Membranes and Electrodes for Vanadium Redox Flow Batteries

Liuyue Cao

M.E. in Material Science and Engineering

A thesis submitted in fulﬁlment of the requirements for the degree of Doctor of Philosophy

School of Chemical Engineering The University of New South Wales Sydney, Australia

December 2018 PLEASE TYPE THE UNIVERSITY OF NEW SOUTH WALES Thesis/Dissertation Sheet

Surname or Family name: Cao

Firstname: Liuyue Other name/s:

Abbreviation for degree as given in the University calendar: PhD

School: School of Chemical Engineering Faculty: Engineering

Title: Membrane and Electrode Materials for Vanadium Redox Flow Batteries

Abstract 350 words maximum: (PLEASE TYPE)

An Important Issue that has limited the more widespread deployment of the vanadium redox flowbattery (VRFB) technology to date Is the relatively high capital cost. Economic analysis reveals that a high current design with decreased stack size and low--0ost alternative membranes can dramatically reduce the cell cost. To address this, the thesis has (1) evaluated three alternative commercial ion exchange membranes by experimental measurement and mathematic simulation, (2) studied the vanadium redox reaction kinetics on various electrode surfaces to elucidate the effect of surface area and surface functional groups, and (3) developed two approaches to reduce the activation overpotential of the carbon paper electrodes in "zero-gap" cell architecture for high power densities operation. The main results Include: (1) The chemical and mechanical stability, vanadium penneation rates and cell perfonnance of Fumasep� FAP-450 (anion exchange membrane), Fumasep� F930-rfd (cation exchange membrane) and VB2 (cation exchange membrane) were evaluated in the VRFB. Simulations were conducted with experimental data to predict the capacity loss and thennal behaviour of the cells during long-tenn operation. Of the membranes tested, the cation exchange membrane VB2 was found to be the best in tenns of stability, vanadium penneation rates and cell efficiencies. In addition, the simulation results suggest that low and balanced penneations rates between V'·Not and �N02+ ions are beneficial for reduced capacity loss and heat generation. (2) The reaction of the vo2•1 VO2• redox couple with a focus on the effects of electrode surfaceroughness and function groups produced from the pretreatment on the electrochemical behaviour of glassy carbon in vo2•1 VOt solutions was Investigated. Electrochemically scanning to high potentials leads to both surface roughness Increase and oxygen functional group fonnation, but the depression effect of oxygen functional groups on the electrode activity is more dominant. However, electrochemical reduction to negative potentials can reduce the oxygen functional groups and retain the surface area so that the electro-activity of the electrodes can be recovered or even enhanced. Mechanical polishing Introduces more defects which can act as active sites for the vo2•1 VOt redox reaction. (3) Two simple methods to reduce the activation overpotential of thin carbon paper electrodes In the "zero-gap" VRFB architecture were developed. The MoO3 nanosheets, fanned by either directly decorating onto the carbon paper electrode or Indirectly after adding MoO.2· Into the circulating electrolyte, act as active sites for fast electrochemical reactions especially the V2•N3• redox reaction. Both methods Improved the cell perfonnance greatly and achieve similar voltage efficiencies of 85.4%, 81.2% and 78% at current densities of 100, 125 and 150 mAcm·2. Within the tested current density range, the highest discharging power density was around 200mWcm-2 within the voltage range of 0.6-1.7V.

Declaration relating to disposition of project thesis/dissertation

I hereby grant to the University of New South Wales or its agents the right to archive and to make available my thesis or dissertationin whole or in partin the University libraries in all forms of media. now or here after known, subject to the provisions of the Copyright Act 1968. I retain all property rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.

I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstracts International (this is applicable to doctoral theses only).

The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be made in writing. Requests for a longer period of restriction may be considered in exceptional circumstances and re uire the a roval of the Dean of Graduate Research.

FOR OFFICE USE ONLY Date of completion of requirements for Award: Dedicated To My Family ORIGINALITY STATEMENT

‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’ COPYRIGHT STATEMENT

‘I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation. I also authorise University Microfilms to use the 350 word abstract of my thesis in Dissertation Abstract International (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.'

AUTHENTICITY STATEMENT

‘I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.’ Abstract

An important issue that has limited the more widespread deployment of the vanadium redox ﬂow battery(VRFB) technology to date is the relatively high capital cost. Economic analysis reveals that a high power density design with decreased stack size and low-cost alternative membranes can dramatically reduce the cell cost. To address this issue, this thesis has (1) evaluated three alternative commercial ion exchange membranes by experimental measurement and mathematic simulation, (2) studied the vanadium redox reaction kinetics on various electrode surfaces to elucidate the eﬀect of surface area and surface functional groups and (3) developed two approaches to reduce the activation overpotential of the carbon paper electrodes in zero-gap cell architecture for high power density operation. The main results include:

(1) The chemical and mechanical stability, vanadium permeation rates and cell

performance of Fumasepr FAP-450 (anion exchange membrane, 50 µm),

Fumasepr F930-rfd (perﬂuorosulphonic acid cation exchange membrane, 30 µm) and VB2 (perﬂuorosulphonic cation exchange membrane, 100 µm) were evaluated in the VRFB. The stability tests were carried out by immersing

+ the membranes in strong acidic and oxidative VO2 solutions for more than 800 days and measuring the dimensional size and weight at intervals. The three membranes exhibited diﬀerent degrees of swelling and adsorption of the electrolyte. Membrane permeation measurements indicated the following

i Abstract ii

2+ + 2+ 3+ trends for the four vanadium ions: VO >VO2 >V >V for the anion

2+ 3+ + 2+ exchange membrane and VO >V >VO2 >V for the cation exchange membranes. Simulations were conducted with the experimental data to predict the capacity loss and thermal behaviour in the long term. The simulation results suggest that more balanced diﬀusion rates between the

2+ + 3+ 2+ V /VO2 and V /VO couples can diminish the imbalance during the charge and discharge respectively. Moreover, the trends in permeation rates

3+ + 2+ 2+ + 3+ 2+ 2+ such as V

2+ + V and VO2 ions are required. Of all the membranes tested, the cation exchange membrane VB2 (V-Fuel Pty Ltd, Sydney Australia) was found to be the best in terms of chemical and dimensional stability, vanadium permeation rates and cell energy eﬃciencies. In addition, the simulation results show that the VB2 membrane which possesses low and balanced permeation rates

2+ + 3+ 2+ between V /VO2 and V /VO ions has the lowest capacity loss and heat generation from self-discharge reactions.

2+ + (2) The kinetics of the VO /VO2 redox reactions were investigated on glassy

2+ + cabon in VO /VO2 solutions with a focus on the eﬀects of electrode surface roughness and surface function groups produced by electrode pre-treatment

2+ + on the electrochemical behaviour of glassy carbon in VO /VO2 solutions was investigated. Electrochemically scanning the electrode to high potentials leads to the increasing surface roughness and formation of oxygen functional groups. The positive eﬀect of increasing surface area and the depression eﬀect of oxygen functional groups formed at higher potential (>1.3 V vs. Abstract iii

Hg/Hg2SO4) such as C-O-C and C=O on the electrode activity are in a competitive relationship and their dominance is highly dependent on the inherent surface status. Electrochemical reduction by scanning to negative potentials can reduce the oxygen functional groups while retaining the surface area so that the electro-activity of the electrodes can be recovered or even enhanced. In addition, mechanical polishing introduces more defects which

2+ + can act as active sites for the VO /VO2 redox reaction.

(3) Thin carbon paper electrodes with serpentine flow fields were utilized to con- struct a zero-gap architecture for reduced ohmic resistance and low pressure drop. Two approaches were developed to modify the carbon paper electrodes and decrease the activation overpotential. The first method involved thermal

treatment of carbon paper, followed by impregnation in (NH4)6Mo7O24·4H2O solution and calcination. The scanning electron microscopy (SEM) and X-ray diﬀraction (XRD) results conﬁrmed that the treated carbon paper was suc-

cessfully decorated with MoO3 microﬂakes. This modiﬁcation was shown to enhance the electrochemical activity of both half-cell reactions, especially the V2+/V3+ reaction. The charge-discharge cell cycling tests also demonstrated

a great improvement in voltage eﬃciency of the cell using the MoO3 decorated carbon paper electrodes (voltage eﬃciency values averaged at 85.4%, 81.2% and 78% at 100, 125 and 150 mA·cm−2) compared to the cell employing the thermally treated carbon paper (79.7%, 76.2% and 73.8% at 100, 125

−2 and 150 mA·cm ). However, the morphology of the MoO3 on the modiﬁed carbon paper electrodes changed from microﬂakes to nanosheets after cycling

and the amount of MoO3 was signiﬁcantly reduced, which could be attributed to the loss of catalyst by electrolyte ﬂushing and/or the dissolution and re-deposition of the catalysts during cycling. The second method of adding

2− 2− traces of MoO4 into the electrolyte was thus proposed and the MoO4 ions Abstract iv

in the electrolyte were shown to form MoO3 nanosheets on the surface of carbon ﬁbres after charge-discharge cycles, providing active sites for the redox reactions. Both methods improved the cell performance greatly and achieved similar voltage eﬃciencies of 85.4%, 81.2% and 78% at current densities of 100, 125 and 150 mA·cm−2. Within the tested current density range, the highest discharging power density was around 200 mW·cm−2 within the voltage range

of 0.6∼1.7 V. This study not only manifested the catalytic ability of MoO3

2+ + 2+ 3+ for VO /VO2 and V /V redox reactions but also exhibited the great potential of employing electrolyte additive as catalysts for VRFBs. Acknowledgments

Foremost, I would like to express my sincere gratitude to my supervisors Emeritus Professor Maria Skyllas-Kazacos and Dr. Da-Wei Wang for their continuous support, guidance and encouragement in all the time of research and thesis writing. Maria is not only a good mentor, imparting her knowledge and sharing her experience, but also a respectful model in ethics, values and attitudes. Da-Wei is a young passionate researcher who always encourage and support new ideas. I feel extremely privileged to have such great supervisors for my doctoral study. I am also deeply grateful to Prof. Rose Amal and Dr. Yun Hau Ng from Particles and Catalysis Research Group for support in synthesis equipments, Dr. Chris Menictas and Bruce Oliver from Mechanical Engineering for help in manufacturing and modifying the cell, Dr. Bill Bin Gong, Rabeya Akter and Dorothy Yu from Solid State and Elemental Analysis Unit, Mark Wainwright Analytical Centre for assistance in material characterisation, and the staﬀ from School of Chemical Engineering including but not limited to John Starling, Victor Wong, Manadalena Hermawan, Hung Chau, Ik Lau, Robert Chan and Ee Meen Iliﬀe for their all kinds of help during my PhD journey. My colleagues Yifeng Li, Marc-Antoni Goulet, Kefeng Xiao, Yanlansen Cui, Chaoqi Shen, Yitao Yan, Ruigang Wang, Ju Sun, Yunhui Lin, Yuheng Tian, Roong Jien Wong, Cui Ying, Xuelian Tian, Hoi Ying Chung, Salina Jantarang and Chen Deng have been very supportive and some of them made time for me out of their

v Acknowledgments vi hectic schedules when I needed their help. Besides, each of them has a sparkling personality which I can learn a lot from. I would like to take this opportunity to thank each of them, who have made my PhD a wonderful experience. Thanks to the Commonwealth’s contribution and the Australian Government Research Training Program Scholarship. The University International Postgraduate Award has significantly relieved the financial burden for me and made my research life more focused. Last but certainly not least, I would like to thank my family. No matter how tough this doctoral journey is, they are always there and offering enormous love and power to me. List of Publications

Journal Papers

1. L. Cao, M. Skyllas-Kazacos and D. Wang. (2017) “Modiﬁcation Based on

MoO3 as Electrocatalysts for High Power Density Vanadium Redox Flow Batteries,” ChemElectroChem, 4 (8), 1836–1839.

2. L. Cao, A. Kronander, A. Tang, D. Wang and M. Skyllas-Kazacos. (2016) “Membrane Permeability Rates of Vanadium Ions and Their Eﬀects on Tem- perature Variation in Vanadium Redox Batteries,” Energies, 9 (12), 1058.

3. L. Cao, M. Skyllas-Kazacos and D. Wang. (2016) “Eﬀects of Surface Pretreat- ment of Glassy Carbon on the Electrochemical Behavior of V(IV)/V(V) Redox Reaction,” Journal of the Electrochemical Society, 163 (7), A1164–A1174.

4. L. Cao, M. Skyllas-Kazacos and D. Wang. (2015) “Electrode Modiﬁcation and Electrocatalysis for Redox Flow Battery (RFB) Applications,” Energy Storage Science and Technology, 4 (5), 433–457.

5. M. Skyllas-Kazacos, L. Cao, M. Kazacos, N. Kausar and A. Mousa. (2016) “Vanadium Electrolyte Studies for the Vanadium Redox Battery–A Review,” ChemSusChem, 9 (13), 1521–1543.

6. L. Cao, M. Skyllas-Kazacos, C. Menictas and J. Noack. (2018) “A Review of Electrolyte Additives and Impurities in Vanadium Redox Flow Batteries,”

vii List of Publications viii

Journal of Energy Chemistry, 27 (5), 1269–1291.

7. L. Cao, M. Skyllas-Kazacos and D. Wang. (2018) “Solar Redox Flow Batteries: Mechanism, Design, and Measurement,” 2 (8–9), 1800031. Conference

1. L. Cao, A. Kronader, A. Tang, D. Wang and M. Skyllas-Kazacos. (2018) “Vanadium Permeability Rates across the Membrane and Their Thermal Eﬀects on Vanadium Redox Flow Batteries,” (Poster) in The 3rd Future Energy Conference, Sydney, Australia.

2. L. Cao, M. Skyllas-Kazacos and D. Wang. (2017) “MoO3 as Catalysts for High Power Vanadium Redox Flow Batteries,” (Oral Presentation and Poster) in The 8th International Flow Battery Forum, Manchester, United Kingdom.

3. L. Cao, M. Skyllas-Kazacos and D. Wang. (2016) “Activation of Carbonaceous Electrodes for Vanadium Redox Systems,”(Oral Presentation) in The 2nd Future Energy Conference, Sydney, Australia. Table of Contents

Abstracti

Acknowledgmentsv

List of Publications vii

Table of Contents ix

List of Figures xiii

List of Tables xxi

1 Introduction1 1.1 Objectives...... 5 1.2 Thesis Structure...... 5

2 Literature Review7 2.1 Membrane Development...... 7 2.1.1 Membranes Researches at UNSW...... 8 2.1.2 Cation Exchange Membranes...... 14

2.1.2.1 Naﬁonr membrane...... 14 2.1.2.2 Other Perﬂuorinated Cation Exchange Membranes 15 2.1.2.3 Hydrocarbon Based Cation Exchange Membranes. 16

ix Table of Contents x

2.1.3 Anion Exchange Membranes...... 17 2.1.4 Amphoteric Ion Exchange Membranes...... 20 2.1.5 Porous Membranes...... 23 2.1.6 Summary for Membrane Development and Research Gaps. 24 2.2 Electrodes and electrocatalysts...... 25 2.2.1 Material selection...... 26 2.2.2 Electrode modiﬁcation and electrocatalysis...... 27 2.2.2.1 Introduction of oxygen functional groups...... 28 2.2.2.2 Metals and metal oxide deposition...... 33 2.2.2.3 Mesoporous carbon and carbon nanomaterials... 37 2.2.2.4 Electrolyte additives as electrocatalysts...... 48 2.2.2.5 Photocatalysts in all-vanadium photoelectrochemical cell...... 50 2.2.2.6 Structural design...... 51 2.2.2.7 Catalyst-coated membrane...... 55 2.2.3 Summary of Electrodes and Electrocatalysis and Research Gaps...... 56

3 Membrane Evaluation 59 3.1 Problem Statement...... 59 3.2 Experimental...... 65 3.2.1 Electrolyte Preparation...... 65 3.2.2 Stability Test...... 67 3.2.3 Diﬀusion Test and Area Resistance...... 67 3.2.4 Single Cell Setup...... 70 3.2.5 Cycling Test...... 71 3.2.6 Thermal Behavior Studies...... 71 3.3 Results and Discussion...... 72 Table of Contents xi

3.3.1 Membrane Stability...... 72 3.3.2 Permeation Rate Determination of Vanadium Ions...... 76 3.3.3 Cycling Tests...... 94 3.3.4 Thermal Simulation...... 104 3.4 Summary...... 111

4 Vanadium redox kinetics on various electrode surfaces 113 4.1 Problem Statement...... 114 4.2 Experimental...... 125 4.2.1 Working electrodes...... 125 4.2.2 Electrolyte...... 126 4.2.3 Surface treatment...... 126 4.2.4 Electrochemical measurements...... 127 4.2.5 Structural characterization...... 128 4.3 Results and discussion...... 128 4.3.1 Effects of electrochemical treatment on glassy carbon plate electrodes...... 128 4.3.2 Effects of electrochemical treatment on glassy carbon electrodes144 4.3.3 Effect of electrolyte SOC...... 156 4.4 Summary...... 162

5 Electrocatalysts for high power density VRFBs 164 5.1 Problem Statement...... 164 5.2 Experimental...... 176 5.2.1 Electrode preparation...... 176 5.2.2 Electrolyte Preparation...... 177 5.2.3 Structural Characterization...... 177 5.2.4 Electrochemical measurement...... 177 Table of Contents xii

5.3 Results and Discussion...... 179

5.3.1 Synthesis of MoO3-CP...... 179 5.3.2 Electrochemical Measurement...... 182 5.3.3 Single Cell Test...... 185 5.3.4 Structural Characterization after cycling...... 192 5.4 Summary...... 197

6 Conclusions and Recommendations 198 6.1 Summary and Conclusions...... 198 6.2 Recommended Future Work...... 201 6.2.1 Membrane Evaluation...... 201 6.2.2 Reaction kinetics...... 202 6.2.3 Membrane Electrode Assemblies...... 202

Bibliography 204

Appendix A Model Development 245 A.1 Dynamic Model Development...... 245 A.2 Thermal Model Development...... 247 A.3 Sensitivity Analysis...... 254 List of Figures

1.1 Conventional scheme of a VRFB...... 2 1.2 Cost analysis of 1 MW/4 MW·h VRFB system [6]...... 4 1.3 The cost at diﬀerent current densities and membrane prices [7]...4

2.1 Examples of hydrocarbon based cation exchange membranes.... 16 2.2 Scheme for the amphiphilic block polymer membrane [64]..... 17 2.3 Some examples of anion exchange membranes...... 19 2.4 Scheme for the mSPAEK-6F-co-x%BI membrane [81]...... 22 2.5 The catalytic mechanism of C−OH groups toward the positive side reaction [95]...... 28 2.6 The catalytic mechanism of C−OH groups in the negative half-cell reaction [103]...... 30

2.7 The catalytic mechanism schematic diagram of CeO2 [124]..... 37 2.8 A schematic diagram of NPL orientation (FF:ﬂow ﬁeld, mem:membrane) [134]...... 41 2.9 Schematic mechanism of the C=O functional groups towards vanadium redox reactions [140]...... 44

2+ + 2.10 The profound mechanism of carbon nanowalls VO /VO2 reactions 44 2.11 The architecture of zero-gap RFB [176]...... 52 2.12 Scheme of a simpliﬁed electrode assembly [181]...... 54

xiii List of Figures xiv

3.1 Apparatus for determination of permeation rate...... 69 3.2 The membrane surface in (a)static solutions and (b)flowing solutions 69 3.3 (a)Perfluorosulfonic acid, (b) scheme of Nafion membrane structure (The scheme was modified based on Ref [206])...... 73 3.4 Effect of immersion time on the membrane length in 1.6 mol·L−1

+ −1 VO2 in 4.2 mol·L total sulphate/bisulphate solution...... 74 3.5 Eﬀect of immersion time on the membrane width in 1.6 mol·L−1

+ −1 VO2 in 4.2 mol·L total sulphate/bisulphate solution...... 75 3.6 Eﬀect of immersion time on the membrane thickness in 1.6 mol·L−1

+ −1 VO2 in 4.2 mol·L total sulphate/bisulphate solution...... 75 3.7 Eﬀect of immersion time on the membrane weight in 1.6 mol·L−1

+ −1 VO2 in 4.2 mol·L total sulphate/bisulphate solution...... 76 3.8 Concentration of V2+ in the enrichment cell with three membranes. 78 3.9 Concentration of V3+ in the enrichment cell with three membranes. 78 3.10 Concentration of VO2+ in the enrichment cell with three membranes. 79

+ 3.11 Concentration of VO2 in the enrichment cell with three membranes. 79 3.12 Plots of concentration functions vs. time for V2+ in the enrichment cell. The dot line is the fitting curve for the data obtained after 180 min...... 87 3.13 Plots of concentration functions vs. time for V3+ in the enrichment cell. The dot line is the fitting curve for the data obtained after 180 min...... 88 3.14 Plots of concentration functions vs. time for VO2+ in the enrichment cell. The dot line is the fitting curve for the data obtained after 180 min...... 88 List of Figures xv

+ 3.15 Plots of concentration functions vs. time for VO2 in the enrichment cell. The dot line is the ﬁtting curve for the data obtained after 180 min...... 89 3.16 Membrane conductivity vs. permeation rate of (a)V2+, (b)V3+,

2+ + (c)VO and (d)VO2 ...... 94 3.17 A comparison of experimental and simulated charge-discharge curves for the single flow battery with FAP450 membrane...... 96 3.18 A comparison of experimental and simulated charge-discharge curves for the single flow battery with F930 membrane...... 97 3.19 A comparison of experimental and simulated charge-discharge curves for the single flow battery with VB2 membrane...... 97 3.20 Simulated discharge capacity of the VRFBs with FAP450, VB2 and

F930 in this work and Naﬁonr 115 in work [194]...... 99 3.21 Simulated concentration of vanadium ions in the VRFB with FAP450 membrane charged at 1.65V...... 102 3.22 Simulated concentration of vanadium ions in the VRFB with VB2 membrane charged at 1.65V...... 102 3.23 Simulated concentration of vanadium ions in the VRFB with F930 membrane charged at 1.65V...... 103

3.24 Simulated concentration of vanadium ions in the VRFB with Naﬁonr 115 membrane charged at 1.65V...... 103 3.25 The simulated concentration in the middle cell of the VRFB stack

with (a)FAP450, (b)VB2, (c)F930 and (d)Naﬁonr 115 membrane at the ﬁrst day...... 106 3.26 The simulated vanadium concentration in the VRFB tanks with

(a)FAP450, (b)VB2, (c)F930 and (d)Naﬁonr 115 membrane at the ﬁrst day...... 107 List of Figures xvi

3.27 The simulated temperature of the middle cell, tank of the VRFB

with (a)FAP450, (b)VB2, (c)F930 and (d)Naﬁonr 115 membrane at the ﬁrst day...... 108 3.28 The simulated concentration in the middle cell of the VRFB stack

with (a)FAP450, (b)VB2, (c)F930 and (d)Naﬁonr 115 membrane at the ﬁrst week...... 109 3.29 The simulated concentration in the VRFB tanks with (a)FAP450,

(b)VB2, (c)F930 and (d)Naﬁonr 115 membrane at the ﬁrst week.. 110 3.30 The simulated temperature of middle cell, tank of the VRFB with

(a)FAP450, (b)VB2, (c)F930 and (d)Naﬁonr 115 membrane at the ﬁrst week...... 111

2+ + 4.1 The anodic treatment of GCP4000 in 0.5 M VO +0.5 M VO2 in 2.5 M total sulphate/bisulphate solution...... 129

2+ + 4.2 The cathodic treatment of GCP4000 in 0.5 M VO +0.5 M VO2 in 2.5 M total sulphate/bisulphate solution...... 130 4.3 The CV of GCP4000 after anodic treatment in 0.5 M VO2++0.5 M

+ VO2 in 2.5 M total sulphate/bisulphate solution...... 131 4.4 The CV of GCP4000 after cathodic treatment in 0.5 M VO2++0.5 M

+ VO2 in 2.5 M total sulphate/bisulphate solution...... 132 4.5 SEM images of GCP electrodes treated in 1 M V4.5+ in 2.5 M sulfate solution after oxidation by CV scans (a) GCP4000 0.7 V, (b)GCP4000 1.0V, (c) GCP4000 1.3V, (d) GCP4000 1.6 V and reduced after oxidization to 1.6V by CV scans of (e) GCP4000 0.2V, (f) GCP4000 0.1 V, (g) GCP4000 0.4 V, (h) GCP4000 0.7 V and (i) GCP4000 1.0 V...... 135 4.6 Carbon 1s region of GCP4000 after anodic treatment (a) GCP4000 0.7V and (b) GCP4000 1.6V...... 136 List of Figures xvii

4.7 Carbon 1s region of GCP4000 after cathodic treatment (a) GCP4000 0.2V and (b) GCP4000 -1.0V...... 137 4.8 Oxygen 1s region of GCP4000 after anodic treatment (a) GCP4000 0.7V and (b) GCP4000 1.6V...... 138 4.9 Oxygen 1s region of GCP4000 after cathodic treatment (a) GCP4000 0.2V and (b) GCP4000 -1.0V...... 139 4.10 The atom percentage of (a) C1s and (b) O1s in various chemical states on the glassy carbon plate surfaces after CV scan treatments 142 4.11 The scheme of the relationship between the groove depth/pore size and diffuse layer...... 145 4.12 The pretreatment CV scans at 100 mV·s−1 by varying (a) upper potential limits and (b) lower potential limits at glassy carbon electrodes GC600, GC1200 and GC4000...... 146 4.13 The modelled Nyquist impedance plots of GC600 after (a)anodic and (b)cathodic treatment...... 149 4.14 The modelled Nyquist impedance plots of GC1200 after (a)anodic and (b)cathodic treatment...... 150 4.15 The modelled Nyquist impedance plots of GC4000 after (a)anodic and (b)cathodic treatment...... 151 4.16 The estimated charge transfer resistance of GC600, GC1200 and GC4000...... 152 4.17 The fifth scan curve at scan rate of 100 mV·s−1 scanned firstly to 0.7 V and then to 0.2 V on glassy carbon electrodes GC600 in 0.5 M

2+ + VO +0.5 M VO2 in 2.5 M total sulphate/bisulphate solution after pretreatment of CV scans within various potential ranges...... 153 List of Figures xviii

4.18 The ﬁfth scan curve at scan rate of 100 mV·s−1 scanned ﬁrstly to 0.7 V and then to 0.2 V on glassy carbon electrodes GC1200 in 0.5 M

2+ + VO +0.5 M VO2 in 2.5 M total sulphate/bisulphate solution after pretreatment of CV scans within various potential ranges...... 154 4.19 The ﬁfth scan curve at scan rate of 100 mV·s−1 scanned ﬁrstly to 0.7 V and then to 0.2 V on glassy carbon electrodes GC4000 in 0.5 M

2+ + VO +0.5 M VO2 in 2.5 M total sulphate/bisulphate solution after pretreatment of CV scans within various potential ranges...... 155 4.20 The ﬁfth CV curve at scan rate of 100 mV·s-1 scanned ﬁrstly to 1.3 V and then to - 0.3 V on GC4000 electrodes in (a) 1 M VO2+ in 2.5

2+ + M sulphate/bisulphate solution and (b) 0.5 M VO +0.5 MVO2 in 2.5 M total sulphate/bisulphate solution...... 158 4.21 The modelled Nyquist impedance plots of GC4000 after CV scans

2+ 2+ + in (a) 1 M VO in 2.5 M sulfate solution and (b) 1M VO /VO2 in 2.5 M sulfate solution...... 160 4.22 (a) The ﬁfth CV scan curves at scan rate of 100 mV·s-1 scanned ﬁrstly to 1.3 V and then to −0.3 V and (b) the corresponding modeled Nyquist plots in 1 M VO2+ in 2.5 M sulfate solution on glassy carbon electrodes...... 161

2− 5.1 Synthesis methods of MoO3-CP and samples with CP with MoO4 electrolyte additive...... 176 5.2 Components of modiﬁed serpentine ﬂow cell...... 178 5.3 SEM images of pristine CP...... 180 5.4 TGA results of pristine CP...... 181 5.5 SEM images of 650-4h-CP...... 182

5.6 SEM images of MoO3-CP...... 183 List of Figures xix

5.7 XRD patterns of the pristine CP, 650-4h-CP, MoO3-CP and PDF card 05-0508 and 35-0609...... 184

5.8 The ﬁfth CV curves of the pristine CP, 650-4h-CP, MoO3-CP and 2− 2+ 650 -4h-CP with MoO4 additive in 1 M VO in 2.62 M total sulphate solution...... 186

2+ 5.9 Five CV cycles of MoO3-CP in 1 M VO in 2.62 M total sulphate solution...... 187

5.10 The Nyquist plots of the 650-4h-CP, MoO3-CP and 650-4h-CP

2− 2+ with MoO4 additive in 1 M VO in 2.62 M total sulphate solution 187 5.11 The Nyquist plots of the single serpentine cell with 650-4h-CP, 2− MoO3-CP and 650 -4h-CP with MoO4 additive after charging to 1.8V...... 188 5.12 The ﬁrst charge curves with 650-4h-CP at constant current density of 100, 125, 150 mA·cm−2 ...... 189

5.13 The first charge curves with MoO3-CP at constant current density of 100, 125, 150 mA·cm−2 ...... 190 2− 5.14 The first charge curves with 650 -4h-CP with MoO4 additive at constant current density of 100, 125, 150 mA·cm−2 ...... 190 5.15 The voltage efficiencies of single serpentine cell with 650-4h-CP, 2− MoO3-CP and 650 -4h-CP with MoO4 additive at constant current density of 100, 125, 150, and 100 mA·cm−2 ...... 191 5.16 The peak power densities of single serpentine cell with 650-4h- 2− CP, MoO3-CP and 650 -4h-CP with MoO4 additive at constant current density of 100, 125, 150 mA·cm−2 ...... 192 2− 5.17 SEM images of (a)MoO3-CP and (b)650 -4h-CP with MoO4 additive after the single cell test...... 193 List of Figures xx

2− 5.18 EDS images of (a) MoO3-CP and (b) 650 -4h-CP with MoO4 additive after the single cell test...... 194

5.19 Mo spectrum XPS analysis of (a)MoO3-CP and (b)650-4h-CP

2− with MoO4 additive after the single cell test...... 196

A.1 A comparison of simulated charge-discharge curves by changing different diffusion coefficients of FAP450 membrane...... 256 A.2 A comparison of simulated discharge capacity by changing different diffusion coefficients of FAP450 membrane...... 257 A.3 The simulated concentrations of the species in the tanks with dif-

fusion coeﬃcients of (a)FAP450, (b)FAP450-2k2, (c)FAP450-2k3,

(d)FAP450-2k4 and (e)FAP450-2k5 at the ﬁrst week...... 258 A.4 The simulated temperature of the VRFB systems with diﬀusion coef-

ﬁcients of (a) FAP450, (b) FAP450-2k2, (c)FAP450-2k3, (d)FAP450-

2k4 and (e)FAP450-2k5 at the ﬁrst week...... 259 List of Tables

3.1 Concentration profiles in the diffusion test of FAP450 membrane.. 80 3.2 Concentration profiles in the diffusion test of F930 membrane... 82 3.3 Concentration profiles in the diffusion test of VB2 membrane.... 84 3.4 Thickness L (µm) and diffusion coefficients k (dm2·s-1) for different vanadium ions across membranes...... 90 3.5 Area resistance and ionic conductivity of membranes...... 93 3.6 Experimental data from cycling tests. CE: coulombic efficiency; VE: voltage efficiency; and EE: energy efficiency...... 96 3.7 Specification of the VRFB in the dynamic model...... 98 3.8 Parameters in the thermal simulation model...... 105

2+ + 4.1 Published kinetic studies for VO /VO2 redox reaction on carbonaceous electrodes...... 115 4.2 Published kinetic studies for V2+/V3+ redox reaction on carbonaceous electrodes...... 121 4.3 Binding energy (BE) values (eV) of C1s for various pretreated glassy carbon plates (atom percentage values given in parentheses).... 140 4.4 Binding energy (BE) values (eV) of O1s for various pretreated glassy carbon plates (atom percentage values given in parentheses).... 140 4.5 Atom ratio of O/C for various pretreated glassy carbon plates... 140

xxi List of Tables xxii

5.1 Summaries of metal- or metal oxide- based electrocatalysts for VRFBs (A: Electrode area (cm-2), f: Flow rate (mLmin-1), j: Current density (mA·cm-2), CE, VE and EE: (%))...... 169

0 2 −1 A.1 Prefactors kx (m ·s ) of vanadium ion x across the membrane... 254 A.2 Diﬀusion coeﬃcients (dm2 · s−1) for simulations...... 255 Chapter 1

Introduction

The widespread implementation of renewable energy is becoming an increasingly important issue around the world. Because of the intrinsic intermittent and unpredictable properties of renewable energy resources such as wind and solar however, suitable energy storage and supply systems are needed to make the grid more robust and reliable. Electrochemical energy storage systems such as batteries and fuel cells are especially suited to the storage of electrical energy since they involve the direct conversion of chemical into electrical energy, allowing very high energy efficiencies to be achieved. Amongst all of the available battery technologies, Redox Flow Batteries (RFBs) allow the greatest flexibility and versatility and have therefore attracted the most interest in recent years. The RFB is a kind of secondary battery system in which energy is stored in liquids contained in external electrolyte reservoirs with inert electrodes that only supply the reaction sites for the redox couple reactions. This unique configuration enables the system to have the power rating dependent on the size of reactors while the energy capacity relies on the size of tanks and the concentration of dissolved active materials, thereby achieving the separation of energy and power which is difficulty in enclosed solid-state battery

1 1. Introduction 2 systems. RFBs thus oﬀer many advantages over conventional batteries, including independent sizing of power and capacity, simple battery structure, long cycle life and low capital and maintenance costs. The development of RFBs started in the 1970s with the NASA project on Fe/Cr RFB in HCl systems [1]. Numerous redox couples have since been studied in both aqueous and non-aqueous systems [2–4]. Among all the RFB systems, vanadium redox ﬂow batteries (VRFBs) have gained the most extensive commercializations because the single active element strategy in the two half-cells can avoid the cross- contamination problems of other RFB systems. The schematic of a VRFB is shown

2+ + in Figure 1.1. This system utilizes the redox reactions between the VO /VO2 and V2+/V3+ couples as the positive and negative half-cell reactions. Therefore, even if the vanadium ions travel across the membrane to the other side of the cell, there is no cross-contamination of the electrolytes and the system can always be rebalanced by re-mixing and re-charging without sacriﬁcing any capacity. Capacity loss associated with side reactions such as H2 or O2 evolution can also be restored by chemical or electrochemical rebalancing, thereby enabling very long cycle life.

Figure 1.1: Conventional scheme of a VRFB

According to the DOE Global Energy Storage Database [5], around the world, fifty VRFB projects at >5 kW scale are under contract, construction or operation (246.5 MW rate power in total) and eight VRFB projects have been announced that 1. Introduction 3 will contribute to an extra 2.7 MW in total power rating. The largest of all VRFB installations is the 200 MW/800 MWh battery currently being constructed by Rongke Power and Prudent Energy in Zhangbei, Hebei, China (Zhangbei National Wind and Solar Energy Storage and Transmission Demonstration Project V). Even so, many investors and industrial groups are still watching and waiting to see how the VRFB will compare against the Li-ion technology that is currently being implemented in many large-scale energy storage installations around the world. The slow uptake of the technology has not been due to the relatively low energy density of VRFBs (20∼30 W·h·kg−1 or 30∼40 W·h·L−1) as this is not an obstacle for stationary applications where size and weight are not an issue. The main hindryance to date is the relatively high capital cost and the uncertainty of potential economic benefits in many grid-scale electricity markets where power prices have been quite low. The same problem has also impacted on the up-take of the Li-ion battery and other energy storage technologies in most parts of the world where electricity prices continue to remain relatively low. Considerable effort is therefore continuing to reduce the cost of these technologies so that they can compete against conventional power generation. As shown in Figure 1.2[6], in a 1 MW/4 MW ·h VRFB system, membranes occupy the second largest percent in the total capital cost. Furthermore, according to the cost analysis of Zhang and coworkers [7] (Figure 1.3), it can be concluded that the reduction of membrane price can dramatically decrease the capital cost while at higher current densities, a lower capital cost can be obtained with a fixed price membrane. Therefore, in order to magnify the economic appeal of the VRFB technology, efforts should be made to identify cheaper, high performance membranes and to operate at higher current densities by improving the power density of the VRFB system. 1. Introduction 4

Figure 1.2: Cost analysis of 1 MW/4 MW·h VRFB system [6]

Figure 1.3: The cost at diﬀerent current densities and membrane prices [7] 1. Introduction 5

1.1 Objectives

Despite the advantages associated with the elimination of cross-contamination in VRFBs, the wider commercialization of the VRFBs in large energy storage, particularly in conjunction with renewable energy sources, can only be accomplished with reduced capital cost and improved power density. Therefore, this thesis focuses on:

(1) Evaluate low-cost commercial membranes with good properties for VRFB systems;

(2) Investigate the relationship between electrode surface properties and the kinetics of the vanadium redox couple reactions.

(3) Investigate and develop electrode modiﬁcation methods to improve the power density of the VRFB.

The three objectives are accomplished in three separate chapters in this thesis. The ﬁrst objective is discussed in Chapter3. The second one is addressed in Chapter4 and the last goal is achieved in Chapter5.

1.2 Thesis Structure

Based on the set goals, the structure for the rest of the thesis is organized as follows: In Chapter2, a literature review of membranes, electrodes and electrocatalysis is presented. At the end of each review, the current research status is summarized and the gaps are highlighted. In Chapter3, three commercial membranes are systematically evaluated in terms of stability, permeability, cell efficiencies and thermal behavior prediction.In order 1. Introduction 6 to achieve more reliable experimental data, the existing methods for determining the permeability rates of different vanadium ions are modified. The obtained results are then applied in a simulation model to predict the thermal behavior of VRFBs with different membrane under various scenarios.

2+ + In Chapter4, the electrochemical behaviour of the VO /VO2 redox couple at various electrode surfaces is investigated for better understanding of the eﬀect of electrode surface properties on the redox kinetics. The electrode surfaces are varied by physical polishing and/or electrochemical treatment and the electrochemical

2+ + response of the VO /VO2 redox reactions are recorded by cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS). The morphologies and elemental analysis of the electrode surface are utilized to further understand the relationship between the electrode surface status and the kinetics of the redox reactions.

In Chapter5, a novel MoO 3-based modification approach is proposed for high power density systems. A small flow cell with serpentine-type flow fields and thin carbon paper electrodes is employed to reduce the ohmic resistance and enhance the mass transfer. Decorating the thermal treated carbon paper with MoO3 nanoparticles and adding MoO3-based additive into the electrolyte are utilized to improve the cell performance at high current densities. The thesis is concluded in Chapter6 with a summary of key results and contributions. In addition, a discussion of some possible extensions to this work is presented. Chapter 2

Literature Review

This chapter provides a brief review of the literature on membrane and electrode development for VRFB applications. Section 2.1 describes the development of membranes in VRFB applications over the past thirty years. Section 2.2 reviews the progress of electrode materials and electrocatalysis with the main content reproduced from a review paper by the author in the Energy Storage Science and Technology journal. The research gaps for each part are summarized at the end of each section.

2.1 Membrane Development

As one of the most important components in VRFBs, the membrane not only acts as the separator of the catholyte and anolyte but also allows the transfer of the charge-balancing ions to complete the circuit. An ideal membrane for VRFBs should exhibit good stability in strong acidic and oxidative solutions, small area resistance, high ionic conductivity, low permeability of vanadium ions, negligible water transfer and low price. Membrane for the VRFB has been studied and developed for more than three

7 2. Literature Review 8 decades. In the first ten years, most of the screening and research work was carried out by Skyllas-Kazacos and co-workers who originally pioneered the VRFB at UNSW Sydney. By the late 1990s other researchers began to notice this field and began new research programs that have made further numerous contributions. Hence, the initial membrane researches at UNSW will be firstly introduced. The subsequent development will be elaborated based on different membrane types in the following sections.

2.1.1 Membranes Researches at UNSW

In the first all-vanadium redox flow cell prototype built by Skyllas-Kazacos and co-workers [8], a sulphonated polyethylene membrane was used. The average coulombic efficiency (CE) was shown to be more than 90% at 3 mA·cm−2 when the electolyte was 0.1 M vanadium in 2 M H2SO4. However, after maintaining a constant open-circuit voltage (OCV) after full charging for more than 70 h, the membrane gradually deteriorated with changes in colour changing and increase in resistivity. Despite this, the possibility of ion exchange membrane in VRFBs was successfully demonstrated. Later, a polystyrene sulfonic acid cation exchange membrane [9] was used in a cell employing 1.5 M vanadium solutions. With the improved cell design, the charging current density could be increased to 40 mA·cm−2 while the polystyrene sulfonic acid cation exchange membrane allowed a CE of 90% and voltage efficiency(VE) of 81% to be achieved in the SOC range of 10∼90%. It can be thus concluded that the membrane material played a significant role in the battery performance and cycle life. A series of membrane evaluation and modification work was therefore initiated to understand the behaviour of the membrane in the VRFB and to identify the critical properties required for VRFB applications. A range of new experimental procedures had to be developed 2. Literature Review 9 to characterize different membranes and separator materials. First, a diffusion test [10] was designed to determine the vanadium ion permeability rates across the membrane. This involved placing a blank solution of Na2SO4 in H2SO4 solution on one side of the membrane and a vanadium (in specific valance) in H2SO4 solution on the other side of a device similar to a VRFB cell. By recording the concentration of vanadium ions in the Na2SO4 solution at different times and incorporating them into the Fick’s First Law of Diffusion, they could obtain the permeability rates of vanadium ions quantitatively. The membrane resistance measurement tests involved the design of a device in which the membrane was sandwiched between two-half cells containing 1 M VO2+ solution and graphite electrodes. With these tests, they [10] evaluated a range of separators and membranes including microporous PVC separator, ultramicroporous filter membrane Duraporer GMV, cation exchange membranes including RAIr, Nafionr N423, Nafionr N324, Gellmanr NFWA, and Selmionr CMV and ion-selective membrane Aquaporer. Except for the

Aquaporer membrane which could not be wetted, the two Nafionr membranes exhibited the lowest VO2+ ion permeabilities while Gellmanr had the highest (It has to be mentioned that the comparison of permeability data among different reports should be treated with caution as different measurement or calculation methods were applied, as discussed in Section 3.1. Furthermore, temperature is rarely mentioned and this can have a significant effect on measured diffusion coefficients). In addition, during the diffusion test, PVC and GMV experienced a dramatic volumetric transfer and RAIr and NFWA membranes were found not stable in the vanadium electrolyte. Finally, Selmionr CMV was identified as the most suitable membrane candidate due to its relatively low area resistance compared with the other membranes(1.9 Ω·cm2 in the begining and 1.6 Ω·cm2 after three month use).

This Selmionr CMV was subsequently adopted in a 1kW VRFB project [11]. 2. Literature Review 10

The CE and VE could reach 98.4% and 82.5% respectively at the current density of 40 mA·cm−2 and the membrane performed consistently during the 100 charge- discharge cycles for several months. However, discrepancy can be found in the report by Ang from the same group [12] that an increasing of CMV membrane resisvity from 2.9 to 4 Ω·cm2 after 30 days. Various factors can be contributed to the deteriorated performance and additional fouling caused by other external conditions such as elevated temperature and evaporated solution by N2 bubbling might be one of them. Many screened membranes were observed to deteriorate after extended cycling in the VRFB and this was attributed to the oxidation of the

+ membrane material by the highly corrosive VO2 species in the charged positive half-cell electrolyte, giving rise to the formation of the reduced VO2+ species in the solution. The rate of VO2+ formation could thus be used as an indicator for membrane degradation. In order to evaluate the long-term stability of the membranes in vanadium electrolyte more eﬀectively, the group developed several techniques [13,14] including

+ (i) soaking the preweighed membrane in 0.1 M VO2 solution and then determining the concentration of the reduced VO2+ ions in the solution by ultraviolet absorption

+ spectrometry periodically and (ii) soaking the membrane in 2 M VO2 solution and measuring the area resistance, VO2+ ion permeability and ion exchange capacities at intervals. By using these methods they conﬁrmed the stability of two anion exchange membranes, Selemionr AMV and Selemionr Type-II. Although Naﬁonr

+ was found to have the best long-term chemical stability in the VO2 solution, its high cost was thought to be prohibitive for commercial application in the VRFB.

Furthermore, Nafionr also exhibited a high degree of swelling when soaked in the vanadium electrolyte. This not only led to high water transfer during operation, but also created significant problems during the assembly of large cell stacks. Nafionr was therefore set aside in preference for other membranes and separators. 2. Literature Review 11

During their screening work, the UNSW group found that when using ion- exchange membranes in the cell, they could observe volumetric transfer from one half-cell to the other during the operation of the VRFBs and the transfer direction highly depended on the type of membrane [15–18]. For anion exchange membranes, the net electrolyte transfer was from the positive into the negative half-cell, while for cation exchange membranes, the net transfer was from negative to positive. They [16] discovered that the hydrogen ion concentrations gradient was the main reason for water transfer in the case of cation exchange membrane and the sulphate ion concentration across the membrane influenced the transfer in the anion exchange membrane case. In an operating cell, the variation of ionic strengths at different SOCs will lead to osmotic water transfer across the membrane; the permeation of vanadium ions due to the concentration gradients and electric field would also carry the hydrated water across the membrane [17]. These processes were found to be related to the ion exchange capacity of membranes. So different membrane modification methods were developed to reduce water transfer during cell operation. Methods such as sulfonation of anion exchange membranes [15] and pretreatment in polyelectrolytes [18] were shown to improve the water transfer properties of different membranes. Having observed that the direction of solution transfer is the opposite for anion and cation exchange membranes, Chieng [13] proposed the use of both types of membrane in alternate cell in a multi-cell stack to offset the ion and water transfer processes and reduce the net solution transfer. With a Selemionr AMV and a

Flemionr (Asahi Glass Co. Japan) membranes in a 2-cell stack, only slight net volumetric transfer to the positive half-cell was observed over around 80 cycles (360 operation hours). The UNSW group [13] also choose to apply an alternative approach that involved incorporating both anion and cation exchange groups into the pores of the same membrane in order to balance the diﬀerent ion exchange 2. Literature Review 12 processes and minimize water transfer in the cell.

Apart from ion exchange membranes, Daramicr attracted their attention due to its low cost and good chemical stability. Like most traditional microporous separators, Daramicr does not possess ion-exchange capabilities. Furthermore, the relatively large pore size cannot block the crossover of vanadium ions. Therefore, modifications were needed to reduce the pore size or improve the ion selectivity. Ang et al [12] studied the effect of three different modification methods for

Daramicr on the VRFB performance. First they applied the Grimes and Bellow method [19] by adding polyelectrolyte with high molecular weight such as sodium polystyrenesulphonate (PSSS) into the electrolyte as ionic barrier to select the ions. Such modiﬁcation improved the CE by around 4∼5% and the VE by around

2%. They also tried soaking Daramicr microporous separators in divinybenzene monomer in methanol solution and boiling in sodium persulphate solutions. How- ever, this method failed to show any significant enhancement. Their third attempt was to immerse the separators in a mixed solution of divinylbenzene, ether and boron trifluoride etherate and the modified membrane gave a increase of 7.5% in CE and VE. Furthermore, Ang [12] discovered that the severe solvent transfer in the system with microporous separators was mainly due to the osmotic pressure difference between the catholyte and anolyte during operation and it could be diminished by adding an appropriate amount of sulphuric acid to adjust the ionic strength. Chieng and co-workers [20] also tried various methods and found soaking the membrane in Amberlite in alcohol solution and crosslinking with divinyl benzene (DVB) can deliver a CE of 94% and VE of 86% at the current density of 40 mA·cm−2. They further optimized the recipe for higher selectivity and lower area resistance [21]. It was believed that the ion exchange resin can enter into the porous structure to decrease the pore size. However, via detailed characterization including resitivity, 2. Literature Review 13 diffusivity, scanning electron microscopy(SEM), thermo gravimetric analysis(TGA), pore size distribution, ion exchange capacity(IEC), Infrared Spectroscopy (IR), and 13C nuclear magnetic resonance(NMR), Mohammadi and co-workers [22] found that the size of Amberlite was much larger than the pore size and the ion exchange capability did not vary much after the modification. The presence of Amberlite might just catalyze the polymerisaztion but the polymerized DVB did form a network within the microposous structure and reduced the pore size. Soon they [23] found that sulfonating the DVB-crosslinked Daramic can diminish the pore size significantly to 0.02 µm from 0.1 µm and cation exchange groups were also sucessfully incorporated. Another approach proposed by the group [24] was to incorporate PSSS into Daramicr and then crosslink with DVB. This method was shown to alleviate the solvent transfer and capacity loss issues. The area resistivity were also decreased to 1.36 Ω·cm2 and the energy efficiency (EE) reached 77% at 40 mA·cm−2. The group attempted to scale up this Daramic modification technique for the production of larger sheets for use in multi-kW stacks, however, the non-uniformity of the manufactured Daramicr sheets proved to be a major problem. By the mid-1990s, other lower cost commercial membranes became available, so the team’s efforts returned to the further screening and characterization of these new materials. By the early 2000s, several membrane research groups around the world turned their attention to the VRFB and to the development of novel membranes specifically for use in the VRFB. Despite the high cost of Nafionr, many groups chose to apply membrane modification to Nafionr membranes in order to improve their properties in the VRFB. These research activities will be discussed in the following sections. 2. Literature Review 14

2.1.2 Cation Exchange Membranes

2.1.2.1 Naﬁonr membrane

Nafionr membrane is a polytetrafluoroethylene(PTFE)/perfluorosulfonic acid copolymer licensed by DuPont Company. The tetrafluoroethylene backbone render the membrane high chemical stability while the acidic sulfonic groups endow it excellent cation conductivity. Due to its superior properties, many VRFB projects employ the Nafionr membrane, such as the 1 kW and 10 kW VRFB by Dalian Institute of Chemical Physics (DICP) [25], the 1 kW VRFB with mixed acid by Pacific Northwest National Laboratory (PNNL) and UniEnergy Technologies [26], etc. However, apart from the high cost, swelling and permeation issues haunt

Nafionr so that many modification studies were carried out. Similar to microporous membrane modification, the methods used for Nafion involved either blocking the pores or enhance the ion selectivity with charged species. For example, coating

Nafionr with polyelectrolyte [27–29] and multi-layered species with alternate charges [30] were proven to be useful in confining the swelling behavior and decreasing of the vanadium diffusion across the membrane, which also agrees with the Donnan effect. Inorganic oxides materials such as SiO2 [31–35], TiO2 [36–38], graphene oxide [39–41], etc. were also found to be able to reduce the vanadium diffusion across the membrane and solvent transfer. Nevertheless, in an investigation of the Nafion/SiO2 membrane [42], PNNL did not recommend the SiO2 modification as a long-term solution. They found that SiO2 particles interacted with the sulfonic functional group of the Nafionr membrane and the vanadium ions via hydrogen bonds. Although the competition with vanadium ions for the sulfonic acid groups hindered the crossover of the vanadium ions, the SiO2 condensed after long operation under strong acidic environment, which reduced the interaction between SiO2 and 2. Literature Review 15

-SO3- groups, disabling the modification function. Simple polymer blending with polymers such as polyvinyidene fluoride (PVDF) [43] was studied as well and showed drastic improvement in the water transfer and ion selectivity. It should be noted that some of the modifications need to be carefully evaluated due to the possibility of increased area resistance [28]. Furthermore, as PNNL [42] pointed out, the long term effects should be also taken into consideration to avoid any performance decay.

