydrogen evolution. 4 Volcano plot (Figure 3-1) shows the relationship of exchange current density of HER and the M-H bond strength. The optimum M-H bond strength for HER lies near platinum and it represents that the too low or too strong M-H bond strengths would not favor HER. 5 As we reported in Chapter 2, the presence of cadmium in the electrode surface during iron deposition, HER exchange current density can be lowered due to poor M-H bond strength. Further, the Tafel slope of HER differs based on the rate-determining step and from metal to metal, reaction mechanism can deviate. 6 Figure 3-1: Volcano plot of Hydrogen Evolution on metals 5 71 Table 3-1: HER overpotential and Tafel slopes of HER on different metals 2, 3, 7 Cathodic metal HER overpotential 2 in 1 N HCl at 25 o C / V at 10 mA cm -2 Tafel slope of HER, V/ decade Platinized Platinum 0.03 0.02 3 Solid Platinum 0.39 0.029 2 Solid Nickel 0.42 0.1 2 Solid Iron 0.53 0.12 2 Solid Silver 0.66 0.07 2 Solid Zinc 0.75 0.12 8 Solid Copper 0.75 0.1 2 Solid Cadmium 1.2 0.25 3 3.1.2 Iron Deposition Kinetics Catalytic and non-catalytic pathways of iron deposition are proposed in literature. 9-13 Using galvanostatic transients and steady-state studies of iron deposition and dissolution, Bockris, Drazic and Despic proposed a non-catalytic (consecutive) mechanism. 9, 13 In this mechanism, a two-step electron reduction is proposed via a FeOH intermediate (Eq 3-6 to 3-8). The first electron reduction step (Eq. 3.7) is the rate-determining step. 9, 10 In a second mechanism (catalytic mechanism) (Eqs.3.6, 3.9 to 3.10) of iron deposition proposed by Heusler, the first two steps in (Eqs. 3-6 and 3-7) are similar to the non-catalytic mechanism, but the rate- determining step involves adsorbed FeOH (Eq. 3.9) acting as a catalytic-surface intermediate to facilitate the second electron reduction process. 9 . Further, in both mechanisms, as FeOH can be adsorbed to the electrode surface, iron deposition and HER are competitive. Fe 2+ + OH - D FeOH + (3-6) FeOH + + e- à FeOHads (3-7) 72 FeOHads + e- D Fe + OH - (3-8a) or FeOHads + H + + e- D Fe + H 2O (3-8b) FeOHads + FeOH + + 2e- àFeOHads + Fe + OH - (3.9a) FeOHads + FeOH + + H + + 2e- àFeOHads + Fe + H 2O (3-9b) Bockris reported that hydrogen evolution exchange current density on iron shows no effect in the presence of iron (II) in the electrolyte. However, he proposed a relationship of ∂log ioHER/ ∂pH = -0.5. 12 Although, iron chloride containing, redox flow cell operation can be achieved at higher concentrations of iron (II) as high as 3.25 M, hardly any data is reported on the kinetics of iron deposition and HER at higher iron (II) concentrations. Therefore, in this study, a wide range of concentrations of iron (II) from 0.6 to 3.0 M was used. This range of concentrations represents the state of charge of the anolyte of the all-iron redox flow cell from 20-100%. Changing the concentration of iron (II), electrolyte, we studied the changes in the kinetics of HER and iron deposition. 3.2 Experimental Conditions The electrolyte used in the experiments was iron (II) chloride (0.6 to 3.0 M) in ammonium chloride (2 M). Conductivity and electrolyte pH were measured using a Thermo Scientific™ Orion Star™ A212 Conductivity Benchtop Meter, and two pH-sensing glass electrodes from Thermo Scientific™ (Orion-ROSS™ Ultra pH/ATC Triode), respectively. EIS data were obtained from 10 KHz to 1 mHz frequency at -0.7 V vs. NHE and experimental data were fitted into a proposed equivalent circuit using ZSimpWin software. EIS at the higher frequency (10 KHz) at open-circuit voltage was used to determine the ohmic potential drop as described in Chapter 2 section 2.2.3. Coulombic efficiency at different current densities was measured on a graphite cylindrical electrode as described previously (Chapter 2 section 2.2.1). The experimental conditions for the coulombic efficiency measurements are listed in Table 3-2. 73 Table 3-2: Experiment Conditions Experimental property Condition Deposition charge 60 Coulombs Charging rate 10-50 mA/ cm 2 Discharging rate 20 mA/ cm 2 Stirring rate 750 rpm Temperature 298 K Electrolyte volume 200 mL Electrode area 23 cm 2 3.3 Results and Discussion 3.3.1 Electrolyte Properties As chloride ions give higher electrical conductivity compared to sulfate electrolytes, chloride plating baths are used in metal deposition processes. 14, 15 Both diffusion and electrical conductance in electrolyte solutions involve the motion of ions. In conduction, positive and negative ions move in opposite directions. 16 Conductivity data in literature related to electrolytes of iron flow batteries show that iron (II) chloride solutions have three times higher conductivity to that of sulfate solutions. 17, 18 In our experiments, we used ammonium chloride as a supporting electrolyte which could increase the conductivity of the iron (II) solutions. Addition of 2 M ammonium chloride increased the conductivity of the electrolyte by 1.5 times compared to the bare iron (II) chloride solutions. However, both sulfate and chloride electrolytes show an initial increase of the conductivity from very low concentrations to higher to achieve a maximum around 1-1.5 M range (Figure 3-2A). The initial increase in the conductivity is due to ionic strength improvement (Figure 3-2C). With further increase in the concentration of iron (II), a drop of conductivity is observed and a similar trend is reported by Savinell et al. 18 . Even though, the ionic strength of solutions increases upon the increase of iron (II) concentration (Figure 3-2C), due to ion-association in the electrolyte, the 74 conductivity drops. Electrolyte pH decreases with increasing the concentration of iron (II) (Figure 3-2B). Figure 3-2:Electrolyte properties at different iron (II) concentrations A: Conductivity, B: pH and C: ionic strength (l -0 M and q- 2 M Ammonium Chloride electrolyte) When iron (II) chloride is dissolved in water, iron (II) ions stay as a hexa-aqua complex with chloride ions staying separated from the metal ions as counter ions (Scheme 1, Equilibrium I). According to Bronsted’s postulations, multi-valent metal ions can undergo hydrolysis to replace water molecules by hydroxyl ions resulting in a decrease in pH (Scheme 1, Equilibrium II). This process can lower the activity and activity coefficient of the metal ion. There are mathematical models such as the Debye-Huckel model, hydration model of Stokes and Robinson, and the Bromley equation to calculate activity coefficients for dilute and concentrated electrolytes. 19 Measurement of the reversible electrode potential is also a widely-practiced method for 1 2 3 4 8 12 FeCl 2 Concentration, M Ionic Strength, M no NH 4 Cl with NH 4 Cl B A C 0.5 1 1.5 50 100 150 200 250 (Concentration) 1/2 , M Conductivity, mS/cm -0.3 0 0.3 0.6 1.5 2 2.5 3 3.5 Log(Concentration), Concentration in M pH 75 determining ionic activity. 20 Randall has reported that activity coefficient values of ferrous solutions from 0.001 to 1M concentrations of iron(II) chloride using the relationship of log [activity coefficient ] and the square root of the ionic strength. 21 In the present study, we have used pH measurements to quantify the activity of iron (II) as a function of concentration (Figure 3-3A) and we observed that the activity and activity coefficient increases with iron (II) concentration as the ionic strength increases (Figure 3-3B) (See Appendix B1 for the detailed Calculation). Scheme 1: Different equilibria in iron chloride solutions Figure 3-3: (A) Activity of different species in the electrolyte and (B) Activity coefficient of Fe 2+ in the electrolyte containing iron(II) chloride (0.6 to 3 M) and ammonium chloride (0 or 2 M);(n- FeOH + = H + , l-Fe 2+ ,ê-Cl- and t- Fe-Cl complex) Also, interactive opposite charges can be in ion-pairing equilibrium in iron (II) chloride solutions as shown in scheme 1 (Equilibrium I and II). Ion pairs are formed in electrolytes when opposite charges can associate together (partially) within a critical distance from each other. The net 4 8 12 Ionic Strength, M 0.3 0.6 0.9 Activity Coefficient, Fe 2+ no NH 4 Cl with NH 4 Cl B A 0.6 1 2 2.5 3 2 4 6 FeCl 2 Conecntration, M Activity 76 charge of the ion pair is poor and leads to change the electrostatic interactions of the fully dissociate forms. 22, 23 With the increase of the ionic strength of the solution when we add more ammonium chloride, ion-pairing should be less. However, this de-shielding process competes with the complexation reaction that adds the chloride into the coordination shell. This process will reduce the amount of chloride available for conduction and hence reduce the conductivity (Figure 3-2A). Upon addition of more chloride ions (2 M ammonium chloride), the decrease of activity of iron (II) arise due to the iron(II)-chloro complexation (Figure 3-2, Scheme 1, Equilibrium II). 24 The decrease of the hydrogen ion concentration in ammonium chloride medium (Figure 3-2B) suggests that the activity of iron(II) and the iron-hydroxo complex is lower than that in ammonium chloride free medium. Drop of the activity coefficient of iron (II) when ammonium chloride is added verifies that upon addition of extra chloride ions, more iron-chloro complexes can be formed (Figure 3- 3B). As the activity of iron (II) could be calculated using pH measurement by using the formation constant from the literature, activity of iron (II)-chloro complexes can also be predicted (See appendix B1 for the calculation). 24 These calculations show that, with increasing concentration of iron (II) in the electrolyte, the extent of formation of the iron (II)-chloro complex can also be higher (Figure 3-3A). Therefore, in an ammonium chloride supportive electrolyte, iron deposition reaction can be expected to happen from different species of iron (II) such as Fe(H 2O) n 2+ , Fe(OH)(H 2O) n + and Fe(H 2O) n-1Cl + . 25 However, the quantity of each species present in the electrolyte is a function of the bulk concentration of iron (II) chloride (C Fe2+) and the ionic strength. The equilibrium potential (E eq) of iron deposition reaction given by the Nernst equation (Eq 3-10) can be affected in the presence of complexes. The equilibrium potential of iron deposition reaction in a complexed solution (E com) can be calculated using the formation constant of the complex (K m) as shown in the Eq. 3-11. > +? => $ − 34 F2 @1- ; ' 2+FG 5 (3-10) > H$I => +? − JK.;M F @$NO I (3-11) 77 Where R = gas constant, F = Faraday constant, T = temperature and at room temperature, RT/F=59.16 This study shows that from a simple pH measurement, we can understand how the deviation of iron (II) concentration can influence the kinetics of the iron deposition and hydrogen evolution reactions. 3.3.2 Dependence of Coulombic Efficiency of Iron Deposition on the Concentration of Iron (II) Figure 3-4: Relationship of coulombic efficiency with A: Current density and electrode potential and B: State of charge of the anolyte from 0 to 80% (in 3-0.6 M FeCl 2 and 2 M NH 4Cl). 93 -0.55 95 -0.6 50 97 40 -0.65 99 30 -0.7 20 -0.75 10 Current Density, mA/cm 2 Electrode Potential, V vs NHE % Coulombic Efficiency 93 94 95 96 97 98 0.6 M 1 M 2 M 2.5 M 3 M A 96 98 100 0 20 40 60 80 %Coulombic Efficiency % State of Charge 40 mA/cm 2 20 mA/cm 2 B 78 Coulombic efficiency increases with increasing the concentrations of ion (II). Electrode potential of iron deposition increases with current density and decreases with increasing concentration (Figure 3-4). The dependence of coulombic efficiency on current density is steeper at lower current densities compared to higher current densities. At a given iron (II) concentration higher rates of iron deposition could be achieved by increasing the overpotential. However, we observed that at lower electrode potentials higher coulombic efficiency could be realized with higher concentrations of iron (II). Both electrode potential and overpotential of iron deposition were lower in 3 M iron (II) chloride solutions. In the range of current density from 10-50 mA/cm 2 , coulombic efficiencies higher than 98% can be observed even though the higher concentration of iron (II) results in lowering of the electrolyte pH. With increasing the iron (II) concentration, we can expect the concentration overpotentials to be lower for iron deposition for any particular rate. Similar behavior of an increase of coulombic efficiency and lowering of the overpotential in cobalt deposition is reported by Kongtein et al. 14, 15 They have described that water activity decreases with increasing the salt concentration. 14 Thus, it may lower the electrode potential of iron deposition leading to achieve higher voltage efficiency of the battery. Improved coulombic efficiency at higher bulk and surface pH (discussed in Chapter two section 2.4.2) can also be due to lower equilibrium concentration of protons in the electrolyte near the electrode surface and generation of more FeOH + because of the concentration gradient in the diffusion layer. More FeOH + can enhance the rate of the first electron reduction process. Also, the equilibrium potentials for iron deposition and HER are a function of the activity of the iron (II) and the degree of the hydrolysis. The second electron reduction process during iron deposition can generate OH - near the electrode surface and the increase of the pH can drive the formation of FeOH + to drive the iron deposition reaction to happen. However, the observation of more negative potentials at higher states of charge or lower concentration of iron (II) suggests that a higher value of current for hydrogen evolution will be sustained compared to higher states of charge where the bulk concentration of iron (II) is higher. The decrease of the coulombic efficiency at 20 mA/cm 2 with increasing the SOC is consistent with this negative shift of the electrode potential (Figure 3-4B Green line). 79 Even though higher electrode potential is required for charging at a low iron (II) concentration, we can expect that the coulombic efficiency to be stable at higher current densities because of the increase of the surface pH (discussed in Chapter 2 section 2.3.4) due to decreased hydrolysis. We observe a coulombic efficiency of 98 ±1% from 0 to 80% SOC at 40 mA/cm 2 (Figure 3-4B Blue line). The present study reveals that even though coulombic efficiency is higher at higher pH, achieving better kinetics along with improved coulombic efficiency at higher iron (II) concentration and lower pH is more beneficial. 3.3.3 Effect of Adsorptive Corrosion-Inhibitors for the Overall Performance of the Iron Deposition Process Figure 3-5: Polarization curves of hydrogen evolution on iron at 10 mV/s scan rate (in 2M in ammonium chloride solution) Different electrolyte additives can help to minimize HER. There are a number of organic molecules studied as corrosion inhibitors for iron in acidic media. 26-29 Addition of these organic molecules inhibits hydrogen evolution by minimizing the surface coverage of protons. In the present study, we studied cysteine, a well-known corrosion inhibitor. 27, 30 Polarization studies of the hydrogen evolution in ammonium chloride solutions shows that cysteine can minimize HER on iron and reduce the rate of iron dissolution (Figure 3-5). The inhibition effect is considerably -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 1.E-05 1.E-04 1.E-03 1.E-02 1.E-01 1.E+00 Potential, V vs. NHE Current density, A/cm 2 Cysteine 20mM Blank 80 higher near the corrosion potential. When the cathodic current density is increased, we can observe that HER current density is not significantly affected by the presence of cysteine. Even though cysteine works as a better additive for minimizing HER near the corrosion potential, adsorption of non-conducting inhibitors can hinder the charge transfer processes of iron deposition. Observation of poor coulombic efficiency (Figure 3-6) of iron deposition in the presence of 20 mM cysteine verifies that the adsorption of cysteine on the electrode surface can hinder surface processes of electron transfer due to competitive surface adsorption. The shift of the iron deposition electrode potential with cysteine addition in both cathodic and anodic charge-discharge curves (Figure 3-6A) imply that surface adsorption of cysteine creates a kinetic barrier for charge transfer processes. The overpotential increase during the iron deposition could increase the partial current density of HER at a given pH and we could therefore observe a lower coulombic efficiency of iron deposition compared to the iron plating bath electrolyte with no inhibitors (Figure 3-6). More importantly, these observations verify that the effect of hindering of iron deposition kinetics is higher compared to that of HER in the presence of cysteine additives. Thus, investigation of inhibitors with favorable kinetics of iron deposition is important for improving the efficiency of the system. Figure 3-6: Iron deposition and dissolution with cysteine additives 0.0 0.1 0.2 0.3 0.4 80 85 90 95 100 0 20 40 60 Iron deposition Overpotential, V %Coulombic Efficiency Current Denisty, mA/cm 2 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0 50 100 150 IR Corrected Electrode Potential at the Iron Electrode, V vs. NHE Time, s 20 mM Cysteine 81 3.3.4 Kinetics of Iron Deposition and Hydrogen Evolution on Iron in Iron Chloride Solutions To analyze the kinetics of iron deposition and hydrogen evolution during iron deposition, the partial current densities of each of these reactions should be calculated. Since the total cathodic current during charging (I tot) is divided between iron deposition and hydrogen evolution, partial current densities for the foregoing processes can be extracted from the total current using the coulombic efficiency values (Eqs. 3-11 & 3-12). We maintained the same current density during discharging (anodic stripping) so that we could make comparisons. CE=IFED/Itot (3-11) Itot=IHER+IFED (3-12) where I FED and I HER are partial current densities of iron deposition and HER. The logarithm of these partial current density values vs. overpotential may then be plotted separately to generate the Tafel lines for both these competitive reactions (Figure 3-7). In Tafel plots, we can see that Tafel slopes for HER and iron deposition vary considerably depending on the concentration and the pH of the solution. At any iron (II) concentration, as we increase the current density, the overpotentials of iron deposition increase. With increasing the concentration of iron (II), lower overpotential for iron deposition can be observed, attributable to the enhanced exchange current density at higher concentrations. However, with increasing the current density, due to concentration decrease at the electrode surface (as explained in surface pH section in Chapter 2), overpotential of both HER and iron deposition will increase. Further, in electrolytes with low concentrations of iron (II), as the pH is higher, the mechanism of charge transfer for HER can change. Bockris et al. reported that HER from proton will be minimized at higher current densities and HER from water would start to happen. 12 The change of Tafel slopes of iron deposition in concentrations of iron (II) from 0.6 and 1 M at higher current densities can also be due to the concentration decrease which is known as mass transfer limitation. Therefore, the current density values can be corrected for mass transport overpotential losses to obtain the true charge-transfer current density. Accordingly, the Butler- Volmer equation can be modified based on the Nernst Diffusion Layer Model for steady-state 82 mass transport. Thus, by using the limiting current density (I lim,) the Tafel equation for HER and iron deposition can be corrected for mass transport losses (Eq.3-13 & 3-14). From these Tafel plots, the exchange current densities for HER (i 0 HER) and iron deposition (i 0 FED) and Tafel slopes can be determined (Table 3-3). η=-[RT/ (a nF) ln(I o)]- [RT/ (a n F) ln(I/ (a s /a*))] (3-13) h=-[RT/(a nF) ln(I o)]- [RT/(a nF) ln(I/(1- I/Ilim,c))] (3-14) Where a s is the surface activity, a* is the bulk activity, a is the transfer coefficient, η is the observed overpotential, R is the gas constant and T is the absolute temperature. Figure 3-7: Tafel lines of A: Iron deposition and B: Hydrogen evolution reactions on iron from 0.6 to 3 M iron (II) chloride (q-0.6, l-1, ¢-2, t-2.5 and ê-3M) and 2 M ammonium chloride. An iron rod dipped in iron (II) containing electrolyte can have a mixed electrode potential due to conjugate reactions of HER and metal deposition simultaneously possible at the surface of the electrode. However, due to the decrease of the bulk concentration of iron (II) during charging, the exchange current densities and Tafel slopes of iron deposition and HER can vary. Iron deposition and dissolution kinetic studies reported in literature explain that impurities and pH of the electrolytes are key factors for determining kinetics of iron deposition. 11 Bockris et al. studied the electrode kinetics of iron deposition and dissolution kinetics in different concentrations of iron (II) (up to 0.5 M), pH values and in different anion solutions and reported anodic and cathodic Tafel slopes of RT/2F. 12 In the present study, with increasing the iron (II) concentration, the Tafel slope of iron deposition gradually decreases with the increase of -2.2 -2 -1.8 -1.6 -1.4 -1.2 -0.25 -0.2 -0.15 -0.1 log(Current Density), A/cm 2 Overpotential FED , V -3.6 -3.4 -3.2 -3 -2.8 -2.6 -2.4 -0.55 -0.5 -0.45 log(Current Density), A/cm 2 Overpotential HER , V B A 83 concentration from 0.6 to 3 M. A linear relationship between Tafel slope of iron deposition and electrolyte pH could be discovered that a rise in pH by 1 unit increases the Tafel slope by 32 mV/decade. That represents a change of the symmetry factor of iron deposition reaction from 0.26 to 0.54 when the concentration is raised from 0.6 to 3 M. Table 3-3: Tafel parameters of iron deposition and hydrogen evolution reactions Iron (II) Chloride Concentration, M pH Tafel slope mV/dec Exchange current density, A/cm 2 FED HER i o FED x10 5 i o HER x10 8 0.6 3.21 115 110 57 5.34 1.0 2.90 93 119 35 8.73 2.0 2.60 69 96 16 0.75 2.5 2.21 65 110 7.1 1.87 3.0 2.00 55 104 7.8 1.05 According to literature data reported by Bockris, the exchange current density of hydrogen evolution on iron varies from 10 -8 to 10 -5 A/cm 2 . 12, 13 We found that when iron (II) concentration is changed from 1 M (pH=2.9) to 3 M (pH=2.0), i 0 HER decreased by 8 fold. Further, the i 0 HER on iron in 3 M iron (II) chloride is 50 times lower compared to that of the reported value in 0.01 N HCl. 31 Such changes could be attributed to the strong adsorption of chloride or FeOH + on the iron surface. 3.3.5 Impedance Spectroscopy and Mechanisms of Iron Deposition and Hydrogen Evolution on Iron in Iron Chloride Solutions In both Volmer and Heyrovsky mechanisms of HER, the reaction rate is a function of the activity of water. 15 Thus, literature reports that in the presence of higher concentrations of metal ions, even though the concentration of protons is high, the activity of water is changed, and it leads to lower the exchange current density of HER. 15 Volmer and Tafel steps in HER show typical 84 Tafel slopes of 120 and 30 mV/ decade respectively. The Heyrovsky mechanism can show a typical Tafel slope of 40 mV/decade and at higher coverage (q H>0.6) of protons a Tafel slope of 120 mV/decade. 6 In the studied range of the concentrations of iron (II), the Tafel slope and the symmetry factor of HER were found to be 108±12 mV/decade and 0.28± 0.03 respectively. However, with no further studies it is difficult to attribute the mechanism for HER to either the Volmer or the Heyrovsky mechanism. Figure 3-8: A and B: EIS of iron deposition reaction in 0.6 and 3.0 M iron (II) chloride solution, C: Experimental (red) and fitted (green) EIS data with a proposed equivalence circuit (D) of iron deposition process in 0.6 M iron (II) chloride solution (Area of the electrode is 23 cm 2 ). Electrochemical impedance spectroscopy can give better understanding on complex reaction pathways. When low-frequency inductive loops are observed they can be ascribed to faradaic reactions that involve adsorbed intermediate species. The impedance response of such systems 5 0 0 0.2 0.4 0.6 0 0.2 0.4 -5 0.6 Z real , ohm log(Frequency), Hz -Z img , ohm -4 -2 0 2 4 log(Frequency), Hz -20 -10 0 10 20 30 40 Phase of Z, deg 3 M 0.6 M A B C D -4 -2 0 2 4 log(Frequency), Hz -20 -10 0 10 20 30 40 Phase of Z, deg 3 M 0.6 M 0 0.2 0.4 0.6 0.8 Z real , ohm 0 0.2 0.4 0.6 0.8 -Z img , ohm 1000 100 10 0.1 0.01 0.001 Exp Fit R ohm R ct R f R f 85 can be explained using inductor equivalents in the electrical circuit (Figure 3-8D). The inductive impedance (Z L) scales directly with frequency and can be represented by Eq. 3-15. Z L= 2 p f L (3-15) Where f=frequency in Hertz and L=inductance in Henry. Table 3-4: EIS data of iron deposition at -0.7 V vs. NHE at different ion (II) activities Iron (II) Chloride Concentration, M R ohm, Ω cm 2 R ct, Ω cm 2 C dl x10 4 , C cm -2 R f Ω cm 2 L 1, Henry cm 2 R f Ω cm 2 L 2, Henry cm 2 0.6 3.3 12.4 1.6 5.0 17.4 0.7 3.0 3.0 4.3 5.1 3.4 2.3 6.7 0.5 1.3 Keddam reported an inductive behavior in EIS (measured between 10 -2 and 10 4 Hz) during iron dissolution in acidic media and verified that iron dissolution occurs via an adsorbed species at the electrode surface. 32, 33 It is believed that the iron deposition mechanism is analogous to the anodic dissolution and electrodeposition of polyvalent metals occurring stepwise via a metal hydroxide intermediate. 9, 11, 12, 32-36 According to the mechanism proposed by Bockris et al. or Heusler, iron deposition occurs in two steps wherein the first step, FeOH + species can be reduced and adsorbed on the metal surface and adsorbed FeOH can be further reduced or facilitate another FeOH + reduction to metallic iron (Figure 3-9 and Eqs. 3-8 to 3-9). Since, it is well known that adsorbed species can undergo reactions faster than the species in solution can, rate-determining step of iron deposition cannot be the second electron reduction step when mechanism Bockris et al. is followed. 37 This could be verified from the observation of lower reactance values for the faradaic reactions associated with the adsorbed intermediates than the charge transfer resistance through capacitive loop (Table 3-4) in our experiments. Further, I could observe a lower R ct at higher iron (II) concentration (Figure 3-8A), identifying pH and concentration of FeOH + as rate governing factors for iron deposition. From the changes of R ct values (Figure 3-8A) and changes of Tafel slopes (Figure 3-7A) of iron deposition with 86 changing activities of iron (II), we can assume that iron deposition mechanism can also vary with the activity as well. In addition to adsorbed FeOH + , other adsorbed species such as H + ads, Cl - ads, inhibitors and intermediates of iron deposition (Figure 3-9) can also impede the electron flow. Inductive loops in EIS representing the adsorbed species are used as indirect methods for detecting intermediates in electrochemical reactions at the electrode surface. 35, 38 During metal deposition multiple-inductive loops can appear and some of them can be distorted with desorption of some adsorbed inhibiting species which are associated with the nucleation and growth of the metal deposit. 35, 36, 39 Compared to sulfate electrolytes, chloride baths give lower HER during iron deposition due to adsorption of chloride species as HER inhibitors. 25 Further, Darko et al. proposed that in chloride solutions, iron deposition can occur via iron hydroxy or iron hydroxy-chloro complexes. 25 Iron dissolution in chloride solution via two parallel mechanisms with intermediates of iron-chloro complexes and non-chloro complexes is also reported in literature. 40-42 Even though the focus of the present study was not to investigate the independent effect of chloride on EIS, the additional inductive behavior shown in our data at slightly lower frequencies (Figure 3-8, between 0.01 to 1 Hz) can be attributed to adsorption/desorption chloride during iron deposition. 39 Additionally, we can observe a decrease of the impedance associated with the smaller inductive loop with increasing the concentration of the iron (II) chloride (Table 3-4). To increase the rate of the first-electron reduction step in iron deposition, other than the requirement of higher concentration of FeOH + , the adsorbates of FeOH should be desorbed from the electrode surface promptly. Relaxation of these adsorbed species depends on the interface composition (Figure 3-9). 33 High-frequency resistance, R ohm, decreases with increasing the electrolyte pH (Table 3-4) which is consistent with the increased conductivity (Figure 3-2B) suggesting the lower extent of formation of FeOH + at lower iron (II) concentrations. Additionally, we could observe an increase of the inductance during iron deposition (Table 3-4 and Figure 3- 8C) when iron (II) concentration is decreased or pH is increased indicating a slower rate of adsorption/desorption processes at the iron surface. These results are similar to the effect of 87 current density and electrode potential on electro-crystallization of nickel reported by Epelboin. 36 Figure 3-9: Nature of the electrode surface and (A) non-catalytic and (B) catalytic mechanisms of iron deposition (d -diffusion layer thickness, q -coverage and In -Inhibitors) However, as shown in step 2 in Figure 3-9, OH - can be a reaction intermediate from the second electron reduction step which is released from the adsorbed FeOH (Figure 3-9A) or the two- electron reduction of FeOH + (Figure 3-9B). As with decreasing the iron (II) concentration, the only species whose concentration is increased is hydroxyl ions and we can assume that the increased kinetics at lower pH is due to fast desorption of hydroxyl ions generated at the surface after the second electron reduction. Additionally, desorption of these OH - ions from the electrode surface can open sites for new FeOH adsorption which are associated with the first- electron reduction process. Also, we can expect desorbed OH - to drive the iron (II) hydrolysis reaction to generate more FeOH + near the electrode surface. (A) (B) 88 While according to the proposed mechanisms by Bockris et al., FeOH + and H + are present in two different steps (Eqs. 3-7, 3-8b and 3-9b)), experimental findings of Tafel parameters and EIS data analysis of iron deposition for different concentrations of iron (II) from 0.6 to 3.0 M, suggest that both concentration and pH will both notably increase the iron deposition rate. We can predict that the rate is limited at higher pH as OH - desorption plays a crucial role at the electrode surface. EIS with steady-state kinetics studies reveals that iron deposition reaction is a function of bulk FeOH + concentration and at a given bulk concentration, the rate of the second electron reduction reaction will be governed by the activity of OH - near the electrode surface. 3.3.6 Evans Diagram Construction Using Tafel slopes, exchange current densities, open-circuit potentials and equilibrium potential calculations (Table 3-3 and Appendix B2) we can construct Evans diagrams of the iron deposition and HER during iron deposition process. At all concentrations exchange current densities for HER are 10 3 fold lower than that for iron deposition. Tafel slopes of iron deposition are lower compared to HER. A representative Evans diagram of metal deposition/ corrosion and HER/HOR at different iron(II) concentrations from 1 to 3 M can be established as Figure 3-10A. Bockris reported that near the corrosion potential, HER can occur via proton adsorption and with increasing the current density or electrode potential, a limiting current of HER from protons can lead to change the hydrogen evolution reaction to occur from water adsorption. Near the equilibrium potential of iron corrosion (E corr), a lower Tafel slope of HER (20 mV/decade) compared to that at higher overpotentials could be observed in our experiments when the iron electrode was cathodically polarized from open-circuit to more negative potentials (Figure 3- 10C). This observation is similar for all the concentrations of iron (II) from 0.6 to 3 M. Regardless of the concentration of iron (II) or chloride ions, our observation of two different Tafel slopes of HER is consistent with Bockris’s suggestion of different mechanisms of HER in different potential regions. Therefore, a modified Evans Diagram over the range of potentials, from E corr to higher electrode potentials, can be constructed as Figure 3-10B. 89 Figure 3-10: Representative Evan Diagrams of iron deposition and hydrogen evolution in iron (II) chloride solutions in higher concentrations from 1-3 M with A: HER at higher potentials and B: HER at higher potentials and near E corr (I HL=limiting current density of HER from proton) and C: Potentiodynamic curves of iron deposition using 0.6 to 3 M iron (II) chloride (p-0.6, l-1, ¢- 2, t-2.5 and ê-3M) and 2 M ammonium chloride. 