Novel Electrolyte Energy Storage Systems Investigators Allen J. Bard, Professor, Chemistry, The University of Texas at Austin; Brent Bennett, Graduate Researcher, Mechanical Engineering, The University of Texas at Austin; Jinho Chang, Graduate Researcher, Chemistry, The University of Texas at Austin; Netzahualcoyotl Arroyo Curras, Graduate Researcher, Chemistry, The University of Texas at Austin; Robert Villwock, Associate Director, Center for Electrochemistry, The University of Texas at Austin. Abstract

To enhance reliability of the electric grid while simultaneously incorporating renewable power sources into it, there is a pressing need for electrical energy storage, to increase the capability for dispatch and to accommodate the variable nature of those resources. There is, at present, very little energy storage on the grid, due in part to the high capital costs associated with electrochemical energy storage and the lack of flexibility in siting for other technologies.

In this project, we approach enabling widespread deployment of grid-based storage by lowering the cost of such a system. We are reexamining the fundamentals of redox flow battery (RFB) technology and engaging in an effort in which new active redox couples are discovered and optimized, in pursuit of high efficiency and lower capital costs. We seek transformative changes in the construction and composition of the RFB.

Major research accomplishments during the past year include: a) a description of the total reaction pathway for the two-electron Sn(IV)/(II) redox reactions in bromide solutions, including identification of the main species and Sn(III) intermediates; and b) the first evidence of reaction intermediates in the bromide/bromine reaction. Both of these accomplishments advance the fundamental science necessary for pushing tin-bromine and other bromine-based RFB systems toward commercialization. c) The application of very high energy-density RFB systems, based on systems with very high concentrations of suitable redox couples, including a solvent-bromine system and the first investigations of the electrochemistry of Sn(IV)Cl4 liquid, both undiluted and in solution with another solvent and d) fundamental studies of how potentials of redox couples (e.g. of Fe, Co, and Mn) can be shifted to increase voltage output with various novel ligands, and the bench-scale demonstration of a novel alkaline Fe/Co redox flow battery.

Introduction Efficient, cost-effective energy storage is vital to the effort to fully integrate

renewable power sources into the electric utility grid. While compressed-air and pumped-hydro storage plants hold the promise of large-scale economical storage, they both require special sites. To date, redox flow batteries (RFB) have shown promise, but are at present far too expensive to be effectively deployed. The objective of this research it to identify new electrolyte systems and cell designs that allow drastic cost reductions (removing this key barrier) while maintaining the high efficiency and ease of operation that are the hallmarks of RFB systems. Our approach brings together expert researchers with skills in chemistry, material science and characterization, electrochemical

engineering, and mathematical modeling. We will advance new materials and electrochemistry for RFB systems and develop guidelines for scaling up to utility-scale configurations.

Time-of-day pricing on the grid, mandates for renewable power sources, and the accompanying intermittency of those renewable sources are creating demand for electrical energy storage. Energy needs to be stored efficiently, and to accommodate several hours of continuous energy accumulation and release to the grid.

The requirements for large-scale electrical energy storage systems are quite different from existing battery systems. While batteries for portable and transportation applications place a premium on weight and volume, stationary energy storage systems have considerably less stringent requirements. Backup power systems support telecommunications and data centers, but are generally not expected to survive large numbers of charge/discharge cycles. For flow battery systems, one can specify independently the size of the electrochemical reactor (power capacity) and the size of the storage tanks of the free-flowing electrolyte streams (energy capacity). The ability to deliver the active material to the electrode surface by convection ensures that one can bypass mass-transport limitations that curtail the energy density of conventional batteries with solid-phase active materials. Moreover, since the charge and discharge cycles of most RFBs do not involve phase changes, the cycle life and stability is much higher than conventional batteries.

Redox couples for flow-battery applications, represented by reaction in Equation 1,

O + ne R (1)

must satisfy a number of requirements: (i) both forms, O and R, must be highly soluble to minimize the storage volume and mass and to allow high mass transfer rates and current densities during charging and discharging; (ii) the formal potential, Eo’, of one couple must be highly positive, and Eo’ of the other highly negative to maximize the cell voltage and energy density; (iii) the heterogeneous reaction rate for the charging and discharging reactions at the inert electrodes should be rapid so that the electrode reactions occur at their mass transfer controlled rates; (iv) both O and R should be stable during generation and storage, and this stability pertains to reactions with solvent, electrolyte, atmosphere, and electrode materials, and, for metal complexes, stability with respect to ligand loss; (v) the materials should be safe, inexpensive, and abundant; (vi) the couple should not be corrosive and react with cell materials, or the storage vessel.