2.1.2.2 Other Perﬂuorinated Cation Exchange Membranes

Due to the chemical stability requirement, perfluorinated membranes are normally preferred. Given the high price of Nafionr membrane, an other fluorinated polymer, PVDF appears to be an interesting alternative. Luo and co-workers [44] utilized a solution-grafting method to prepare the PVDF-graft-poly(styrene sulfonic acid) (PVDF-g-PSSA) membrane for the VRFBs. The permeation rates of V3+, VO2+

+ and VO2 ions of the newly developed membranes were significantly lower than that of Nafionr 117 membrane. The CE and EE of the charge-discharge tests also exhibited higher values than Nafionr 117 in the current density range of 10 to 60 mA·cm−2. No obvious capacity decay was observed in the cell with the PVDF-g-PSSA-22 membrane for over 200 cycles. Qiu et al. [45] also grafted styrene onto PVDF to introduce the ion exchange function, but they added maleic anhydride (MAn) to copolymerize with styrene to reduce the radiation damage during grafting. The obtained PVDF-g-PSSA-co-PMAc membrane demonstrated

3+ 2+ + much lower V , VO and VO2 ion permeability as well as better self-discharge performance. 2. Literature Review 16

2.1.2.3 Hydrocarbon Based Cation Exchange Membranes

Various sulfonated aromatic polymers are also investigated as backbone materials for VRFB membranes including poly(arylene thioether) [46], poly(arylene ether sulfone) [47], poly(arylene ether ketone)s (SPAEK) [48], poly(imide)s (SPI) [38, 49–51], ﬂuorinated poly(arylene ether)s (SFPAE) [52, 53], poly(phthalazinone ether)s (SPPEK) [54], poly (ether ether ketone) (SPEEK) [55–63], etc (See examples in Figure 2.1). These sulfonated membranes were claimed to have lower vanadium ion permeation and higher CE than Naﬁonr membranes.

Figure 2.1: Examples of hydrocarbon based cation exchange membranes

However, the improvement in ion selectivity are usually accompanied with a decline in mechanical or chemical stability due to the increased degree of sulfonation. Therefore, researchers proposed to add fillers into the polymer matrix to keep the balance. For instance, SiO2 [62], TiO2 [38], ZrO2 [51,63], graphene [60] or graphene oxides [61] were reported to reduce the vanadium ion diffusion across the membrane 2. Literature Review 17 and improve the mechanical strength at the same time after incorporation within the polyer matrix. In addition, block copolymer membrane was expected to possess superior properties than alternating or random copolymersystems. Wang and co-workers from EIC Laboratories, Inc. [64] proposed the use of an amphiphilic block copolymer to form an interpenetrating network with both highly hydrophobic unsulfonated and highly hydrophilic sulfonated regions. They prepared an amphiphilic block copolymer PSP with hydrophobic PAEK and hydrophilic SPAEK blocks as an example (see Figure 2.2). Such phase separation was somehow like the Nafionr structure, thus ensuring strong mechanical stability as well as high ionic conductivity.

The obtained membrane was much stiffer than Nafionr membrane and also showed around one order of magnitude higher Young’s modulus. It also demonstrated dramatically reduced permeability rate of VO2+ ion as well as significantly higher CE. Nevertheless, the membrane area resistance was 1.5 Ω·cm2 which was almost double that of the Nafionr membrane.

Figure 2.2: Scheme for the amphiphilic block polymer membrane [64]

2.1.3 Anion Exchange Membranes

According to the Donnan eﬀect, anion exchange membranes (AEM) are supposed to give the minimum vanadium permeation rates by the repelling force produced by the cationic groups. However, due to the cationic groups, the charge carriers 2. Literature Review 18

2− − in VRFBs with AEMs would be mainly SO4 or HSO4 with a small contribution from H+ (due to its small size and extremely high mobility). Therefore, the low ionic conductivity of AEMs would be a major concern compared to CEMs. In the mid-1990s, Kashima-Kita Electric Power Corporation [65] developed a speciﬁc polysulphone type anion membrane with low resistivity to overcome the

+ strong oxidative environment in VO2 solution and applied it in their 2 kW, 10 kW and 200 kW batteries and they performed stably for over 300, 1000 and 650 cycles. The CE, VE and EE of the demonstrated 200 kW battery with such anion exchange membranes were 93%, 86% and 80%, respectively. Probably due to commercial conﬁdentiality, no further detail about the membrane was disclosed. Hwang and co-workers [66] crosslinked the new Selemion anion exchange membrane (Type II-b) with the aid of accelerated electron radiation and the highest CE, VE and EE obtained were 93.5%, 87.7% and 82% at 60 mA·cm−2. The parameters during crosslinking was very crucial in chemical and mechanical stability in vanadium electrolyte and the best results presented in the report was for 8 cycles. DICP developed several new anion exchange membranes for VRFBs, including Chloromethylated poly(phthalazinone ether sulfone)(CMPPES) membrane [67], CMPPES immobilized with trimethylamine(TMA) and ethylenediamine(EDA)/TMA membrane (QAPPES) [68],chloromethylated/quaternized poly(phthalazinone ether ketone) (CMPPEK) anion exchange membrane [69], quaternized poly(phthalazinone ether ketone ketone) (QAPPEKK) membrane [70], quaternized poly(tetramethyl diphenyl ether sulfone) (QAPES) membrane [71], etc (See examples in Figure 2.3). Although they shared slightly lower VE, all of the developed membranes exhibited higher CE than the Naﬁonr 117 or 115 membrane. Some of the membranes can even achieve CE as high as 99.4% at current density of 80 mA·cm−2 for over 30 days

(100 cycles), providing a higher EE than Naﬁonr membrane [70]. To ensure suﬃ- cient membrane conductivity as well as mechanical strength, they [72] developed 2. Literature Review 19 a polysulfone(PSF)/PVDF composite membrane through a dication cross-linking method. This new membrane showed great promise for practical VRFB applications with the ability of delivering CE of 99%, EE of 84% at 80 mA·cm−2 for over 900 cycles.

Figure 2.3: Some examples of anion exchange membranes 2. Literature Review 20

2.1.4 Amphoteric Ion Exchange Membranes

Due to the conflict between vanadium ion permeability and ionic conductivity in cation or anion exchange membranes, amphoteric ion exchange membranes (AIEM) with both cation and anion exchange groups were proposed to achieve high conductivity and low vanadium ion diffusivity. Mohammadi et al [15] originally proposed the incorporation of cation exchange groups into an anion exchange membrane to partially neutralize its anion exchange capacity and reduce water and solution transfer. Sukkar et al subsequently introduced anion exchange groups into cation exchange membranes to achieve similar effects [18].The incorporation of anion exchange functional groups into cation exchange membranes was more recently reported by other groups [27,28,30]. Since 2009, a group from Peking University has specifically developed several AIEMs for VRFBs based on the grafting method. Their first AIEM utilized PVDF as matrix film. They then grafted with styrene (St) and dimethylaminoethyl methacrylate (DMAEMA) under γ-ray irradiation and subsequently performed sulfonation and protonation to achieve both cation and anion exchange capacity. They also pointed out that grafting yield influences the membrane properties significantly and a grafting yield of 26.1% was believed to give the best for the performance in the VRFB. Later they further developed a poly(ethylene-co- tetrafluoroethylene)(ETFE)-based AIEM [73]. Different from the previous grafting method, this time they first grafted the ETFE with St and directly sulfonated the obtained cation ETFE-g-PS membrane to produce ETFE-g-PSSA. The DMAEMA was introduced into this membrane during the second grafting step. After further protonation, an AIEM ETFE-g-PSSA-PDMAEMA was achieved. Compared to

Nafionr 117, this AIEM can provide much better self-discharge performance as well as CE (7.7% higher). However, this AIEM still had around 4% lower VE 2. Literature Review 21 than Nafionr which was probably due to the inferior ionic conductivity. Simi- lar AIEMs were developed by the same group via α-methyl styrene (AMS) and DMAEMA grafting into PVDF membrane [74] and poly(St-co-DMAEMA) into PVDF membrane [75]. Although all these AIEM successfully reduce the vanadium ion crossover, the chemical and mechanical stability in vanadium electrolyte is still one major issue. Other research groups also tried various methods on different polymer materials. For example, Wang and co-workers [76] applied an one-pot direct polymerization method to synthesize the AIEM with pendant quaternary ammonium groups and adjusted the sulfonic acid group and ammonium groups by simply varing the comonomer feed ratio. This membrane did reduce the VO2+ ion permeability rate but the proton conductivity, water uptake as well as weight loss showed inferior properties compared with Nafionr membrane. Li and co-workers [77] selected nontoxic 2-aminoethanesulfonic acid (taurine) to provide the cation and anion exchange capacity to the brominated fluorinated poly(aryl ether oxadiazole) polymer chain and prepare the AIEM by solution-casting. The results showed that the VO2+ ion permeation across the membrane was sucessfully hindered and the self-discharge was greatly reduced compared to Nafionr 115 membrane. The acidic resistivity of the AIEMs under different temperature was evaluated in 2 M

H2SO4 and the conductivities after 7 days of soaking remained around 89%, 80% and 81% of their initial values. As a well-known vanadium complexing agents, amidoximes were selected as the anion exhcange group to supress the vanadium transfer across the membrane. Lee and co-workers [78] applied the pore-filling and cross-linking technique to introduce the ionic conductors vinyl sulfonic acid and (vinylbenzyl)trimethylammonium chloride into the porous polyethylene (PE) substrate film. This AIEM exhibited similar CE, VE and EE to Nafionr 117 over 130 cycles at 50 mA·cm−2. Nibel and co-workers [79] utilized sulfonic acid 2. Literature Review 22 and amidoxime groups as the amphoteric exchange capacity. The capacity fading was minimized. Capacity loss of 3% over 122 cycles and higher VE than Nafionr 117 and 212. Recently, the same group [80] grafted vinylpyridine(VP) monomers into an ETFE film and further functionalized it via alkylation, sulfonation and protonation. The synthesized AIEM not only reduced the vanadium crossover but also achieve comparable ionic conductivity as Nafionr 117. Higher EE, lower capacity loss and reduced water transfer were obtained and good stability over 40 cycles were demonstrated. It is worth pointing out an AIEM developed by Liao and co-workers [81] which possesses an ultra-low vanadium ion diffusion rate and good ionic conductivity at the same time. The structure of this AIEM can be seen in Figure 2.4. The positively charged imidazole rings can repel the positive charged vanadium ions and the crosslinking can reduce the channels to further hinder the transportation of vanadium ions. The oxidatation resistant -CF3 groups and benzimidaozle structures were expected to enhance the chemical and mechanical stability of the membrane. The charge-discharge test results showed that such AIEM can reach a CE more than 99.0% compared with 96.4% for Nafionr 117 while the VE (90.7%) was the same as Nafionr 117 at a current density of 30 mA·cm−2. This high CE was maintained for over 220 cycles with only slight fluctuations.

Figure 2.4: Scheme for the mSPAEK-6F-co-x%BI membrane [81] 2. Literature Review 23

2.1.5 Porous Membranes

Porous separators usually have relatively large pore sizes as well as thicknesses. Unlike ion-exchange membranes, porous separators do not have ion-exchange capabilities but separate the ions based on the diﬀerent transport speeds between the ionic species (similar to the mechanism of chromatography). The variation of Stokes radii and charge densities leads to diﬀerent traveling time so that the ion selectivity can be adjusted by altering the pore size as well as the thickness. As reported by

Chieng [13] and Mohammadi [24], the intrinsic pore size of the Daramicr separator is too large for vanadium ions and modification is required. Tian and co-workers [82] studied the incorporation of polytetrafluoroethylene(PTFE)/perfluorosulfonic acid copolymer(Nafion) solution with Daramicr separator and claimed the impregnated separator can provide a suitable IEC for the VRFB, thus reducing the vanadium ion diffusion as well as improving the proton conductivity. Nanofiltration separators with smaller pore size at nano-scales are another type of porous membrane that have gained attention in the VRFB applications. This kind of membrane was originally used in water treatment or other food industry to separate polyvalent ions. DICP [83] first introduced this concept into VRFB applications and prepared a polyacrylonitrile (PAN) nanofiltration separator and a polysulfone/SPEEK (PSF/SPEEK) nanofiltration separator by phase inversion methods. The pore size distribution can be readily controlled by polymer concentration and co-solvents. The experimental results confirmed that the smaller pore size can increase the ionic selectivity of the membrane and contribute to higher CE. A comparable CE of 95% and EE of 76% (at 80 mA·cm−2) with Nafionr was obtained in their first attempt. Later they [84] further reduced the pore size of the nanofiltration membrane by introducing SiO2 into the pores.

The SiO2 was incorporated by in-situ hydrolysis of tetraethylorthosilicate (TEOS). 2. Literature Review 24

The modified nanofiltration membrane successfully improved the CE to 98% and EE to 79% at 80 mA·cm−2. They subsequently developed two more nanofiltration separators, poly (ether sulfone) (PES) membrane [85] and PVDF ultrafiltration membrane [86] but neither of them achieved CE of more than 95% at 80 mA·cm−2.

2.1.6 Summary for Membrane Development and Research

Gaps

Numerous efforts have been devoted to the development and modification of membranes for VRFB applications. The work is always challenged by the trade-off between ionic conductivity and vanadium permeability, while long-term stability has been difficult to achieve with many novel materials in the literature. Although many membranes were reported to be comparable with the benchmark, Nafionr in certain properties, few can guarantee their long-term performance in practical

VRFB systems. The modification of Nafionr membrane might seem a good approach but the high cost is still a concern. Another problem in the selection of membrane is that the membrane evaluation methods are not standardized and direct comparison between the results from different research groups is impossible. For example, there are many research groups using ion selectivity, or the ratio of conductivity to permeability of VO2+ ion to compare the performance of membranes. Although this ratio has some value in assessing membrane properties, it has limited practical value when evaluating membranes for the VRB and other flow battery systems where ion corss-over can lead to capacity loss. More specifically, the reasons include: (1) the permeabilities of

2+ 3+ 2+ + V , V , VO and VO2 ions across the membrane are different and a knowledge of each of the diffusion coefficients allows detailed modelling of capacity loss to be undertaken in order to schedule electrolyte rebalancing to restore capacity during long term operation; (2) as pointed previously, the measurement methods 2. Literature Review 25 of permeability vary from lab to lab and the permeabilities reported by different groups cannot be directly compared; (3) not all groups have measured the ionic conductivity of membrane and the methods or conditions for the determination of ionic conductivity are different which would again lead to discrepancies in the results [54,83,87,88]. Therefore, the literature review only demonstrates the results as published in the reports without additional processing to avoid any misleading interpretation. Therefore, identifying or developing a new membrane alternative is still a top priority for further commercialization of the VRFB. Prediction of the membrane and cell performance during extended cycling in kW or MW-scale VRFB systems is essential and thus the acquisition of detailed and reliable data from the lab-scale experiment is extremely important for modelling and simulation studies that can help to predict cell behavior during long-term operaion. Most published papers report permeabilities for VO2+ species only, yet the diffusion coefficients of all four vanadium ions must be known for any modelling and simulation studies of practical VRFB systems under different operating conditions.

2.2 Electrodes and electrocatalysts

The electrodes provide the inert surface where the charge-discharge redox couple reactions take place. Although the electrodes in the VRFBs do not take part in the actual charge-discharge reactions, their electrocatalytic properties will strongly affect the kinetics of the redox couple reactions that will in turn influence the voltage and energy efficiency of the VRFB. The electrode material must therefore exhibit fast kinetics for the redox couple reactions, but inhibit undesirable side reactions that could produce hydrogen and oxygen at the negative and positive electrodes respectively during charging. Poor selectivity will not only reduce the 2. Literature Review 26 coulombic efficiency of the VRFB, but could lead to excessive gassing during charging that could lead to safety hazards in the case of hydrogen evolution, or deterioration of the positive electrode material that could potentially reduce cycle life. The electrode is thus a critical component of the RFB that will not only determine the efficiency of the battery, but potentially its cycle life as well. The following sections will provide a detailed overview of the research and development of efficient electrode materials for VRFB applications. This will include early screening studies of electrode materials for different flow battery systems, electrode modifications and electrocatalysis for high power density VRFB systems.

2.2.1 Material selection

The first study on electrode material selection for the VRFB was published in 1987 by Rychcik and Skyllas-Kazacos [89]. Cyclic voltammetry method with saturated calomel (SCE) electrode as reference electrode and graphite rod as counter electrode was first used to investigate the electrochemical behaviour of different materials in VOSO4 in H2SO4 solutions. Then charge-discharge tests were performed in a static cell to further evaluate the electrode performance. They found that most metals are unsuitable for use in acidic solutions since they will dissolve at the high anodic potentials experienced at the positive electrode during charging. Some inert metals such as lead and titanium were found to be passivated in the

2+ + sulphuric acid supporting electrolyte within the VO /VO2 potential range. Gold electrodes showed relatively poor electrochemical reversibility, while dimensionally stable anode (DSA), which consists of noble metal oxides coated on titanium, presented good performance during several charge-discharge cycles. DSAs were however excluded on the basis of cost and long-term stability under continuous charge-discharge cycling was also expected to be poor. 2. Literature Review 27

Carbon and graphite were therefore selected as the most promising electrode materials for the acidic vanadium electrolytes in the VRFB. Although carbon and graphite materials undergo gradual corrosion with oxygen evolution during overcharge at the positive electrode, it was found that these materials exhibited excellent chemical and electrochemical stability during normal battery operation, allowing long cycle life to be achieved with good cell voltage control.

2.2.2 Electrode modiﬁcation and electrocatalysis

Early studies showed considerable variation in electrochemical performance when different graphite paper or felts were used. Skyllas-Kazacos and co-workers compared two typical rayon- and polyacrylonitrile (PAN) - based graphite felts respectively by evaluation and comparison of their physical properties and electrochemical performance [90]. They pointed out that various factors such as surface functional groups, electrical conductivity, surface area, microstructure, hydrophilicity and precursor materials affected their electrochemical performance. Blasi et al [91], Melke et al [92] and Schweiss [93] subsequently tried to correlate the performance of different carbonaceous electrodes and the effect of thermal, chemical and electrochemical treatment with the graphitic degrees. They found that carbon electrodes with less graphitic order tended to show less or even negative effect after treatment but electrodes with higher graphitic character showed much better electrochemical behaviour. Even with the same carbon material, different sites can also show different response for vanadium redox reactions. By examining the kinetics on basal-exposed and edge-exposed graphite foil and HOPG electrodes, Pour et al [94] proved that

2+ 3+ 2+ + more edge sites can signiﬁcantly contribute to V /V and VO /VO2 redox reactions especially when the vanadium concentration is low. In addition to the intrinsic properties of carbonaceous materials, researchers are investigating the approaches such as introducing functional groups, catalysts 2. Literature Review 28 and additives intensively to modify the electroactivity of the electrodes, to reduce the activation polarization for the charging and discharging reactions in VRFB systems.

2.2.2.1 Introduction of oxygen functional groups

In the early 1990s, Sun and Skyllas-Kazacos [95] thermally treated the graphite electrode materials at a range of temperatures and for diﬀerent durations. Using XPS analysis, they found functional groups such as C-OH and C=O increased and suggested they were the contributors to the improved performance of the VRFB with treated graphite felt. They also pointed out that the C-OH groups on the electrode surface could be the active sites and a catalysing mechanism for the positive side reaction was hypothesized as illustrated in Figure 2.5.

Figure 2.5: The catalytic mechanism of C−OH groups toward the positive side reaction [95]

In this mechanism, ion exchange occurs ﬁrstly between VO2+ ions from the electrolyte and H+ of the phenolic functional groups on the electrode surface. Electrons from VO2+ transfer to the electrode with an O atom joining VO2+ to

+ + form VO2 on the electrode surface simultaneously. Finally, the VO2 ion on the graphite surface is replaced by H+ from the bulk solution and forms free ions back in solution. As pointed out by the group, this is a hypothetical mechanism based on the ob- 2. Literature Review 29 served correlation between surface functional groups produced by thermal treatment and the reduced cell resistance as well as increased cell efficiencies. However, more evidence such as the surface functional groups on the graphite fibres after cycling would be helpful to support or contradict this hypothesis. In fact, researchers have revealed that electrochemical treatment would alter the electrochemical activity of the electrodes and cell cycling is one kind of electrochemical treatments that could alter the surface functional groups depending on the potentials used. The strong acidity of the vanadium electrolyte and the significant oxidising nature of

+ the VO2 ions have the possibility to change the surface status of the carbonaceous electrodes. Moreover, V(IV) and V(V) do not only exist in the forms of VO2+ and

+ VO2 ions respectively in the sulphuric acid solutions, but also in the form of other 2+ 2+ + complexes such as VO(H2O)5 , VO(SO4)2 and VOHSO4 , etc. for V(IV) and − 3− − 4+ 2+ VO2SO4 , VO2(SO4)2 , VO3 , V2O3 and V2O4 , etc. for V(V) [96–102]. There- fore, the proposed mechanism oﬀers one of the possibilities which might explain the improved electrochemical activity found in the published reports. The temperature and thermal time were found to be critical to the surface treatment and the best results were obtained for a thermal treatment time of thirty hours at a temperature of 400. At higher treatment temperatures, weight loss associated with mechanical degradation of the carbon due to carbon dioxide formation was observed. Later, the UNSW group [103] used sulphuric acid of diﬀerent concentrations and mixed acid of H2SO4 and HNO3 to treat the graphite felt. Dramatic improvement of the electroactivity was observed as a result of the increased amount of the surface functional groups C-OH and C=O. The C-OH groups obtained after the treatment were thought to contribute to the increased hydrophilicity of the felt as well as the increasing electroactivity. A catalytic mechanism for the negative half-cell reaction was also proposed as shown in Figure 2.6. Similar to the catalytic mechanism for 2. Literature Review 30 the positive half-cell reaction hypothesized by the same group, other complexes forms of the V2+ and V3+ ions in the sulphuric acid solution were not taken into consideration.

Figure 2.6: The catalytic mechanism of C−OH groups in the negative half-cell reaction [103]

V3+ ions thus exchange with H+ of the phenol groups in the ﬁrst step. Electron transfer takes place subsequently and -V+ forms at the end of the C−O bond. This then exchanges with H+ and diﬀuses into the solution. The formation of these oxygen functional groups were thus shown to provide active sites for the vanadium redox couple reactions, while also enhancing wettability of the carbon electrodes through increased hydrogen bonding with the active sites [95,103]. Although stable under normal charging and discharging conditions in the cell however, both carbon and graphite will slowly oxidise and disintegrate if exposed to very high anodic voltages during extensive overcharge and oxygen evolution at the positive electrode. Careful voltage control is therefore required during operation of the VRFB to prevent oxidation of the carbon positive electrodes and maintain long cycle life. Skyllas-Kazacos and co-workers [104] thus investigated the electrochemical behaviour of the positive electrode in the VRFB under overcharge conditions and examined the surface functionality of the graphite felt electrodes. It was reported that after overcharge, the volume resistivity increased slightly, but the cell resistivity increased by 1.35 Ω·cm2 which can be attributed to the loss of electrochemical activity. The XPS analysis results showed that four types of oxygen groups can be formed during the overcharge process, i.e., hydroxide groups (C-OH), ether groups (C-O-C), carbonyl groups (C=O), ether carboxyl groups 2. Literature Review 31

(COOH), ester group (COOR) as well as carbonate groups (-CO3-). It was shown that with increasing overcharge time, more higher oxidation-state groups were produced on the surface such as C=O, COOH and COOR. It seemed to suggest those higher oxides could inhibit the electroactivity of the carbon surface, however the role of various oxygen groups in catalysing the vanadium redox reactions still remains controversial. Since the original work was published by Skyllas-Kazacos and co-workers, various other methods such as electrochemical oxidation [105], wet-chemical treatment [106, 107], corona discharge [108], plasma etching [109, 110] and gamma irradiation [109], etc have been explored to enhance the performance of the VRFBs by introducing oxygen functional groups onto the surface of carbonaceous electrodes and the inﬂuence of content and types of oxygen groups and the variation of surface nature after treatment has been discussed. Li et al [111] hydroxylated carbon paper with diﬀerent content of hydroxyl groups in mixed acids of H2SO4

2+ + and HNO3 and confirmed that -OH did have a positive effect on both VO /VO2 and V2+/V3+ reactions. But when the content of -OH exceeded 11.9%, the carbon paper surface would be damaged seriously, resulting in a negative effect in the electrochemical performance. Kim et al [109] investigated three treatment methods with different purposes to reveal the correlation of the surface area, functional groups and the electrochemical performance of carbon felts. Mild oxidation was utilized to increase the oxygen functional groups as well as the surface area, while oxygen plasma treatment was employed to change the felt surface physically and γ-ray irradiation was used to introduce oxygen groups without much disturbance of the surface area. After comparison with the data of surface area, oxygen group type and the electrochemical behaviour of the treated felt electrodes, it was concluded that voltage efficiency (VE) of VRFB was closely related to surface area and that coulombic efficiency (CE) was affected by the surface functional groups of carbon 2. Literature Review 32 felts. The phenolic (C-O) groups were found to be beneficial to the improvement of electroactivity towards vanadium redox couples. It should also be mentioned that a further limitation of the surface analysis method used to identify the surface functional groups produced after cell cycling is the fact that the electrodes need to be removed, washed and dried before being placed under vacuum in the XPS apparatus. These processes can significantly alter the surface chemistry of the carbon electrodes and give erroneous measurements. Despite these limitations however, most studies by various researchers have continued to use the same approach to assess the surface functional groups of the carbon electrodes when investigating different electrode pre-treatments and have proposed similar mechanisms for the vanadium electrode reactions. Oxygen groups can affect the performance in a number of ways. On the one hand, oxidation of the surface of carbon or graphite electrode can produce carbonyl, carboxylic and phenolic groups. Some hydrophilic groups attached to the carbonaceous surface can modify the wettability of the electrolyte to the electrode, increase the effective surface area and facilitate the charge transfer in return. On the other hand, those functional groups can influence the electrical conductivity of electrodes depending on the carbon atom arrangement order in the original electrode material and the degree of oxidation. In addition, excess oxygen groups can also put electrodes at a risk of disintegration because of easier carbon monoxide and carbon dioxide evolution. Overall therefore, a careful trade-off should be made during modification by surface oxidation, while good voltage control is also needed during charging to prevent over-oxidation of the carbon electrode surface that could lead to undesirable surface functional groups or carbon dioxide evolution. 2. Literature Review 33

2.2.2.2 Metals and metal oxide deposition

As mentioned before, many noble metals have excellent electrochemical activity for the vanadium redox couple reactions in VRFBs. However, in consideration of the cost, they are more suitable for use in modiﬁcation of the carbonaceous electrode surface rather than acting as the actual electrode itself. In the patent issued in 1987, Skyllas-Kazacos claimed that similar eﬀects can be achieved by using carbon electrodes impregnated with the corresponding metals or metal ions [112]. For the positive electrode, impregnation with Au, Mn, Pt, Ir, Ru, Os, Re, Rh, Sb, Te, Pb

2+ + and/or Ag ion was proposed to enhance the kinetics of the VO /VO2 reaction, while for the negative electrode, impregnation with Pb, Bi, Tl, Hg, Cd, In, Ag, Be, Ga, Sb, As, Zn, Ca and/or Mg ions were suggested for inhibition of hydrogen evolution and/or catalysis of the negative V2+/V3+ reaction. Each of these ions was introduced as an additive to the electrolyte and its incorporation on the electrode surface was thought to occur as a metallic deposit formed during cycling at negative potentials, or as an oxide at positive potentials, these providing active sites for the reactions. Later, Sun and Skyllas-Kazacos [113] partially oxidised graphite fibres by cation-exchange with surface functional groups in vanadium acidic solutions containing Pt4+ , Pd2+, Au4+, Mn2+, Te4+, In3+ and Ir3+ and then vacuum drying . Results showed that ions like Mn2+, Te4+ and In3+ did not exhibit much catalytic effect while Pt4+ , Pd2+ and Au4+ enhanced hydrogen evolution. On the other hand, Ir3+ modification showed the greatest improvement in the vanadium redox couple reversibility. In a later study, Wang et al [114] subsequently coated carbon felts with Ir metal (99.77%, by mass) by repeatedly immersing in H2IrCl6 ethanol solution and thermally treating in the air at 450 for 15 min. Using this method, Ir metal demonstrated good coherence to the graphite fibres even after 50 charge-

2+ + discharge cycles. The overpotential of VO /VO2 reaction and the cell resistance 2. Literature Review 34 was decreased signiﬁcantly because of the activation of Ir metal. The EE was also increased by 5.2%. However, since noble metals are always accompanied by the disadvantage of high cost and increased hydrogen evolution at the negative electrode during charging, researchers have continued to ﬁnd cheaper replacements. Santamara and co-workers [115] utilized Bi nanoparticles to modify the positive electrode of VRFB

−1 by immersing the felt into a saturated solution of Bi2O3 in 0.01 mol·L HCl. It was reported that excellent electrochemical performance in terms of increased peak current densities, better reversibility (peak separation of 0.05 V at 1 mV·s−1) and long term stability (100 scans) can be obtained at the same time. Transition metal oxides can be applied as electrocatalysts due to their ability to change between their different valency states [116] and their potential ability to absorb reactive species as active centres [117]. As mentioned before, in the early screening of electrode materials for VRFBs, a titanium electrode with iridium oxide on the surface (known as dimensional stability anodes, DSA) was tested by Skyllas-Kazacos and co-worker and showed excellent electrochemical performance and stability [89]. Later Kim et al [118] filed a patent on graphite/DSA assembled electrode by rolling method for VRFBs in 2013 and claimed it can enhance the electrode lifetime, chemical stability, power density, energy efficiency and cycle performance. Again, high cost became the main factor hindering its broader application. Therefore, decorating carbonaceous electrodes with transition metal of low cost and high catalytic ability may be a promising way for modification.

Kim et al also introduce Mn3O4 onto carbon felts by a hydrothermal method. Carbon felt was immersed with manganese acetate solution in an autoclave and heated at 200 for 12 hours. After this reaction, Mn3O4 was successfully coated onto the surface of the carbon felt. However, these nanoparticles could not remain stable during the single cell test. They therefore heat-treated the modiﬁed carbon 2. Literature Review 35 electrodes at 500 in Ar atmosphere after the hydrothermal reaction. This time Mn3O4 proved to be well attached even after more than 20 charge-discharge cycles. The CV results showed that after Mn3O4 modiﬁcation, the carbon felts demonstrated better reversibility and electrochemical activity toward both the

2+ 3+ 2+ + V /V and VO /VO2 reactions, but the more remarkable improvement was

2+ + for the VO /VO2 reaction. In the cycle test, the vanadium redox flow cell employing Mn3O4 modified carbon felts as both positive and negative electrodes exhibited increasingly higher CE, VE and EE with the increasing cycle number than the cell using bare carbon felts. The researchers also compared the function of Mn3O4 by using the modified electrode only as the positive or the negative

2+ + electrode and found that Mn3O4 was more eﬀective in VO /VO2 reaction. They believed the improvement of electrochemical performance with Mn3O4 was due to the better hydrophilicity of the carbon felt surface and the lower activation barrier for the vanadium redox reactions [119]. Afterwards, several kinds of nano-sized metal oxides were introduced to the surface of graphite or carbon felt such as

WO3 [120] and Nb2O5 [121] by similar hydrothermal method. However, although diﬀerent kinds of metal oxide crystals of high quality can be prepared by using the hydrothermal method, the demand of high vapour pressure during the process makes it unsuitable for modiﬁcation on a large scale. In the patent of Skyllas-Kazacos and co-workers, Pb ions displayed good catalytic ability for the positive VRFB half-cell reaction and high overpotential for oxygen evolution. Moreover, Pb ions can suppress hydrogen evolution as well [122].

Accordingly, Wu et al employed pulse electrodepostion to coat PbO2 particles

(2 µm, a mixture of α-PbO2 and β-PbO2) onto the surface of graphite felt as a positive electrode for VRFBs. From their research, there was no obvious change

2+ + in the peak potential but an increase in peak current for the VO /VO2 redox reaction could be observed in the cyclic voltametric (CV) results. The charge 2. Literature Review 36 transfer resistance obtained from the electrochemical impedance spectroscopy (EIS) measurements of PbO2-modified electrode was 1.85 Ω less and an increase of 2.6% in VE was demonstrated in the single cell cycle test at 70 mA·cm−2. In the later charge-discharge test at different current densities and long-term cycle test (30 cycles), the modified cell maintained a good performance. It can be seen that pulse electrodeposition is a good way to coat metal crystals stably onto the surface of graphite felts, but the improvement of the electrochemical performance is limited.

More investigation in crystal size and structure of PbO2 might be needed [123].

CeO2 modification has captured the attention of researchers recently in a variety of electrochemical applications because of its abundance and the transformation between Ce3+ and Ce4+ oxidation state which may promote redox reactions. Fur- thermore, the oxygen vacancies in the fluorite structure offer CeO2 higher oxygen

2+ + mobility, which might be advantageous for the VO /VO2 reactions. Thus, Xi et al [124] prepared CeO2 decorated graphite felts by precipitation and calcination. Enhanced hydrophilicity, improved reversibility, increased voltage and energy eﬃ- ciencies and good stability in single cell test were demonstrated after modiﬁcation. A catalytic mechanism was proposed as described by Equation (2.1) and (2.2) and Figure 2.7. During the transformation between Ce4+ and Ce3+ (Equation (2.1)), Ce active sites are occupied by adsorbed hydroxyl group (Equation (2.2)). Then ion exchanges occurrs between vanadium ions and hydrogen ions. With the electron

2+ + transfer, VO converts to VO2 at the surface and diﬀuses into the solution. The abundant hydroxyl groups formed on the surface of CeO2 accelerates the electron and oxygen transfer faster, resulting in a higher electrochemical activity.

CeO2 CeO2−x + 1/2O2 (06x6 0.5) (2.1)

+ − CeO2−x + H2O CeO2−x-OH + H + e (2.2) 2. Literature Review 37

Figure 2.7: The catalytic mechanism schematic diagram of CeO2 [124]

In general, it is promising to look for high electroactive metals or metal oxides of low cost to decorate the carbonaceous electrode in a way that is easy for industrial large-scale production.

2.2.2.3 Mesoporous carbon and carbon nanomaterials

Mesoporous carbons and nanomaterials are regarded as promising candidates to catalyse the vanadium redox reactions due to their high surface area, good chemical stability under acidic conditions and flexibility of functionality by introducing oxygen or nitrogenous groups, loading metal or metal oxide nanoparticles and doping different elements. Mesoporous carbon Porous carbons have been proposed as promising electrodes for electrochemical devices due to their great catalytic properties and light density. A rapid screening method was developed by Walsh et al [125] to evaluate activated carbon particles (ACPs) for RFB electrodes by mounting the powders into a thin-layer packed bed electrode. They measured the CV behaviour of this electrode in aqueous sulphuric acid and from the charge envelope in the non-faradic region at controlled scan rates they estimated the capacitance per unit area of various ACPs. The significant specific capacitance value they obtained indicated that almost all the micropores within the ACPs were utilized. Afterwards, they carried out repetitive potential step experiment to characterize the chemical-electrochemical reaction rate and 2. Literature Review 38 constant current cycling to record the overpotential variation with time increasing. The facilitated faradaic reaction rate suggested that the porosity was beneficial for the reaction but the main contribution to the real active surface was from the larger meso/macropores. In the report of Shao et al [126], mesoporous carbon was synthesized by the soft-template method to investigate its electrochemical behaviour towards the

2+ + VO /VO2 reaction. However, its onset potential was even lower than graphite and no obvious redox peak could be observed in the CV measurement. The larger arc radius in the high frequency range also indicated the reaction resistance of mesoporous carbon is greater than graphite. It seemed from this report that pure mesoporous carbon cannot catalyse the vanadium redox reactions. However, several other published articles showed great improvement in electrochemical properties by coating carbon paper with mesoporous carbon.

Zhang et al [117] used Naﬁonr as binder to coat a commercial super activated carbon (SAC) with high surface area (2900 m2·g−1) and pore cubage (1.588 mL·g−1) onto a SGL carbon paper. The electro-catalytic ability towards both positive and negative reactions was notably improved in terms of peak separations and peak currents. The charge transfer resistance was reduced remarkably as shown in the EIS measurement. The VE of the single VRFB was increased from 69.7% to 79.5% after coating with SAC. They owed this improvement to more active sites provided by SAC because of its high surface area, large pore volume and less graphitic structure. They further used SAC as supporting material to make

WO3/SAC composite catalysts for the same carbon paper and more improvement was obtained. In a recent article published by researchers from Singapore [127], mesoporous carbon exhibited superior performance in the VRFB with multicouple reactions. By using coconut shell as precursor, they prepared mesoporous activated carbon 2. Literature Review 39 with BET speciﬁc surface area of 1652 m2·g−1 about and coated them onto the

Torayr carbon paper with PVDF as binder. The raw Torayr carbon paper showed

2+ + great electrochemical activity to VO /VO2 reaction but negligible function to V2+/V3+ reaction. After coating with the bio-mass derived mesoporous carbon, the modified electrode demonstrated significant improvement in catalysing the negative reactions. Moreover, the authors claimed that the reaction peak of V3+/VO2+ can also be observed when the scan rate was not greater than 15 mV·s−1, which indicates a good reaction kinetics and excellent reversibility. There are still debates however as to whether the observed new peaks are associated with the V3+/VO2+ couple, or with surface oxide reactions on the carbon. This modified electrode was however further tested in a static vanadium redox batteries and a 31% increase in the EE was achieved. The high surface area and mesoporosity derived from coconut shell were believed to be the major contributors to the improvement in the electrochemical performance. Compared to graphite felt, carbon paper is much thinner and lacks enough active surface area. Mesoporous carbon can provide a greater surface area and enough flow channels for the electrolyte solution as well. Therefore, this method can be expected to obtain great enhancement in the electrochemical properties of carbon paper, which may be a promising electrode for VRFB with zero-gap architecture (see Section 2.2.2.6). Carbon nanotubes Zhu et al were the first to report the electrochemical behaviour of carbon nanotubes (CNTs) in vanadium electrolytes. They found pure CNTs were not so active towards the redox reactions of vanadium ions, but if CNTs were composited with graphite when their content was less than 5% (by mass), the reaction of

2+ + VO /VO2 was highly catalysed compared to the bare graphite. The CV results of a graphite electrode modified with CNTs under different conditions demonstrated 2. Literature Review 40 that the anodic reactions became reversible on the electrode surface with 5% (by mass) CNTs treated at 200 in vacuum [128]. Later, several investigations were carried out using CNTs of different types [129–131] and functionalities [132,133]. Yan et al [132] coated a glassy carbon electrodes (GCEs) with MWCNTs, hydroxyl MWCNTs and carboxyl MWCNTs respectively and the peak separation

2+ + values of the positive VO /VO2 reaction were in the order: hydroxyl MWC- NTs/GCE < carboxyl MWCNTs/GCE < MWCNTs/GCE < GCE whereas the peak current order was carboxyl MWCNTs/GCE < hydroxyl MWCNTs/GCE < MWCNTs/GCE < GCE. Although the peak separation for carboxyl MWC- NTs/GCE was little higher, given that the peak current of carboxyl MWCNTs/GCE was almost three times of that of hydroxyl MWCNTs/GCE, they claimed that carboxyl MWCNTs/GCE had the best catalytic activity among those samples. In the static VRFB test, the cell with carboxyl MWCNTs modified graphite felt exhibited higher capacity and stable performance under different current densities. They believed that carboxyl functional groups were more beneficial than hydroxyl groups for the oxygen transfer of the positive reaction which is possibly the critical process in the overall mechanism. It is noteworthy that Mench and co-workers fabricated nanoporous layers (NPL) consisting of multiwall CNTs (MWCNTs) on carbon paper (without any binder) and examined the polarization behaviour. They filtered the supernatant of the sodium dodecyl sulfate (SDS) MWCNT solution to prepare a thin film (about 4 µm) of MWCNTs and then placed this layer (NPL) at four different positions of a single-serpentine flow channel VRFB (as shown in Figure 2.8). By comparing

2+ the polarization data for charging in the VO /H2SO4 solution under constant current (0.5 A˚) and then constant voltage (1.8 V) until the current was smaller than 5 mA, the greatest improvement was observed when NPL was placed onto the surface of negative electrode facing the flow field (labeled as -ve NPL|FF), where 2. Literature Review 41 the power density and cell voltage were increased by 8% and 65 mV respectively. They claimed that the enhancement was attributed to the increased active surface area of NPL since the 40 mV improvement in cell voltage cannot be attributed to ohimc loss only [134]. This requires further verification however. Since it was the first time that a discrete carbon NPL was used in VRFBs, more investigation is still needed.

Figure 2.8: A schematic diagram of NPL orientation (FF:ﬂow ﬁeld, mem:membrane) [134]

Although modification with CNTs can increase the surface area and electrical conductivity of the electrode and thus achieve a better electrochemical performance for the vanadium redox reactions, however, residual metal (such as Ni-based catalyst derived from the primary synthesis of CNTs) may lead to hydrogen evolution in the negative side of the cell and this requires further long-term investigation. Graphene-based carbon Graphene was first introduced into the modification of electrodes for VRFBs by Tsai et al [135]. They mixed different contents of thermally reduced graphene 2. Literature Review 42 oxide with graphite and polyvinylidene fluoride (PVDF), coated the mixture onto a Pt foil and measured their CV performance. It showed that the reversibility of

2+ + VO /VO2 reaction was enhanced when the content of graphene was less than 5% (by mass) and the best improvement was obtained when the content was 3% (by mass). The electrochemical behaviour of graphene oxides (GrO) dried under diﬀerent temperatures was characterized by Han et al [136]. It was found that

2+ + 2+ 3+ graphene oxides can improve not only the VO /VO2 but also the V /V redox couple reaction. Though GrO dried under 50 possessed the lowest conductivity and the smallest BET specific surface area, it still showed the best enhancement in CV behaviour. The author thus attributed this effect to the abundant oxygen groups of GrO dried under 50 . Subsequently, Blanco and co-workers published a series of articles based on graphene in VRFBs [137–139]. Firstly, they prepared graphene which was thermally reduced from graphite oxide (GO) at different temperatures and compared the CV behaviour with GO. They believed the restoration of sp2 domains after thermal reduction leads to the remarkable improvement of TRGO [137]. Afterwards, they thermally reduced two kinds of GO into graphene (TRGO) and found the TRGO with higher sp2 content showed better modifications as positive electrode in terms of catalytic ability, charge transfer resistance, CE and EE [136]. They also examined the performance of two kinds of graphene, TRGO and TRGrO, which were thermally reduced from graphite oxide and graphene oxide respectively. The result showed that TRGO presented lower onset potential, smaller peak potential separation and higher peak current densities. Again, the author claimed it was the better restored graphitic structure that renders the better performance for the TRGO than TRGrO [139]. Thus it can be seen that electrical conductivity (influenced by the sp2 content or graphitic structures) is critical to the electrochemical behaviour in VRFBs. 2. Literature Review 43

Regarding the role of oxygen functional groups and electrical conductivity, Yan et al [140] reduced GO electrochemically to different extents. By applying various negative reduction potential from -0.8 V to -1.6 V, they obtained ERGO (electrochemically reduced graphene oxide) with different O/C ratios. Through the comparison of CV results, ERGO obtained at -1.4 V demonstrated the best catalytic activity toward both positive and negative reactions. The content of C-O (including C-O-C and C-OH) decreased dramatically when the applied reduction potential was -0.8 V, but the electrochemical catalytic performance revealed by CV showed much improvement. Thus C-O functional groups were confirmed to be not so essential in the catalysing process. For the same reason, C=O (containing COOH, C=O, COOR) started to be reduced at -1.2 V, whereas the ERGO obtained when the reduction potential was lower than -1.2 V showed slightly increase in the

2+ + peak separation of VO /VO2 reaction. Because with the increasing of negative reduction potential, carbon atoms became more ordered, which implied better electrical conductivity, the reduction of C=O was believed to cause the decay of electroactivity. Therefore, the significance of C=O in the positive redox process was clearly confirmed. In terms of the negative reactions, C-O also presented negligible effect, whereas the hydrogen evolution was activated, when the reduction potential was below -1.2 V. A mechanism of the function of C=O towards vanadium redox reaction was proposed (Figure 2.9). 2. Literature Review 44

Figure 2.9: Schematic mechanism of the C=O functional groups towards vanadium redox reactions [140]

As discussed before, edge planes in graphite layers are believed to be more active than basal planes due to the greater reactivity of the edge defects in the carbon structure. Recently Blanco et al [141] used radio frequency plasma enhanced chemical vapour deposition (RF-PECVD) to synthesize carbon nanowalls, graphene layers aligned vertically on a substrate (Figure 2.10), as positive electrodes for

2+ + VRFBs. The overall electrochemical responses to the VO /VO2 were good and more optimization can be achieved through adjusting the intermediate thickness and edge number of the nanowalls .

2+ + Figure 2.10: The profound mechanism of carbon nanowalls VO /VO2 reactions

Similarly, Cho et al [142] tried dry-ice assisted ball milling method to selectively functionalize the edge of graphene without bare defects in the basal plane and compared its cyclic voltammetry and cycle behaviour with that of conventional hydrazine-reduced graphene oxide. The former electrode exhibited superior elec- 2. Literature Review 45 troactivity and cycle performance toward vanadium redox reactions compared with the latter. It was claimed that graphene materials with oxygen functionalized-edges and defect-free basal planes were the main contributors to the enhanced peak current density and charge-discharge transfer. Nitrogen doping and nitrogenous groups Doping is a method typically applied in the semiconductor industry by introducing impurities into an ultra-pure semiconductor material to adjust the electrical properties. Recently, however, nitrogen- doping of carbon materials have been proved successfully in catalysing many electrochemical reactions [143]. Shao et al [126] were the first to extend this method to the VRFB system. They heat-treated the mesoporous carbon in NH3 environment at a temperature of 850 to realize the doping of nitrogen atoms and achieved a much better electrochemical performance with the modified electrode than the untreated graphite, which might be attributed to three factors. Firstly, the five electrons of nitrogen atoms would lead to extra charge at the π bonds in the graphene layers and thereby both the basicity and electrical conductivity of carbon are increased. Secondly, the chemisorption mode of V/O atoms might be varied, thus lowering the activation energy for V-O bond formation and breaking. Lastly, the hydrophilicity increased after N-doping. Subsequently, several studies by similar method to obtain better electrocatalytic carbon electrode materials for VRFB were reported [133,144,145]. Nevertheless, one disadvantage of this method is that NH3 is toxic and the prevention of gas leakage can be complex and difficult. Other researchers [146–148] employed a hydrothermal method to introduce nitrogenous groups to carbon materials and found these modified electrodes pos-

2+ + sessed excellent wettability and catalytic activity for the VO /VO2 redox couple reaction. Specifically, Zhang et al [146] and Liu et al [149] found that among the four types of nitrogenous groups, pyridinic-N, pyrrolic-N, quaternary-N and 2. Literature Review 46 oxidic-N, the third one is the main contributor to the enhanced properties. Notably, the hydrothermal method is a much milder method and requires lower temperature, but its industrial application would meet some challenges. Recently, zein, the major protein that exists in corn, with several advantages such as high abundance, low cost, and self-assembly ability for hierarchical structures, etc. was used to synthesize highly active, electrocatalysts for VRFBs via a facile solution method. Zein powder and carbon black (CB) were mixed in ethanol and water solution and then dried at 60 to wrap CB with zein film. After a high temperature treatment at 700∼900 in Ar atmosphere, the obtained electrocatalysts were characterized as exhibiting high surface area, large quantities of oxygen groups, nitrogen-doped coating networks as active sites and high electron conductivity. The graphite felt with these N-doped CB particles exhibited much higher electroactivity than untreated graphite felt (GF), oxidized-GF as well as CB coated-GF [150]. Dopamine, another environmental friendly source of nitrogen, was proposed by Kim et al [151] recently to coat onto graphite felt with abundant amine groups. Pristine graphite felt was immersed in dopamine solution (pH of 8.5) at 45 to complete the in-situ polymerization. Then the felt was thermally treated at 900 in Ar for pyrolysing. Compared to oxidized graphite felt, the N-doped felt gave a higher discharge capacity and EE in the VRFB test cell when the current density was 50 to 150 mA·cm−2 . Introducing nitrogenous group via bio-derived sources thus showed great promise in the modification of the carbon electrodes for VRFBs due to their cost-effectiveness, safety and eco-friendliness. Nanocarbon composites Hybrid carbon-based composites, for instance, graphite/CNTs [128], graphite/GO [152], GO/CNTs [153], CNT/CNFs [154, 155], are capable of constructing different dimensional structures (pores or channels), modulating the contents of functional groups and electrical conductivity, and have thus been investigated extensively 2. Literature Review 47 by various groups for VRFB catalysis. The composites were prepared by either physically mixing or in-situ chemical vapour deposition (CVD) method. In addition, carbon nanomaterials can be good substrate materials to support metal or metal oxide nanoparticles to catalyse the vanadium redox reactions [155–162].