3.4 Chapter Three Conclusions This study provides an understanding of the effect of electrolyte composition for the performance of all-iron redox flow battery in terms of iron (II) hydrolysis, iron (II)-chloro complex formation, and the kinetics of HER and iron deposition. Increased concentration of iron (II) improved the activity coefficient of iron (II) leading to improve the kinetics of the iron deposition and we could verify that improved kinetics of iron deposition lowered the rate of HER. In chloride electrolytes at a pH between 2.00 to 3.21, the Tafel slope for iron deposition varies from 55 to 115 mV/ decade at A -7 -6 -5 -4 -3 -2 -1 log(Charging Current Density), A/cm 2 -0.6 -0.5 -0.4 -0.3 -0.2 Overpotential HER , V 0.6M 1M 2M 2.5M 3M Log ià + Potentialà + I oFED I oHER E oFED E oHER 4 3 1 2 B C 1’ Log ià + Potentialà + I corr I oFED I oHER E oFED E corr E oHER 1. H 2 O + e - à ½ H 2 + HO - 1’. H 3 O + + e - à ½ H 2 + H 2 2. ½ H 2 + HO - à H 2 O + e - 2’. ½ H 2 + H 2 Oà H 3 O + + e - 3. Fe 2+ + 2e - à Fe 4. Fe à Fe 2+ + 2e - 4 3 1 1’ I HL 2’ 90 25 o C, indicating a pH-dependent reaction mechanism for iron deposition. This study could provide insights to verify the two-step reaction mechanism via intermediates of iron hydroxy or iron hydroxy chloro-complexes. When the mechanism proposed by Bockris is considered, the first electron reduction can be recognized as the rate determining step. Additionally, we investigated that hydroxyl ion desorption is vital for FeOH adsorption and the faster formation of FeOH + near the electrode surface is recognized as a key factor for the improved kinetics. By combining the surface pH increase discussed in Chapter 2 and the results of improved kinetics discussed in this Chapter, we show that from 0 to 80% state-of-charge, at the charging current density of 40 mA/cm 2 , a steady coulombic efficiency of 98±1% could be observed in an all-iron redox flow cell. 91 3.5 Chapter Three References 1. Jayathilake B.S., Z. B., Plichta E.J., Hendrickson M.A., Narayanan S. R. In Optimizing Coulombic Efficiency of All-Iron Redox-Flow Cell, 48th Power Sources Conference, Denver, CO, 2018; p 71-74. 2. Bockris, J. M., Electrolytic polarisation—I. The overpotential of hydrogen on some less common metals at high current densities. Influence of current density and time. Transactions of the Faraday Society 1947, 43, 417-429. 3. Bockris, J. O. M., Recent Developments in the Study of Hydrogen Overpotential. Chemical Reviews 1948, 43 (3), 525-577. 4. Conway, B.; Bockris, J. O. M., Electrolytic hydrogen evolution kinetics and its relation to the electronic and adsorptive properties of the metal. The Journal of Chemical Physics 1957, 26 (3), 532-541. 5. Roger, I.; Shipman, M. A.; Symes, M. D., Earth-abundant catalysts for electrochemical and photoelectrochemical water splitting. Nature Reviews Chemistry 2017, 1, 0003. 6. Shinagawa, T.; Garcia-Esparza, A. T.; Takanabe, K., Insight on Tafel slopes from a microkinetic analysis of aqueous electrocatalysis for energy conversion. Scientific reports 2015, 5, 13801. 7. Gaida, B., Electroplating Science: Fundamental chemistry, Electrochemistry, Physics and Electricity with particular reference to the needs of electroplating students. Draper: Teddington, 1970. 8. Gabe, D., The centenary of Tafel's equation. Transactions of the IMF 2005, 83 (3), 121- 124. 9. Hilbert, F.; Miyoshi, Y.; Eichkorn, G.; Lorenz, W. J., Correlations between the Kinetics of Electrolytic Dissolution and Deposition of Iron: I . The Anodic Dissolution of Iron. J. Electrochem. Soc. 1971, 118 (12), 1919-1926. 10. El Miligy, A. A.; Hilbert, F.; Lorenz, W. J., Kinetics of Iron Deposition on a Rotating Platinum Disk Electrode. Journal of The Electrochemical Society 1973, 120 (2), 247-251. 92 11. Bockris, J. O. M.; Drazic, D., The kinetics of deposition and dissolution of iron: Effect of alloying impurities. Electrochimica Acta 1962, 7 (3), 293-313. 12. Bockris, J. M.; Drazic, D.; Despic, A., The electrode kinetics of the deposition and dissolution of iron. Electrochim. Acta 1961, 4 (2-4), 325-361. 13. Bockris, J. M.; Kita, H., Analysis of galvanostatic transients and application to the iron electrode reaction. Journal of the Electrochemical Society 1961, 108 (7), 676-685. 14. Kongstein, O. E.; Haarberg, G. M.; Thonstad, J., Current efficiency and kinetics of cobalt electrodeposition in acid chloride solutions. Part I: The influence of current density, pH and temperature. Journal of Applied Electrochemistry 2007, 37 (6), 669-674. 15. Kongstein, O. E.; Haarberg, G. M.; Thonstad, J., Current efficiency and kinetics of cobalt electrodeposition in acid chloride solutions. Part II: the influence of chloride and sulphate concentrations. Journal of Applied Electrochemistry 2007, 37 (6), 675-680. 16. Robinson, R.; Stokes, R., Electrolyte Solutions Second Revised Edition Dover Publications. Mineola, NY 2002. 17. Tucker, M. C.; Phillips, A.; Weber, A. Z., All-Iron Redox Flow Battery Tailored for Off- Grid Portable Applications. ChemSusChem 2015, 8 (23), 3996-4004. 18. Hruska, L.; Savinell, R., Investigation of Factors Affecting Performance of the Iron-Redox Battery. J. Electrochem. Soc. 1981, 128 (1), 18-25. 19. Lee, M.-S., Chemical equilibria in ferrous chloride acid solution. Metals and Materials International 2004, 10 (4), 387-392. 20. MacInnes, D.; Brown, A. S., The Determination of Activity Coefficients from the Potentials of Concentration Cells with Transference. Chemical Reviews 1936, 18 (2), 335-348. 21. Randall, M.; Frandsen, M., The standard electrode potential of iron and the activity coefficient of ferrous chloride. Journal of the American Chemical Society 1932, 54 (1), 47-54. 22. Marcus, Y.; Hefter, G., Ion pairing. Chemical reviews 2006, 106 (11), 4585-4621. 23. Wells, C. F.; Salam, M. A., Hydrolysis of Ferrous Ions : a Kinetic Method for the Determination of the Fe(II) Species. Nature 1965, 205, 690. 24. Heinrich, C. A.; Seward, T. M., A spectrophotometric study of aqueous iron (II) chloride complexing from 25 to 200°C. Geochimica et Cosmochimica Acta 1990, 54 (8), 2207-2221. 93 25. Grujicic, D.; Pesic, B., Iron nucleation mechanisms on vitreous carbon during electrodeposition from sulfate and chloride solutions. Electrochimica acta 2005, 50 (22), 4405- 4418. 26. Finšgar, M.; Jackson, J., Application of corrosion inhibitors for steels in acidic media for the oil and gas industry: A review. Corrosion Science 2014, 86, 17-41. 27. Hluchan, V.; Wheeler, B. L.; Hackerman, N., Amino acids as corrosion inhibitors in hydrochloric acid solutions. Materials and Corrosion 1988, 39 (11), 512-517. 28. Ferreira, E. S.; Giacomelli, C.; Giacomelli, F. C.; Spinelli, A., Evaluation of the inhibitor effect of l-ascorbic acid on the corrosion of mild steel. Mater. Chem. Phy. 2004, 83 (1), 129-134. 29. Vishwanatham, S.; Haldar, N., Furfuryl alcohol as corrosion inhibitor for N80 steel in hydrochloric acid. Corrosion Science 2008, 50 (11), 2999-3004. 30. Zheng, J.; Yang, T.; Zhou, J.; Xu, M.; Zhang, X.; Rao, Z., Elimination of a free cysteine by creation of a disulfide bond increases the activity and stability of Candida boidinii formate dehydrogenase. Applied and Environmental Microbiology 2017, 83 (2), e02624-16. 31. Pentland, N.; Bockris, J. M.; Sheldon, E., Hydrogen evolution reaction on copper, gold, molybdenum, palladium, rhodium, and iron mechanism and measurement technique under high purity conditions. Journal of The Electrochemical Society 1957, 104 (3), 182-194. 32. Keddam, M.; Mattos, O. R.; Takenouti, H., Reaction model for iron dissolution studied by electrode impedance II. Determination of the reaction model. Journal of The Electrochemical Society 1981, 128 (2), 266-274. 33. Keddam, M.; Mottos, O. R.; Takenouti, H., Reaction model for iron dissolution studied by electrode impedance I. Experimental results and reaction model. J. Electrochem. Soc. 1981, 128 (2), 257-266. 34. Heusler, K.; Cartledge, G., The influence of iodide ions and carbon monoxide on the anodic dissolution of active iron. Journal of the electrochemical society 1961, 108 (8), 732-740. 35. Epelboin, I.; Ksouri, M.; Wiart, R., On a model for the electrocrystallization of zinc involving an autocatalytic step. Journal of The Electrochemical Society 1975, 122 (9), 1206-1214. 36. Epelboin, I.; Wiart, R., Mechanism of the electrocrystallization of nickel and cobalt in acidic solution. Journal of the Electrochemical Society 1971, 118 (10), 1577-1582. 94 37. Brett, C. M. A., Electrochemistry : principles, methods, and applications. Oxford University Press: Oxford ;, 1993. 38. Amin, M. A.; Khaled, K.; Mohsen, Q.; Arida, H., A study of the inhibition of iron corrosion in HCl solutions by some amino acids. Corrosion Science 2010, 52 (5), 1684-1695. 39. Wiart, R., Elementary steps of electrodeposition analysed by means of impedance spectroscopy. Electrochimica Acta 1990, 35 (10), 1587-1593. 40. Kuo, H.; Nobe, K., Electrodissolution kinetics of iron in chloride solutions VI. concentrated acidic solutions. Journal of the Electrochemical Society 1978, 125 (6), 853-860. 41. Itagaki, M.; Tagaki, M.; Watanabe, K., Active dissolution mechanisms of iron using EIS with channel flow double electrode. Influences of chloride and fluoride ions. Electrochimica Acta 1996, 41 (7), 1201-1207. 42. Bech-Nielsen, G., The anodic dissolution of iron—VII: A detailed kinetic model for the two coupled, parallel anodic reactions. Electrochimica Acta 1976, 21 (8), 627-636. 95 Critical Analysis of the Performance of Lithium-Sulfur Batteries with Mixed Conduction Membrane 4.1 Background The operation of the lithium-sulfur (Li–S) battery involves lithium dissolution and lithium plating at the negative electrode, and conversion of elemental sulfur to lithium sulfide at the positive electrode via a step-wise multi-electron process. These processes at the positive and negative electrode present technical challenges for the industrial application of the Li-S battery as a rechargeable energy storage system. There are a number of studies describing these charge- discharge processes in the Li-S battery . 1, 2 In this chapter, we focus on studying and analyzing the discharge process at the positive electrode with the goal of understanding the factors affecting utilization and rate capability. 4.1.1 Features of Discharging Curve During discharging of the Li-S cell, at least five regions can be identified on the discharge curve pertaining to the reactions of sulfur (Scheme 1). 1, 3 The first step is the formation of soluble higher order polysulfide from elemental sulfur, given by the reaction (I) S 8 0 àS 8 2- . S 8 2- anions are dissolved into electrolyte and lead to two reduction reactions in liquid phase as (II) S 8 2- àS 6 2- and (III) S 6 2- àS 4 2- , also sometimes represented as S 8 2 àS 4 2- . 3 A quarter of the theoretical discharge capacity of the sulfur electrode is due to formation of soluble polysulfides. Kinetics of these reactions can be expected to be faster than those reactions that involve two solids. The final two steps of the Li-S discharging process result in the formation of the disulfide and sulfide, shown as (IV) S 4 2- àLi 2S 2 and (V) Li 2S 2 à Li 2S. Often the deposition of Li 2S 2 and Li 2S on cathode from soluble polysulfides is reported as a combined process and is represented as S 4 2- àLi 2S . 4, 5 Steps IV and account for three-fourths of the theoretical capacity. The last step is known to have the largest overpotential losses because of the sluggish diffusion of lithium ion in the solid materials and poor conductivity of Li 2S. 96 Scheme 1: Charging and discharging curves of Li-S battery from the reference 6 According to Scheme 1, plateaus on the voltage-capacity curves are associated with the transformations involving two solid phases (Steps IV and V). Where solid transforms into a solution or if the products and reactants are both in solution (as in Steps I, II and III), the discharge curves are sloping. Polysulfide species, S 8 2− , S 6 2− and S 4 2− are soluble in the electrolyte and the equilibrium potential of the reduction reactions involved with these species’ changes with the activity of the dissolved species (Eqs. 4-3 and 4-4). > P Q /ST F P Q $ = F.U V W) ST X /ST (4-1) > ST F P U /ST F P WTY ST F P F Z = F.; V W) ST X /ST (4-2) >=> ; $ [ + 34 F2 @1 [P Q F< ] _ [P M F< ] U E 1 o’ = 2.32 V vs. Li + /Li 7 (4-3) >=> F $ [ + 34 F2 @1 [P M F< ] F [P U F< ] _ E 1 o’ = 2.22 V vs. Li + /Li 7 (4-4) Since, Li 2S 2 and Li 2S are insoluble, a solid-state reaction can be anticipated. In this region the Nernst potential does not vary with the state-of-charge. However, a steep drop in the voltage can be observed upon formation of a non-conductive layer of Li 2S. Li 2S formation can decrease the effective-conductive surface area of the cathode and hinder further discharge of the cell. 97 4.1.2 Polysulfide Shuttling One of the most significant issues in Li-S battery technology is the shuttling of soluble polysulfides between the cathode and the anode during both charging and discharging of the battery. 1, 2, 7 A cell continues to accept charge indefinitely and the discharging is reduced as a consequence of the shuttling. Thus, coulombic efficiency of the system is poor. When the soluble polysulfides react with lithium metal and form insoluble sulfides at the anode, there are other implications to the cell performance. Therefore, different approaches to prevent the polysulfide shuttling are being researched upon. 8-12 Dr. Derek Moy and Dr. S. R. Narayanan at USC introduced a mixed-conduction membrane (MCM) as a viable barrier for polysulfide movement between the lithium and sulfur electrodes. 8 Moy proposed that lithium-ion movement occurs through MCM via electrochemical intercalation/de-intercalation processes. 8 « Charge transfer, « Diffusion and ® Li + Intercalation Figure 4-1: Lithium ion transport via (A) and (B): dense, (C): porous and (D): porous and non- porous layered MCMs In the present study, attempts to validate the intercalation mechanism of lithium ion transport through MCM. To this end we have tested MCMs of different porosity levels and surface area for their effect on the polysulfide movement (Figure 4-1, C and D). Further, to enhance the rate if transfer of lithium ion to and from the cathode via MCM, a favorable cell configuration was Cathode Anode MCM with porous and non-porous layers Li + Polypropylene separator Li + Li + Li + Cathode Anode Dense MCM Li + Polypropylene separator Li + Li + Li + Cathode Anode MCM Li + Polypropylene separator Li + Li + Li + Cathode Anode MCM Li + Polypropylene separator Li + Li + (A) (B) (C) (D) 98 tested (Figure 4-1B). In this configuration, the cathode and MCM are assembled together without a barrier separator. Thus, lithium-ion diffusion and access to the cathode can expected to be improved compared to the cell configuration with a separator between the MCM and the cathode. 4.1.3 Rate Dependent Li2S Formation Three quarters of Li-S discharge capacity come from the formation of Li 2S from S 4 2- polysulfide via formation of the intermediate Li 2S 2. Conductivity of Li 2S is reported as 10 -14 S/cm. 3 Thus, current flow even through a very thin film of Li 2S is expected to be negligible due to the high resistivity. 3 It is recognized that this poor conductivity and chemical reversibility of Li 2S are also main reasons for low material utilization and poor rechargeability of the of Li-S system. 13 Chiang et al. has proposed that the surface reaction kinetics govern the Li 2S formation. 14 Nucleation on conductive substrate and 3D growth are considered as main pathways of Li 2S deposition (Figure 4-2). 13, 14 With the 2D growth of Li 2S, coverage increases at triple phase boundaries formed by the precipitate, substrate and electrolyte. At a monolayer of full 2D coverage, electrode potential can reach the voltage cut-off of the cell operation due to a large IR drop. Further, the morphology of Li 2S is dependent on the rate of discharge (Figure 4-2). 14 Li 2S formation at higher rates can particle growth as a film can decrease the discharge capacity. 3, 5, 15 However, in contrast, at lower rates larger size particles can be formed during Li 2S formation leading to a precipitate and thus higher capacity will be retrieved before complete coverage of the cathode surface occurs. 3, 14 The same phenomenon occurs in a lithium-air cell during Li 2O 2 precipitation where a film-like layer can be formed at higher rates. 15 Porosity and volume of the cathode construction play a key role in the nucleation rate and 3D growth. Also these properties affect the mechanical stability during Li 2S formation due to volume expansion upon sulfur lithiation. 6 When the cathode structure is covered with non-conducting Li 2S, major current flow (i) during discharging can happen through uncovered part and the Tafel equation for the Li 2S formation can be modified based on the coverage (q) as per Eq. 4-5. T=T ` (;−a)+(,( C0 S 2 34 ) (4-5) 99 Where i 0= exchange current density, R=gas constant, F=faraday constant, T= temperature, a is the transfer coefficient and h L is the overpotential. Figure 4-2: Effect of discharging rate for Li 2S deposition 4.1.4 Characterization of Li-S System Using EIS In a real cell, the predicted cell voltage can deviate from the theoretical values due to a number of factors such as viscosity, conductivity, cathode construction, solid-electrolyte interphase properties, mass transport and charge transport kinetics. 1, 2, 16, 17 Sulfur and its various discharge products show poor ionic and electronic conductivities and as a result, we observe a large polarization loss that reduces the energy efficiency of the battery. The poor conductivity also arises from the poor dissolution of polysulfides in electrolyte and the formation of an insoluble insulating layer on the surface of the cathode during deep discharge. The latter factor leads to a poor active material utilization. Electrochemical impedance spectroscopy, in-situ X-ray diffraction spectroscopy, in-operando optical imaging and a number of other analytical tools are used to analyze the features of the voltage vs. capacity curves during charge and discharge curves of Li-S batteries. 1, 18, 19 EIS is a valuable tool for studying electrochemical processes and chemical and physical changes occurring at the electrode surface. Literature shows that the discharge curves can be analyzed based on EIS results. 18, 20-22 Electrical equivalent circuits (Figure 4-3) for the Li-S cells contain at least four resistance features namely, electrolyte resistance (R ohm), Increasing the discharging time Higher current density Lower current density Nucleation center Li2S deposition Conductive Cathode surface End of discharge High 3D growth Low 3D growth Slow nucleation rate Fast nucleation rate 100 charge transfer resistance (R ct), Warburg impedance (W o) due to diffusion of ions and capacitance (C i) to compensate the non-ideal behavior of electrodes. 18, 20, 23 Additionally, interphase contact resistance (R int) has also been discussed in some reports. 18 Figure 4-3: Electrical equivalent circuit for Li-S system 4.2 Experiments and Methods Sulfur electrodes and single layer MCM membranes were fabricated as described in Derek Moy’s Ph.D. dissertation and publications. 8, 24 Layered-MCM was fabricated using three layers of single membranes (15 microns) as explained below. A slurry of LCO membrane coating was prepared by stirring a mixture of lithium cobalt oxide (LCO,7.6 g), N-methyl-2-pyrrolidone (NMP, 7 mL), polyvinylidene fluoride (PVDF, 0.4 g) in a glass beaker. To prevent solvent evaporation during mixing, the top of the beaker was closed using a glass petri dish. Initially NMP and PVDF were mixed for 30 min until the PVDF dissolved in NMP to form a clear solution. Then LCO was added and mixed for 18 hours using 3 magnetic stirring bars. LCO membranes (thickness of 15-75 um) were coated on a clean aluminum foil on the heater using a doctor blade at a speed setting of 4-5. The coating was dried at 80 o C (setting of 4.5 on the vacuum oven) for 5 hours. Then the dried coating was placed in a conventional oven for 16 hours. Dried coatings were initially hot-pressed as shown in Table 4-1. Then the aluminum layer was removed using freshly prepared 10 M potassium hydroxide (56 g/100mL) solution which was slightly warm. Excess potassium hydroxide was removed by rinsing in DI water for 3 times and soaking in DI water for 24 hours. Then the cleaned membranes were oven dried for 16 hours to remove moisture. Single layers were combined together as shown in Table 4-2 and hot-pressed at high temperature (180 o C) but at a low pressure (15 lbs/ cm 2 ). SEM images of single layers of mixed conduction membranes were collected using FEI Nova NanoSEM 450 microscope. C 1 R ct R ohm R int C 2 101 Table 4-1: Single-layer mixed conduction membrane preparation Single Membrane label Initial Pressing KOH treatment A not Treated after the initial oven drying step B 1243 lbs/cm 2 (Step II-1) Treated after initial hot-pressing step Table 4-2: Multi-layered mixed conduction membrane preparation Assembly Layers (see Table 4-1) Final pressing temperature and time Expected properties of LCO membrane ABA 30 o C, 30 min followed by 180 o C, 10 min High surface area for intercalation Medium to high non-porosity in the middle layer AAA 180 o C, 10 min High surface area for intercalation High porosity in all the layers BBB 180 o C, 10 min Low surface area for intercalation High non-porosity in all the layers 102 4.3 Results and discussion 4.3.1 Polysulfide Shuttle Hindrance and Charge Transfer via Lithium Ion Intercalation Through MCMs Verification of Derek Moy’s approach of the use of MCM membrane to improve the Li-S discharging capacity is shown in Figure 4-4 A. About 11-16% improvement in the utilization of sulfur during discharge confirms that more of the soluble polysulfides remain at the cathode side during the discharge process for the cell configuration with MCM. SEM images in Figure 4-5 show that smaller particles are fused into bigger sizes (about 50%) when the membranes were hot-pressed while bigger pores (about 2-3 times) can be seen in the membranes that were not pressed. Additionally, SEM images show that compared to mixing using magnetic stirring, mixing with ball milling gives smaller particle sizes. Thus, surface area of the membrane can be higher when the MCM was prepared using ball milling. With higher porosity and smaller particle size in AAA membrane, higher surface area of MCM can be expected. Thus, improved lithium ion intercalation can be expected with AAA type MCM during the operation of the cell. Better voltage efficiency, lower impedance at low frequency and higher utilization in the cell with AAA type MCM compared to the cell with no-membrane l suggest that the overall reduction kinetics of Li-S system during discharging improved with the introduction MCM. Due to the higher surface area in AAA type MCM with higher porosity, higher rate of lithium ion intercalation was predicted. However, poor charging features in AAA membrane similar to no-membrane case indicated that higher porosity in MCM can lead to polysulfide movement during the cell operation. To overcome this, we introduced an advanced sandwich layered membrane with porous layers at the outside and non-porous MCM layer at the middle. This, ABA type MCM with improved surface area and polysulfide shuttling barrier gave voltages in discharging step similar or better to the configuration without the membrane. Further, well defined features during both charging and discharging could be observed in the cells that are constructed with ABA layered MCM. These features verify that little or no polysulfide shuttled with this MCM configuration. However, BBB type MCM gave slightly lower cell voltage indicating poor discharge kinetics due to lower surface area for the lithium ion intercalation. Changes of the 103 voltages of 1 st and 2 nd discharge plateaus in layered membrane containing cells can be account for the changes in initial impedance at low frequency (Figure 4-4B). We note that with the increasing the impedance, cell voltage of 1 st and 2 nd discharging plateaus shift to more negative potentials. As all three cells with MCM membranes show better utilization of S cathode than a cell without the membrane, we can verify that polysulfide shuttling is hindered with MCM. However, 1/8 th of the capacity loss during S 4 2- D Li 2S transformation in both charging and discharging curves with non-porous MCM verifies that other than to the polysulfide shuttling, utilization of active material suffers from factors such as poor conductivity of Li 2S. Figure 4-4 (A): Discharging curves and (B): EIS of Li-S with KB cathodes using different layered membranes, Cell parameters: 15 µm thick cathode, 2 cm 2 of cell area, KB carbon with 63.3% S, 100 µL of electrolyte volume (extra 40 µL for ABA and BBB type MCM) and at C/64 rate. Li 2 S 8 / Li 2 S 6 Li 2 S 6 / Li 2 S 4 Li 2 S 4 / Li 2 S 2 Li 2 S 2 / Li 2 S 0 500 1000 1500 2000 Z real , ohm 0 500 1000 1500 2000 -Z img , ohm E157-no MCM E164-MCM-AAA E168-MCM-ABA E169-MCM-BBB (A) (B) No MCM Porous MCM Dense MCM Layered MCM Dense MCM 104 Figure 4-5: (A,C,F and H ) SEM images of pressed, (B,D,G and I) SEM images of not pressed and (E) Visible appearance of 15 µm thick single layer of MCM (A-E are thickness of 15 µm, slurry prepared using mechanical stirring and potassium hydroxide treated to remove the aluminum support and F-I are thickness of 75 µm, slurry prepared using ball milling and no potassium hydroxide treatment). 105 4.3.2 Improved Initial Cycling with Mixed Conduction Membrane Figure 4-6: Coulombic efficiency and specific discharge capacity of Li-S cells with MCM (●E101) and with no MCM (▼E157). Cell parameters: 15 µm thick cathode, 2 cm 2 of cell area, KB carbon with 63.3% S, 100 µL of electrolyte volume and at C/64 rate. In addition to the improved capacity in the initial discharging, MCM gives the advantage of capacity retention with cycling. Derek Moy has thoroughly studied the capacity retention of MCM with AB cathodes. 8 Figure 4-6 shows that initial cycling of a Li-S cells and verifies that MCM can facilitate the capacity retention in KB cathodes. For initial eight cycles, utilization higher than 850 mAh/ g could be achieved at higher 90% of coulombic efficiency with MCM membrane. Contrastingly, a similar type of cathode assembled with no MCM membrane Figure 4-6 shows more than two-fold loss of its initial capacity in the subsequent discharge cycles. Further, in the absence of MCM, the coulombic efficiency was half of that of a cell with MCM. This observation implies that the main loss of discharge capacity in cells in the absence of MCM is due to polysulfide shuttling in the charging step. After the sudden capacity loss from initial to 1 st cycle in the cells with no MCM, we can observe that the rate of capacity loss is lower compared to that in a cell with MCM. Even though the rate of operation is the same from cycle to cycle, an increase of the voltage gap between charging and discharging plateaus can be observed (Appendix C1, Figure C1-1). Further, in cells without MCM cell, even though the polysulfides are free to move across the separator, we could observe a similar shift of the charging and discharging plateaus voltages (Appendix C1, Figure C1-2). Combined observation of the voltage shifts in the cells containing porous MCM membrane and no MCM membrane 1 2 3 4 5 6 7 8 Cycle number 20 40 60 80 100 %Coulombic Efficiency 200 400 600 800 1000 1200 1400 1600 Specific Capacity, mAh/g CE with MCM CE with no MCM Cap with MCM Cap with no MCM With MCM With no MCM 106 suggested that there can be a common kinetic barrier for redox reaction with increasing the cycling. It can be the S redistribution issues at the cathode such as non-reversible Li 2S formation or consequence of polysulfide shuttle such as Li 2S formation at the lithium anode. 3, 18 As discussed above, polysulfide shuttling complicates both charging and discharging properties of the cell. Therefore, for simplicity in the present analysis, we focus on the discharge process and evaluate the voltage shifts of the discharge plateaus. 4.3.3 Analysis of Initial EIS of Li-S Cells Figure 4-7 provides a summary of low frequency impedance and initial discharge capacity of cells assembled in different configurations. These cells were prepared with cathodes using AB, CNT and KB carbon supports and may contain single or layered MCM. Further, as described in the introduction, in some cells, the cathodes were assembled together with MCM (configuration shown in Figure 4-1B). This Figure shows that even though the utilization of S at lower rates are higher in KB cathodes, most of these cells have significantly higher impedance values. Higher impedance at lower frequency suggests that rate capability of these cells is likely to be poor due at higher rates. Distinguishing the impedance as mass transport or charge transport requires further investigations such as equivalent circuit fittings to the experimental EIS data and galvanostatic EIS measurements at different states-of-charge. AB cells show the lowest impedance among studied cells. However, utilization of AB electrodes was lower than KB and CNT electrodes. Further, we found that AB: KB mixed carbon electrodes show the combined effect of impedance and utilization during the initial discharge. There is a general trend showing an increase of the initial discharging capacity with decreasing the impedance in a similar type of cells (lines shown in Figure 4-7). Difference of impedance prior to discharge is significant among different cathodes, and we could find that the differences in membrane construction and cell configuration with membrane gives differences in impedance and utilization. However, all the cells analyzed in Figure 4-7 show a higher impedance compared to literature reports. Thus, we can expect that the higher impedance is due to a common factor such as poor structure of carbon-supported cathode, poor wettability of electrodes or formation of precipitation from the soluble species. 107 Figure 4-7: Initial Specific Capacity vs. EIS (At the frequency of 0.1 Hz) of different Li-S cell configurations before cycling I could find similar higher R ct in literature reported by Xiong et al. showing significantly higher charge transfer resistance in Li- Li symmetric cells assembled with lithium nitrate additive. 25 They found that when polysulfides are present in the electrolyte, lithium ion migration can be higher. 25 Compared to the AB-S/ Li cells prepared by Derek Moy, I observed a five times higher R ct with MCM, Li/ Li symmetric cells. 8, 24 However, in our cells (shown in Figure 4-7 ) we could see 5-100 times higher impedance at low frequency compared to Derek’s data. Therefore, we can conclude that the main contribution for the increased resistance in these cells does not come from lithium anode but from the cathode and other components of the cell. 4.3.4 Optimum Electrolyte for Better Ion Transport To achieve high energy density in the Li-S system, we need to reduce the amount of the redox- inactive material including the electrolyte. Thus a ratio of 4:1 weight/weight of electrolyte/sulfur was chosen as the maximum ratio. 6 Electrolytes consisting of dissolved lithium salt in an organic 0 200 400 600 800 1000 1200 1400 1600 0 500 1000 1500 2000 2500 3000 3500 Initial Specific Dischraging capacity, mA/g Intial Impedance at lower frequency, ohm KB 15 um KB 15 um with single LCO KB 15 um with layered LCO semi-cracked KB 15 um with extra spacer KB:AB 15 um with single LCO KB:AB 15 um with layerd LCO AB 15 um with single LCO AB 15 um with layered LCO AB 50 um AB 50 um with extra spacer CNT 15 um with single LCO 108 solvent provide lithium ion transport between the anode and cathode. Even though an adequate distribution of electrolyte is required to facilitate a continuous charge transport, an excess of electrolyte can affect the battery performance due to the solubility of higher order polysulfides. The most significant issue of using excess electrolyte is the capacity loss due to the transfer of dissolved polysulfide from the cathode to the anode which leads a permanent damage to the battery through the formation of non-conducting Li 2S at the anode surface. Even when barrier layers are introduced to overcome the soluble polysulfide transfer from cathode to anode, dissolved polysulfide movement within the cathode structure itself can change the distribution of sulfur in the cathode. Due to this maldistribution, sulfur can lose the access to the conducting part of the cathode. Further, increased dissolution of higher order polysulfide can increase the viscosity of the electrolyte. Thus, calculation of the optimum volume of electrolyte to compensate for the losses due to the viscosity factor is also significant to achieving a reliable operation of the battery throughout a single charge-discharge cycle. Additionally, Derek Moy discussed that electrolyte depletion due to the reactivity with lithium metal and water. 24 He reported a higher R ct in a Li-Li symmetric cell due to depleted electrolyte loss of charge transport capability. 24 Murugan et al. also reported that higher R ct is due to interfacial resistance which arises from the poor wettability at the interface. 12 Therefore, achieving low impedance requires better interface wettability. Zheng et al. reported 20 µl per 1 mg of sulfur as the optimum volume of electrolyte to achieve the optimum conditions of viscosity, wetting capability, lithium ion diffusion, inhibition of the lithium metal corrosion, polysulfide dissolution and Li 2S precipitation. 