Background There are many redox flow battery chemistries, including iron-chromium, zinc-

bromine, cerium-zinc, iron-vanadium, vanadium air, and all-vanadium. The all-vanadium RFB uses vanadium (which can exist as +2, +3, +4, or +5 ions) in both electrolytes. This eliminates some problems with crossover from the positive to the negative side, because the negative half-cell uses the V(II)/(III) redox couple, and the positive half-cell uses the V(IV)/(V) redox couple. This system is now commercial, with the technology controlled by Prudent Energy (Bethesda, MD).

Several new redox flow battery and hybrid flow battery technologies are moving toward commercialization, and the field is rapidly expanding, both in market size and in technology development. According to a recent report,[1] rapid growth is expected for the entire advanced battery market, of which RFBs represent a leading edge. The market is expected to double each year over the next five years, and continue growing steadily beyond that. The report predicts total worldwide capacity of advanced batteries for utility-scale energy storage will reach nearly 16 GW by 2022.

For example, EnerVault (Sunnyvale, CA) recently announced a $15 million round of funding, and has plans to install a demonstration unit of an iron-chromium RFB system at a scale of 250 kW/1 MWh in the Central Valley of California. Zinc-bromine hybrid flow batteries are being commercialized by several companies, including Premium Power (North Reading, MA), Primus Power (Hayward, CA), RedFlow Limited (Brisbane, Australia), and ZBB Energy Corporation (Menomonee Falls, WI). Aquion Energy (Pittsburg, PA) is commercializing a hybrid sodium-water battery developed by Jay Whitacre at Carnegie Mellon University. That technology is now in volume production. Other companies include Cellstrom (EU, vanadium), EnStorage (Israel, hydrogen-bromine), and Deeya Energy (Fremont, CA, iron-chromium).

The major thrusts of our research have been directed at discovering new redox couples for flow batteries that are less expensive than current systems (e.g. the vanadium RFB) and work under different conditions, for example in extremely alkaline solutions that show lower corrosion effects. As discussed below we have been successful in working out guidelines for that work, and have demonstrated a novel Fe/Co alkaline system. We have also been interested in couples that show multielectron transfers (thereby increasing the energy efficiency) while still maintaining good electrochemical properties, and we present studies of the Sn(IV)/(II)-bromine system, where the reaction mechanism has been elucidated with the detection of intermediates, while investigations of electrocatalysis with these couples continue. Finally, we have begun work on systems that could be described as “electrochemical fuels,” by using liquids that themselves show redox properties. These are capable of much higher energy density storage than conventional RFBs, and we demonstrate the principles with the solvent-bromine system.

Results Multielectron Redox Couples for RFB Applications — Tin/Bromine System

Most of the research on redox couples for flow batteries involves one electron transferred (i.e., n=1 in Equation 1). However, by employing multielectron couples the storage density could be multiplied. A problem with such couples is that the mechanism of the half-reaction is more complex, and this may result in kinetic complications in the charge and discharge cycles. We are focusing first on tin-based systems for redox flow batteries. Sn(IV) bromide in an acidic bromide medium is a promising candidate because both redox couples, Sn(IV)/Sn(II) and Br2/Br- are dissolved in same solution, eliminating concerns of cross contamination. This work was encouraged by the design of a new tin-bromine RFB system with Sn(IV)/Sn(II) as an anode and Br2/Br- as a cathode.[2]

The fundamental understanding of the kinetics and mechanism of such electrode reactions (where n=2) is very poor compared to the rather advanced state of n=1

reactions. With a view toward discovering ways to improve poor electrochemical reversibility, progress is being made toward a complete understanding of the mechanism of the Sn(IV)/Sn(II) redox reaction. A key question has been whether the preferred reaction pathway is (a) two electrons simultaneously transferred interfacially through tunneling, or (b) two stepwise one-electron transfers occurring via a Sn(III) intermediate. We have sought to find and quantify a Sn(III) intermediate through electrochemical analysis methods such as fast-scan cyclic voltammetry and scanning electrochemical microscopy (SECM), and use these techniques to elucidate the mechanism and rate constants for multielectron reaction pathways.