Apart from monometallic nanocatalysts, nano-sized metal alloys such as CuPt3 [133,156], and RuSe [157] can also reduce polarization and improve the reversibility

2+ + of the VO /VO2 reaction. As reported by Flox and coworkers, Cu-Pt alloy demonstrated better electroactivity than Pt on graphene and it was probably due to the Pt-Cu bicomponent catalytic activity and its enhanced active surface area [133,157]. Considering that chalcogenide (S, Se and Te) modified metals can effectively catalyse some electrochemical reactions [158], the catalytic effect of RuSe

2+ + into the VO /VO2 reaction was explored by Cui et al with reduced graphene oxide as the dispersion substrate. Results showed that RuSe can remarkably enhanced the reversibility and reduced the polarization of a VRFB single cell [157].

TiO2 possesses good water adsorbing ability and was thus investigated by

Hsueueh and Shieu et al. They used a hydrothermal method to synthesize the TiO2 nanoparticles that were then mixed with CB particles to produce the composite electrodes. It was reported that the wettability was increased with the increasing content of TiO2. Besides, both the speciﬁc capacitance of the electrode and the eﬃciency of the cell with the TiO2/CB layer placed between the membrane and the negative carbon felt electrodes were promoted [160]. Later, the same group published another electrochemical study of these electrodes at high current density

−2 (200 mA·cm ) and found good potential in the application of TiO2/CB as the negative electrocatalyst [161]. Briefly, compositing can be regarded as a good way to enhance the electrochemical response of an electrode to vanadium reactions because of their numerous 2. Literature Review 48 possibilities and potential synergetic effects. More work could be done to find a more optimized combination.

2.2.2.4 Electrolyte additives as electrocatalysts

The effect of electrolyte additives was first disclosed by Skyllas-Kazacos and co- workers in a patent which was filed in 1987 [122]. In the patent, Au, Mn, Pt, Ir, Ru, Os, Re, Rh, Sb, Te, Pb and/or Ag salts were claimed to be effective in the

2+ + kinetic improvement of the VO /VO2 redox reaction when added in the ppm (1×10−6) level. Experimental evidence such as the reduction of the anodic and cathodic peak separation in the cyclic voltammograms (CV) at a glassy carbon electrode after adding AuCl3, PbCl2, InCl3, Sb2O3, TeCl4, SbCl3, Borax and PdCl2 was presented. Among these additives, Au and Sb exhibited higher catalytic eﬀect than Pb, Te and borax. Due to the nature of patents, the inventors did not provide any experimental data of the other claimed additives. Nor did they discuss the additive catalytic mechanisms in detail. But it was believed that during a long-term cycling, the oxides of these metal species can be gradually deposited at the surface

2+ + of the carbonaceous electrode to provide electro-active sites for the VO /VO2 redox reaction. An investigation of the electrode surface before and after the CV measurement would be useful to conﬁrm this hypothesis. In the same patent [122], they also claimed that hydrogen evolution at the negative electrode could be inhibited by adding traces (1×10−6) of Pb, Bi, Tl, Hg, Cd, In, Ag, Be, Ga, Sb, As, Zn, Ca and/or Mg. However, in the example section, they only compared the negative potential CV scans in the electrolyte with AuCl3,

PbCl2, InCl3, Sb2O3, TeCl4, SbCl3, Borax and PdCl2. Among them, only Pb, Sb, Te and borax showed remarkable increase in the hydrogen evolution overvoltage. The real effect of the rest additives a such as Bi, Tl, Hg, Cd, In, Ag, Be, Ga, As, Zn, Ca and/or Mg still needs to be confirmed by further evidence. The reasons 2. Literature Review 49 for their hydrogen suppressing effect in the negative V2+/V3+ electrolyte is also worthy of investigation.

In 2013, Wang et al added a small amount of BiCl3 directly into the electrolyte

2+ + and conducted a series of CV test. It showed that VO /VO2 reaction was hardly aﬀected while the reversibility of V2+/V3+ reaction was enhanced by the Bi additives. Besides, after several cycles in the charge-discharge test, Bi nanoparticles were found to be attached only to the negative carbon electrode. They believed it was due to the standard potential of Bi/Bi3+ which lies between that of the

2+ + 2+ 3+ VO /VO2 and V /V reactions, Bi metal would be formed before the reduction of V3+ to V2+ and it was suggested that Bi metal rather than Bi ions contributes to the improved kinetics of V2+/V3+ reaction. The investigation also showed the charge and discharge capacity, VE and EE were greatly enhanced with the Bi additives. In addition, no obvious decay in EE was observed after 50 cycles under the current density of 50 mA·cm−2 [163]. Recently in the study of Sb3+ additives in VRFB systems, similar effect was also achieved [164]. In a recent study Park and co-workers [165] reported on the influence of Li+, Na+, K+, Mg2+, Cr3+, Mn2+, Fe2+, Fe3+, Co2+, Ni2+, Cu2+, Zn2+ and Mo6+ impurities on the vanadium reactions. By comparing cyclic voltammetric behaviours, they found Li+, Cr3+ and Ni2+ ions decreased the V2+/V3+ reaction kinetics while K+, Mn2+, Fe2+, Fe3+, Co2+ and Cu2+ ions hindered the diffusion of V3+ ions. Moreover, Cu2+ and Mo6+ were found to form sedimentation in both the positive and negative solutions when the concentration was 0.001 mol· L−1. Comparatively,

+ 2+ 2+ 3+ 2+ + Na and Mg ions exhibited negligible eﬀect on both V /V and VO /VO2 redox couple reactions. However, it has to be mentioned that some of the ions were added into the vanadium electrolyte in the form of chlorides, such as LiCl, KCl,

MgCl2, CrCl3, MnCl2, CoCl2, NiCl2, CuCl2 while others were added in the sulphate form such as ZnSO4, or as metal oxides such as MoO6. The difference in anions 2. Literature Review 50 was not taken into consideration when the results were compared. As reported by the PNNL group [166], vanadium ions in HCl solution will behave differently from vanadium ions in H2SO4 solutions. So the effect of the aforementioned metal ions should be acknowledged with their counter ions. Electrolyte additives were also applied in VRFB systems as stabilizing agents by the UNSW group to inhibit precipitation of vanadium species in sulphuric acid for higher energy density [162, 167, 168]. These additives and stabilizing agents included glucose, inositol and other organic materials containing groups of -OH, =O, -NH and -SH [168]. These additives were further studied by groups from Central South University who found that some organic compounds such as D-sorbitol and inositol can also promote the electrochemical activity of vanadium redox couple.

+ Nevertheless, the improvement is not so significant and their oxidation by VO2 is a major problem for long-term stability [169,170]. It seems that adding metal additives is more efficient than organic ones in improving the electrochemical performance of VRFBs. Moreover, compared to the other metal-based modification, adding metal ions in electrolyte is quite simple. Therefore, it is worthwhile to further explore the effect of metal additives, confirm the stability of the cell performance and ensure no adverse effects of the additives in the electrolyte at the same time.

2.2.2.5 Photocatalysts in all-vanadium photoelectrochemical cell

When a semiconductor and an electrode are electrically connected and inserted into solutions which can be either the same or diﬀerent but with an ionic conducting separator, the semiconductor can absorb light energy higher than its bandgap between the conduction and valance band to produce hole (n-type) or electron (p-type) at the surface exposed to the solution under illumination. If the redox potenital of the species dissolved in the solution lies between the conduction and 2. Literature Review 51 valance band of the semiconductor, the hole or electron at the surface can oxidize or reduce the corresponding species in the solution. Using such unique properties, semiconductors can be used to photocatalyze the reactions of redox couples under illumination and build photoelectrochemical cells. Liu and co-workers [171] from University of Texas discovered the hole scavenging of VO2+ ions on semiconductors when they were investigating the photo-response of

2+ TiO2 and TiO2/WO3 electrodes in the solution containing VO ions. Considering

2+ + 2+ 3+ the redox potentials of VO /VO2 and V /V locate just between the conduction edge (-0.5 V vs. NHE) and valance band edge (2.73 V vs. NHE), they [172] assembled a photoelectrochemical cell using TiO2 as the semiconductor electrode,

r 2+ + Naﬁon 117 as the membrane, Pt as the counter electrode, VO /VO2 and

2+ 3+ V /V couples in H2SO4 solution as the electrolytes contacting TiO2 and Pt respectively. This photoelectrochemical cell achieved around 12% light to electricity

−1 eﬃciency with 0.01 mol·L vanadium electrolyte. Similarly, TiO2/WO3 tandem electrode [173] was employed to an all-vanadium photoelectrochemical cell and a higher eﬃciency was obtained. They also found that the TiO2/WO3 [174] photoelectrode can produce current not only under light irradiation but also in darkness. The decomposition reaction of hydrogen tungsten bronze (HxWO3) which was formed during under illumination was claimed to be the reason of dark current.

2.2.2.6 Structural design

Apart from ohmic losses and activation overpotential, mass transport polarization is also an important issue that can affect voltage efficiency as well as the pressure drop through the cell stack that in turn influences pumping energy losses. Increasing felt compression and/or the anode-cathode distance will reduce the ohmic resistance of the felt and of the cell respectively, but this in turn increases pressure drop and pumping energy losses that adversely affects the overall energy efficiency of the 2. Literature Review 52 cell. Several investigations have been carried out with regard to the influence of the thickness, porosity and compression of carbon or graphite felt on the overall performance of VRFB [175]. Most suitable commercial carbon and graphite felts tend to be 3∼5 mm thick, so high ohmic losses are experienced due to the relatively large anode-cathode distance in the conventional type of cell architecture. In recent years however, the concept of the zero-gap cell architecture (Figure 2.11) has been explored by several groups, adapting similar designs as used in fuel cells [176,177].

Figure 2.11: The architecture of zero-gap RFB [176]

United Technologies Research Centre (UTRC) [178, 179] utilized two porous carbonaceous materials, i.e. 2.54 mm thick SGL felt and 0.42 mm thick CP-A carbon paper as electrodes materials to investigate the effect of flow through and interdigitated flow channel configurations. Firstly they found that although the porosity of these two materials did not show much difference (97% and 85% respectively), the density varied dramatically (0.0063 and 0.29 g·cm−3), leading to significant differences in resistance. For example, when contacted with graphite plates incorporating with interdigitated flow fields (IDFF) and compressed at 20%, the bulk area resistances were 76 and 9 mWcm2 for the carbon felt and carbon paper respectively. Further studies showed that the combination of thin carbon paper and 2. Literature Review 53

IDFF can achieve comparable pressure drop with that of thick felt and flat plates, ensuring reasonable flow rates of the vanadium electrolyte to the carbon fibres. Subsequent polarization and scale-up analysis were conducted and they concluded that thick electrodes were essential for the design with flat plates to produce acceptable pressure drop and sufficient mass transport but this was accompanied by large ohmic losses; on the other hand, thinner carbon paper electrodes with IDFF configuration can be a promising approach if this is accompanied by fast kinetics, as this design not only maintains the mass transport but will also reduces the ohmic losses due to the reduced interelectrode gap and higher carbon fiber density [179]. In the case of some active carbon papers, the internal surface area is high because they are loaded with nanoparticles. Other carbon papers have much lower effective surface area however and require surface treatment to enhance their electrochemical activity. However, the voltage drop through the layer is dramatically improved with a thinner carbon structure, with a resultant improvement in the local electrochemical potential at the far reaches of in the porous carbon layer. This is fundamentally a transport phenomenon that manifests itself as a pseudo-ohmic effect in the polarization curve but not in the resistance measurement. By employing a very thin carbon paper as the active layer in the cell, an almost zero anode-cathode distance can be achieved. The incorporation of serpentine or interdigitated flow channels in the graphite current collector however, allows high flow-rates to be maintained with minimal pressure drop through the cell. While dramatically reducing ohmic polarisation losses however, the use of very thin carbon or graphite papers that have a greatly reduced surface area, leads to increased activation polarisation at high operating current densities. Therefore, Zhou and co-workers [180] activate the carbon paper electrodes with KOH to produce a dual-scale porous structure to further increase the surface area. The original 3D 2. Literature Review 54 network of the carbon fibers provided the pathways (large pores of ∼ 10 µm ) for fast electrolyte flow and the nano-scale pores (∼ 5 nm) formed by KOH activation on the surface of carbon fibers served as the active sites for the redox reactions

2+ + 2+ 3+ −1 of VO /VO2 and V /V couple. With a flow rate of 15 mL·s , the energy efficiencies were reported to be 82∼88% at current densities of 200∼400 mA·cm−2, which exhibited the highest VRFB performances in the published reports. An other concern of using thin carbon paper is that the increased dead volume will lead to declining energy efficiency as well. To address this problem, Liu et al published a patent on a new designed electrode structure which consists of a membrane, graphite papers, graphite felts with multi-strip structures and graphite polar plates (Figure 2.12). Because of the large surface area of the graphite felts and the filling of the cavity of the channel, the concentration overpotential and dead volume can be decreased. An increased energy efficiency can therefore be achieved in the VRFB [181]. However, the pressure drop in this structure was not reported, although it would be expected that the filling of the channel cavity with felt will once again increase pressure drop, negating the benefits of the flow channels.

Figure 2.12: Scheme of a simpliﬁed electrode assembly [181] 2. Literature Review 55

2.2.2.7 Catalyst-coated membrane

Early in 1979, Pelligri et al filed a patent describing a redox accumulator for liquid phase electrochemical devices to reduce the ohmic drops and to improve the power density. In the accumulator, anode and cathode were separated by an ion exchange membrane coated with thin, porous, electrically conductive and relatively chemical stable powders. In their example using Cr6+/Cr3+ redox couples, graphite and platinum black were mixed to make the electrode and then bonded to an anionic membrane. When the flow rate was 10 cm·min−1 and charging current density was 0.1 A·cm−2, the charge was extended to approximately 40 W·h·kg−1 in the chromium system. The power density reached 5.8 kW·m2 and the energy efficiency was around 90% [182]. Later in 1993, a patent on membrane-electrode assemblies (MEA) for electrochemical cells was published by Swathirajan et al. Two layers of electrodes and ion conductive materials were coated onto or partially embedded into the membrane. The first layer of the electrode was mainly composed of finely divided (90∼110A)˚ carbon powders with better hydrophilicity, water retention properties and a pH value around 6 7, while the second layer is made of catalytic particles supported on/in carbon particles with smaller size (60∼80 A),˚ higher pH (8∼10) and less wettability. This kind of design was aimed at enhanced catalyst utilization [183]. Actually, MEA or catalyst-coated membranes (CCM) have been widely reported especially for fuel cells [184]. However, only a few studies have been carried out for VRFBs. Zhang et al recently introduced the concept of the CCM for VRFBs [185].

Since WO3/SAC has been proven to be catalytically active for both the positive and negative half-cell reactions and has demonstrated good performance in a single

flow cell when coated onto carbon paper [117], Zhang fabricated a WO3/SAC- coated membrane by the spraying method with Nafionr solution as binder. The performance of the CCM was tested in a cell in which the WO3/SAC catalyst 2. Literature Review 56 layer was applied between the graphite felt and membrane, and the VE and EE obtained at the current density of 120 mA·cm−2 were 81.3% and 76.9% respectively. However, when the cell was assembled using current collectors (graphite plates with carved flow field), the CCM, the VE and EE increased to 85.9% and 81.3% at the same current density. Besides, the almost constant values of EE during 300 charge and discharge cycles confirmed the good stability of the CCM [185]. It is believed that CCMs can enhance the catalyst utilization and extend the interface between catalyst and membrane, and in comparison with coating catalysts on graphite felt or paper, coating on the membrane is proposed to form a compacted catalyst layer, providing better mechanical stability and facilitating the electrochemical reactions. Nevertheless, more reliable evidence is still needed for CCMs and MEA systems in the VRFB. This is therefore an area where significant performance improvements could be achieved, allowing high power density VRFBs to be developed, leading to considerable stack cost savings as well.

2.2.3 Summary of Electrodes and Electrocatalysis and Re-

search Gaps

Currently, there is no perfect modification and electrocatalytic method but effort continues to be devoted by research groups all around the world to the development of highly electroactive electrodes that can achieve high power densities in RFB systems. It is clear that an ideal electrode for RFBs should combine high electrochemical activity and high active surface area to minimize activation polarization, with excellent electrical conductivity to reduce ohmic losses, good porosity or channel structure to decrease mass-transfer polarization and pumping energy losses, good chemical, mechanical and electrochemical stability in the flowing acidic electrolyte, good wettability and reasonable cost. Introducing a certain amount of surface oxygen functional groups can enhance 2. Literature Review 57 the wettability and electroactivity of carbonaceous electrodes, but under some conditions, this can introduce problems such as inferior electrical conductivity and electrochemical stability (under overcharge conditions). In addition, facile and controllable oxidizing methods are still necessary in order to obtain reproducible results. Many metal and metal oxides exhibit excellent catalytic activity towards the vanadium redox reactions, however, hydrogen evolution overpotential, cost- effectiveness, mechanical and chemical stability and scalablility of the synthesis methods should never be ignored when characterizing the modified-electrodes. Carbon nanomaterials can be flexibly tailored in terms of surface area, functional groups, electrical conductivity and compositing with other materials to achieve better electrochemical performance. Doping heterologous elements like N can promote the electrical conductivity, hydrophilicity as well as electroactivity of carbonaceous electrodes, but more environmentally-friendly methods are needed. Other dopant elements such as P, B and S can also be explored. Adding metallic additives directly into the electrolyte is a simple way to catalyse the reactions, but potential side effects such as increased gassing rates should be avoided. More importantly, it is difficult to compare published work to identify the best catalysts for vanadium redox reactions, since in each study, improvements were characterized by different methods or under varying conditions. For example, cell geometry, electrolyte composition, flow rate and temperature are rarely defined. Evaluation studies under standardised conditions are therefore essential. The zero-gap cell structure recently proposed for the VRFB and shown to achieve dramatic improvements in power density and cell efficiencies is a direction worthy of further investigation for scale-up in practical systems. Despite its commercial implementation in many stationary applications to date, further improvements in the performance of VRFB with higher power density, 2. Literature Review 58 lower cost and reduced stack size will help to expand its potential use in a wider range of applications in the future. Chapter 3

Membrane Evaluation

In this chapter, three commercial ion exchange membranes, Fumasepr FAP-450

(anion exchange membrane), Fumasepr F930-rfd (cation exchange membrane) and VB2 (cation exchange membrane) were evaluated for VRFBs. The membrane

+ stability in the highly oxidizing VO2 electrolyte over 800 days was ﬁrst investigated. Secondly, a modiﬁed approach was developed to determine the permeability rates

2+ 3+ 2+ + of V , V , VO and VO2 ions across the membrane more accurately. The experimental results were later applied into a simulation model and the thermal behaviour of a 40 kWh VRFB system under speciﬁc operating scenario was predicted. Most content of this chapter is reproduced from a published paper by the author to Energies.

3.1 Problem Statement

As reviewed previously in Section 2.1, remarkable progress has been made in the development and modiﬁcation of membranes for VRFBs. Although a number of new membranes are currently being used in commercial VRFB systems, apart from Naﬁonr, very little information has been published on the properties and

59 3. Membrane Evaluation 60 performance of these materials and their long-term behaviour is not well understood. There is still a desire to produce or identify new reliable membranes with good stability, low vanadium permeability, high ionic conductivity and lower cost. Fur- thermore, most of the membrane evaluation work in the literature was performed in lab-scale cells and only under short-term cycling. To understand the long-term behaviour in large-scale VRFBs, more eﬀort is needed to further improve the membrane evaluation methods as well as the performance prediction techniques. The ideal ion exchange membrane for redox ﬂow batteries (RFBs) only transfers

+ + − H or other non-reacting ions (such as Na , HSO4 , etc.) to complete the internal circuit while separating the catholyte and anolyte to avoid capacity loss and self- discharge. However, the vanadium ions in the catholyte and anolyte will also diffuse across the membrane leading to loss of coulombic efficiency, the extent of which depends on the vanadium ion selectivity of the membrane. Because of their different size and charge, the four vanadium ions in the VRFB electrolyte will have different diffusion rates and over extended operations, this will result in a build-up of vanadium in one half-cell and dilution in the other half-cell. This leads to a gradual loss of capacity that can be restored by periodically remixing the two-half-cell electrolytes. Apart from the differential diffusion of vanadium ions across the membrane, other side reactions can also lead to capacity loss. At high states of charge (SOC) and at high current densities, oxygen evolution on the positive electrode and hydrogen generation on the negative electrode can occur during charging. Any non-synchronous gassing reactions will contribute to an imbalance between the SOC of the catholyte and anolyte, thus reducing cell capacity. Air oxidation of V2+ ions in the negative half-cell solution will also lead to an imbalance in the SOC of the two half-cell solutions, causing loss of capacity that can only be restored by chemical or electrochemical rebalancing. Measures of charging the cell within 3. Membrane Evaluation 61 a reasonable SOC range, or by employing constant current followed by constant voltage charging can minimize problems associated with gassing side reactions while the impact of air oxidation can be minimized by reducing the exposed area of anolyte to the air and/or de-aerating the negative reservoir. Nevertheless, the diffusion of vanadium ions across the membrane still remains an important factor in determining cell performance. Self-discharge occurs when

+ 2+ 3+ 2+ the VO2 and VO from the positive electrolyte react with the V and V ions from the negative electrolyte as given by the following reactions: In the positive cell:

2+ + + 2+ V + 2VO2 + 2H 3VO + H2O (3.1)

2+ 2+ + 3+ V + VO + 2H 2V + H2O (3.2)

3+ + 2+ V + VO2 2VO (3.3)

In the negative cell:

+ 2+ + 3+ VO2 + 2V + 4H 3V + 2H2O (3.4)

+ 3+ 2+ VO2 + V 2VO (3.5)

2+ 2+ + 3+ VO + V + 2H 2V + H2O (3.6)

Unequal diﬀusion rates of vanadium ions across the membrane can cause a disparity between the SOC of the anolyte and catholyte, thereby leading to capacity fade. Moreover, the exothermic self-discharge reactions can lead to elevated temperature within the cell, especially when the pumps are turned oﬀ during stand-by periods [186,187]. Although equipped with a built-in cooling system− the circulating electrolyte solutions, the possibility of thermal precipitation of

+ concentrated VO2 in the positive half-cell during static or low flow rate situations needs to be avoided. Most research groups evaluating or developing novel membranes for the VRFB have tended to limit their permeation measurements only to the VO2+ species in the 3. Membrane Evaluation 62 discharged positive half-cell solution [28,30,32,71,188 –192] and very few studies have included the diffusion coefficients of the other three vanadium ions [10,193–195]. A knowledge of accurate and validated permeation rates of all four vanadium ions is however vital when selecting membranes for the VRFB and predicting the electrochemical and thermal behaviour under various operating and climatic circumstances by mathematical modelling. To date, complete diffusion coefficient data has only been published for Asahi Selemion CMV [10,193], Tokuyama Soda

CMS and CMX [193], Nafionr 115 [194], and silica nanocomposite AEM [195] membranes. Apart from Nafionr, none of these membranes is currently being used in commercial VRFB systems, so their practical value is limited. The aim of this chapter is thus to evaluate some of the new membranes that are being commercialized for VRFB applications so that their performance can be predicted under a wide range of operating conditions. A second objective of this chapter is to improve the currently used methodology for measuring membrane diffusion coefficients of vanadium ions in the VRFB in order to obtain more reliable data. Experimental techniques used to monitor and interpret the movement of vanadium ions across the membrane in an operating cell as a function of time are subject to a number of interferences that lead to large errors. The process of ion crossover can involve simultaneous diffusion due to the concentration gradient across the membrane, miationgration under external electrical field and osmotic convection associated with the different ionic strengths of the solutions in each half-cell [196]. Several theories have been applied to simulate the ion transport in and across the membrane. Fick’s law [197] is the governing law for diffusion processes driven by a concentration gradient. This process is dependent on temperature and obeys the Arrhenius equation [198]. Grossmith et al. [10] firstly applied Fick’s first law to experimentally measure the diffusion of vanadium ions across a number of membranes using the assumption 3. Membrane Evaluation 63 of one dimensional steady state diffusion. This theory was subsequently used to measure the permeation rate of vanadium ions especially the VO2+ ion by many research groups [28,30–32,44,71,188 –192,194,195,199]. Heintz et al. [200] attempted to employ the Maxwell-Stefan theory to include the electrical field as well as the influence of frictional force between ions based on the ion exchange equilibria data. Later they calculated the ion exchange rate of vanadium ions

+ with H3O ions in the membrane with the combination of Donnan equilibrium and diﬀusion experiment but neglected the frictional eﬀects [193]. Compared to Fick’slaw, such a method is more comprehensive but also complicated for even binary systems and is limited to only low concentrations of vanadium ions. Chen et al. [201] subsequently incorporated the Nernst-Planck equation to determine

2+ + the permeation rates of VO and VO2 ion across cation exchange membranes. The membrane was regarded as charged and the diffusion of vanadium ions was attributed to the concentration gradient as well as the potential difference across the membrane. In fact, the obtained Nernst-Plank equation is in the same form as Fick’s diffusion equation and the Donnan potential in their experiment (where

+ vanadium ions exchanged only with H3O ) can be undermined when exchanging with ions of the same charge, which is closer to reality in an operating VRFB. To apply these theories, the importance of experimental conditions should not be ignored. Typically, a dialysis cell separated by a membrane is used for the diffusion tests. Vanadium and blank solutions are filled in each side of the membrane as the depletion and enrichment components. The permeation rates are then calculated based on the variation in concentration of vanadium ion in the enrichment solution as a function of time. First of all, the measurement should be carried out with fresh membranes and/or samples treated according to the manufacturers instructions. The operating history may give rise to swelling, fouling and even degradation of the membrane at different levels that vary with 3. Membrane Evaluation 64 time [193,200–204]. For example, the water content inside the membrane will affect the interaction between the vanadium cations and sulfonic acid group in cation exchange membrane such as Nafionr, thus leading to adsorption and fouling to various degrees [201,203,204]; the strong acidic electrolyte and the strong oxidative

+ VO2 ions might react with the polymers, altering the chemical structure of the original membranes [14]. Secondly, the hydration shell of the vanadium ions and proton are prone to carrying water molecules across the membrane due to the concentration gradient, while differences in ionic strength between the depletion and enrichment solutions can give rise to significant water transfer that could interfere with the active ion diffusion processes across the membrane. The blank solutions should thus be carefully selected for each of the different vanadium ions to balance the ionic strength and osmotic pressure. The electrochemical potential difference at the membrane/solution interface can also be minimized by exchanging with ions of the same valence. Regarding the cell setup, the static dialysis cell [199] tends to form a diffusive layer on either side of the membrane. Consequently, the vanadium concentration near the surface of the membrane will be much lower than that of the bulk solution, giving rise to concentration polarization and adding unnecessary experimental error to the measurement. Mixing the solution with a magnetic bar [30–32,44,189,193] or circulating the solution by pumps [10,188,194,201] can minimize concentration polarization effects and better represent the real scenario. In the latter circumstance, heating of the solutions by the running pumps can cause temperature variations that should be monitored and controlled to ensure accuracy of the experimental data. Experimental errors can also be introduced by the analytical method used to measure concentration changes as a function of time. Various methods including potentiometric titration [189,194,201], Ultraviolet (UV)-visible spectroscopy [10,32, 3. Membrane Evaluation 65

188,189,193,194], inductively coupled plasma-atomic emission spectroscopy/mass spectroscopy (ICP-AES/MS) [30,44,195, 199] and atomic absorption spectroscopy (AAS) [193] can be used to determine the concentration of vanadium ions in the enrichment solution. The former two methods are highly dependent on the oxidation state of vanadium ions while the latter two can be used to analyze the atomic constituents. Since V2+ and diluted V3+ solutions can be easily oxidized by air, ICP-AES/MS and AAS are better choices for ex-situ concentration determination. In this chapter, the permeation rates of the four vanadium ions across the three kinds of commercial ion exchange membranes were determined by a modified diffusion test based on Fick’s first law. The dimensional stability and performance of each of the membranes was also determined and the experimental diffusion data obtained was then incorporated in the dynamic models based on the work of Tang et al [186,187] and Yan et al [205] to simulate their effects on thermal behaviour under specific operating conditions.

3.2 Experimental

3.2.1 Electrolyte Preparation

The vanadium in sulfuric acid electrolyte was prepared by dissolving V2O3 and

V2O5 powder (EVRAZ Stratcor, Hot Springs, AR, USA) in H2SO4 solution. Take the preparation of 2 mol·L−1 V in 5 mol·L−1 total sulfate/bisulfate solution for

−1 example. Firstly, a 5 mol·L H2SO4 aqueous solution was prepared by adding

98 wt% H2SO4 (Sigma-Aldrich, Sydney, Australia) slowly into milli-Q water (Merk Millipore, Melbourne, Australia). The acid solution was later heated to 80 and then 224.82 g of V2O3 powder was added. The electrolyte was stirred until most of the powder had dissolved. 90.94 g of V2O5 powder was then slowly added into the suspension. After all the powders were dissolved, the solution, comprising a 50:50 3. Membrane Evaluation 66 mixture of V3+ and VO2+ (referred to as V3.5+) was cooled down and distilled water was added to bring the volume to its original level. It was then decanted and collected for subsequent analysis or processing. The vanadium electrolytes of diﬀerent valence states were obtained by electrolysis and mixing. Speciﬁcally, equal volume of the vanadium in sulfuric acid solution was poured into the reservoirs of an electrochemical cell (described in the later Section 3.2.4). A constant current density of 40 mA·cm−2 was used

+ to charge the cell with a voltage limit of 1.8 V. The catholyte would yield VO2 (yellow solution) and the anolyte would produce V2+ (violet solution). If the initial vanadium electrolyte was not balanced (here balanced electrolyte refers to the initial vanadium electrolyte consisting of equal amount of V3+ and VO2+ ions),

+ 2+ the electrolyte of either half-cell would reach VO2 or V ﬁrst. After collecting the yellow or violet solution from one reservoir, this was reﬁlled with the initial V3.5+ electrolyte and charging continued until the other side reached the violet or yellow colour of the V2+ and VO2+ oxidation states respectively. This procedure

+ 2+ was repeated several times until suﬃcient volumes of the VO2 or V solutions

2+ 3+ + were obtained. The VO and V solution were prepared by mixing the VO2 or V2+ solution in a volume ratio of 2:1 and 1:2, respectively. ICP-AES (Optima 7300 ICP-OES Spectrometers, PerkinElmer, Waltham, MA, USA) was utilized to determine the concentration of V and S in the solutions and 2 vol% HNO3 solution was used to dilute the samples to reach the equipment detection limit (1-100 ppm) (The dilution solution should also be carefully selected and other dilution solutions such as H2O, HCl or H2SO4 in this case will inﬂuence the accuracy of ICP analysis results). 3. Membrane Evaluation 67

3.2.2 Stability Test

The stability tests were carried out by immersing the fresh membranes in 300

−1 + -1 mL a 1.6 mol·L VO2 solution in 2.6 mol·L total sulfate/bisulfate at room temperature for more than 800 days. Periodically, the membranes were taken out and the surplus electrolyte on the membrane surfaces was wiped with tissue paper. Subsequently, the length, width and thickness were measured as indicators of the dimensional stability. The weight variation was also determined to evaluate

+ the membrane stability in the highly oxidizing VO2 solution and to estimate the electrolyte adsorption of the membranes.

3.2.3 Diﬀusion Test and Area Resistance

The setup for permeation rate measurement was designed in house as a cell with circulating vanadium electrolyte and blank solution of the same volume on each side of the membrane, as shown in Figure 3.1. 70 mL vanadium solution and corresponding blank solution were filled in the depletion and enrichment side respectively. The vanadium ion concentrations at the beginning of the measurement were set as 0 and 1 mol·L−1 in the enrichment and the depletion side respectively to create considerable concentration difference across the membrane. In order to equalize the ionic strength of the solutions and minimize osmotic pressure effects and solvent transfer across the membrane, the compositions of the permeating and enrichment solutions for each of the vanadium ions were as follows: V2+ ion:

−1 −1 −1 −1 1 mol·L VSO4 + 1.6 mol·L H2SO4 ||1 mol·L MgSO4 + 1.6 mol·L H2SO4 V3+ ion:

−1 −1 −1 −1 0.5 mol·L V2(SO4)3 + 1.1 mol·L H2SO4 ||0.5 mol·L Fe2(SO4)3 + 1.1 mol·L

H2SO4 3. Membrane Evaluation 68

VO2+ ion:

−1 −1 −1 −1 1 mol·L VOSO4 + 1.6 mol·L H2SO4 ||1 mol·L MgSO4 + 1.6 mol·L H2SO4

+ VO2 ion:

−1 −1 −1 −1 0.5 mol·L (VO2)2SO4 + 2.1 mol·L H2SO4 ||0.5 mol·L K2SO4 + 2.1 mol·L

H2SO4

The chemicals used to prepare the blank solutions including K2SO4, MgSO4 and

Fe2(SO4)3 were purchased from Sigma-Aldrich, Sydney, Australia. Inert gas pro- tection was provided during the permeation rate measurements for V2+ solution. The reservoirs were shielded from the light with Aluminium foil when using Fe3+ in the test cell to avoid photodecomposition of Fe2(SO4)3. Balancing the osmotic pressure of the blank and test solution is critical in preventing significant water transfer across the membrane that introduces large errors in the measurement of diffusion coefficients. The UNSW group proposed the addition of MgSO4 into the blank solution to balance the ionic strength for VO2+ permeability studies. This is not effective for other vanadium ions that have a different cation charge.

2+ 3+ + Other salts were thus considered for the determination of the V , V and VO2 permeabilities. It is certain that K+, Mg2+ and Fe3+ ions will diffuse into the other side during the diffusion test, but it is not clear to what extent the presence of these cations will alter the properties of the membrane. It is however expected that the effect can be minimised by reducing the duration of the test in order to minimise the uptake of any introduced cations. It is known that a diffusion layer will form on the surface of the membrane if the solutions remain static (Figure 3.2a) but this can be effectively removed by circulating the solution (Figure 3.2b). To minimize any errors associated with concentration polarization, the diffusion test cell employed pumps to circulate the solution continuously. At regular intervals, a 1 ml sample was withdrawn from the blank solution and measured by ICP-AES to determine the concentration of 3. Membrane Evaluation 69

Figure 3.1: Apparatus for determination of permeation rate the vanadium ions within an accuracy of 99%. The temperature of the solution in the enrichment reservoir was monitored during the entire diﬀusion test. For each diﬀusion test, a new dry membrane was used as received.

(a) (b)

Figure 3.2: The membrane surface in (a)static solutions and (b)ﬂowing solutions

The membrane area resistance was assessed by electrochemical impedance spectroscopy (EIS) with a 5 mV sinusoidal perturbation within the frequency range of 800∼1 kHz in a house-made Swagelok-type two-electrode cell. The distance 3. Membrane Evaluation 70 between the membrane and electrode surface was kept 0.5 mm by placing a piece of gasket (KLINGER SIL C-8200) on each side of the membrane. The eﬀective area of the membrane was 0.16 cm2 (diameter 4.50 mm). The membrane area resistance (AR) was calculated using the following equation,

AR = (R1 − R0) × A (3.7) where A is the exposed area of membrane, and R1 and R0 are the X-intercept of the Nyquist plot of the EIS results of the cell with and without membrane. The ionic conductivity (σ) can also be obtained by

σ = L/AR (3.8) where L is the thickness of the membrane.

3.2.4 Single Cell Setup

The components of the single VRFB were fabricated in-house. One piece of 6 mm thick and 25 cm2 graphite felt was employed on each side of the cell as the anode and cathode. Three kinds of commercial ion exchange membranes Fumasep® FAP-450, Fumasep® F930-rfd, (Fumatech, Bietigheim-Bissingen, Germany) and VB2 (V- Fuel, Sydney, Australia) were employed as the separators. These three membranes were denoted as FAP450, F930 and VB2 in the following parts. Gaskets were used to adjust the compression of the electrodes to around 66.7% of the original thickness. Vanadium electrolyte (1.6 mol·L−1 vanadium in 4.2 mol·L−1 total sulphate solution) in each side of the reservoir was circulated through each half-cell by two MD-10-230GS01 Iwaki Magnet Pumps (Iwaki Pumps, Sydney, Australia). 3. Membrane Evaluation 71

3.2.5 Cycling Test

Charge and discharge curves were performed by a RePower Battery Test System (RePower, Shenzhen, China) within the voltage range of 1.0 V and 1.65 V at a constant current density of 40 mA·cm−2. The voltage range is set to minimize the influence of gassing side reactions. However, H2 evolution was sometimes observed in the negative half-cell at the end of the charging cycle. Nitrogen gas was bubbled through the solution before the cycling tests to remove the oxygen in the solution and reservoir. ParafilmrM (Sigma-Aldrich, Sydney, Australia) was then placed on top of the reservoir to minimize water evaporation and air oxidation of the anolyte. Coulombic efficiency (CE), voltage efficiency (VE) and energy efficiency (EE) values were calculated based on the following equations: R Qdischarge Idischarge(t)dt CE(%) = × 100(%) = R × 100(%) (3.9) Qcharge Icharge(t)dt R Edischarge V (t)Idischargedt EE(%) = × 100(%) = R × 100(%) (3.10) Echarge V (t)Ichargedt EE VE(%) = × 100(%) (3.11) CE where Q and E represents the capacity (Ah) and energy(Wh). I and V are the current(A) and voltage(V) during charge and discharge processes.

3.2.6 Thermal Behavior Studies

Thermal simulation studies were conducted using the models developed by Tang et al [187] and Yan et al [205]. Some of the main model equations and parameters are described in AppendixA. Any parameters diﬀerent from the references are speciﬁed in Section 3.3.4. 3. Membrane Evaluation 72

3.3 Results and Discussion

3.3.1 Membrane Stability

Three ion exchange membranes, Fumasep® FAP-450 , Fumasep® F930-rfd, and VB2 (V-Fuel Pty Ltd, Australia) (denoted as FAP450, F930 and VB2 in the following sections) were selected for this study. As mentioned before, membranes for the VRFB should be able to endure the strongly acidic solutions and the strong

+ oxidative inﬂuence of the VO2 ions in the charged positive half-cell electrolyte. The

−1 + −1 three membranes were therefore immersed in 1.6 mol·L VO2 in 4.2 mol·L total sulphate/bisulphate solutions to investigate their long-term stability. Fumasepr FAP-450 is an anion exchange membrane based on sterically hindered aromatic polyamine which has been used in commercial VRFBs (Cellcube) produced by

Gildemeister (Würzburg,Germany). Fumasepr F930-rfd is a reinforced sulfonated tetrafluoroethylene based fluoropolymer-copolymer cation exchange membrane, while VB2 is a cast perfluorosulfonic acid (PFSA)-based cation exchange membrane. Due to commercial confidentiality, the exact membrane structures of many commercial materials are not provided but the latter two cation exchange membranes are believed to share similar structures of perfluorosulfonic acid (Figure 3.3) which are the key materials of Nafion membrane. The dimensional variations including length, width and thickness as well as weight were measured over a duration of more than 800 days and the results are presented in Figures 3.4∼3.7. The scattering of results reflects the distortion of the samples such as wrinkling and curling during the test that made it difficult to obtain accurate dimensional sizes, and the possible experimental errors in the weight measurement caused by the incomplete removal of the electrolyte attached to the surface of the membrane. All the membranes here showed greater change 3. Membrane Evaluation 73

Figure 3.3: (a)Perfluorosulfonic acid, (b) scheme of Nafion membrane structure (The scheme was modified based on Ref [206]) in both dimensional size and weight in the initial stages of the experiment but then remained relatively stable in the later periods. The length and width of the anion exchange membrane FAP450 increased by around 5∼7% while the thickness remained relatively constant. By comparison, the length of the cation exchange membrane F930 increased slightly while the width shrank by around 6%. Moreover, the thickness was nearly double the initial size at the end of experiment. Such disparity in the dimensional sizes might be associated with the anisotropy of the membrane or the reinforcement. Comparatively, VB2 exhibited the best dimensional stability among the three tested samples and less than 3% variation in all the dimension parameters were observed. In terms of the weight measurement, normally three weight should be measured to distinguish weight loss due to chemical degradation and weight gain due to electrolyte uptake, i.e., the initial weight of the dry membrane before any electrolyte soaking (m0), the weight of wet membrane after immersion and wiped by tissue paper (m1), and the weight after removing any electrolyte or moisture within the polymer membrane (m2). Thus the weight loss can be calculated as (m2−m0) and the electrolyte uptake as (m1−m2). However, the ideal way to remove any vanadium is to soak the membrane in acid, wash in water and dry. But this would have altered the samples and interrupted the long-term immersion studies. Therefore we 3. Membrane Evaluation 74

measured m0 and m1 after different soaking time to demonstrate the combined effect of membrane degradation and electrolyte uptake, (m2−m0)+(m1−m2). During these long-term experiments, it would be expected that solution uptake would have reached steady state in the initial stages, while subsequent weight loss measurement are associated with membrane degradation. As shown in Figure 3.7, FAP450, F930 and VB2 showed around 60%, 28% and 15% increase in the weight in the beginning of study and no significant weight loss was observed beyond the 300

+ days, suggesting good chemical stabilities of all three membranes in VO2 solutions.

1 2 0

1 1 5

1 1 0 ) m m

( 1 0 5

h t g n e 1 0 0 L

F A P 4 5 0 9 5 F 9 3 0 V B 2

9 0 0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 T i m e ( d a y )

−1 + Figure 3.4: Eﬀect of immersion time on the membrane length in 1.6 mol·L VO2 in 4.2 mol·L−1 total sulphate/bisulphate solution 3. Membrane Evaluation 75

1 1 0

1 0 5

1 0 0 ) m m

( 9 5

h t d i

W 9 0

8 5 F A P 4 5 0 F 9 3 0 V B 2 8 0 0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 T i m e ( d a y )

−1 + Figure 3.5: Eﬀect of immersion time on the membrane width in 1.6 mol·L VO2 in 4.2 mol·L−1 total sulphate/bisulphate solution

0 . 1 5

0 . 1 0 ) m m (

s s

e n k c i

h 0 . 0 5 T

F A P 4 5 0 F 9 3 0 V B 2 0 . 0 0 0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 T i m e ( d a y )

Figure 3.6: Eﬀect of immersion time on the membrane thickness in 1.6 mol·L−1 + −1 VO2 in 4.2 mol·L total sulphate/bisulphate solution 3. Membrane Evaluation 76

2 . 0

1 . 5 ) g (

t h g i

e 1 . 0

0 . 5 F A P 4 5 0 F 9 3 0 V B 2 0 . 0 0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 T i m e ( d a y )

−1 + Figure 3.7: Eﬀect of immersion time on the membrane weight in 1.6 mol·L VO2 in 4.2 mol·L−1 total sulphate/bisulphate solution

3.3.2 Permeation Rate Determination of Vanadium Ions

Given the membrane swelling and electrolyte adsorption issue observed in the stability test, a new piece of dry membrane was employed in the permeation rate measurement of each ion in order to minimize errors associated with changes in the membrane pore dimensions between experiments. The solution was circulated by pumps to reduce the concentration polarization. Osmotic pressure was also minimized by carefully selecting the permeating and enrichment solution compositions (see details in section 3.2.3). As presented in Figures 3.8∼3.11, the concentrations of transported vanadium ions increase as the diffusion time increases for all systems. The concentrations of different species in the depletion solution, membrane and enrichment solution at the beginning and the end of the diffusion tests are also listed in Table 3.1∼3.3 3. Membrane Evaluation 77 for comparison. It should be noted that it is difficult to determine the actual

2− − + concentrations of SO4 , HSO4 and H ions due to the uncertainty of the dissociation level of H2SO4 in the actual vanadium solutions where a number of ionic equilibrium reactions are occurring. Furthermore, the concentrations of the counter metal ions were assumed to be similar to that of the transported vanadium ions since the ionic strengths of two half-cell species were set the same at the beginning of the tests and osmotic pressure and solvent transfer were minimized. To the best of the authors knowledge, no report has been published on the interaction between vanadium ions and these counter metal ions inside the membrane. Considering that vanadium ions of four oxidation states are transferring across the membrane concurrently in actual practice, the utilization of metal ions of diﬀerent valences is acceptable. Furthermore, judging from the comparison of the permeability results with other groups in the following parts, even if there is some interaction, the eﬀect is negligible on the magnitude of the permeation rates. In addition, according to

− 2− the Donnan Equilibrium, HSO4 and SO4 are more likely to move across the anion exchange membrane FAP450 than the vanadium cations and counter metal ions while cation exchange membranes such as F930 and VB2 are expected to show the opposite trend. 3. Membrane Evaluation 78

0 . 1 0 2 + V F A P 4 5 0 0 . 0 8 F 9 3 0 V B 2 ) 1 - L l

o 0 . 0 6 m (

n o

i t a

r 0 . 0 4 t n e c n o

C 0 . 0 2

0 . 0 0 0 2 0 0 0 0 4 0 0 0 0 6 0 0 0 0 8 0 0 0 0 1 0 0 0 0 0 T i m e ( s )

Figure 3.8: Concentration of V2+ in the enrichment cell with three membranes.

0 . 0 8 V 3 + F 4 5 0 F 9 3 0 0 . 0 6 V B 2 ) 1 - L l o m (

n 0 . 0 4 o

i t a r t n e c n

o 0 . 0 2 C

0 . 0 0 0 5 0 0 0 1 0 0 0 0 1 5 0 0 0 2 0 0 0 0 2 5 0 0 0 3 0 0 0 0 3 5 0 0 0 4 0 0 0 0 T i m e ( s )

Figure 3.9: Concentration of V3+ in the enrichment cell with three membranes. 3. Membrane Evaluation 79

0 . 3 0 2 + V O F A P 4 5 0 0 . 2 5 F 9 3 0 V B 2 ) 1 -

L 0 . 2 0 l o m (

n 0 . 1 5 o

i t a r t n

e 0 . 1 0 c n o C 0 . 0 5

0 . 0 0 0 2 0 0 0 0 4 0 0 0 0 6 0 0 0 0 8 0 0 0 0 1 0 0 0 0 0 T i m e ( s )

Figure 3.10: Concentration of VO2+ in the enrichment cell with three membranes.