26 In layered ABA type MCM cells, a better capacity retention can be observed in the initial cycling at lower rates (Figure 4-8). Even though, polysulfide shuttling to lithium anode could be lowered with a non-porous LCO layer when a minimum amount of electrolyte is used, a higher impedance at lower frequency (Appendix C2, Figure C2-1) was observed. This observation indicated that lithium ion transfer via intercalation is impeded due to poor wettability of the porous layer of MCM. At lower electrolyte volume a significant change in the viscosity can be predicted. Thus, during discharging, other than to precipitation of the viscous polysulfide at the cathode, some 109 of this precipitate can be deposited inside the pores of the ABA membrane (Figure 4-8C). This may lower the active surface area at the cathode and MCM interface for the lithium ion intercalation process. As a result, mass transport or ion transport impedance can be higher. Therefore, the operation of the Li-S cell can be hindered as shown in Figure 4-8. To minimize the impedance, increase due to viscosity increase, we may need sufficient electrolyte to wet electrodes as well the MCM. Figure 4-8: Effect of discharging capacity and features with lower volume of electrolyte; Cell parameters: 15 µm thick cathode, 2 cm 2 of cell area, KB carbon with 63.3% S, 100 µL of electrolyte volume and at C/64 rate. However, in a similar cell construction, addition of extra 40 µL electrolyte to wet the porous layers in ABA layered LCO membrane showed that membrane wettability is significant to 1 2 3 4 5 6 Cycle number 20 40 60 80 100 %Coulombic Efficiency 200 400 600 800 1000 1200 1400 1600 Specific Capacity, mAh/g PS(l) Cathode Anode MCM with porous layer Precipitated polysulfide inside pores (A) (B) (C) 110 achieve better lithium ion intercalation in the sandwich layer of porous and non-porous MCM. This hypothesis was verified by the improved EIS (decrease of the low frequency impedance) in the presence of extra electrolyte (Figure 4-9A, ABA type LCO). Further, better charging discharging features and superior utilization could be achieved with improved wettability (Red curve in Figure 4-4). The observation of the shift of the discharging plateau (theoretically 2.1 V vs. Li + /Li) to 100 mV more negative with lower amount of electrolyte (Red curve in Figure 4-8B) compared to the discharging plateau with adequate electrolyte (Red curve in Figure 4-4) can be due to the impedance with the precipitation of polysulfide resulting from increased viscosity. Based on the literature, R ct is better in the presence of dissolved polysulfides in the electrolyte, Derek’s experiments of higher R ct with lower electrolyte, and improved charging/discharging features and EIS in adequate electrolyte, we can conclude that observed higher impedance values in most of our experiments could be a function of the electrolyte distribution and not inadequacy of the electrolyte amount. 12, 24 4.3.5 Different Types of Cell Constructions with MCM using KB, AB and Mixed AB and KB as Carbon sources for Cathodes Table 4-3: Properties of Acetylene Black and Ketjen Black carbon particles Carbon type Pore size, nm Pore volume 27 , cm 3 /g Surface area 27 , m 2 / g Particle size, nm Acetylene black 11.4 27 , 2.5 0.16 56 27 40, 27 100 28 Ketjen black 7.7 27 1.17 818 27 34-40 29 Cathode, anode and separator are the three main components of a Li-S cell. Cathode surface area, conductive carbon pore size and distribution, and sulfur loading are key parameters that determine the utilization, degradation and rate-capability of the system. Larger pore volumes in cathode can allow increased sulfur loading. However, it might only be beneficial if all the loaded sulfur can be contacted by the carbon support as sulfur is a very poor electronic conductor. Thus, a high-performance sulfur electrode requires optimal surface area and pore 111 size distribution of carbon. In the present study Acetylene Black (AB) and Ketjen Black (KB) were used as conductive carbon supports for the cathode. Both of these materials are small carbon particles and their properties are tabulated in Table 4-3. We also tested a 1:1 mixture of AB and KB was also used as a cathode support. Other than fabrication of improved electrodes, cell construction has also a significant impact on the performance of the Li-S cell. Among different cell constructions, we noticed that performance was better when the cathode and MCM were placed adjacent to each other (that we termed “shorted” cathode cells) as shown in Figure 4- 1B. Figure 4-9: EIS of different LI-S cell constructions with (A) Ketjen Black cathode, (B) Mixed carbon (Acetylene Black: Ketjen Black, 1:1) cathodes and (C) Acetylene Black cathodes. 0 20 40 60 80 100 120 140 160 180 200 Z real , ohm 0 20 40 60 80 100 120 140 160 180 200 -Z img , ohm E182-LCO E189-LCO shorted E190-ABA 0 500 1000 1500 2000 Z real , ohm 0 500 1000 1500 2000 -Z img , ohm E191-LCO E101-LCO shorted E168-LCO ABA 0 50 100 150 200 250 300 350 400 Z real , ohm 0 50 100 150 200 250 300 350 400 Z img , ohm E195A-LCO E194A-LCO shorted E193A-ABA (A) (B) (C) 112 In all the three different cathodes, the lowest initial R ct could be observed with the “shorted” cathode cells (Figure 4-9). This observation verified that lithium intercalation was improved when MCM is placed near the cathode with no polypropylene barrier (Figure 4-1B). As we have verified earlier that layered MCM can give better intercalation (in Section 4.3.1) we have tested three different cathode structures with ABA type layered MCM. Figure 4-10: 3-D structures of carbon supported -sulfur cathodes (A) KB, (B) AB and (C) mixed carbon KB:AB and 2D structure of a single pore created with carbon particles (D) KB, (E) AB and (F) mixed carbon KB:AB (C - ● and ❍, S -n, PVDF -g and Li + -n). Due to the higher pore volume and higher surface area, the cathodes fabricated with KB were the least dense. Thus, at a given S:C w/w ratio, S loading per unit volume is lower in KB cathodes compared to that with AB. In our experiments, 2-fold lower weight of cathode for the same area in KB compared to that in AB signified that the KB electrodes occupy a larger volume and can be resistive for issues associated with volume expansion in the cathode during discharging. When we synthesize porous carbon supports, initial sulfur loading, and electrolyte can be distributed in pores which are created among carbon particles (Figure 4-10). However, cathode structures with nano-pores of KB particles can allow sulfur loading and electrolyte distribution into particle pores (Figure 4-10 A, C, D and F). These features are absent in AB cathodes and only external pore structures can be expected. Thus, volume expansion and changes the 113 mechanical properties due to lithiation is more predictable in AB cathodes compared to that in KB. 4.3.6 Electrolyte Distribution Due to the higher sorption properties of KB, coating of KB electrodes requires twice of NMP solvent compared to the amount of NMP required for AB electrode coating. With the heat treatment for drying the cathodes, evaporation of solvent could lead to the formation of cracks in the cathode which can give poor connectivity between the KB particles as a conductive carbon support. A similar observation is reported by Dong Ping et al. using KB and they explained that volume shrinkage during drying process to lead to a cracked cathode structure. 30 To overcome this, we introduced a blotting technique to remove the excess of solvent. 31 However, according to Derek Moy’s experimental evidence of poor performance with pressed cathodes, we can assume that the blotting technique may give similar effects to pressing and reduce the thickness of the cathode and the performance of the cell by decreasing the porosity of the cathode and increasing the particle density. 24 On the other hand, as coating of AB electrodes requires lower amount of solvent, we can expect a better connectivity within the carbon network when the solvent is dried. These observations suggest that the electrolyte uptake for wetting KB cathode can be higher than that for AB. At a given electrolyte: sulfur ratio, the wettability of cathode can be better with AB cathode due to less volume and lower surface area. Thus, in KB electrodes, reduced sulfur species may not be dissolved well in the electrolyte near the electrode surface and semi- solid or precipitates can be formed inside KB pores as adequate electrolyte is not available inside pores. This re-distribution can increase the internal resistance due to poor conductivity and mass transport. Such differences could account for the ten times higher initial impedance at low frequency in cells with KB cathodes compared to the cells with AB cathodes. Poor construction of the cathode working against an optimal electrolyte distribution can be a major reason for the higher impedance in most of our cathodes among a number of other reasons that were discussed in the sections above. Even though, to lower the impedance we should focus on preparing higher surface area electrodes, fast movement of lithium ion in the electrolyte is also 114 a requirement for achieving better efficiencies. Thus, higher initial impedance was an evidence for poor rate capability. When polysulfides are dissolved, electrolyte viscosity increases and as a consequence the impedance of the Li-S system increases. 4.3.7 Material Utilization and Polysulfide Distribution KB electrodes show high utilization as high as 85-90 % of the theoretical capacity at lower rates and it can be due to its high surface area and average size pore distribution compared to AB (Table 4-3Table 4-3). The higher surface area and pore volume in KB can provide room for dissolved polysulfide to stay inside hierarchical pore structures. Thus, movement of soluble polysulfide can be slower, and a lower shuttle current could be expected with KB electrodes due to conservation of polysulfide inside the KB particle pores. We could realize significantly higher initial discharging capacity in KB with single layer or layered MCMs (Figure 7-12 B). Even with no MCM (Figure 7-1) the capacity is higher than that in AB cathode with MCM (Figure 7- 12). Therefore, we can conclude that the KB cathode structure itself allows soluble polysulfide to stay within the cathode structure. This hypothesis is supported by the experimental observation of 99% polysulfide adsorption when Ketjen Black is added to a polysulfide containing solution. 32 The same properties of KB were used by Xiaoming et al. to modify the separator material to create a physical barrier for polysulfide movement to anode. 33, 34 With the benefit of KB, a higher sulfur utilization (1430 mAh/g) at 0.1 C was reported. 33, 34 Further, a “cauliflower” like cathode structure created with KB for higher sulfur loading of 14 mg/cm 2 and utilization of 1300 mAh/g were reported at 0.01C rate. 35 Thus, the assumption of residing polysulfides inside KB structures as the reason for achieving higher utilization of S is supported by the literature on KB as a cathode material or barrier layer for preventing polysulfide shuttling. 33-36 However, in AB cathode structures, faster polysulfide movement within cathode can be expected due to the absence of nano-pores. Therefore, the properties of the MCM may be more important in determining the cell performance of AB cells by preventing the polysulfide shuttle towards the anode. 115 4.3.8 Influence of Different Carbon Materials on Rate-Capability The voltage difference between different discharge curves is attributable to the differences in impedance. As shown in Table 4-4, the initial impedance is higher for KB containing cells (E101) than cells with AB-containing cathodes. The next highest impedance comes from mixed carbon cathodes. Due to higher impedance of KB-containing cathodes, significant shifts of the voltage during charge and discharge can be observed with the variation of rate (Figure 4-11 A and B). Literature reports suggest that with increasing the rate, the first plateau capacity in the discharge curve remains constant and the second plateau capacity of the same curve decreases with increasing the discharging rate. We noticed the same trend in all our discharge curves in both AB and KB cathodes. 14 Also, we found that the capacity in the 1 st plateau and that in sloping region are similar at any current density (Figure 4-11). However, the capacity of the 2 nd plateau does not show a direct relationship to the 1 st plateau and capacity from the soluble region. In addition to lowering the discharging capacity, a shift of the cell voltage in the entire discharge curve toward a lower cell voltage could be observed with increasing the discharging rate. A similar observation of a potential shift is reported in literature in Li-S coin cell and pouch cell experiments. 3, 37 From operando spectroscopic studies, Sun et al. verified that the distribution of polysulfide can control the step of formation of Li 2S 2 and Li 2S. 10, 19 In our studies, due to higher solvent sorption by the carbon structure of KB, mobile electrolyte volume can become lower. Thus, a poor conductivity of lithium ion inside cathode pores can be expected. At higher discharge rates if the diffusion of lithium ions in the electrolyte is not fast enough, higher overpotentials can be expected during the reduction of the dissolved sulfur species. The dominant potential shift is in the initial region of the discharge curves with increasing the rate in KB compared to that in AB cathodes (Figure 4-11) is consistent with the explanation of poor solubility of polysulfides in the electrolyte. Thus, we can assume that in KB cathodes at lower electrolyte volume, discharging can happen into a saturated electrolyte through a solid-solid process as proposed by Zheng et 116 al. 38 The observation of higher impedance in our KB cathodes can be due to poor mass transport of ions which limits overall rate of charge transfer over the thickness of electrode. 33 During precipitation, the nucleation step requires higher energy to overcome the surface adsorption energy barrier. 3 S 4 2- anion adsorption on conductive carbon support before being reduced to Li 2S may involve adsorption in the inner Helmholtz plane. 3 The observation of the ‘dip’ like feature in discharging curves between the 1 st discharge slope and the 2 nd discharging plateau is often attributed to the initial nucleation of lithium sulfide in crystalline phases which has a higher energy barrier compared to the 3D growth of Li 2S. 13, 14 This feature is very clearly observed in AB cathodes in all the discharging current densities. The first slope region corresponds to the increase of viscosity of the electrolyte by the formation of soluble polysulfides. However, AB electrodes show a ‘curve’ like behavior from the beginning to end of the 2 nd plateau region. A declining or inclining slope during charging or discharging can be expected when there is a Nernst potential change due to activity changes of the redox active materials. But in KB cathodes, this feature is only visible in the initial discharge curve. The overpotential drop with S 4 2- polysulfide consumption is low (lack of dip feature) in KB cathodes, and a 2 nd discharging plateau which is almost parallel to the capacity axis was observed with KB. Chiang et al. verified that Li 2S deposition on multi-walled CNT at higher rates leads to a higher nucleation rate with a uniform coating while lower rates precipitates a fewer number of micro- meter sized larger particles of Li 2S. 14 This observation indicates a 3D growth at the three-phase boundary. 14 Further, these observations are explained by overpotential-dependent nucleation rates. 14 Experimentally, we could observe a linear relationship between the discharge capacity and the overpotential in the 2 nd plateau (Figure 4-12). This trend is true for both AB and KB cathodes. Overpotential for the second discharge plateau (h L) vs. log discharging current density shows an exponential relationship (Figure 4-13) and the higher overpotential observed in the KB cathode than that in AB suggests poor kinetics of Li 2S formation step. According to Derek Moy’s studies, we can see that discharge plateaus appear at higher potentials as 2.1 V vs. Li + / Li at C/20 and C/10 and at least at 2 V vs. Li + / Li at C/5 rates in both AB and Vulcan XC- 72 cells. 8, 24 In our experiments, observation of electrode potentials was 1.7 V vs. Li + / Li at C/12 117 and 1.8 V vs. Li + / Li at C/11 for KB and AB electrodes respectively shows charge transport kinetics and/or mass transport in both cathode materials are hindered during Li 2S formation. Due to the poor contact distribution of the electrolyte with the cathode substrate, the low- density cathode structures like KB can show higher overpotential for the precipitation process of Li 2S than that in AB. Overpotential dependent nucleation rate and 3D growth usually lower the discharging capacity due to fast surface coverage by insulating Li 2S. With significantly higher surface area and electrode volume in KB, we can assume that the surface concentration of active sites for nucleation is higher in KB cathodes. Thus, at higher overpotentials, utilization of S can still be expected to be better in KB than AB. This assumption is verified by the observation of higher specific capacity in the second discharge plateau with KB electrodes compared to AB at a given overpotential (Figure 4-12). Also, the ratio of the initial region capacity to the second plateau capacity is higher with KB cathodes than that with AB cathodes. However, at a given overpotential, AB cathodes show higher discharging rates compared to that with KB cathodes (Figure 4-13). This might be due to the presence of lower number of pores in AB than that in KB that increases the accessibility of the electrolyte to wet the cathode structure. However, AB cathodes shows higher sulfur loading per area at a fixed ratio of the carbon to sulfur compared to that in KB cathodes. Figure 4-11 C and D show that AB cathode allows us to achieve a C/5 rate before reaching the 1.5 V vs. Li+/Li voltage cut-off while KB electrodes reach the 1.5 V vs. Li + /Li voltage cut-off even at C/8 rate even well before the appearance of the 2 nd discharge plateau. In AB cells, a sloping discharge curve can be observed from the middle of the second plateau toward the end of the discharge. This observation reflects a non-conducting Li 2S growth impeding the discharging or changes in the electrolyte properties. However, unlike in KB cathode structure, in AB electrolyte and polysulfide distribution occurs only within the porous structures created between the carbon particles. Thus, higher concentration of dissolved polysulfides within the pore volume can be expected in AB cathodes. Further, surface adsorption of soluble polysulfides can be lower in AB and the distinct dip feature discussed above in AB can be due to the noticeable electrolyte decrease of viscosity 118 during S 4 2- polysulfide consumption as well. With galvanostatic EIS analysis at different SOC and different rates, Zhang et al. simulated the discharging curves of Li-S at different rates and they proposed that rate limitation in Li-S system mainly comes from the transport limitation. 37 Table 4-4: Cathode properties of different carbon supports Figure 4-11: Discharging rate capability of Ketjen Black (A and C) and Acetylene Black (B and D) cathodes (15 µm). Cell Carbon type S Loading, mg Electrolyte volume, µL MCM type Impedance at low frequency, ohm R ohm, ohm 101 KB 3.7 100 75 µm, shorted ~570 or less ~45 189 AB 7.0 100 15 µm, shorted ~60 or less ~20 168 KB 3.83 140 ABA ~1200 ~30 190 AB 7.0 140 ABA ~95 ~20 Cathode thickness: 15 µm, Cell area: 2 cm 2 , S %: 63.3% 0 200 400 600 800 1000 1200 1400 1600 Specific Capacity, mAh/g 1.4 1.6 1.8 1.9 2 2.1 2.2 2.3 2.4 2.6 2.8 Voltage, V vs Li + /Li 5 Rate =C/ 11 Rate =C/ 15 Rate =C/ 23 Rate =C/ 29 Rate =C/ 39 Rate =C/ 1st 0.3 mA 0.3 mA 0.4 mA 0.5 mA 0.8 mA 1.1 mA 2.2 mA 0 200 400 600 800 1000 1200 1400 1600 Specific Capacity, mAh/g 1.4 1.6 1.8 1.9 2 2.1 2.2 2.3 2.4 2.6 2.8 Voltage, V vs Li + /Li 8 Rate =C/ 9 Rate =C/ 12 Rate =C/ 21 Rate =C/ 31 Rate =C/ 62 Rate =C/ 0.1 mA initial 0.1 mA 0.2 mA 0.3 mA 0.5 mA 0.7 mA 0.8 mA 10 20 30 40 Cycle number 50 60 70 80 90 100 %Coulombic Efficiency 200 400 600 800 1000 1200 1400 1600 Specific Capacity, mAh/g Cycling E101 with 75 um LCO, shorted, 3.7 mg S loading in KB 0.1mA 0.3 mA 0.5 mA 0.7 mA 10 20 30 40 Cycle number 50 60 70 80 90 100 %Coulombic Efficiency 200 400 600 800 1000 1200 1400 1600 Specific Capacity, mAh/g Cycling E189 with 15 um LCO (1000 lbs/cm 2 , KOH treated, shorted), 7 mg S loading in AB 0.3mA 0.4 mA 0.5 mA 0.8 mA 1.1 mA 2.2 mA (A) (B) (C) (D) 119 Figure 4-12: Specific capacity at second plateau vs. Overpotential in Ketjen Black (p) Acetylene Black cathodes (¨). Figure 4-13: Analysis of kinetics of S 4 2- to Li 2S conversion and utilization in Ketjen Black (p) Acetylene Black cathodes (¨). Further, in the present study we studied rate capability with layered MCMs achieving improved intercalation while preventing the shuttling of polysulfides (A and B). A Li-S cells with AB cathodes and layered MCM membranes show poor sulfur utilization (Figure 4-14B) compared to that with Li-S cells with AB cathodes and single layered MCMs in shorted cell configurations (discussed above in Figure 4-11D). KB with layered membranes shows higher utilization at lower rates. But rate capability is poor. 0.2 0.1 0.3 y = -359.83x + 276.35 R² = 0.9578 y = -716.98x + 651.89 R² = 0.9741 0 100 200 300 400 500 600 700 0 0.1 0.2 0.3 0.4 0.5 0.6 Specific capacity in the second plateau, mAh/g Overpotential, V AB KB y = 0.728x + 0.0259 R² = 0.9999 y = 0.2365x + 0.0454 R² = 0.9984 0 0.1 0.2 0.3 0.4 0.5 0.6 0.1 1 10 Overpotential, V Discharging Current, mA KB AB 120 Figure 4-14: Discharging rate capability with different Li-S cell construction with layered MCM and (A) KB cathodes and (B) AB cathodes. Figure 4-15: Analysis of kinetics of S 4 2- to Li 2S conversion and utilization in Ketjen Black (p) and Acetylene Black cathodes (¨) with layered MCM. After reaching 1.5 V vs. Li+/Li cut-off at a high discharge rate in AB cathodes, we studied further discharging capability at lower rates. The capability of discharging the remaining Li 2S 2 at a lower rate (0.1 mA) improved coulombic efficiency by 5% and 10% at initial discharge current of 0.3 and 0.4 mA, respectively (Appendix C4). This observation also suggested that even though capacity remains within the cathode structure, R ct or R m can be the barrier for achieving higher utilization of S at higher discharge rates. Capability of further discharging at a lower rate after 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0 Overpotential, V log(Discharging Current), mA AB KB 200 400 600 800 1000 1200 1400 1600 Specific Capacity, mAh/g 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 Voltage, V vs Li + /Li Discharging Rate capability of E168 with ABA LCO, 3.8 mg S loading in KB 15 um Rate =C/64 Rate =C/21 Rate =C/13 Rate =C/11 Rate =C/64 Rate =C/21 Rate =C/13 Rate =C/11 Rate =C/64 Rate =C/21 Rate =C/13 Rate =C/11 0.1 mA initial 0.3 mA 0.5 mA 0.6 mA 0 200 400 600 800 1000 1200 1400 1600 Specific Capacity, mAh/g 1.4 1.6 1.8 1.9 2 2.1 2.2 2.3 2.4 2.6 2.8 Voltage, V vs Li + /Li Discharging Rate capability of E190 with ABA LCO membrane, 7 mg S in 15 um AB 29 Rate =C/ 39 Rate =C/ 117 Rate =C/ 0.1 mA 0.3 mA 0.4 mA (A) (B) 121 discharging at higher rates verified that the utilization at higher rate is limited by the potential drops. 4.3.9 Rate Capability Studies with Mixed Carbon Cathodes Even though better utilization by capturing of polysulfide within cathode structure could be achieved with KB electrodes, the poor lithium ion and lower polysulfides transport rates within the electrolyte, achieving higher rates is still quite challenging with only KB-containing cathodes. On the other hand, although, only AB containing cathodes show lower impedance, fast nucleation minimizes the 3D growth of Li 2S leading to poor utilization. EIS analysis has verified that kinetics of Li-S system in AB cathodes can be improved with the addition of KB in the separator as a second current collector. 34 Other than the higher utilization, outstanding rate capabilities from 0.1 C to 6 C were reported by Xiaoming et al. in Li-S cells with AB cathodes and conductive separators coated with Ketjen Black nano-particles. 33, 34 Thus, anticipating that mixed carbon cathode structures can give better rate capability and utilization compared to the individual carbons, we fabricated cathodes using a 1:1 mixture of AB and KB and the next section discusses experimental results of these mixed carbon cathodes. Table 4-5: Cathode properties of mixed carbon supports with MCM In the first attempt, ours studies focused on 1:1: mixture of KB and AB for the mixed carbon cathode. However, in mixed carbon cathodes, both layered and single membrane MCM showed better utilization at lower rates (Figure 4-16A and B). Capacity values and discharging curve features compared well with the performance of only KB containing cathodes at lower rates. At lower rates of discharge in mixed carbon electrodes we can see a plateau similar to Cell S loading, mg Electrolyte volume, µL MCM type Impedance at low frequency, ohm R ohm, ohm 193A 6.3 140 ABA ~250 or less ~25 195A 6.3 100 B ~250 or less ~20 Carbon composition: 1:1 of KB: AB, Cell thickness: 15 µm, Cell area: 2 cm 2 , S %: 63.3% 122 that in KB electrodes. However, with increasing the rate, the discharge curve of the mixed carbon electrodes looks more like AB electrodes. As we predicted, with the mixed carbons, utilization is slightly better than AB-only electrodes. Also, rate capability is better than that with KB-only electrodes. Overpotential at the second discharge plateau vs. current density shows a logarithmic relationship (Figure 4-17) in both single and multi-layered MCM regardless of the cathode construction. The results in this plot suggests that better kinetics can be achieved with AB-containing cathodes. Figure 4-16: Discharging rate capability with different Li-S cell construction with mixed carbon cathodes with (A) layered MCM and (B) with single MCM 0 200 400 600 800 1000 1200 1400 1600 Specific Capacity, mAh/g 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 Voltage, V vs Li + /Li Discharging Rate capability of E193A with ABA LCO membrane, 6.3 mg S loading in KB:AB 106 Rate =C/ 35 Rate =C/ 26 Rate =C/ 21 Rate =C/ 10 Rate =C/ 5 Rate =C/ 0.1 mA 0.3 mA 0.4 mA 0.5 mA 1.1 mA 2.2 mA 0 200 400 600 800 1000 1200 1400 1600 Specific Capacity, mAh/g 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 Voltage, V vs Li + /Li Discharging Rate capability of E195A with 15 um LCO membrane, 6.3 mg S loading in KB:AB 106 Rate =C/ 35 Rate =C/ 26 Rate =C/ 21 Rate =C/ 10 Rate =C/ 5 Rate =C/ 106 Rate =C/ 35 Rate =C/ 26 Rate =C/ 21 Rate =C/ 10 Rate =C/ 5 Rate =C/ 0.1 mA 0.3 mA 0.4 mA 0.5 mA 1.1 mA 2.2 mA (A) (B) 123 Figure 4-17: Analysis of kinetics of S 4 2- to Li 2S conversion with MCM single layer (¢) and ABA layered (●). 4.3.10 Summary of Critical Features in Cell Designing To achieve the commercial deployment of lithium-sulfur cell technology, certain experimental design parameters should be met. Values of these parameters can vary with the applications. However, as a general application, Manthiram and co-workers reviewed literature thoroughly and established merit of Figures for critical parameters in Li-S system (Table 4-6). 2 Table 4-6: Cell Parameters 2 Parameter Quantity Sulfur content by weight 70 % Areal sulfur loading 6.5 mg/ cm 2 Coulombic efficiency 98 % Electrolyte/Sulfur ratio 5 uL/mg Capacity at C-rate of C/10 1000 mAh/ g Capacity at C-rate of C/5 800 mAh/ g Capacity retention after 200 cycles 81.25 % y = 0.1195ln(x) + 0.2959 R² = 0.9522 y = 0.1732ln(x) + 0.4482 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.1 1 10 Overpotential, V log(Discharging Current), mA 124 Even though our experimental data shows better utilization at C/60 rate (0.1 mA/cm 2 ), discharge capacity is still lower than 600 mAh/g at C/10 rates. Also, compared to the literature, we can notice kinetics and/or transport are significantly poorer as discharging plateaus appear at higher overpotentials. In addition to them, significantly higher impedance also suggests that there is a primary issue of cell construction and achieving comparable data to literature requires further performance improvements. 4.4 Chapter Four Conclusions In this study we verified Derek Moy’s claims of prevention of polysulfide shuttling by using MCM. By changing the porosity of MCM, 11 to 16% improved utilization of sulfur in initial discharging could be achieved in cells compared to the cells without MCM. Initial cycling data showed that about 20% improvement of coulombic efficiency and 1/3 rd improvement of sulfur utilization was achieved with the MCM membrane. Further, by using different layers and different porosity of MCM, we could validate intercalation process of lithium ion via MCM to facilitate charge transport within the Li-S cell. EIS studies of the Li-S cells showed that when MCM and the cathode are shorted the lowest impedance can be achieved suggesting that better rate capability can be achieved with “shorted” cathode configurations. Among the different carbon materials studied in the present study, AB cathodes showed the lowest impedance at low frequency while KB containing cells showed the best utilization due to their higher surface area. Thus, mixed carbons were identified as a suitable cathode material to receive benefits of both carbon materials. Layered MCM with porous and non-porous layers and mixed carbon cathodes showed better rate capability due to improved utilization, better kinetics of Li 2S formation and enhanced lithium ion intercalation across MCM. 125 4.5 Chapter Four References 1. Zhao, E.; Nie, K.; Yu, X.; Hu, Y. S.; Wang, F.; Xiao, J.; Li, H.; Huang, X., Advanced Characterization Techniques in Promoting Mechanism Understanding for Lithium–Sulfur Batteries. Advanced Functional Materials 2018, 1707543. 2. Chung, S. H.; Chang, C. H.; Manthiram, A., Progress on the Critical Parameters for Lithium–Sulfur Batteries to be Practically Viable. Advanced Functional Materials 2018, 1801188. 3. Ren, Y. X.; Zhao, T. S.; Liu, M.; Tan, P.; Zeng, Y. K., Modeling of lithium-sulfur batteries incorporating the effect of Li2S precipitation. Journal of Power Sources 2016, 336, 115-125. 4. Zheng, J.; Gu, M.; Wang, C.; Zuo, P.; Koech, P. K.; Zhang, J.-G.; Liu, J.; Xiao, J., Controlled Nucleation and Growth Process of Li2S2/Li2S in Lithium-Sulfur Batteries. Journal of The Electrochemical Society 2013, 160 (11), A1992-A1996. 5. Andrei, P.; Shen, C.; Zheng, J. P., Theoretical and experimental analysis of precipitation and solubility effects in lithium-sulfur batteries. Electrochimica Acta 2018, 284, 469-484. 6. Fang, R.; Zhao, S.; Sun, Z.; Wang, D.-W.; Cheng, H.-M.; Li, F., More Reliable Lithium- Sulfur Batteries: Status, Solutions and Prospects. Advanced Materials 2017, 29 (48), 1606823. 7. Moy, D.; Manivannan, A.; Narayanan, S., Direct measurement of polysulfide shuttle current: A window into understanding the performance of lithium-sulfur cells. Journal of the electrochemical society 2015, 162 (1), A1-A7. 8. Moy, D.; Narayanan, S., Mixed conduction membranes suppress the polysulfide shuttle in lithium-sulfur batteries. Journal of The Electrochemical Society 2017, 164 (4), A560-A566. 9. Jeong, T.-G.; Lee, Y.-S.; Cho, B. W.; Kim, Y.-T.; Jung, H.-G.; Chung, K. Y., Improved performance of dual-conducting polymer-coated sulfur composite with high sulfur utilization for lithium-sulfur batteries. Journal of Alloys and Compounds 2018, 742, 868-876. 10. Sun, Z.; Zhang, J.; Yin, L.; Hu, G.; Fang, R.; Cheng, H.-M.; Li, F., Conductive porous vanadium nitride/graphene composite as chemical anchor of polysulfides for lithium-sulfur batteries. Nature communications 2017, 8, 14627. 11. Su, Y.-S.; Manthiram, A., Lithium–sulphur batteries with a microporous carbon paper as a bifunctional interlayer. Nature communications 2012, 3, 1166. 126 12. Din, M. M. U.; Murugan, R., Garnet structured solid fast Li+ conductor as polysulfide shuttle inhibitor in Li-S battery. Electrochemistry Communications 2018, 93, 109-113. 13. Marmorstein, D.; Yu, T. H.; Striebel, K. A.; McLarnon, F. R.; Hou, J.; Cairns, E. J., Electrochemical performance of lithium/sulfur cells with three different polymer electrolytes. Journal of Power Sources 2000, 89 (2), 219-226. 14. Fan, F. Y.; Carter, W. C.; Chiang, Y.-M., Mechanism and Kinetics of Li2S Precipitation in Lithium–Sulfur Batteries. Advanced Materials 2015, 27 (35), 5203-5209. 15. Lau, S.; Archer, L. A., Nucleation and Growth of Lithium Peroxide in the Li–O2 Battery. Nano Letters 2015, 15 (9), 5995-6002. 16. Blomgren, G. E., The Development and Future of Lithium Ion Batteries. Journal of The Electrochemical Society 2017, 164 (1), A5019-A5025. 17. Yan, J.; Liu, X.; Li, B., Capacity fade analysis of sulfur cathodes in lithium–sulfur batteries. Advanced Science 2016, 3 (12), 1600101. 18. Deng, Z.; Zhang, Z.; Lai, Y.; Liu, J.; Li, J.; Liu, Y., Electrochemical impedance spectroscopy study of a lithium/sulfur battery: modeling and analysis of capacity fading. Journal of The Electrochemical Society 2013, 160 (4), A553-A558. 19. Sun, Y.; Seh, Z. W.; Li, W.; Yao, H.; Zheng, G.; Cui, Y., In-operando optical imaging of temporal and spatial distribution of polysulfides in lithium-sulfur batteries. Nano Energy 2015, 11, 579-586. 20. Kolosnitsyn, V.; Kuzmina, E.; Karaseva, E.; Mochalov, S., A study of the electrochemical processes in lithium–sulphur cells by impedance spectroscopy. Journal of Power Sources 2011, 196 (3), 1478-1482. 21. Yuan, L.; Qiu, X.; Chen, L.; Zhu, W., New insight into the discharge process of sulfur cathode by electrochemical impedance spectroscopy. Journal of Power Sources 2009, 189 (1), 127-132. 22. Chen, C.-F.; Mistry, A.; Mukherjee, P. P., Probing Impedance and Microstructure Evolution in Lithium–Sulfur Battery Electrodes. The Journal of Physical Chemistry C 2017, 121 (39), 21206-21216. 127 23. Cañas, N. A.; Hirose, K.; Pascucci, B.; Wagner, N.; Friedrich, K. A.; Hiesgen, R., Investigations of lithium–sulfur batteries using electrochemical impedance spectroscopy. Electrochimica Acta 2013, 97, 42-51. 24. Moy, D. Advancing High Energy Lithium-Sulfur Batteries – Understanding and Solving Key Degradation Issues Affecting Cycle Life. University of Sourthern California, 2017. 25. Xiong, S.; Xie, K.; Diao, Y.; Hong, X., Characterization of the solid electrolyte interphase on lithium anode for preventing the shuttle mechanism in lithium–sulfur batteries. Journal of Power Sources 2014, 246, 840-845. 26. Zheng, J.; Lv, D.; Gu, M.; Wang, C.; Zhang, J.