We report the detection of a Sn(III) intermediate, Sn(III)Br4-, during Sn(II)Br4

2-

/Sn(IV)Br62- oxidation, building upon our previous work in which Sn(III)Br6

3- was detected during Sn(IV)Br6

2-/Sn(II)Br42- reduction. Using these new data, we propose a

mechanism for Sn(II)/Sn(IV) oxidation.

Fast scan cyclic voltammetry (FSCV) and scanning electrochemical microscopy (SECM) were used to detect Sn(III). FSCV has been successfully used with other half-reactions to observe two single-electron-transfer steps in overall two-electron-transfer reactions, allowing kinetic discrimination of the two electrochemical steps.[3] FSCV is also a powerful tool for detecting short-lived intermediates because the intermediates can be re-oxidized/reduced before undergoing homogeneous reactions to form more stabilized structures. This detection is possible when the characteristic time of cyclic voltammetry is shorter than homogenous reaction time of the intermediates.[4] SECM methods have been developed and used for the collection of intermediates.[5] In SECM, a 100% collection efficiency (CE) can be achieved in pure diffusion-controlled redox mediators at a reasonably close distance between a tip and substrate; this CE is much higher than that obtained from a rotating ring disk electrode (normally, 30–35%). Therefore, SECM can collect even traces of any generated intermediates.[6]

A one-electron-transfer step in the Sn(II)/Sn(III) redox reaction was observed by FSCV, using scan rates of ! ! 10 V/s. From tip generation/ substrate collection (TG/SC) mode in SECM, no further fast decomposition reactions of Sn(II)Br4

2- and Sn(IV)Br62-

were confirmed. In SECM studies, the intermediate Sn(III)Br4-, generated from the tip,

was collected by the reducing it to Sn(II)Br42- on the substrate at d~1 "m while holding

Esub = 0 V, where Sn(IV)Br62- was not reduced. The rate of the disproportionation and

bromide addition reactions of Sn(III)Br4- was estimated to be 1.0!109 M-1s-1 and

3.0!103 s-1, respectively, by a digital simulation based on SECM experiments. From the experimental evidence of the Sn(III) species in the Sn(IV)/Sn(II) redox reaction, different reaction pathways between Sn(IV)/Sn(II) reduction and oxidation, via the Sn(III) intermediates, were proposed, and the irreversibility of Sn(IV)/Sn(II) redox reaction was explained by the difference of the reduction potentials between Sn(IV)Br6

2-/Sn(III)Br63-

and Sn(III)Br4-/Sn(II)Br4

2-.

In the study of Sn(IV)/Sn(II) reduction previously reported, the main species of Sn(IV) and Sn(II) in 2 M HBr + 4 M NaBr solution were reported to be Sn(IV)Br6

2- and Sn(II)Br4

2-, respectively, based on Raman studies,[7,8] and the stability constants of

Sn(II) with Br-.[9] Based on our cyclic voltammetry and SECM experiments, we propose the following Sn(II)/Sn(IV) oxidation reaction pathway:

Sn(III)Br4- + e- Sn(II)Br4

2- (2) 2 Sn(III)Br4

- Sn(II)Br42- + Sn(IV)Br6

2- (3)

Sn(III)Br4- + Br- Sn(III)Br5

2- (4) Sn(IV)Br5

- + e- Sn(III)Br52- (5)

2 Sn(III)Br52- Sn(II)Br4

2- + Sn(IV)Br62- (6)

Sn(IV)Br5- + Br- Sn(IV)Br6

2- (7)

The new anodic peak with one electron transfer and a small reversal peak in the CVs

with fast scans and the collected current in SECM at Esub = 0 V are attributed to Sn(III)Br4

-/Sn(II)Br42- redox reaction (2). The Sn(III)Br4

- can either disproportionate to Sn(II)Br4

2- and Sn(IV)Br62- (3) or become Sn(III)Br5

2- with the addition of one bromide (4). Sn(III)Br5

2- can be oxidized to Sn(IV)Br5- with one electron transfer, which

was not observed in our fast scan CVs, (5) or disproportionate to Sn(II)Br42- and

Sn(IV)Br62- (6). Finally, Sn(IV)Br5

- goes to Sn(IV)Br62- with the addition of one

bromide (7).