0 . 0 6 + V O 2 F A P 4 5 0 0 . 0 5 F 9 3 0 V B 2 ) 1 -

L 0 . 0 4 l o m (

n 0 . 0 3 o

i t a r t n

e 0 . 0 2 c n o C 0 . 0 1

0 . 0 0 0 5 0 0 0 1 0 0 0 0 1 5 0 0 0 2 0 0 0 0 2 5 0 0 0 3 0 0 0 0 3 5 0 0 0 4 0 0 0 0 T i m e ( s )

+ Figure 3.11: Concentration of VO2 in the enrichment cell with three membranes. 3. Membrane Evaluation 80 − 4 − 4 − 4 − 4 − 4 + HSO + HSO + HSO + 0.090 M + + + 0.026 M + 2+ + HSO + HSO 3+ + H + H + H + + − − − 2 4 2 4 2 4 + H + H − − + SO + SO 2 4 2 4 + SO Continued on next page 2+ 2+ 3+ + SO + SO 2+ 3+ Enrichment solution 1 M Mg approx. 0.910 M Mg 1 M Fe approx. 0.974 M Fe V V 1 M Mg + + ) ) − 4 − 4 2+ − 4 − 4 − 4 3+ + Mg + Fe + HSO + HSO + HSO + HSO + HSO 2+ 3+ − − − − − 2 4 2 4 V V 2 4 2 4 2 4 + SO + SO + SO + SO + SO + + + + + Membrane H H traces of ( H H traces of ( H + + − 4 − 4 − 4 + + + H + HSO + H + HSO + HSO + approx. + approx. + − + + − 2 4 2 4 2+ 3+ + H V V + H + H − 2 4 − − + SO + SO 2 4 2 4 2+ 3+ + SO + SO + SO 2+ 2+ 3+ O V − 4 V − 4 V Table 3.1: Concentration profiles in the diffusion test of FAP450 membrane Depletion solution 1 M approx. 0.910 M 0.090 M Mg HSO 1 M approx. 0.974 M 0.026 M Fe HSO 1 M 2+ 2+ O 3+ t = 0 s t = 86400 s t = 0 s t = 32400 s t = 0 s V V V 3. Membrane Evaluation 81 − 4 − 4 + HSO + 0.097 M − 4 + + HSO + 0.058 M ˇ 2+ + + + H − + H + HSO 2 4 − + 2 4 + SO + H + + SO − 2 4 2+ O approx. 0.944 M K Enrichment solution approx. 0.903 M Mg V 1 M K + SO ) + + 2+ ) − 4 − 4 + − 4 + Mg + K + 2 + HSO + HSO 2+ + HSO O O − − − 2 4 2 4 V V 2 4 + SO + SO + SO + + + H H traces of ( Membrane H traces of ( Continued from previous page + + − 4 + + H + ap- − + H 2 4 + HSO 2+ + approx. + − O 2 4 2+ Table 3.1 – V + SO O + H V + − 2 4 + SO 2+ + SO 2+ − 4 O − 4 V 1 M approx. 0.944 M prox. 0.058 M K Depletion solution approx. 0.903 M 0.097 M Mg HSO + HSO + 2 O t = 0 s t = 32400 s t = 82800 s V 3. Membrane Evaluation 82 − 4 − 4 − 4 − 4 + HSO + HSO +0.062 M + 0.003 M + + 3+ 2+ + HSO + HSO + H + H + + − − 2 4 2 4 + H + H − − + SO 2 4 2 4 + SO Continued on next page 2+ 3+ + SO + SO 2+ 3+ Enrichment solution 1 M Mg approx. 0.997 M Mg 1 M Fe approx. 0.938 M Fe V V ) ) + + + + − 4 − 4 − − 2 4 2 4 2+ 3+ + HSO + HSO + Fe − − + Mg 2 4 2 4 3+ 2+ V V ) ) + − 4 − 4 + + traces of (SO + traces of (SO + + + + Membrane H HSO H traces of (SO H HSO H traces of (SO + + − 4 − 4 + + + H + H + HSO + HSO + approx. + approx. − + + − 2 4 2 4 2+ 3+ V V + H + H − − + SO + SO 2 4 2 4 2+ 3+ + SO + SO 2+ 3+ Table 3.2: Concentration profiles in the diffusion test of F930 membrane V − 4 V − 4 Depletion solution 1 M approx. 0.997 M 0.003 M Mg HSO 1 M approx. 0.938 M 0.062 M Fe HSO 2+ 2+ O 3+ t = 0 s t = 90294 s t = 0 s t = 32400 s V V V 3. Membrane Evaluation 83 − 4 − 4 − 4 + HSO + HSO + 0.255 M + − 4 + + HSO + 0.017 M ˇ 2+ + + + H + H − 2 4 − + H + HSO 2 4 − + 2 4 + SO + SO + H 2+ + + SO − 2 4 2+ O approx. 0.983 M K Enrichment solution approx. 0.745 M Mg V 1 M K 1 M Mg + SO ) + + + − 4 − − 2 4 2 4 2+ + HSO ) + traces of + Mg − 4 − 2 4 + 2+ O V + HSO ) ) − 4 − 4 − + + traces of (SO + traces of (SO + ˇ + K 2 4 + + + + H HSO H Membrane H HSO H traces of (SO (SO Continued from previous page + + − 4 − 4 + + H + ap- − + H 2 4 + HSO + HSO 2+ + approx. + + − O 2 4 2+ Table 3.2 – V + SO O + H + H V + − − 2 4 2 4 + SO 2+ + SO + SO 2+ 2+ − 4 O O − 4 V V 1 M approx. 0.983 M prox. 0.017 M K Depletion solution 1 M approx. 0.745 M 0.255 M Mg HSO + HSO + 2 O t = 0 s t = 32400 s t = 0 s t = 86400 s V 3. Membrane Evaluation 84 − 4 − 4 − 4 − 4 + HSO + HSO +0.018 M + 0.002 M + + 3+ 2+ + HSO + HSO + H + H + + − − 2 4 2 4 + H + H − − + SO 2 4 2 4 + SO Continued on next page 2+ 3+ + SO + SO 2+ 3+ Enrichment solution 1 M Mg approx. 0.998 M Mg 1 M Fe approx. 0.982 M Fe V V ) ) + + + + − 4 − 4 2+ − − 2 4 2 4 3+ + HSO + HSO + Mg + Fe − − 2 4 2 4 2+ 3+ V V ) ) + + − 4 − 4 + traces of (SO + traces of (SO + + + + Membrane H HSO H traces of (SO H HSO H traces of (SO + + − 4 − 4 + + + H + H + HSO + HSO + approx. + approx. − + + − 2 4 2 4 2+ 3+ V V + H + H − − + SO + SO 2 4 2 4 2+ 3+ + SO + SO 2+ 3+ Table 3.3: Concentration profiles in the diffusion test of VB2 membrane V − 4 V − 4 Depletion solution 1 M approx. 0.998 M 0.002 M Mg HSO 1 M approx. 0.982 M 0.018 M Fe HSO 2+ 2+ O 3+ t = 0 s t = 21540 s t = 0 s t = 32400 s V V V 3. Membrane Evaluation 85 − 4 2+ − 4 O − 4 V + HSO + HSO + 0.003M + − 4 + + HSO + + H + + H − 2 4 + 0.080 M − + HSO 2 4 + H 2+ + − 2 4 + SO + SO + H 2+ + − + SO 2 4 + 2 O approx. 0.997 M K Enrichment solution 0.920 M Mg V 1 M K 1 M Mg + SO ) + + + − 4 − − 2 4 2 4 2+ + HSO ) + traces of + Mg − 4 − 2 4 + 2+ O V + HSO ) ) − 4 − 4 − + + traces of (SO + traces of (SO + ˇ + K 2 4 + + + + H HSO H Membrane H HSO H traces of (SO (SO Continued from previous page + + − 4 − 4 + + + H + H + HSO + HSO + approx. + approx. − + + − 2 4 2 4 2+ 2+ Table 3.3 – O O + H + H V V − − 2 4 2 4 + SO + M SO 2+ + + SO + SO 2+ 2+ O O − 4 − 4 V V 1 M approx. 0.997 M 0.003 M K HSO Depletion solution 1 M approx. 0.920 M 0.080 M Mg HSO + 2 O t = 0 s t = 21600 s t = 0 s t = 86400 s V 3. Membrane Evaluation 86

However, due to the heat generated by the working pumps and the convection with the varying ambient atmosphere, the temperature of electrolyte fluctuated in the first 3 h, but then remained stable at 30±2 afterwards. Therefore, only the concentration values obtained after 180 min were used for the calculation of permeation rates. According to the Fick’s first law under the assumption of one-dimensional steady state: C − C J = −k B A (3.12) L where J is the diffusion flux of the amount of vanadium ions per unit area per unit time, mol· dm-2·s-1, k is the diffusion coefficient for vanadium ions in unit of

2 -1 dm ·s , CB and CA are the concentration of vanadium ions contained in vanadium electrolyte and blank solution in the unit of mol·L−1 respectively, L is the thickness of the membrane in the unit of dm, k/L is defined as the permeation rate in dm·s-1 (Inconsistencies can be found in the definition and unit of permeability in literatures [28,30–32,44,71,188 –192,194,195,199,208], so the term ”permeation rates” is used to define k/L in the thesis), the amount of vanadium ions diffused across the membrane per unit time should be equal to the accumulation rate of vanadium ions in the blank solution side: dC −J × A = V A (3.13) A dt

2 where A is the exposed area of the membrane in dm , VA is the volume of the blank solution in dm3. The amount of vanadium ions transferred from the depletion side should be equal to that of the vanadium ions diﬀused into the enrichment side. Therefore,

VA(CA − CA0) = −VB(CB − CB0) (3.14) where CA0 and CB0 are the initial concentration of vanadium ions in the blank solution and vanadium electrolyte respectively. 3. Membrane Evaluation 87

Hence, Equations (3.13) and (3.14) can be combined and give:

dCA Ak VA VA = [CB0 + CA0 − (1 + )CA] (3.15) dt VAL VB VB

Since VA = VB = V,CA0 = 0, the equation can be simpliﬁed to:

dC Ak A = (C + C − 2C ) (3.16) dt VL B0 A0 A

Integrating gives:

2Ak ln(C − 2C ) − ln(C ) = − t (3.17) B0 A B0 VL

According to Equation (3.17), a plot of ln(CB0 − 2CA) vs. t should give a straight line with slope equal to -2Ak/(VL). The experimental data were thus plotted in Figures 3.12∼3.13 and the diffusion coefficient k and permeation rate k/L were calculated from the slope of the line of best fit.

0 . 0 0 2 + V

- 0 . 0 5 ) 0 B

C - 0 . 1 0 ( n l - ) A

C 2

- - 0 . 1 5 0 B C ( n

l F A P 4 5 0 - 0 . 2 0 F 9 3 0 V B 2 F i t t i n g - 0 . 2 5 0 2 0 0 0 0 4 0 0 0 0 6 0 0 0 0 8 0 0 0 0 1 0 0 0 0 0 T i m e ( s )

Figure 3.12: Plots of concentration functions vs. time for V2+ in the enrichment cell. The dot line is the ﬁtting curve for the data obtained after 180 min. 3. Membrane Evaluation 88

0 . 0 0 V 3 + ) 0 - 0 . 0 5 B C ( n l - )

A C 2 - 0

B - 0 . 1 0 C (

n F A P 4 5 0 l F 9 3 0 V B 2 F i t t i n g - 0 . 1 5 0 1 0 0 0 0 2 0 0 0 0 3 0 0 0 0 4 0 0 0 0 T i m e ( s )

Figure 3.13: Plots of concentration functions vs. time for V3+ in the enrichment cell. The dot line is the ﬁtting curve for the data obtained after 180 min.

0 . 0 2 + V O - 0 . 1

- 0 . 2 ) 0

B - 0 . 3 C ( n l -

) - 0 . 4 A

C 2 - 0

B - 0 . 5 C ( n l - 0 . 6 F A P 4 5 0 F 9 3 0 - 0 . 7 V B 2 F i t t i n g - 0 . 8 0 2 0 0 0 0 4 0 0 0 0 6 0 0 0 0 8 0 0 0 0 1 0 0 0 0 0 T i m e ( s )

Figure 3.14: Plots of concentration functions vs. time for VO2+ in the enrichment cell. The dot line is the ﬁtting curve for the data obtained after 180 min. 3. Membrane Evaluation 89

0 . 0 0 V O + 2 )

0 - 0 . 0 5 B C ( n l - ) A

C 2 - 0 B - 0 . 1 0 C ( n l F A P 4 5 0 F 9 3 0 V B 2 F i t t i n g - 0 . 1 5 0 1 0 0 0 0 2 0 0 0 0 3 0 0 0 0 4 0 0 0 0 T i m e ( s )

+ Figure 3.15: Plots of concentration functions vs. time for VO2 in the enrichment cell. The dot line is the ﬁtting curve for the data obtained after 180 min.

The diffusion coefficient values are thus converted into permeation rates of vanadium ions here by dividing by the thickness of the membrane and the results are presented in Table 3.4. The corresponding published data for Nafionr 115 [194] is also included for comparison. It needs to be pointed out however, that this comparison may not be meaningful since the Nafionr 115 measurements were obtained under different conditions. In their experiment [194], only H2SO4 solution was employed as the blank solution for all measurements of four vanadium ions, which can cause water transfer deriving from the imbalanced ionic strength and osmotic pressure.

It can be observed that the results quite agree with the reported data of Nafionr 115 [194] in terms of magnitude. Specifically, the permeation rates of different vanadium ions across the anion exchange membrane FAP450 are in the order of

2+ + 2+ 3+ 2+ VO >VO2 >V >V . F930 and VB2 both have much lower diﬀusivities of V 3. Membrane Evaluation 90

Table 3.4: Thickness L (µm) and diffusion coefficients k (dm2·s-1) for different vanadium ions across membranes

2+ 3+ 2+ + Membrane Thickness k(V ) k(V ) k(VO ) k(VO2 ) FAP450∗ 50 1.61 × 10−10 1.08 × 10−10 1.98 × 10−10 1.92 × 10−10 F930∗ 30 3.24 × 10−12 1.86 × 10−10 3.83 × 10−10 5.11 × 10−11 VB2∗ 100 2.76 × 10−11 1.84 × 10−10 3.05 × 10−10 4.03 × 10−11 Naﬁonr 115∗∗ 127 8.76 × 10−10 3.22 × 10−10 6.82 × 10−10 5.89 × 10−10 2+ 3+ 2+ + Membrane k/L(V ) k/L(V ) k/L(VO ) k/L(VO2 ) FAP450∗ 3.23 × 10−7 2.16 × 10−7 3.96 × 10−7 3.83 × 10−7 F930∗ 1.08 × 10−8 6.20 × 10−7 1.28 × 10−6 1.70 × 10−7 VB2∗ 2.76 × 10−8 1.84 × 10−7 3.05 × 10−7 4.03 × 10−8 Naﬁonr 115∗∗ 6.90 × 10−7 2.54 × 10−7 5.37 × 10−7 4.64 × 10−7

∗The values were obtained under 30 ± 2 in this work; ∗∗ The values of Naﬁonr 115 are from Reference [194].

+ 3+ 2+ and VO2 , while relatively higher values for V and VO . As mentioned in Section 3.1, the transport of vanadium ions under actual charging-discharging situations is driven by multiple forces, including both external and internal electrical field, osmotic convection and concentration gradient. This was initially discussed and studied by Fuller et al [213] in 1992, Chieng [13] in 1993, Mohammadi et al [16] in 1997, Sukkar et al [17] in 2003 and more recently by Mench et al [209–212]. In the current study, only the diffusion of the vanadium ions across the membrane is considered. While this is the dominant transport mechanism when the cell is in standby mode, other transport processes will contribute to ion diffusion during charge-discharge cycling. The presence of the electric field will cause migration of ions across the membrane. Although most of the current in this case would be carried by the more mobile hydrogen ions, migration will have some influence on vanadium ion transfer across the membrane. Water in the aqueous vanadium solution is also likely to transfer across the membrane [16,17,213] due to the osmotic pressure effect induced by the different ionic strengths of the 3. Membrane Evaluation 91 positive and negative half-cell solutions. This water transfer can drag electrolyte across the membrane, leading to additional vanadium ion crossovers. The diffusion experiments here do not take into account the effects of electric field, nor do they account for differences and changes in ionic strength of the two half-cell electrolytes during charge-discharge cycling. However, the internal electrical field produced by the ion exchange functional groups still exists. In addition, apart from membrane conductivity, the membrane thickness also has an inextricable relationship with the transfer of vanadium ions [214]. Therefore, the diffusion coefficients obtained in this experimental design also contains some parasitic transport through the membrane diffusion path. Furthermore, according to Fuller and Newman, the water content of the membrane would also affect the transport behaviour [213]. Hydration can lead to a change in the water content of the membrane during the transport processes, which can in turn affect the diffusion of vanadium ions. Based on Donnan equilibrium theory [215], the positive functional groups of anion exchange membrane have stronger repulsive force against the positive charged vanadium ions with higher valance such as V3+ ions and allow easier passage of

+ 2+ 2+ lower charged VO2 ions. However, in the meantime, VO can form VO(H2O)5 in the moderately acidic media of the vanadium solutions. The hydration shell can lead to accelerated movement of VO2+ ions across the membrane [201, 203, 216]. Another explanation is the known ion pairing between vanadyl and sulphate ions that gives rise to the neutral VO−SO4 ion pair [217] that is not repelled by the charged fixed groups in the pores of the ion exchange membranes. It is believed that the synergetic effect of Donnan exclusion, hydration and ion pairing causes the different diffusion behaviour of the four vanadium ions. By comparison, the two cation exchange membranes According to Yamamoto [218], the higher-valent positively charged ions have larger affinity for negative functional groups in the cation exchange membrane. Therefore, the high diffusion 3. Membrane Evaluation 92 coefficients of trivalent and divalent vanadium ions are reasonable. Moreover, the hydration and ion pairing of VO2+ ions as discussed above can contribute to higher diffusion coefficient than V3+ ions. This would explain why the permeation rates

2+ 3+ + 2+ of vanadium ions are in the order of VO >V >VO2 >V across both F930 and VB2 cation exchange membranes. Apart from the factors discussed above, different membrane manufacturing methods could also give rise to various diffusion behaviour of the vanadium ions. For instance, the extruded PFSA membrane must convert the SO2F to the form of SO3K with KOH and dimethyl sulfoxide (CH3)2SO solution firstly and need a subsequent acid exchange with HNO3 to achieve the final SO3H form. Membranes produced by extrusion are normally thicker than 125 µm [219–221]. The solution cast membrane is prepared by dispersing the polymer to form the base film followed by a drying process without any extra chemical conversion process. The control of thickness and uniformity is better in solution-casting method compared to the extrusion method [221]. However, based on the work published in 2016 [222], the manufacturing process of the membrane has a comparatively weak effect on the permeability but pretreatment methods can alter the membrane properties significantly. Given that all the membranes that studied here are used as received, further work could be performed to identify the effect of pretreatment on the permeability and optimize the pretreatment methods for different types of membranes. The area resistance and ionic conductivity of FAP450, VB2 and F930 were measured in 1.6 mol·L−1 V3.5+ in 4.2 mol·L−1 total sulphate solution, as shown in Table 3.5. Among the three tested membranes, F930 exhibited the lowest area resistance of 0.13 Ω·cm2 while FAP450 and VB2 showed the same value of 0.14 Ω·cm2. Nevertheless, if taking the thickness into account, the ionic conductivity results of the three membranes were in the order of VB2 > FAP450 > F930. It is generally known that cation exchange membranes normally possess higher ionic 3. Membrane Evaluation 93 conductivity in VRFB systems than anion exchange membranes, so the higher value of VB2 over FAP450 is expected. However, regarding the lower σ of F930, the reinforcement of the membrane might a possible explanation since this restricts the membrane expansion and swelling that can affect conductivity.

Table 3.5: Area resistance and ionic conductivity of membranes

2 2 −1 Membrane L (mm) A (cm )R0 (Ω) R1 (Ω) AR (Ω·cm ) σ (S·cm ) FAP450 0.05 0.16 1.34 2.23 0.14 0.04 VB2 0.10 0.16 1.34 2.25 0.14 0.07 F930 0.03 0.16 1.34 2.15 0.13 0.02

Figure 3.16 depicts the plot of the membrane conductivity against the vanadium ion permeation rate. The slope of the dashed line reflects the ion selectivity of the H+ ions over the corresponding vanadium ions. It is obvious that the slopes of the same membrane vary depending on the type of vanadium ions. Moreover, they do not follow the same order for all the four vanadium ions, which suggests the measurement of the permeation rate of a single vanadium ion (typically VO2+ ion by many research groups as reviewed previously) is insufficient and the ion selectivity calculated based on only one vanadium ion does not necessarily represent the ion selectivity of H+ over other vanadium ions, re-confirming the importance of measuring all the permeabilities of four vanadium ions for a comprehensive membrane evaluation. 3. Membrane Evaluation 94

Figure 3.16: Membrane conductivity vs. permeation rate of (a)V2+, (b)V3+, 2+ + (c)VO and (d)VO2

3.3.3 Cycling Tests

As mentioned in Section 3.1, various factors can decide the membrane performance in the VRFBs. Nevertheless, the single cell test is a simple and straightforward way to reveal the synergetic effect of membrane properties such as the permeation rates and ionic conductivity. For example, high permeation rates of vanadium ions can cause significant self-discharge and lead to decreased coulombic efficiency while low ionic conductivity will result in reduced voltage efficiency. A comparison of the voltage efficiencies was obtained with each membrane, when all other cell components are the same and when the cell is tested under identical conditions, provides an in-situ comparison of the ionic conductivities of the membranes in the cell. 3. Membrane Evaluation 95

Therefore, four charge-discharge cycles were carried out in a lab-scale VRFB with electrode area of 25 cm2 and three different membranes within the voltage range of 1.0∼1.65 V at constant current of 40 mA·cm-2 to evaluate the short-term performance. The experimental parameters and the average cell efficiencies of four cycles based on Equations (3.9)∼(3.11) are listed in Table 3.6. The voltage-time profiles of the four cycles are also demonstrated in the solid line in Figures 3.17∼3.19. It can be found that both the anion and the cation exchange membranes have good and stable performance in terms of voltage efficiencies and coulombic efficiency over the four cycles, indicating reasonable permeation rates of vanadium ions and ionic conductivity of the three membranes (The inconsistency between the orders of the VE results and ionic conductivity in Table 3.5 might be caused by different batches used in the cell assembly).

2+ + As mentioned before, it should be noted that in a charged cell, V and VO2 will react in the membrane to form V3+ and VO2+ but this will be diffusion controlled and the relative magnitudes of the diffusion coefficients will still influence the rate of diffusion of each ion into and thus across the membrane. It is therefore useful to know the relative diffusion coefficients for each of the vanadium ions for each membrane since this will influence the differential rate of diffusion across the membrane and the subsequent self-discharge rate and capacity loss during operation of the VRFB. Hence, the experimental permeation rates of four vanadium ions obtained with different membranes were applied into a dynamic model developed by Ao Tang [223]. The simulated curves were plotted in the dash lines in Figures 3.17∼3.19 to compare with the experimental data. It can be seen that the trend, magnitude of voltages and time match very well between the experimental and simulated data. The slight differences between the curves can be derived from many other

2+ factors during the charge-discharge cycles, including air oxidation of V ions, H2 3. Membrane Evaluation 96

Table 3.6: Experimental data from cycling tests. CE: coulombic efficiency; VE: voltage efficiency; and EE: energy efficiency

Membrane C V A L CE VE EE (mol·L−1) (mL) (cm2) (mm) (%) (%) (%)

FAP450 1.6 100 25 0.05 96.2 82.9 79.8 VB2 1.6 90 25 0.10 97.0 85.0 82.5 F930 1.6 80 25 0.03 97.7 81.7 79.8

1 . 7 F A P 4 5 0

1 . 6

1 . 5

) 1 . 4 V (

g 1 . 3 a

t l o

V 1 . 2

1 . 1

1 . 0 E x p e r i m e n t a l 0 . 9 S i m u l a t e d

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0 1 4 0 0 1 6 0 0 1 8 0 0 T i m e ( m i n )

Figure 3.17: A comparison of experimental and simulated charge-discharge curves for the single ﬂow battery with FAP450 membrane

evolution in the negative solution, O2 production in the positive electrolyte and volumetric transfer. In order to study the eﬀect of vanadium permeation on the expected capacity loss and eliminate the inﬂuences of other factors, the dynamic model [223] was employed to simulate the constant cycling performance for 100 cycles with the parameters given in Table 3.7. Applying the experimental permeation rates of vanadium ions across FAP450, 3. Membrane Evaluation 97

1 . 7 F 9 3 0

1 . 6

1 . 5

) 1 . 4 V (

g 1 . 3 a

t l o

V 1 . 2

1 . 1

1 . 0 E x p e r i m e n t a l 0 . 9 S i m u l a t e d

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0 1 4 0 0 T i m e ( m i n )

Figure 3.18: A comparison of experimental and simulated charge-discharge curves for the single ﬂow battery with F930 membrane

1 . 7 V B 2

1 . 6

1 . 5

) 1 . 4 V (

g 1 . 3 a

t l o

V 1 . 2

1 . 1

1 . 0 E x p e r i m e n t a l 0 . 9 S i m u l a t e d

0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0 1 2 0 0 1 4 0 0 1 6 0 0 1 8 0 0 T i m e ( m i n )

Figure 3.19: A comparison of experimental and simulated charge-discharge curves for the single ﬂow battery with VB2 membrane 3. Membrane Evaluation 98

Table 3.7: Speciﬁcation of the VRFB in the dynamic model

Parameters Value

Number of Cells 1 Total Vanadium Concentration 1.6 mol·L-1 Tank Volume(Each Side) 100 mL Temperature 303.15 K Current Density 40 mA·cm-2 Electrode Surface 25 cm2 Membrane Area 25 cm2 Cell Resistance 2 Ωcm2 Formal Cell Potential 1.4 V Gas Constant 8.314 J·mol-1·K-1 Number of Electrons Transferred in Reaction 1 Faraday’s Constant 96485 C·mol-1 Voltage Upper limit 1.65 V Voltage Lower Limit 1.0 V

VB2 and F930 membranes in this work as well as the results of Naﬁonr 115 membrane from reference [194] into the dynamic model, the capacity variation as a function of time for 100 charge-discharge cycles are shown in Figure 3.20. As can be seen from the simulated results, the capacity loss resulting from vanadium ion crossing over in the absence of other factors is quite signiﬁcant at the beginning, but reaches steady states gradually. The discharge capacity for the VRFB with

FAP450, VB2, F930 and Nafionr 115 at the end of 100 cycles decreases to 58.5%, 82.6%, 71.2% and 87.8% of the initial capacity respectively. It is interesting to note that Nafionr 115 [194] has the same order of magnitude for all vanadium ions as FAP450 but greater absolute values than FAP450 membrane and the trends of capacity variation for these two membranes are totally different. Furthermore, VB2

2+ + 3+ 2+ and F930 both have diﬀusion coeﬃcients in the order of V

4 ) h A (

y t i 3 c a p a C

e g

r 2 a h c s i

D F A P 4 5 0 1 V B 2 F 9 3 0 N a f i o n 1 1 5 0 0 2 0 4 0 6 0 8 0 1 0 0 C y c l e N u m b e r

Figure 3.20: Simulated discharge capacity of the VRFBs with FAP450, VB2 and F930 in this work and Naﬁonr 115 in work [194] speed by changing the total concentration of vanadium ions in each half-cell. Figures 3.21∼3.24 give the simulated concentrations of four vanadium ions at the charged state (E=1.65 V) in the VRFB for FAP450, VB2, F930 and Naﬁonr

+ 2+ 115 membranes. For the FAP450 membrane, both vanadium ions VO2 and VO from the positive half-cell have relatively higher permeation rates than the V2+ and V3+ ions from the negative half cell. Consequently, self-discharge reactions (Reactions (3.4)∼(3.6)) can occur to a greater extent, resulting in the build-up of V3+ ions and much reduced V2+ in the negative half-cell electrolyte. For VB2

2+ + and F930 membranes, the diﬀusion coeﬃcients of V and VO2 ions are both around one order of magnitude lower than that of V3+ and VO2+ ions. This can dramatically reduce the self-dicharge reactions in the charged stage. Additionally,

2+ + 2+ 2+ VB2 has more balanced diﬀusion rates for the V /VO2 and V /VO couples compared with F930. Therefore, the eﬀects of self-discharge reactions at both cell 3. Membrane Evaluation 100 voltages are less pronounced for the VRFB with the VB2 membrane. Regarding

Nafionr 115, the reported diffusion coefficients [194] of four vanadium ions are all within the same order of magnitude. Moreover, the particular order of permeation

3+ + 2+ 2+ 2+ rates (i.e., V < VO2 < VO < V ) leads to greater V ions migration to the positive electrolyte in the charged state while more VO2+ ions move to the negative solutions in the discharge state. Therefore, the reduced vanadium concentration in the negative cell at higher potential can be partially compensated during discharging.

This also explains why VRFB with Naﬁonr 115 can reach steady state earlier than with other membranes. It can be thus concluded that more balanced diﬀusion rates

2+ + 3+ 2+ between the V /VO2 and V /VO couples can diminish the imbalance during charge and discharge respectively. Furthermore, the speciﬁc order of permeation

3+ + 2+ 2+ + 3+ 2+ 2+ rates as V

2+ + 3+ 2+ in the order of V

2+ + 2+ 3+ of V /VO2 and VO /V plays a more critical role than their absolute values. Absolute values will however play a significant role in the magnitude of the thermal effects in the cell during cycling and standby. In order to gain an understanding of the sensitivity of the model parameters on the simulation results, however, some additional simulations were carried out in which the diffusion coefficients of three of the vanadium ions was kept the same as before, but the fourth was varied by doubling the value and the results were analysed in Appendix A.3. 3. Membrane Evaluation 102

1 . 6

1 . 2 ) 1 - L l o m (

n o

i 0 . 8 t a

r t

n 2 + e

c V n 3 + o V C 0 . 4 V O 2 + V O + 2

0 . 0 0 2 0 4 0 6 0 8 0 1 0 0 C y c l e N u m b e r

Figure 3.21: Simulated concentration of vanadium ions in the VRFB with FAP450 membrane charged at 1.65V

1 . 6

1 . 4

1 . 2 ) 1 - 2 + L l V o 1 . 0 3 + m

( V

n 2 +

o V O i 0 . 8 t + a

r V O

t 2 n

e 0 . 6 c n o

C 0 . 4

0 . 2

0 . 0 0 2 0 4 0 6 0 8 0 1 0 0 C y c l e N u m b e r

Figure 3.22: Simulated concentration of vanadium ions in the VRFB with VB2 membrane charged at 1.65V 3. Membrane Evaluation 103

1 . 6

1 . 2 ) 1 - L l o m (

n o

i 0 . 8 t a

r t n

e 2 + c V n

o 3 +

C 0 . 4 V V O 2 + V O + 2

0 . 0 0 2 0 4 0 6 0 8 0 1 0 0 C y c l e N u m b e r

Figure 3.23: Simulated concentration of vanadium ions in the VRFB with F930 membrane charged at 1.65V

1 . 6

1 . 2 ) 1 - L l

o 2 +

m V ( 3 + n V o

i 0 . 8

t 2 + a

V O r

t + n V O e 2 c n o

C 0 . 4

0 . 0 0 2 0 4 0 6 0 8 0 1 0 0 C y c l e N u m b e r

Figure 3.24: Simulated concentration of vanadium ions in the VRFB with Naﬁonr 115 membrane charged at 1.65V 3. Membrane Evaluation 104

3.3.4 Thermal Simulation

To study the eﬀect of membrane permeation rates on the thermal behaviour of the VRFB, a 40 kWh system was simulated in an residential power arbitrage scenario as described similar to Reference [205]: The VRFB battery starts charging at 22:00 with an operating range of 20∼80% SOC. Before used for peak shaving during the peak hours 14:00∼20:00, the battery experiences an idle period when the pumps are switched oﬀ. The surrounding air temperature is assumed to be varied according to the Equation (3.18). All the other parameters are presented in Table 3.8.

π 5π T = 10sin( t − ) + 20 (3.18) air 12 6 where Tair is the temperature of surrounding air in ; t is the time in hour and 22:00 is the when t=0. The concentration and temperature variation of the tanks and stack employing were simulated with the thermal model developed by Tang et al [187] and Yan et al [205]. The incorporation of permeation rates of four vanadium ions into the concentration and temperature variation during the operation is demonstrated in the mathematical equations in Appendix A.2. The simulated results of the

VRFBs with FAP450, VB2, F930 and Naﬁonr 115 in the ﬁrst day of operation are presented in Figures 3.25∼3.27. Taking FAP450 membrane (Figure 3.25a) for example, the concentration of V2+

+ 3+ 2+ and VO2 ions increase during charging while the V and VO ions decrease as expected. When the battery enters the standby mode and the pumps are switched oﬀ, the SOC of the electrolyte inside the stack is high, so the self-discharge reactions

2+ + (3.1) and (3.4) lead to a drop in the concentrations of V and VO2 within the stack while the V3+ and VO2+ ions increase. Due to the disparity of the permeation

2+ + 2+ rates of the V and VO2 ions across the FAP450 membrane, charged V ions 3. Membrane Evaluation 105

Table 3.8: Parameters in the thermal simulation model

Parameters Value

Number of Cells 40 Vanadium concentration 2.0 mol·L−1 Capacity 40 kWh Cell Resistance for Charge Process 2.0 Ωcm2 Cell Resistance for Discharge Process 2.1 Ωcm2 Minimum Air Temperature 283.3K Maximum Air Temperature 303.3K Dishcarging Period in 24 h format 14:00 − 20:00 Charge Current 50 A Discharge Current 75 A Lower SOC 20% Upper SOC 80% Flow Factor for charging 6 Flow Factor for discharging 6 Tank Shape Cylindrical Tank Material Polypropylene Tank Height 1 m Tank Wall Thickness 1cm Thickness of Half-Cell Cavity 0.3 cm Length of Half-Cell Cavity (Parallel to the Flow) 30 cm Width of Half-Cell Cavity (Vertical to the Flow) 50 cm Flow Frame Edge Width 5.0 cm End Plate Material Steel + Polypropylene End Plate Thickness 1+1 cm

+ deplete faster than the VO2 ions. Since no self-discharge reactions occur in the tanks, the tank concentrations during the standby period remain constant until discharging starts. During the idle period, the self-discharge reactions in the stack are the only heat source, which can increase the temperature in the stack above 40 (see Figure 3.27a). The consequence of such a temperature increase

+ cannot be understated as VO2 ions tends to precipitate at temperature above 40 if the concentration is high, thereby increasing the risk of blockage in the 3. Membrane Evaluation 106

2.5 2

1.8

2 1.6

1.4

1.5 1.2 V 2 V V 1 3 2 V V 4 1 3 0.8 V V 5

Concentration (M) 4 Concentration (M) V 0.6 5

0.5 0.4

0.2

0 0 22:00 07:00 14:00 17:00 22:00 22:00 07:00 14:00 17:00 22:00 Time Time (a) FAP450 (b) VB2

2.5 2.5

2 2

1.5 1.5 V 2 V 2 V 3 V 3 V 1 1 4 V 4 V 5 Concentration (M) V Concentration (M) 5

0.5 0.5

0 0 22:00 07:00 14:00 17:00 22:00 22:00 07:00 14:00 17:00 22:00 Time Time (c) F930 (d) Naﬁonr 115

Figure 3.25: The simulated concentration in the middle cell of the VRFB stack with (a)FAP450, (b)VB2, (c)F930 and (d)Naﬁonr 115 membrane at the ﬁrst day cell stacks. However, in this case, the temperature increase is accompanied by a

+ decrease in VO2 ion concentration (below 0.5 M in this case) in the stack, so no precipitation would be expected in the present situation. Furthermore, a drop in stack temperature can be achieved by turning on the pumps to allow the electrolyte to be circulated so as to dissipate the heat to the environment or into the tanks, as observed in the temperature profile in Figure 3.27a. Among all the membranes simulated here, FAP450 and Nafionr 115 exhibit the largest heat generation and can even reach 45 within the stack when the pumps are turned off, while VB2 shows the best performance with the temperature only rising to less than 26 . 3. Membrane Evaluation 107

1.8 1.8

1.6 1.6

1.4 1.4

1.2 1.2 V 2 V 1 3 1 V 4 0.8 V 0.8 5 Concentration (M) Concentration (M)

0.6 0.6 V 2 V 3 0.4 0.4 V 4 V 5 0.2 0.2 22:00 07:00 14:00 17:00 22:00 22:00 07:00 14:00 17:00 22:00 Time Time (a) FAP450 (b) VB2

1.8 1.8

1.6 1.6

1.4 1.4

1.2 1.2 V V 2 2 V V 3 3 1 1 V V 4 4 V V 5 0.8 5 0.8 Concentration (M) Concentration (M)

0.6 0.6

0.4 0.4

0.2 0.2 22:00 07:00 14:00 17:00 22:00 22:00 07:00 14:00 17:00 22:00 Time Time (c) F930 (d) Naﬁonr 115

Figure 3.26: The simulated vanadium concentration in the VRFB tanks with (a)FAP450, (b)VB2, (c)F930 and (d)Naﬁonr 115 membrane at the ﬁrst day

The peak temperature for the four membranes during the ﬁrst cycle is in the order

VB2 < F930 < FAP450 < Naﬁonr 115. As the charged negative and positive electrolytes have the higher content of

2+ + V and VO2 ions respectively, the self-discharge reactions are mainly driven by the diﬀusion of these two ions. Therefore, Naﬁonr 115, which has the higher

2+ + values of V and VO2 ion permeation rates, can contribute to a greater level of exothermic self-discharge reactions. A similar explanation can also be applied to the VRFB with FAP450 membranes. Although F930 has relatively lower V2+

+ and VO2 diffusion coefficients, the difference between these two values is almost 3. Membrane Evaluation 108

50 30 Cell 28 45 Tank Air 26 40 24 C) C)

o 35 o 22

30 20

18 25 Temperature ( Temperature ( 16 20 14 15 Cell 12 Tank Air 10 10 22:00 07:00 14:00 17:00 22:00 22:00 07:00 14:00 17:00 22:00 Time Time (a) FAP450 (b) VB2

45 50 Cell Tank 40 45 Air

40 35 C) C)

o o 35 30 30 25 25 Temperature ( Temperature ( 20 20

15 15 Cell Tank Air 10 10 22:00 07:00 14:00 17:00 22:00 22:00 07:00 14:00 17:00 22:00 Time Time (c) F930 (d) Naﬁonr 115

Figure 3.27: The simulated temperature of the middle cell, tank of the VRFB with (a)FAP450, (b)VB2, (c)F930 and (d)Naﬁonr 115 membrane at the ﬁrst day one order of magnitude. Furthermore, the smaller thickness of this membrane compared with the other materials gives rise to higher permeation rates and greater heat generation during the standby period. The VB2 membrane, being the thickest of the three commercial membranes investigated in this work, shows the lower degree of heat generation as expected. The trends of concentration and temperature in the stack and tank for the seven day period were also simulated and are illustrated in Figures 3.28∼3.30. The growing imbalance between the SOC of two half-cells is attributed to the discrepancy among the permeation rates of the four vanadium ions. The decreasing 3. Membrane Evaluation 109

2.5 2

1.8

2 1.6

1.4

1.5 1.2

1 0.8 Concentration (M) Concentration (M) 0.6 V 2 V 0.5 0.4 2 V V 3 3 V V 4 0.2 4 V V 5 5 0 0 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 Time (Day) Time (Day) (a) FAP450 (b) VB2

2.5 2.5

2 2

1.5 1.5

1 1 Concentration (M) Concentration (M)

V V 2 0.5 2 0.5 V V 3 3 V V 4 4 V V 5 5 0 0 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 Time (Day) Time (Day) (c) F930 (d) Naﬁonr 115

Figure 3.28: The simulated concentration in the middle cell of the VRFB stack with (a)FAP450, (b)VB2, (c)F930 and (d)Nafionr 115 membrane at the first week capacity can be thus predicted by the model and regular re-balancing procedures can be scheduled. The stack temperature increases during the standby stage due to the considerable self-discharge reactions, but drops again once the pumps are turned on and fresh solution enters from the tanks. The tank temperature is seen to fluctuate around 25 . VB2 is still seen to give the least heat during the stand-by period because of the lower permeation rates. Furthermore, the smaller variation in the diffusion coefficients of the four vanadium ions, also gives rise to the lowest imbalance between the ion concentrations in each half-cell. It can thus be inferred that in order to avoid dramatic heat accumulation in the stack when the pumps 3. Membrane Evaluation 110

1.8 1.8

1.6 1.6

1.4 1.4

1.2 1.2

1 1

0.8 0.8 Concentration (M) Concentration (M)

0.6 V 0.6 V 2 2 V V 3 3 0.4 V 0.4 V 4 4 V V 5 5 0.2 0.2 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 Time (Day) Time (Day) (a) FAP450 (b) VB2

2 1.8

1.8 1.6

1.6 1.4

1.4 1.2 1.2 1 1 0.8

Concentration (M) 0.8 Concentration (M)

V 0.6 V 0.6 2 2 V V 3 3 0.4 V 0.4 V 4 4 V V 5 5 0.2 0.2 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 Time (Day) Time (Day) (c) F930 (d) Naﬁonr 115

Figure 3.29: The simulated concentration in the VRFB tanks with (a)FAP450, (b)VB2, (c)F930 and (d)Nafionr 115 membrane at the first week are turned off during standby, the ideal membrane should possess low and balanced diffusion coefficients for all four vanadium ions or at least for the charged V2+ and

+ VO2 ions. It should be pointed out however, that while small errors in the values of the calculated permeation rates will have only a minor eﬀect on the predicted heat generation, the impact on the simulated concentration changes in each half-cell over time might be quite signiﬁcant. The suggested capacity trends given in these simulations utilizing the experimental permeation rates should therefore be treated with caution. 3. Membrane Evaluation 111

50 30

28 45 26 40 24 C) C)

o 35 o 22

30 20

18 25 Temperature ( Temperature ( 16 20 14 15 Cell Cell 12 Tank Tank Air Air 10 10 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 Time (Day) Time (Day) (a) FAP450 (b) VB2

50 55

45 50

45 40

40 C) C)

o 35 o 35 30 30 25

Temperature ( Temperature ( 25

20 20

15 Cell 15 Cell Tank Tank Air Air 10 10 0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 Time (Day) Time (Day) (c) F930 (d) Naﬁonr 115

Figure 3.30: The simulated temperature of middle cell, tank of the VRFB with (a)FAP450, (b)VB2, (c)F930 and (d)Naﬁonr 115 membrane at the ﬁrst week

3.4 Summary

In this chapter, three commercial ion exchange membranes FAP450, VB2, F930 were evaluated experimentally in terms of stability, permeation rates and cell eﬃciencies. The stability test showed that all three membranes experienced some swelling in dimensions and adsorption of electrolyte after soaking in the strong oxidative

+ VO2 solutions. A modified methodology was developed for the determination of vanadium ion diffusion coefficients. The anion exchange membrane FAP450, showed a similar order of magnitude of permeation rates for all four vanadium ions, 3. Membrane Evaluation 112 while in the case of the cation exchange membranes F930 and VB2 the permeation

2+ + 3+ of V and VO2 ions was almost an order of magnitude lower than those of V and VO2+. Of the two cation exchange membranes, the thicker VB2 exhibited the lowest permeation rates as would be expected. The constant current charge- discharge tests with VRFBs over four cycles showed that these three membranes can deliver high cell efficiencies and good stability during short-term cycling. Furthermore, the longer-term performance of each membrane was evaluated with the aid of simulation models. To simplify, dynamic models incorporating the effect of ion crossover was applied to investigate the capacity loss and thermal behaviour of VRFBs. The simulated results showed that the VB2 membrane can retain capacity comparable to Nafionr 115 membrane, but the other two FAP450 and F930 membranes had more severe capacity loss. The simulation results also imply

2+ + 3+ 2+ that balanced permeation rates between V /VO2 and V /VO couples with

3+ + 2+ 2+ + 2+ 2+ 2+ speciﬁc order of V

2+ + SOCs, lower order of magnitudes of the permeation rates of V and VO2 ions are essential. This requirement was satisﬁed by the thicker VB2 membrane in the present study. Chapter 4

Vanadium redox kinetics on various electrode surfaces

In this chapter, a systematic approach was utilized to study the reactions of the

2+ + VO /VO2 redox couple with a focus on the eﬀects of electrode surface roughness and function groups derived from the pre-treatment on the electrochemical

2+ + behaviour of glassy carbon in VO /VO2 solutions. Firstly, 600-grit, 1200-grit and 4000-grit SiC sandpapers were used to change the roughness of the electrode surfaces. Electrochemical preatment by cyclic voltammetry (CV) within diﬀerent potential ranges were applied to alter oxygen functional groups on the electrode surface. CV and electrochemical impedance spectroscopy (EIS) were then used

2+ + to investigate the electrode behaviour in 1M VO /VO2 (1:1) solution on the glassy carbon electrodes. Scanning electron microscopy (SEM) was employed to characterize the surface roughness. X-ray photoelectron spectroscopy (XPS) was adopted to analyze the surface oxygen functional groups. The contents of this chapter are reproduced from a published paper by the author in the Journal of the Electrochemical Society. The reaction kinetics of V2+/V3+ couple had been studied by M. A. Goulet working in the same group and published in reference [224].

113 4. Vanadium redox kinetics on various electrode surfaces 114

4.1 Problem Statement

The kinetic study of the vanadium redox reactions is essential and critical to the performance of the cell especially with regard to the power density. As summarized in Tables 4.1 and 4.2, however, discrepancies can be easily found in the

2+ 3+ 2+ + published studies of V /V and VO /VO2 redox reactions, not only in relation to the magnitude of the calculated rate constants but also in the identification of the faster half-cell reaction. Apart from experimental conditions including the type of electrode material used, the electrolyte species and concentrations, solution pH and measuring temperature, the surface condition of the electrodes also imposes a significant impact on the electrochemical response and kinetics of the vanadium redox couples. Despite the importance of these parameters on the kinetic measurements, most publications have failed to specify all of the experimental details, especially the preparation or pretreatment of the electrodes used. 4. Vanadium redox kinetics on various electrode surfaces 115 -1 cm 6 − 10 × 750 − Methods Rate Constant CV 1500 Cyclic Voltam- metry (CV) CV and rotating- disc voltammetry (RDV) Continued on next page redox reaction on carbonaceous electrodes − − − + 2 4 in O /VO S + 2 4 2 2+ O O H S V 2 H 4 O S 2 0.05 M V in 3 M 0.05 M V in 1 M H 1.8 M polish- 3 O 2 Al m µ 1200-grit SiC sandpaper polishing and distilled water cleaning 600- and 800-grit SiC sandpaper, and then 0.3 ing, distilled water cation Mechanical polishing [ 227 ] washing and ultrasoni- Table 4.1: Published kinetic studies for VO Electrode Surface Treatment ElectrolyteGlassy carbon Temperature Graphite and reticu- lated vitreous carbon [ 226 ] Graphite reinforcement carbon (GRC) (GC) [ 225 ] 4. Vanadium redox kinetics on various electrode surfaces 116 -1 cm 6 − 10 3 × < Methods RateCV Constant CV 0.3 Continued on next page − in and + 2 2+ O 4 O V O S V 2 H 90 mM 110 mM 1 M Continued from previous page µ Table 4.1 – polishing and 3 O 2 Al 600- and 1200-grit paper polishing, Buehler polisher at 100 rpm m ultrasonic bath − − − with 1, 0.3 and 0.05 Electrode Surface Treatment ElectrolyteHighly ordered Temperature rolytic py- graphite Graphite rotating disk electrode (HOPG) [ 228 ] (RDE) [ 229 ] 4. Vanadium redox kinetics on various electrode surfaces 117 -1 cm 11 6 − 10 × 10 1470 × Methods RateCV Constant 850 CV 0.813 LSV 1.2 CV, linear scan voltammetry (LSV) Continued on next page − − − − in in in 4 4 4 O O O S S S O4 O4 O4 O O O S S S V V V 2 2 2 4 H H H O S 2 0.05 M V in 1 M H 2.0 M 2.0 M 2.0 M 2 M 2 M 2 M Continued from previous page Table 4.1 – 1000-grit paper polishing, ultrasonication in SiC paper down to 5 min at 300rpm, ultrasonication and air dry − − 2000 grit polishing for water and air dry Electrode Surface Treatment ElectrolytePlastic formed Temperature carbon Graphite rod RDE [ 232 ] Graphite [ 233 ] (PFC) [ 230 ] (GR) [ 231 ] 4. Vanadium redox kinetics on various electrode surfaces 118 -1 cm 6 33 − ± 10 × Methods RateCV, Constant EIS 176 LSV 0.0168 Microfluid cell 10 Continued on next page − − 20 CV− 2130 in in in 1 4 4 and 4 4 + 2 O O 2+ O O S S O O S S 4 2 O O 2 V V O H H V V 4 S 2 O H S 2 50 mM 50 mM M 0.1 M 0.5 M 1.2 M 1.7 M V in 4 M H 3.0 M Continued from previous page slurry and 7 slurry polish- Table 4.1 – 3 2 O 2 O Si Al − 800-, 2400- and 4000- grit SiC sandpaper, 50 nm nm ing SiC paper down to nel polishing and alcohol rinsing Heat treatment 2000-grit and then flan- RDE [ 235 ] Electrode Surface Treatment ElectrolyteMulti-walled carbon Temperature nanotubes on GC [ 234 ] GR [ 236 ] Carbon paper (MWCNTs) (CP) [ 237 ] 4. Vanadium redox kinetics on various electrode surfaces 119 -1 cm 6 − 10 × Methods RateCV Constant 534 Static cell 0.5 CV, EISCV, EISCV, EIS 44.8 CV, EIS 430 CV, EIS 28.1 160 220 Continued on next page − Room temperature − − − − − and 2+ 2+ 2+ 2+ 2+ in 2 + 2 O O O O O 2+ and 2.5 V V V V V O O : : : : : 4 4 V + 2 + 2 + 2 + 2 + 2 V O O S O O O O O S 2 2 V V V V V H H 0.1 M 0.1 M M 1.0 M V inM 2.5 M HCl Continued from previous page Table 4.1 – 1200 and 2000 CCR/R SiC paper polishing and 5 min ultrasonication − Anodic treatment 1:1 Electrode Surface Treatment ElectrolyteGraphite [ 238 ] Temperature CP [ 239 ] GC [ 240 ]GC [ 240 ] Anodic treatmentCP [ 240 ] Cathoodic treatmentCP [ 240 ] 1:1 1:1 Anodic treatmentCarbon Xero- Cathoodicgel treatment [ 240 ] 1:1 1:1 4. Vanadium redox kinetics on various electrode surfaces 120 -1 cm 6 − 10 × − Methods RateCV, Constant EISCV, EIS 1183 CV, EIS 36.0 170 CV, EIS, analytical flow cell − − − − in 4 2+ 2+ 2+ O O O 5+ . V V V : : : 4 3 + 2 + 2 + 2 O V S O O O 2 V V V H 0.1 M M Continued from previous page , rinsed + 2 Table 4.1 – O V 3+ / V / 2+ 2+ O V Immersedand then in immersed in Cathodic treatment 1:1 V Electrode Surface Treatment ElectrolyteCarbon Xero- gel [ 240 ] Temperature Fiber [ 240 ]Fiber [ 240 ] Anodic treatmentCP [ 224 ] Cathodic treatment 1:1 1:1 4. Vanadium redox kinetics on various electrode surfaces 121 -1 cm 6 − 10 × Methods Rate Constant CVCV 480 1 CV 530 Continued on next page 25 CV− − 17.1 − redox reaction on carbonaceous electrodes 3+ 4 in in O and /V S 3 3+ 2 atom 2+ 2+ V Cl H − V V -3 -3 4 4 4 4 O O O O S 2 S S S 2 2 2 0.12 M 0.5 M and 1.04 M 0.05 M V in 1 M H 0.875 g 0.805g-atom H 0.05 M V in 1 M H Na (V) dm (V) dm - α m µ polishing and 3 O 2 1200-grit SiC sandpaper and 0.3 Al distilled water risiing Immersed in the test solution for 9h − Table 4.2: Published kinetic studies for V GRC [ 227 ] Mechanical Polishing Electrode Surface Treatment ElectrolyteGlassy carbon Temperature GC [ 241 ] PFC [ 230 ] (GC) [ 225 ] 4. Vanadium redox kinetics on various electrode surfaces 122 -1 7 cm 6 10 − × 10 × Microfluid cell 0.1 Methods RateLSV Constant CV 2.81 86.4 Static cell 0.01 CV, EISCV, EISCV, EIS 136 CV, EIS 13.8 27.7 1.13 Continued on next page − − Room temperature − − − − in 4 and 4 O 2+ in 2 M 3+ 3+ 3+ 3+ S O and 2.5 V S V V V V 4 O 3+ : : : : 2 V O V H 4 2+ 2+ 2+ 2+ S 2 O V V V V S H 2 − − 0.1 M 0.5 M 0.2 M 0.2 M H 1.0 M V inM 2.5 M HCl Continued from previous page Table 4.2 – 1200CCR/R and paper polishing and 5-min ultrasonication − − 2000CCR/R SiC CP [ 237 ] Heat treatment Electrode Surface Treatment ElectrolyteRDE [ 235 ] Temperature Graphite [ 238 ] CP [ 239 ] GC [ 240 ]GC [ 240 ] Anodic treatmentCP [ 240 ] Cathoodic treatmentCP [ 240 ] 1:1 1:1 Anodic treatment Cathoodic treatment 1:1 1:1 4. Vanadium redox kinetics on various electrode surfaces 123 -1 cm 6 − 10 × − Methods RateCV, Constant EISCV, EIS 369 CV, EIS 28.2 CV, EIS 76.3 8.60 CV, EIS, analytical flow cell − − − − − in 4 3+ 3+ 3+ 3+ 5+ . 4 2 V V V V : : : : O V S 2+ 2+ 2+ 2+ 2 V V V V H 0.1 M M Continued from previous page , rinsed + 2 Table 4.2 – O V 3+ / V / 2+ O 2+ Immersedand then in immersed in Anodic treatment 1:1 Cathodic treatment 1:1 V V Electrode Surface Treatment ElectrolyteCarbon Xero- gel [ 240 ] Temperature Carbon Xero- gel [ 240 ] Fiber [ 240 ]Fiber [ 240 ] Anodic treatmentCP [ 224 ] Cathodic treatment 1:1 1:1 4. Vanadium redox kinetics on various electrode surfaces 124

In the original studies of the vanadium redox couple reactions by Sum and Skyllas-Kazacos [225], surface polishing was shown to dramatically aﬀect the observed cyclic voltammetric response of a glassy carbon electrode in the vanadium

2+ + redox couple solutions. The authors found that the reversibility of the VO /VO2 reaction on glassy carbon electrodes polished with 1200-grit sandpaper alone, was enhanced dramatically compared to that on the electrodes polished with 1200- grit sandpaper and 0.3 µm alumina. Although they speculated that mechanical polishing SiC paper leads to the improved activity of the glassy carbon electrodes, they did not provide any evidence as to the roughness variation or any oxygen functional groups introduced by polishing. Polishing is known to have the capability to activate carbonaceous electrodes for various reactions [241,242]. The possible reasons can be explained as i) removal of contaminants from the surface, ii) increase of electrode area by roughening, iii) more exposure of fresh carbon edges, nanoparticles and decfects that can be active sites for electron transfer and iv) introduction of surface functional groups as potential electron mediators. The applied potential in electrochemical treatment or measurement is also a critical factor that will inﬂuence the electrochemical response of electrodes [243–248] especially when the potential is higher than the equilibrium oxidation potential of the redox couples because it tends to bring about surface oxidation and corrosion as well as the generation of oxygen functional groups. The surface status of the electrode can be changed completely, leading to divergent electrochemical activity. Hence, this chapter aims to build up a systematic way to study the reactions

2+ + of the VO /VO2 redox couple and mainly focuses on the eﬀect of surface roughness and functional groups on the electrochemical behaviour of glassy carbon

2+ + in VO /VO2 solutions. 600-grit, 1200-grit and 4000-grit SiC sandpapers were utilized to change the roughness of the electrode surfaces. Cyclic voltammetry (CV) 4. Vanadium redox kinetics on various electrode surfaces 125 within diﬀerent potential ranges were applied in advance to alter oxygen functional groups and roughness of the electrode surfaces. CV and EIS were then used to investigate the electrode behaviour of glassy carbon disc and glassy carbon plate

2+ + electrodes in 1 M VO /VO2 solution containing 2.5 M total sulphate/bisulphate 2− − (SO4 + HSO4 ). Scanning electron microscope (SEM) and X-ray photoelectron spectroscopy (XPS) were employed to characterize the surface roughness and surface functional groups. The eﬀects of surface area and oxygen functional groups on the electrochemical behaviour of glassy carbon electrodes are discussed, as is the eﬀect of vanadium oxidation state or electrolyte state of charge (SOC).