-G.; Liu, J.; Xiao, J., How to obtain reproducible results for lithium sulfur batteries? Journal of The Electrochemical Society 2013, 160 (11), A2288-A2292. 27. Ding, N.; Chien, S. W.; Hor, T. A.; Lum, R.; Zong, Y.; Liu, Z., Influence of carbon pore size on the discharge capacity of Li–O 2 batteries. Journal of Materials Chemistry A 2014, 2 (31), 12433-12441. 28. Zhang, B.; Lai, C.; Zhou, Z.; Gao, X. P., Preparation and electrochemical properties of sulfur–acetylene black composites as cathode materials. Electrochimica Acta 2009, 54 (14), 3708-3713. 29. Lion Specialty Chemicals Co., L. A. r. r. KETJENBLACK Highly Electro-Conductive Carbon Black. https://www.lion-specialty-chem.co.jp/en/product/carbon/carbon01.htm. 30. Lv, D.; Zheng, J.; Li, Q.; Xie, X.; Ferrara, S.; Nie, Z.; Mehdi, L. B.; Browning, N. D.; Zhang, J. G.; Graff, G. L., High energy density lithium–sulfur batteries: Challenges of thick sulfur cathodes. Advanced Energy Materials 2015, 5 (16), 1402290. 31. Irshad, A.; Jayathilake, B. S.; Elizalde-Segovia, R.; Moy, D.; Plichta, E. J.; Hendrickson, M. A.; Narayanan, S., Lithium-Sulfur Battery with a Robust Sulfur Electrode and Mixed Conduction Membrane, 48th Power Sources Conference, Denver, CO, 2018; P456-459. 32. Jozwiuk, A.; Sommer, H.; Janek, J.; Brezesinski, T., Fair performance comparison of different carbon blacks in lithium–sulfur batteries with practical mass loadings – Simple design competes with complex cathode architecture. Journal of Power Sources 2015, 296, 454-461. 128 33. Tang, H.; Yao, S.; Shen, X.; Xi, X.; Xiao, K., Lithium–Sulfur Batteries with High Rate and Cycle Performance by using Multilayered Separators coated with Ketjen Black. Energy Technology 2017, 5 (4), 623-628. 34. Zhao, D.; Qian, X.; Jin, L.; Yang, X.; Wang, S.; Shen, X.; Yao, S.; Rao, D.; Zhou, Y.; Xi, X., Separator modified by Ketjen black for enhanced electrochemical performance of lithium– sulfur batteries. RSC Advances 2016, 6 (17), 13680-13685. 35. Ma, Y.; Zhang, H.; Wu, B.; Wang, M.; Li, X.; Zhang, H., Lithium Sulfur Primary Battery with Super High Energy Density: Based on the Cauliflower-like Structured C/S Cathode. Scientific Reports 2015, 5, 14949. 36. Song, J.; Yu, Z.; Xu, T.; Chen, S.; Sohn, H.; Regula, M.; Wang, D., Flexible freestanding sandwich-structured sulfur cathode with superior performance for lithium–sulfur batteries. Journal of Materials Chemistry A 2014, 2 (23), 8623-8627. 37. Zhang, T.; Marinescu, M.; Walus, S.; Kovacik, P.; Offer, G., What Limits the Rate Capability of Li-S Batteries during Discharge: Charge Transfer or Mass Transfer? Journal of The Electrochemical Society 2018, 165 (1), A6001-A6004. 38. Shen, C.; Xie, J.; Zhang, M.; Andrei, P.; Hendrickson, M.; Plichta, E. J.; Zheng, J. P., Understanding the role of lithium polysulfide solubility in limiting lithium-sulfur cell capacity. Electrochimica Acta 2017, 248, 90-97. 129 Direct Electrochemical Detection and Quantification of NAD + and NADH Using an Unmodified Carbon-Fiber Microelectrode 5.1 Background Nicotinamide adenine dinucleotides in reduced form (NADH) and oxidized form (NAD + ) are termed co-factors in numerous enzyme-catalyzed redox processes such as, photosynthesis, glycolysis, fatty acid synthesis, citric acid cycle, and amino acid metabolism. 1-3 In both prokaryotes and eukaryotes the reduced form NADH carries the electrons. 1-3 Measuring the concentration of the oxidized and reduced species can be used to determine the cell activity and cell fate across a broad range of diseases such as cancer, diabetes and neurodegeneration. 3 We expect that developing a simple, fast and direct detection technique for NAD + and NADH would be very useful. 2 Enzymatic activity is directly observable by following the conversion of the substrate into products. 4,5 Photometry, fluorescence spectroscopy, turbidity, luminometry and electrochemical analysis are some the common methods of enzyme kinetics. 4-6 For redox reactions catalyzed by enzymes where a co-factor such as NAD + /NADH is involved, the consumption and generation of the cofactors may be measured to deduce the rate of substrate transformation. 7, 8 However, many of these methods are time-consuming, labor-intensive and require large sample volumes. One widely-practiced approach is monitoring NADH by the light absorbance at 340 nm. 6 Colorimetric analysis offers high sensitivity in detecting lower concentrations. 9 However, the absorbance in the UV region can be influenced by many external factors such as the concentration of the sample, pH, temperature and the solvent. 9 The solvent can change the positions of the absorption peaks due to interactions with the sample molecules. Higher concentrations can also create interactions among sample molecules and position shifts and changes to the shape of the absorption peaks can be observed. Thus, the linear relationship between the concentration and the intensity of the absorption peak can be distorted with higher concentration. Indirect methods of determining NADH concentration by a redox reaction with 130 other molecules that generate intensely-colored products or by shuttling of mediator molecules are in use. 10 However, these approaches involve complex formation and require additional reagents. Thus, direct-electrochemical detection can be a promising method to measure regenerated or synthesized cofactors without consuming the sample. Further, in addition to detecting the cofactors of enzymes, analytical methods of detecting various enzymatic substrates such as formate and ethanol by monitoring the redox activity of biological cofactors have also been previously studied. 11 Bio-electrochemical detection is based mainly on the electrochemical oxidation of NADH cofactors at the electrode (Eq 5-1). They are particularly useful in the indirect detection of several non-electroactive substrates. 3, 11 Gang et al. have shown that unmodified screen-printed carbon electrodes can be used to indirectly, but accurately analyze the concentration of ethanol in plasma by detecting NAD cofactors. 12 NADH àNAD + + H + +2e - (5-1) Microelectrodes offer advantages over macroelectrodes in electroanalytical measurements: (1) Microelectrode analysis suitable for resistive media because the measured currents are small and therefore the ohmic voltage drop between the working electrode and the reference electrode can be minimized. 13 Thus, dilute solutions of low ionic conductivity can be used. This is the key element in the successful application of microelectrode in bio-electrochemistry as it gives the advantage of working in a diversity of biologically used buffer solutions. (2) In biological measurements, sample volumes in the range of picoliter to microliter are used. Under these conditions, electrochemical methods offer benefits over spectroscopic methods by eliminating the dependence of signal intensity on path length. Electrochemical methods rely on the detection of current arising from electron transfer at a surface and current in the range of picoamperes can be reliably measured. (3) Due to the reduced size of the microelectrode, a high spatial resolution can be achieved. (4) Further, compared to conventional electrochemical methods which are restricted to milliseconds or longer time scales, microelectrode methods can be used to study ultrafast electrochemical events which occur in the nanoseconds to microseconds time scale. 131 Due to these advantages of microelectrodes, they are applied in many areas including environmental, biomedical and material science. Amperometric detection of cellular cholesterol at a microelectrode by positioning the electrode adjacent to the biological cell and contacting the cell is a notable biomedical analytical application of microelectrode. 14 Screening of NAD- dependent enzymes using carbon paste electrodes, modified carbon electrodes with a redox mediator such as phenazine, and boron-doped diamond microelectrodes are promising for detection of NADH oxidation in the millisecond time scale. 15-17 In-vivo biochemistry, measurements of chemical events can benefit from a high degree of spatial and temporal resolution as well as a high degree of sensitivity and selectivity. An in-situ measurement of enzymatic activity using NAD + / NADH ratios can be obtained if both oxidation and reduction currents are detectable using a single electrode. Carbon-based electrode materials have been studied for NADH oxidation and low overpotentials are reported. 18 In this study, we have developed a simple method of using an unmodified, commercially- available carbon fiber microelectrode (CFME) to detect and quantify both NADH and NAD + (Figure 5-1). In-situ detection of regenerated or synthesized cofactors is used for quantifying NAD-dependent enzyme activity and substrate concentration. Figure 5-1: Microelectrode detection of NAD + / NADH to investigate Bio-catalytic redox reactions (R=adenine dinucleotide) 132 5.2 Methods 5.2.1 Material and Reagents NAD + , NADH, NAM and imidazole were purchased from Alfa-Aesar. Tris(hydroxymethyl)aminomethane hydrochloride (tris- HCl), 2-(N-morpholino) ethanesulfonate (MES), MES-NaOH, formate dehydrogenase, alcohol dehydrogenase, sodium formate were obtained from Sigma-Aldrich. Sodium dihydrogen phosphate, Disodium hydrogen phosphate, D 2O and 100% pure Ethanol were purchased from VWR. The anion-exchange membranes (A901) were manufactured by Tokuyama, Japan. 5.2.2 Apparatus All the electrochemical experiments were achieved with Par-four-electrode potentiostat. Carbon fiber microelectrodes, glassy carbon disk electrode, Ag/AgCl reference electrode and platinum wire were obtained from BASi. Graphite materials were purchased from graphite store (catalog numbers CM001006 and GT001672). HNMR was obtained from Varian 600 MHz instrument. 5.2.3 Preparation of Standard Solutions Different buffer solutions (0.1 M) were prepared using Sodium phosphate, Tris or MEM. pH of the electrolytes was adjusted using 1 M NaOH dissolved in the same buffer solution. NADH and NAD + calibration solutions were prepared by adding NAD + and NADH to phosphate buffer solutions. 5.2.4 Electrochemical Generation of NADH NAD + was reduced at -0.9 V vs. Ag/AgCl using a chronoamperometric technique. A two chamber-three-electrode system consisting a working electrode (platinum wire or graphite felt), a platinum wire counter electrode and a silver/silver chloride reference electrode were used to electrochemically synthesize NADH. Counter electrode chamber (0.1 M phosphate buffer) and 133 working electrode chamber (20 mM NAD + in 0.1 M phosphate buffer) were separated using AEM A901 membrane during electrolysis. 5.2.5 Electrochemical Detection Cyclic voltammetry experiments were performed at a range of scan rates from 25-250 mV s -1 (- 1.4 to 1 V vs. Ag/AgCl ). Experiments were carried out in a three-electrode micro- electrochemical cell with carbon working electrode, platinum wire counter electrode and a silver/silver chloride reference electrode. Prior to cyclic voltammetrric analysis, argon gas was bubbled to de-areate the solution. Enzyme kinetics were determined from CV studies of a solution containing an enzyme (FDH or ADH), the substrate (sodium formate or ethanol) and NAD + over 1 hour. Electrochemically generated NADH was analyzed using in-situ detection using microelectrode. 5.2.6 Validation of the Electrochemical Detection Electrochemically generated NADH and substrate concentrations were determined using 1 H NMR with 512 scans. Imidazole was used as an internal standard to quantify NADH. 5.3 Results and Discussion 5.3.1 Cyclic Voltammetry Detection of NAD + and NADH To investigate the electrochemical detection of NAD + and NADH, different carbon materials were used in this study. Figure 5-2 shows cyclic voltammograms (CV) of 5 mM NAD + and 5mM NADH on carbon fiber and glassy carbon microdisk (GCMD) electrodes. GCMD gives a conventional CV as shown in Figure 5-2B. Widely separated oxidation and reduction peaks of NAD + /NADH redox couple indicates that NAD + and NADH redox reactions are highly irreversible on carbon electrodes. When performing a 2 nd scan on GCMD, we observe a slight decrease of the peak height and the appearance of a smaller oxidation peak around 380 mV vs. Ag/AgCl. However, the CFME shows distinct oxidation and reduction features over multiple scans (Figure 5-2A) for NADH and NAD + respectively. Reproducible diffusion-limited 134 current observation on carbon fiber microelectrode shows that these features can be used to qualitatively and quantitatively analyze both NAD + and NADH in a sample. Figure 5-2: Cyclic voltammograms of equimolar (5 mM) NAD + and NADH in 0.5 M phosphate buffer on A: Glassy carbon micro disk electrode and B Carbon fiber microelectrode (at scan rate of 25 mV/ s) 5.3.2 Calibration of the Microelectrode and Quantification of NAD + and NADH Equation 5-2 shows the diffusion-controlled limiting current at a microelectrode: I d = γ n F D C* r o (5-2) -2.E-05 0.E+00 2.E-05 -1.5 -1 -0.5 0 0.5 1 Current, A Potential, V vs. Ag/AgCl 1st cycle 2nd cycle (B) -3E-09 -2E-09 -1E-09 0 1E-09 2E-09 3E-09 -1.5 -1 -0.5 0 0.5 1 Current, A Potential, V vs. Ag/AgCl (A) 135 where γ= 4 for disk electrodes, n= number of electrons transferred, F= Faraday constant, D= diffusion coefficient of the electroactive specie, C*= bulk concentration of the electroactive specie and r o = radius of the electrode (5.5 µm). Figure 5-3: Calibration curve of NAD + and NADH detection on microelectrode (0.25 to 20 mM) To calibrate the current response to concentration, we carried out the cyclic voltammetry of a series of concentrations of NAD + and NADH in sodium phosphate buffer (Figure 5-3) using the carbon fiber microelectrode. Further, we conducted voltammetry of NAD + using CFME in three different buffer solutions (Figure 5-4) and the diffusion coefficients of NAD species in each media are presented in Table 5-1. In buffer solutions, the diffusion coefficient of NADH is about 2 times higher than that of NAD + . However, in the MES buffer, the difference of the diffusion coefficients of NAD and NADH is lower compared to that in tris and phosphate buffers. -10 -8 -6 -4 -2 0 2 4 6 8 10 0 2 4 6 8 10 12 14 16 18 20 Current, nA NAD + or NADH Concentration, mM 136 Figure 5-4: Detection of A: NAD+ and B: NADH (5 mM) in MES, Phosphate buffer and Tris buffer at pH 7.00 Table 5-1: Diffusion coefficients of NAD species in three different buffer solutions Buffer solution NAD + , Diffusion coefficient x 10 -6 , cm 2 /s NADH, Diffusion coefficient x 10 -6 , cm 2 /s Sodium phosphate 0.9 1.8 MES 0.8 1.3 Tris-HCl 1.1 2.1 5.3.3 Detection Limit Enhancement and Detection of Low Volume Samples The lowest concentration used in the calibration plot using a single microelectrode in Figure 5- 4 is 250 µM of NAD + and NADH. However, the current response for NAD cofactors can be enhanced in the presence of a higher number of microelectrodes (Figure 5-5). The assembly of electrodes did not change the microelectrode behavior. Use of two microelectrodes shows 2.3±0.1 times higher current response while 3 microelectrodes together showed 4.3±0.1 times higher currents for both NAD + and NADH detection. Therefore, this study implies that the detection limit can be enhanced by combining a number of microelectrodes. -4.E-09 -2.E-09 0.E+00 -1.5 -1.4 -1.3 -1.2 -1.1 -1 Current, A Potential, V vs. Ag/AgCl phosphate MES Tris 0.E+00 3.E-09 6.E-09 0.5 0.6 0.7 0.8 0.9 1 Current, A Potential, V vs. Ag/AgCl phosphate MES Tris 137 Figure 5-5: Relative current response variation with increasing the number of electrodes Figure 5-6: Detection of 7.5 mM NAD + in 0.5 M Phosphate Buffer Reduction current of NAD + at -1.2 V vs. Ag/AgCl shows a limiting current of 2.8±0.1 nA in different sample volumes from 0.05-10.0 mL (Figure 5-6). Therefore, microelectrode-based analysis of NAD + and NADH can be a capable method for detecting NAD + concentrations in lower sample volumes. Microfluidic containing BDD chip electrode is previously reported for oxidation of NADH and it could be used to detect NADH in a range of concentrations from 10- 100 µM. 17 Since, the analysis of samples in low-volumes as low as nano-liter scale is reported, we can propose herein that a microfluidic device containing an array of carbon fiber electrodes can be a promising sensor for low-cost detection of NAD cofactor in small volumes. 1.0 2.0 3.0 4.0 -2 -1 0 1 2 Reductuon current, nA log volume, (volume in mL) Reduction current 2.8± 0.1 0.0 1.0 2.0 3.0 4.0 5.0 0 1 2 3 Relative current response per unit concentration Number of electrodes NAD+ NADH 138 5.3.4 Dependence of pH Figure 5-7 shows the reduction currents of NAD + (5 mM at 25 mV s -1 scan rate) at different pH values. Even though NAD redox chemistry involves a proton transfer or a hydride transfer, the reduction potential on carbon fiber microelectrode is not changed significantly by the electrolyte pH in the range 6.5-10.0. However, a decrease of the diffusion-limited current can be observed with increasing the pH of the electrolyte. Figure 5-7: Current response of NAD + (5 mM) reduction at different pH in the region of 6.5-10.0 in 0.5 M phosphate buffer solution (pH adjusted using NaOH) 5.3.5 Observing Enzymatic Reaction by the Measuring the Consumption and Generation of NAD Cofactors The microelectrode allows observing convection independent currents. Further, the steady- state rate of diffusion to a microelectrode is high; this permits to study of rapid electron transfer, and fast coupled chemical reactions using steady-state techniques. Therefore, other than to the assaying of NAD + / NADH, we could use this technique for analyzing enzymatic reactions. To verify this simple technique for studying NAD-dependent enzyme reactions we studied the bio- catalytic oxidations of formate to carbon dioxide and ethanol to acetaldehyde using FDH and ADH respectively in the presence of NAD + cofactor in a three-electrode electrochemical cell using a carbon fiber microelectrode. Figure 5-8 shows that the rate of enzymatic reaction increases with enzyme concentration. When enzymatic reactions are studied at higher cofactor 1.50 1.70 1.90 2.10 6 7 8 9 10 Reductuon current, nA pH 139 or substrate concentrations in bio-electrochemical applications, when we use absorbance-based measurements, we need to dilute the sample for two reasons. (1) The sample volume is not enough for filling the cuvette and (2) The sample concentration is higher than the linear range of concentration vs. absorbance. Dilution can introduce systematic errors. In the present technique, as a wide range of concentration can be measured, the errors associated with dilution can be remarkably minimized. Further, this method allows the in-situ analysis of bio- electrochemical systems and the analysis can be achieved without consuming the sample. Table 5-2 shows the variation of the NAD dependence enzymatic reaction rates when the cofactor concentrations were varied. Figure 5-8: FDH enzyme catalyzed formate oxidation (100 mM Sodium formate and 5 mM NAD + ) Table 5-2: Rate of different dehydrogenase enzymatic reactions Enzyme Type Enzyme substrate Initial NAD + concentration, mM Rate of substrate oxidation, mM per hour Formate Dehydrogenase (1µM) Formate (100 mM) 4.96 1.26 10.07 2.58 30.06 3.12 Alcohol Dehydrogenase (0.02 µM) Ethanol (20%) 4.00 5.04 8.00 10.14 0 1 2 3 4 5 6 0 10 20 30 40 50 60 Formate oxidation, mM Time, min 1 uM FDH 0.5 uM FDH 140 5.3.6 Detection of the Electrochemical Generation of NADH Figure 5-9 A: Microelectrode Detection of NAD + electrolyzed sample B: H-NMR peaks of NADH from NAD + in the electrolyzed sample on platinum electrode and C: Correlation of H-NMR analysis and microelectrode analysis. Electro-synthesis of enzyme active NADH is a two-electron reduction process which is associated with a protonation step. Preparing enzyme-active NADH is challenging due to the formation of an inactive dimer and isomers of NADH from the 1 st reduction process. However, a successful electrochemical pathway for enzyme active NADH regeneration using platinum electrodes at -0.7 to -1.0 V vs. Ag/AgCl was previously reported by Yun et al. 19 In an electrochemical generation process of NADH, the microelectrode can be used for in-situ detection of the cofactor. 0.E+00 2.E-09 4.E-09 6.E-09 8.E-09 0.4 0.6 0.8 1 Current, A Potential, V vs. Ag/AgCl 4 hour 3 hour A B Detection in 4 hour: 11.6 mM Detection in 3 hour: 8.3 mM 0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14 Microelectrode method, mM NMR assay, mM C 141 In this present study analysis of NADH using a carbon fiber microelectrode verified that NAD + reduction at the platinum electrode can produce NADH. Figure 5-9 shows the generation of NADH on platinum wire electrode. The appearance of the oxidation peak at 500 mV vs. Ag/AgCl on microelectrode can confirm that NADH can be electro-synthesized and this observation could be verified qualitatively and quantitatively using NMR analysis of NADH (Figure 5-9B). A quartet appearing at a chemical shift of 2.4-2.7 ppm in NMR spectra is characteristic of NADH. 20 Further, electroactive electrodes for NADH synthesis could be screened using microelectrode analysis of NADH. Analysis of NADH using NMR or microelectrode analysis shows that enzyme- active NADH could be generated using neither glassy carbon nor graphite electrodes. The microelectrode analysis can to detect the differences in the NADH produced at a carbon electrode versus a platinum electrode. However, NAD + electrolyzed sample on graphite electrode shows a cyclic voltammogram (Figure 5-10A) with a diffusion-limited oxidation current at 250 mV lower to the NADH oxidation peak. This analysis confirms that unmodified low-cost carbon electrodes cannot be used for the electrochemical generation of NADH. It has been previously reported that (NAD) 2 dimer can undergo electrochemical oxidation producing two NAD + molecules at lower oxidation potentials compared to NADH. 21 The inability of NADH generation on graphite electrode could be further confirmed using cyclic voltammograms of NAD + and NADH on graphite electrodes. In the initial cycle in NADH containing samples, potential scan from the open circuit potential to more positive values shows only a single oxidation peak which is characteristic to NADH oxidation (Figure 5-10B). When we scanned back to negative potentials, a reproducible reduction peak could be observed in NADH samples. This peak is due to NAD + reduction and the same peak appears in CVs of samples containing only NAD + . From the 2 nd cycle onwards the initial NADH peak height is reduced and another oxidation peak at 300 mV lower from the initial NADH oxidation peak could be detected. Samples containing only NAD + also show the second oxidation peak around 200 mV vs. Ag/AgCl. These observations suggest that NAD + reduction on graphite electrode yields another oxidizable product which can be attributed to the (NAD) 2 dimer. 142 Figure 5-10 A: Microelectrode Detection of NAD + electrolyzed sample (electrolysis using graphite felt) and B: Cyclic voltammogram of NAD + and NADH (5 mM) using graphite rod in 0.1 M phosphate buffer at pH 7. 5.3.7 Detection of NAD + Analogue Molecules Using Microelectrode Figure 5-11: Cyclic voltammogram of NAD + and NAM in phosphate buffer solution. Nicotinamide mononucleotide (NAM) has a similar redox active site as NAD + . Cyclic voltammogram of NAM (Figure 5-11) shows that both nucleotides behave similarly on reduction at carbon fiber microelectrode. The observation of a higher reduction current for NAM reflects that NAM has relatively higher diffusion coefficient in the electrolyte. The higher diffusion -6.E-09 -3.E-09 0.E+00 3.E-09 -1.5 -1 -0.5 0 0.5 1 Cureent, A Potential, V vs. Ag/AgCl NAD+ NAM 0.E+00 3.E-09 0 0.2 0.4 0.6 0.8 1 Current, A Potential, V vs. Ag/AgCl A -4.E-04 0.E+00 4.E-04 -2 -1 0 1 2 Current, A Potential, V vs. Ag/AgCl NADH NAD+ B (NAD) 2 oxidation NADH oxidation 143 coefficient of NAM compared to NAD + may be attributed to the lower molecular weight of NAM. 5.4 Chapter Five Conclusions In this study, we demonstrated the use of carbon fiber microelectrodes for quantitative and qualitative assessment of NAD + and NADH cofactors. Enzymatic reactions of alcohol dehydrogenase and formate dehydrogenase with NAD cofactors were evaluated using the oxidation and reduction currents on a carbon fiber microelectrode. The technique is fast, does not degrade the sample and can be used in situ. We have tested this principle at various pH values as well. These advantages suggest that the electrochemical method using the micro- electrode can be used as NAD cofactor sensor to replace the colorimetric technique where higher concentrations of NAD cofactors are used. Other than the detection and quantification of cofactors, this technique permits the investigation of the effect of different substrate concentrations used in NAD-dependent enzymatic reactions. Sensitivity could be increased by combining a number of microelectrodes together. This method can be further developed for studying the enzyme kinetics at different temperatures, pH and media, and also for identifying the enzyme inhibitors. 144 5.5 Chapter Five References 1. Musameh, M.; Wang, J.; Merkoci, A.; Lin, Y., Low-potential stable NADH detection at carbon-nanotube-modified glassy carbon electrodes. Electrochemistry Communications 2002, 4 (10), 743-746. 2. Li, J.; Sun, Q.; Mao, Y.; Bai, Z.; Ning, X.; Zheng, J., Sensitive and low-potential detection of NADH based on boronic acid functionalized multi-walled carbon nanotubes coupling with an electrocatalysis. Journal of Electroanalytical Chemistry 2017, 794, 1-7. 3. Lobo, M. J.; Miranda, A. J.; Tuñón, P., Amperometric biosensors based on NAD (P)- dependent dehydrogenase enzymes. Electroanalysis 1997, 9 (3), 191-202. 4. Bisswanger, H., Enzyme assays. Perspectives in Science 2014, 1 (1), 41-55. 5. Reymond, J.-L.; Fluxa, V. S.; Maillard, N., Enzyme assays. Chemical Communications 2008, (1), 46. 6. Vázquez, M. J.; Ashman, S.; Ramón, F.; Calvo, D.; Bardera, A.; Martín, J. J.; Rüdiger, M.; Tew, D.; Domínguez, J. M., Utilization of Substrate-Induced Quenching for Screening Targets Promoting NADH and NADPH Consumption. Journal of Biomolecular Screening 2006, 11 (1), 75-81. 7. Yoshida, N.; Uchida, E.; Katsuragi, T.; Tani, Y., A Novel NAD-Dependent Dehydrogenase, Highly Specific for 1,5-Anhydro-d-Glucitol, from Trichoderma longibrachiatum Strain 11-3. Applied and Environmental Microbiology 2003, 69 (5), 2603-2607. 8. Yoo, Y. J.; Feng, Y.; Kim, Y. H.; Yagonia, C. F. J., Enzymes for Environment. In Fundamentals of Enzyme Engineering, Springer Netherlands: Dordrecht, 2017; pp 189-203. 9. Ito, H.; Terai, T.; Hanaoka, K.; Ueno, T.; Komatsu, T.; Nagano, T.; Urano, Y., Detection of NAD(P)H-dependent enzyme activity with dynamic luminescence quenching of terbium complexes. Chemical Communications 2015, 51 (39), 8319-8322. 10. Korzeniewski, C.; Callewaert, D. M., An enzyme-release assay for natural cytotoxicity. Journal of Immunological Methods 1983, 64 (3), 313-320. 145 11. Hung Tzang, C.; Yuan, R.; Yang, M., Voltammetric biosensors for the determination of formate and glucose-6-phosphate based on the measurement of dehydrogenase-generated NADH and NADPH. Biosensors and Bioelectronics 2001, 16 (3), 211-219. 12. Tian, G.; Zhang, X.-Q.; Zhu, M.-S.; Zhang, Z.; Shi, Z.-H.; Ding, M., Quantification of ethanol in plasma by electrochemical detection with an unmodified screen printed carbon electrode. Scientific reports 2016, 6, 23569. 13. Forster, R. J.; Keyes, T. E., Ultramicroelectrodes. In Handbook of Electrochemistry, Elsevier: 2007; pp 155-171. 14. Devadoss, A.; Burgess, J. D., Steady-State Detection of Cholesterol Contained in the Plasma Membrane of a Single Cell Using Lipid Bilayer-Modified Microelectrodes Incorporating Cholesterol Oxidase. Journal of the American Chemical Society 2004, 126 (33), 10214-10215. 15. Abdellaoui, S.; Bekhouche, M.; Noiriel, A.; Henkens, R.; Bonaventura, C.; Blum, L. J.; Doumèche, B., Rapid electrochemical screening of NAD-dependent dehydrogenases in a 96- well format. Chemical Communications 2013, 49 (51), 5781-5783. 16. Abdellaoui, S. n.; Noiriel, A.; Henkens, R.; Bonaventura, C.; Blum, L. J.; Doumèche, B., A 96-well electrochemical method for the screening of enzymatic activities. Analytical chemistry 2013, 85 (7), 3690-3697. 17. Oyobiki, R.; Kato, T.; Katayama, M.; Sugitani, A.; Watanabe, T.; Einaga, Y.; Matsumoto, Y.; Horisawa, K.; Doi, N., Toward High-Throughput Screening of NAD(P)- Dependent Oxidoreductases Using Boron-Doped Diamond Microelectrodes and Microfluidic Devices. Analytical Chemistry 2014, 86 (19), 9570-9575. 18. Blandón-Naranjo, L.; Della Pelle, F.; Vázquez, M. V.; Gallego, J.; Santamaría, A.; Alzate- Tobón, M.; Compagnone, D., Electrochemical Behaviour of Microwave-Assisted Oxidized MWCNTs Based Disposable Electrodes: Proposal of a NADH Electrochemical Sensor. Electroanalysis. 19. Yun, S.-E.; Taya, M.; Tone, S., Direct reduction of NAD+ by electrochemical procedure and application of the regenerated NADH to enzyme reaction. Biotechnology Letters 1994, 16 (10), 1053-1058. 146 20. Mostad, S. B.; Glasfeld, A., Using high field NMR to determine dehydrogenase stereospecificity with respect to NADH: An undergraduate biochemistry lab. J. Chem. Educ 1993, 70 (6), 504. 21. Carelli, V.; Liberatore, F.; Casini, A.; Mondelli, R.; Arnone, A.; Carelli, I.; Rotilio, G.; Mavelli, I., Dimers of nicotinamide adenine dinucleotide: New evidence for the structure and the involvement in an enzymatic redox process. Bioorganic Chemistry 1980, 9 (3), 342-351. 147 Efficient and Selective Electrochemically- Driven Enzyme-Catalyzed Reduction of Carbon Dioxide to Formate using Formate Dehydrogenase and an Artificial Cofactor 6.1 Background This Chapter is an extended discussion from the conference presentation at the Electrochemical Society Meeting, National Harbor, MD in October 2017 1 and a part of the discussion from our manuscript published in Accounts of Chemical Research in February 2019. 2 Also, this Chapter contains the collaborative work of computational studies with Prof. Vaidehi Nagarajan and Dr. Supriyo Bhattacharya from the Department of Computational and Quantitative Medicine, Beckman Research Institute of the City of Hope, 1500 E. Duarte Road, Duarte CA 91010. As I have mentioned in Chapter 1, generally, enzyme-catalyzed processes are known for their selectivity. 3 Thus, in this study, the question we seek to answer is can we exploit the high level of selectivity of the processes in enzymes to achieve high energy efficiency for formate production. To this end, we recognize ways of using formate dehydrogenase to produce formate from carbon dioxide. 4-9 The enzyme formate dehydrogenase (FDH) is involved in metabolic processes in micro-organisms and plants, performing the oxidation of formate to carbon dioxide. 9 However, our goal was to force metal-independent FDH to produce formate from carbon dioxide, the reverse of the metabolic reaction. Chapter 1 section 1.2.7, discusses that the metal-independent FDH relies on cofactors such as nicotinamide adenine dinucleotide (phosphate) (NAD(P) + / NAD(P)H). However, the electrochemical generation of the reduced form of NAD using low-cost electrodes is challenging (Chapter 5 section 5.3.6) and the reverse reaction of FDH with reduced form of NAD is significantly slower than the forward reaction. Therefore, in the current study, our focus is on replacing NAD cofactor using methyl viologen as an artificial cofactor for the metal-independent type of FDH to achieve efficient CO 2 reduction. 148 6.1.1 Replacing NADH with an Artificial Redox Cofactor, Methyl Viologen We overcame the above-mentioned challenges with NADH by using an artificial redox cofactor, methyl viologen (MV •+ / MV 2+ ), as per Eq. 6-1 (Figure 6-1). CO2 + 2MV •+ +H + ⇆ 2MV 2+ + HCOO - (6-1) Figure 6-1: Enzyme-catalyzed CO 2 reduction using metal independent FDH and MV cofactor There are isolated reports of reversing formate oxidation with the methyl viologen radical cation (MV •+ ) as the reducing cofactor instead of NADH. 10-13 In some cases, MV •+ was used as a mediator (or reducing agent) to generate NADH. 5, 7 Specifically, we noted the following advantages of using MV to carry out the reaction in Eq. 6-1: (1) MV •+ provides a favorable thermodynamic driving force for the reduction of carbon dioxide to formate, unlike NADH (Appendix D1). The standard electrode potential of the MV 2+ / MV •+ couple is 0.120 V more negative to the NAD + /NADH couple. 14 Thus, the equilibrium constant for Eq. 2 is about 10,000 times larger than with NADH (Appendix D1). (2) Our studies have proved that the oxidation of formate to carbon dioxide does not occur readily with MV 2+ . Thus, the overall yield of formate would be preserved without the loss from re-oxidation. 13 FDH catalyzed 149 (3) MV •+ can be generated efficiently by the electrochemical reduction of MV 2+ at a carbon electrode as per Eq. 6-2 without the need for any external reducing agent, a distinct advantage for continuous production. The direct electron-transfer process is fast without significant energy losses. Previous reports of using dithionite for the reduction of MV 2+ to MV •+ add other byproducts to the reaction mixture that are not desirable for the subsequent enzymatic process; the direct electrochemical production of MV •+ avoids such contamination entirely. 