Figure 1: A schematic description of the total reaction pathway in the Sn(IV)/Sn(II) redox reactions; black circles show main species in 2 M HBr / 4 M NaBr solution and blue circles show the Sn(III) intermediates detected by FSCV and SECM.

Figure 1 shows a schematic description of the reaction pathway for both Sn(IV)/Sn(II) reduction (red) and oxidation (green). The proposed mechanism explains

(II) (III) (IV)

4Br-

5Br-

6Br-

Br- c

oord

inat

ion

Sn oxidation state

Oxidation Reduction

the large irreversibility in Sn(IV)/Sn(II) redox reaction. In the reduction, Sn(IV)Br62- is

reduced to Sn(II)Br42- through the Sn(III) intermediates, Sn(III)Br6

3- and Sn(III)Br52-. The

reduced Sn(II)Br42- cannot be oxidized via the same route as that in the reduction because

Sn(II)Br42- is thermodynamically the most stable form among the Sn(II)Brx

2-x species in 2 M HBr + 4 M NaBr. It must first be oxidized to Sn(III)Br4

-. Therefore, the irreversibility is based on the difference in the reduction potentials between Sn(IV)Br6

2-

/Sn(III)Br63- and Sn(III)Br4

-/Sn(II)Br42-.

Study of the Bromide/Bromine Reaction in Nonaqueous Solvents Nonaqueous solvents offer the potential to create redox flow batteries with higher

energy densities, new cell designs, and less expensive materials, which together could lead to the lower capital costs and speedier commercialization. Accordingly, interest in nonaqueous solvents for redox flow batteries has grown in recent years.[10] With these research trends in mind and a desire to reach well beyond the limitations of current technologies, we have begun investigating the use of the bromide/bromine redox couple in nonaqueous solvents. This redox couple has the advantages of a large positive redox potential (~1 V vs. NHE), a two-electron-transfer reaction, and high solubility in many solvents.

Figure 2: Cyclic voltammogram of the bromide/bromine redox couple in a nonaqueous solvent. The first anodic wave is bromide oxidation to tribromide and the second wave is tribromide oxidation to bromine.

In aqueous systems, the bromide/bromine redox couple is an efficient and reversible

reaction. Hence, it has been the subject of much study for redox flow batteries and has found a commercial application in the zinc/bromine flow battery.[11] However, in most nonaqueous solvents, this reaction is a complex two-step process through a stable intermediate tribromide ion, and is highly irreversible (see Figure 2). Although this reaction has been investigated for over 50 years, no consensus has been reached on the

reaction mechanism. We are studying the mechanism in detail and seeking to uncover the individual reaction steps in order to find ways to make reaction more reversible and efficient.

Our nonaqueous solvent was chosen to be relatively nonvolatile and perfectly miscible with bromine. Quaternary ammonium salts were chosen for the bromide and supporting electrolyte due to their stability and solubility. Our initial investigation of the oxidation of bromide to tribromide and the reduction of tribromide to bromide has revealed two new insights. First, fast-scan cyclic voltammetry (FSCV) and scanning electrochemical microscopy (SECM) have revealed that an adsorbed bromine radical (Br2

-) is an intermediate product between bromide and tribromide. This finding suggests that the reaction mechanism follows an electrochemical-chemical (ECC) pattern. That is, the first oxidation step is followed by the addition of a bromide ion to form the Br2

- radical, which then picks up a bromine atom to form tribromide. Second, using SECM we have found evidence of a different intermediate during the reduction of tribromide to bromide, suggesting that the reduction follows an electrochemical-electrochemical (ECEC) mechanism. No previous studies of the bromide/bromine reaction in nonaqueous solvents have provided evidence of these intermediates or of the elementary reaction steps.