4.2 Experimental

4.2.1 Working electrodes

The glassy carbon (GC) electrode used in this work was CHI 104, supplied by CH Instruments, Inc with an exposed area of 0.096cm2. Glassy carbon plate (GCP) electrodes were prepared from glassy carbon plate (supplied by Mitsubishi Chemicals) covered with epoxy resin to isolate the contact from the electrolyte, while leaving an exposed area of 0.25 cm2 on one ﬂat side. While it would be desirable to use carbon paper or felt in the kinetic studies, these are not good electrode materials for electrochemical studies. The reasons for this are: (1) their surface area cannot be accurately determined; (2) they are not uniform and (3) their surface cannot be readily reproduced. A good carbonaceous model electrode is therefore desirable. The reason to choose glassy carbon (GC) as working electrode instead of other carbonaceous material such as graphite is that GC usually possesses a wider electrochemical window due to its higher water oxidation or reduction overpotential. Comparatively, graphite is porous and is also vulnerable and tends to delaminate at high potentials. Considering that electrochemical pretreatment 4. Vanadium redox kinetics on various electrode surfaces 126 under diﬀerent potential ranges is one of the main subjects in this work, GC is a better choice. Furthermore, graphite is commonly contaminated with metal impurities, GC also exhibits advantages at the purity level.

4.2.2 Electrolyte

The preparation of the electrolyte was similar to the method described in Sec- tion 3.2.1. The electrolyte used in the studies was a solution of 0.5 M VO2++0.5

+ M VO2 (1:1) in 2.5 M total sulphate/bisulphate unless otherwise speciﬁed. A solution of 1 M VO2+ in 2.5 M total sulphate was used to investigate the eﬀect of vanadium oxidation state. This was prepared by mixing the corresponding V2+

+ and VO2 solutions with a volume ratio of 1:2.

4.2.3 Surface treatment

GC and GCP electrodes were polished manually with SiC sandpaper of 4000-grit, 1200-grit and 600-grit (Bunnings, Sydney, Australia) for 30 s respectively to alter the surface roughness. All electrodes were shaken vigorously in milli-Q water and then rinsed with ethanol after polishing to remove any impurities introduced by the

2+ + sandpaper. CV was conducted in 1 M VO /VO2 (1:1) in 2.5 M total sulphate solution at a scan rate of 100 mV·s−1 within different potential ranges to vary the functional groups as well as the roughness on the electrode surface. For oxidation of the surface, the electrodes were first scanned from the open circuit voltage (OCV) to a fixed potential 0.2 V (vs. Hg/Hg2SO4 reference electrode in saturated K2SO4 solution, same as the following potentials unless specified), then to upper limiting potential such as 0.7, 1.0, 1.3 and 1.6 V, and finally back to the OCV to complete the cycle. In a separate set of experiments, the electrodes were first oxidized at 1.6 V and then scanned in the opposite direction from a fixed upper potential limit of 0.7 V to various lower potential limits of 0.2, −0.1, −0.4, −0.7 and −1.0 V. 4. Vanadium redox kinetics on various electrode surfaces 127

To simplify the description, samples were named according to their material type (GC or GCP), polishing conditions and the limiting potential of the anodic and cathodic treatment. For example, glassy carbon electrodes and glassy carbon plate electrodes which were only treated with sandpaper down to 4000-grit are noted as GC4000 and GCP4000 respectively; the glassy carbon electrode polished with 1200-grit sandpaper and then oxidized by scanning anodically within the range of 0.2∼1.6 V is denoted as GC1200 1.6V and the glassy carbon electrode polished with 4000-grit sandpaper and subjected to a reduction of 0.7∼-1.0 V is marked as GCP4000 -1.0V.

4.2.4 Electrochemical measurements

Electrochemical measurements were performed at room temperature in a three- electrode single compartment cell with a cylindrical graphite rod as counter electrode and a Hg/Hg2SO4 (in saturated K2SO4 solution) reference electrode. A Bio-VSP potentiostat (Bio Logic Science Instruments, Seyssinet-Pariset, France) was used to control the electrochemical parameters and record the electrode responses. Five CV scans were tested within the range of 0.7∼0.2 V (ﬁrst scanned to 0.7 V and then back to 0.2 V) for each sample. A relaxation at OCV for 2 min was set between each scan to ensure that the electrode/electrolyte interface had recovered before the next measurement. Only the ﬁfth scan is displayed in the graphs presented in the Results and Discussion Section 4.3. Electrochemical Impedance Spectroscopy (EIS) was implemented at OCV with a sine amplitude of 10 mV and a frequency range from 20 kHz to 10 mHz. The charge transfer resistance was calculated by modelling the equivalent circuit. 4. Vanadium redox kinetics on various electrode surfaces 128

4.2.5 Structural characterization

The surface morphology of the treated electrodes was recorded by scanning electron microscopy (SEM) using a FEI Nova NanoSEM 450 (FEI company, Oregon, United States) with a voltage of 5.00kV, a spot size of 3.0 and a working distance of 6.2 mm. A Thermo ESCALAB 250Xi X-ray photoelectron spectroscopy (XPS)

(Thermo Scientific, Sydney, Australia ) with monochromated Al Kα (1486.68 eV) at 13 kV×12 mA was employed to analyze the surface chemical state of the electrodes. The spectrometer binding energy scale was calibrated according to Au 4f7 (83.96 eV), Ag 3d5 (368.21 eV) and Cu 2p3 (932.62 eV). For detailed C1s and O1s spectral analysis, mixed 30/70 Lorentzian and Gaussian line shapes and non-linear Shirley background subtraction was adopted during the curve fitting. Due to the difficulty of sample preparation from the glassy carbon electrode for SEM and XPS analysis, all characterizations were performed on the surface of glassy carbon plate electrodes and no element other than C and O has been found.

4.3 Results and discussion

4.3.1 Eﬀects of electrochemical treatment on glassy car-

bon plate electrodes

A glassy carbon plate (GCP) electrode polished with 4000-grit sandpaper was selected to act as the working electrode to study the inﬂuence of electrochemical treatment. Figure 4.1 shows the eﬀect of anodic treatment on the CV response of the

2+ + GCP4000 electrode in 0.5 M VO +0.5 M VO2 in 2.5 M total sulphate/bisulphate solution. The arrow shows the initial scan direction. It can be clearly seen that during the anodic oxidation treatment of the GCP4000, especially when the

2+ + treatment potential was greater than 1.0 V, apart from the VO to VO2 oxidation 4. Vanadium redox kinetics on various electrode surfaces 129

1 5 0 G C P 4 0 0 0

1 0 0

5 0 ) 2 - m c

m 0 (

0 . 2 ~ 0 . 7 V - 5 0 0 . 2 ~ 1 . 0 V 0 . 2 ~ 1 . 3 V 0 . 2 ~ 1 . 6 V - 1 0 0 0 . 0 0 . 5 1 . 0 1 . 5 2 . 0

E ( V ) v s . H g / H g S O 2 4

2+ + Figure 4.1: The anodic treatment of GCP4000 in 0.5 M VO +0.5 M VO2 in 2.5 M total sulphate/bisulphate solution peak, other small oxidation peaks can also be observed, suggesting some oxygen functional groups are developed during this treatment. After oxidation, cathodic scans were conducted on GCP4000 within various potential ranges, as shown in

2+ + Figure 4.2. No obvious peaks were observed except for the VO /VO2 redox reaction peaks. 4. Vanadium redox kinetics on various electrode surfaces 130

1 5 0 G C P 4 0 0 0 0 . 7 ~ 0 . 2 V 1 0 0 0 . 7 ~ - 0 . 1 V 0 . 7 ~ - 0 . 4 V 0 . 7 ~ - 0 . 7 V 0 . 7 ~ - 1 . 0 V 5 0 ) 2 - m

c A 0 m (

- 5 0

- 1 0 0 - 1 . 0 - 0 . 5 0 . 0 0 . 5 1 . 0

E ( V ) v s . H g / H g S O 2 4

2+ + Figure 4.2: The cathodic treatment of GCP4000 in 0.5 M VO +0.5 M VO2 in 2.5 M total sulphate/bisulphate solution

Five CV scans within the same range of 0.7∼0.2 V were performed after the electrochemical treatment of GC4000 and the fifth scans are presented in Figures 4.3 and 4.4. In both figures, the peak potentials remain almost the same. The peak current densities increase when the oxidation potential limit is changed from 0.7 to 1.3 V and then decrease slightly when the potential is further extended to 1.6 V. However, on the cathodically scanned GCP4000 electrodes, the peak current densities increase firstly when the lower potential limit is decreased from 0.2 to −0.1 V, then remain almost the same from −0.1 to −0.7 V and finally increase greatly when the negative potential limit reaches −1.0 V. These results show similar trends with those reported by Bourke [240, 247, 249] and are believed to be influenced by two main factors, i.e., surface area and oxygen functional groups, which will be discussed in the following sections. It needs to point out that the formation of vanadium complexes [150,250] can also affect the CV results especially 4. Vanadium redox kinetics on various electrode surfaces 131

2+ + with in-situ anodic treatment in concentrated mixtures of VO /VO2 electrolyte. In contrast to other anodic treatments such as the application of electrochemical potential pulses [240, 247, 249] however, in the present study, the pre-treatment used involved only one CV scan to a high potential, followed by reversal to a lower potential limit and finishing at the OCV. Any vanadium complexes that may have formed at higher potentials are expected to have been reduced to some extent therefore. Furthermore, the CV pretreatment and measurements were run at a scan rate of 100 mV·s-1 and the electrode rested for 2 min between scans. Any residual vanadium complexes formed at high potential should have diffused into the bulk solution and should not therefore affect the reaction kinetics significantly.

1 0 0 G C P 4 0 0 0 0 . 2 ~ 0 . 7 V 0 . 2 ~ 1 . 0 V 5 0 0 . 2 ~ 1 . 3 V 0 . 2 ~ 1 . 6 V ) 2 -

m 0 c

A m (

- 5 0

- 1 0 0 0 . 2 0 . 3 0 . 4 0 . 5 0 . 6 0 . 7 E ( V ) v s . H g / H g S O 2 4

Figure 4.3: The CV of GCP4000 after anodic treatment in 0.5 M VO2++0.5 M + VO2 in 2.5 M total sulphate/bisulphate solution

SEM was carried out to study the surface morphology of the GCP electrodes after electrochemical treatment, as shown in Figure 4.5. The inserted lower magniﬁcation image in Figure 4.5a shows obvious straight lines which are the 4. Vanadium redox kinetics on various electrode surfaces 132

1 0 0 G C P 4 0 0 0 0 . 7 ~ 0 . 2 V 0 . 7 ~ - 0 . 1 V 5 0 0 . 7 ~ - 0 . 4 V 0 . 7 ~ - 0 . 7 V

) 0 . 7 ~ - 1 . 0 V 2 - m c 0 A

m (

- 5 0

- 1 0 0 0 . 2 0 . 3 0 . 4 0 . 5 0 . 6 0 . 7 E ( V ) v s . H g / H g S O 2 4

Figure 4.4: The CV of GCP4000 after cathodic treatment in 0.5 M VO2++0.5 M + VO2 in 2.5 M total sulphate/bisulphate solution scratches produced during the SiC polishing. From the higher magnification image of Figure 4.5b, it can be seen that these scratches are more susceptible to corrosion than the planar region. Glassy carbon has a microstructure consisting of tangled curved non-graphitic carbon ribbons [251,252] and polishing can introduce structural defects such as by disrupting the sp2 C conjugation in the fragments of carbon planes, creating corrugations in the nanoscales and exposing more edge sites leading to higher surface energy and enhanced electroactivity than just basal and clean surfaces. This is in agreement with other studies of active sites on various carbon electrodes in vanadium solutions [136,150,234,253,254]. Moreover, it can be clearly seen that corrosion has gradually expanded over the whole surface of the GCP electrodes in line with the enhancement of the oxidation potential limit, especially when it reached 1.3 V. This was also observed by Neffe [243] and Liu [255] in the electrochemical corrosion study of glassy carbon in sulfuric 4. Vanadium redox kinetics on various electrode surfaces 133 acid solution [243] and graphite electrodes in VO2+ solutions [255]. In the case of reduction treatment, no significant changes of roughness can be found. It can thus be concluded that electrochemical oxidation can increase the surface area of the electrodes even at potentials before the oxygen evolution potential while reduction does not cause further physical variation to the oxidized surface. Defects derived from the mechanical polishing process are more likely to be electrochemically attacked. X-ray photoelectron spectroscopy (XPS) was performed on the electrodes after anodic treatment such as GCP4000 0.7V and GCP4000 1.6V and the one after cathodic treatment including GCP4000 0.2V and GCP4000 −1.0V. The high−resolution C1s and O1s spectra are displayed in Figures 4.6∼4.9 , respectively. Six peaks are deconvoluted in the C1s spectra at around 284.48 eV, 284.78 eV, 286.28 eV, 287.78 eV, 289.08 eV and 290.98 eV, which arise from C=C [256], C- C[256],C-OR (including C-OH and C-O-C) [257], C=O [257], COOR (including COOH and COOC) [257] and π − π* shakeup satellite contributions [258] respectively. The O1s signal is also resolved into three peaks at approximately 532.68 eV (alcohol and phenyl groups) [259], 531.08 eV(carbonyl groups and C-O-C) [257,259] and 534.08 eV(carboxyl groups) [259]. The calculated percentages of the C1s and O1s as well as the atomic ratios are listed in Tables 4.3∼4.5. It should be pointed out that the values of O1s may not be accurate due to the complexity of the chemical status of O1s under different circumstances [257–262], but at least, they can show a general trend. For better comparison, the atomic percentage of different types of functional groups are also plotted against the treating limiting potential, as shown in Fig- ure 4.10. It needs to be mentioned that the percentage of O1s in Figure 4.10b was calculated by the functional group content multiplied by corresponding O/C ratio. For example, the percentage of alcohol and phenyl groups of GCP4000 0.7V is 4. Vanadium redox kinetics on various electrode surfaces 134 equal to 65.65% times 0.11. 4. Vanadium redox kinetics on various electrode surfaces 135 1.0 V 0.7 V and (i) GCP4000 1.6 V and reduced after oxidization to 1.6V by CV in 2.5 M sulfate solution after oxidation by CV scans (a) 0.4 V, (h) GCP4000 4.5+ 1.3V, (d) GCP4000 0.1 V, (g) GCP4000 1.0V, (c) GCP4000 0.2V, (f) GCP4000 0.7 V, (b)GCP4000 Figure 4.5: SEM images ofGCP4000 GCP electrodes treated inscans 1 of M (e) V GCP4000 4. Vanadium redox kinetics on various electrode surfaces 136

( a ) C 1 s 0 . 2 ~ 0 . 7 V

C = C ) . u . a (

y t i s n

e t C - C n I C - O R

p - p * C O O H C = O

2 9 2 2 9 0 2 8 8 2 8 6 2 8 4 2 8 2 B i n d i n g e n e r g y ( e V )

( b ) C 1 s 0 . 2 ~ 1 . 6 V )

. C = C u . a (

y t

i C - C

s n e

t C - O R n I

C O O H C = O p - p *

2 9 2 2 9 0 2 8 8 2 8 6 2 8 4 2 8 2 B i n d i n g e n e r g y ( e V )

Figure 4.6: Carbon 1s region of GCP4000 after anodic treatment (a) GCP4000 0.7V and (b) GCP4000 1.6V 4. Vanadium redox kinetics on various electrode surfaces 137

( a ) C 1 s 0 . 7 ~ 0 . 2 V

C = C ) . u . a (

y t i

s C - C n e t n I C - O R

C O O H C = O p - p *

2 9 2 2 9 0 2 8 8 2 8 6 2 8 4 2 8 2 B i n d i n g e n e r g y ( e V )

( b ) C 1 s 0 . 7 ~ - 1 . 0 V

C = C ) . u . a (

y t i

s C - C n e t n I C - O R

C O O H p - p * C = O

2 9 2 2 9 0 2 8 8 2 8 6 2 8 4 2 8 2 B i n d i n g e n e r g y ( e V )

Figure 4.7: Carbon 1s region of GCP4000 after cathodic treatment (a) GCP4000 0.2V and (b) GCP4000 -1.0V 4. Vanadium redox kinetics on various electrode surfaces 138

( a ) O 1 s 0 . 2 ~ 0 . 7 V a ) . u . a (

y t i

s n e t

n b I g

5 3 6 5 3 4 5 3 2 5 3 0 5 2 8 B i n d i n g e n e r g y ( e V )

( b ) O 1 s 0 . 2 ~ 1 . 6 V ) . a u . a (

y t i

s b n e

t g n I

5 3 6 5 3 4 5 3 2 5 3 0 5 2 8 B i n d i n g e n e r g y ( e V )

Figure 4.8: Oxygen 1s region of GCP4000 after anodic treatment (a) GCP4000 0.7V and (b) GCP4000 1.6V 4. Vanadium redox kinetics on various electrode surfaces 139

( a ) O 1 s 0 . 7 ~ 0 . 2 V ) . a u . a (

y t i

s g

n b e t n I

5 3 6 5 3 4 5 3 2 5 3 0 5 2 8 B i n d i n g e n e r g y ( e V )

( b ) O 1 s 0 . 7 ~ - 1 . 0 V

a ) . u . g a (

t b i

s n e t n I

5 3 6 5 3 4 5 3 2 5 3 0 5 2 8 B i n d i n g e n e r g y ( e V )

Figure 4.9: Oxygen 1s region of GCP4000 after cathodic treatment (a) GCP4000 0.2V and (b) GCP4000 -1.0V

From Figure 4.10, it can be found that when the oxidizing potential changes 4. Vanadium redox kinetics on various electrode surfaces 140

Table 4.3: Binding energy (BE) values (eV) of C1s for various pretreated glassy carbon plates (atom percentage values given in parentheses)

BE Samples (eV) 0.2∼0.7 V 0.2∼1.6V 0.7∼0.2 V 0.7∼-1.0 V C=C 284.48 (42.09%) 284.48 (26.44%) 284.48 (30.44%) 284.48 (36.35%) C-C 284.78 (34.25%) 284.78 (42.23%) 284.78 (43.62%) 284.78 (40.23%) C-OR 286.28 (16.12%) 286.38 (20.79%) 286.38 (16.02%) 286.38 (13.04%) C=O 287.78 (2.39%) 287.78 (4.50%) 287.78 (3.48%) 287.78 (3.15%) COOR 289.08 (1.68%) 288.98 (3.36%) 288.98 (4.17%) 288.98 (3.55%) π-π* 290.98 (3.47%) 290.88 (2.66%) 291.08 (2.27%) 290.98 (3.68%)

Table 4.4: Binding energy (BE) values (eV) of O1s for various pretreated glassy carbon plates (atom percentage values given in parentheses)

BE Samples (eV) 0.2∼0.7 V 0.2∼1.6 V 0.7∼0.2 V 0.7∼-1.0 V α1 532.68 (65.65%) 532.68 (36.04%) 532.38 (40.34%) 532.48 (48.69%) β2 531.48 (22.9%) 531.68 (37.59%) 531.88 (32.43%) 531.28 (24.09%) γ3 533.68 (11.45%) 533.28 (26.38%) 533.38 (27.22%) 533.58 (27.21%)

1α arises from alcohol and phenyl groups; 2 β arises from carbonyl groups and C-O-C; 3 γ arises from carboxyl groups.

Table 4.5: Atom ratio of O/C for various pretreated glassy carbon plates

Atom Ratio 0.2∼0.7 V 0.2∼1.6 V 0.7∼0.2 V 0.7∼-1.0 V

O/C 0.11 0.20 0.18 0.14 4. Vanadium redox kinetics on various electrode surfaces 141 from 0.7 to 1.6 V, the total amount of oxygen functional groups goes up remarkably, and then decreases after reduction. Speciﬁcally, the carbonyl groups, C-O-C and carboxyl groups increase signiﬁcantly during oxidation, and after reduction a great decrease in carbonyl groups and C-O-C can be observed. Furthermore, compared to GCP4000 0.7V, GCP4000 -1.0V shows a great increase in COOH while other groups change slightly. As suggested above, electrochemical oxidation of GCP electrodes at potentials exceeding 0.7 V can lead to an increase in surface roughness which starts mainly at the mechanically produced defect sites, accompanied by the formation of oxygen functional groups such as C-OH, C-O-C, C=O as well as COOH. In addition, electrochemical reduction can decrease the amount of the oxygen functional groups and largely reduces groups such as C=O and C-O-C. Combined with the CV scan results after treatment, we can assume that the increased peak current densities at treatment potentials no greater than 1.3 V can be attributed to the increasing surface area. However, when the potential is increased to 1.6 V, the inhibiting role of oxygen functional groups in the electrochemical activity of the glassy carbon

2+ + electrode towards the VO /VO2 reactions becomes more evident, resulting in a decline in peak current densities. Furthermore, since the surface roughness of the electrode remained similar to that oxidized to 1.6 V, the enhanced current densities of the reduced GCP electrodes are considered to be caused by the decrease of oxygen functional groups especially C=O and C-O-C. It is known that controversy always exists in regard to the role of oxygen functional groups in the electro-activity of carbonaceous electrodes towards the vanadium redox couples. Sun and Skyllas-Kazacos [95] thermally treated graphite felt and attributed the improved cell eﬃciency of the VRFB to the increase in the C-O and C=O functional groups. They also found that during the acid treatment of graphite felt electrodes, that the oxygen functional groups produced provided active 4. Vanadium redox kinetics on various electrode surfaces 142

2 5 ( a ) R e d u c t i o n O x i d a t i o n

2 0

) 1 5 % (

g C - O R a t

n C = O

e 1 0 c C O O R r e P

0 - 1 . 0 - 0 . 5 0 . 0 0 . 5 1 . 0 1 . 5 2 . 0 E ( V ) v s . H g / H g S O 2 4

1 0 . 0 ( b ) R e d u c t i o n O x i d a t i o n

7 . 5 ) % (

g 5 . 0 a t

n e c r e P 2 . 5 a b g 0 . 0 - 1 . 0 - 0 . 5 0 . 0 0 . 5 1 . 0 1 . 5 2 . 0 E / V v s . H g / H g S O 2 4

Figure 4.10: The atom percentage of (a) C1s and (b) O1s in various chemical states on the glassy carbon plate surfaces after CV scan treatments sites for both the positive and negative half-cell charge and discharge reactions [113]. They also suggested that the enhanced cell efficiencies were a result of improved 4. Vanadium redox kinetics on various electrode surfaces 143 wettability associated with the formation of these functional groups. However, in their investigation on overcharge of the VRFBs, a decay in the cell efficiency was accompanied by an increasing amount of higher oxidation-state groups, i.e. C=O, COOH and COOR [104]. Yue and co-workers hydroxylated carbon fibers by acid treatment, increasing the content of hydroxyl groups from the original 3.8%

2+ 3+ 2+ + to 14.3% and obtained enhanced activity towards the V /V and VO /VO2 reactions [263]. In a study of electrochemically reduced graphene oxide by Li et al, they proposed that C=O functional groups were more likely to provide reactive sites for the vanadium reactions than C-O groups [140]. The discrepancy in the findings relating to the effects of functional groups and the specific role of oxygen functional groups can be attributed to several reasons:(i) the complexity of oxidation that introduces not only oxygen functional groups but also other variables, (ii) the difficulty in precise deconvolution of the XPS results of O1s and

2+ + probably (iii) various effects of the same functional groups on VO /VO2 and V2+/V3+ reactions and (iv) errors associated with changes in the surface chemistry when the electrodes are removed from solution and placed under inert atmosphere in the XPS instrument. The latter is likely to cause significant changes in surface chemistry that could explain the large inconsistencies in the results reported by various groups. Based on our experimental findings however, after eliminating the effect of surface roughness, partial reduction of the carbon surface is definitely associated with an increased peak current, suggesting that C=O and C-O-C may

2+ + be associated with the reduced electrochemical activity for the VO /VO2 redox couple reaction. 4. Vanadium redox kinetics on various electrode surfaces 144

4.3.2 Eﬀects of electrochemical treatment on glassy car-

bon electrodes

A series of CV scans on glassy carbon electrodes polished with 600, 1200 and 400 grit SiC paper is shown in Figure 4.12a for the oxidizing potential limit varying from 0.7 to 1.6 V. As can be seen, the peak current densities of these GC electrodes decrease in the order of GC600 > GC1200 > GC4000 while the peak potential separation becomes wider with decreased surface roughness, providing clear evidence that increased roughness can greatly enhance the surface activity of electrodes towards

2+ + the VO /VO2 reactions. In fact, the polishing process is accompanied by a change in the microstructure and roughness as well as removing adsorbed species (which are absorbed before contacting the electrolyte). It is well-known in microthrowing (electroplating) that the diffusion can take place within the groove when the depth is long enough whereas the electroactive species more likely to diffuse only to the interfacial corrugations when the depth is less than the thickness of the diffusion layer [264, 265]. In other words, with thinner roughness depth than the diffuse layer, less active species are absorbed and thus will be rapidly depleted so that concentration overpotential will limit reactions at such sites [265]. In addition, the size of the pores that are created by polishing can influence the electroactive surface area as well. When the pores are too fine, they cannot allow ions to get in while large pores might absorb ions oriented along the longest dimension of the pore, leading to a longer diffusion path [266]. Therefore, polishing to alter the surface microstructure by producing ”right” groove depths and pore sizes can contribute to increased electroactive area and improve the surface activity of the electrode (See Figure 4.11). CVs were also performed on these electrodes over different potential ranges to oxidize and reduce the glassy carbon (GC) surface and the results are presented in 4. Vanadium redox kinetics on various electrode surfaces 145

Figure 4.11: The scheme of the relationship between the groove depth/pore size and diﬀuse layer

Figure 4.12b. For the oxidation of all GC electrodes, the peak current densities

2+ + of the VO /VO2 redox couple do not decrease until the upper potential limit is greater than 1.0 V. This is a result of the competition between the two main factors, i.e. roughness and functional groups, in the process of electrochemical oxidation. Although electrochemical oxidation can increase the roughness of the surface of electrodes and thus increase the surface area, some oxygen functional groups introduced during the oxidation may in turn, hinder the electron transfer . When the upper potential limit reaches 1.0 V therefore, the inhibiting eﬀect of some oxygen functional groups could become the dominant factor rather than the contribution of increased surface area. Meanwhile, electrochemical reduction is expected to retain surface roughness of the oxidized electrodes, and only reduces the oxygen functional groups. Hence, the observed increase in peak current densities after scanning to potentials more negative than -0.1V, proves that reduction of oxygen functional

2+ + groups can be beneficial for the VO /VO2 redox reactions. A distinct feature of these figures is that the peak potential separation for GC600 and GC1200 remain almost the same regardless of the treatment potential, demonstrating that the combination effect of oxygen functional groups and surface roughness variation during oxidation and reduction is more significant for smoother GC electrodes. Besides,

2+ + the diﬀerent behaviour of GCP4000 and GC4000 in VO /VO2 solution under 4. Vanadium redox kinetics on various electrode surfaces 146

( a ) 6 0

3 0 2 - m c A

m 0 / j

- 3 0 G C 6 0 0 G C 1 2 0 0 G C 4 0 0 0

- 6 0 0 . 0 0 . 5 1 . 0 1 . 5 2 . 0 E / V v s . H g / H g S O 2 4

( b ) 6 0

3 0 ) 2 - m c A

m 0 (

- 3 0 G C 6 0 0 G C 1 2 0 0 G C 4 0 0 0 - 6 0 - 1 . 0 - 0 . 5 0 . 0 0 . 5 1 . 0 E ( V ) v s . H g / H g S O 2 4

Figure 4.12: The pretreatment CV scans at 100 mV·s−1 by varying (a) upper potential limits and (b) lower potential limits at glassy carbon electrodes GC600, GC1200 and GC4000 the same treatment conditions is likely due to a different surface microstructure so the same polishing and electrochemical oxidation process introduced much fewer 4. Vanadium redox kinetics on various electrode surfaces 147 defects (mechanical) or active sites (electrochemical) on GC4000 than GCP4000, leading to considerable variation in the ratio of the true to apparent surface area and to a different electrochemical activity. After each pre-treatment was applied, EIS measurements were performed and the Nyquist impedance plots were modeled according to the Randles circuit as illustrated in Figures 4.13-4.15. All the curves consisted of a straight line associated with the Warburg impedance in the lower frequency range and a circular arc whose radius depends on the charge transfer resistance (Rct) in the higher frequency region. The Rct was quantitatively estimated and plotted against the treatment potential limit in Figure 4.16. The Rct of the electrodes treated by CV under the same potential range are still in the order GC600 < GC1200 < GC4000, which proves again that rougher surfaces can facilitate charge transfer at the electrode/electrolyte interface. The results also show that Rct increases along with the oxidation potential and a rapid growth can be observed when the potential changed from 1.3 to 1.6 V. In other words, an oxidation by the CV scan to 1.6 V can significantly deactivate the electrode. The following reduction treatment demonstrated its effectiveness in recovering and accelerating the charge transfer

2+ + process in the VO /VO2 electrolyte especially when the potential reaches -0.7 V.

It is also noticeable from the slope of the Rct variation with the treatment potential. At the same time, however, the slopes in oxidation are always slightly larger than for reduction, probably because oxidation can lead to roughness variation of the electrode surface, which can accelerate the charge transfer processes. In this case, the hindered charge transfer processes on the GC surface resulted mainly from the suppression effects of the surface functional groups. The recovery and accelerating effect is attributed mainly to the reduction of certain oxygen functional groups because reduction retains the roughness of the pre-oxidised GC electrodes. Another interesting feature is that the influence of electrochemical treatment is most evident 4. Vanadium redox kinetics on various electrode surfaces 148 for GC4000. This is probably because GC4000 has a smoother surface than GC600 and GC1200. Fewer defects by mechanical polishing are therefore exposed during the treatment so that less roughness variation can occur. Hence, the inhibition role of oxygen functional groups becomes more critical. 4. Vanadium redox kinetics on various electrode surfaces 149

2 5 ( a ) G C 6 0 0 0 . 2 ~ 0 . 7 V 0 . 2 ~ 1 . 0 V 0 . 2 ~ 1 . 3 V 2 0 0 . 2 ~ 1 . 6 V

1 5 ) W (

Z 1 0 -

0 0 1 0 2 0 3 0 4 0 5 0 Z ' ( W )

2 5 ( b ) G C 6 0 0 0 . 7 ~ 0 . 2 V 0 . 7 ~ - 0 . 1 V 0 . 7 ~ - 0 . 4 V 2 0 0 . 7 ~ - 0 . 7 V 0 . 7 ~ - 1 . 0 V

) 1 5 W (

" Z

- 1 0

0 0 1 0 2 0 3 0 4 0 5 0 Z ' ( W )

Figure 4.13: The modelled Nyquist impedance plots of GC600 after (a)anodic and (b)cathodic treatment 4. Vanadium redox kinetics on various electrode surfaces 150

2 5 ( a ) G C 1 2 0 0 0 . 2 ~ 0 . 7 V 0 . 2 ~ 1 . 0 V 0 . 2 ~ 1 . 3 V 2 0 0 . 2 ~ 1 . 6 V

1 5 ) W ( " Z

- 1 0

0 0 1 0 2 0 3 0 4 0 5 0

Z ' ( W )

2 5 ( b ) G C 1 2 0 0 0 . 7 ~ 0 . 2 V 0 . 7 ~ - 0 . 1 V 0 . 7 ~ - 0 . 4 V 2 0 0 . 7 ~ - 0 . 7 V 0 . 7 ~ - 1 . 0 V

1 5 ) W ( " Z

- 1 0

0 0 5 1 0 1 5 2 0 2 5 3 0 3 5 4 0 4 5 5 0 Z ' ( W )

Figure 4.14: The modelled Nyquist impedance plots of GC1200 after (a)anodic and (b)cathodic treatment 4. Vanadium redox kinetics on various electrode surfaces 151

1 0 0 ( a ) G C 4 0 0 0 0 . 2 ~ 0 . 7 V 0 . 2 ~ 1 . 0 V 0 . 2 ~ 1 . 3 V 0 . 2 ~ 1 . 6 V 7 5 ) W (

5 " 0 Z

2 5

0 0 5 0 1 0 0 1 5 0 2 0 0

Z ' ( W )

1 0 0 ( b ) G C 4 0 0 0 0 . 7 ~ 0 . 2 V 0 . 7 ~ - 0 . 1 V 0 . 7 ~ - 0 . 4 V 7 5 0 . 7 ~ - 0 . 7 V 0 . 7 ~ - 1 . 0 V ) W (

5 " 0 Z -

2 5

0 0 5 0 1 0 0 1 5 0 2 0 0 Z ' ( W )

Figure 4.15: The modelled Nyquist impedance plots of GC4000 after (a)anodic and (b)cathodic treatment 4. Vanadium redox kinetics on various electrode surfaces 152

1 5 0 R e d u c t i o n O x i d a t i o n

1 0 0 G C 6 0 0

) G C 1 2 0 0 W (

G C 4 0 0 0

t c R 5 0

0 - 1 . 0 - 0 . 5 0 . 0 0 . 5 1 . 0 1 . 5 E ( V ) v s . H g / H g S O 2 4

Figure 4.16: The estimated charge transfer resistance of GC600, GC1200 and GC4000

Five scans within the range of 0.7∼0.2 V were performed after the electrochemical treatment and the ﬁfth scans are presented in Figures 4.17-4.19. Consistent with the Rct values discussed above, the CV curves exhibit obvious decreases in electrochemical activity after oxidation and enhanced performance after reduction. This supports the suggestion by Bourke and co-workers [240] that the kinetics of

2+ + VO /VO2 reaction is hindered by anodic polarization but improved by cathodic treatment. Again, the greatest enhancement in either electrochemical reversibility or peak current densities by oxidation and then reduction is observed on GC4000. The reason is considered to be the same as discussed above. 4. Vanadium redox kinetics on various electrode surfaces 153

7 5 ( a ) G C 6 0 0 0 . 2 ~ 0 . 7 V 5 0 0 . 2 ~ 1 . 0 V 0 . 2 ~ 1 . 3 V 0 . 2 ~ 1 . 6 V

) 2 5 2 - m c A

m 0 (

- 2 5

- 5 0

0 . 2 0 . 3 0 . 4 0 . 5 0 . 6 0 . 7 E ( V ) v s . H g / H g S O 2 4

7 5 ( b ) G C 6 0 0 0 . 7 ~ 0 . 2 V 0 . 7 ~ - 0 . 1 V 5 0 0 . 7 ~ - 0 . 4 V 0 . 7 ~ - 0 . 7 V 0 . 7 ~ - 1 . 0 V

) 2 5 2 - m c A

m 0 (

- 2 5

- 5 0

0 . 2 0 . 3 0 . 4 0 . 5 0 . 6 0 . 7 E ( V ) v s . H g / H g S O 2 4

Figure 4.17: The ﬁfth scan curve at scan rate of 100 mV·s−1 scanned ﬁrstly to 0.7 V and then to 0.2 V on glassy carbon electrodes GC600 in 0.5 M VO2++0.5 M + VO2 in 2.5 M total sulphate/bisulphate solution after pretreatment of CV scans within various potential ranges 4. Vanadium redox kinetics on various electrode surfaces 154

7 5 ( a ) G C 1 2 0 0 0 . 2 ~ 0 . 7 V 0 . 2 ~ 1 . 0 V 5 0 0 . 2 ~ 1 . 3 V 0 . 2 ~ 1 . 6 V ) 2

- 2 5 m c A

m (

j 0

- 2 5

- 5 0 0 . 2 0 . 3 0 . 4 0 . 5 0 . 6 0 . 7 E / V v s . H g / H g S O 2 4

7 5 ( b ) G C 1 2 0 0 0 . 7 ~ 0 . 2 V 5 0 0 . 7 ~ - 0 . 1 V 0 . 7 ~ - 0 . 4 V 0 . 7 ~ - 0 . 7 V

) 2 5 0 . 7 ~ - 1 . 0 V 2 - m c A

m 0 (

- 2 5

- 5 0

0 . 2 0 . 3 0 . 4 0 . 5 0 . 6 0 . 7 E ( V ) v s . H g / H g S O 2 4

Figure 4.18: The ﬁfth scan curve at scan rate of 100 mV·s−1 scanned ﬁrstly to 0.7 V and then to 0.2 V on glassy carbon electrodes GC1200 in 0.5 M VO2++0.5 M + VO2 in 2.5 M total sulphate/bisulphate solution after pretreatment of CV scans within various potential ranges 4. Vanadium redox kinetics on various electrode surfaces 155

7 5 ( a ) G C 4 0 0 0 0 . 2 ~ 0 . 7 V 5 0 0 . 2 ~ 1 . 0 V 0 . 2 ~ 1 . 3 V 0 . 2 ~ 1 . 6 V

) 2 5 2 - m c A

m 0 (

- 2 5

- 5 0

0 . 2 0 . 3 0 . 4 0 . 5 0 . 6 0 . 7 E ( V ) v s . H g / H g S O 2 4

7 5 ( b ) G C 4 0 0 0 0 . 7 ~ 0 . 2 V 5 0 0 . 7 ~ - 0 . 1 V 0 . 7 ~ - 0 . 4 V 0 . 7 ~ - 0 . 7 V

) 2 5 0 . 7 ~ - 1 . 0 V 2 - m c A

m 0 (

- 2 5

- 5 0

0 . 2 0 . 3 0 . 4 0 . 5 0 . 6 0 . 7 E ( V ) v s . H g / H g S O 2 4

Figure 4.19: The ﬁfth scan curve at scan rate of 100 mV·s−1 scanned ﬁrstly to 0.7 V and then to 0.2 V on glassy carbon electrodes GC4000 in 0.5 M VO2++0.5 M + VO2 in 2.5 M total sulphate/bisulphate solution after pretreatment of CV scans within various potential ranges 4. Vanadium redox kinetics on various electrode surfaces 156

4.3.3 Eﬀect of electrolyte SOC

Chen and coworkers [238] carried out a series of experiments on the kinetics of vanadium redox couples in diluted solutions of different SOCs and found that both rate constant and diffusion coefficient change with SOC. The effect of electrolyte SOC is believed to become more significant in concentrated vanadium solutions because the formation of vanadium complexes and the oxidizing ability of V(V) solutions cannot be ignored.

2+ + The CV curves of VO /VO2 reaction on GC4000 at a scan rate of 100

-1 2+ 2+ + mV·s within the potential range -0.8∼1.3 V in VO and VO /VO2 solutions (corresponding to the positive electrolyte of the vanadium redox battery at 0% and 50% SOC respectively) are presented in Figure 4.20. As is shown in Figure 4.20a,

2+ + 2+ the VO to VO2 oxidation peak occurs in VO solution at a potential of 0.90 V and the reduction at -0.60 V (4E = 1.5V). By comparison, within the same

2+ + potential range, the oxidation and reduction peaks in 0.5 M VO +0.5 M VO2 in 2.5 M total sulphate/bisulphate solution occur at 0.64 V and -0.07 V respectively (4E = 0.71 V). The smaller peak potential separation indicates that the reversibility

2+ + 2+ + of VO /VO2 redox couple reactions in the VO /VO2 solution is much better than that in VO2+ solution. It seems to disagree with our previous experimental results where oxidation leads to decreased electrochemical activity. In fact, in the discussion of electrodes by diﬀerent polishing treatment and electrochemical treatment, it is pointed out that both surface area and oxygen functional groups can aﬀect the electrochemical activity and certain oxygen functional groups are

2+ + unfavourable for the redox reaction of VO /VO2 . The results here suggest that

+ 2+ + the VO2 ions in the VO /VO2 solution altered the surface states of the glassy carbon electrode. The electrochemical activity was improved after the partial oxidation as the increase of active sites is more dominant in the competition 4. Vanadium redox kinetics on various electrode surfaces 157 between the increased surface area and the undesired oxygen functional group generated by oxidation. 4. Vanadium redox kinetics on various electrode surfaces 158

5 0 ( a ) G C 4 0 0 0 i n V O 2 +

2 5 ) 2 - m

c 0 A

m (

- 2 5 0 . 9 0 V - 0 . 6 0 V

- 5 0 - 1 . 0 - 0 . 5 0 . 0 0 . 5 1 . 0 1 . 5

E ( V ) v s . H g / H g S O 2 4

5 0 ( b ) G C 4 0 0 0 i n V O 2 + / V O + 2

2 5 ) 2 - m

c 0 A

m (

- 2 5 0 . 6 4 V

- 0 . 0 7 V - 5 0 - 1 . 0 - 0 . 5 0 . 0 0 . 5 1 . 0 1 . 5

E ( V ) v s . H g / H g S O 2 4

Figure 4.20: The ﬁfth CV curve at scan rate of 100 mV·s-1 scanned ﬁrstly to 1.3 V and then to - 0.3 V on GC4000 electrodes in (a) 1 M VO2+ in 2.5 M 2+ + sulphate/bisulphate solution and (b) 0.5 M VO +0.5 MVO2 in 2.5 M total sulphate/bisulphate solution

2+ + This is further demonstrated by the Nyquist plots for the VO /VO2 redox 4. Vanadium redox kinetics on various electrode surfaces 159

2+ 2+ + couple on the GC4000 electrode in solutions of pure VO and VO /VO2 as shown in Figure 4.21. Both plots show the same features with capacitive semicircles in the high-frequency region and straight lines in the low-frequency region [267]. The diﬀerences between the diameters of the semicircles in Figure 4.20a and b clearly show that the charge transfer resistance in VO2+ solution is signiﬁcantly

2+ + larger than that in the VO /VO2 solution [267]. The variation of the Warburg tails also indicates the distinguishing diﬀusion rate in diﬀerent solutions. It has

+ been veriﬁed by various researchers [96–102] that VO2 ions in sulfuric acid can exist

+ − 3− − 4+ 2+ 2+ as VO2 , VO2SO4 , VO2(SO4)2 , VO3 ,V2O3 and V2O4 , etc. while VO can

2+ 2+ 2+ + exist as VO , VO(H2O)5 , VO(SO4)2 and VOHSO4 . Furthermore, as shown

2+ + 2+ by Kazacos et al [268] different VO /VO2 complexes can form in mixed VO + and VO2 solutions which could account for the varied diffusion behaviour in the two solutions. As mentioned before, another possible reason for the different electrode response

2+ + to the VO /VO2 redox couple reactions in concentrated vanadium solution of + diﬀerent oxidation states or SOC, is the oxidizing eﬀect of the VO2 solutions. This was demonstrated by CV experiments performed with glassy carbon electrodes

+ polished with 4000-grit sandpaper with and without immersion in VO2 solution. The glassy carbon electrode was polished with 4000-grit sandpaper followed with

+ dipping in VO2 solution for 10s and rinsing with distilled water and its CV response in VO2+ solution was compared to that subjected only to mechanical polishing. The

2+ + CV curves in Figure 4.22a show that the peak potentials of the VO /VO2 redox + reactions peaks shift with the VO2 dipping pretreatment. Moreover, the charge transfer resistance indicated by the diameter of the semicircles in Figure 4.22b

+ shows an obvious increase after the GC4000 electrode was treated with VO2 for + 10 s, which reveals again the strong oxidizing eﬀect of VO2 solution that changes the surface functional groups and the electrochemical activity of the electrode in 4. Vanadium redox kinetics on various electrode surfaces 160

2+ + the VO /VO2 solution.