11-13 MV 2+ + e - ® MV •+ E o = -0.44 V vs. NHE (6-2) Cognizant of these enabling advantages of MV, we set out to address following central questions: I. Can we use MV effectively to reverse the normal function of FDH and generate formate from carbon dioxide? II. Can we gain mechanistic insight using molecular dynamics (MD) simulations into the effectiveness of MV as a redox cofactor for FDH in the production of formate from carbon dioxide, and can we specifically predict the relative binding energies of the cofactors with FDH? III. Can the electrochemical regeneration of the artificial redox cofactor be coupled with the enzymatic conversion by FDH to achieve a continuous and efficient process for the production of formate from carbon dioxide? To answer these questions, we have studied the FDH-catalyzed reduction of carbon dioxide (Eq. 6-3) in various electrochemical reactor configurations. Compared to the comprehensive work done by Amao et al. 10-13 for developing photo-catalytic pathways with sacrificial agents, the present study reports electrochemical approaches with water as the electron donor (Eq. 6-3) in carbon dioxide reduction using FDH and MV. Typically, these reactors had four functional sections: (i) a chamber for the reduction of carbon dioxide to formate by FDH and MV •+ (Eq. 6- 1), (ii) an electrode where the reduced form of redox cofactor, MV •+ was regenerated electrochemically (Eq. 6-2), (iii) an electrode where the conjugate electrochemical oxidation of water to oxygen occurred (Eq.6-4) and (iv) a region where the formate was separated from the FDH catalyzed 150 reaction mixture. These electrochemical compartments were separated by specifically selected cation-exchange membranes (PEM) or anion-exchange membranes (AEM) to allow for the transport of protons or hydroxide ions and for the exclusion of anions or cations as required. CO 2 + H 2O ® HCOO - +H + + ½ O 2 (6-3) H 2O ® ½ O 2 + 2H + +2e - E o = 1.23 V vs. NHE (6-4) 6.2 Materials and Methods 6.2.1 Materials All chemicals were used were of the analytical grade and used as received without further purification. Methyl viologen dichloride hydrate (98%), FDH, tris(hydroxymethyl)aminomethane hydrochloride (tris- HCl), 2-(N-morpholino) ethanesulfonate (MES), MES-NaOH were purchased from Sigma-Aldrich. NAD + , NADH, imidazole, sodium hydrogen phosphate and sodium dihydrogen phosphate were obtained from Alfa-Aesar, Sodium hydroxide was obtained from VWR and the anion-exchange membranes (A901) were manufactured by Tokuyama, Japan. Films of Nafion® 117 and non-teflonized Toray ® papers were obtained from the Fuel Cell Store. All the membranes were immersed in sodium phosphate buffer (0.5 M at pH 6.6) for 12-14 hours prior to use. Glassy carbon rod electrodes (diameter 0.2 cm, length 5 cm) were acquired from Tokai Carbon Co. Ltd, Japan. Impervious graphite (catalog number GT001672) was obtained from the Graphitestore. Glassy carbon disk working electrodes (3 mm diameter) were obtained from BASi. 6.2.2 Cyclic Voltammetry of Cofactor Solutions Cyclic voltammetric measurements were made using 5 mM cofactor in “tris” buffer (0.1 M), phosphate buffer (0.1M) and MES buffer (0.1 M) at 25-200 mV S -1 scan rates using a three- electrode system consisting of a glassy carbon disk working electrode (BASi 3 mm), platinum wire counter electrode and Ag/AgCl reference electrode (Pine Instruments). 151 6.2.3 Bulk Electrolysis of Cofactor Solutions to Prepare MV •+ Electrochemical reduction of MV 2+ and NAD + was carried out using potential-controlled chronoamperometry in a two-chamber and three-electrode system using various types of carbon working electrodes (non-teflonized Toray ® paper, graphite rod electrode, glassy-carbon rod electrode), a platinum wire counter electrode and a silver-silver chloride reference electrode. The counter electrode chamber was separated from the working electrode chamber by ion exchange membranes. In all cases, MV 2+ reduction was conducted at -0.44 V vs. NHE while NAD + was reduced at -0.8 to -1 V vs. NHE. All the electrochemical reactions were carried out inside a sealed plastic enclosure to minimize extraneous oxygen diffusion to the system. Argon gas was bubbled through the MV 2+ solution for 10 minutes prior to beginning all the electrochemical studies. 6.2.4 Carbon Dioxide Reduction Using FDH and the Reduced Form of Methyl Viologen FDH (1-3 uM), methyl viologen (1- 40 mM) and sodium bicarbonate (10-100 mM) were added, and carbon dioxide gas at atmospheric pressure was bubbled into the reaction mixture. The reduced form of methyl viologen was supplied by electrochemical generation. Please note that in all our experiments, we used sodium phosphate buffer (0.1 M) at the pH value of 6.6 in which FDH is optimally active and carbon dioxide is almost entirely in the form of the bicarbonate anion. Therefore, in all the CO 2 reduction experiments we have used bicarbonate (0.1 M) to achieve a high concentration of carbon dioxide (Eq.6-5), although we nominally refer to the process as the electrochemical reduction of carbon dioxide. H 3O + + HCO 3 - D CO 2 + 2H 2O (6-5) 6.2.5 Formate Oxidation to Carbon Dioxide Using FDH and MV 2+ FDH (4.3 µM) and sodium formate (10- 45 mM) along with methyl viologen (20 mM) were added into the phosphate buffer (0.5 M, pH=7, 2 mL), and argon gas was bubbled during the reaction. 152 6.2.6 NMR Analysis 1 H NMR spectra were recorded on a Varian 600 MHz NMR instrument and acquired at 512 scans using D 2O as a solvent. The addition of a known amount of imidazole as an internal standard to the NMR sample allowed us to quantitate the concentration of formate. 6.2.7 Computational Studies Ligand docking and MD simulation were studied using the techniques described in Appendix D2. MV •+ was docked to the FDH active site (the docking box in Figure D2 in the Appendix D2) using the crystal structure of the FDH bound to NAD + and azide ion (PDB ID: 5DN9). 15 The difference in binding free energy of the products and reactants in FDH, with MV •+ and NADH as cofactors were calculated using the Molecular Mechanics Generalized Born Surface Area method (MMGBSA). 6.3 Results 6.3.1 Electrochemical Reversibility of NAD + / NADH and MV 2+ / MV •+ Electrochemical regeneration is a necessary step for the continuous production of formate. The cyclic voltammetry measurements of NAD + / NADH at a glassy carbon electrode in phosphate buffer (Chapter 5 and Appendix D3), showed that the reversibility of the NAD + /NADH couple on a glassy carbon electrode is poor. Firstly, NAD + is not reduced to NADH at carbon electrodes until -1.0 V vs. NHE, although the formal potential in pH 7 is about -0.320 V. 16 Our experimental studies of bulk electrolysis of NAD + containing solutions using various types of carbon electrodes (Toray paper, glassy carbon and graphite electrodes), produced an intense yellow color indicating the generation of a (NAD) 2 dimer (Chapter 5). 17 Due to the generation of these various products, NADH generation in pure enzyme active form at carbon electrodes presents a challenge (Appendix D3). 18 Thus, we determined that the electrochemical re- generation of NADH would be impractical for conducting the reverse reaction of carbon dioxide to formate using FDH. 153 Scheme 6-1: Different oxidation states of methyl viologen Figure 6-2: Cyclic Voltammograms of MV in A: Phosphate buffer (⎯), B: MES buffer (--) and C: Tris buffer (…) at pH 6.6 Table 6-1: Electrochemical properties of MV in different buffer solutions In MES In phosphate In Tris E p,c mV vs. NHE -456 -457 -440 E p,a mV vs. NHE -373 -403 -402 ½(E p,a+E p,c) , mV vs. NHE -415 -430 -421 ∆E p, mV 83 52 38 Diffusion Coefficient of MV 2+ (x10 6 ), cm 2 s -1 Microelectrode 5.07 4.71 3.30 Micro-disk electrode 3.01 2.81 2.80 pKa 6.10 7.21 8.07 -4.E-05 -2.E-05 0.E+00 2.E-05 4.E-05 -0.6 -0.5 -0.4 -0.3 -0.2 Current, A Potential, V vs. NHE 154 Unlike NAD + / NADH, methyl viologen was found to be electrochemically reversible. MV has well known electrochemistry spanning three different oxidation states. The MV di-cation (MV 2+ ), which is the most stable form, can be reduced in two one-electron steps as shown in Scheme 1. The cyclic voltammogram of redox reactions between MV di-cation and MV free radical cation in three different buffers at pH 6.6 (Figure 6-2, Table 6-1) demonstrate the reversibility of MV increased from MES to phosphate to the Tris buffer. To have a higher buffer capacity and pH closer to that desirable for the enzymatic reaction, we chose the phosphate buffer, although the highest electrochemical reversibility of MV was observed in the Tris buffer. The electrochemical conversion of MV di-cation into reduced form and the detection of different oxidation states of MV during bulk electrolysis is shown as Appendix D4. 6.3.2 Carbon Dioxide Reduction Using FDH and the Reduced Form of Methyl Viologen In the configuration shown in Figure 6-3, the reduced form of methyl viologen, MV •+ was electro- generated at a porous carbon fiber paper electrode (Toray ® Inc.) held at -0.44 V vs. the normal hydrogen electrode (NHE) while oxygen evolution occurred on a platinum electrode in chamber A. For every mole of methyl viologen that was reduced we can expect 0.25 moles of oxygen to be evolved. Electrochemically-produced, intensely-blue-colored MV •+ was allowed to react with FDH and carbon dioxide to produce formate in chamber B (Figure 6-3B). The rate of formate production as followed by 1 H NMR, increased with the concentration of MV •+ , and reaching a maximum value at a concentration of 20 mM. Correspondingly, the steady-state current required to support the electro-generation of MV •+ also increased and reached a plateau (Figure 6-3C, Appendix D5). The rate of formate production at various concentrations of MV •+ followed the Michaelis-Menten type kinetics. These results demonstrated the technical feasibility of producing formate from carbon dioxide using FDH and electrochemically-generated MV •+ . Further, we proved that the MV •+ was a viable cofactor to reverse the natural function of FDH and produce formate from carbon dioxide. 155 Figure 6-3: Electrochemical formate generation: (A) Reactor configuration, (B) Experimental set- up and (C) Rate of formate generation for various concentrations of MV •+ , and the continuous current density for producing MV •+ at an applied potential of -0.44 V vs. NHE using 3-40 mM MV 2+ , 1.2 µM FDH, 15 mL buffer solution in each chamber and Tokuyama A901 AEM. 6.3.3 Methyl Viologen is a Unidirectional Cofactor for FDH 6.3.3.1 Experimental Evidences While there are some reports in the literature to suggest that the oxidation of formate to carbon dioxide does not occur with FDH and the MV 2+ , 13 it was important for us to verify this property of MV 2+ . We combined FDH with the MV 2+ and formate in various concentrations and analyzed for the decrease in concentration of formate. No change in the concentration formate was detectable even after 30 hours (Figure 6-4). UV-visible spectroscopic studies also showed no H 2 O O 2 MV 2+ MV .+ CO 2 HCOO - Platinum electrode Toray Paper Electrode Enzyme FDH Enzyme-catalyzed CO 2 Reduction Electrochemical generation of Artificial co- factor AEM Chamber A Chamber B Ag/AgCl reference electrode Toray Paper Electrode Platinum electrode Chamber B Chamber A MV and Enzyme containing buffer solution Buffer solution AEM + - Power source (A) (B) OH - 156 decrease in the concentration of the MV 2+ nor the appearance of MV •+ (Appendix D6). Thus, we proved that MV 2+ is not a viable cofactor for the oxidation of formate to carbon dioxide, unlike NAD + that oxidizes formate to carbon dioxide quite efficiently. Therefore, the yield of formate from the reduction of carbon dioxide would be preserved without loss by re-oxidation by using MV as the artificial cofactor. Figure 6-4: Effect of 20 mM MV 2+ on the concentration of sodium formate (11 to 44 mM) in the presence of 4.3 µM FDH. The unique property of MV as a redox cofactor that supports only carbon dioxide reduction is crucial for achieving the high efficiency of formate production. This behavior is consistent with the favorable thermodynamic driving force for carbon dioxide reduction over formate oxidation using MV, as determined from the equilibrium constant value. 19 14, 16, 20 Although the equilibrium constant favored the reduction of CO 2 to formate with MV •+ , we expect the binding properties of MV to FDH to also play a key role in facilitating the catalytic reaction. We also want to rationalize the differences between MV •+ and NADH as redox cofactors, with MV •+ being just an electron donor while NADH being a hydride donor. Further, unlike NADH, two molecules of MV •+ are needed to reduce CO 2 to formate, and it is not evident where the second molecule of MV •+ would bind in FDH. Thus, we found it is important to understand the molecular interactions of the two cofactors with FDH to explain the differences 157 in their effectiveness. Therefore, we have collaborated with a team from the Beckman Research Institute of the City of Hope, Duarte, CA to conduct ligand docking and MD simulations to study the binding of MV 2+ /MV •+ and the NAD + /NADH redox couples to FDH in the presence of bicarbonate and formate. 6.3.3.2 Binding Free Energy Recapitulates the Thermodynamic Feasibility of the Bicarbonate Conversion by Methyl Viologen and FDH Our goal in this section is to provide mechanistic insights into the binding energetics of the reactants and products shown in Eq. 6-6 (the net reaction of Eqs. 6-1&6-5 at neutral pH) below and study the effect of the protein dynamics on the stability and flexibility of the reactants and products. 2MV •+ + H 3O + + HCO 3 - ® 2MV 2+ + OH - + HCOO - + H 2O (6-6) In MD simulation as described in the Appendix D2, starting from the crystal structure of the FDH bound to NAD + and azide ion (PDB ID: 5DN9), MV •+ was docked to the FDH active site (the docking box in Figure D6 in the Appendix D2). We replaced the azide ion with the HCO 3 - (Figure 6-5A) and docked H 3O + to the active site with MV •+ and HCO 3 - present. The best pose for H 3O + was located between MV •+ and HCO 3 - (Figure 6-5A). R258, the highly-conserved residue in the FDH family, and H311 which was implicated by mutagenesis to be involved in the catalytic activity of FDH, are both present clamping the HCO 3 - in our structural model. 37 MV •+ occupies and makes similar residue contacts as the nicotinamide moiety of NADH. These consistencies in the docked structures indirectly validate our structural models. We used the same docked positions and converted the reactants to products, MV 2+ and HCOO - , in the place of MV •+ and HCO 3 - , respectively. The OH - was docked separately in the presence of the rest of the products. Subsequently, we performed atomistic MD simulations on FDH bound to MV and NADH as cofactors, in presence of the reactants and products. During MD simulations, the HCO 3 - mostly stayed sandwiched between H311 and R258, as can be seen from the distance distribution (Figure 6-5B). However, in the MD simulations of NADH-bound FDH, the bicarbonate ion is distant from both H311 and R258 (distances 10.5 Å and 8.5 Å) and oriented away from the plane of the nicotine ring (Figure 6-5C). The HCO 3 - is also highly dynamic in the presence of NADH FDH catalyzed 158 compared to MV •+ (Figure 6-5D), suggesting its instability in NADH-bound FDH. Thus, the forward reaction of bicarbonate to formate is disfavored by NADH compared to MV •+ . Figure 6-5: Orientation of bicarbonate ion in the representative conformation from the highest populated cluster in the MD ensemble in the presence of (A) hydronium ions and MV •+ (green), and (C) NADH (green); major active site residues shown in magenta; (B, D) probability density of bicarbonate position as function of distance with R258 and H311 as sampled during the MD simulations of (B) MV •+ bound FDH and (D) NADH bound FDH; for each residue, the minimum distance of the oxygen atoms in bicarbonate with the sidechain polar nitrogen atoms was measured; (E) Difference in binding free energy between the reactants and products of bicarbonate reduction in FDH in the presence of MV •+ and NADH. Using the Molecular Mechanics Generalized Born Surface Area method (MMGBSA), we calculated the difference in binding free energy of the products and reactants in FDH, with MV •+ and NADH as cofactors. As shown in Figure 6-5E, with MV •+ as cofactor, the binding free energy favors the binding of reactants by 25 kcal/mol compared to the products. The weaker binding free energy of the products in presence of MV 2+ will facilitate the forward reaction of 2 2.5 3 Bicarbonate - R258 distance, Å 2 4 6 8 10 Bicarbonate - H311 distance, Å 0 0.5 1 1.5 2 2.5 3 3.5 4 7 8 9 10 11 Bicarbonate - R258 distance, Å 8 9 10 11 12 13 Bicarbonate - H311 distance, Å 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 2.8Å 3.0Å 3.1Å 3.3Å R258 H311 MV ⦁+ HCO 3 - H 3 O + (A) (C) (B) (D) (E) -15 -10 -5 0 5 10 15 20 25 30 Binding free energy difference (products – reactants), kcal/mol MV ⦁ + , HCO 3 - , H 3 O + à MV 2+ , HCOO - , OH - , H 2 O NADH, HCO 3 - à NAD + , HCOO - NADH HCO 3 - 159 bicarbonate reduction by efficient removal of the reaction products from the active site of FDH. In contrast, the products in the NADH-based reaction (NAD + and formate) show better binding energy than the reactants, indicating high stability of the reaction products, which hinders the forward reaction. These results validated the thermodynamic nature of the HCO 3 - reduction with MV •+ . We have used these simulation results to provide further insights into the mechanism of action of MV in the FDH catalysis. 6.3.3.3 Relative Stability and Flexibility of the Reactants and Products in FDH We clustered the conformations obtained from the MD simulations of the reactants and products bound to FDH independently by their root mean square deviation (RMSD) in coordinates (Appendix D2). We used the most populated conformation cluster for all further analysis presented in this section. We observed that the distances between the HCO 3 - , H 3O + and MV •+ are optimal for facilitating the proton-coupled electron transfer to the HCO 3 - ion. During the MD simulations of FDH bound to MV •+ , HCO 3 - and H 3O + , the bicarbonate ion is held in place by a hydrogen bond and salt bridge to H311 and R258, respectively. The hydronium ion is stabilized between the bicarbonate and the pyridine ring of MV •+ and is 4Å away from the nitrogen of the pyridine. This facilitates the conversion of bicarbonate to formate by stabilizing the reactants (Figure 6-6A). In contrast, in the simulations of MV-bound FDH (Figure 6-6B), the products hydroxyl and formate ions leave the binding site after 6 ns and 50 ns respectively of the MD simulations. These simulations indicate that for MV, the reactants binding is favored over that of the products. 160 Figure 6-6: Comparison of the reactants and products orientation in the FDH active site in MV facilitated bicarbonate reduction; (A) orientation of bicarbonate and hydronium ions in presence of MV •+ (green); residues making hydrogen bonds with the reactants (blue); (B) Orientation of the products formate and hydroxyl ions in presence of MV 2+ (green); residues that formed hydrogen bonds with the reactants in the MV •+ bound FDH (blue); these hydrogen bonds are disrupted upon product formation, as shown by the increased distances with the formate. 6.3.3.4 Involvement of A Second MV•+ in the Catalytic Reaction The two electrons required for converting bicarbonate to formate, should come from two molecules of MV •+ involved in the reaction. 38 To identify the binding site of the second molecule of MV •+ we docked a second MV •+ to FDH in the presence of the first MV •+ , HCO 3 - and H 3O + in the active site and performed MD simulations of the docked conformation. The second MV •+ molecule bound at the entry of the active site of FDH (Figure 6-7A), and it stayed stable in this position during the MD simulations. As shown in Figure 6-7B, several aromatic and hydrophobic residues in this surface pocket interact with the second MV •+ . Residues H232 and Y358 pi-stack with the pyridine rings of MV •+ . The distance between the two MV •+ molecules is about 8.5 Å. We speculate that the aromatic residues could be involved in electron transfer to the HCO 3 - . Based on the structure of the FDH binding pocket, we can expect that Y358 that pi stacks with the second MV•+ to transfer electrons on to the neighboring D96. D96 is involved in a salt bridge network with R258 and bicarbonate as shown in Figure 6-7C. Although the distance 8.3Å 9.3Å 5.1Å W284 H311 G259 HCOO - OH - MV 2+ W284 H311 G259 HCO 3 - H 3 O + MV ⦁+ 2.9Å 3.4Å 3.1Å (A) (B) 161 between Y358 and D96 is 5.5 Å, we can expect that the bridging water molecules between these two residues (as observed in the MD simulations) can facilitate the electron transfer. Figure 6-7: (A) Cross section of FDH showing the binding site of two bound MV •+ molecules (green); the aromatic residues making contact with the second MV •+ in the secondary pocket (magenta); the major active site residues involved in catalysis (blue); reactants bicarbonate and hydronium (green); (B) Magnified view of the FDH cross section near the active and secondary sites and (C) Possible electron transfer pathway from the second MV •+ molecule. (C) Y358 D96 R258 bicarbonate H 3 O + Second MV . + molecule Bridging water H311 162 6.3.4 Continuous Reduction of Carbon Dioxide to Formate in Two- Chamber and Three-Chamber Electrolyzer Configurations 6.3.4.1 Formate Accumulation under Continuous Steady-State Carbon Dioxide Reduction The experimental results in Figure 6-3C and Figure 6-4 and the MD simulations confirmed the feasibility of the continuous accumulation of formate by the reduction of carbon dioxide using FDH and the artificial cofactor, MV •+ . In a two-chamber electrochemical reactor with a PEM (Figure 6-8) we could demonstrate the continuous production and accumulation of formate in the enzymatic chamber. The concentration of formate in chamber B (Figure 6-8 A) increased linearly (Figure 6-8 B) with time, consistent with the constant current used for the production of MV •+ (Appendix D7). The continuous electro-regeneration of MV •+ occurred at -0.44 V vs. NHE, followed by the FDH-catalyzed carbon dioxide reduction to formate. Since the equilibrium potential value for carbon dioxide reduction to formate at pH 6.6 is -0.42 V vs. NHE (Appendix D1) we can conclude that the conversion to formate using FDH and MV can be achieved at an overpotential of 20 mV. This overpotential loss is very small compared to the several hundreds of millivolts of overpotential encountered during carbon dioxide reduction at metal electrodes. We found that in the configuration presented in Figure 6-5A with a PEM, MV •+ and MV 2+ could crossover from the enzyme chamber to the counter electrode, as the PEM allows for the facile transport of cations (Appendix D8). Consequently, the concentration of MV •+ and MV 2+ in the enzyme chamber decreased with time. Therefore, we took further steps to prevent the crossover of MV •+ and MV 2+ to the counter electrode chamber by introducing an AEM between the oxygen-evolving counter electrode and the MV reducing electrode (Figure 6-9). 163 Figure 6-8: Electrochemical formate generation: (A) Illustration of the two-chamber electrochemical reactor and (B) Results of continuous accumulation of formate from electrochemical reduction of carbon dioxide in the two-chamber reactor at an applied potential of -0.44 V vs. NHE using 40 mM MV 2+ , 2.6 µM FDH, 7.5 mL buffer solution in each chamber and Nafion® 117 PEM. H 2 O O 2 MV 2+ MV .+ CO 2 HCOO - Platinum electrode Toray Paper Electrode Enzyme FDH Enzyme-catalyzed CO 2 Reduction Electrochemical generation of Artificial co- factor H + PEM Chamber A Chamber B + - Power source (A) 164 Figure 6-9: Electrochemical formate generation (A) Illustration and (B) Experimental setup showing diffusion of MV across PEM and (C) Steady-state CO 2 reduction in the novel three chamber reactor at an applied potential of -0.44 V vs. NHE using 40 mM MV 2+ , 2.4 µM FDH, 7.5 mL buffer solution in each chamber, Tokuyama A901 AEM and Nafion® 117 PEM. Such a membrane would block the transport of cations, allowing only anions to be transported. Further, in this new configuration (Figure 6-6A), we also introduced a PEM between the enzyme chamber where formate was produced and chamber B where the MV •+ was produced. This additional separation ensured that formate did not leave the enzyme chamber. Thereby, the (A) (B) Platinum Electrode Toray paper Electrode Ag/AgCl reference electrode Chamber B MV in buffer solution PEM Chamber C Enzyme in buffer solution Chamber A Buffer solution AEM (B) 165 MV 2+ and the MV •+ could freely diffuse between chambers B and C through the PEM (Figure 6- 6B) and participate in the conversion of carbon dioxide to formate. We verified that no MV 2+ or MV •+ crossed over to chamber A, even after 21 hours of electrolysis (Figure 6B). While the rate of transport of MV •+ into chamber C was dependent on the membrane thickness and the concentration of MV •+ in chamber B, the steady-state generation of formate in chamber C was governed by the rate of the carbon dioxide reduction reaction with FDH and the MV •+ . Under steady-state conditions, the current required to maintain the production of MV •+ in chamber B is equal to both the mass flux of MV •+ into chamber C and the rate of consumption of MV •+ in chamber C for the conversion of carbon dioxide to formate. Thus, upon the addition of FDH and carbon dioxide to chamber C, we could observe an immediate jump in the value of the reduction current, indicating a sudden increase in the rate of consumption of MV •+ resulting from the reduction of carbon dioxide (Figure 6-6C). In 2 hours after introducing FDH and CO 2 into chamber C, the formate concentration rose to 0.72±0.01 mM. The rate of carbon dioxide reduction per unit concentration of the enzyme was calculated to be 75±3 h -1 (Appendix D9) and this value compares well with the k cat value of FDH that has been reported by Ikeyama et al. (61 h -1 , based on the Michaelis-Menten equation) in the photochemical reduction of carbon dioxide using FDH and MV •+ . 12 In all the studies discussed here, the regeneration of the MV •+ was carried out at just a few millivolts away from the reversible potential for the electrochemical reduction of carbon dioxide to formate. Thus, the conversion to formate could be achieved at high energy efficiency. Energy losses that result from the oxygen evolution reaction are small with catalytic electrodes based on iron-doped nickel hydroxide, well-studied by us and others. 21, 22 6.3.5 Formate Yield By comparing the rate of production of formate with the total charge passed for the reduction of MV, the yield of formate was determined to be 61±1% (Table 6-2). As no product other than formate is generated, we attributed the reduced yield to the diffusion of oxygen from chamber A to chamber B that results in the parasitic oxidation of the MV •+ . We estimated the oxygen 166 crossover rate to be 0.43±0.01 micromoles/cm 2 /hour. For attaining 100% of coulombic efficiency (CE) in this reactor we must avoid completely the diffusion of oxygen from chamber A to chamber B in addition to the loss of formate by crossover. Table 6-2: Yield of the formate in the three-compartment cell shown in Figure 6-6. 6.4 Chapter Six Conclusions We have demonstrated a novel approach to the continuous reduction of carbon dioxide to formate by reversing the normal function of metal-independent FDH in the wild type, using an electrochemically-generated artificial cofactor, the MV •+ . This method of electrochemical reduction of carbon dioxide relies on MV being an effective redox cofactor for the reduction of carbon dioxide but not for the oxidation of formate. Calculation of binding free energies from atomistic MD simulations showed that the wild-type FDH binds to the MV •+ , bicarbonate, and the hydronium ion in preference to the products, namely, the MV 2+ , and formate. Additionally, we observed that the cofactor NADH does not favor binding of the reactant, bicarbonate. The MD simulation results also provided insights into how a second MV •+ could bind at the entrance to the active site of FDH and provide the second electron required for the reaction to be completed. We have demonstrated the continuous production of formate at high energy efficiency and yield by carrying out the reactions in a novel three-compartment cell configuration. By the careful selection of ion-exchange membranes to separate these compartments we could preserve the yields of MV •+ and formate. Expected formate concentration, mM Detected formate concentration, mM Formate yield % 1.180±0.005 0.72±0.01 61±1 167 By the electro-regeneration of MV •+ at -0.44 V vs. NHE, we could produce formate at just 20 millivolts negative to the reversible electrode potential for carbon dioxide reduction to formate. Our results are in sharp contrast to the large overpotentials of -800 mV to -1000 mV required on metal catalysts. We anticipate the insights from the electrochemical studies and the MD simulations to be useful in re-designing FDH and the artificial cofactor to achieve even higher rates of conversion. 168 6.5 Chapter Six References 1. Jayathilake, B. S.; Narayanan, S. In Bio-Electrocatalytic CO 2 Reduction into Formate Using Metal-Independent Formate Dehydrogenase from Candida boidinii (Yeast), Meeting Abstracts, The Electrochemical Society: 2017; pp 1995-1995. 2. Jayathilake, B. S.; Bhattacharya, S.; Vaidehi, N.; Narayanan, S. R., Efficient and Selective Electrochemically Driven Enzyme-Catalyzed Reduction of Carbon Dioxide to Formate using Formate Dehydrogenase and an Artificial Cofactor. Accounts of Chemical Research 2019. 3. Blumenfeld, L. A.; Tikhonov, A. N., Principles of Enzyme Catalysis. In Biophysical Thermodynamics of Intracellular Processes, Springer: 1994; pp 86-111. 4. Hwang, H.; Yeon, Y. J.; Lee, S.; Choe, H.; Jang, M. G.; Cho, D. H.; Park, S.; Kim, Y. H., Electro-biocatalytic production of formate from carbon dioxide using an oxygen-stable whole cell biocatalyst. Bioresource technology 2015, 185, 35-39. 5. Parkinson, B. A.; Weaver, P. F., Photoelectrochemical pumping of enzymatic CO 2 reduction. Nature 1984, 309 (5964), 148-149. 6. Peck, H. D.; Gest, H., Formic Dehydrogenase and the Hydrogenlyase Enzyme Complex in Coli-aerogenes Bacteria. Journal of Bacteriology 1957, 73 (6), 706-721. 7. Reda, T.; Plugge, C. M.; Abram, N. J.; Hirst, J., Reversible interconversion of carbon dioxide and formate by an electroactive enzyme. Proceedings of the National Academy of Sciences 2008, 105 (31), 10654-10658. 8. Choe, H.; Ha, J. M.; Joo, J. C.; Kim, H.; Yoon, H.-J.; Kim, S.; Son, S. H.; Gengan, R. M.; Jeon, S. T.; Chang, R., Structural insights into the efficient CO 2-reducing activity of an NAD-dependent formate dehydrogenase from Thiobacillus sp. KNK65MA. Acta Crystallographica Section D: Biological Crystallography 2015, 71 (2), 313-323. 9. Davison, D. C., Studies on plant formic dehydrogenase. Biochemical Journal 1951, 49 (4), 520. 10. Ikeyama, S.; Amao, Y., The effect of the functional ionic group of the viologen derivative on visible-light driven CO 2 reduction to formic acid with the system consisting of water-soluble zinc porphyrin and formate dehydrogenase. Photochemical & Photobiological Sciences 2018. 11. Ikeyama, S.; Amao, Y., An Artificial Co-enzyme Based on the Viologen Skeleton for Highly Efficient CO 2 Reduction to Formic Acid with Formate Dehydrogenase. ChemCatChem 2017, 9 (5), 833-838. 12. Ikeyama, S.; Amao, Y., Novel Artificial Coenzyme Based on the Viologen Derivative for CO2 Reduction Biocatalyst Formate Dehydrogenase. Chemistry Letters 2016, 45 (11), 1259- 1261. 13. Amao, Y.; Ikeyama, S., Discovery of the Reduced Form of Methylviologen Activating Formate Dehydrogenase in the Catalytic Conversion of Carbon Dioxide to Formic Acid. Chemistry Letters 2015, 44 (9), 1182-1184. 14. Karyakin, A. A.; Ivanova, Y. N.; Karyakina, E. E., Equilibrium (NAD + /NADH) potential on poly(Neutral Red) modified electrode. Electrochemistry Communications 2003, 5 (8), 677-680. 169 15. Guo, Q.; Gakhar, L.; Wickersham, K.; Francis, K.; Vardi-Kilshtain, A.; Major, D. T.; Cheatum, C. M.; Kohen, A., Structural and Kinetic Studies of Formate Dehydrogenase from Candida boidinii. Biochemistry 2016, 55 (19), 2760-71. 16. Berg, J.; Tymoczko, J.; Stryer, L.; Stryer, L., Biochemistry, Ed 5th. WH Freeman, New York: 2002. 17. Jaegfeldt, H., A study of the products formed in the electrochemical reduction of nicotinamide-adenine-dinucleotide. Journal of Electroanalytical Chemistry and Interfacial Electrochemistry 1981, 128, 355-370. 18. Ali, I.; Soomro, B.; Omanovic, S., Electrochemical regeneration of NADH on a glassy carbon electrode surface: The influence of electrolysis potential. Electrochemistry communications 2011, 13 (6), 562-565. 19. Rusching, U.; Müller, U.; Willnow, P.; Höpner, T., CO 2 reduction to formate by NADH catalysed by formate dehydrogenase from Pseudomonas oxalaticus. The FEBS Journal 1976, 70 (2), 325-330. 20. Ito, M.; Kuwana, T., Spectroelectrochemical study of indirect reduction of triphosphopyridine nucleotide: I. Methyl viologen, ferredoxin-TPN-reductase and TPN. Journal of Electroanalytical Chemistry and Interfacial Electrochemistry 1971, 32 (3), 415-425. 21. Mitra, D.; Narayanan, S. R., A Stable and Electrocatalytic Iron Electrode for Oxygen Evolution in Alkaline Water Electrolysis. Topics in Catalysis 2018, 61 (7), 591-600. 22. Mitra, D.; Trinh, P.; Malkhandi, S.; Mecklenburg, M.; Heald, S.; Balasubramanian, M.; Narayanan, S., An Efficient and Robust Surface-Modified Iron Electrode for Oxygen Evolution in Alkaline Water Electrolysis. Journal of The Electrochemical Society 2018, 165 (5), F392- F400. 170 171 Performance Advancements for the Continuous Bio-Electrochemical Generation, Conservation and Separation of Formate from Carbon Dioxide using FDH and Artificial Cofactor Systems 7.1 Background Figure 7-1: Factors affecting in a bio-catalytic pathway of formate generation from carbon dioxide reduction The desirable properties of a catalyst for ERC are low overpotential, high product selectivity and no hydrogen evolution reaction (discussed in Chapters One and Six) .1 Chapter Six describes an enzymatic synthesis of formate, as a product of indirect-electrochemical CO 2 reduction achieved using artificial co-factors at low-cost carbon electrodes with no HER. Although enzymes remain the only reported reversible electro-catalysts for the reduction of CO 2 to carbon monoxide and formate with high product selectivity compared to other techniques, the use of enzyme-based systems at the industrial scale is highly challenging. 