Alkaline Fe/Co Redox Flow Battery Progress has been made on the development of an alkaline redox flow battery. Our

motivation to develop such a device came from the idea that an alkaline electrolyte (such as aqueous NaOH) can achieve conductivities as high as 414 mS/cm,[12] a value that is comparable to the conductivity of 1 M H2SO4. Using an alkaline electrolyte offers the following advantages: a) the conductivity of the electrolyte can be adjusted to minimize undesired shunts within cell stacks; b) alkaline solutions are less corrosive with typical materials of construction than acids; c) ligand-modified redox couples can be used that are chemically stable and reversible, offering good cyclability and high efficiency during the electron transfer reactions.

Ligands can be used to improve solubility and to tune the formal equilibrium potential over wide potential ranges for electrolyte-phase redox couples. Increased solubility allows for higher electrolyte concentrations, increases the energy storage per unit volume, and reduces variability in the potential profile. By suitable choice of ligand the formal potential of a transition-metal ion couple can be shifted in the desired direction. Moreover, such complexes may show improved characteristics with respect to stability in comparison to the uncomplexed species.

The chemical principles related to the formation and properties of metal complexes are well developed and many potentially useful ligands have been reported and compiled.[13] This new work leverages our previous investigations of the electrochemical behavior of promising transition-metal ion complexes with a variety of ligands, including measurements of the formal potential, the electrode kinetics, reversability and the stability of reactants and products of the complexes in solution.[14] So far, we have identified fifteen chemically stable, electrochemically reversible/quasi-reversible couples that are potential candidates for an alkaline flow battery. These

chemistries have been extensively characterized and are compatible with carbon-felt electrodes. An example of two of such chemistries is shown in Figure 3.

Figure 3. Cyclic voltammograms of candidate redox couples for an alkaline redox flow battery. The concentrations of both couples are 10 mM in Mn+ ions.

Also, a first test has been performed in a flow cell using concentrations of 0.6 M of each redox couple in solution. Our preliminary measurements gave an output voltage of 0.95 V and output currents in the order of 150 mA (carbon felt electrodes, 4.7 cm x 4.7 cm x 0.5 mm, 11.05 cm3, " = 0.1 g/cm3).

Progress We expect that the fundamental leaps in chemistry, materials, and engineering

knowledge provided by this work will unlock the ability to build and operate commercial grid-scale storage for intermittent renewable energy sources. This type of large-scale storage is a natural complement to intermittent renewable generation methods, and to the degree we can enable those methods, we can directly reduce greenhouse gas emissions.

The research proposed in this work will lead to fundamental understanding of the controlling processes in flow batteries, enabling modifications in their chemistry so that they can deliver efficient energy storage. The chemistry developed will inform related efforts and enable advances in the development of these devices and materials. Specifically, we are developing new understanding of how to tailor redox couples, and quantification of competing factors that can tune the redox potentials. Our research considers the eventual large-scale application by working with both aqueous and nonaqueous systems, earth-abundant materials, and optimizing for manufacturability and cost. The insights developed will inform characterization and design efforts, and enable significant advances, including guidelines for scale-up.

This work could offer guidelines for enhancing the electrochemical reversibility of the stannous/stannic redox reaction in the newly proposed Sn/Br RFB. New RFB systems being discovered and developed could provide new technology options for energy storage, enabling wider deployment of renewable energy sources on the electrical utility grid.

Future Plans We continue to seek the following: (1) synthesize new ligands and complexes as

electrolytes with tailored thermodynamic potentials, (2) screen these candidate materials under a variety of conditions on carbon electrodes, and (3) characterize the most promising materials in larger electrochemical cells. We will continue to use a variety of electrochemical tools to elucidate the mechanism of these reactions for RFBs, understand the rate-determining steps and discover, by scanning electrochemical microscopy rapid screening methods, electrocatalysts that promote this reaction. Armed with this new information, flow cells utilizing this reaction will be tested.

While good redox couples have a number of requirements in terms of stability, solubility, cost, and dielectric constant, which make inorganic liquids like SnCl4 challenging, there are a number of organic liquids that are promising. These will be studied using a variety of electrochemical tools and useful couples will be tested in flow cell configurations. This is a promising new area of inquiry, with far-reaching implications for design of flow battery systems. We will identify promising organic-liquid redox couples, evaluate their performance as flow battery energy carriers using electroanalytical tools, downselect the best candidates for evaluation in lab-scale flow battery systems, and scale up to obtain relevant process data.