6 0 0 ( a ) G C 4 0 0 0 i n V O 2 +

5 0 0

4 0 0 ) W (

" 3 0 0 Z

2 0 0

1 0 0

0 0 5 0 0 1 0 0 0 1 5 0 0 Z ' ( W )

1 2 ( b ) G C 4 0 0 0 i n V O 2 + / V O + 2

1 0

8 ) W (

" 6 Z -

0 0 2 0 4 0 6 0 Z ' ( W )

Figure 4.21: The modelled Nyquist impedance plots of GC4000 after CV scans in 2+ 2+ + (a) 1 M VO in 2.5 M sulfate solution and (b) 1M VO /VO2 in 2.5 M sulfate solution 4. Vanadium redox kinetics on various electrode surfaces 161

1 0 0 0 ( b ) G C 4 0 0 0 G C 4 0 0 0 d i p p e d i n V ( V ) s o l u t i o n f o r 1 0 s

7 5 0 ) W ( 5 0 0 " Z

2 5 0

0 0 5 0 0 1 0 0 0 1 5 0 0 2 0 0 0 Z ' ( W )

5 0 ( a ) - 0 . 6 0 V 0 . 9 0 V

2 5 ) 2 - m

c 0 A

m (

- 2 5 - 0 . 7 1 V 0 . 9 2 V

G C 4 0 0 0 G C 4 0 0 0 d i p p e d i n V O + s o l u t i o n f o r 1 0 s 2 - 5 0 - 1 . 0 - 0 . 5 0 . 0 0 . 5 1 . 0 1 . 5 E ( V ) v s H g / H g S O 2 4

Figure 4.22: (a) The ﬁfth CV scan curves at scan rate of 100 mV·s-1 scanned ﬁrstly to 1.3 V and then to −0.3 V and (b) the corresponding modeled Nyquist plots in 1 M VO2+ in 2.5 M sulfate solution on glassy carbon electrodes 4. Vanadium redox kinetics on various electrode surfaces 162

4.4 Summary

Electrochemical oxidation produced by scanning to high potentials can increase the surface roughness of glassy carbon material, while mechanical polishing produces defects that provide active sites to facilitate electrochemical oxidation. Formation of various C-O, C=O and COOH functional groups were characterized by XPS after the oxidation treatment. Both increased surface area and development of functional groups can inﬂuence the electrochemical behaviour of the glassy carbon electrode

2+ + towards the VO /VO2 redox reactions. Increasing surface roughness of glassy carbon plate electrodes during oxidation at potentials up to 1.3 V is advantageous for promoting electrochemical activity, while oxygen functional groups especially C-O-C and C=O, formed at higher potentials can yield inactive sites for the

2+ + VO /VO2 reaction. Thus during electrochemical oxidation, the two factors are in a competitive relationship that depends highly on the inherent surface status of the electrodes. When the electrode is smoother with fewer defects, the role of oxygen functional groups is more dominant than the variation of surface area, resulting in the deactivation of the electrodes, and vice versa. Cathodic treatment to negative potentials can reduce the oxygen functional groups but retain the surface roughness produced during oxidation. Consequently, electrochemical reduction treatment can always recover and even promote the activity of oxidized electrodes. When the electrochemical treatment is conducted under the same conditions, the rougher surfaces of the glassy carbon electrodes

2+ + possess better electrochemical activity towards the VO /VO2 reactions. These results highlight the importance of specifying the electrode preparation method used in electrode kinetic studies of vanadium redox couple reactions. Knowledge of the previous polishing and electrochemical treatment history is critical when describing or comparing the electrochemical behaviour of various 4. Vanadium redox kinetics on various electrode surfaces 163 electrodes in the vanadium electrolytes, especially when applying various catalytic coatings to enhance the kinetics of the vanadium redox couple reactions. The results presented here have shown that increases in peak current and reduced peak potential separation may simply be associated with the increased surface area provided by such coatings. Another factor that should not be neglected is the strong

+ oxidizing eﬀect of VO2 species. The surface status of electrodes will probably be varied in-situ when the electrochemical characterizations are carried out in

+ solutions containing concentrated VO2 . The present study has focussed on glassy carbon electrodes, however, further studies are still needed to better understand the role of surface area and functional groups on other carbonaceous electrode materials such as graphite plates, graphite or carbon fibers and on catalytic coatings such as carbon nanotubes so that the most appropriate treatment can be applied to enhance the kinetics of both the positive and negative half-cell reactions of the VRFB. This will be of particular significance in the novel cell architectures that replace the high surface area graphite felts with thin graphite papers and interdigitated flow fields in high power density flow battery designs. Chapter 5

Electrocatalysts for high power density VRFBs

This chapter provides two feasible approaches for modifying carbonaceous electrodes to reduce activation overpotential for vanadium redox reactions, i.e., decorating the

2− electrodes with non-noble metal oxides MoO3 and adding MoO4 into electrolyte. These two methods were employed in a single serpentine vanadium redox ﬂow cell and demonstrated great promise in the use of zero-gap cell structure with electrocatalysts for high power density VRFB applications. The content of this chapter is reproduced from a research paper by the author in ChemElectroChem with some supplementary information.

5.1 Problem Statement

As discussed previously, VRFB technology provides a promising alternative for large-scale energy storage and conversion (ESC) due to the independence of power rating and energy capacity as well as the elimination of cross-contamination issues. However, relatively high capital costs associated with the price of the

164 5. Electrocatalysts for high power density VRFBs 165 vanadium electrolyte and the membrane have to date limited its more extensive commercialization in countries where electricity is still relatively cheap. One approach to decreasing the cost of the battery is to increase power density so that the size of battery stack and corresponding material costs can be significantly reduced. The electrodes in VRFB systems do not directly participate as reactants in the cell reactions but provide the reaction sites for the electron transfer processes. The electrode activity and cell geometry govern the power density as well as the energy efficiency by influencing the magnitude of all components of the cell voltage, i.e., ohmic, concentration and activation polarization. In conventional VRFB cells that employ thick porous graphite felts, the high surface area of the graphite felt leads to very low effective current density at the electrode surface, so that activation polarization is minimal. This only applies if the felt is fully wetted however, so that the active surface area is maximised. Certain pretreatments have been shown to enhance the hydrophilicity and electrochemical activity of the graphite felt materials used in conventional porous electrode flow-through cells, but the thickness of these felts increases the anode-cathode distance and the ohmic resistance of the cell, reducing the power density. Traditional VRFB cell designs that use thick graphite felt electrodes are therefore limited by ohmic resistance that is strongly governed by the thickness of the graphite felt and the resulting anode-cathode distance. This can be reduced by increasing the compression of the felt, but this in turn increases the pressure drop and pumping energy required. The high pressure drop in the flow-through design (felt electrodes with flat bipolar plates) can cause a considerable energy loss by the pumps if higher current and power densities are required. In the zero-gap configuration thin carbon paper electrodes and serpentine or interdigitated flow channels are used to minimize the ohmic losses and enhance the mass transport at the same time [176,177,180]. 5. Electrocatalysts for high power density VRFBs 166

In fact, carbon paper is usually too low in permeability so it cannot be used in flow-through configurations but it is a good candidate in flow-by architecture as it has higher hydraulic conduction through the thickness direction when compared to carbon felt. This means the electrolyte can jump over the channel ribs to the electrode/membrane interface [179]. However, the reduced surface area of thin carbon paper, compared to the traditional 3∼5 mm thick porous graphite felt electrodes in VRFBs, can result in larger polarization especially at high current densities. Therefore, it is critically important to reduce the overpotential especially the charge transfer polarization of the electrode material for better performance at higher current densities in new zero-gap cell designs. Researchers have attempted to develop electrode modification and activation for electrodes for VRFBs including physical and/or chemical treatment, incorporation of nanoparticles and structural modification, as summarized in Chapter2. Unfortunately, only a few of these methods can actually provide a leap forward in power densities and be suitable for industrial scale production. Thermal treatment in air at 400 for 30 h [95] and chemical treatment in hot and concentrated HNO3 and H2SO4 [103] which were developed by Skyllas-Kazacos group in early 1990s are still regarded as the most effective and economic ways to activate the carbon electrode materials. Nevertheless, the extended heating and strong oxidizing and corrosive acid solutions in the above methods lack commercial appeal. Thermal, chemical and electrochemical treatments have been used to activate carbon felts and papers by introducing more surface oxide or hydroxide functional groups that are believed to provide active sites for the vanadium redox couple reactions. If the surface concentrations of the oxygen functional groups is too high however, adjacent groups will combine to release CO2 molecules that cause the carbonaceous material to disintegrate. An alternative method to increase the active sites on carbon electrodes is therefore needed to further enhance the electrochemical 5. Electrocatalysts for high power density VRFBs 167 activity of these materials without damaging the surface. Carbonaceous electrodes modified by metal or metal oxides have also been intensively investigated especially in recent ten years (See Table 5.1). Among them, Pacific Northwest National Laboratory (PNNL) first achieved great performance at high current density (beyond 100 mA·cm−2) by adding Bi3+ into the electrolyte. With the additive of Bi3+, the voltage efficiency can reach around 87% and 80.4% at 100 mA·cm−2 and 150 mA·cm−2 respectively. However, this single cell test was carried out in vanadium electrolyte in HCl supporting solution instead of conventional H2SO4 solution [163]. After that, the performance of electrodes modified by WO3 [120], W-

3+ doped Nb2O5 [121], TiO2 [269], CeO2 [124] and ZrO2 [180] and Sb additive [164] at current densities higher than 100 mA·cm−2 has been studied and the highest

−2 voltage eﬃciencies at 100 mA·cm was around 85% obtained on CeO2-graphite felt [124] and ZrO2-graphite felt [180].

Non-noble metal oxide MoO3 with diﬀerent crystalline forms (α-MoO3, β-MoO3,

ε-MoO3 and h-MoO3) has been widely used in various technological applications due to properties such as wide bandgap, easy transformation between different oxidation states, abundant oxygen vacancies on the surface and high oxygen evolution overpotential. [270–274] Especially in the field of catalysis, MoO3 has been used for various reactions including hydrogenation and dehydrogenation [275], epoxidation [276], isomerization [277], polymerization [278], etherification [279], addition and disproportionation [280]. In addition, molybdenum is a common impurity in vanadium electrolyte [281]. The procedure used to eliminate this element is quite complicated due to the similarity in properties between the molybdenum and vanadium. However, so far there is no reported investigation of the effect of

MoO3 or vanadium electrolytes containing molybdenum impurity on the redox reactions in VRFBs. Two approaches are described in this chapter to investigate the eﬀect of Mo in 5. Electrocatalysts for high power density VRFBs 168

2+ 3+ 2+ + catalyzing the V /V and VO /VO2 redox reactions: 1) synthesizing MoO3- 2− decorated carbon paper by soaking and calcination and 2) directly adding MoO4 into vanadium electrolyte. Both methods showed reduced activation overpoten-

2+ 3+ 2+ + tial for the V /V and VO /VO2 redox reactions and great improvement in voltage efficiency in single serpentine flow cell cycling at high current densities. The enhanced electrochemical activity of carbon papers towards vanadium redox reactions by applying MoO3-based electrocatalyst either in the form of MoO3 nanoparticles initially on electrode surfaces or in the form of additive in both anolyte and catholyte and its low cost as well as non-toxicity make this material a promising electrocatalyst material for VRFB. 5. Electrocatalysts for high power density VRFBs 169 ), f: Flow rate -2 10 - 60 67.1 87.2 58.5 4 O Continued on next page S 2 0.5 M V in 2 M H ------ ethanol 6 in 0.01 M 3 Cl O Ir 2 2 in air for 15min; H Bi ), CE, VE and EE: (%)) -2 cm · in air for 30 h, Electrolyte additives - -in air for 3 h - - - - - 400 solution soaking; 450 repeat 8 times and then annealair at 1h 450 in Saturated solution of HCl under vacuum for 3 h, then 450 , , 3+ 4+ In Au , , 4+ 2+ Te ), j: Current density (mA Pd -1 -graphene 19 ------, 3 , , etc [ 122 ] 2+ Pt 4+ 3+ Ir-Carbon Felt [ 114 ] Ir Cu ElectrocatalystsPt Modification MethodsMn Electrolyte A fBi-Graphite Felt(GF) j[ 115 ] CE VE EE Table 5.1: Summaries of metal- or metal oxide- based electrocatalysts for VRFBs (A: Electrode area (cm (mLmin 5. Electrocatalysts for high power density VRFBs 170 9 - 60 95.1 81.8 78.1 25 20 20 53.50 81.00 43.33 4 4 O O Continued on next page S S 2 2 1 M V in 1 M 1.5 M V in 3 M H ------H + /C 3 6 O Cl W Pt 2 for 30 min H 2 + N 12 h room tem- . Spray 2 4 in N vacuum oven and N H Br B 42 + activate carbon + Na H 19 Continued from previous page + 24 C for 2 h in 3 O for 10 h; spray Pt/MWCNTs 7 Cl in air) GF W Ir 6 for 4 h, filtration, dry, calcination ) , dried at 100 4 O H 2 Table 5.1 – N H MWCNTs/ NaOH + methanol thermal-reflux at 80 at 300 GO + perature, dry, 400 ( then 550 with Nafion ink onto thermal treated with Nafion ink onto CP (300 /C-carbon pa- 3 O Ir-Graphene [ 283 ] W Electrocatalysts Modification MethodsPt/MWCNTs- GF [ 282 ] Electrolyte A f j CE VEper(CP) [ 117 ] EE 5. Electrocatalysts for high power density VRFBs 171 73 84.7 ∼ ∼ 88 91.5 ∼ ∼ 80 96 ∼ ∼ - - 40 9 - 20 25 20 20 82.8 85.6 70.7 4 4 4 O O O Continued on next page S S S 2 2 2 2 M V in 2.5 M 1 M V in 1 M 1.2 M V in 3 M H H H + so- 6 Cl Mn 2 ) Pt 2 2 O H 3 + H 2 C O for 6h in vacuum 2 thermal reflux at H H for 12 h, wash and dry O to electrolyte 3 3 Continued from previous page H for 5 h under Ar C O 2 In ; spray Pt/C Nafion ink onto car- Table 5.1 – Hydrothermal in 1 M ( lution at 200 under vaccum room temperature and then 500 Carbon black + NaOH + 80 bon felt, dry at 65 oven -carbon 4 -GF [ 285 ] Adding O 3 3+ In Electrocatalysts Modification MethodsMn felt [ 119 ] ElectrolytePt/C-carbon felt [ 284 ] A f j CE VE EE 5. Electrocatalysts for high power density VRFBs 172 84 ∼ 87.5 ∼ 96 ∼ 25 20 20 80.6 90.010 72.5 20 100 /5 M 4 O S O 4 3 V O Continued on next page S Cl 2 Bi ------1 M V inH 1 M 2 M HCl and 0.1 M - - 20 20 97.3 87.9 85.5 3 O N H / 3 ) + NaOH /C Nafion 3 3 2 O O O N Se Ti under Ar for 3h 2 Na Continued from previous page overnight; spray in air for 30 min in air for 3 h, Bi( additive in the electrolyte 3+ Table 5.1 – Bi Sol-gel method with precusor of TnBT ethylene glycol solution, dry in oven at ink onto carbon felt GO in ethylene glycol solution + Ruthe- nium trichloride + thermal reflux at 160 450 solution electrodeposition at -0.2 V 300 450 (vs. Ag/AgCl/3.5M KCl) dry and then -carbon C / 2 O Ti Bi-GF [ 163 ] Electrocatalysts Modification Methodsfelt [ 286 ] Electrolyte A f j CE VE EE Bi-GF [ 287 ] RuSe/RGO [ 157 ] 5. Electrocatalysts for high power density VRFBs 173 83 ∼ 84 ∼ 97 ∼ 5 20 100 12 500 100 99.78 72.19 72.03 in 5 4 in the O 4 S O O S 2 V in the nega- Continued on next page H 2 2 M M HCl 1 M VM in 3 positive side and tive side H in for and 41 O + HCl + 12 4 W O )] solution at 10 ) O 4 W 2 2 H H ( N Na 2 ) 4 O 2 Continued from previous page C + NaCl solution at 180 for 48 h. Dry and 500 4 O )[NbO( 2 4 C 2 Table 5.1 – H Hydrothermal in Hydrothermal in ( 4 h; wash and dry in air Ar for 2 h. 170 (NH GF 5 O 2 Nb -GF [ 120 ] 3 O W Electrocatalysts Modification Methods Electrolyte AW-dopped- f[ 121 ] j CE VE EE 5. Electrocatalysts for high power density VRFBs 174 25 30 200 80.0 73.0 65.4 12 - 80 99.7 78.3 78.1 posi- 4 O S 4 2 O in the nega- H Continued on next page S 2 2 3 M V inH 2 M 0.5 M VM in 3 tive solution and H tive solution and 2 /C ink 2 O O Ti Ti for 4 h with and carbon black 2 O Ti Continued from previous page in air for 30 h and then electrode- Table 5.1 – Sol-gel process to produce then dry in vacuumovernight. oven Make at 300 by thermal reflux at 80 NaOH and ethanol. Spray onto the carbon felt and then dried65 at 400 position -GF [ 123 ] /C-GF [ 269 ] 2 2 O O Ti Pb Electrocatalysts Modification Methods Electrolyte A f j CE VE EE 5. Electrocatalysts for high power density VRFBs 175 70 78 ∼ ∼ 85 72.5 ∼ ∼ 96.6 ∼ 25 - 100 87.9 84.130 74 500 100 4 4 O O S S 2 2 2 M V in 2 M 1.2 M V in 3 M - 25 - 100 92 H H 4 3 ) ) in 3 3 O O N N dry for 12 2 for 5 h into electrolyte N 2 3 Continued from previous page N precipitation;70 precipitation. Dry at 70 in in air for 10 h, Ce( O O in air for 10 h; Zr( 2 2 H H · · 3 3 H H Table 5.1 – N for 2 h under h; 500 N air for 12 h. Then calcination at 600 420 400 -GF [ 124 ] -GF [ 288 ] -GF [ 164 ] Adding SbCl 2 2 O 3+ O Zr Ce Sb Electrocatalysts Modification Methods Electrolyte A f j CE VE EE 5. Electrocatalysts for high power density VRFBs 176

5.2 Experimental

5.2.1 Electrode preparation

Thermal treatments are conventionally carried out in air at temperatures no greater than 400 for 30 h in order to avoid any mechanical degradation of the carbon felt associated with carbon dioxide formation. Treatment at higher temperature can reduce treatment time, but it is important to protect the carbon from excessive oxidation in air. In this study, the carbon paper (Toray 120, 370 µm) was thus buried in a bed of carbon black and heated in a muﬄe oven in air at 650 for 4 h.

To decorate with MoO3, ammonium molybdate tetrahydrate (NH4)6Mo7O24·4H2O (Sigma-Aldrich, Sydney, Australia) was dissolved in deionized water (3.8 mg·mL-1). The previously thermally treated carbon paper was then soaked in this solution while keeping the weight ratio of (NH4)6Mo7O24·4H2O to carbon as 1:4, and dried at 80 . This was subsequently calcined at 650 for 2 h in a bed of carbon black to produce the MoO3-CP electrode.

2− Figure 5.1: Synthesis methods of MoO3-CP and samples with CP with MoO4 electrolyte additive 5. Electrocatalysts for high power density VRFBs 177

5.2.2 Electrolyte Preparation

The method of preparing vanadium electrolyte was described in previous Sec-

2− tion 3.2.1. To prepare the vanadium electrolyte with MoO4 additive, the required amount of Na2MoO4·2H2O (Sigma-Aldrich, Sydney, Australia) was ﬁrst dissolved in distilled water and then the vanadium electrolyte was added into the Na2MoO4 solution to achieve the set concentration. The sequence of chemical or solution addition is critical to avoid precipitation and the maximum concentration of Na2MoO4 in 1.6 M V3.5+ (i.e., 0.5 M V3+ + 0.5 M VO2+) in 4.2 M total sulphate/bisulfate solution was 0.01 M.

5.2.3 Structural Characterization

The surface morphology of the carbon paper samples was recorded by scanning electron microscopy (SEM) using a FEI Nova NanoSEM 450 (FEI company, Oregon, United States) with an accelerating voltage of 5 kV. The compositional difference across the sample surface was recorded by a Bruker SDD-EDS detector(Bruker, Singarpore). X-ray diffraction (XRD) analysis was carried out using a PANalytical X’pert Multipurpose X-ray Diffraction System (PANalytical, Sydney, Australia) operated at 40 kV, 40 mA with CuKα radiation in the scanning range of 5∼90°. X- Ray photoelectron spectroscopy (XPS) was conducted on ESCALAB250Xi (Thermo Scientific, UK) with mono-chromated Al Kα as X-Ray source (spot size of 500 µm) at power of 150 W and vacuum better than 2 × 10−9 mBar.

5.2.4 Electrochemical measurement

Cyclic voltammetry (CV): CV measurements were conducted using a single compartment three-electrode cell and a Bio-VSP potentiostat from Bio Logic (Bio Logic Science Instruments, Seyssinet-Pariset, France). Carbon paper strips of 5. Electrocatalysts for high power density VRFBs 178

4 mm×30 mm dimension were employed as working electrodes. Resin was used to mask the strip so that only 16 mm2 of the working electrode was exposed to the electrolyte. Hg/Hg2SO4 (in sat. K2SO4) and graphite rod were used as reference and counter electrodes respectively. The CV test were carried out in 1 M VO2+ in 2.6 M total sulphate solution with or without additive at a scan rate of 10 mV·s-1. Five scans were initially conducted starting from the open circuit voltage (OCV=

-0.15 V) to an upper voltage limit of 1.0 V vs. Hg/Hg2SO4 and then to the lower limit of -1.5 V vs. Hg/Hg2SO4. Serpentine flow cell setup: The inlet and outlet ports of an off-the-shelf Direct Methanol Fuel Cell Single Stack (FuelCellStore, Texas, United States) with serpentine flow field were modified to accommodate the acidic vanadium solutions. The components of the serpentine cell are shown in Figure 5.2. Two layers of carbon paper samples were employed on each side as anodes or cathodes (area = 10.24cm2). A 30 µm thick VB1 membrane (V-Fuel Pty Ltd, Sydney, Australia) was used as separator. Gaskets were used to adjust the compression of the electrodes to around 67.6% of the original thickness.

Figure 5.2: Components of modiﬁed serpentine ﬂow cell

Vanadium electrolyte (1.6 M V3+ in 4.2 M sulphate solution and 1.6 M VO2+ in 4.2 M sulphate solution) was circulated through negative and positive half-cell using two MD-10-230GS01 Iwaki Magnet Pumps (Iwaki Pumps, Castle Hill, Australia) at a fixed flow rate of around 20 mL·min-1. Nitrogen was purged for 30 min 5. Electrocatalysts for high power density VRFBs 179 to deaerate the negative electrolyte and two layers of Parafilm (Sigma-Aldrich, Sydney, Australia) were used to cover the top of negative reservoir. The 100% states of charge (SOC) electrolyte was preprared by charging the cell at 50mA·cm-2

+ to 1.8V, where the positive electrolyte turned to VO2 (yellow) and the negative electrolyte to V2+ (violet). The electrochemical impedance spectroscopy (EIS) tests were conducted on the serpentine cell at an SOC close to 100% (OCV=1.54 V) with amplitude of 10 mV within the frequency range of 10 kHz to 300 mHz. The cell cycling tests were carried out using a RePower Battery Test System at diﬀerent current densities. All experiments were carried out at room temperature and the electrolyte temperature was measured and maintained at 35±1 . Each electrochemical measurement was repeated at least once to conﬁrm reproducibility.

5.3 Results and Discussion

5.3.1 Synthesis of MoO3-CP

The pristine CP used in this study was the commercialized Toray-20 and as the SEM shows(Figure 5.3), carbon fibres are well interconnected with a diameter of around 6∼ 8 µm. The woven structure can facilitate the mass transport of the electrolyte and also ensure sufficient electrical conduction as well as mechanical strength. In Figure 5.3a, it can also be observed that some coating layers are wrapped around the carbon fibres, which are assumed to be the Teflon wet proofing of this product. A TGA analysis as shown in Figure 5.4 also confirmed the pristine CP contains around 20% polymer coating, making the material highly hydrophobic. Although this is desirable for fuel cell application, the the hydrophobic property is not favoured for VRFBs. Instead, a hydrophilic surface with better wet-ability for vanadium electrolyte is desired. As the polymer is shown to decompose completely at around 580 and the residual carbon starts losing significant weight above 5. Electrocatalysts for high power density VRFBs 180

Figure 5.3: SEM images of pristine CP

600 , a temperature of 650 was chosen to treat the CP for the removal of the hydrophobic coating and mild oxidation of the fibre surface. This was done under a thin layer of carbon black particles to protect the CP from decomposition in air. The morphologies of thermally treated 650-4h-CP are demonstrated in Fig- ure 5.5. It can be seen that the coating appears carbonized and the surface of carbon fibres more rougher, which was expected to be beneficial for the adsorption 5. Electrocatalysts for high power density VRFBs 181

1 0 0

8 0 2 0 % p o l y m e r c o a t i n g ) % (

t 8 0 % c a r b o n n 6 0 e c r e

h 4 0 g i e W 2 0 P r i s t i n e C P

0 0 2 0 0 4 0 0 6 0 0 8 0 0 1 0 0 0

T e m p e r a t u r e ( ° C )

Figure 5.4: TGA results of pristine CP of ions containing molybdenum. After a series of procedures including soaking the thermally treated CP in (NH4)6Mo7O24 solution, drying in air at 80 and calcin- ing in a bed of carbon black at 650 for another 2 h, the MoO3 was successfully synthesized. Figure 5.6 shows the typical microstructure of the MoO3-CP prepared in this study. The carbon paper still retains the network of carbon ﬁbres but is loaded uniformly with MoO3 micro-ﬂakes.

To further conﬁrm the microstructure of the MoO3 micro-ﬂakes, XRD analysis was employed and a comparison with that of the pristine CP and 650-4h-CP is shown in Figure 5.7. The XRD pattern of 650-4h-CP exhibits almost identical

XRD patterns as that of pristine CP. In the results of the MoO3 sample, apart from the peaks attributed to carbon, the rest is in good agreement with the orthorhombic phase of MoO3. 5. Electrocatalysts for high power density VRFBs 182

Figure 5.5: SEM images of 650-4h-CP

5.3.2 Electrochemical Measurement

CV measurements in 1 M VO2+in 2.62 M total sulphate solution were performed at the scan rate of 10mV·s-1 to evaluate the electrochemical behavior of diﬀerent samples. The ﬁfth scans starting from the open circuit voltage (OCV) to -1.5 V and then to 1.0 V are presented in Figure 5.8. Within the tested voltage range, the 5. Electrocatalysts for high power density VRFBs 183

Figure 5.6: SEM images of MoO3-CP pristine CP failed to show complete oxidation or reduction peaks. However, after a mild thermal treatment in air at 650 for 4 h, two sets of redox peaks centered at -0.99 V and 0.42 V can be clearly observed from the CV results of 650-4h-CP. The cathodic peak located at -1.15 V is associated with the production of V2+ ions and that at -0.83V corresponds to the reoxidation of V2+ to V3+. The peaks

+ 2+ at 0.49 V and 0.35 V belong to the oxidation to VO2 and reduction to VO 5. Electrocatalysts for high power density VRFBs 184

∆ Pristine-CP

∆

∆ 650°C-4h-CP )

. ∆ u . a ( ∆ MoO -CP y 3 s i t n e t

♦

n ♦ ♦ I ♦ ♦ ♦ ♦ ♦ ♦ ∆

PDF#05-0508

PDF#35-0609

2 0 4 0 6 0 8 0 2q (° )

Figure 5.7: XRD patterns of the pristine CP, 650-4h-CP, MoO3-CP and PDF card 05-0508 and 35-0609 respectively. The results proved that the mild oxidation of CP at 650 can signiﬁcantly improve the electrochemical activity of the CP for the vanadium redox couple reactions.

It should be pointed out that in the first cycle of MoO3-CP in the CV measurement, as shown in Figure 5.9, there were some additional minor peaks, i.e., three reduction peaks at -260, -629 and -832 mV and one oxidation peak at -387 mV. These peaks were believed to be the redox transformation of molybdenum (VI) to (V), (V) to (VI), (VI) to (III) and (III) to (IV) [289].However, after the first cycle, 5. Electrocatalysts for high power density VRFBs 185 no peaks for molybdenum can be found and the CV curves did not vary significantly. As the peak current is mainly dominated by the vanadium ion diffusion to the electrode surface and since the accurate electrode surface area cannot be readily determined, the peak current in this work cannot adequately reflect the electrochemical performance. However, the peak separation potential (4Ep) is still a good indicator of electrochemical reversibility. The values of 4Ep associated 2+ + with VO /VO2 redox reaction for 650 -4h-CP, MoO3-CP and 650 -4h-CP 2− with MoO4 -additive are 144, 130 and 111 mV respectively, suggesting the loading of MoO3 slightly improved the electrochemical activity of 650-4h-CP for the

2+ + 2− VO /VO2 redox reaction and adding MoO4 into the electrolyte can achieve similar enhancement. On the other hand, the 4Ep of the peaks representing V2+/V3+ oxidation and reduction obtained on electrodes MoO3-CP and 650-4h-CP with

2− MoO4 additive are 171mV and 203mV, exhibiting greatly reduced overpotential for the V2+/V3+ redox reactions compared to electrodes 650-4h-CP (306 mV) tested in electrolyte without additive. The EIS results of the three samples are shown in Figure 5.10. All the curves are composed of a semi-circle which represents the charge transfer resistance and a straight line which relates to the Warburg impedance. The comparison of diameters of the semi-circles also show a much reduced charge transfer resistance of the two molybdenum-modiﬁed carbon papers compared with the thermally treated CP.

Furthermore, the similar size of the semi-circles of the MoO3-CP and the 650-4h-

2− CP with MoO4 additive is also consistent with the similar CV behaviour discussed before.

5.3.3 Single Cell Test

To evaluate the electrochemical performance in the zero-gap VRFBs, single cells with serpentine ﬂow ﬁeld were assembled with the 650-4h-CP, MoO3-CP and 5. Electrocatalysts for high power density VRFBs 186

(mA) I

-10

Pristine CP

-20

650 C-4h-CP

MoO -CP

-30 3

650 C-4h-CP with MoO additive

-40

-1.5 -1.0 -0.5 0.0 0.5 1.0

E (V) vs. Hg/Hg SO

2 4

Figure 5.8: The ﬁfth CV curves of the pristine CP, 650-4h-CP, MoO3-CP and 2− 2+ 650 -4h-CP with MoO4 additive in 1 M VO in 2.62 M total sulphate solution

2− -2 650 -4h-CP with MoO4 additive samples. The electrode area was 10.24 cm ). A 30 µm thick VB1 perﬂuorosulfonic acid membrane was used as separator, 50 mL of 1.6 M V3+/VO2+ in 4.2 M total sulfate solution was employed as initial electrolyte in each half-cell reservoir. EIS measurements with amplitude of 10 mV in the frequency range of 10 kHz to 300 mHz were performed when the cell was charged to 1.8 V at a current density of 50 mA·cm-2 (close to 100% SOC, OCV around 1.54 V). The results are shown in Figure 5.11. It is apparent that the three Nyquist impedance curves have almost identical X-intercepts in the high frequency regions which indicate similar ohmic resistances of the three cells. The smaller semi-circles of the serpentine cells with 2− MoO3-CP and 650 -4h-CP with MoO4 additive showed lower charge transfer 5. Electrocatalysts for high power density VRFBs 187

1 0 3 0 ) A m

(

2 0 I 0

1 0 - 0 . 8 - 0 . 6 - 0 . 4 - 0 . 2 E ( V ) v s . H g / H g S O 2 4 0 ) A

m - 1 0 (

I - 2 0 1 s t c y c l e 2 n d c y c l e - 3 0 3 r d c y c l e 4 t h c y c l e - 4 0 5 t h c y c l e

- 5 0 - 1 . 5 - 1 . 0 - 0 . 5 0 . 0 0 . 5 1 . 0

E ( V ) v s . H g / H g S O 2 4

2+ Figure 5.9: Five CV cycles of MoO3-CP in 1 M VO in 2.62 M total sulphate solution

5 . 0 6 5 0 °C - 4 h - C P M o O - C P 3 6 5 0 °C - 4 h - C P w i t h M o O 2 - a d d i t i v e 4 ) W (

2 . 5 "

Z -

0 . 0 0 5 1 0 1 5

Z ' ( W )

Figure 5.10: The Nyquist plots of the 650-4h-CP, MoO3-CP and 650-4h-CP 2− 2+ with MoO4 additive in 1 M VO in 2.62 M total sulphate solution 5. Electrocatalysts for high power density VRFBs 188

0 . 0 4 6 5 0 °C - 4 h - C P M o O - C P 3 6 5 0 °C - 4 h - C P w i t h M o O 2 - a d d i t i v e 4

) W (

" 0 . 0 2

Z -

0 . 0 0 0 . 0 0 0 . 0 5 0 . 1 0 0 . 1 5

Z ' ( W )

Figure 5.11: The Nyquist plots of the single serpentine cell with 650-4h-CP, 2− MoO3-CP and 650 -4h-CP with MoO4 additive after charging to 1.8V

2− resistances compared with the cell using 650 -4h-CP tested without MoO4 2− additive. These results show that modification with MoO3 or adding MoO4 into the electrolyte are both promising methods to reduce the activation overpotential in vanadium redox flow batteries. Galvanostatic charge-discharge tests at 100, 125 and 150 mA·cm-2 were also conducted to evaluate the performance of the serpentine cells at high current densities with a flow rate of 20 mLmin-1. The operating voltage range was set at 0.6∼1.7 V to minimize oxygen or hydrogen evolution. Figures 5.12 ∼ 5.14 present the charge and discharge profiles of the serpentine cells with 650-4h-CP, MoO3- 2− CP and 650 -4h-CP with MoO4 under different current densities respectively. A reduction of charge and discharge capacity with increasing current density can be observed in all serpentine cells, which can be attributed to the increased voltage losses at the higher currents that results in a lower final SOC during charging (and higher final SOC during discharging) to the set voltage limits. In the meantime, a distinctly reduced overpotential can be seen in both the charge and discharge steps of the modified cells compared with the cell using 650-4h-CP, further proving 5. Electrocatalysts for high power density VRFBs 189

6 5 0 ° C - 4 h - C P 1 . 6

1 . 4 ) V

( 1 . 2

e g

a t l o

V 1 . 0

1 0 0 m A c m - 2 0 . 8 1 2 5 m A c m - 2 1 5 0 m A c m - 2 0 . 6 0 1 0 2 0 3 0 4 0 S p e c i f i c C a p a c i t y ( A h L - 1 )

Figure 5.12: The first charge curves with 650-4h-CP at constant current density of 100, 125, 150 mA·cm−2 the enhanced activity of the electrodes due to the modification, since the ohmic resistances were almost identical and flow rates were fixed. Figure 5.15 demonstrates the voltage efficiency (VE) of the serpentine cells with different electrodes at various current densities. No significant variations can be observed in VE between the cells with MoO3 coatings and 650-4h-CP

2− with MoO4 additive, but a significant enhancement of 4∼6% is observed over the entire current range tested, compared with the cell employing the 650-4h- CP. Specifically, the two modified serpentine cells showed VE values averaged 85.4%, 81.2%, and 78% while the cell with 650-4h-CP showed 79.7%, 76.2% and 73.8% VE at current densities of 100, 125 and 150 mA·cm−2 respectively. In addition, after continuous tests for twelve cycles at current densities ranging from 100 to 150 mA·cm−2 (i.e., four cycles at each current density), the subsequent four cycles at 100 mA·cm−2 exhibited similar VE compared to the initial four cycles at 100 mA·cm−2, suggesting relatively good catalytic stabilities of the two 5. Electrocatalysts for high power density VRFBs 190

6 5 0 ° C - 4 h - C P w i t h M o O 2 - a d d i t i v e 1 . 6 4

1 . 4 ) V

( 1 . 2

e g

a t l o

V 1 . 0

1 0 0 m A c m - 2 0 . 8 1 2 5 m A c m - 2 1 5 0 m A c m - 2 0 . 6 0 1 0 2 0 3 0 4 0 S p e c i f i c C a p a c i t y ( A h L - 1 )

Figure 5.13: The ﬁrst charge curves with MoO3-CP at constant current density of 100, 125, 150 mA·cm−2

M o O - C P 1 . 6 3

1 . 4 ) V

( 1 . 2

e g

a t l o

V 1 . 0

1 0 0 m A c m - 2 0 . 8 - 2 1 2 5 m A c m 1 5 0 m A c m - 2 0 . 6 0 1 0 2 0 3 0 4 0 S p e c i f i c C a p a c i t y ( A h L - 1 )

2− Figure 5.14: The ﬁrst charge curves with 650 -4h-CP with MoO4 additive at constant current density of 100, 125, 150 mA·cm−2 5. Electrocatalysts for high power density VRFBs 191

1 0 0

8 0 1 0 0 1 2 5 1 5 0 1 0 0 6 0 ) %

(

E 4 0 V

6 5 0 °C - 4 h - C P 2 0 M o O - C P 3 6 5 0 °C - 4 h - C P w i t h M o O 2 - a d d i t i v e 4 0 0 4 8 1 2 1 6 C y c l e N u m b e r

Figure 5.15: The voltage efficiencies of single serpentine cell with 650-4h-CP, 2− MoO3-CP and 650 -4h-CP with MoO4 additive at constant current density of 100, 125, 150, and 100 mA·cm−2 modification methods. Further long-term cycling would be desirable however, and and this could be the subject of further study. Compared with the VE values reported for metal-based modification (Table 5.1), the performance obtained in this work is quite competitive at current densities higher than 100 mA·cm−2. Figure 5.16 compares the discharging power density of the three VRFBs at 100%

SOC. It can be seen that the VRFBs modiﬁed by the MoO3-based methods deliver almost identical power densities over the tested current densities and voltage ranges.

2− The highest power densities obtained in this work with the MoO3-CP and MoO4 additive are 199.65 and 199.95 mW·cm−2 respectively, which is comparable with the performance reported in other serpentine vanadium redox ﬂow batteries [176]. 5. Electrocatalysts for high power density VRFBs 192

2 - m c 2 0 0 W m /

C O S

1 8 0 % 0 0 1

t a

y t i s

n 1 6 0 e

D ° 6 5 0 C - 4 h - C P r e M o O - C P

w 3 o 2 - P

6 5 0 °C - 4 h - C P w i t h M o O

c 4

i 1 4 0 n a d d i t i v e a v l a

G 1 0 0 1 1 0 1 2 0 1 3 0 1 4 0 1 5 0 C u r r e n t D e n s i t y / m A c m - 2

Figure 5.16: The peak power densities of single serpentine cell with 650-4h-CP, 2− MoO3-CP and 650 -4h-CP with MoO4 additive at constant current density of 100, 125, 150 mA·cm−2

5.3.4 Structural Characterization after cycling

The morphologies of the modified electrodes after the cycling tests were also characterized and the results are shown in Figure 5.17. The amount of particles decorated on the surface of MoO3-CP after the cycling test was greatly reduced and their shape turned to ultra-thin nanosheets (Figure 5.17a). It is likely that some large microflakes were washed away by the flowing electrolyte and some experienced electrochemical redox reactions and re-deposition onto the surface of the carbon paper (as shown in the first CV curve of MoO3-CP in Figure 5.9). It is interesting 2− to observe that 650 -4h-CP with MoO4 additive after cycling (Figure 5.17b) had similar nanosheets on the fibre surface as the MoO3-CP. The EDS images shown in Figure 5.18 were taken exactly where the SEM images were taken. It can be seen that the catalysts containing molybdenum were uniformly dispersed in the carbon paper samples. 5. Electrocatalysts for high power density VRFBs 193

2− Figure 5.17: SEM images of (a)MoO3-CP and (b)650 -4h-CP with MoO4 additive after the single cell test 5. Electrocatalysts for high power density VRFBs 194

2− Figure 5.18: EDS images of (a) MoO3-CP and (b) 650 -4h-CP with MoO4 additive after the single cell test

The X-Ray photoelectron spectroscopy (XPS) results show that after charge- discharge tests and washing with distilled water MoO3-CP and 650-4h-CP with

2− MoO4 additive had a Mo:C atomic ratio of around 0.2% and 0.8% respectively. 5. Electrocatalysts for high power density VRFBs 195

Additionally, the state of Mo in MoO3-CP after cycling was multivalent Mo(IV) 2− and Mo(VI) while in 650 -4h-CP with MoO4 additive this was pure Mo(VI) (Figure 5.19 a and b). More studies on the catalytic mechanism and optimization of catalysts loading are needed to understand this phenomenon.

Nevertheless, from Figure 5.17 and 5.18, it can be seen that in both MoO3-

2− decorated CP and MoO4 electrolyte additive cases, MoO3-based nanosheets were attached onto the surface of carbon fibres even after cycling test at different current densities. It is reasonable to assume the increased internal surface area due to the attached MoO3 is the main contributor for the improved electrochemical activity of the electrodes. To be more specific, the MoO3-based nanosheets increase the total surface concentration of the active sites available for the electrochemical reactions, thus enhancing the electrochemical activity of the carbon paper electrodes. It is thus proposed that the MoO3 based particles or nanosheets introduce a higher concentration of surface oxide groups than is possible with carbonaceous materials that disintegrte when the surface oxide concentration exceeds a critical level. Further studies are however needed to support the above hypothesis. 5. Electrocatalysts for high power density VRFBs 196

( a ) M o O - C P a f t e r c y c l i n g 2 6 0 3

M o ( I V ) 3 d 3 / 2 2 4 0 s c r e e n e d M o ( V I ) 3 d 4 d n 3 / 2 M o ( I V ) 3 d

) 5 / 2 s (

u n s c r e e n e d s t 2 2 0 M o ( I V ) 3 d n

5 / 2

u M o ( V I ) o s c r e e n e d 0 C 3 d 4 d 3 / 2 2 0 0

1 8 0

2 3 8 2 3 6 2 3 4 2 3 2 2 3 0 2 2 8 2 2 6

B i n d i n g E n e r g y ( e V )

4 5 0 ( b ) 6 5 0 ° C - 4 h - C P w i t h M o O 2 - a f t e r c y c l i n g 4 M o ( V I ) 3 d 4 0 0 5 / 2

3 5 0

) M o ( V I ) 3 d s 3 / 2 (

s t 3 0 0 n

u o C 2 5 0

2 0 0

1 5 0 2 4 2 2 4 0 2 3 8 2 3 6 2 3 4 2 3 2 2 3 0 2 2 8

B i n d i n g E n e r g y ( e V )

Figure 5.19: Mo spectrum XPS analysis of (a)MoO3-CP and (b)650-4h-CP with 2− MoO4 additive after the single cell test 5. Electrocatalysts for high power density VRFBs 197

5.4 Summary

In summary, two feasible approaches for modifying carbonaceous electrodes to reduce activation overpotential for vanadium redox reactions are described here.

The MoO3 decorated carbon paper features a three dimensional carbonaceous network with good dispersion of MoO3 mirco-ﬂakes. During cycling the microﬂakes experienced some redox reactions and turned to thin nanosheets which provide active sites and act as electrocatalysts for the vanadium oxidation and reduction

2− reactions. The simple addition of MoO4 into the electrolyte can achieve similar properties. A single serpentine vanadium redox ﬂow cell employing the MoO3- CP electrodes and electrolyte additive showed great improvement in performance with increasing VE at high current densities. The averaged VE values obtained were 85.4%, 81.2% and 78% at current densities of 100, 125 and 150 mA·cm−2, which show great potential for electro-activation for the VRFB compared with other reported results so far. The highest discharging power density of around 200 mW·cm−2 in this work within the voltage range of 0.6∼1.7 V was obtained

2− −2 with the MoO3-CP and 650C-4h-CP with MoO4 additive at 150 mA·cm . The modiﬁcation based on non-noble MoO3 can thus be considered as a promising approach for high power density performance in VRFBs. Chapter 6

Conclusions and Recommendations

6.1 Summary and Conclusions

Among the available large scale battery technologies, the VRFB possesses a range of intrinsic merits such as decoupled power rating and energy capacity, 100% depth of discharge, long cycle life, non flammable chemistry, simple battery design and low system maintenance cost, making it an excellent option for a wide range of grid-onnected and off-grid applications. Many kW- or MW-scale VRFB systems have been successfully installed and are currently operating in electric power systems around the world. But similar to other large scale battery technologies, its market penetration is still relatively small, especially in places where the electricity prices are quite cheap. For a VRFB with specified power/energy requirements, the membrane price and the operating current density play a vital role in the capital cost. Therefore, in order to magnify the economic appeal of this technology, identifying a low-cost, high performance membrane, understanding the vanadium redox reaction kinetics as well as improving the power density of VRFBs are

198 6. Conclusions and Recommendations 199 extremely important. The objective of the study was to provide possible solutions for the above mentioned aspects.

In Chapter3, three commercial ion exchange membranes Fumasep r FAP-450

(anion exchange membrane), Fumasepr F930-rfd (cation exchange membrane) and VB2 (cation exchange membrane, 100 µm) were evahe VRFB. The mem-

+ brane stability in VO2 electrolyte over 800 days was ﬁrst investigated. All three membranes swelled to some extent and absorbed some of the electrolyte after

+ immersion in the strong acidic and oxidative VO2 solution. With regard to the permeability rates of vanadium ions across the membrane, the existing method was modified to obtain more accurate and reliable data. The reliability of the modified method was further confirmed by the comparison between the experimental and the modelled charge-discharge results using the obtained permeability rates. The simulated results suggest that membranes with more balanced permeation rates

2+ + 3+ 2+ between the V /VO2 and V /VO couples and diﬀusion coﬃcients in the order

3+ + 2+ 2+ + 2+ 2+ 2+ of V

2+ + ulation results show that lower V and VO2 ion permeability rates are desired to lower the heat generation of the stack during standby periods after charging when pumps are turned oﬀ. In this study, the VB2 cation exchange membrane showed the best overall performance. This can be attributed to its low degree of swelling in the vanadium electrolyte and its greater thickness compared to some of the other membranes. In Chapte4, a systematic approach was utilized to study the reactions of the

2+ + VO /VO2 redox couple with a focus on the eﬀects of electrode surface roughness and functional groups derived from pre-treatment on the electrochemical behaviour

2+ + of glassy carbon in VO /VO2 solutions. The results show that electrochemical oxidation produced by scanning to high potentials can increase the surface roughness 6. Conclusions and Recommendations 200 of glassy carbon material and form oxygen functional groups, while mechanical polishing produces defects that provide active sites to facilitate electrochemical oxidation. The role of the oxygen functional groups and the surface area are highly dependent on the inherent surface status of the electrodes. The smoother the electrode is, the fewer defects it possesses and the more dominant will be the eﬀect of oxygen functional groups than that of surface area variation. Cathodic treatment to negative potentials can reduce the oxygen functional groups while retaining the surface area after the electrochemical oxidation, thus recovering or even enhancing the activity of electrodes. In addition, a rougher electrode surface appears to be more reactive after experiencing the same electrochemical treatment. In Chapter5, a zero-gap cell structure with thin carbon paper electrodes was employed to evaluate selected electrocatalysts for high power density VRFBs. Approaches including decorating the carbon paper electrodes with non-nobel metal

2− oxides MoO3 and adding its related ions MoO4 into the electrolyte were developed to reduce activation polarization. Both methods can achieve a similar enhancement in electrochemical activity of the carbon paper electrodes in the cell. The average voltage eﬃciencies reached 85.4%, 81.2% and 78% at current densities of 100, 125 and 150 mA·cm−2. The highest discharging power density of around 200 mW·cm−2 within the voltage range of 0.6∼1.7 V was obtained at 150 mA·cm−2 with both the MoO3-decorated carbon paper and thermally treated carbon paper (650 for

2− 4h) with MoO4 additive in the electrolyte. The based on the non-noble MoO3 can thus be considered as a promising approach for achieving high power density performance in VRFBs. This study thus demonstrated the great potential of introducing the electrocatalysts as an electrolyte additive as an option to improve the power density since the process is simple and achieves the similar improvement as the corresponding electrode modiﬁcation. The main contribution of this study is in the potential to enhance the economic 6. Conclusions and Recommendations 201 viability of the VRFB by appropriate membrane material selection and power density improvement. Speciﬁcally, the present study has:

(1) Identiﬁed and assessed low-cost, high performance commercial membranes for the VRFB;

(2) Gained greater insights into the factors that aﬀect the kinetics of the vanadium redox reactions;

(3) Demonstrated two novel and simple approaches to improve the power density of the VRFB.

6.2 Recommended Future Work

In this ﬁnal section, some suggestions on future research directions are proposed.

6.2.1 Membrane Evaluation

The membrane permeability studies in this thesis were only carried out at temperature of 30 and the thermal behaviour of VRFB systems was predicted by assuming all the vandium ions share the same activation energy in Arrhenius Equa- tion (see Equation (A.2) in AppendixA). It was also assumed that the activation energy for membrane permeation is the same for all membranes, which is not necessarily accurate. More detailed studies of membrane permeability rates for selected membranes over a wider range of temperatures are therefore recommended since they would enable the Arrhenius parameters to be determined for each membrane and each vanadium ion, allowing more reliable simulations to be conducted over diﬀerent operating conditions and for extended charge-discharge cycling. Secondly, evaluating membrane performance on counting cycles might not be conclusive. Extended charge-discharge cycling tests over a wider temperature range 6. Conclusions and Recommendations 202 are also recommended for more information of the membrane under unfavourable temperature conditions or heavy-duty cycling applications. Moreover, diﬀusion tests after periodic cycling could also be conducted to study any changes in membrane properties and behaviour during long-term operation. Last but not least, similar to the operating history, pretreatment may also give rise to swelling, fouling and even degradation of the membrane. Various pretreatment methods could also be performed to investigate any impact on membrane properties.

6.2.2 Reaction kinetics

Different roles of oxygen functional groups in the vanadium redox reactions can be seen on the surface of glassy carbon electrodes in Chapter4 and on carbon paper electrodes in Chapter5. As reviewed previously, controversy and disagreement remain over the effect of oxygen functional groups. Many factors including the types of electrode raw material, pretreatment methods, the electrochemical history, and post-treatment procedures can significantly alter the electrode surface status, increasing the difficulty to identify the ”real” oxygen functional groups that are active or detrimental to the vanadium redox couple reactions. More methods should be developed for in-situ characterization and certain mathematical models could developed to assist the understanding of the interaction between oxygen functional groups and vanadium ions at the electrode-electrolyte interface.