172 The practical realization of ERC using bio-catalytic systems requires improvements to charge- transfer kinetics and mass transport. Additionally, maintaining the physiological conditions like pH and temperature is required to achieve optimum enzyme functioning (Figure 7 1) during ERC. Various factors in the enzymatic chamber such as the activity of the enzyme, effective cofactor concentration, the pressure of CO 2, concentration of bicarbonate and the transfer of the cofactor to the active site of the enzyme can limit the rate of formate generation in FDH- catalyzed CO 2 reduction. Compared to an instantaneous batch-wise process, in a continuous process of CO 2 conversion, in addition to the generation rate of products, approaches of product conservation and removal are also essential. 7.2 Thermodynamics and Kinetics of Bio-catalytic Carbon Dioxide Reduction Reaction Kinetics of bio-catalytic CO 2 reduction reaction is usually determined by the activity of the enzyme. Binding of the cofactors and the substrates and proton-coupled charge transfer at the active site of the enzyme are the key kinetic factors. The rate of electro-reduction of the cofactor could also vary with the electrode materials and thereby affect the efficiency losses and rates of the overall reaction during a continuous operation. For the electrochemically-generated cofactor, the reducing power of the cofactor is a function of the standard reduction potential of the cofactor. Thus, the applied electrode potential can govern the rate of CO 2 reduction. In the bio-catalytic CO 2 conversion pathway, the use of cofactors with standard reduction potentials negative to that of the equilibrium potential of the carbon dioxide/formate reaction present favorable thermodynamics for the CO 2 reduction reaction. However, the more negative electrode potential can increase the overall cell voltage, reduce the efficiency and increase the cost of electricity for formate synthesis. Therefore, identifying the optimal co-factor based on thermodynamics, enzyme kinetics, and operating cost, is essential for the commercial application of bio-catalytic pathways for the synthesis of organic substances from carbon dioxide. 173 7.2.1 Mass Transport in Bio-catalytic Carbon Dioxide Reduction Reaction Membrane-based flow cells and microfluidic reactors are two different types of reactors used in CO 2 reduction research. 2 However, in industrial use, an enhanced flow of the reactants is required to overcome limitations of mass transport. Poor transport of the cofactor, carbon dioxide and protons to the electrode surface and the active site of the enzyme can lower the rate of the charge transfer process. In an experimental set-up with the configuration A (Chapter six Figure 6-9), the concentration of MV in the enzymatic chamber is a function of the rate of electro-reduction, diffusion and re-oxidation. In this Chapter, we discuss the diffusion of materials across membranes. In practice, diffusion is found to be a common rate-limiting mass transport process in many electrolysis systems. 7.2.1.1 The Rate of Diffusion in Electrochemical Systems Diffusion of cofactor and formate into the oxygen evolution chamber will result in re-oxidation of the formate and reduced form of the cofactor. Similarly, the diffusion of oxygen into the enzyme reactor chamber can lead to re-oxidation. The diffusion rate of formate, molecular oxygen or the cofactor across polymer electrolyte membranes depends on the concentration gradient, intrinsic membrane properties bulk pH of the solutions and the pH near the membrane. The steady-state rate of diffusion is governed by Fick’s first law. Fick’s first law shows that the diffusional flux, J o (amount of substance per unit area per unit time) is related to the concentration gradient across the membrane as per Eq. 7-1. b $ =−c- d' $ d( 5 (7-1) Where D: diffusion coefficient. However, when formate concentration increases with electrolysis time (t), Fick's second law (Eq. 7-2) can be used to determine the product cross-over or separation rates. - d' $ de 5 ( =c- d F ' $ d( F 5 e (7-2) 174 7.2.1.2 Ion-Exchange Membranes Since the diffusion of reactants and products can occur readily through the membrane separators in the reactor, the choice of membranes is important for the design of a reactor for long-term continuous CO 2 reduction. Membrane properties such as type, area, thickness, ion- conductivity, equivalence weight, wetting capability, etc. must be considered in the design of efficient ECR systems. 3 Cation-exchange, anion-exchange and bipolar membranes are three different categories of ion-exchange membranes that were tested in the CO 2 reduction reactors (Figure 7 2). 2, 4 These membranes resemble the semi-permeable membranes in biological systems and other electrochemical devices that enable selective ion-transport or avoid crossover. In ECR, the transport of charge-carrying ions can be controlled by the selective conductivity of the membrane. Proton-exchange membranes facilitate moving of cations across the membrane and anion-exchange membranes allow anion transport through the membrane. However, bipolar membranes are composed of two distinct layers of cation and anion exchange materials. Because of this unique characteristic, bipolar membranes are able to electrically separate water into protons and hydroxyl ions (Figure 7 2C) at the interface of two layers of the membrane. Figure 7-2: Different types of ion-exchange membranes A: Proton-exchange membrane (PEM), B: Anion-exchange membrane (AEM) and C: Bipolar membrane (BPM) used in CO 2 reduction A B C AEM OH - OH - OH - OH - + + + + + + + - + BPM OH - H + OH - H 2 O H 2 O H + + + + + - - - - - - + 175 7.2.2 Coulombic Efficiency in Bio-catalytic Carbon Dioxide Reduction Reaction In an electrochemical system, as discussed in Chapter One, the coulombic efficiency of the main reaction can be reduced by the occurrence of a parallel parasitic reaction. In the electrochemical carbon dioxide reduction reaction, hydrogen evolution is a parasitic or competitive reaction. However, when carbon electrodes are used hydrogen evolution is negligible as the hydrogen evolution reaction has a high overpotential. In the indirect-electrochemical carbon dioxide conversion system using FDH that is being pursued in this thesis work, we recognize that the recombination of molecular oxygen generated at the positive electrode with the reduced form of the cofactor (MV •+ ) is a competitive reaction for the CO 2 reduction reaction (Eq 7-3). 4MV •+ + 2H 2O + O 2 à 4MV 2+ + 4OH - (7-3) Therefore, the cross-over of the cofactor to the OER chamber and the movement of molecular oxygen (generated in the positive side electrode) to the cofactor-reduction chamber needs to be minimized to improve the coulombic efficiency. In addition, if products of CO 2 reduction can be transferred to the surface of the positive electrode, direct re-oxidation will occur and the coulombic efficiency of the CO 2 conversion process will be reduced. Therefore, product conservation and regulation of the diffusion rates of the products across ion-exchange membranes is an important consideration in designing membrane-based electrolysis cells. 7.2.3 Focus of Chapter Seven As discussed above, the optimum conditions for the continuous reduction of CO 2 into formate are based on the rate of formate generation, conservation, product separation, recyclability of the system, energy efficiency and the cost. Therefore, a system designing can be improved through the optimization of the reactor and electrode design, catalyst, selection of membranes and electrolyte. In the present study, we explore performance improvements to the FDH- catalyzed CO 2 reduction system in continuous process using different polymer electrolyte membranes. Additionally, as bio-catalytic processes are highly sensitive to pH, in this study, we 176 have discussed a system of achieving a continuous CO 2 reduction in a pH balanced system. In- situ separation of formate is also examined. Further, directions to minimize oxygen-MV •+ recombination are discussed to improve coulombic efficiency. Also, a number of alternate candidates artificial cofactors have been studied for CO 2 reduction using FDH. 7.3 Experimental 7.3.1 Materials Materials used in these experiments are similar to those described in section 6.2.1. Additionally, riboflavin-5-monophosphate, 2,7-AQDS, ethyl viologen and benzyl viologen were purchased from Sigma-Aldrich. Neutral red was obtained from TCI America. Magnesium strips were purchased from Amazon. 7.3.2 Studying MV and Formate Diffusion Across PEMs We studied the formate diffusion across different polymer electrolyte membranes. Two glass compartments were assembled together using a pinch clamp and separated using proton exchange membranes (PEMs) of area 0.64 cm 2 . The various types of PEMs included, Nafion Ò 117, 211, Fumasep E 750, F 1460 and F 1850). Compartment 1 contained 5 mL of 20 mM methyl viologen di-cation in 0.5 M phosphate buffer at pH 6.6 and compartment 2 contained 5 mL of 0.5 M phosphate buffer at pH 6.6. Both compartments were magnetically stirred for 24 hours and the MV concentrations in each compartment were analyzed by obtaining cyclic voltammograms using a carbon fiber microelectrode. In the study of formate diffusion, compartment 1 contained 100 mM sodium formate and HNMR was used to detect diffused formate to compartment 2. 7.3.3 Uptake of MV into PEMs Pieces of PEMs (1x 2 cm 2 ) were dipped separately in phosphate buffer solution (2 mL of 0.5 M at pH 6.6) containing the methyl viologen di-cation (20 mM). Each mixture was magnetically 177 stirred for 48 hours and remaining MV concentration in each mixture were analyzed by obtaining cyclic voltammograms using carbon fiber microelectrode. 7.3.4 Bipolar Membrane Fabrication A thin layer of Nafion ionomer solution (5%) was painted on a Tokuyama A901 membrane to fabricate the Nafion-coated AEM membrane. Membranes were dried in air for 24 hours. Tokuyama A901 and Nafion 325 membranes were stuck together using the Nafion ionomer solution. Membranes were dried in air for 2 hours and pressed for 30 min at 9 lbs/cm 2 . 7.3.5 Carbon Dioxide Reduction using FDH and the Reduced form of MV FDH, cofactor and sodium bicarbonate were added, and carbon dioxide gas at atmospheric pressure was bubbled into the reaction mixture. The reduced form of methyl viologen was supplied by electrochemical generation. 7.3.6 Screening of the Artificial Cofactors CV of a selected artificial cofactors were studied using a glassy carbon micro-disk electrode at 25 mV/s scan rate in sodium phosphate buffer at pH 7 to obtain the formal potential for oxidation and reduction at carbon electrodes. 178 7.4 Results and Discussion 7.4.1 Replacing PEM and AEM Separators with a Bipolar Membrane. Figure 7-3: CO 2 reduction using a bipolar membrane (reactor configuration B) The use of bipolar membranes (BPM) in CO 2 electrolyzers is known to prevent product cross- over and maintain a constant pH in long-term experiments. 2 In conventional BPMs, there is a 2D planar junction interface between the anion- and cation-exchange layers. 4 Therefore, the ion- transport across BPM is facilitated by the dissociation of water at the interface. During electrolysis the cofactor is reduced at the negative electrode, protons move towards the negative electrode and OH - ions move towards the positive electrode. At the positive electrode, water is oxidized to molecular oxygen. In our system, we expect that the BPM can prevent both MV 2+ and formate cross over from enzymatic chamber to OER chamber. Therefore, the coulombic efficiency of CO 2 reduction can be expected to be better than in the single membrane experiments. Table 7-1 shows that the rate of CO 2 reduction with a bipolar membrane expressed as a turn-over frequency is 156 hr 1 which is four times higher than with an AEM and 2.5 times higher compared to that with a PEM. From the observation of the three-fold lower current density of cofactor reduction in the experiment using a BPM compared to that using an AEM and a PEM, we can expect a lower H 2 O ½ O 2 2MV 2+ 2MV .+ HCO 3 - HCOO - Carbon Electrode Enzyme FDH Enzyme-catalyzed CO 2 Reduction Electrochemical generation of Artificial co-factor BPM OH - Chamber A Chamber B Oxygen evolution catalyst + - Power source H + H + OH - H 2 O OH - OH - H 2 O H + H + 179 rate of formate generation. However, NMR studies show that the use of the BPM conserves the generated formate in the enzymatic chamber at a slightly higher rate than obtained in the reactor configuration A which is explained in Chapter Six section 6.3.4. Also, improved formate yield could be observed in configuration B with the BPM (Table 7 1) compared to single- membrane experiments. This observation verifies that movements of cofactor and formate towards OER and the re-oxidation of them at the OER electrode can be prevented by using a BPM. Table 7-1: CO 2 reduction using different membranes and configurations (40 mM, MV 2+ ) Configuration label CO 2 reduction approach TOF±3, hr -1 Current density ±0.05, mA/cm 2 FDH, µM Volume, mL %Formate Yield A OER/AEM/MVR/ PEM/FDH 150 0.22 2.4 7.5 61 B OER/BPM (Nafion-coated on AEM)/MVR, FDH 156 0.30 2.4 7.5 46 C OER/AEM/MVR, FDH 62 0.95 1.2 15 6 D OER/PEM/MVR, FDH 40 0.88 2.6 7.5 4 7.4.2 pH Balancing During Bio-Electrocatalytic Carbon Dioxide Reduction Electrons for the indirect CO 2 reduction using bio-catalysts is provided by the OER reaction where OH - ions are converted to oxygen and protons. The protons are transported to the enzymatic chamber across the membranes to generate formate. As the OER reaction produces 180 two equivalents of protons from water, the continuous reduction of CO 2 to formic acid can be supported. However, Reda et al. has shown that a range of pH from 6.5 - 7.0 gives the optimum value for CO 2 reduction using metal-dependent FDH extracted from S. fumaroxidans. 5 Thus, the pH of the enzymatic chamber should be maintained near physiological pH to achieve the optimum activity of the enzyme. There are reports in the literature on the use of phosphate buffer and bicarbonate solutions with pH in the range of 6-7.4 for electrochemical studies of CO 2 reduction using FDH from Candida boidinii. 6-9 In our experiments we use buffer solutions to maintain a constant pH for a short-term. Bipolar membranes usually give a constant pH across an electrochemical cell. 4 The reactor configuration A is also similar to an expanded bipolar membrane system and we can expect to have a constant pH during CO 2 reduction. Carbon dioxide dissolution in water will generate bicarbonate as per Eq. 7-4 & 7-5. At near- neutral pH, bicarbonate dominates because of the pKa of bicarbonate is 6.3. However, as bicarbonate dissolution in water can give hydroxyl ions, achieving continuous CO 2 reduction using bicarbonate dissolved in phosphate buffer solutions can be conducted at a balanced pH (Eq.7-6). When the pH of the solution exceeds 10, bicarbonate is converted to carbonate (Eq. 7-5). CO 2 + + H 2O à H 2CO 3 (7-4) CO 3 2− + 2H 2O ⇌ HCO 3 − + H 2O + OH − ⇌ H 2CO 3 + 2OH − (7-5) 2HCO 3 - à 2HCOO - + O2 (7-6) For any of the above cell configurations bicarbonate yields a pH-balanced CO 2 reduction if formate is not lost from the enzymatic chamber to the oxygen production chamber. Three different schemes with different membranes facilitating ion transport to achieve equilibria and pH balance in the reactor are discussed in next section. Additionally, these reactor designs include formate separation units. 7.4.3 In-situ Separation of Formate for Recycling the Reaction Mixture Industrial production of formate using FDH and cofactor will require an additional step of separation of formate from the reaction mixture consisting of the enzyme and the buffer. This 181 separation does not involve the enzyme in a metal electrode based direct electrochemical catalytic CO 2 reduction. Separation of formate has two benefits namely, purifying formate for using as a raw material or feedstock for production of multi-carbon compounds, and for the repeated recycling of the electrolyte mixture for long-term use. The 3-chambers approach we proposed in Chapter Six (Configuration A) requires an external formate separation and electrolyte regeneration unit such as the anion-exchange column. However, achieving continuous CO 2 reduction can be more beneficial if we can separate the formate in situ. Therefore, in the present study we propose to add a chamber separated by an AEM to achieve formate separation. Formate ion-transfer via anion-exchange membranes in diffusion dialysis is used for recovery of acids. 10 Both configurations A and B, can be extended with in-situ formate separation using the extra chamber (chamber D in Figure 7-4 and chamber C in Figure 7-5). Figure 7-5 is a “compressed” version of Figure 7-4 where the bipolar membrane replaces the combination of both AEM and PEM. There can be a passive diffusion of bicarbonate and formate across the second AEM (between the enzymatic chamber and the formate separation unit). Properties of AEM are tabulated in Appendix E1. Another novel configuration for achieving CO 2 reduction in membrane cells is shown in Figure 7-6. In this configuration, formate ions generated in the enzymatic chamber can be exchanged with bicarbonate ions supplied to the central chamber and the central chamber can accumulate formate during electrolysis. In this configuration, we could achieve both MV reduction and CO 2 conversion in a single chamber while preventing MV cross over to either central chamber or OER chamber. The PEM separates, the OER chamber from the central chamber and AEM separates the enzymatic chamber. Therefore, during electrolysis proton transfer occurs from OER to the central chamber. During electrolysis, OH - ions and formate ions can be transferred from the MV reduction chamber to the central chamber as charge carriers. The disadvantage of this configuration is internal resistance can be higher due to two membranes and two layers of electrolytes between OER and MV reduction electrodes. Further, the rate of CO 2 reduction reaction can highly depend on the conductivity of the charge carriers via AEM, PEM and 182 electrolyte layer in the central chamber. However, compared to the other two schemes (Configuration E and F), as the cost for an additional AEM can be removed, Configuration G is more viable. Formate and OH - ions can also be passively exchanged across AEM. However, in all three configurations, as bicarbonate is continuously supplied and formate is taken away, we can expect a continuous CO 2 reduction process with no pH change. Water transfer can be expected due to osmotic pressure balance between adjacent chambers. Figure 7 6-B shows that even though the rate of formate generation is constant from the beginning to end, formate accumulation in the central chamber increases exponentially with the time due to the increased concentration gradient of formate across the AEM from the MV reduction chamber to the central chamber. The increased rate of formate generation in this novel cell configuration is higher as the previously described experiment with configuration due to the minimized crossover of cofactor and formate to OER. The rate of the CO 2 reduction reaction is determined by the effective concentration of the MV in the reduced form in the FDH chamber. Other than of achieving the separation of formate, having a central chamber can minimize the recombination of the MV reduced form and molecular oxygen leading to a higher effective concentration of MV reduced form in this novel system. Figure 7-4: Proposed CO 2 reduction scheme (configuration E) for achieving constant pH and formate separation (a modified design from reactor configuration A) ½ O 2 2MV 2+ 2MV . + 2MV 2+ ½ O 2 2MV . + HCO 3 - HCOO - 2H + AEM 2OH - PEM AEM HCO 3 - HCO 3 - H 2 O OH - FDH Oxygen evolution catalyst Carbon Electrode Enzymes Electrochemical generation of Artificial co-factor Enzyme-catalyzed CO 2 reduction H 2 O H 2 O HCOO - Formate separation unit + - Power source Chamber B Chamber C Chamber A 183 Figure 7-5: CO 2 reduction A: scheme and B: experimental set-up (configuration F) for achieving constant pH and formate separation (a modified design from reactor configuration B) The transfer of MV from the MV reduction chamber to the carbon dioxide reduction chamber (FDH containing chamber) in the scheme given in Figure 7-4 depends on the diffusion rate. Therefore, we studied MV diffusion rates in different PEMs to identify the optimum membrane. The screening of membranes showed that the use of Nafion Ò membranes can be more effective in transferring MV from MV reducing chamber to the enzymatic chamber (discussed below in section 7.3.4). Figure 7-6: A: Bio-catalytic CO 2 reduction using in-situ formate separation using the reactor configuration G and B: Formate accumulation in the separation chamber with time A 0.0 0.5 1.0 1.5 0 10 20 30 Diffused Formate accumulation, mM Time, hr B O 2 2MV 2+ 2MV . + ½ O 2 HCOO - H + AEM OH - PEM Oxygen evolution catalyst Carbon Electrode Enzyme FDH H 2 O H + OH - H + H + HCO 3 - HCO 3 - OH - H 2 O HCO 3 - OH - H 2 O HCOO - Formate separation unit Electrochemical generation of Artificial co- factor and Enzymatic reaction unit + - Power source Chamber A Chamber B Chamber C A B H 2 O ½ O 2 2MV 2+ 2MV .+ HCO 3 - HCOO - Carbon Electrode Enzyme FDH Enzyme-catalyzed CO 2 Reduction Electrochemical generation of Artificial co-factor BPM OH - Chamber A Chamber B Oxygen evolution catalyst + - H + H + OH - H 2 O OH - OH - H 2 O H + H + HCO 3 - HCO 3 - AEM Formate separation unit HCOO - Chamber C Power source 184 CO 2 reduction experiments using different reactor configurations with formate separation unit show that the configuration G is 5% more efficient in the separation of formate (in 21 hrs) than that in the configuration F (Table 7 2). In the configuration G, in addition to passive ion exchange of formate with bicarbonate, formate transport can happen via AEM as charge carriers from the MV reduction chamber to the central chamber during OER and MV reduction at two electrodes. With time more formate can be generated and accumulated in this configuration. In Configuration G, 6.5 mM of formate could be detected in the enzymatic chamber at the end of 21 hours. When the generation of formate is higher, a higher contribution of formate as charge carriers can also be expected. Therefore, the separation of formate can be efficient when a higher formate concentration is generated. Table 7-2: CO 2 reduction process combined with an in-situ formate separation However, the conservation of formate in the enzymatic chamber and separation of formate via AEM is dependent on the diffusion properties of products and reactants. In the next sections, I describe membrane screening experimental results that we tested to select the optimum diffusion conditions for achieving the CO 2 reduction. 7.4.4 Screening PEMs Based on Diffusion 7.4.4.1 Effective Concentration of MV Using different PEMs in the absence of electrolysis, we studied the decrease in concentration of MV in compartment 1 due to diffusion. Experimental results of the diffusion of MV into Configuration label CO2 reduction approach Rate formate generation mM/ FDH mM, hr -1 Formate separation % F OER/BPM/MVR, FDH / AEM/ Buffer 24 9 G OER/PEM/Buffer/ AEM/ MVR, FDH 114 14 185 compartment 2 across 5 different PEMs showed that Fumasep F-1850 membrane can provide the highest effective MV concentration in the original chamber (compartment 1) due to a low rate of cross-over. Overall, observations presented in Table 7-3 and Figure 7-7 prove that the cyclability of the cofactor for long-term conversion of CO 2 cannot be achieved when we use PEMs to seperate the chambers for MV reduction and oxygen evolution reaction. Effective concentrations of MV during CO 2 reduction in single membrane experiments were quantified using NMR and results showed that by using Nafion Ò 117 (Appendix D7), a significant loss of MV could be observed at the beginning of electrolysis. Nafion Ò 211 shows a gradual drop with time (Figure 7 8). The rate of the concentration drop of MV in Nafion Ò 211 during CO 2 reduction experiment was similar to the rate found in the independent studies of MV diffusion. Studying the MV diffusion across PEMs (Table 7 3) suggest that the use of Nafion Ò 211 could give the highest effective MV concentration in the enzymatic chamber when we use the configuration A or E (Figure 7 4) for achieving CO 2 reduction. Figure 7-7: Methyl viologen diffusion across different PEMs However, the sudden drop of MV concentration in the experiment with Nafion Ò 117 and visible evidence of blue coloration in the membrane after electrolysis (Appendix D7-3) verifies that MV can be incorporated into PEMs. The MV di-cation incorporation analysis within 48 hours (Table 7 3) indicates that the incorporation of MV into PEMs depends mainly on the volume of the membrane. MV cation absorption (normalized to thickness) into hydrocarbon membranes 106 100 175 134 130 5 38 5 18 45 89 63 20 48 25 0 50 100 150 200 Nafion 211 Nafion 117 F 1850 E 750 F 1460 MV number of umoles Compartment 2 membrane Compartment 1 186 (Fumapem E-750 and F-1460), which contains PEEK, is higher compared to fluorinated membranes (Nafion Ò 211, 117 and Fumasep F-1850). Table 7-3: Incorporation of MV di-cation into different PEMs Membrane Thickness, µm MV 2+ absorption– normalized to thickness, mmol cm -3 Diffusion coefficient x 10 8 , cm 2 / s Nafion Ò 211 25.4 0.79 2.06 Nafion Ò 117 183 0.78 10.40 Fumasep F-1850 50 0.77 0.92 Fumapem E-750 50 0.87 2.18 Fumasep F-1460 60 0.95 1.34 Figure 7-8: Methyl viologen concentration in the enzymatic chamber vs. electrolyzing time 7.4.4.2 Studying Formate Diffusion Across PEMs Formic acid diffusion across AEMs and BPMs have been studied by others in electrodialysis and bipolar membrane electrodialysis. 10, 11 If there are formate incorporation and cross-over into/across PEMs, similar to the observation of MV adsorption and cross-over of the products of CO 2 reduction, we can expect a lower formate concentration compared to the actual 0 5 10 15 20 25 30 35 40 45 0 10 20 30 40 50 60 MV dication in MV reduction chamber, mM Time, hr MV diffusion rate: 5.8 umol/ hr/ cm 2 187 concentration of formate generated from the enzymatic reaction. To study this hypothesis, independently, we measured the cross-over of formate across different cation-exchange membranes. Even though, the movement of formate ions via PEM cannot be anticipated as PEMs usually rejects anions, our experimental findings from independent studies (Table 7-4) show that formate can cross-over via PEMs near neutral pH. These findings are not surprising with the observation of four times lowering of the formate accumulation in the CO 2 reduction experiment with Nafion Ò 211 membrane (TOF=11 hr -1 ) compared to that with Nafion Ò 117 experiment (TOF=40 hr -1 ). The formic acid concentration in bulk is a function of the dissociation constant and pH. Therefore, the formic acid concentration at neutral pH can be very low (Eq. 7-6 & 7- 7). However, being a salt of a weak acid, formate can associate with protons at the interface of the membrane and form formic acid (reverse reaction of Eq. 7-6). Formic acid is an uncharged molecule and thus can be transported across a cation-exchange membrane to the OER chamber by diffusion. As there is a concentration gradient from the enzymatic chamber to OER chamber, diffusion can be expected. Further, in CO 2 reduction reaction experiment using configuration B and C, we can expect formate re-oxidation into CO 2 at the platinum wire electrode in the OER chamber. Therefore, the diffusion gradient will persist and the accumulation rate of formate in the enzymatic chamber will be low when formate crossover is present. HCOOH D HCOO - + H + (7-6) gh= [iX][j. ] [ij] =10 -3.75 (7-7) Nafion Ò 211 and 117 show higher rates of formate crossover (Figure 7-9 and Table 7-4) compared to the other cation-exchange membranes. As Nafion Ò 211 is thinner, formate cross- over rate can be higher and the concentration of formate accumulation in the enzymatic chamber from the CO 2 reduction experiment is lower. The membrane based on s-PEEK, E-750 (50 µm) and E 1460 (60 µm), provide a moderate cross-over of formate. Nevertheless, experimental observation of F1850 shows the lowest cross-over of formate compared to other four membranes in the present study. F 1850 which 188 is a fluorinated cation-exchange membrane (thickness 50 µm) with a low resistance is reported to a have high blocking capability of vanadium ions and high stability in an acidic environment. Also, it has shown a low cross-over rate for organic molecules in organic redox flow cells compared to Nafion Ò 117. 12 Figure 7-9: Formate Diffusion via different PEMs Table 7-4: Formate cross over via different PEMs Membrane Rate, mM hr -1 Thickness, µm Diffusion coefficient x 10 8 , cm 2 / s Nafion Ò 211 0.451 25.40 2.49 Nafion Ò 117 0.215 183.00 8.54 E-750 0.053 50.00 0.58 F-1850 0.002 50.00 0.02 F-1460 0.041 60.00 0.53 0 10 20 30 40 50 0 50 100 150 200 Diffused Formate Concentration, mM Time, hr F 1460 F-1850 Nafion 117 Nafion 211 E 750 189 Also, the poor performances in the reactors similar to configuration F with different BPMs such as the Nafion-coated AEM, the in-house created bipolar membrane (Figure 7-9) can be due to formate cross-over as neutral molecules via BPMs. Therefore, formate conservation should be considered as another factor when we need to screen ion exchange membranes to use in CO 2 electrolyzers. Figure 7-10: In-house created bipolar membrane by combining Tokuyama A-901 and Nafion Ò 325 membranes 7.4.4.3 Optimal Membranes and Configuration for Achieving CO2 Reduction with Advanced Performance These studies reveal that the loss of the coulombic efficiency in the experiments using PEM can be due to formate cross-over as well. F1850 may be down-selected as the best single membrane for a configuration D as it gives the lowest cross-over rate for both formate and MV. Even though higher rates of MV transfer are required for achieving CO 2 reduction using configuration A, PEM selection criteria based on the effective methyl viologen diffusion can no longer be valid as the best membrane for MV diffusion can give the lowest ability to for formate conservation in the enzymatic chamber. However, configuration G with a central formate separation unit can be a viable method using the F1850 membrane as formate can be conserved from moving to the OER chamber. Obtaining a higher effective concentration of MV in the enzymatic chamber and conservation of formate by preventing the cross-over to OER, and formate separation using AEM can be possible according to the configuration F if an appropriate BPM can be synthesized to conserve both ions. The required membrane properties of synthesizing of BPMs can be decided from the understanding of both methyl viologen and formate diffusion rates. 190 7.4.5 Approaches to Improving the Coulombic Efficiency Earlier, we discussed that in addition to the product cross-over, molecular oxygen diffusion from the OER chamber to MV reduction chamber via ion-exchange membranes is also a major factor for lowering the performance of the energy conversion. The lower rate of formate generation with Nafion Ò 211 compared to 117 can also be due to higher oxygen diffusion across the thinner membrane which can also lower the effective concentration of the reduced form of MV. Consequently, the rate of CO 2 reduction can be lowered. Independent studies of the CE for the production of MV •+ (without any CO 2 reduction) using different electrolyte membranes (Table 7-5) verified that the reaction of the MV •+ with in-situ generated molecular oxygen is the biggest parasitic loss leading to the reduction of CE of CO 2 reduction. The formate yield in the CO 2 reduction experiment, after accounting for the loss of MV •+ by reaction with oxygen, was re- calculated to be 97±1%. By introducing a glass frit that restricted the oxygen transport, we could increase the CE of MV •+ generation to 96±1% (Table 7-5). A glass frit in conjunction with AEM to separate chamber A and B (configuration A, Figure 6-9) can be a viable approach. In yet another modification to realize 100% CE, we completely avoided the oxygen evolution reaction by using a magnesium metal strip as the counter electrode. Thus, instead of oxygen evolution, magnesium dissolution to magnesium hydroxide was the reaction at the counter electrode reaction while MV was reduced at the working electrode. Under these conditions, the CE of MV reduction was 98±1%. While the use of a magnesium strip proved the ability to reach high coulombic efficiencies, magnesium is a sacrificial electrode and is not recommended because of the effects of dissolved magnesium ions on the electrolyte pH and deactivation of the enzyme. Table 7-5: Coulombic Efficiency of Methyl Viologen Reduction Counter electrode reaction Separator Coulombic Efficiency±1 % Oxygen evolution AEM 901 63 Oxygen evolution Glass frit 96 Magnesium dissolution AEM 901 98 191 7.4.6 Step-wise Reaction of MV •+ with FDH and Carbon Dioxide We recognize that attaining 100% of coulombic efficiency in the reactor configurations A-G, we must avoid the diffusion of oxygen from chamber A to chamber B and others. In this manner, the concentration of MV •+ is not lost to recombination with oxygen. To verify that the reaction rates can indeed be higher in the absence of recombination with molecular oxygen, we conducted a two-step batch reactor experiment (Figure 7-11) where we produced a known amount of electro-reduced methyl viologen free-radical cation and in the second step we combined this solution of MV •+ in a separate chamber with FDH and carbon dioxide to produce formate. Under these conditions, we observed an apparent TOF value of 620±3 hr -1 which is four times higher than the best TOF values obtained here with the continuously operating membrane reactors. Figure 7-11: Schematic diagram of electrochemical reduction of MV using an anion exchange membrane and FDH catalyzed CO 2 reduction 7.4.7 Screening Redox-Active Cofactors Screening of artificial cofactors with improved redox properties to replace the expensive NAD + / NADH cofactors remain is a promising area of research as bio-catalytic reaction pathways of material synthesis is attractive. Blue N-3 (dual presence of a blue anthraquinone chromophore and an N-substituted nicotinamide ring) and naphthalene based synthetic precursors are studied as artificial cofactors for dehydrogenases. 