We intend to fully characterize the reaction steps between bromide and tribromide. We will then use these same techniques to investigate the oxidation of tribromide to bromine, which is the second step of the bromide/bromine reaction. Once we gain a full understanding of the reaction mechanism, we intend to create a prototype flow cell using tribromide/bromine at the positive electrode and nitrobenzene or another redox active liquid at the negative electrode.

Publications and Presentations 1. Chang, J., Leonard, K. C., Cho, S. K., & Bard, A. J. Examining Ultramicroelectrodes for Scanning

Electrochemical Microscopy by White Light Vertical Scanning Interferometry and Filling Recessed Tips by Electrodeposition of Gold Anal. Chem. 84, 5159–5163 (2012).

2. Chang, J. Detection of Intermediate Sn(III) on Gold electrode through Fast Scan Cyclic Voltammetry and Scanning Electrochemical Microscopy, 4th Annual Center for Electrochemistry Workshop, Poster Session, February 11, 2012, Austin, TX.

3. Chang, J. Detection of Intermediate Sn(III) in Sn(II)/Sn(IV) Oxidation on Gold electrode Through Fast Scan Cyclic Voltammetry (FSCV) and Scanning Electrochemical Microscopy (SECM), 5th Annual Center for Electrochemistry Workshop, Poster Session, February 9, 2013, Austin, TX.

4. Bennett, B. Electrochemical Study of the Bromide/Bromine Reaction in Nitrobenzene, 5th Annual Center for Electrochemistry Workshop, Poster Session, February 9, 2013, Austin, TX.

Manuscripts in preparation

5. Chang, J. & Bard, A. J. Detection of Sn(III) intermediates in Sn(IV)/Sn(II) Redox reaction on Gold electrode through Fast Scan Cyclic Voltammetry and Scanning Electrochemical Microscopy. I. Sn(IV)/Sn(II) Reduction via Sn(III) and II. Sn(II)/Sn(IV) Oxidation via Sn(III), in preparation.

6. Chang, J. & Bard, A. J. In-situ Electrochemical Quantification of Pb UPD on Polycrystalline Au by Surface Interrogation mode in Scanning Electrochemical Microscopy (SECM), in preparation.

7. Chang, J., Bennett, B. & Bard, A. J. Br-/Br3- Oxidation in Nitrobenzene: Evidence of Bromine anion radical, Br2-# through Fast Scan Cyclic Voltammetry (FSCV) and Scanning Electrochemical Microscopy (SECM), in preparation.

References

1. Advanced Batteries for Utility-Scale Energy Storage Applications, Navigant Research, July 2012, http://www.navigantresearch.com/research/advanced-batteries-for-utility-scale-energy-storage-applications (accessed April 29, 2013).

2. (a) Skyllas-Kazacos, M., Chakrabarti, M. H., Hajimolana, S. A., Mjalli, F. S., & Saleem, M. Progress in Flow Battery Research and Development J. Electrochem. Soc.158, R55-R79 (2011). (b) Skyllas-Kazacos, M. Novel Vanadium Chloride/Polyhalide Redox Flow Battery J. Power Sources 124, 299-302 (2003).

3. (a) Pierce, D. T. & Geiger, W. E. Electrochemical Kinetic Discrimination of the Single-Electron-Transfer Events of a Two-Electron-Transfer Reaction: Cyclic Voltammetry of the Reduction of the bis(Hexamethylbenzene)ruthenium Dication J. Am. Chem. Soc..114, 6063-6073 (1992). (b) Stoll, M. E., Belanzoni, P., Calhorda, M. J., Drew, M. G. B., Félix, V., Geiger, W. E., Gamelas, C. A., Gonçalves, I. S., Romão, C. C., & Veiros, L. F. Stepwise Hapticity Changes in Sequential One-Electron Redox Reactions of Indenyl-Molybdenum Complexes:$ Combined Electrochemical, ESR, X-ray, and Theoretical Studies J. Am. Chem. Soc. 123, 10595-10606 (2001). (c) Kaim, W., Reinhardt, R., Greulich, S., & Fiedler, J. Resolving the Two-Electron Process for the Couple [(C5Me5)M(N^N)Cl]+/[(C5Me5)M(N^N)] (M=Rh, Ir) into Two One-Electron Steps Using the 2,2’-Azobis(pyridine) N^N Ligand, Fast Scan Cyclovoltammetry, and Spectroelectrochemistry:$ Detection of Radicals instead of MII Intermediates Organometallics 22, 2240-2244 (2003).