6.2.3 Membrane Electrode Assemblies

As shown in Chapter5, the electrocatalysts coated on carbon paper electrodes are prone to be flushed away by the circulating electrolyte and thereby the catalysts utilization efficiency is low. Adding the electrocatalyst ions into the electrolyte has been shown to be superior in terms of simplicity and catalyst utilization 6. Conclusions and Recommendations 203 efficiency. In addition to the above two approaches, membrane electrode assemblies (MEAs), which have been widely used in fuel cell applications, are believed to offer inherent potential for high power density VRFBs. As described in Section 2.2, the MEA is an assembled stack of catalyst coated membrane and electrodes, and the electrocatalysts are usually partially embedded into the membrane, thus increasing the catalyst utilization as well as the interface between the electrocatalysts and membrane. Further studies involving the use of electrocatalysts in a MEA structure are therefore desirable since this is expected to offer better mechanical stability and facilitation of the electrochemical reactions in the VRFB. Bibliography

[1] L. H. Thaller, “Electrically rechargeable redox ﬂow cells,” tech. rep., Lewis Research Center, Cleveland, Ohio, 1974.

[2] R. Ye, D. Henkensmeier, S. J. Yoon, Z. Huang, D. K. Kim, Z. Chang, S. Kim, and R. Chen, “Redox ﬂow batteries for energy storage: a technology review,” Journal of Electrochemical Energy Conversion and Storage, vol. 15, no. 1, p. 010801, 2018.

[3] P. Leung, A. Shah, L. Sanz, C. Flox, J. Morante, Q. Xu, M. Mohamed, C. P. de Le´on,and F. Walsh, “Recent developments in organic redox ﬂow batteries: A critical review,” Journal of Power Sources, vol. 360, pp. 243–283, 2017.

[4] J. Winsberg, T. Hagemann, T. Janoschka, M. D. Hager, and U. S. Schu- bert, “Redox-ﬂow batteries: From metals to organic redox-active materials,” Angewandte Chemie International Edition, vol. 56, no. 3, pp. 686–711, 2017.

[5] “Doe global energy storage database,” Jan. 2018. https://www.energystorageexchange.org/.

[6] E. S. Hanley, G. Amarandei, and B. A. Glowacki, “Potential of redox ﬂow batteries and hydrogen as integrated storage for decentralized energy systems,” Energy & Fuels, vol. 30, no. 2, pp. 1477–1486, 2016.

204 Bibliography 205

[7] M. Zhang, M. Moore, J. Watson, T. A. Zawodzinski, and R. M. Counce, “Cap- ital cost sensitivity analysis of an all-vanadium redox-ﬂow battery,” Journal of The Electrochemical Society, vol. 159, no. 8, pp. A1183–A1188, 2012.

[8] M. Skyllas-Kazacos, M. Rychcik, R. G. Robins, A. Fane, and M. Green, “New all-vanadium redox ﬂow cell,” J. Electrochem. Soc.;(United States), vol. 133, 1986.

[9] M. Skyllas-Kazacos and F. Grossmith, “Eﬃcient vanadium redox ﬂow cell,” Journal of the Electrochemical Society, vol. 134, no. 12, pp. 2950–2953, 1987.

[10] F. Grossmith, P. Llewellyn, A. Fane, and M. Skyllas-Kazacos, “Evaluation of membranes for all-vanadium redox cell,” in Stationary energy storage: load leveling and remote applications, 1988.

[11] M. Skyllas-Kazacos, D. Kasherman, D. Hong, and M. Kazacos, “Characteris- tics and performance of 1 kW UNSW vanadium redox battery,” Journal of Power Sources, vol. 35, no. 4, pp. 399–404, 1991.

[12] C. P. H. Ang, “Optimization of the performance of an all-vanadium redox ﬂow cell through membrane evaluation,” Master’s thesis, School of Chemical Engineering and Industrial Chemistry, UNSW. Australia, 1990.

[13] S. Chieng, Membrane processes and membrane modiﬁcation for redox ﬂow battery applications. PhD thesis, University of New South Wales, 1993.

[14] T. Mohammadi and M. S. Kazacos, “Evaluation of the chemical stability of some membranes in vanadium solution,” Journal of applied electrochemistry, vol. 27, no. 2, pp. 153–160, 1997. Bibliography 206

[15] T. Mohammadi and M. S. Kazacos, “Modiﬁcation of anion-exchange membranes for vanadium redox ﬂow battery applications,” Journal of Power Sources, vol. 63, no. 2, pp. 179–186, 1996.

[16] T. Mohammadi, S. Chieng, and M. S. Kazacos, “Water transport study across commercial ion exchange membranes in the vanadium redox ﬂow battery,” Journal of membrane science, vol. 133, no. 2, pp. 151–159, 1997.

[17] T. Sukkar and M. Skyllas-Kazacos, “Water transfer behaviour across cation exchange membranes in the vanadium redox battery,” Journal of Membrane Science, vol. 222, no. 1, pp. 235–247, 2003.

[18] T. Sukkar and M. Skyllas-Kazacos, “Modiﬁcation of membranes using polyelectrolytes to improve water transfer properties in the vanadium redox battery,” Journal of Membrane Science, vol. 222, no. 1, pp. 249–264, 2003.

[19] P. G. Grimes and R. J. Bellows, “Method of improving the ion selectivity of membranes,” Dec. 21 1982. US Patent 4,365,009.

[20] S. Chieng, M. Kazacos, and M. Skyllas-Kazacos, “Modiﬁcation of daramic, microporous separator, for redox ﬂow battery applications,” Journal of membrane science, vol. 75, no. 1-2, pp. 81–91, 1992.

[21] S. Chieng, M. Kazacos, and M. Skyllas-Kazacos, “Preparation and evaluation of composite membrane for vanadium redox battery applications,” Journal of power sources, vol. 39, no. 1, pp. 11–19, 1992.

[22] T. Mohammadi and M. Skyllas-Kazacos, “Characterisation of novel composite membrane for redox ﬂow battery applications,” Journal of Membrane Science, vol. 98, no. 1-2, pp. 77–87, 1995. Bibliography 207

[23] T. Mohammadi and M. Skyllas-Kazacos, “Preparation of sulfonated composite membrane for vanadium redox ﬂow battery applications,” Journal of Membrane Science, vol. 107, no. 1-2, pp. 35–45, 1995.

[24] T. Mohammadi and M. Skyllas-Kazacos, “Use of polyelectrolyte for incorporation of ion-exchange groups in composite membranes for vanadium redox ﬂow battery applications,” Journal of power sources, vol. 56, no. 1, pp. 91–96, 1995.

[25] P. Zhao, H. Zhang, H. Zhou, J. Chen, S. Gao, and B. Yi, “Characteristics and performance of 10kW class all-vanadium redox-ﬂow battery stack,” Journal of Power Sources, vol. 162, no. 2, pp. 1416–1420, 2006.

[26] S. Kim, E. Thomsen, G. Xia, Z. Nie, J. Bao, K. Recknagle, W. Wang, V. Viswanathan, Q. Luo, X. Wei, et al., “1 kW/1 kWh advanced vanadium redox ﬂow battery utilizing mixed acid electrolytes,” Journal of Power Sources, vol. 237, pp. 300–309, 2013.

[27] J. Zeng, C. Jiang, Y. Wang, J. Chen, S. Zhu, B. Zhao, and R. Wang, “Studies on polypyrrole modified nafion membrane for vanadium redox flow battery,” Electrochemistry Communications, vol. 10, no. 3, pp. 372–375, 2008.

[28] Q. Luo, H. Zhang, J. Chen, P. Qian, and Y. Zhai, “Modification of Nafion membrane using interfacial polymerization for vanadium redox flow battery applications,” Journal of Membrane Science, vol. 311, no. 1, pp. 98 – 103, 2008.

[29] B. Schwenzer, S. Kim, M. Vijayakumar, Z. Yang, and J. Liu, “Correlation of structural differences between Nafion/polyaniline and Nafion/polypyrrole composite membranes and observed transport properties,” Journal of membrane science, vol. 372, no. 1, pp. 11–19, 2011. Bibliography 208

[30] J. Xi, Z. Wu, X. Teng, Y. Zhao, L. Chen, and X. Qiu, “Self-assembled polyelectrolyte multilayer modified nafion membrane with suppressed vanadium ion crossover for vanadium redox flow batteries,” J. Mater. Chem., vol. 18, pp. 1232–1238, 2008.

[31] J. Xi, Z. Wu, X. Qiu, and L. Chen, “Naﬁon/SiO2 hybrid membrane for vanadium redox ﬂow battery,” Journal of Power Sources, vol. 166, no. 2, pp. 531 – 536, 2007.

[32] X. Teng, Y. Zhao, J. Xi, Z. Wu, X. Qiu, and L. Chen, “Nafion/organically modified silicate hybrids membrane for vanadium redox flow battery,” Journal of Power Sources, vol. 189, no. 2, pp. 1240 – 1246, 2009.

[33] D. Schulte, J. Drillkens, B. Schulte, and D. Sauer, “Naﬁon hybrid membranes for use in redox ﬂow batteries,” Journal of the Electrochemical Society, vol. 157, no. 9, pp. A989–A992, 2010.

[34] C.-H. Lin, M.-C. Yang, and H.-J. Wei, “Amino-silica modified Nafion membrane for vanadium redox flow battery,” Journal of Power Sources, vol. 282, pp. 562–571, 2015.

[35] E. M. Davis, J. Kim, V. P. Oleshko, K. A. Page, and C. L. Soles, “Uncovering

the structure of Naﬁon-SiO2 hybrid ionomer membranes for prospective large- scale energy storage devices,” Advanced Functional Materials, vol. 25, no. 26, pp. 4064–4075, 2015.

[36] X. Teng, Y. Zhao, J. Xi, Z. Wu, X. Qiu, and L. Chen, “Naﬁon/organic

silica modiﬁed TiO2 composite membrane for vanadium redox ﬂow battery via in situ sol–gel reactions,” Journal of Membrane Science, vol. 341, no. 1, pp. 149–154, 2009. Bibliography 209

[37] N. Wang, S. Peng, D. Lu, S. Liu, Y. Liu, and K. Huang, “Naﬁon/TiO2 hybrid membrane fabricated via hydrothermal method for vanadium redox battery,” Journal of Solid State Electrochemistry, vol. 16, no. 4, pp. 1577–1584, 2012.

[38] J. Li, Y. Zhang, and L. Wang, “Preparation and characterization of sulfonated

polyimide/TiO2 composite membrane for vanadium redox ﬂow battery,” Journal of Solid State Electrochemistry, vol. 18, no. 3, pp. 729–737, 2014.

[39] L. Yu, F. Lin, L. Xu, and J. Xi, “A recast Naﬁon/graphene oxide composite membrane for advanced vanadium redox ﬂow batteries,” RSC Advances, vol. 6, no. 5, pp. 3756–3763, 2016.

[40] B. G. Kim, T. H. Han, and C. G. Cho, “Sulfonated graphene oxide/Naﬁon composite membrane for vanadium redox ﬂow battery,” Journal of nanoscience and nanotechnology, vol. 14, no. 12, pp. 9073–9077, 2014.

[41] K. J. Lee and Y. H. Chu, “Preparation of the graphene oxide (GO)/Naﬁon composite membrane for the vanadium redox ﬂow battery (VRB) system,” Vacuum, vol. 107, pp. 269–276, 2014.

[42] M. Vijayakumar, B. Schwenzer, S. Kim, Z. Yang, S. Thevuthasan, J. Liu,

G. L. Graff, and J. Hu, “Investigation of local environments in Nafion-SiO2 composite membranes used in vanadium redox flow batteries,” Solid State Nuclear Magnetic Resonance, vol. 42, pp. 71–80, 2012.

[43] Z. Mai, H. Zhang, X. Li, S. Xiao, and H. Zhang, “Nafion/polyvinylidene fluoride blend membranes with improved ion selectivity for vanadium redox flow battery application,” Journal of Power Sources, vol. 196, no. 13, pp. 5737– 5741, 2011.

[44] X. Luo, Z. Lu, J. Xi, Z. Wu, W. Zhu, L. Chen, and X. Qiu, “Inﬂuences of permeation of vanadium ions through PVDF-g-PSSA membranes on Bibliography 210

performances of vanadium redox ﬂow batteries,” The Journal of Physical Chemistry B, vol. 109, no. 43, pp. 20310–20314, 2005. PMID: 16853627.

[45] J. Qiu, L. Zhao, M. Zhai, J. Ni, H. Zhou, J. Peng, J. Li, and G. Wei, “Pre- irradiation grafting of styrene and maleic anhydride onto pvdf membrane and subsequent sulfonation for application in vanadium redox batteries,” Journal of power sources, vol. 177, no. 2, pp. 617–623, 2008.

[46] D. Chen, S. Wang, M. Xiao, and Y. Meng, “Synthesis and characterization of novel sulfonated poly (arylene thioether) ionomers for vanadium redox ﬂow battery applications,” Energy & Environmental Science, vol. 3, no. 5, pp. 622–628, 2010.

[47] D. Chen, S. Wang, M. Xiao, and Y. Meng, “Synthesis and properties of novel sulfonated poly (arylene ether sulfone) ionomers for vanadium redox ﬂow battery,” Energy Conversion and Management, vol. 51, no. 12, pp. 2816–2824, 2010.

[48] M. A. Aziz, K. Oh, and S. Shanmugam, “A sulfonated poly (arylene ether ketone)/polyoxometalate–graphene oxide composite: a highly ion selective membrane for all vanadium redox ﬂow batteries,” Chemical Communications, vol. 53, no. 5, pp. 917–920, 2017.

[49] M. Yue, Y. Zhang, and L. Wang, “Sulfonated polyimide/chitosan composite membrane for vanadium redox ﬂow battery: membrane preparation, characterization, and single cell performance,” Journal of Applied Polymer Science, vol. 127, no. 5, pp. 4150–4159, 2013.

[50] Y. Zhang, J. Li, L. Wang, and S. Zhang, “Sulfonated polyimide/AlOOH composite membranes with decreased vanadium permeability and increased Bibliography 211

stability for vanadium redox ﬂow battery,” Journal of Solid State Electro- chemistry, vol. 18, no. 12, pp. 3479–3490, 2014.

[51] J. Li, Y. Zhang, S. Zhang, X. Huang, and L. Wang, “Novel sulfonated

polyimide/ZrO2 composite membrane as a separator of vanadium redox ﬂow battery,” Polymers for Advanced Technologies, vol. 25, no. 12, pp. 1610–1615, 2014.

[52] D. Chen, M. A. Hickner, E. Agar, and E. C. Kumbur, “Optimizing membrane thickness for vanadium redox ﬂow batteries,” Journal of membrane science, vol. 437, pp. 108–113, 2013.

[53] D. Chen, S. Kim, V. Sprenkle, and M. A. Hickner, “Composite blend polymer membranes with increased proton selectivity and lifetime for vanadium redox ﬂow batteries,” Journal of Power Sources, vol. 231, pp. 301–306, 2013.

[54] N. Wang, J. Yu, Z. Zhou, D. Fang, S. Liu, and Y. Liu, “SPPEK/TPA composite membrane as a separator of vanadium redox ﬂow battery,” Journal of membrane science, vol. 437, pp. 114–121, 2013.

[55] S. Winardi, S. C. Raghu, M. O. Oo, Q. Yan, N. Wai, T. M. Lim, and M. Skyllas-Kazacos, “Sulfonated poly (ether ether ketone)-based proton exchange membranes for vanadium redox battery applications,” Journal of Membrane Science, vol. 450, pp. 313–322, 2014.

[56] Z. Li, W. Dai, L. Yu, L. Liu, J. Xi, X. Qiu, and L. Chen, “Properties investigation of sulfonated poly (ether ether ketone)/polyacrylonitrile acid– base blend membrane for vanadium redox ﬂow battery application,” ACS applied materials & interfaces, vol. 6, no. 21, pp. 18885–18893, 2014.

[57] Z. Li, J. Xi, H. Zhou, L. Liu, Z. Wu, X. Qiu, and L. Chen, “Preparation and characterization of sulfonated poly (ether ether ketone)/poly (vinylidene Bibliography 212

ﬂuoride) blend membrane for vanadium redox ﬂow battery application,” Journal of Power Sources, vol. 237, pp. 132–140, 2013.

[58] Z. Li, L. Liu, L. Yu, L. Wang, J. Xi, X. Qiu, and L. Chen, “Characteri- zation of sulfonated poly (ether ether ketone)/poly (vinylidene fluoride-co- hexafluoropropylene) composite membrane for vanadium redox flow battery application,” Journal of Power Sources, vol. 272, pp. 427–435, 2014.

[59] S. Liu, L. Wang, Y. Ding, B. Liu, X. Han, and Y. Song, “Novel sulfonated poly (ether ether keton)/polyetherimide acid-base blend membranes for vanadium redox ﬂow battery applications,” Electrochimica Acta, vol. 130, pp. 90–96, 2014.

[60] W. Dai, L. Yu, Z. Li, J. Yan, L. Liu, J. Xi, and X. Qiu, “Sulfonated poly (ether ether ketone)/graphene composite membrane for vanadium redox ﬂow battery,” Electrochimica Acta, vol. 132, pp. 200–207, 2014.

[61] W. Dai, Y. Shen, Z. Li, L. Yu, J. Xi, and X. Qiu, “SPEEK/Graphene oxide nanocomposite membranes with superior cyclability for highly eﬃcient vanadium redox ﬂow battery,” Journal of Materials Chemistry A, vol. 2, no. 31, pp. 12423–12432, 2014.

[62] Z. Li, W. Dai, L. Yu, J. Xi, X. Qiu, and L. Chen, “Sulfonated poly (ether ether ketone)/mesoporous silica hybrid membrane for high performance vanadium redox ﬂow battery,” Journal of Power Sources, vol. 257, pp. 221–229, 2014.

[63] J. Pan, S. Wang, M. Xiao, M. Hickner, and Y. Meng, “Layered zirconium phosphate sulfophenylphosphonates reinforced sulfonated poly (ﬂuorenyl ether ketone) hybrid membranes with high proton conductivity and low vanadium ion permeability,” Journal of membrane science, vol. 443, pp. 19– 27, 2013. Bibliography 213

[64] F. Wang, J. M. Sylvia, M. M. Jacob, and D. Peramunage, “Amphiphilic block copolymer membrane for vanadium redox ﬂow battery,” Journal of Power Sources, vol. 242, pp. 575–580, 2013.

[65] A. Shibata and K. Sato, “Development of vanadium redox ﬂow battery for electricity storage,” Power Engineering Journal, vol. 13, no. 3, pp. 130–135, 1999.

[66] G.-J. Hwang and H. Ohya, “Crosslinking of anion exchange membrane by accelerated electron radiation as a separator for the all-vanadium redox ﬂow battery,” Journal of Membrane Science, vol. 132, no. 1, pp. 55–61, 1997.

[67] D. Xing, S. Zhang, C. Yin, C. Yan, and X. Jian, “Preparation and characterization of chloromethylated/quanternized poly (phthalazinone ether sulfone) anion exchange membrane,” Materials Science and Engineering: B, vol. 157, no. 1, pp. 1–5, 2009.

[68] D. Xing, S. Zhang, C. Yin, B. Zhang, and X. Jian, “Eﬀect of amination agent on the properties of quaternized poly (phthalazinone ether sulfone) anion exchange membrane for vanadium redox ﬂow battery application,” Journal of Membrane Science, vol. 354, no. 1, pp. 68–73, 2010.

[69] S. Zhang, C. Yin, D. Xing, D. Yang, and X. Jian, “Preparation of chloromethylated/quaternized poly (phthalazinone ether ketone) anion exchange membrane materials for vanadium redox ﬂow battery applications,” Journal of Membrane Science, vol. 363, no. 1, pp. 243–249, 2010.

[70] B. Zhang, S. Zhang, D. Xing, R. Han, C. Yin, and X. Jian, “Quaternized poly (phthalazinone ether ketone ketone) anion exchange membrane with low permeability of vanadium ions for vanadium redox ﬂow battery application,” Journal of Power Sources, vol. 217, pp. 296–302, 2012. Bibliography 214

[71] Z. Mai, H. Zhang, H. Zhang, W. Xu, W. Wei, H. Na, and X. Li, “Anion- conductive membranes with ultralow vanadium permeability and excellent performance in vanadium ﬂow batteries,” ChemSusChem, vol. 6, no. 2, pp. 328–335, 2013.

[72] F. Zhang, H. Zhang, and C. Qu, “A dication cross-linked composite anion- exchange membrane for all-vanadium ﬂow battery applications,” Chem- SusChem, vol. 6, no. 12, pp. 2290–2298, 2013.

[73] J. Qiu, M. Zhai, J. Chen, Y. Wang, J. Peng, L. Xu, J. Li, and G. Wei, “Performance of vanadium redox ﬂow battery with a novel amphoteric ion exchange membrane synthesized by two-step grafting method,” Journal of Membrane Science, vol. 342, no. 1, pp. 215–220, 2009.

[74] G. Hu, Y. Wang, J. Ma, J. Qiu, J. Peng, J. Li, and M. Zhai, “A novel amphoteric ion exchange membrane synthesized by radiation-induced grafting α-methylstyrene and N, N-dimethylaminoethyl methacrylate for vanadium redox ﬂow battery application,” Journal of membrane science, vol. 407, pp. 184–192, 2012.

[75] J. Ma, Y. Wang, J. Peng, J. Qiu, L. Xu, J. Li, and M. Zhai, “Designing a new process to prepare amphoteric ion exchange membrane with well-distributed grafted chains for vanadium redox ﬂow battery,” Journal of membrane science, vol. 419, pp. 1–8, 2012.

[76] Y. Wang, S. Wang, M. Xiao, S. Song, D. Han, M. A. Hickner, and Y. Meng, “Amphoteric ion exchange membrane synthesized by direct polymerization for vanadium redox ﬂow battery application,” international journal of hydrogen energy, vol. 39, no. 28, pp. 16123–16131, 2014. Bibliography 215

[77] C.-P. Li, S.-M. Zhang, S.-B. Wang, X.-F. Xie, and C.-S. Deng, “Study of chemical modiﬁed amphoteric ion exchange membrane from taurine,” ECS Electrochemistry Letters, vol. 3, no. 10, pp. A102–A104, 2014.

[78] M. Lee, H. Kang, J. Jeon, Y. Choi, and Y. Yoon, “A novel amphoteric ion-exchange membrane prepared by the pore-ﬁlling technique for vanadium redox ﬂow batteries,” RSC Advances, vol. 6, no. 67, pp. 63023–63029, 2016.

[79] O. Nibel, T. J. Schmidt, and L. Gubler, “Bifunctional ion-conducting polymer electrolyte for the vanadium redox ﬂow battery with high selectivity,” Journal of The Electrochemical Society, vol. 163, no. 13, pp. A2563–A2570, 2016.

[80] O. Nibel, T. Rojek, T. J. Schmidt, and L. Gubler, “Amphoteric ion-exchange membranes with signiﬁcantly improved vanadium barrier properties for all- vanadium redox ﬂow batteries.,” ChemSusChem, 2017.

[81] J. Liao, M. Lu, Y. Chu, and J. Wang, “Ultra-low vanadium ion diﬀusion amphoteric ion-exchange membranes for all-vanadium redox ﬂow batteries,” Journal of Power Sources, vol. 282, pp. 241–247, 2015.

[82] B. Tian, C. Yan, and F. Wang, “Proton conducting composite membrane from Daramic/Naﬁon for vanadium redox ﬂow battery,” Journal of membrane science, vol. 234, no. 1, pp. 51–54, 2004.

[83] H. Zhang, H. Zhang, X. Li, Z. Mai, and J. Zhang, “Nanoﬁltration (NF) membranes: the next generation separators for all vanadium redox ﬂow batteries (VRBs)?,” Energy & Environmental Science, vol. 4, no. 5, pp. 1676– 1679, 2011.

[84] H. Zhang, H. Zhang, X. Li, Z. Mai, and W. Wei, “Silica modified nanofiltration membranes with improved selectivity for redox flow battery application,” Energy & Environmental Science, vol. 5, no. 4, pp. 6299–6303, 2012. Bibliography 216

[85] Y. Li, H. Zhang, X. Li, H. Zhang, and W. Wei, “Porous poly (ether sulfone) membranes with tunable morphology: Fabrication and their application for vanadium ﬂow battery,” Journal of Power Sources, vol. 233, pp. 202–208, 2013.

[86] W. Wei, H. Zhang, X. Li, H. Zhang, Y. Li, and I. Vankelecom, “Hydrophobic asymmetric ultraﬁltration pvdf membranes: an alternative separator for VFB with excellent stability,” Physical Chemistry Chemical Physics, vol. 15, no. 6, pp. 1766–1771, 2013.

[87] G.-J. Hwang and H. Ohya, “Preparation of cation exchange membrane as a separator for the all-vanadium redox ﬂow battery,” Journal of membrane science, vol. 120, no. 1, pp. 55–67, 1996.

[88] X. Wei, Z. Nie, Q. Luo, B. Li, B. Chen, K. Simmons, V. Sprenkle, and W. Wang, “Nanoporous polytetraﬂuoroethylene/silica composite separator as a high-performance all-vanadium redox ﬂow battery membrane,” Advanced Energy Materials, vol. 3, no. 9, pp. 1215–1220, 2013.

[89] M. Rychcik and M. Skyllas-Kazacos, “Evaluation of electrode materials for vanadium redox cell,” Journal of Power Sources, vol. 19, no. 1, pp. 45–54, 1987.

[90] S. Zhong, C. Padeste, M. Kazacos, and M. Skyllas-Kazacos, “Compari- son of the physical, chemical and electrochemical properties of rayon-and polyacrylonitrile-based graphite felt electrodes,” Journal of Power sources, vol. 45, no. 1, pp. 29–41, 1993.

[91] A. Di Blasi, O. Di Blasi, N. Briguglio, A. Aricò,D. Sebastián,M. Lázaro, G. Monforte, and V. Antonucci, “Investigation of several graphite-based Bibliography 217

electrodes for vanadium redox ﬂow cell,” Journal of Power Sources, vol. 227, pp. 15–23, 2013.

[92] J. Melke, P. Jakes, J. Langner, L. Riekehr, U. Kunz, Z. Zhao-Karger, A. Nefe- dov, H. Sezen, C. W¨oll,H. Ehrenberg, et al., “Carbon materials for the positive electrode in all-vanadium redox ﬂow batteries,” Carbon, vol. 78, pp. 220–230, 2014.

[93] R. Schweiss, “Influence of bulk fibre properties of pan-based carbon felts on their performance in vanadium redox flow batteries,” Journal of Power Sources, vol. 278, pp. 308–313, 2015.

[94] N. Pour, D. G. Kwabi, T. Carney, R. M. Darling, M. L. Perry, and Y. Shao- Horn, “Inﬂuence of edge-and basal-plane sites on the vanadium redox kinetics for ﬂow batteries,” The Journal of Physical Chemistry C, vol. 119, no. 10, pp. 5311–5318, 2015.

[95] B. Sun and M. Skyllas-Kazacos, “Modiﬁcation of graphite electrode materials for vanadium redox ﬂow battery application I. thermal treatment,” Electrochimica Acta, vol. 37, no. 7, pp. 1253–1260, 1992.

[96] N. Kausar, Studies of V (IV) and V (V) species in vanadium cell electrolyte. PhD thesis, University of New South Wales, 2002.

[97] N. Kausar, R. Howe, and M. Skyllas-Kazacos, “Raman spectroscopy studies of concentrated vanadium redox battery positive electrolytes,” Journal of Applied Electrochemistry, vol. 31, no. 12, pp. 1327–1332, 2001.

[98] A. Ivakin and E. Voronova, “Spectrophotometric study of vanadium (IV) sulfate complexes,” Zhurnal Neorganicheskoi Khimii, vol. 18, no. 7, pp. 1809– 1812, 1973. Bibliography 218

[99] M. Vijayakumar, L. Li, G. Graﬀ, J. Liu, H. Zhang, Z. Yang, and J. Z. Hu, “Towards understanding the poor thermal stability of V5+ electrolyte solution in vanadium redox ﬂow batteries,” Journal of Power Sources, vol. 196, no. 7, pp. 3669–3672, 2011.

[100] S. C. Furman and C. S. Garner, “Absorption spectra of vanadium (III) and vanadium (IV) ions in complexing and non-complexing media,” Journal of the American Chemical Society, vol. 72, no. 4, pp. 1785–1789, 1950.

[101] v. H. Strehlow and H. Wendt, “Fast ionic reactions in solution. IV. The formation of the vanadyl sulfate complex in aqueous solution,” Inorganic Chemistry, vol. 2, no. 1, pp. 6–10, 1963.

[102] P. Blanc, C. Madic, and J. Launay, “Spectrophotometric identiﬁcation of a mixed-valence cation-cation complex between aquadioxovanadium (V) and aquaoxovanadium (IV) ions in perchloric, sulfuric, and hydrochloric acid media,” Inorganic Chemistry, vol. 21, no. 8, pp. 2923–2928, 1982.

[103] B. Sun and M. Skyllas-Kazacos, “Chemical modiﬁcation of graphite electrode materials for vanadium redox ﬂow battery applicationpart II. acid treatments,” Electrochimica Acta, vol. 37, no. 13, pp. 2459–2465, 1992.

[104] F. Mohammadi, P. Timbrell, S. Zhong, C. Padeste, and M. Skyllas-Kazacos, “Overcharge in the vanadium redox battery and changes in electrical resistivity and surface functionality of graphite-felt electrodes,” Journal of Power sources, vol. 52, no. 1, pp. 61–68, 1994.

[105] W. Zhang, J. Xi, Z. Li, H. Zhou, L. Liu, Z. Wu, and X. Qiu, “Electrochemical

2+ + activation of graphite felt electrode for VO /VO2 redox couple application,” Electrochimica Acta, vol. 89, pp. 429–435, 2013. Bibliography 219

[106] Y. Men and T. Sun, “Carbon felts electrode treated in diﬀerent weak acid solutions through electrochemical oxidation method for all vanadium redox ﬂow battery,” Int. J. Electrochem. Sci, vol. 7, pp. 3482–3488, 2012.

[107] H. Lawrence, Jones K. Xiao, “Improved electrode for ﬂow batteries,” Jan. 2014. International patent WO/2014/149192.

[108] K. J. Kim, S.-W. Lee, T. Yim, J.-G. Kim, J. W. Choi, J. H. Kim, M.-S. Park, and Y.-J. Kim, “A new strategy for integrating abundant oxygen functional groups into carbon felt electrode for vanadium redox ﬂow batteries,” Scientiﬁc reports, vol. 4, p. 6906, 2014.

[109] K. J. Kim, Y.-J. Kim, J.-H. Kim, and M.-S. Park, “The effects of surface modification on carbon felt electrodes for use in vanadium redox flow batteries,” Materials chemistry and physics, vol. 131, no. 1, pp. 547–553, 2011.

[110] E.-M. Hammer, B. Berger, and L. Komsiyska, “Improvement of the performance of graphite felt electrodes for vanadium-redox-ﬂow-batteries by plasma treatment,” International Journal of Renewable Energy Development, vol. 3, no. 1, p. 7, 2014.

[111] W. W. Li, Y. Q. Chu, and C. A. Ma, “Highly hydroxylated graphite felts used as electrodes for a vanadium redox ﬂow battery,” in Advanced Materials Research, vol. 936, pp. 471–475, Trans Tech Publ, 2014.

[112] M. Skyllas-Kazacos, “All-vanadium redox battery and additives,” 1989. WO Patent App. PCT/AU1988/000,472.

[113] B. Sun and M. Skyllas-Kazakos, “Chemical modiﬁcation and electrochemical behaviour of graphite ﬁbre in acidic vanadium solution,” Electrochimica Acta, vol. 36, no. 3-4, pp. 513–517, 1991. Bibliography 220

[114] W. Wang and X. Wang, “Investigation of ir-modiﬁed carbon felt as the positive electrode of an all-vanadium redox ﬂow battery,” Electrochimica Acta, vol. 52, no. 24, pp. 6755–6762, 2007.

[115] Z. González,A. Sánchez, C. Blanco, M. Granda, R. Menéndez,and R. Santa- mar´ıa,“Enhanced performance of a Bi-modified graphite felt as the positive electrode of a vanadium redox flow battery,” Electrochemistry Communica- tions, vol. 13, no. 12, pp. 1379–1382, 2011.

[116] A. Wieckowski, Interfacial electrochemistry: theory: experiment, and applications. CRC Press, 1999.

[117] C. Yao, H. Zhang, T. Liu, X. Li, and Z. Liu, “Carbon paper coated with supported tungsten trioxide as novel electrode for all-vanadium ﬂow battery,” Journal of Power Sources, vol. 218, pp. 455–461, 2012.

[118] H. S. Kim, B. W. Cho, Y.-J. Kim, and K. J. Kim, “Graphite/DSA assembled electrode for redox ﬂow battery, method of manufacturing the same and redox ﬂow battery including the same,” Aug. 27 2013. US Patent 8,518,572.

[119] K. J. Kim, M.-S. Park, J.-H. Kim, U. Hwang, N. J. Lee, G. Jeong, and Y.-J.

Kim, “Novel catalytic eﬀects of Mn3O4 for all vanadium redox ﬂow batteries,” Chemical Communications, vol. 48, no. 44, pp. 5455–5457, 2012.

[120] Y. Shen, H. Xu, P. Xu, X. Wu, Y. Dong, and L. Lu, “Electrochemical catalytic

2+ + activity of tungsten trioxide-modiﬁed graphite felt toward VO /VO2 redox reaction,” Electrochimica Acta, vol. 132, pp. 37–41, 2014.

[121] B. Li, M. Gu, Z. Nie, X. Wei, C. Wang, V. Sprenkle, and W. Wang, “Nanorod niobium oxide as powerful catalysts for an all vanadium redox ﬂow battery,” Nano letters, vol. 14, no. 1, pp. 158–165, 2013. Bibliography 221

[122] M. Skyllas-Kazacos, “All-vanadium redox battery and additives,” Dec. 1987. International patent WO 89/05526.

[123] X. Wu, H. Xu, L. Lu, H. Zhao, J. Fu, Y. Shen, P. Xu, and Y. Dong, “PbO2- modiﬁed graphite felt as the positive electrode for an all-vanadium redox ﬂow battery,” Journal of Power Sources, vol. 250, pp. 274–278, 2014.

[124] H. Zhou, J. Xi, Z. Li, Z. Zhang, L. Yu, L. Liu, X. Qiu, and L. Chen, “CeO2 decorated graphite felt as a high-performance electrode for vanadium redox ﬂow batteries,” RSC Advances, vol. 4, no. 106, pp. 61912–61918, 2014.

[125] G. Radford, J. Cox, R. Wills, and F. Walsh, “Electrochemical characterisation of activated carbon particles used in redox ﬂow battery electrodes,” Journal of Power Sources, vol. 185, no. 2, pp. 1499–1504, 2008.

[126] Y. Shao, X. Wang, M. Engelhard, C. Wang, S. Dai, J. Liu, Z. Yang, and Y. Lin, “Nitrogen-doped mesoporous carbon for energy storage in vanadium redox ﬂow batteries,” Journal of Power Sources, vol. 195, no. 13, pp. 4375– 4379, 2010.

[127] M. Ulaganathan, A. Jain, V. Aravindan, S. Jayaraman, W. C. Ling, T. M. Lim, M. P. Srinivasan, Q. Yan, and S. Madhavi, “Bio-mass derived mesoporous carbon as superior electrode in all vanadium redox ﬂow battery with multicouple reactions,” Journal of Power Sources, vol. 274, pp. 846–850, 2015.

[128] H. Zhu, Y. Zhang, L. Yue, W. Li, G. Li, D. Shu, and H. Chen, “Graphite– carbon nanotube composite electrodes for all vanadium redox ﬂow battery,” Journal of Power Sources, vol. 184, no. 2, pp. 637–640, 2008.

[129] X. Rui, A. Parasuraman, W. Liu, D. H. Sim, H. H. Hng, Q. Yan, T. M. Lim, and M. Skyllas-Kazacos, “Functionalized single-walled carbon nanotubes with Bibliography 222

enhanced electrocatalytic activity for redox reactions in vanadium bromide redox ﬂow batteries,” Carbon, vol. 64, pp. 464–471, 2013.

[130] W. Li, J. Liu, and C. Yan, “The electrochemical catalytic activity of single-

+ 3+ 2+ walled carbon nanotubes towards VO2+/VO2 and V /V redox pairs for an all vanadium redox ﬂow battery,” Electrochimica Acta, vol. 79, pp. 102–108, 2012.

[131] G. Wei, C. Jia, J. Liu, and C. Yan, “Carbon felt supported carbon nanotubes catalysts composite electrode for vanadium redox ﬂow battery application,” Journal of Power Sources, vol. 220, pp. 185–192, 2012.

[132] W. Li, J. Liu, and C. Yan, “Multi-walled carbon nanotubes used as an

2+ + electrode reaction catalyst for VO /VO2 for a vanadium redox ﬂow battery,” Carbon, vol. 49, no. 11, pp. 3463–3470, 2011.

[133] C. Flox, J. Rubio-Garc´ıa, M. Skoumal, T. Andreu, and J. R. Morante,

“Thermo–chemical treatments based on NH3/O2 for improved graphite-based ﬁber electrodes in vanadium redox ﬂow batteries,” Carbon, vol. 60, pp. 280– 288, 2013.

[134] M. Manahan, Q. Liu, M. Gross, and M. Mench, “Carbon nanoporous layer for reaction location management and performance enhancement in all-vanadium redox ﬂow batteries,” Journal of Power Sources, vol. 222, pp. 498–502, 2013.

[135] H.-M. Tsai, S.-Y. Yang, C.-C. M. Ma, and X. Xie, “Preparation and electrochemical properties of graphene-modiﬁed electrodes for all-vanadium redox ﬂow batteries,” Electroanalysis, vol. 23, no. 9, pp. 2139–2143, 2011.

[136] P. Han, H. Wang, Z. Liu, X. Chen, W. Ma, J. Yao, Y. Zhu, and G. Cui, “Graphene oxide nanoplatelets as excellent electrochemical active materials for Bibliography 223

2+ + 2+ 3+ VO /VO2 and V /V redox couples for a vanadium redox ﬂow battery,” Carbon, vol. 49, no. 2, pp. 693–700, 2011.

[137] Z. González, C. Botas, P. Alvarez,´ S. Roldán,C. Blanco, R. Santamar´ıa, M. Granda, and R. Menéndez,“Thermally reduced graphite oxide as positive electrode in vanadium redox flow batteries,” Carbon, vol. 50, no. 3, pp. 828– 834, 2012.

[138] Z. González,C. Botas, C. Blanco, R. Santamar´ıa,M. Granda, P. Alvarez,´ and R. Menéndez,“Graphite oxide-based graphene materials as positive electrodes in vanadium redox flow batteries,” Journal of Power Sources, vol. 241, pp. 349–354, 2013.

[139] Z. Gonz´alez,C. Botas, C. Blanco, R. Santamar´ıa,M. Granda, P. Alvarez,´ and R. Men´endez,“Thermally reduced graphite and graphene oxides in VRFBs,” Nano Energy, vol. 2, no. 6, pp. 1322–1328, 2013.

[140] W. Li, J. Liu, and C. Yan, “Reduced graphene oxide with tunable C/O ratio and its activity towards vanadium redox pairs for an all vanadium redox ﬂow battery,” Carbon, vol. 55, pp. 313–320, 2013.

[141] Z. González,S. Vizireanu, G. Dinescu, C. Blanco, and R. Santamar´ıa,“Carbon nanowalls thin films as nanostructured electrode materials in vanadium redox flow batteries,” Nano Energy, vol. 1, no. 6, pp. 833–839, 2012.

[142] M. Park, I.-Y. Jeon, J. Ryu, J.-B. Baek, and J. Cho, “Exploration of the eﬀective location of surface oxygen defects in graphene-based electrocatalysts for all-vanadium redox-ﬂow batteries,” Advanced Energy Materials, vol. 5, no. 5, 2015.

[143] R. I. Jafri, N. Rajalakshmi, and S. Ramaprabhu, “Nitrogen doped graphene nanoplatelets as catalyst support for oxygen reduction reaction in proton Bibliography 224

exchange membrane fuel cell,” Journal of Materials Chemistry, vol. 20, no. 34, pp. 7114–7117, 2010.

[144] S. Wang, X. Zhao, T. Cochell, and A. Manthiram, “Nitrogen-doped carbon nanotube/graphite felts as advanced electrode materials for vanadium redox ﬂow batteries,” The journal of physical chemistry letters, vol. 3, no. 16, pp. 2164–2167, 2012.

[145] Z. He, L. Shi, J. Shen, Z. He, and S. Liu, “Eﬀects of nitrogen doping on the electrochemical performance of graphite felts for vanadium redox ﬂow batteries,” International Journal of Energy Research, vol. 39, no. 5, pp. 709– 716, 2015.

[146] L. Shi, S. Liu, Z. He, and J. Shen, “Nitrogen-doped graphene: Eﬀects of nitrogen species on the properties of the vanadium redox ﬂow battery,” Electrochimica Acta, vol. 138, pp. 93–100, 2014.

[147] Z. He, A. Su, C. Gao, Z. Zhou, C. Pan, and S. Liu, “Carbon paper modiﬁed by hydrothermal ammoniated treatment for vanadium redox battery,” Ionics, vol. 19, no. 7, pp. 1021–1026, 2013.

[148] T. Wu, K. Huang, S. Liu, S. Zhuang, D. Fang, S. Li, D. Lu, and A. Su, “Hydrothermal ammoniated treatment of PAN-graphite felt for vanadium redox ﬂow battery,” Journal of solid state electrochemistry, vol. 16, no. 2, pp. 579–585, 2012.

[149] J. Jin, X. Fu, Q. Liu, Y. Liu, Z. Wei, K. Niu, and J. Zhang, “Identifying the

2+ + active site in nitrogen-doped graphene for the VO /VO2 redox reaction,” ACS nano, vol. 7, no. 6, pp. 4764–4773, 2013.

[150] M. Park, J. Ryu, Y. Kim, and J. Cho, “Corn protein-derived nitrogen-doped carbon materials with oxygen-rich functional groups: a highly eﬃcient elec- Bibliography 225

trocatalyst for all-vanadium redox ﬂow batteries,” Energy & Environmental Science, vol. 7, no. 11, pp. 3727–3735, 2014.

[151] H. Lee, H.J. Kim, “Nitrogen-doped carbons as electrode materials for all- vanadium redox ﬂow batteries,” in The International Flow Battery Forum, 2015.

[152] W. Li, J. Liu, and C. Yan, “Graphite–graphite oxide composite electrode for vanadium redox ﬂow battery,” Electrochimica Acta, vol. 56, no. 14, pp. 5290– 5294, 2011.

[153] P. Han, Y. Yue, Z. Liu, W. Xu, L. Zhang, H. Xu, S. Dong, and G. Cui, “Graphene oxide nanosheets/multi-walled carbon nanotubes hybrid as an

2+ + excellent electrocatalytic material towards VO /VO2 redox couples for vanadium redox ﬂow batteries,” Energy & Environmental Science, vol. 4, no. 11, pp. 4710–4717, 2011.

[154] M. Park, Y.-j. Jung, J. Kim, H. i. Lee, and J. Cho, “Synergistic effect of carbon nanofiber/nanotube composite catalyst on carbon felt electrode for high-performance all-vanadium redox flow battery,” Nano letters, vol. 13, no. 10, pp. 4833–4839, 2013.

[155] G. Wei, X. Fan, J. Liu, and C. Yan, “Electrospun carbon nanoﬁbers/electrocatalyst hybrids as asymmetric electrodes for vanadium redox ﬂow battery,” Journal of Power Sources, vol. 281, pp. 1–6, 2015.

[156] C. Flox, J. Rubio-Garcia, R. Nafria, R. Zamani, M. Skoumal, T. Andreu,

J. Arbiol, A. Cabot, and J. R. Morante, “Active nano-CuPt3 electrocatalyst supported on graphene for enhancing reactions at the cathode in all-vanadium redox ﬂow batteries,” Carbon, vol. 50, no. 6, pp. 2372–2374, 2012. Bibliography 226

[157] P. Han, X. Wang, L. Zhang, T. Wang, J. Yao, C. Huang, L. Gu, and G. Cui,

2+ + “RuSe/reduced graphene oxide: an eﬃcient electrocatalyst for VO /VO2 redox couples in vanadium redox ﬂow batteries,” RSC Advances, vol. 4, no. 39, pp. 20379–20381, 2014.

[158] N. A. Vante, W. Jaegermann, H. Tributsch, W. H¨onle,and K. Yvon, “Elec- trocatalysis of oxygen reduction by chalcogenides containing mixed transition metal clusters,” Journal of the American Chemical Society, vol. 109, no. 11, pp. 3251–3257, 1987.

[159] P. J. Kulesza, K. Miecznikowski, B. Baranowska, M. Skunik, S. Fiechter, P. Bogdanoﬀ, and I. Dorbandt, “Tungsten oxide as active matrix for dispersed

carbon-supported RuSex nanoparticles: enhancement of the electrocatalytic oxygen reduction,” Electrochemistry communications, vol. 8, no. 5, pp. 904– 908, 2006.

[160] T.-M. Tseng, R.-H. Huang, C.-Y. Huang, K.-L. Hsueh, and F.-S. Shieu, “Improvement of titanium dioxide addition on carbon black composite for negative electrode in vanadium redox ﬂow battery,” Journal of The Electro- chemical Society, vol. 160, no. 8, pp. A1269–A1275, 2013.

[161] T.-M. Tseng, R.-H. Huang, C.-Y. Huang, C.-C. Liu, K.-L. Hsueh, and F.-S. Shieu, “Carbon felt coated with titanium dioxide/carbon black composite as negative electrode for vanadium redox ﬂow battery,” Journal of The Electrochemical Society, vol. 161, no. 6, pp. A1132–A1138, 2014.

[162] M. S. Kazacos and M. Kazacos, “Stabilized vanadium electrolyte solutions for all-vanadium redox cells and batteries,” May 13 2003. US Patent 6,562,514.

[163] B. Li, M. Gu, Z. Nie, Y. Shao, Q. Luo, X. Wei, X. Li, J. Xiao, C. Wang, V. Sprenkle, et al., “Bismuth nanoparticle decorating graphite felt as a high- Bibliography 227

performance electrode for an all-vanadium redox ﬂow battery,” Nano letters, vol. 13, no. 3, pp. 1330–1335, 2013.

[164] J. Shen, S. Liu, Z. He, and L. Shi, “Inﬂuence of antimony ions in negative electrolyte on the electrochemical performance of vanadium redox ﬂow batteries,” Electrochimica Acta, vol. 151, pp. 297–305, 2015.

[165] J. H. Park, J. J. Park, H. J. Lee, B. S. Min, and J. H. Yang, “Inﬂuence of metal impurities or additives in the electrolyte of a vanadium redox ﬂow battery,” Journal of The Electrochemical Society, vol. 165, no. 7, pp. A1263–A1268, 2018.

[166] S. Kim, M. Vijayakumar, W. Wang, J. Zhang, B. Chen, Z. Nie, F. Chen, J. Hu, L. Li, and Z. Yang, “Chloride supporting electrolytes for all-vanadium redox ﬂow batteries,” Physical Chemistry Chemical Physics, vol. 13, no. 40, pp. 18186–18193, 2011.

[167] M. Kazacos and M. Kazacos, “High energy density vanadium electrolyte solutions, methods of preparation thereof and all-vanadium redox cells and batteries containing high energy vanadium electrolyte solutions,” Aug. 22 2002. US Patent 20030143456A1.

[168] M. Skyllas-Kazacos, C. Peng, and M. Cheng, “Evaluation of precipitation inhibitors for supersaturated vanadyl electrolytes for the vanadium redox battery,” Electrochemical and solid-state letters, vol. 2, no. 3, pp. 121–122, 1999.

[169] S. Li, K. Huang, S. Liu, D. Fang, X. Wu, D. Lu, and T. Wu, “Eﬀect of organic additives on positive electrolyte for vanadium redox battery,” Electrochimica Acta, vol. 56, no. 16, pp. 5483–5487, 2011. Bibliography 228

[170] X. Wu, S. Liu, N. Wang, S. Peng, and Z. He, “Inﬂuence of organic additives on electrochemical properties of the positive electrolyte for all-vanadium redox ﬂow battery,” Electrochimica Acta, vol. 78, pp. 475–482, 2012.

[171] D. Liu, F. Liu, and J. Liu, “Eﬀect of vanadium redox species on photoelec-

trochemical behavior of TiO2 and TiO2/WO3 photo-electrodes,” Journal of Power Sources, vol. 213, pp. 78–82, 2012.

[172] Z. Wei, D. Liu, C. Hsu, and F. Liu, “All-vanadium redox photoelectrochemical cell: an approach to store solar energy,” Electrochemistry Communications, vol. 45, pp. 79–82, 2014.

[173] D. Liu, Z. Wei, C.-j. Hsu, Y. Shen, and F. Liu, “Eﬃcient solar energy storage

using a TiO2/WO3 tandem photoelectrode in an all-vanadium photoelectrochemical cell,” Electrochimica Acta, vol. 136, pp. 435–441, 2014.

[174] D. Liu, W. Zi, S. D. Sajjad, C. Hsu, Y. Shen, M. Wei, and F. Liu, “Reversible electron storage in an all-vanadium photoelectrochemical storage cell: synergy between vanadium redox and hybrid photocatalyst,” Acs Catalysis, vol. 5, no. 4, pp. 2632–2639, 2015.

[175] S.-K. Park, J. Shim, J. H. Yang, C.-S. Jin, B. S. Lee, Y.-S. Lee, K.-H. Shin, and J.-D. Jeon, “The inﬂuence of compressed carbon felt electrodes on the performance of a vanadium redox ﬂow battery,” Electrochimica Acta, vol. 116, pp. 447–452, 2014.