13 Tanja et al. reported that biomimetic synthetic MV 2+ MV . + MV 2+ O 2 MV . + CO 2 HCOO - AEM OH - Toray paper electrode Enzyme FDH Enzyme-catalyzed CO 2 Reduction Reaction Chamber Electrochemical generation of Artificial co-factor H 2 O + - Power source Platinum Electrode 192 cofactors using substituted nicotinamide can perform better than natural cofactors in bio- catalytic reactions. 14 To investigate electrochemically recyclable bio-mimetic cofactors for bio- catalytic reactions, understanding of their redox reversibility is also important. Therefore, as the initial step of screening candidates for replacing NAD + /NADH, we have studied cyclic voltammograms of selected small molecules. Figure 7-12 shows highly reversible CVs and Table 7-6 provides electrochemical properties of these candidate molecules. Figure 7-12: Cyclic voltammogram of candidate artificial cofactors in phosphate buffer at pH 7 A: Different viologen molecules, B: Bio-compatible organic molecules and C: 2,7-AQDS Ikeyama et al. have studied a number of viologen derivatives for CO 2 reduction (Appendix E2) and reported that 1,1’-diaminoethyl-4,4’- bipyridinium (AEV) can show the highest photocatalytic CO 2 reduction using FDH. 15 To achieve a steady-state CO 2 reduction, we need B A C -0.6 -0.4 -0.2 0 0.2 Potential, V vs NHE -40 -30 -20 -10 0 10 20 30 Current, uA Amino Ethyl Viologen Ethyl Viologen Benzyl Viologen -1 -0.5 0 0.5 1 Potential, V vs NHE -60 -40 -20 0 20 40 60 Current, uA Riboflavin Neutral Red Methylene Blue -0.8 -0.6 -0.4 -0.2 0 0.2 Potential, V vs NHE -200 -150 -100 -50 0 50 100 Current, uA 2,7-AQDS 193 to electrochemically reduce the cofactor using low-cost carbon electrodes and the rates of reduction can be improved having a high concentration of the cofactors. Compared to methyl viologen, ethyl viologen has a lower solubility. However, except for ethyl viologen, all other artificial cofactors which we have screened in the present study show more positive reduction potentials to methyl viologen. These observations indicate that thermodynamic favorability for CO 2 reduction with these candidates can be lower than MV. Even though cofactors with less negative reduction potential can lower thermodynamic driving force, the positive shift of the applied electrode potential can improve the overall conversion energy efficiency. Also, as the electrode potential can control the concentration ratio of the reduced and oxidized form of the cofactor, if thermodynamics are slightly favorable, still we can attempt to achieve CO 2 reduction. However, favorable binding of the cofactor and substrate to the enzyme active site is required for achieving the CO 2 reduction. Sayan Kar from Professor Prakash’s group synthesized AEV for the present study based on the synthesis procedure given by Ikeyama et al. 16 AEV shows a similar CV to neural red (Figure 7- 12). In previous studies of photocatalytic CO 2 reduction by Ikeyama et al. reported that dithionite reduced AEV can show 10 and 4 times higher formate generation compared to NADH and MV respectively. 16 We also attempted CO 2 reduction using FDH and electrochemically reduced AEV. In an electrochemical cell using a PEM and two chambers, the rate of CO 2 reduction using AEV is 46 hr -1 which is slightly higher to that using MV. Also, Ikeyama et al. proposed that higher positive charge on the AEV molecule compared to MV can facilitate better binding to the active site of FDH. As the reduction potential of EAV is 40 mV lower in AEV, we can predict that the equilibrium constant of the CO 2 reduction using AEV is lower compared to that with MV. However, upon application of -0.44V vs. NHE in both systems, AEV can give higher concentrations of the reduced form than that of MV. Therefore, the higher, concentration of the reduced form of viologen can compensate for the lower standard thermodynamic driving force for CO 2 reaction. Even with lower thermodynamic favorability, as AEV can give slightly higher formate yield, we can predict that binding of AEV to FDH can be as favorable as MV. Further studies will be required to understand the binding of the AEV to the enzyme. 194 For the continuous CO 2 reduction and conservation of the cofactor in the enzymatic chamber, we can assume that diffusion of AEV can occur to the OER chamber across PEM. NMR studies show a cross-over rate of 16 µmol/hr/cm 2 for AEV which 3 times is higher compared to the diffusion of MV across PEM. Despite the higher size of AEV, the diffusion rates can be higher due to the presence of four positive charges in the molecule. Further analysis of AEV based CO 2 reduction is required to verify the similar or enhanced performance of AEV even with a lower standard reduction potential. Benzyl viologen shows a slightly positive reduction potential to AEV. However, due to the larger size and similar di-positive charge of BV to MV, we can expect less cross-over. Neutral red and methylene blue are widely used in biochemistry as they have higher bio- compatibility. 17-19 Riboflavin is also a highly compatible molecule with bio-catalysts due to structural similarities to the flavin mononucleotide. However, as riboflavin and methylene blue show lower reduction potentials which can be lower than the standard reduction potential of CO 2, we haven’t attempted to achieve CO 2 reduction. And neutral red shows a significant cross- over in both acidic and neutral media across AEM (reactor configuration C, Figure 7-11). Thus, the use of the neutral red without immobilizing to the electrode can be problematic in a long- term and continuous CO 2 reduction process Figure 7-13: Cross-over testing of neutral red across an AEM AQDS and its derivatives that are used with different enzymes as a mediator for facilitating charge transfer in bio-sensor-based studies, can be viable candidates as they show more negative redox potentials and reversible redox chemistry (Figure 7-10C). 20, 21 We attempted CO 2 reduction using 2-7, AQDS in a reactor similar to configuration C at -0.44 V vs. NHE. However, 195 as AQDS itself shows smaller peaks in the region of formate peak in 1 H NMR (Appendix E3), reliable detection of formate from CO 2 reduction experiments was not successful. Further experiments with sensitive detection techniques are required to conclude the behavior of 2,7- AQDS as a reliable cofactor for CO 2 reduction. All these molecules except 2,7-AQDS and riboflavin show a net cationic charge and we can assume it might help to bind to the active site of the FDH, as we have discussed in our previous chapter, having a cationic group similar to nicotinamide ring in NAD + may be helpful for the binding. However, if the electron transfer happens not via binding, we can still try anionic reversible redox couples with negative charges. Under these conditions, the cross-over of both the co-factor and the final product can be minimized using PEMs to separate OER chamber and co-factor chamber. 196 Table 7-6: Electrochemical properties of candidate artificial cofactors Candidate artificial cofactor Molecular structure ½(Ep,a+Ep,c), mV vs. NHE Diffusion Coefficienx10 6 , cm 2 / S Ethyl Viologen -441 18 Amino ethyl Viologen -303 3.4 Benzyl Viologen -297 17 Neutral red -326 0.02 Methylene Blue 108 25 Riboflavin 5’ monophosphate sodium salt -196 0.25 2,7-AQDS -265 1.6 N + N + N + N + N N N NH 2 H Cl S N N N + Cl - Na + Na + N NH O N OH HO OH N O P O O O - O - Na + Na + S O O O - S O O - O O O N + H 2 N N + NH 2 Br - Br - 197 7.5 Chapter Seven Conclusions In this chapter, a variety of factors for improving the performance of bio-electrocatalytic were examined. We have found that the selection of membranes is critical to achieving the continuous operation of the system and improving the coulombic efficiency. The F-1850 membrane was identified to give the lowest cross-over of both formate and methyl viologen cofactor. Elimination of the molecular oxygen recombination with the reduced form of the MV was identified as the key direction of improving the coulombic efficiency of the system. We have shown that yields as high as 97±1 % can be realized by avoiding adventitious re-oxidation of MV •+ by molecular oxygen. Among the several reactor configurations studied in the present discussion, the batch-wise CO 2 reduction approach showed the best %yield of formate. However, two novel configurations with an in-situ formate separation were identified as reliable approaches for achieving pH-balance and long-term continuous CO 2 reduction. Out of several screened artificial cofactors, amino ethyl viologen (AEV) could be verified as an effective artificial cofactor for CO 2 reduction using FDH. The CO 2 reduction rate was comparable to MV while using the electrochemically reduced form of AEV. 198 7.6 Chapter Seven References 1. Jones, J. P.; Prakash, G.; Olah, G. A., Electrochemical CO 2 reduction: recent advances and current trends. Israel Journal of Chemistry 2014, 54 (10), 1451-1466. 2. Weekes, D. M.; Salvatore, D. A.; Reyes, A.; Huang, A.; Berlinguette, C. P., Electrolytic CO2 Reduction in a Flow Cell. Accounts of Chemical Research 2018, 51 (4), 910-918. 3. Endrődi, B.; Bencsik, G.; Darvas, F.; Jones, R.; Rajeshwar, K.; Janáky, C., Continuous- flow electroreduction of carbon dioxide. Progress in Energy and Combustion Science 2017, 62, 133-154. 4. Li, Y. C.; Yan, Z.; Hitt, J.; Wycisk, R.; Pintauro, P. N.; Mallouk, T. E., Bipolar Membranes Inhibit Product Crossover in CO 2 Electrolysis Cells. Advanced Sustainable Systems 2018, 2 (4), 1700187. 5. Reda, T.; Plugge, C. M.; Abram, N. J.; Hirst, J., Reversible interconversion of carbon dioxide and formate by an electroactive enzyme. Proceedings of the National Academy of Sciences 2008, 105 (31), 10654-10658. 6. Srikanth, S.; Maesen, M.; Dominguez-Benetton, X.; Vanbroekhoven, K.; Pant, D., Enzymatic electrosynthesis of formate through CO 2 sequestration/reduction in a bioelectrochemical system (BES). Bioresource technology 2014, 165, 350-354. 7. Srikanth, S.; Alvarez Gallego, Y.; Vanbroekhoven, K.; Pant, D., Enzymatic Electrosynthesis of Formic Acid through CO2 Reduction in Bioelectrochemical System (BES): Effect of Immobilization and Carbonic Anhydrase Addition. ChemPhysChem 2017. 8. Zhang, L.; Ong, J.; Liu, J.; Li, S. F. Y., Enzymatic electrosynthesis of formate from CO 2 reduction in a hybrid biofuel cell system. Renewable Energy 2017, 108, 581-588. 9. Kim, S.; Kim, M. K.; Lee, S. H.; Yoon, S.; Jung, K.-D., Conversion of CO2 to formate in an electroenzymatic cell using Candida boidinii formate dehydrogenase. Journal of Molecular Catalysis B: Enzymatic 2014, 102, 9-15. 10. Güler Akgemci, E.; Ersöz, M.; Atalay, T., Transport of Formic Acid Through Anion Exchange Membranes by Diffusion Dialysis and Electro-Electro Dialysis. Separation Science and Technology 2005, 39 (1), 165-184. 11. Ferrer, J. S. J.; Laborie, S.; Durand, G.; Rakib, M., Formic acid regeneration by electromembrane processes. Journal of Membrane Science 2006, 280 (1), 509-516. 12. Murali, A.; Nirmalchandar, A.; Krishnamoorthy, S.; Hoober-Burkhardt, L.; Yang, B.; Soloveichik, G.; Prakash, G. K. S.; Narayanan, S. R., Understanding and Mitigating Capacity Fade in Aqueous Organic Redox Flow Batteries. Journal of The Electrochemical Society 2018, 165 (7), A1193-A1203. 13. McLoughlin, S. B.; Lowe, C. R., An enzymatically active artificial redox coenzyme based on a synthetic dye template. Enzyme and Microbial Technology 1997, 20 (1), 2-11. 14. Knaus, T.; Paul, C. E.; Levy, C. W.; de Vries, S.; Mutti, F. G.; Hollmann, F.; Scrutton, N. S., Better than Nature: Nicotinamide Biomimetics That Outperform Natural Coenzymes. Journal of the American Chemical Society 2016, 138 (3), 1033-1039. 199 15. Ikeyama, S.; Abe, R.; Shiotani, S.; Amao, Y., Novel Artificial Coenzyme Based on Reduced Form of Diquat for Formate Dehydrogenase in the Catalytic Conversion of CO2 to Formic Acid. Chemistry Letters 2016, 45 (8), 907-909. 16. Ikeyama, S.; Amao, Y., Novel Artificial Coenzyme Based on the Viologen Derivative for CO2 Reduction Biocatalyst Formate Dehydrogenase. Chemistry Letters 2016, 45 (11), 1259- 1261. 17. Heli, H.; Bathaie, S. Z.; Mousavi, M. F., Electrochemical investigation of neutral red binding to DNA at the surface. Electrochemistry Communications 2004, 6 (11), 1114-1118. 18. Karyakin, A. A.; Ivanova, Y. N.; Karyakina, E. E., Equilibrium (NAD + /NADH) potential on poly(Neutral Red) modified electrode. Electrochemistry Communications 2003, 5 (8), 677-680. 19. Zhan, R.; Song, S.; Liu, Y.; Dong, S., Mechanisms of methylene blue electrode processes studied by in situ electron paramagnetic resonance and ultraviolet–visible spectroelectrochemistry. Journal of the Chemical Society, Faraday Transactions 1990, 86 (18), 3125-3127. 20. dos Santos, A. B.; Cervantes, F. J.; van Lier, J. B., Azo dye reduction by thermophilic anaerobic granular sludge, and the impact of the redox mediator anthraquinone-2,6- disulfonate (AQDS) on the reductive biochemical transformation. Applied Microbiology and Biotechnology 2004, 64 (1), 62-69. 21. dos Santos, A. B.; Cervantes, F. J.; Yaya-Beas, R. E.; van Lier, J. B., Effect of redox mediator, AQDS, on the decolourisation of a reactive azo dye containing triazine group in a thermophilic anaerobic EGSB reactor. Enzyme and Microbial Technology 2003, 33 (7), 942- 951. 200 201 Future Directions 8.1 All-Iron Redox flow Cell In Chapter Two, I have discussed in detail the challenges of the all-iron redox flow cell due to the parasitic hydrogen evolution reaction (HER). Also, I have evaluated a number of factors directed at minimizing the HER. But minimizing the last 2% of coulombic efficiency loss is challenging. However, without achieving 100% coulombic efficiency, the technical viability of the system is limited. Thus, future work should be focused on total suppression of HER and its effects. Some of the directions towards addressing this challenge are discussed in the following sections. 8.1.1 Electrolyte Recombination To overcome the electrolyte composition changes and pH changes, an electrolyte recombination unit can be introduced. Such a system can allow the all-iron redox flow cell to operate for a larger number of cycles even with inefficiencies. The net reaction of the recombination reactor is proton generation from molecular hydrogen, and the reduction of iron (III) reduction to iron (II) as per Eq. 8.1. H2 + 2Fe 3+ à 2Fe 2+ + 2H + (8-1) In a single cell reactor (Figure 8-1), combining electrolytes in both the anolyte and catholyte chambers at the end of a discharge step can allow the recombination reaction of iron (III) with hydrogen gas to happen. However, this process requires expensive catalysts like platinum. After the recombination of electrolytes, the composition and pH value of the electrolytes can be the same as the initial values of them and we can split the electrolyte into two chambers to perform 202 charge storage again. A two-chamber reactor separated by an AEM can be another type of electrolyte regeneration unit (Figure 8-2). This unit can be a hydrogen fuel cell with iron (III). Circulation of electrolytes in a flow cell type fuel cell with same pumping units may allow continuous and fast re-modification of the electrolyte composition. Figure 8-1: A single chamber reactor for H 2 and iron(III) recombination A: schematic diagram with all iron redox flow cell, B: Experimental setup of the recombination reactor, C: Reaction mixture before recombination, D: Reaction mixture after recombination and E: Teflon treated, Platinum deposited Reticulated Vitreous Carbon Foam as the catalyst. B C D E A 203 Figure 8-2: A two-chamber reactor for H 2 and iron (III) recombination with all-iron redox flow cell A: schematic diagram and B: Experimental setup of the two-chamber reactor 8.1.2 Lowering HER to 0% Studying different methods to totally suppress hydrogen evolution is required to operate all- iron redox flow cell in economically viable level as platinum-based catalysts are expensive. Different electrolyte additives (Table 8.1) can help to suppress HER. In Chapter Three we discussed the mechanisms of iron deposition. However, investigating competitive surface processes can give a better understanding to suppress HER completely. 204 Table 8.1: Different types of additives used in all-iron redox flow cell in our studies Type of Electrolyte additives Selection criteria Examples Electrolyte side Expected Performance Effect Adsorptive additives Forms an adsorbed layer at the electrode inhibiting hydrogen evolution Pyrazole Cysteine Ascorbic acid Guanidium chloride Plating The increase of the charging efficiency by lowering the surface coverage for HER Kinetic inhibitors High hydrogen overpotential Cadmium chloride Plating The increase of the charging efficiency by increasing HER overpotential Buffering agents Buffering capability at pH lower than 3 Ascorbic acid Plating pH enhancement due to HER can be minimized Chelating ligands Weak binding ability with iron (ii) and Fe(III) Very little shift of the redox potential of the of Fe 2+ /Fe 3+ couple Ionization at pH 3 or below Citric acid Oxalic acid Succinic acid Glycine Malonic acid Formic acid Both Increase the solubility of iron species at elevated pH values by the formation of iron-ligand complexes Ionic strength enhancers Increasing the ionic strength Not lowering the concentration of the redox active species Ammonium Chloride Sodium Chloride Both Enhance the solubility of the electrolyte by shifting the hydroxide precipitation pH to a higher value 205 8.1.3 Other Directions As elevated temperature showed improved kinetics, the operation of the flow cell at higher temperatures is required to translate half-cell results into a full-cell. Implementing the re- balancing schemes and demonstrating continuous operation, addressing voltage losses through improved cell design and developing the stack configuration for a continuous long-term operation are also various other future activities of importance. Introducing novel AEM to prevent the complex transfer or using better sulfate ion transporting AEM and replace the chloride system with sulfate systems can be viable approaches to avoid the formation and transfer of the iron complexes. Combining iron (II)/ iron(III) with different negative side material with no HER can be an extension of the iron flow cells. 8.2 Lithium-Sulfur System Chapter Four provides an analysis of the lithium-sulfur project. Achieving higher utilization, capacity retention, cycle life, rate capability and low-cost are the main challenges. Even though we introduced layers of MCM to improve the utilization, layered structures using different porosity and thickness of membranes for optimum intercalation and polysulfide shuttling barrier should be studied. Mixed conduction membrane modification with KB as a conductive membrane can also be proposed when mixed carbons or AB cathodes are used. Mixed carbon, layered MCM, and “shorted” cathode cells are improved cell configurations for achieving better utilization, improved rate capability, and higher coulombic efficiency. Thus, EIS of the cell with different states of charge and comparison of discharging capacity with different 206 cell constructions can be useful to improve the understanding of how mixed and shorted cells provide better utilization. Even though with thin Ketjen black cathode designs we could see a better utilization, higher impedance of these cells suggest that R charge transfer or R mass transport can be poor due to the lower distribution of the electrolyte in the porous structure. Thus, the pore structure modification using secondary carbon is important, and changes in utilization and rate capability should be evaluated. 8.3 Bio-catalytic Carbon Dioxide Reduction In Chapter Six and Seven, bio-catalytic carbon dioxide conversion using formate dehydrogenase and methyl viologen cofactor has been discussed. However, the mechanism of the carbon dioxide reduction using FDH and MV is still unclear. Thus, future studies can be focused on studying the reaction mechanism. 8.3.1 Evolving the Mechanism of FDH based CO2 reduction The binding of MV free radical to FDH, CO 2 binding to FDH, two-electron transfer, proton transfer, formate generation, MV di-cation release from the FDH binding site, and formate release from FDH active site are different stages in the FDH-catalyzed reduction of CO 2. Exploring these stages as can help in elucidating the reaction mechanism of FDH-catalyzed CO 2 reduction by the MV free radical cation. 207 8.3.1.1 NMR Studies According to the NMR study of MV inclusion into cucurbituril, we can see that the beta proton of MV can be moved to the more higher field due to the addition of electron density with the inclusion. 1 Insertion of MV into p-sulfonatocalix arenes shows shifts of NMR peaks of both protons in MV to the higher field. 2 Thus, we can predict that in the presence of FDH in equivalence quantity the NMR peaks can be shifted to higher fields. However, the chemical shift can be higher or different with MV free radical cation as it can be bound strongly than MV di- cation. These changes can be studied with different amounts of enzyme. 8.3.1.2 Cyclic Voltammetry Cyclic voltammetry studies in the cucurbituril study show a lowered current density for included MV and shifts of reduction potentials. 1 A similar effect is shown in the studies of insertion of MV into p-sulfonatocalix arenes. 2 Similarly, we can study the CV of MV in the presence and absence of FDH to see the binding effect. 8.3.1.3 EPR Studies Debora et al. have studied the photo-induced electron transfer process at the surface of TiO 2 nanoparticles in ethanol using FT-EPR. 3 The rate constant of the photo-induced electron- transfer reaction was derived from the time profile of the FT-EPR spectrum of the MV radical. Further, the same authors established that the electron-transfer process generates chemically- induced dynamic electron polarization by performing FT-EPR measurements on coumarin 343/ MV 2+ in the presence of the TEMPONE(d16) nitroxide radical. 3 Further, Yoon and Kochi studied the superoxide electron transfer with methyl viologen in zeolite using EPR. 4 They have observed 208 a diminished EPR peak of MV radical cation when molecular oxygen is introduced to the system. 4 In our studies, EPR can be used to study MV free radical disappearance with formate generation. 8.3.1.4 Experimental Techniques for Mechanism Elucidation We can attempt to detect intermediates and elucidate the mechanism. Figure 8-3 shows two possible mechanisms. If electron transfer at the active site of FDH occurs in two steps, a bicarbonate-formate radical anion which is an intermediate according to the proposed mechanism 1 (as shown in Figure 8-3A) should be detectable by EPR. Also, 13 C NMR of bicarbonate can be used for intermediate detection. Computational analysis of the stability of reactant and product binding can be used to predict the mechanism. Figure 8-3: Proposed mechanisms of bicarbonate reduction into formate using Methyl viologen radical cations. A: Via a bicarbonate radical intermediate and B: Via no bicarbonate radical intermediates We propose another mechanism as shown in Figure 8-3B without radical intermediates of bicarbonate. To verify this, we need to study the electron transfer capability to the main active site of the FDH from the second methyl viologen. In our manuscript now accepted for the H O H H N N C O HO O N N C O H HO O C H O O + OH - H C O H HO O MV 2+ + H 2 O + + MV 2+ H O H H N N C O HO O N N C O H HO O C H O O + OH - + H 2 O + MV 2+ A B 209 publication in Accounts of Chemical Research, January 2019, we have speculated that the second methyl viologen is bound to a position from where electron transfer may occur via interactions of the pi-stack overlap of aromatic amino acid residues. However, further experimental and theoretical studies are required to verify this mechanism as well. 8.3.2 Ex-Situ Separation of Formate from the CO2 Reduction Reaction Mixture In Chapter Seven, I have discussed that other than to the rate of formate generation, formate accumulation and separation approaches are also important for the industrial production of the FDH catalyzed-CO 2 reduction process. Formate separation from the electrolyte has two objectives namely, (1) Regeneration of the electrolyte and (2) Purification of the formate prior to use as a raw material. Even though we introduce three different reactor configurations for achieving the in-situ separation of the formate, studying the ex-situ separation is also worthwhile from a cost standpoint. As a widely practiced method in separation chemistry, ion exchange columns can be used to separate formate anion and MV cations. The capacity of an ion-exchange column can be increased by increasing the number of functional groups per unit weight of the resin. The equilibrium constant (K A, E) for the reaction of ion exchange (Eq. 8-2) is called the selectivity coefficient (Eq. 8-3). yA m x- + xE r y- D y A r x- + xE m y- (8-2) KA,E= [A r x- ] y [E m y- ] x / [A m x- ] y [E r y- ] x (8-3) 210 Properties of the solute can determine the affinity of different ions. Electro-selectivity is higher with increasing the charge, ions with smaller solvated size show greater binding, and selectivity coefficient increases with the degree of polarizability. These properties should be considered for developing a suitable ion-exchange column or for screening commercially-available ion exchange columns. 8.3.3 Cofactor Immobilization To avoid the separation requirement for the recovery of the cofactor recovery, we can try to immobilize methyl viologen or viologen derivatives and achieve CO 2 reduction. Literature provides various methods of methyl viologen immobilization including electropolymerization, 5 immobilization in the zeolite, 4 in manganese oxide 6 or in carrageenan hydrogel 7 . Further, there are studies of entrapping methyl viologen in Nafion. 8-10 Also designing biosensors using enzyme and methyl viologen co-immobilization has been reported. 5, 11 These methods may be evaluated to facilitate co-factor separation. 211 8.4 Chapter Eight References 1. Kim, H.-J.; Jeon, W. S.; Ko, Y. H.; Kim, K., Inclusion of methylviologen in cucurbit [7] uril. Proceedings of the National Academy of Sciences 2002, 99 (8), 5007-5011. 2. Guo, D.-S.; Wang, L.-H.; Liu, Y., Highly effective binding of methyl viologen dication and its radical cation by p-sulfonatocalix [4, 5] arenes. The Journal of organic chemistry 2007, 72 (20), 7775-7778. 3. Martino, D. M.; van Willigen, H.; Spitler, M. T., FT-EPR Study of Photoinduced Electron Transfer at the Surface of TiO 2 Nanoparticles. The Journal of Physical Chemistry B 1997, 101 (44), 8914-8919. 4. Yoon, K.; Kochi, J., Direct observation of superoxide electron transfer with viologens by immobilization in zeolite. Journal of the American Chemical Society 1988, 110 (19), 6586-6588. 5. Cosnier, S.; Galland, B.; Innocent, C., New electropolymerizable amphiphilic viologens for the immobilization and electrical wiring of a nitrate reductase. Journal of Electroanalytical Chemistry 1997, 433 (1-2), 113-119. 6. Nakayama, M.; Hoyashita, R.; Komatsu, H.; Muneyama, E.; Shoda, K.; Kunishige, A., Immobilization of methylviologen between well-ordered multilayers of manganese oxide during their electrochemical assembly. Langmuir 2007, 23 (6), 3462-3465. 7. Rillema, D. P.; Edwards, A. K.; Perine, S. C.; Crumbliss, A. L., Electrochemistry and photocurrents of the tris (bipyridine) ruthenium (II) and methyl viologen cations immobilized in carrageenan hydrogel. Inorganic Chemistry 1991, 30 (23), 4421-4425. 212 8. Barroso-Fernandez, B.; Theresa Lee-Alvarez, M.; Seliskar, C. J.; Heineman, W. R., Electrochemical behavior of methyl viologen at graphite electrodes modified with Nafion sol– gel composite. Analytica Chimica Acta 1998, 370 (2), 221-230. 9. Krishnan, M.; White, J. R.; Fox, M. A.; Bard, A. J., Integrated chemical systems: photocatalysis at semiconductors incorporated into the polymer (Nafion)/mediator systems. Journal of the American Chemical Society 1983, 105 (23), 7002-7003. 10. Rabani, J.; Behar, D., Radiation-Induced Immobilization of Ruthenium Tris (bipyridine) and Methyl Viologen in Nafion Layers. Ionic Exchange, Mobilities, and Illumination Studies. The Journal of Physical Chemistry 1995, 99 (29), 11531-11536. 11. Da Silva, S.; Shan, D.; Cosnier, S., Improvement of biosensor performances for nitrate determination using a new hydrophilic poly (pyrrole-viologen) film. Sensors and Actuators B: Chemical 2004, 103 (1), 397-402. 0.25 mM to 5mM NAD+ and NADH in 1M NaF using scan rate 250mvs.-1 I Publications and Presentations Journal Articles 1. Jayathilake, B.S.; Bhattacharya, S.; Vaidehi, N.; Narayanan, S. R., Efficient and Selective Electrochemically Driven Enzyme-Catalyzed Reduction of Carbon Dioxide to Formate using Formate Dehydrogenase and an Artificial Cofactor. Accounts of Chemical Research 2019. DOI: 10.1021/acs.accounts.8b00551 2. Jayathilake, B. S.; Plichta, E. J.; Hendrickson, M. A.; Narayanan, S. R., Improvements to the Coulombic Efficiency of the Iron Electrode for an All-Iron Redox-Flow Battery. Journal of The Electrochemical Society 2018, 165 (9), A1630-A1638. Conference Proceedings 1. Buddhinie S. Jayathilake, Billal Zayat, Edward J. Plichta, Mary A. Hendrickson, S R. Narayanan, Optimizing Coulombic Efficiency of All-Iron Redox-Flow Cell (Page 71-74), 48 th Power Sources Conference Proceeding, June 2018 2. Ahamed Irshad, Buddhinie S. Jayathilake, Rodrigo Elizalde-Segovia, Derek Moy, Edward. J. Plichta, Mary. A. Hendrickson and S. R. Narayanan, Lithium-Sulfur Battery with a Robust Sulfur Electrode and Mixed Conduction Membrane, (Page 456-459), 48 th Power Sources Conference Proceeding, June 2018 Conference Presentations 1. B. S. Jayathilake (University of Southern California), E. J. Plichta, M. A. Hendrickson (Army Power Division, RDER-CCA) and S. R. Narayanan (University of Southern California), 48th Power Sources Conference, Denver, CO, USA, June 11, 2018, Paper No. 5.2 2. Ahamed Irshad, Buddhinie S. Jayathilake, Rodrigo Elizalde-Segovia, Derek Moy (University of Southern California), Edward. J. Plichta, Mary. A. Hendrickson (Army Power Division, RDER-CCA) and S. R. Narayanan (University of Southern California), 48th Power Sources Conference, Denver, CO, USA, June 13, 2018, Paper No. 26.5 II 3. B. S. Jayathilake and S. R. Narayanan (University of Southern California), 232nd Meeting of the Electrochemical Society, National Harbor, MD, USA, October 4, 2017, Abstract#1995 4. B. S. Jayathilake, A. Manohar (University of Southern California), E. J. Plichta, M. A. Hendrickson (Army Power Division, RDER-CCA), and S. R. Narayanan (University of Southern California), 231st Biannual Spring Meeting of the Electrochemical Society, New Orleans, LA, USA, May 30, 2017, Abstract #99825 III Appendix Appendix A Supporting Information for Chapter 2 A1. Calculation of the equilibrium potential shift at pH 3 due to iron (II) ascorbate complex formation ! "#$ =! &' − 59.16 2 /012 log2=log6 7 +/019 9= [;< = ] [; ? <]+[;<−] 69= [;< = ][; @ ] [; ? <] ∆! =!B0C−!9D=− EF.GH ? /012 (A-1) IV A2. Calculation of the bulk and electrode surface pH Iron hydrolysis can form iron (II) hydroxide. pH of an iron(II) containing solution (Concentration =C* Fe(II)) can be calculated using the conditional hydrolysis constant (K h ’ ) of iron(II). Fe 2+ + H 2O ⇆ Fe(OH) + + H 3O + (A-2) K h= a Fe(OH)+ *a H3O+/ a Fe2+ (A-3) If C* Fe(II) is the formal concentration of iron (II) chloride, mass balance requires that the sum of the various iron species equals the formal concentration of iron(II) chloride, viz., C* Fe(II)= C Fe(OH)++ C Fe2+ (A-4) Thus, K h= g FeOH+ (C* Fe(II) - C Fe2+ ) *a H3O+/ g Fe2+C Fe2+ (A-5) pH = log g FeOH+ (C* Fe(II) - C Fe2+ ) - log K h g Fe2+C Fe2+ (A-6) Approximating all activity coefficients to be 1 pH = log [(C* Fe(II) - C Fe2+ )/(K h C Fe2+)] (A-7) (C* Fe(II) - C Fe2+ ) would be approximately equal to C H3O+ when K h << C* Fe(II) This assumption is valid because the concentration of iron (II) chloride are in the range of 0.6 to 3.25 M whereas K h is of the order of 10 -6 . Under these conditions, pH = log [(C H3O+)/(K h C Fe2+)] (A-8) or, approximating the numerical values of C H3O+ to a H3O+, pH = - ½ log K h C Fe2+ (A-9) Surface pH under passage of current. When electrochemical deposition of iron undergoes at the electrode surface, according to the First Fick's law, diffusion of the ions controls the material transfer to the electrode surface and the deposition rate. The flux, J Fe(II) of the electroactive substance, Fe(II) toward the cathode surface is proportional to the gradient of the iron(II) concentrations between the bulk (C* Fe(II)) and at the electrode surface (C s (Fe(II)). J (Fe(II)) = D [(C* Fe(II)-C s (Fe(II))/ d] (A-10) V Where D = diffusion coefficient of iron(II) and d = the diffusion layer thickness The flux can be related to the observed current density(I) formed by moving ions as follows; I= J (Fe(II)) n F = [ n F D (C * Fe(II)-C s (Fe(II)) ]/ d (A-11) Where n= number of electrons and F=Faraday constant The surface concentration of ions varies with the current density and faradic efficiency(Ñ). C s Fe(II) = C * Fe(II) –(I α d / n F D) (A-12) From the derivation above, we know the relationship between pH and the concentration iron (II) chloride. Therefore, Surface pH = - (1/2) log (C s Fe(II) K h ) (A-13) At a given current density, the surface pH can be calculated as follows; Surface pH = - (1/2) log [(C * Fe(II) - (Iα d / n F D))K h ] (A-14) VI A3. Calculation of the diffusion layer thickness (δ) Considering a convective boundary layer, the diffusion layer thickness is δ=5.0L MN O# P G/? (A-15) Diffusion layer thickness at low flow velocities calculated using Levich equation δ= R S/T U.H?L VW X P S/Y M ZS/[ (A-16) u o,=flow velocity, v=kinematic viscosity of iron(II) chloride= 7.34 x10 -7 m s -1 , x= distance downstream from the start of the boundary layer, r=radial distance from the rotational axis,, assume r=x=1 cm for this calculation, D=diffusion coefficient=5.5x10 -6 cm 2 s -1 0.00 0.10 0 2 4 6 8 10 12 Diffusion layer thickness, mm 1/[Flow Velocity] 2 , (Flow Velocity in m/ s) VII A4: Ascorbic acid dissociation with pH in 2 M Ammonium chloride medium (■:HL - , ▲H2L) VIII A5: Tafel parameters for HER and iron deposition at different temperatures. Temperature, ºC Metal electrodeposition on iron Hydrogen evolution on iron Tafel slope, mV/ decade Exchange current density, A/ cm 2 Tafel slope, mV/ decade Exchange current density, A/ cm 2 25 113 5.46x10 -7 280 3.13x10 -4 60 81 8.52x10 -8 325 1.71x10 -3 IX Appendix B Supporting Information for Chapter 3 B1: Activity of FeOH + , Fe 2+ , Cl - and Fe-Cl complex calculation • Activity of iron (II) (aFe 2+ ) in the electrolyte can be calculated from the activity of the protons (aH + = aFeOH + ) and hydrolysis constant (K h) at 25 o C K h=4.6*10 -4 ref1 • Activity of iron (II) chloride complexes (aFeCl + ) in the electrolyte can be calculated from the activity of iron(II) and formation constant of Fe-Cl + (K) at 25 o C K=0.69 ref2 • Activity of chloride ions in the electrolyte can be calculated by subtracting the activity of iron (II) chloride complexes from the total chloride (ClT) • Activity coefficient of a given specie x can be calculated as activity coefficient(ac x)= activity(a x)/Concentration where x=Fe 2+ , FeOH + , FeCl + , Cl - pH Iron (II) Activity Iron (II) Chloride Concentration, M 0.6 3.21 0.001 1.0 2.90 0.003 2.0 2.60 0.014 2.5 2.21 0.083 3.0 2.00 0.217 X Matlab code for calculating the activities and activity coefficients of FeOH + , Fe 2+ , Cl - and Fe-Cl + complex % iron(II) total concentration X2=Concentration; %solution pH Y2=pH; %hydrolysis constant of iron(II) Kh=4.6*10.^(-4); % formation constant of Fe-Cl + K=0.69; %proton activity H2=10.^(-Y2); %product of proton activity and iron hydroxide activity H22=H2.^2; %activity of iron(II) aFe2=H22/Kh; X22=X2.^-1; %activity coefficient of iron(II) ac2=[(X22.*aFe2)]; %total Cl - concentration ClT=[2+(2*X2)]; %activity of Fe-Cl + afecl=K.*aFe2.*ClT.*((1+K.*aFe2).^-1); %activity coefficient of Fe-Cl + acfecl=afecl.*(X2.^-1); %activity of Cl - aCl=(ClT-afecl); %activity of Cl - acCl=aCl.*(ClT.^-1); XI B2: Evan Diagram Construction Table B2: Equilibrium potentials (E eq) calculated based on the activity and mixed potential observed at open circuit conditions (OC) (E corr=Corrosion potential) FeCl 2 concentration, M E corr = E OC, V vs. NHE E eq HER, V vs. NHE E eq FED, V vs. NHE 0.6 -0.485 -0.189 -0.531 1.0 -0.470 -0.171 -0.513 2.0 -0.439 -0.153 -0.495 2.5 -0.405 -0.130 -0.472 3.0 -0.365 -0.118 -0.460 Figure B2-1: Evan diagram of iron deposition in 3 M Figure B2-2: Evan diagram of iron deposition in 2.5 M -8 -7 -6 -5 -4 -3 -2 -1 0 -1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 log(Current Density), A/cm 2 Electrode Potential, V vs NHE HOR Fe2+ +2e- --> Fe IoHER EoFED IoFED Fe-->Fe2+ +2e- Ecorr=OCV EoHER HER Icorr HER near Ecorr -8 -7 -6 -5 -4 -3 -2 -1 0 -1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 log(Current Density), A/cm 2 Electrode Potential, V vs NHE IoHER Fe2+ +2e- --> Fe Fe-->Fe2+ +2e- EoFED IoFED Ecorr=OCV EoHER Icorr HER HER near Ecorr XII B3: References used in Appendix B 1. Wells, C. F.; Salam, M. A., Hydrolysis of Ferrous Ions : a Kinetic Method for the Determination of the Fe(II) Species. Nature 1965, 205, 690. 2. Heinrich, C. A.; Seward, T. M., A spectrophotometric study of aqueous iron (II) chloride complexing from 25 to 200°C. Geochimica et Cosmochimica Acta 1990, 54 (8), 2207- 2221. XIII Appendix C Supporting Information for Chapter 4 C1: Capacity fade with initial discharging Figure C1-1: Specific discharging capacity vs. cell voltage in a Li-S cell with MCM Figure C1-2: Specific discharging capacity vs. cell voltage in a Li-S cell with no MCM XIV C2: Significantly higher impedance with less electrolyte Figure C2-1: EIS and open circuit potential with time in KB cathodes with ABA type MCM with low electrolyte volume Losing the Access of Redox Active Intermediate Species leading to poor cycling As PS can move with electrolyte, the contact of PS with the C electrode can be lower. For example, if PS are moved across the separator from the cathode toward MCM (Figure 1 A and B), contact of them with the carbon electrode will be poor. As a result, charge transfer resistance of soluble PS to insoluble PS can be increased. However as soluble PS can be moved back to cathode, at lower discharging rates we can achieve better utilization. Figure A2-2: PS movement away from cathode and loosing contact with electrode 0 2 4 6 8 10 12 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 -12000 -10000 -8000 -6000 -4000 -2000 0 0 5000 10000 15000 20000 25000 Z real , ohm Voltage, V vs. Li+/Li Z img , ohm Number of days day 01 day 02 day 04 day 07 day 08 day 09 day 11 PS PS Cathode Anode MCM PS Cathode Anode MCM with porous layer (A) (B) XV The diffusion of soluble PS across separator can be a function of soluble PS concentration and time. If we operate at higher rates and generate higher concentration of PS, a sudden increase of the viscosity can lead to precipitate PS in the separator-MCM pocket and can increases the Rct. (if precipitation happen near the cathode, we can expect it to undergo the next charge transfer step). Also, at higher rates when PS are formed in higher amounts and moved away from the cathode, remaining PS will undergo the next electron reduction steps. Even though back movement of PS to cathode happens, back diffusion rate might not be faster as the rate of discharging. Therefore, having a shorted cell with a MCM with a highly non-porous layer and minimum electrolyte in cathode side to improve the accessibility of soluble polysulfides to the cathode surface can be a possible solution. Figure A2-3: Access of PS to cathode when stored between MCM and cathode in shorted cells PS PS Cathode Anode MCM XVI C3: Discharging curve of a Li-S cell with a cathode of CNT Figure C3-1: Charging and discharging curves of Li-S cell with CNT carbon support, 70% S loading and current density at 0.1 mA 200 400 600 800 1000 1200 1400 1600 Specific Capacity, mAh/g 1.6 1.8 2 2.2 2.4 2.6 Voltage, V vs Li + /Li Cycling E85 with LCO, 6.72 mg S loading in CNT initial 1 st 2 nd 1 st XVII C4: Discharging capacity recovery by discharging at lower rates Figure C4-1: Discharging rate capability of Acetylene Black cathodes at (A and B) at 0.3 mA and followed by 0.1 mA and (C and D) at 0.4 mA followed by 0.1 mA 1 2 3 4 5 6 7 8 9 10 Cycle number 80 85 90 95 %Coulombic Efficiency 400 500 600 Specific Capacity, mAh/g Cycling E189 with 15 um LCO (1000 pa/cm2, KOH treated, shorted), 7 mg S loading in AB 39 Rate =C/ 1 2 3 4 5 6 7 8 9 10 Cycle number 80 85 90 95 %Coulombic Efficiency 300 400 Specific Capacity, mAh/g Cycling E189 with 15 um LCO (1000 pa/cm2, KOH treated, shorted), 7 mg S loading in AB 29 Rate =C/ (A) (B) (C) (D) XVIII Appendix D Supporting Information for Chapter 6 D1: Thermodynamics of FDH catalyzed CO2 reduction Thermodynamics of FDH catalyzed CO 2 reduction using MV radical cation Redox reaction of CO 2 reduction to formate by MV radical cation gives a free energy of 15.4 kJ/mol (3.68 kcal/mol) at pH 7 (Eq.D-4). CO 2 + 2H + + 2e - à HCOOH E o = -0.19V vs. NHE (D-1) MV 2+ + e - à MV •+ E o = -0.44 V vs. NHE (D-2) CO 2+ H 2O + 2e - à HCOO - + OH - E o’ [pH=7]= -0.52 V vs. NHE (D-3) (D-3)-(D-2) ∆! =! #\ (D−3)− ! # (D−2)=−0.08 V ∆d # =−ef∆! where n= number of electrons and F= Faraday constant CO 2 + H 2O + 2MV •+ D 2MV 2+ + HCOO - + OH - ∆G o’ = 15.4 KJ/mol (3.68 kcal/mol) (4) Equllibrium constant of CO 2 reduction using MV radical cation is 2x10 -3 . ∆d =∆d # +g h /e6 ij kl lℎn !Dop//pqrpoC , ∆d # = −g h /e6 ij 6 ij = [tuvv Z ][7w ?@] Y [xuv?][t y ][7w •@] Y 9l 2986,6 =n = ∆z W {| =2.0x10 =~ Thermodynamics of FDH catalyzed CO 2 reduction using NADH Redox reaction of CO 2 reduction to formate by NADH gives a free energy of 38.6 KJ/mol (9.23 kcal/mol) at pH 7 (Eq.D-6). NAD + + 2e - + H + à NADH E o’ [pH=7]= -0.32 V vs. NHE (D-5) (D-3)-(D-5) ∆! =! #\ (D−3)− ! # (D−5)=−0.20 V ∆d # =−ef∆! CO 2 + NADH D NAD + + HCOO - ∆G o’ = 38.6 KJ/mol (9.23 kcal/mol) (D-6) Equllibrium constant of CO 2 reduction using the natural co-factor, NADH is 1.7 x10 -7 . XIX ∆d =∆d # +g h /e6 ÄÅ kl lℎn !Dop//pqrpoC ,∆d # = −g h /e6 ÄÅ 6 ÄÅ = [tuvv Z ][ÇÉR y ] [xuv?][ÇÉRt] 9l 2986,6 =n Z∆z W {| =1.7Ö10 =Ü Ratio of K MV/K NAD=10,000 Equllibrium potential at our experimental conditions can be represented as per Eq.D-7 1 . ! #\ =! # − RT nF /e [;ãåå = ][; @ ] ? [;ãå ~ = ] ç é\ =−è.èêë−è.èíìîï−è.èñè óéòô [öõúú = ] [öõú ñ = ] ù (û−ê) ! #\ [ü†°H.H,U.G i †¢£ T Z ,U.UU§ i †¢££ Z ] =! # −0.059(pH)−0.030/01ô [;ãåå = ] [;ãå ~ = ] ù ç é\ =−è.èêë−è.ñëì+è.èß®=−è.ß® © ™´ ¨öç XX D2: Molecular dynamics studies Figure D2: Crystal structure of cbFDH bound to NAD + and azide (PDB ID: 5DN9). The docking box used to dock MV •+ is shown in magenta. Docking of methyl viologen, bicarbonate and hydronium ions to FDH MV •+ was docked to the active site of cbFDH bound to NAD + and azide ion (PDB ID: 5DN9) 2 using single precision Glide. 3, 4 The receptor was prepared and minimized using the Protein Preparation Wizard of the Maestro interface (Schrodinger™LLC.). The hydrogen atoms were added, and protonation state of histidine residues were determined using ProtAssign 5 . The center of mass of the bound NAD+ was chosen as the center of the docking box, as shown in Figure D2. The vdW radii of both the receptor and ligand atoms were scaled by 0.8. The partial charges of the ligands MV •+ , H 3O + and HCO 3 - were calculated using quantum mechanical geometry optimization using the Hartree-Fock theory with 6-31G** basis set and a PBF water solvation model in Jaguar 6 . The docked poses of MV •+ were clustered by RMSD and the best pose was selected based on both low glide score and overlap with the bound NAD + in the crystal structure. Next, we replaced the azide ion in the crystal structure with the bicarbonate ion and docked H 3O + to the active site with MV •+ and HCO 3 - present, using the same docking method as MV •+ . For docking H 3O + , the center of the docking box was placed on the AZIDE NAD + XXI bicarbonate ion. The best pose for H 3O + was selected based on proximity to both bicarbonate and MV •+ . Using the docked structure of FDH bound to the first MV •+ molecule, the second MV •+ was docked using a similar protocol as given above. The structure of the product bound FDH was obtained from the reactant bound FDH by editing the reactant molecules in the Builder module of Maestro. As with the reactants, the partial charges of the product molecules, MV 2+ , formate and OH - ions were also calculated using QM geometry optimization in Jaguar. Molecular dynamics simulations of methyl viologen and NADH bound FDH The structures of MV •+ and MV 2+ bound FDH were obtained from the docking calculations described in the previous section. The structures of NADH and NAD + bound FDH were obtained from the crystal structure of NAD + bound FDH (PDB ID: 5DN9). Both formate and bicarbonate ions were placed at the location of the azide molecule in the crystal structure. The protein structures were solvated in explicit water, and sodium and chloride ions were added to neutralize the system charges. The systems were parameterized using the AMBER FF14SB force field for protein 7 , GAFF2 for ligands 8 and the TIP3P water model for solvent 9 . The simulations were performed using the AMBER16 software package on a GPU cluster comprising NVIDIA P100 GPUs. The systems were initially minimized for 2000 steps using conjugate gradient minimization, followed by gradual heating to 310K over 50 ns in the NVT ensemble. Then, a 20 ns equilibration was performed in the NPT ensemble at 310K and a pressure of 1 atm. During these steps, the protein and ligand heavy atoms were restrained with a force constant of 500 kcal/mol. Next, the system was relaxed while slowly reducing the restraining force over 50 ns (NPT ensemble, 310K, 1 atm). Finally, a 50 ns equilibration was performed on each system without any restraints, before initiating the production simulations. Due to the instability of the OH - and formate ions in MV 2+ bound FDH (they leave the protein cavity within 6 and 50ns of starting unrestrained MD), we restrained these ions inside the binding cavity by imposing distance restraints with R258 and MV 2+ using a flat bottom potential. The oxygen atom of OH - was restrained to be within 6.5Å of the sidechain of R258, and 5Å of the nitrogen of the pyridine ring of MV 2+ . The oxygen atom of formate was restrained to be within 7.5Å of the sidechain of R258 and 5Å of the nitrogen of the pyridine ring of MV 2+ . Likewise, in the NADH bound FDH XXII simulations (bicarbonate ion left the protein cavity in unrestrained MD within 30-80 ns), the bicarbonate ion was restrained to be within 5 and 6.5Å of the R258 and H311 sidechains respectively, and 5Å of the nicotine ring of NADH. Below these cutoff distances, the ions did not experience any force and were free to move. The production runs consisted of 3 independent simulations of 50 ns each in the NPT ensemble, starting from randomly selected velocities from the Maxwell-Boltzmann distribution. Clustering of trajectories and calculation of binding free energy The protein ligand complex conformations obtained from the MD simulations were clustered by protein backbone and ligand heavy atom RMSD using the hierarchical clustering routine in CPPTRAJ 10 , with a maximum RMSD cutoff of 1.5Å per cluster. The binding energy values were calculated using the MMPBSA program, which is part of the AMBERTOOLS17 package. XXIII D3: Electrochemical reduction of NAD + using carbon electrodes The first electron reduction step generates a reactive NAD free radical. 11 Two NAD free radicals can combine through the 4 th carbon of the nicotinamide ring and generate an enzyme inactive dimer. This dimerization process is reported to be faster than the next step of protonation and second electron transfer (Scheme 2A). 12 Further, the second electron reduction step can produce different isomers of NADH. Significantly improved reversibility is observed using platinum electrodes .13 However, the use of platinum as the electrode surface for generating NADH for electrochemical CO 2 reduction would not be economically attractive and hence was not pursued. 11, 14, 15 Figure D3-1: Cyclic voltammogram of NAD + (5 mM) in sodium phosphate buffer at pH 6.6 Scheme D3: NAD + reduction pathway -3.E-05 0.E+00 3.E-05 -1.5 -0.5 0.5 1.5 Current, A Potential, V vs. NHE 1 st Reduction step: NAD + + e - à NAD . Dimerization: Fast* 2NAD . à (NAD)2 (enzyme inactive) Protonation and 2 nd Reduction step: Slow NAD . + H + + e - à 1,4- NADH (enzyme active) NAD . + H + + e - à 1,6- NADH (enzyme inactive) XXIV Figure D3-2: NMR of electrochemically reduce NAD + sample (using glassy carbon) acquired at 512 scans using Varian 600MHzs NMR instrument The characteristic peaks for C-4 protons in NADH, which appear near 2.5-2.75 ppm range (AB quartet due to magnetically and chemically nonequivalent proton splitting) are absent in the HNMR spectrum of the electrochemically synthesized sample in this study. 11 14, 15 No NADH peak XXV D4: Detection of MV di-cation and reduced form during bulk electrolysis Figure D4: The concentration of MV in its different redox states followed by scanning voltammetry using a carbon fiber microelectrode (BASi, 11 μm) at a 25 mVS -1 scan rate. -40 -20 0 20 40 -0.7 -0.5 -0.3 Current, mA/ cm 2 Potential, V vs. NHE at 0 min at 174 min at 400 min Decreasing the reduction current of MV 2+ with bulk electrolysis of MV2+ Increasing the oxidation current of MV free radical with bulk electrolysis of MV 2+ XXVI D5: Steady-state reduction of CO2 using an electrochemical cell assembled with an AEM Figure D5: Steady-state reduction MV (40 mM) in the presence of FDH (1.2 uM) and CO 2 -1.2E-03 -8.0E-04 -4.0E-04 0.0E+00 0 1800 3600 5400 7200 Current density at -0.44 V vs. NHE, A/cm 2 Time, s 3.5 mM 5 mM 10 mM XXVII D6: Formate oxidation using FDH and MV di-cation Figure D6: UV/Vis detection of MV •+ generation during formate oxidation with MV 2+ using FDH FDH (3.6 µM) and sodium formate (0.5 M) were added into the phosphate buffer (0.5 M, pH=7, 3 mL). Argon gas was bubbled before adding methyl viologen (20 mM) to the reaction mixture. No changes in the absorption intensity at 606 nm (characteristic peak to MV •+ ) 16 were observed in 6 hrs (Figure D6). XXVIII D7: Steady-state reduction CO2 reduction Figure D7: CO 2 reduction current at 0.44 V vs. NHE (MV 2+ :40 mM, FDH:2.6 uM) and Nafion® 117 membrane; volume of each chamber 7.5 mL). -4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0 24 48 72 96 120 Current Density, mA/cm 2 Time, Hr XXIX D8: Cross over of MV across proton exchange membrane Figure D8-1: MV concentration in cathodic chamber (Chamber B) during continuous electrolysis Figure D8-2: NMR detection of MV di-cation in oxygen evolution chamber (chamber A) Figure D8-3. Proton exchange membrane incorporated with MV radical cations 0 5 10 15 20 25 30 35 40 45 0 10 20 30 40 50 MV di-cation concentration, mM Time, hr MV peaks Imidazole peak XXX D9: Rate constant (kcat) calculation Rate constant or Turnover number of the enzyme (k cat) was calculated as rate per enzyme active site. k cat = Rate / n *[ET] where Rate=Detected formate concentration/ time n= number of active sites per unit of enzyme [ET] = enzyme concentration XXXI D10: References used in Appendix D 1. Pourbaix, M., Atlas of Electrochemical Equilibria in Aqueous Solutions. National Association of Corrosion Engineers: 1974. 2. Guo, Q.; Gakhar, L.; Wickersham, K.; Francis, K.; Vardi-Kilshtain, A.; Major, D. T.; Cheatum, C. M.; Kohen, A., Structural and Kinetic Studies of Formate Dehydrogenase from Candida boidinii. Biochemistry 2016, 55 (19), 2760-71. 3. Friesner, R. A.; Banks, J. L.; Murphy, R. B.; Halgren, T. A.; Klicic, J. J.; Mainz, D. T.; Repasky, M. P.; Knoll, E. H.; Shelley, M.; Perry, J. K.; Shaw, D. E.; Francis, P.; Shenkin, P. S., Glide: a new approach for rapid, accurate docking and scoring. 1. Method and assessment of docking accuracy. J Med Chem 2004, 47 (7), 1739-49. 4. Halgren, T. A.; Murphy, R. B.; Friesner, R. A.; Beard, H. S.; Frye, L. L.; Pollard, W. T.; Banks, J. L., Glide: A new approach for rapid, accurate docking and scoring. 2. Enrichment factors in database screening. Journal of Medicinal Chemistry 2004, 47 (7), 1750-1759. 5. Sastry, G. M.; Adzhigirey, M.; Day, T.; Annabhimoju, R.; Sherman, W., Protein and ligand preparation: parameters, protocols, and influence on virtual screening enrichments. J Comput Aid Mol Des 2013, 27 (3), 221-234. 6. Bochevarov, A. D.; Harder, E.; Hughes, T. F.; Greenwood, J. R.; Braden, D. A.; Philipp, D. M.; Rinaldo, D.; Halls, M. D.; Zhang, J.; Friesner, R. A., Jaguar: A high-performance quantum chemistry software program with strengths in life and materials sciences. Int J Quantum Chem 2013, 113 (18), 2110-2142. 7. Maier, J. A.; Martinez, C.; Kasavajhala, K.; Wickstrom, L.; Hauser, K. E.; Simmerling, C., ff14SB: Improving the Accuracy of Protein Side Chain and Backbone Parameters from ff99SB. J Chem Theory Comput 2015, 11 (8), 3696-3713. 8. Wang, J. M.; Wolf, R. M.; Caldwell, J. W.; Kollman, P. A.; Case, D. A., Development and testing of a general amber force field. J Comput Chem 2004, 25 (9), 1157-1174. 9. Jorgensen, W. L.; Chandrasekhar, J.; Madura, J. D.; Impey, R. W.; Klein, M. L., Comparison of Simple Potential Functions for Simulating Liquid Water. J Chem Phys 1983, 79 (2), 926-935. XXXII 10. Roe, D. R.; Cheatham, T. E., 3rd, PTRAJ and CPPTRAJ: Software for Processing and Analysis of Molecular Dynamics Trajectory Data. J Chem Theory Comput 2013, 9 (7), 3084-95. 11. Aresta, M.; Dibenedetto, A.; Baran, T.; Angelini, A.; Łabuz, P.; Macyk, W., An integrated photocatalytic/enzymatic system for the reduction of CO 2 to methanol in bioglycerol–water. Beilstein Journal of Organic Chemistry 2014, 10, 2556-2565. 12. Ali, I.; Soomro, B.; Omanovic, S., Electrochemical regeneration of NADH on a glassy carbon electrode surface: The influence of electrolysis potential. Electrochemistry communications 2011, 13 (6), 562-565. 13. Yun, S.-E.; Taya, M.; Tone, S., Direct reduction of NAD+ by electrochemical procedure and application of the regenerated NADH to enzyme reaction. Biotechnology Letters 1994, 16 (10), 1053-1058. 14. Tropp, J.; Redfield, A. G., Proton magnetic resonance of NADH in water-methanol mixtures. Conformational change and behavior of exchangeable proton resonances as a function of temperature. Journal of the American Chemical Society 1980, 102 (2), 534-538. 15. Sarma, R. H.; Kaplan, N. O., 220 MHz Nuclear Magnetic Resonance Spectra of Oxidized and Reduced Pyridine Dinucleotides. Journal of Biological Chemistry 1969, 244 (3), 771-774. 16. Watanabe, T.; Honda, K., Measurement of the Extinction Coefficient of the Methyl Viologen Cation Radical and the Efficiency of Its Formation by Semiconductor Photocatalysis. J Phys Chem-Us 1982, 86 (14), 2617-2619. XXXIII Appendix E Supporting Information for Chapter 7 E1: Tokuyama anion exchange membrane properties Property A201 A901 Counter Ion type Prepared as OH Prepared as OH Thickness (Dry)/ μm 28 11 Ion Exchange capacity/ mmol -1 1g -1 1.8 1.7 Electric Resistance/ Ω cm 2 0.3 0.2 XXXIV E2: Different viologen molecules studied in photocatalytic CO2 reduction using FDH Artificial coenzyme Structure Redox potential/ V vs. NHE Enhancement of photocatalytic CO 2 reduction compared to NADH Methyl Viologen 1 N,N' -dimethyl -4,4'- bipyridinium - 0.446 25, 2 20 1 Diquat 2 1,1'-Ethylene-2,2'- bipyridyldiylium dibromide -0.46 126 2 1,1’-dimethyl-2,2’-bipyr- idinium dichloride 2 -0.79 11 2 1,1’-diaminoethyl-4,4’- bipyridinium salt (DA 2+ ) 3 -0.40 560 3 N + N + N + N + Br - Br - N + N + Cl - Cl - N + H 2 N N + NH 2 Br - Br - XXXV E3: 1 H NMR of 2,7-AQDS in D2O and MES buffer Figure E3: 1 H NMR Spectra of CO 2 reduced sample using 2,7-AQDS as a reduced co-factor. Observation of small HNMR peaks of 2,7-AQDS near the chemical shift of formate. 7 . 7 0 7 . 7 5 7 . 8 0 7 . 8 5 7 . 9 0 7 . 9 5 8 . 0 0 8 . 0 5 8 . 1 0 8 . 1 5 8 . 2 0 8 . 2 5 8 . 3 0 8 . 3 5 8 . 4 0 8 . 4 5 8 . 5 0 8 . 5 5 8 . 6 0 8 . 6 5 f 1 ( p p m ) - 2 0 2 4 6 8 1 0 1 2 1 4 1 6 1 8 2 0 2 2 2 4 2 6 2 8 3 0 3 2 3 4 3 6 3 8 4 0 4 2 HNMR Chemical shift of Formate XXXVI E4: References used in Appendix E 1. Noji, T.; Jin, T.; Nango, M.; Kamiya, N.; Amao, Y., CO2 Photoreduction by Formate Dehydrogenase and a Ru-Complex in a Nanoporous Glass Reactor. ACS Applied Materials & Interfaces 2017, 9 (4), 3260-3265. 2. Ikeyama, S.; Abe, R.; Shiotani, S.; Amao, Y., Novel Artificial Coenzyme Based on Reduced Form of Diquat for Formate Dehydrogenase in the Catalytic Conversion of CO 2 to Formic Acid. Chemistry Letters 2016, 45 (8), 907-909. 3. Ikeyama, S.; Amao, Y., Novel Artificial Coenzyme Based on the Viologen Derivative for CO 2 Reduction Biocatalyst Formate Dehydrogenase. Chemistry Letters 2016, 45 (11), 1259- 1261.
Abstract (if available)
Abstract
This thesis focuses on two main topics in electrochemistry: (1) studying electrochemical energy storage systems and (2) developing electrochemical energy conversion pathways by the reduction of carbon dioxide into useful organic products. The first half of Chapter One provides an introduction to electrochemical energy storage technologies. Different categories of aqueous redox flow battery systems suitable for large-scale energy storage are discussed in detail. At the end of the first half of the introduction, a brief introduction to the lithium-sulfur battery is presented. The rest of Chapter One discusses the electrochemical carbon dioxide reduction. ❧ Among the different redox flow battery technologies, the all-iron redox flow battery is an attractive solution for large-scale energy storage because of the low-cost and eco-friendliness of iron-based materials. A major challenge for realizing a continuously operable all-iron redox flow battery is the parasitic evolution of hydrogen at the iron electrode during the charging step. Results and discussion presented in Chapter Two provide insights for minimizing hydrogen evolution in the all-iron battery system. At a given bulk concentration of iron (II), pH of the electrolyte, temperature, concentration of additives, and current density are recognized as key factors affecting the coulombic efficiency of the battery. Elevation of pH near the electrode surface during electrodeposition plays a significant role in hindering hydrogen evolution. Thus, electrolyte flow rates drastically influence the coulombic efficiency of the all-iron redox flow battery. By operating at 60℃ and a pH of 3 with ascorbic acid and ammonium chloride, we could achieve a coulombic efficiency of 98%. This value of coulombic efficiency is among the highest values reported for the iron electrode of the all-iron flow battery. ❧ Chapter Three discusses the kinetics of electrodeposition of iron and the evolution of hydrogen during the charging of the all-iron redox flow cell. We could verify that the kinetics of iron deposition improved with the higher activity of iron (II), lowering the rate of the hydrogen evolution reaction (HER). Further, the increase of the pH near electrode surface results in increased coulombic efficiency at the higher charging current densities. We show that from 0 to 80% state-of-charge and charging at 30 and 40 mA/cm², a steady coulombic efficiency of 98±1% could be observed due to improved kinetics at the higher concentration of iron (II) or increased surface pH at raised current densities. ❧ Chapter Four delivers a critical analysis of studies of the lithium-sulfur battery. The lithium-sulfur battery is a promising technology that has the prospect of doubling the energy density of lithium-ion batteries due to the high specific capacity of the sulfur electrode. Further, sulfur is also an expensive material and has the potential to lead to a cost-effective battery. In the present study, we attempted to understand the use of the USC invented mixed-conduction membrane to address the issue of the polysulfide shuttle. Initial cycling data showed that about 20% improved coulombic efficiency and 11% to 16% improved material utilization with this novel membrane. Further, by using different layers and different porosity of mixed conduction membranes, we attempted to validate the intercalation process of lithium ion to facilitate charge transfer kinetics of the lithium-sulfur cell. Further, the properties of Ketjen Black as a cathode matrix material for improving polysulfide retention and increasing material utilization are discussed. And the results presented in Chapter Four with a mixture of carbon materials point to approaches for achieving better utilization with improved rates of charge and discharge using cathodes based on Ketjen Black. ❧ Chapters Five, Six and Seven discuss the bio-catalytic pathways for the reduction of carbon dioxide to formate using the enzyme formate dehydrogenase and redox co-factors. Electrochemical analysis of NAD⁺/NADH and measurement of enzyme activity is challenging, and Chapter Five presents a simple electrochemical detection of NAD⁺/NADH using an unmodified carbon fiber microelectrode as an accurate, convenient and low-cost detection technique for studying enzymatic reactions. Diffusion-limited current measurements of reduction and oxidation of NAD⁺ and NADH enables investigation of NAD-dependent enzyme activity and indirect analysis of concentrations of substances like formate and ethanol which are substrates for NAD-dependent enzymes. Further, this method was validated for the in-situ detection of electrochemically-generated NADH. This fast, in-situ determination of NAD⁺ and NADH can be used in bio-electrochemical applications and be further developed to a biosensor. ❧ Chapter Six describes results on the bio-catalytic reduction of carbon dioxide using commercially-available formate dehydrogenase enzyme. We demonstrate an efficient and continuous conversion of carbon dioxide to formate using formate dehydrogenase enzyme derived from Candida boidinii yeast and an electrochemically-generated artificial co-factor, methyl viologen radical cation. Continuous regeneration of this artificial cofactor could be achieved at -0.44 V vs. the Normal Hydrogen Electrode (NHE) leading to a significantly lower overpotential for the reduction of carbon dioxide to formate. Thus, the electrochemically-regenerated methyl viologen radical cation works efficiently as a cofactor to support the proton-coupled electron transfer to the carbon dioxide molecule. With a special electrochemical reactor assembled with three chambers separated by anion and cation exchange membranes, we demonstrate the generation and accumulation of formate and the evolution of molecular oxygen. The proton-exchange membrane allows for the accumulation of formate by preventing loss to electrochemical re-oxidation at the oxygen evolution electrode while the anion-conducting membrane improves the utilization of the cofactor by suppressing its loss by cross-over of the oxygen electrode. This novel three-chamber reactor has been shown as a proof-of-concept for efficient and continuous conversion of carbon dioxide into formate. ❧ In Chapter Seven, we discuss various factors in improving the efficiency of the bio-catalytic carbon dioxide reduction including the role of thermodynamics, kinetics, mass transport and pH balance of the system, and in-situ formate separation on the coulombic efficiency. We also discuss the performance improvements resulting from various cell configurations. ❧ Chapter Eight summarizes the future directions for research on all of the projects discussed in the thesis.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Small organic molecules in all-organic redox flow batteries for grid-scale energy storage
PDF
Studies on iron-chloride redox flow battery for large scale energy storage
PDF
Understanding the role of electrode design in determining the electrochemical performance of high-energy/high-power lithium-ion and lithium-sulfur batteries
PDF
Advancing lithium batteries and related electrochemical technologies for a sustainable future
PDF
Design and characterization of flow batteries for large-scale energy storage
PDF
Understanding the factors affecting the performance of iron and nickel electrodes for alkaline nickel-iron batteries
PDF
Nanostructured silicon anode and sulfur cathode for lithium rechargeable batteries
PDF
Selective fluoroalkylation methods and synthesis of water-soluble organic molecules for organic redox flow batteries
PDF
Design and modification of electrocatalysts for use in fuel cells and CO₂ reduction
PDF
Understanding the structure-property relationship in electrode materials for electrochemical energy storage
PDF
Nanostructure design of sulfur cathodes and lithium metal anodes for lithium-ion batteries
Asset Metadata
Creator
Jayathilake, Buddhinie Srimali
(author)
Core Title
Electrochemical pathways for sustainable energy storage and energy conversion
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Chemistry
Publication Date
02/21/2019
Defense Date
01/22/2019
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
all-iron redox flow battery,artificial cofactor,carbon dioxide reduction,coulombic efficiency,efficiency,electrochemistry,energy conversion,energy storage,enzyme,flow batteries,formate,formate dehydrogenase,hydrogen evolution,iron plating,lithium-sulfur,metal deposition,OAI-PMH Harvest,polysulfide shuttle,reactor,reactors,rechargeable battery
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Narayan, Sri (
committee chair
), Prakash, Surya (
committee member
), Shing, Katherine (
committee member
)
Creator Email
bjayathi@usc.edu,bsjayat@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-c89-126636
Unique identifier
UC11675628
Identifier
etd-Jayathilak-7108.pdf (filename),usctheses-c89-126636 (legacy record id)
Legacy Identifier
etd-Jayathilak-7108.pdf
Dmrecord
126636
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Jayathilake, Buddhinie Srimali
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Access Conditions
The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA
Tags
all-iron redox flow battery
artificial cofactor
carbon dioxide reduction
coulombic efficiency
efficiency
electrochemistry
energy conversion
energy storage
enzyme
flow batteries
formate
formate dehydrogenase
hydrogen evolution
iron plating
lithium-sulfur
metal deposition
polysulfide shuttle
reactor
reactors
rechargeable battery