[176] D. Aaron, Q. Liu, Z. Tang, G. Grim, A. Papandrew, A. Turhan, T. Za- wodzinski, and M. Mench, “Dramatic performance gains in vanadium redox ﬂow batteries through modiﬁed cell architecture,” Journal of Power sources, vol. 206, pp. 450–453, 2012. Bibliography 229

[177] Q. Liu, G. Grim, A. Papandrew, A. Turhan, T. A. Zawodzinski, and M. M. Mench, “High performance vanadium redox ﬂow batteries with optimized electrode conﬁguration and membrane selection,” Journal of The Electrochemical Society, vol. 159, no. 8, pp. A1246–A1252, 2012.

[178] M. L. Perry, R. M. Darling, and R. Zaﬀou, “High power density redox ﬂow battery cells,” ECS Transactions, vol. 53, no. 7, pp. 7–16, 2013.

[179] R. M. Darling and M. L. Perry, “The influence of electrode and channel configurations on flow battery performance,” Journal of The Electrochemical Society, vol. 161, no. 9, pp. A1381–A1387, 2014.

[180] X. Zhou, Y. Zeng, X. Zhu, L. Wei, and T. Zhao, “A high-performance dual- scale porous electrode for vanadium redox ﬂow batteries,” Journal of Power Sources, vol. 325, pp. 329–336, 2016.

[181] M.-H. Liu and K.-Y. Lee, “Electrode structure of vanadium redox ﬂow battery,” Aug. 19 2014. US Patent 8,808,897.

[182] A. Pelligri and P. Spaziante, “Process and accumulator for storing and releasing electrical energy,” UK Patent GB, vol. 2, no. 030, p. 349, 1980.

[183] S. Swathirajan and Y. M. Mikhail, “Membrane-electrode assemblies for electrochemical cells,” Dec. 21 1993. US Patent 5,272,017.

[184] D.-h. Ye and Z.-g. Zhan, “A review on the sealing structures of membrane electrode assembly of proton exchange membrane fuel cells,” Journal of power sources, vol. 231, pp. 285–292, 2013.

[185] C. Yao, H. Zhang, T. Liu, X. Li, and Z. Liu, “Cell architecture upswing based on catalyst coated membrane (CCM) for vanadium ﬂow battery,” Journal of Power Sources, vol. 237, pp. 19–25, 2013. Bibliography 230

[186] A. Tang, S. Ting, J. Bao, and M. Skyllas-Kazacos, “Thermal modelling and simulation of the all-vanadium redox ﬂow battery,” Journal of Power Sources, vol. 203, no. Supplement C, pp. 165 – 176, 2012.

[187] A. Tang, J. Bao, and M. Skyllas-Kazacos, “Thermal modelling of battery conﬁguration and self-discharge reactions in vanadium redox ﬂow battery,” Journal of Power Sources, vol. 216, no. Supplement C, pp. 489 – 501, 2012.

[188] T. Mohammadi and M. Skyllas-Kazacos, “Preparation of sulfonated composite membrane for vanadium redox ﬂow battery applications,” Journal of Membrane Science, vol. 107, no. 1, pp. 35 – 45, 1995.

[189] D. Chen, S. Wang, M. Xiao, and Y. Meng, “Preparation and properties of sulfonated poly(ﬂuorenyl ether ketone) membrane for vanadium redox ﬂow battery application,” Journal of Power Sources, vol. 195, no. 7, pp. 2089 – 2095, 2010.

[190] Q. Luo, H. Zhang, J. Chen, D. You, C. Sun, and Y. Zhang, “Preparation and characterization of Naﬁon/SPEEK layered composite membrane and its application in vanadium redox ﬂow battery,” Journal of Membrane Science, vol. 325, no. 2, pp. 553 – 558, 2008.

[191] Y. Wang, S. Wang, M. Xiao, D. Han, M. A. Hickner, and Y. Meng, “Layer- by-layer self-assembly of PDDA/PSS-SPFEK composite membrane with low vanadium permeability for vanadium redox ﬂow battery,” RSC Adv., vol. 3, pp. 15467–15474, 2013.

[192] C. Jia, J. Liu, and C. Yan, “A signiﬁcantly improved membrane for vanadium redox ﬂow battery,” Journal of Power Sources, vol. 195, no. 13, pp. 4380 – 4383, 2010. Bibliography 231

[193] E. Wiedemann, A. Heintz, and R. Lichtenthaler, “Transport properties of vanadium ions in cation exchange membranes:: Determination of diﬀusion coeﬃcients using a dialysis cell,” Journal of Membrane Science, vol. 141, no. 2, pp. 215 – 221, 1998.

[194] C. Sun, J. Chen, H. Zhang, X. Han, and Q. Luo, “Investigations on transfer of water and vanadium ions across naﬁon membrane in an operating vanadium redox ﬂow battery,” Journal of Power Sources, vol. 195, no. 3, pp. 890 – 897, 2010.

[195] P. Leung, Q. Xu, T. Zhao, L. Zeng, and C. Zhang, “Preparation of silica nanocomposite anion-exchange membranes with low vanadium-ion crossover for vanadium redox ﬂow batteries,” Electrochimica Acta, vol. 105, no. Sup- plement C, pp. 584 – 592, 2013.

[196] Y. Tanaka, “2 - fundamental properties of ion exchange membranes,” in Ion Exchange Membranes (Second Edition) (Y. Tanaka, ed.), pp. 29 – 65, Amsterdam: Elsevier, second edition ed., 2015.

[197] A. Fick, “Ueber diﬀusion,” Annalen der Physik, vol. 170, no. 1, pp. 59–86, 1855.

[198] S. Arrhenius, Uber die Dissociationswarme und den Einﬂuss der Temperatur auf den Dissociationsgrad der Elektrolyte. Leipzig : Wilhelm Leipzig : Wilhelm Engelmann, 1889.

[199] J. Qiu, L. Zhao, M. Zhai, J. Ni, H. Zhou, J. Peng, J. Li, and G. Wei, “Pre- irradiation grafting of styrene and maleic anhydride onto PVDF membrane and subsequent sulfonation for application in vanadium redox batteries,” Journal of Power Sources, vol. 177, no. 2, pp. 617 – 623, 2008. Bibliography 232

[200] A. Heintz, E. Wiedemann, and J. Ziegler, “Ion exchange diﬀusion in elec- tromembranes and its description using the maxwell-stefan formalism,” Jour- nal of Membrane Science, vol. 137, no. 1, pp. 121 – 132, 1997.

[201] J. Chen, B. Wang, and J. Yang, “Adsorption and diﬀusion of VO2+ and

+ VO2 across cation membrane for allvanadium redox ﬂow battery,” Solvent Extraction and Ion Exchange, vol. 27, no. 2, pp. 312–327, 2009.

[202] E. Wiedemann, A. Heintz, and R. Lichtenthaler, “Sorption isotherms of vana-

+ dium with H3O ions in cation exchange membranes,” Journal of Membrane Science, vol. 141, no. 2, pp. 207 – 213, 1998.

[203] M. Vijayakumar, M. Bhuvaneswari, P. Nachimuthu, B. Schwenzer, S. Kim, Z. Yang, J. Liu, G. L. Graff, S. Thevuthasan, and J. Hu, “Spectroscopic investigations of the fouling process on nafion membranes in vanadium redox flow batteries,” Journal of Membrane Science, vol. 366, no. 1, pp. 325 – 334, 2011.

[204] S. Cui and S. J. Paddison, “A molecular dynamics study of the eﬀects of V2+ and V3+ on the local structure of hydrated Naﬁon,” The Journal of Physical Chemistry C, vol. 119, no. 23, pp. 12848–12855, 2015.

[205] Y. Yan, Y. Li, M. Skyllas-Kazacos, and J. Bao, “Modelling and simulation of thermal behaviour of vanadium redox ﬂow battery,” Journal of Power Sources, vol. 322, pp. 116–128, 2016.

[206] Y. S. Sedesheva, V. S. Ivanov, A. I. Wozniak, and A. S. Yegorov, “Proton- exchange membranes based on sulfonated polymers,” Orient J Chem, vol. 32, no. 5, pp. 2283–2296, 2016. Bibliography 233

[207] E. Yang, K.-J. Chae, and I. S. Kim, “Assessment of diﬀerent ceramic ﬁltration membranes as a separator in microbial fuel cells,” Desalination and Water Treatment, vol. 57, no. 58, pp. 28077–28085, 2016.

[208] R. B. Bird, W. E. Stewart, and E. N. Lightfoot, Transport Phenomena. John Wiley & Sons, INC, 2002.

[209] Y. A. Gandomi, T. A. Zawodzinski, and M. M. Mench, “Concentrated solution model of transport in all vanadium redox ﬂow battery membrane separator,” ECS Transactions, vol. 61, no. 13, pp. 23–32, 2014.

[210] Y. A. Gandomi, D. Aaron, and M. Mench, “Coupled membrane transport parameters for ionic species in all-vanadium redox ﬂow batteries,” Electrochimica Acta, vol. 218, pp. 174–190, 2016.

[211] Y. Ashraf Gandomi, D. S. Aaron, and M. M. Mench, “Inﬂuence of membrane equivalent weight and reinforcement on ionic species crossover in all-vanadium redox ﬂow batteries,” Membranes, vol. 7, no. 2, p. 29, 2017.

[212] S. Kim and M. Mench, “Investigation of temperature-driven water transport in polymer electrolyte fuel cell: Thermo-osmosis in membranes,” Journal of Membrane Science, vol. 328, no. 1-2, pp. 113–120, 2009.

[213] T. F. Fuller and J. Newman, “Experimental determination of the transport number of water in naﬁon 117 membrane,” Journal of The Electrochemical Society, vol. 139, no. 5, pp. 1332–1337, 1992.

[214] R. Darling, K. Gallagher, W. Xie, L. Su, and F. Brushett, “Transport property requirements for ﬂow battery separators,” Journal of The Electrochemical Society, vol. 163, no. 1, pp. A5029–A5040, 2016. Bibliography 234

[215] F. Donnan and E. Guggenheim, “Die genaue thermodynamik der mem- brangleichgewichte,” Zeitschrift f¨urPhysikalische Chemie, vol. 162, no. 1, pp. 346–360, 1932.

[216] F. Sepehr and S. J. Paddison, “Eﬀect of sulfuric and triﬂic acids on the hydration of vanadium cations: An ab initio study,” The Journal of Physical Chemistry A, vol. 119, no. 22, pp. 5749–5761, 2015.

[217] v. H. Strehlow and H. Wendt, “Fast ionic reactions in solution. iv. the formation of the vanadyl sulfate complex in aqueous solution,” Inorganic Chemistry, vol. 2, no. 1, pp. 6–10, 1963.

[218] H. Yamamoto, “Mobility of counter ion in ion exchange membrane,” Bulletin of the Society of Sea Water Science, Japan, vol. 22, no. 4, pp. 323–326, 1968.

[219] R. A. Morgan and W. H. Sloan, “Extrusion ﬁnishing of perﬂuorinated copolymers,” Dec. 2 1986. US Patent 4,626,587.

[220] R. A. Smith and M. S. Withers, “Coextruded multilayer cation exchange membranes,” Mar. 20 1984. US Patent 4,437,952.

[221] D. E. Curtin, R. D. Lousenberg, T. J. Henry, P. C. Tangeman, and M. E. Tisack, “Advanced materials for improved pemfc performance and life,” Journal of power Sources, vol. 131, no. 1, pp. 41–48, 2004.

[222] W. Xie, R. M. Darling, and M. L. Perry, “Processing and pretreatment eﬀects on vanadium transport in naﬁon membranes,” Journal of The Electrochemical Society, vol. 163, no. 1, pp. A5084–A5089, 2016.

[223] A. Tang, J. Bao, and M. Skyllas-Kazacos, “Dynamic modelling of the eﬀects of ion diﬀusion and side reactions on the capacity loss for vanadium redox Bibliography 235

ﬂow battery,” Journal of Power Sources, vol. 196, no. 24, pp. 10737–10747, 2011.

[224] M.-A. Goulet, M. Skyllas-Kazacos, and E. Kjeang, “The importance of wetting in carbon paper electrodes for vanadium redox reactions,” Carbon, vol. 101, pp. 390–398, 2016.

[225] E. Sum and M. Skyllas-Kazacos, “A study of the V (II)/V (III) redox couple for redox ﬂow cell applications,” Journal of Power sources, vol. 15, no. 2-3, pp. 179–190, 1985.

[226] S. Zhong and M. Skyllas-Kazacos, “Electrochemical behaviour of vanadium (V)/vanadium (IV) redox couple at graphite electrodes,” Journal of Power Sources, vol. 39, no. 1, pp. 1–9, 1992.

[227] H. Kaneko, K. Nozaki, Y. Wada, T. Aoki, A. Negishi, and M. Kamimoto, “Vanadium redox reactions and carbon electrodes for vanadium redox ﬂow battery,” Electrochimica Acta, vol. 36, no. 7, pp. 1191–1196, 1991.

[228] K. K. Cline, M. T. McDermott, and R. L. McCreery, “Anomalously slow electron transfer at ordered graphite electrodes: inﬂuence of electronic factors and reactive sites,” The Journal of Physical Chemistry, vol. 98, no. 20, pp. 5314–5319, 1994.

[229] M. Gattrell, J. Park, B. MacDougall, J. Apte, S. McCarthy, and C. Wu, “Study of the mechanism of the vanadium 4+/5+ redox reaction in acidic solutions,” Journal of the Electrochemical Society, vol. 151, no. 1, pp. A123– A130, 2004.

[230] T. Yamamura, N. Watanabe, T. Yano, and Y. Shiokawa, “Electron-transfer

3+ 4+ + 2+ 2+ 3+ 2+ + kinetics of Np / Np , NpO2 / NpO2 ,V /V , and VO / VO2 at Bibliography 236

carbon electrodes,” Journal of The Electrochemical Society, vol. 152, no. 4, pp. A830–A836, 2005.

[231] Y. Wen, H. Zhang, P. Qian, H. Zhou, P. Zhao, B. Yi, and Y. Yang, “Studies

3+ 2+ − on iron (Fe / Fe )-complex/bromine (Br2/ Br ) redox ﬂow cell in sodium acetate solution,” Journal of the Electrochemical Society, vol. 153, no. 5, pp. A929–A934, 2006.

[232] C.-H. Wu, H.-Y. Liao, K.-L. Hsueh, and J.-S. Hung, “Study of the kinetics of vanadium redox reaction by rotating disk electrode,” ECS Transactions, vol. 35, no. 32, pp. 11–22, 2011.

[233] H. Liu, Q. Xu, C. Yan, Y. Cao, Y. Qiao, et al., “The eﬀect of temperature on the electrochemical behavior of the V (IV)/V (V) couple on a graphite electrode,” International Journal of Electrochemical Science, 2011.

[234] J. Friedl, C. M. Bauer, A. Rinaldi, and U. Stimming, “Electron transfer

2+ + kinetics of the VO /VO2 reaction on multi-walled carbon nanotubes,” Carbon, vol. 63, no. Supplement C, pp. 228 – 239, 2013.

[235] H.-Y. Yang, K.-L. Hsueh, C.-L. Hsieh, and J.-S. Hung, “Study of the kinetics of vanadium (II)/(III) redox reaction,” ECS Transactions, vol. 50, no. 45, pp. 87–92, 2013.

[236] W. Wang, X. Fan, J. Liu, C. Yan, and C. Zeng, “Temperature-related reaction kinetics of the vanadium (IV)/(V) redox couple in acidic solutions,” RSC Advances, vol. 4, no. 61, pp. 32405–32411, 2014.

[237] M.-A. Goulet, M. Eikerling, and E. Kjeang, “Direct measurement of electrochemical reaction kinetics in ﬂow-through porous electrodes,” Electrochem- istry Communications, vol. 57, pp. 14–17, 2015. Bibliography 237

[238] F. Chen, H. Chen, Q. Sun, J. Liu, C. Yan, and Q. Liu, “Electron-transfer kinetics of V(IV)/V(V) and V(II)/V(III) couples for diﬀerent state of charge,” Int. J. Electrochem. Sci, vol. 10, pp. 725–736, 2015.

[239] E. Hollmann, “Factors inﬂuencing kinetics of electrode processes in vanadium redox ﬂow batteries,” Master’s thesis, University of Tennessee Honors Thesis Projects, 2015.

[240] A. Bourke, M. Miller, R. P. Lynch, X. Gao, J. Landon, J. Wainright, R. Savinell, and D. Buckley, “Electrode kinetics of vanadium ﬂow batteries: contrasting responses of VII-VIII and VIV-VV to electrochemical pretreatment of carbon,” Journal of The Electrochemical Society, vol. 163, no. 1, pp. A5097–A5105, 2016.

[241] G. Oriji, Y. Katayama, and T. Miura, “Investigations on V (IV)/V (V) and V (II)/V (III) redox reactions by various electrochemical methods,” Journal of Power Sources, vol. 139, no. 1, pp. 321–324, 2005.

[242] R. Taylor and A. Humﬀray, “Electrochemical studies on glassy carbon electrodes: Ii. oxygen reduction in solutions of high pH (pH>10),” Journal of Electroanalytical Chemistry and Interfacial Electrochemistry, vol. 64, no. 1, pp. 63–84, 1975.

[243] S. Neﬀe, “Electrochemical corrosion of glasslike carbon in sulfuric acid solution,” Carbon, vol. 26, no. 5, pp. 687–692, 1988.

[244] D. Laser and M. Ariel, “The anodic behavior of glassy carbon in acid solution: A spectroelectrochemical study,” Journal of Electroanalytical Chemistry and Interfacial Electrochemistry, vol. 52, no. 2, pp. 291–303, 1974.

[245] M. Santhiago, C. M. Maroneze, C. C. Silva, M. N. Camargo, and L. T. Kubota, “Electrochemical oxidation of glassy carbon provides similar electro- Bibliography 238

chemical response as graphene oxide prepared by tour or hummers routes,” ChemElectroChem, vol. 2, no. 5, pp. 761–767, 2015.

[246] A. Dekanski, J. Stevanović,R. Stevanović,B. Z.ˇ Nikolić,and V. M. Jovanović, “Glassy carbon electrodes: I. characterization and electrochemical activation,” Carbon, vol. 39, no. 8, pp. 1195–1205, 2001.

[247] A. Bourke, M. Miller, R. Lynch, J. Wainright, R. Savinell, and D. Buckley, “Eﬀect of cathodic and anodic treatments of carbon on the electrode kinetics of VIV /VV oxidation-reduction,” Journal of The Electrochemical Society, vol. 162, no. 8, pp. A1547–A1555, 2015.

[248] A. Bourke, R. P. Lynch, and D. N. Buckley, “Eﬀect of electrode pretreatment

2+ + on the cyclic voltammetry of VO /VO2 at a glassy carbon electrode,” ECS Transactions, vol. 53, no. 30, pp. 59–67, 2013.

[249] A. Bourke, N. Quill, R. P. Lynch, and D. N. Buckley, “Eﬀect of pretreatment

2+ + 2+ 3+ on the rate of the VO /VO2 and V /V reactions at a carbon electrode,” ECS Transactions, vol. 61, no. 37, pp. 15–26, 2014.

[250] D. N. Buckley, X. Gao, R. P. Lynch, N. Quill, and M. J. Leahy, “Towards optical monitoring of vanadium redox ﬂow batteries (VRFBs): an investigation of the underlying spectroscopy,” Journal of The Electrochemical Society, vol. 161, no. 4, pp. A524–A534, 2014.

[251] G. Jenkins and K. Kawamura, “Structure of glassy carbon,” Nature, vol. 231, no. 5299, p. 175, 1971.

[252] A. Craievich, “On the structure of glassy carbon,” Materials Research Bulletin, vol. 11, no. 10, pp. 1249–1255, 1976. Bibliography 239

[253] E.-M. Hammer, B. Berger, and L. Komsiyska, “Improvement of the performance of graphite felt electrodes for vanadium-redox-ﬂow-batteries by plasma treatment,” International Journal of Renewable Energy Development, vol. 3, no. 1, pp. 7–12, 2014.

[254] N. Pour, D. G. Kwabi, T. Carney, R. M. Darling, M. L. Perry, and Y. Shao- Horn, “Inﬂuence of edge-and basal-plane sites on the vanadium redox kinetics for ﬂow batteries,” The Journal of Physical Chemistry C, vol. 119, no. 10, pp. 5311–5318, 2015.

[255] H. Liu, Q. Xu, C. Yan, and Y. Qiao, “Corrosion behavior of a positive graphite electrode in vanadium redox ﬂow battery,” Electrochimica Acta, vol. 56, no. 24, pp. 8783–8790, 2011.

[256] Y. Chang, G. Han, D. Fu, F. Liu, M. Li, Y. Li, and C. Liu, “Like n-doped graphene ﬁlms prepared by hydroxylamine diﬀusion induced assembly and their ultrahigh-rate capacitive properties,” Electrochimica Acta, vol. 115, pp. 461–470, 2014.

[257] G. Ilangovan and K. Chandrasekara Pillai, “Electrochemical and XPS characterization of glassy carbon electrode surface eﬀects on the preparation of a monomeric molybdate (VI)-modiﬁed electrode,” Langmuir, vol. 13, no. 3, pp. 566–575, 1997.

[258] C. Kozlowski and P. M. Sherwood, “X-ray photoelectron-spectroscopic studies of carbon-ﬁbre surfaces. Part 5.the eﬀect of ph on surface oxidation,” Journal of the Chemical Society, Faraday Transactions 1: Physical Chemistry in Condensed Phases, vol. 81, no. 11, pp. 2745–2756, 1985. Bibliography 240

[259] M. D. LeVan, Fundamentals of adsorption: proceedings of the Fifth Interna- tional Conference on Fundamentals of Adsorption, vol. 356. Springer Science & Business Media, 2012.

[260] Y. Wang, Y. Shao, D. W. Matson, J. Li, and Y. Lin, “Nitrogen-doped graphene and its application in electrochemical biosensing,” ACS nano, vol. 4, no. 4, pp. 1790–1798, 2010.

[261] N. D´ıez,A. Sliwak,´ S. Gryglewicz, B. Grzyb, and G. Gryglewicz, “Enhanced reduction of graphene oxide by high-pressure hydrothermal treatment,” RSC Advances, vol. 5, no. 100, pp. 81831–81837, 2015.

[262] D. Yang, A. Velamakanni, G. Bozoklu, S. Park, M. Stoller, R. D. Piner, S. Stankovich, I. Jung, D. A. Field, C. A. Ventrice, et al., “Chemical analysis of graphene oxide ﬁlms after heat and chemical treatments by x-ray photoelectron and micro-raman spectroscopy,” Carbon, vol. 47, no. 1, pp. 145–152, 2009.

[263] L. Yue, W. Li, F. Sun, L. Zhao, and L. Xing, “Highly hydroxylated carbon ﬁbres as electrode materials of all-vanadium redox ﬂow battery,” Carbon, vol. 48, no. 11, pp. 3079–3090, 2010.

[264] R. de Levie, “Fractals and rough electrodes,” Journal of electroanalytical chemistry and interfacial electrochemistry, vol. 281, no. 1-2, pp. 1–21, 1990.

[265] O. Kardos, “Current distribution on microproﬁles,” Plating, vol. 61, no. 2, p. 129, 1974.

[266] C. Largeot, C. Portet, J. Chmiola, P.-L. Taberna, Y. Gogotsi, and P. Simon, “Relation between the ion size and pore size for an electric double-layer capaci- tor,” Journal of the American Chemical Society, vol. 130, no. 9, pp. 2730–2731, 2008. Bibliography 241

[267] A. J. Bard, L. R. Faulkner, J. Leddy, and C. G. Zoski, Electrochemical methods: fundamentals and applications, vol. 2. wiley New York, 1980.

[268] M. Kazacos, M. Cheng, and M. Skyllas-Kazacos, “Vanadium redox cell electrolyte optimization studies,” Journal of Applied electrochemistry, vol. 20, no. 3, pp. 463–467, 1990.

[269] T.-M. Tseng, R.-H. Huang, C.-Y. Huang, C.-C. Liu, K.-L. Hsueh, and F.-S. Shieu, “Carbon felt coated with titanium dioxide/carbon black composite as negative electrode for vanadium redox ﬂow battery,” Journal of The Electrochemical Society, vol. 161, no. 6, pp. A1132–A1138, 2014.

[270] L. Campanella and G. Pistoia, “MoO3: A new electrode material for non- aqueous secondary battery applications,” Journal of The Electrochemical Society, vol. 118, no. 12, pp. 1905–1908, 1971.

[271] D. O. Scanlon, G. W. Watson, D. Payne, G. Atkinson, R. Egdell, and D. Law,

“Theoretical and experimental study of the electronic structures of MoO3 and

MoO2,” The Journal of Physical Chemistry C, vol. 114, no. 10, pp. 4636–4645, 2010.

[272] K. Dewangan, N. N. Sinha, P. K. Sharma, A. C. Pandey, N. Munichandraiah,

and N. Gajbhiye, “Synthesis and characterization of single-crystalline α-MoO3 nanoﬁbers for enhanced Li-ion intercalation applications,” CrystEngComm, vol. 13, no. 3, pp. 927–933, 2011.

[273] A. Chithambararaj, D. B. Mathi, N. R. Yogamalar, and A. C. Bose, “Struc-

tural evolution and phase transition of [NH4]6Mo7O24.4H2O to 2D layered

MoO3−x,” Materials Research Express, vol. 2, no. 5, p. 055004, 2015. Bibliography 242

[274] W. Tang, L. Liu, S. Tian, L. Li, Y. Yue, Y. Wu, and K. Zhu, “Aqueous

supercapacitors of high energy density based on MoO3 nanoplates as anode material,” Chemical communications, vol. 47, no. 36, pp. 10058–10060, 2011.

[275] J. Lipsch and G. Schuit, “The CoO - MoO3 - Al2O3 catalyst: III. Catalytic properties,” Journal of Catalysis, vol. 15, no. 2, pp. 179–189, 1969.

[276] M. Abrantes, T. R. Amarante, M. M. Antunes, S. Gago, F. A. Almeida Paz, I. Margiolaki, A. E. Rodrigues, M. Pillinger, A. A. Valente, and I. S. Goncalves, “Synthesis, structure, and catalytic performance in cyclooctene epoxidation

of a molybdenum oxide/bipyridine hybrid material:{[MoO3 (bipy)][MoO3

(H2O)]} n,” Inorganic chemistry, vol. 49, no. 15, pp. 6865–6873, 2010.

[277] T. Matsuda, H. Shiro, H. Sakagami, and N. Takahashi, “Isomerization of heptane on molybdenum oxides treated with hydrogen,” Catalysis letters, vol. 47, no. 2, pp. 99–103, 1997.

[278] A. W. Scaroni, R. G. Jenkins, and P. L. Walker, “Coke deposition on Co-

Mo/Al2O3 and Co-Mo/C catalysts,” Applied catalysis, vol. 14, pp. 173–183, 1985.

[279] P. Okoye and B. Hameed, “Review on recent progress in catalytic carboxy- lation and acetylation of glycerol as a byproduct of biodiesel production,” Renewable and Sustainable Energy Reviews, vol. 53, pp. 558–574, 2016.

[280] F. El Kady, S. Shaban, and A. A. El Naga, “Catalytic dehydrogenation

of cyclohexene over MoO3/γ-Al2O3 catalysts,” Transition Metal Chemistry, vol. 36, no. 2, pp. 237–244, 2011.

[281] Vanadium electrolyte cost and purity consideration for vanadium redox ﬂow battery application, COM 2014-Conference of Metallurgists Proceedings, 2014. Bibliography 243

[282] R.-H. Huang, C.-H. Sun, T.-m. Tseng, W.-k. Chao, K.-L. Hsueh, and F.-S. Shieu, “Investigation of active electrodes modiﬁed with platinum/multiwalled carbon nanotube for vanadium redox ﬂow battery,” Journal of the Electro- chemical Society, vol. 159, no. 10, pp. A1579–A1586, 2012.

[283] H.-M. Tsai, S.-J. Yang, C.-C. M. Ma, and X. Xie, “Preparation and electrochemical activities of iridium-decorated graphene as the electrode for all-vanadium redox ﬂow batteries,” Electrochimica Acta, vol. 77, pp. 232–236, 2012.

[284] T.-M. Tseng, R.-H. Huang, C.-Y. Huang, K.-L. Hsueh, and F.-S. Shieu, “A kinetic study of the platinum/carbon anode catalyst for vanadium redox ﬂow battery,” Journal of The Electrochemical Society, vol. 160, no. 4, pp. A690– A696, 2013.

[285] Z. He, L. Chen, Y. He, C. Chen, Y. Jiang, Z. He, and S. Liu, “Eﬀect of In3+ ions on the electrochemical performance of the positive electrolyte for vanadium redox ﬂow batteries,” Ionics, vol. 19, no. 12, pp. 1915–1920, 2013.

[286] T.-M. Tseng, R.-H. Huang, C.-Y. Huang, K.-L. Hsueh, and F.-S. Shieu, “Improvement of titanium dioxide addition on carbon black composite for negative electrode in vanadium redox ﬂow battery,” Journal of The Electro- chemical Society, vol. 160, no. 8, pp. A1269–A1275, 2013.

[287] D. J. Suárez,Z. González,C. Blanco, M. Granda, R. Menéndez,and R. San- tamar´ıa,“Graphite felt modified with bismuth nanoparticles as negative electrode in a vanadium redox flow battery,” ChemSusChem, vol. 7, no. 3, pp. 914–918, 2014. Bibliography 244

[288] H. Zhou, Y. Shen, J. Xi, X. Qiu, and L. Chen, “ZrO2-nanoparticle-modified graphite felt: bifunctional effects on vanadium flow batteries,” ACS applied materials & interfaces, vol. 8, no. 24, pp. 15369–15378, 2016.

[289] S. Okashy, M. Noked, T. Zimrin, and D. Aurbach, “The study of activated

carbon/CNT/MoO3 electrodes for aqueous pseudo-capacitors,” Journal of The Electrochemical Society, vol. 160, no. 9, pp. A1489–A1496, 2013. Appendix A

Model Development

A.1 Dynamic Model Development

The constant current charging-discharging dynamic model was developed by Tang et al [223] based on the following assumptions and molar mass balance. Assumptions:

(1) The battery is operated at constant temperature;

(2) Electrolyte is circulated at suﬃcient ﬂow rate;

(3) The self-discharge reactions are instantaneous;

(4) No volumetric change occurs during the cycling test;

(5) The cell reactions rates are determined by Faraday’s electrolysis law;

(6) Side reactions (3.3) and (3.6) can be ignored;

(7) The concentration of H+ remains constant during cycling;

(8) The electrolytes are perfectly mixed;

(9) The cell resistance remain the same over cycling.

245 A. Model Development 246

Molar mass balance during charging-discharging: dc iA c c c V 2 = ± − k 2 S − 2k 5 S − k 4 S (A.1) s dt zF 2 d 5 d 4 d dc iA c c c V 3 = ∓ − k 3 S + 3k 5 S + 2k 4 S (A.2) s dt zF 3 d 5 d 4 d dc iA c c c V 4 = ∓ − k 4 S + 3k 2 S + 2k 3 S (A.3) s dt zF 4 d 2 d 3 d dc iA c c c V 5 = ± − k 5 S − 2k 2 S − k 3 S (A.4) s dt zF 5 d 2 d 3 d where

Vs is the half cell volume (L);

2+ −1 c2 is the concentration of V ions (mol·L );

3+ −1 c3 is the concentration of V ions (mol·L );

2+ −1 c4 is the concentration of VO ions (mol·L );

+ −1 c5 is the concentration of VO2 ion (mol·L ); i is the current density (mA·cm−2); A is the electrode area (cm2); z is the electron transfer number; F is the Faraday’s constant (96485 C·mol−1); d is the membrane thickness (dm); S is the membrane area (dm2);

2+ 2 −1 k2 is the diﬀusion coeﬃcient of V ions across the membrane (dm ·s );

3+ 2 −1 k3 is the diﬀusion coeﬃcient of V ions across the membrane (dm ·s );

2+ 2 −1 k4 is the diﬀusion coeﬃcient of VO ions across the membrane (dm ·s );

+ 2 −1 k5 is the diﬀusion coeﬃcient of VO2 ions across the membrane (dm ·s );

Cell potential: 0 RT c2(t) × c5(t) E(t) = E0 + ln ± ir (A.5) zF c3(t) × c4(t) where E is the cell potential (V); A. Model Development 247

0 E0 is the formal cell potential (1.4 V at 50% SOC of 2 M V in 5 M total sulphate solution); R is the gas constant (8.314 J·mol−1·K−1); r is the cell resistance (Ω·cm2).

A.2 Thermal Model Development

The thermal simulation model associated with the vanadium diﬀusion across the membrane was developed by Yan et al [205] based on the following assumptions and molar mass and energy balance: Assumptions:

(1) The tank is full of electrolyte;

(2) The concentration and temperature in each component are uniformly distributed;

(3) The self-discharge reactions are instantaneous;

(4) No volumetric change occurs during the cycling test;

(5) The cell reactions rates are determined by Faraday´s electrolysis law;

(6) Side reactions (3.3) and (3.6) can be ignored;

(7) The concentration of H+ remains constant during cycling;

(8) The electrolytes are perfectly mixed;

(9) The cell resistance remain the same over cycling.

Molar mass balance: When the VRFB system is under normal charge-discharge reactions or normal self-discharge reactions during the standby case with pumps are oﬀ, A. Model Development 248

V2+ ions:

V dc−n I c−n c+n c+n c 2 = Q (ct − c−n) ± − K 2 S − 2K 5 S − K 4 S (A.6) 2 dt c 2 2 zF 2 d 5 d 4 d N dct X V 2 = Q c−j − NQ ct (A.7) t dt c 2 s 2 j=1

V3+ ions:

V dc+n c 3 = 0 (A.8) 2 dt V dc−n I c−n c+n c+n c 3 = Q (ct − c−n) ∓ − K 3 S + 3K 5 S + 2K 4 S (A.9) 2 dt c 3 3 zF 3 d 5 d 4 d N dct X V 3 = Q c−j − NQ ct (A.10) t dt c 3 s 3 j=1

VO2+ ions:

V dc−n c 4 = 0 (A.11) 2 dt V dc+n I c+n c−n c−n c 4 = Q (ct − c+n) ∓ − K 4 S + 3K 2 S + 2K 3 S (A.12) 2 dt c 4 4 zF 4 d 2 d 3 d N dct X V 4 = Q c+j − NQ ct (A.13) t dt c 4 s 4 j=1

+ VO2 ions: V dc+n I c+n c−n c−n c 5 = Q (ct − c+n) ± − K 5 S − 2K 2 S − K 3 S (A.14) 2 dt c 5 5 zF 5 d 2 d 3 d N dct X V 5 = Q c+j − NQ ct (A.15) t dt c 5 s 5 j=1 where

3 Vc is the volume of total cell stack (m );

3 Vt is the volume of the tank (m );

±n x+ cx is the concentration of vanadium ion V in the nth positive or negative half-cell (mol·m−3);

t x+ −3 cx is the concentration of vanadium ion V in the tank (mol·m );

3 −1 Qc is the volumetric ﬂow rate of the electrolyte in the cell stack; (m s ) A. Model Development 249

3 −1 Qt is the volumetric ﬂow rate of the electrolyte in the cell stack; (m ·s ) N is the number of cells in a stack; I is the current (A); F is the Faraday’s constant (96485 C·mol−1); S is the membrane area (dm2); d is the membrane thickness (dm);

x+ Kx is the diffusion coefficient of V ions across the membrane at different temperature (dm2·s−1).

2+ + If V ions are depleted in the nth cell before VO2 ions during the idle period, the concentrations of vanadium ions in the cell should follow:

V dc−n c 2 = 0 (A.16) 2 dt V dc+n c 3 = 0 (A.17) 2 dt V dc−n c−n c+n c 3 = −K 3 S − K 5 S (A.18) 2 dt 3 d 5 d V dc+n c−n − c+n c−n c+n c 4 = −K 4 4 S + 2K 3 S + 2K 5 S (A.19) 2 dt 4 d 3 d 5 d V dc−n c−n − c+n c 4 = K 4 4 S (A.20) 2 dt 4 d V dc+n c+n c−n c 5 = −K 5 S − K 3 S (A.21) 2 dt 5 d 3 d A. Model Development 250

+ 2+ Similarly, if VO2 ions are depleted prior to V ions, the equation would be:

V dc−n c−n c+n c 2 = −K 2 S − K 4 S (A.22) 2 dt 2 d 4 d V dc+n c+n − c−n c 3 = K 3 3 S (A.23) 2 dt 3 d V dc−n c+n − c−n c−n c+n c 3 = −K 3 3 S + 2K 2 S + 2K 4 S (A.24) 2 dt 3 d 2 d 4 d V dc+n c+n c−n c 4 = −K 4 S − K 2 S (A.25) 2 dt 4 d 2 d V dc−n c 4 = 0 (A.26) 2 dt V dc+n c 5 = 0 (A.27) 2 dt

2+ + If both V and VO2 ions are depleted before discharging starts, no more self-discharge reactions would occur but only the diﬀusion of V3+ and VO2+ ions through the membrane until the concentrations of V3+ and VO2+ are the same in each half-cell:

V c−n c 2 = 0 (A.28) 2 dt V dc+n c−n − c+n c 3 = K 3 3 S (A.29) 2 dt 3 d V dc−n c+n − c−n c 3 = K 3 3 S (A.30) 2 dt 3 d V dc+n c−n − c+n c 4 = K 4 4 S (A.31) 2 dt 4 d V dc−n c+n − c−n c 4 = K 4 4 S (A.32) 2 dt 4 d V dc+n c 5 = 0 (A.33) 2 dt

Energy balance: During normal charge-discharge operation and no vanadium ions are depleted: A. Model Development 251

For the nth cell: dT C ρV n =Q C ρ(T + − T ) + Q C ρ(T − − T ) p c dt c p in n c p in n

+ UxAx(Tn+1 − Tn) + UxAx(Tn−1 − Tn)

+ 2UyAy(Tair − Tn) + 2UzAz(Tair − Tn) (A.34) c−n c−n + K 2 S · (−∆H ) + K 3 S · (−∆H ) 2 d 1 3 d 3 +n +n c4 c5 2 r + K4 S · (−∆H6) + K5 S · (−∆H4) + I d d Ax where

−1 −1 Cp is the speciﬁc heat capacity of electrolyte, 3.2 J·g ·K ; ρ is electrolyte density, 1.354×106 g·m3;

Tn is the nth cell temperature(K);

+ Tin is the temperature (K) of the pipe where electrolyte enters the positive half-cell; − Tin is the temperature (K) of the pipe where electrolyte enters the negative half-cell;

Tair is the time variant ambient temperature (K);

UxAx is the overall heat transfer capability of the nth cell in the x direction (J·K−1·s−1);

UyAy is the overall heat transfer capability of the nth cell in the y direction (J·K−1·s−1);

UzAz is the overall heat transfer capability of the nth cell in the z direction (J·K−1·s−1);

−1 −∆H1 is the enthalpy change of reaction (3.1), -220000 J·mol ;

−1 −∆H3 is the enthalpy change of reaction (3.3), -64000 J·mol ;

−1 −∆H4 is the enthalpy change of reaction (3.4), -246800 J·mol ;

−1 −∆H6 is the enthalpy change of reaction (3.6), -91200 J·mol ;

Qn is the ﬂow factor of the electrolyte in the negative half-cell;

Qp is the ﬂow factor of the electrolyte in the positive half-cell; N is the number of cells in a stack; A. Model Development 252

I is the current (A);

x+ Kx is the diffusion coefficient of V ions across the membrane at different temperature (dm2·s−1).

It should be noted that the Tn−1 for the ﬁrst cell and the Tn+1 for the last cell should be Tair.

+ 2+ During standby period, if VO2 ions are depleted and V ions are still in the stack: dT C ρV n =U A (T − T ) + U A (T − T ) p c dT x x n+1 n x x n−1 n

+ 2UyAy(Tair − Tn) + 2UzAz(Tair − Tn) (A.35) c−n c+n + K 3 S · (−∆H ) + K 5 S · (−∆H ) 3 d 3 5 d 3

+ 2+ If VO2 ions are depleted and V ions still exist in the stack: dT C ρV n =U A (T − T ) + U A (T − T n) p c dT x x n+1 n x x n−1

+ 2UyAy(Tair − Tn) + 2UzAz(Tair − Tn) (A.36) c−n c+n + K 2 S · (−∆H ) + K 4 S · (−∆H ) 2 d 6 4 d 6 For the pipes:

dT + C ρ(NV ) in = Q C ρ(T + − T +) + U (NA )(T − T +) + W (A.37) p p dt s p t in p p air in pump dT − C ρ(NV ) in = Q C ρ(T − − T −) + U (NA )(T − T −) + W (A.38) p p dt s p t in p p air in pump dT + C ρV out = Q C ρ(T − T + ) + U A (T − T + ) (A.39) p p dt c p n outn p p air outn dT − C ρV outn = Q C ρ(T − T − ) + U A (T − T − ) (A.40) p p dt c p n outn p p air outn

Qs = NQc. (A.41) where

+ Toutn is the temperature of the pipe where electrolyte leaves the nth positive half- cell; A. Model Development 253

− Toutn is the temperature of the pipe where electrolyte leaves the nth negative half-cell;

+ Tt is the temperature of the tank storing the electrolyte for postive half-cell; − Tt is the temperature of the tank storing the electrolyte for neagtive half-cell;

−1 −1 UpAp is the overall heat transfer capability of the pipe (J·K ·s );

3 −1 Qs is the volumetric ﬂow rate of stack (m s );

3 −1 Qc is the volumetric ﬂow rate of cell (m s ).

s For the tanks where the electrolyte from each cell merge, the temperature Tout is assumed to be the average of the temperature of all cells:

dT + C ρV t = Q C ρ(T +s − T +) + U A (T − T +) (A.42) p t dt s p out t t t air t dT − C ρV t = Q C ρ(T −s − T −) + U A (T − T −) (A.43) p t dt s p out t t t air t N 1 X T ±s = T ± (A.44) out N outn n=1 where

+s Tout is the average temperature of the merged electrolyte ﬂowing out the all the positive half-cells;

−s Tout is the average temperature of the merged electrolyte flowing out the all the negative half-cells. Electrolyte flow rate: Q I Q = f (A.45) FCvSOC where Q is the volumetric flowrate (m3s−1);

Qf is the ﬂow factor;

Cv is the total vanadium concentration in the tank; SOC is the average state of charge of the electrolyte in two tanks. A. Model Development 254

Arrhenius equation:

Ea 0 − RT Kx = kxe (A.46)

where

x+ Kx is the diffusion coefficient of V ions across the membrane at different temperature (dm2·s−1);

0 x+ 2 −1 kx is the prefactor of V ions (dm ·s );

Ea is the activation energy for diﬀusion. The activation energy was assumed to remain constant (17341 J·mol−1) regard-

0 less of the temperature or type of membrane and the prefactor kx was considered to be the same for all temperatures but varied with the kind of vanadium ions. The permeability rates of vanadium ions across FAP450, VB2 and F930 membranes were obtained at 303.3 K (Table 3.4 in Chapter3). Therefore, the prefactors for each vanadium ions across the membrane were re-calculated and listed in

Table A.1, so as the values for Naﬁonr 115 from reported data [194] at 298.3 K.

0 2 −1 Table A.1: Prefactors kx (m ·s ) of vanadium ion x across the membrane

0 0 0 0 Membrane k2 k3 k4 k5 FAP450 1.57 × 10−9 1.05 × 10−9 1.93 × 10−9 1.87 × 10−9 VB2 2.69 × 10−10 1.80 × 10−9 2.98 × 10−10 3.93 × 10−10 F930 3.16 × 10−11 1.82 × 10−9 3.74 × 10−9 4.99 × 10−10 Naﬁonr 115 9.60 × 10−9 3.53 × 10−9 7.47 × 10−9 6.45 × 10−9

A.3 Sensitivity Analysis

In order to gain an understanding of the sensitivity of the model parameters on the simulation results, however, some additional simulations were carried out in which the diffusion coefficients of three of the vanadium ions was kept the same as before, but the fourth was doubled. The diffusion coefficients of vanadium ions across A. Model Development 255

FAP450 were used an example here as the experimental diffusion coefficients of four vanadium ions are in the similar absolute values and order of magnitudes but showed significant decay in capacity loss. The detailed parameters for sensitivity analysis are listed in Table A.2.

Table A.2: Diﬀusion coeﬃcients (dm2 · s−1) for simulations

2+ 3+ 2+ + Membrane k(V ) k(V ) k(VO ) k(VO2 ) FAP450 1.61 × 10−10 1.08 × 10−10 1.98 × 10−10 1.92 × 10−10 −10 −10 −10 −10 FAP450-2k2 3.22 × 10 1.08 × 10 1.98 × 10 1.92 × 10 −10 −10 −10 −10 FAP450-2k3 1.61 × 10 2.16 × 10 1.98 × 10 1.92 × 10 −10 −10 −10 −10 FAP450-2k4 1.61 × 10 1.08 × 10 3.96 × 10 1.92 × 10 −10 −10 −10 −10 FAP450-2k5 1.61 × 10 1.08 × 10 1.98 × 10 3.84 × 10

The simulated charge-discharge profiles by using the diffusion coefficients in Table A.2 are presented in Figure A.1. Significant differences can be observed after the first two simulated cycles when one of the diffusion coefficients was doubled. The third and fourth cycles when only k2 was doubled exhibited almost identical charge- discharge curves as the ones when k3 was increased to twice the original value and the curves are left-shifted compared to the simulated curve with the experimental diffusion coefficients of FAP450 membrane. Similar phenomenon but right shifting can be found in the cases of doubling k4 and k5. Given that no remarkable left- or right shift was shown in the comparison between the experimental and simulated FAP450 results (Figure 3.17), the experimental diffusion coefficients of FAP450 are believed to be reliable. When the simulation was extended to 100 cycles, a greater difference was revealed and the discharge capacity variation as a function of cycle numbers is shown in Figure A.1. The results again prove the previous conclusion that a balance among the diffusion coefficients of the vanadium ions is more important than the absolute diffusion coefficient values in the capacity retention. Specifically, A. Model Development 256

FAP450 1.6 FAP450-2k2 FAP450-2k3

FAP450-2k4

1.4 FAP450-2k5

1.2 otg (V) Voltage

1.0

0 50000 100000 150000 Time (s)

Figure A.1: A comparison of simulated charge-discharge curves by changing different diffusion coefficients of FAP450 membrane

the single cell with the datasets of FAP450-2k3 and FAP450-2k2 which have closer

2+ + 2+ 3+ k(V )/k(VO2 ) and k(VO )/k(V ) values showed much less capacity decay than those with the parameters of FAP450, FAP450-2k4 and FAP450-2k5. The parameters in Table A.2 were also applied in the thermal simulation models described in Section 3.3.4 and Appendix A.2 to investigate the effect of diffusion coefficients on the concentration (Figure A.3) and temperature (Figure A.4) variations at the first week. It is evident that the systems with membranes of greater differences between

2+ + 2+ 3+ k(V )/k(VO2 ) and k(VO )/k(V ) exhibit faster imbalance in SOCs of the two half-cell electrolytes. Nevertheless, all five simulated temperatures are all in the range of 45 ∼ 55 with slight different variation trends. This is because the heat is mainly generated by Reactions (3.1) and (3.4), and the absolute values of diffusion coefficients determines the temperature but the imbalance of SOCs leads to less A. Model Development 257

4 FAP450 FAP450-2k 3 2 FAP450-2k3 FAP450-2k 2 4 FAP450-2k5 1

icag aaiy(Ah) capacity Discharge 0 0 50 100 Cycle number

Figure A.2: A comparison of simulated discharge capacity by changing different diffusion coefficients of FAP450 membrane

2+ + 2+ + V and/or VO2 formation during charging so that less V and VO2 ions are available to self-discharge. Therefore the heat generation increases in the beginning and then declines after a few days in the cases of FAP450-2k4 and FAP450-2k5. From the above comparison of the simulated results, it can be found that the capacity loss or concentration profiles have a strong dependence on the relative values of the diffusion coefficients of four vanadium ions across the membrane and high accuracy of the diffusion coefficients is required. On the other hand, the peak temperature estimation is less sensitive to the relative values but more dependent on the orders of magnitude of the absolute vanadium diffusion coefficients. Since many parameters can affect the experimental measurement, the adoption of the simulation models to predict the VRFB capacity performance for different membranes should be treated with special caution and more validation methods other than polarization curves should be developed. A. Model Development 258

1.8

1.6

1.4

1.2

0.8 Concentration (M)

0.6 V 2 V 3 0.4 V 4 V 5 0.2 0 1 2 3 4 5 6 7 Time (Day) (a) FAP450

(b) FAP450-2k2 (c) FAP450-2k3

(d) FAP450-2k4 (e) FAP450-2k5

Figure A.3: The simulated concentrations of the species in the tanks with diﬀusion coeﬃcients of (a)FAP450, (b)FAP450-2k2, (c)FAP450-2k3, (d)FAP450-2k4 and

(e)FAP450-2k5 at the ﬁrst week A. Model Development 259

40 C)

o 35

25 Temperature (

15 Cell Tank Air 10 0 1 2 3 4 5 6 7 Time (Day) (a) FAP450

(b) FAP450-2k2 (c) FAP450-2k3

(e) FAP450-2k (d) FAP450-2k4 5

Figure A.4: The simulated temperature of the VRFB systems with diﬀusion coeﬃcients of (a) FAP450, (b) FAP450-2k2, (c)FAP450-2k3, (d)FAP450-2k4 and

(e)FAP450-2k5 at the ﬁrst week