4. (a) Yang, H. & Bard, A. J. The Application of Rapid Scan Cyclic Voltammetry and Digital Simulation to the Study of the Mechanism of Diphenylamine Oxidation, Radical Cation Dimerization, and Polymerization in Acetonitrile J. Electroanal. Chem. Interfacial Electrochem. 306, 87-109 (1991). (b) Yang, H. & Bard, A. J. The Application of Fast Scan Cyclic Voltammetry. Mechanistic Study of the Initial Stage of Electropolymerization of Aniline in Aqueous Solutions J. Electroanal. Chem. 339, 423-449 (1992). (c) Yang, H., Wipf, D. O., & Bard, A. J. Application of Rapid Scan Cyclic Voltammetry to a Study of the Oxidation and Dimerization of N,N-dimethylaniline in Acetonitrile. J. Electroanal. Chem. 331, 913-924 (1992). (d) Baur, J. E., Wang, S., & Brandt, M. C. Fast-Scan Voltammetry of Cyclic Nitroxide Free Radicals Anal. Chem. 68, 3815-21 (1996). (e) Andrieux, C. P., Hapiot, P., & Saveant, & J. M. Fast Chemical Steps Coupled with Outer-sphere Electron Transfers. Application of Fast Scan Voltammetry at Ultramicroelectrodes to the Cleavage of Aromatic Halide Anion Radicals in the Microsecond Lifetime Range J. Phys. Chem. 92, 5987-5992 (1988). (f) Andrieux, C. P., Audebert, P., Hapiot, P., & Saveant, J. M. Observation of the Cation Radicals of Pyrrole and of Some Substituted Pyrroles in Fast Scan Cyclic Voltammetry. Standard Potentials and Lifetimes. J. Am. Chem. Soc. 112, 2439-2440 (1990).

5. (a) Zhou, F. & Bard, A. J. Detection of the Electrohydrodimerization Intermediate Acrylonitrile Radical Anion by Scanning Electrochemical Microscopy J. Am. Chem. Soc. 116, 393-394 (1994). (b) Bi, S., Liu, B., Fan, F. R., & Bard, A. J. Electrochemical Studies of Guanosine in DMF and Detection of Its Radical Cation in a Scanning Electrochemical Microscopy Nanogap Experiment J. Am. Chem. Soc. 127, 3690-3691 (2005).

6. Unwin, P. R. Kinetics of Homogeneous Reactions Coupled to Heterogeneous Electron Transfer. In Scanning Electrochemical Microscopy; Mirkin, M. V. & Bard, A. J., Eds. Marcel Dekker, Inc.: New York, 2001; pp. 244–254.

7. Woodward, L. A. & Anderson, L. E. Raman Spectrum and Structure of the Hexabromostannate

Ion in Aqueous Solution J. Chem. Soc. 1284-1286 (1957). 8. Taylor, M. J. & Coddington, J. M. The Constitution of Aqueous Tin(IV) Chloride and Bromide

Solutions and Solvent Extracts Studied by 119Sn NMR and Vibrational Spectroscopy Polyhedron 11, 1531-1544 (1992).

9. Högfeldt. E. Stability constants of metal-ion complexes, Part A: Inorganic ligands, 1st ed.; IUPAC Chemical Data Series, No. 21; Pergamon Press: 1982; pp 264.

10. Shin, S.-H., Yun, S.-H., & Moon, S.-H. A Review of Current Developments in Non-aqueous Redox Flow Batteries: Characterization of Their Membranes for Design Perspective RSC Adv. (2013). DOI: 10.1039/C3RA00115F

11. ZBB Energy Corporation, http://www.zbbenergy.com/ (accessed April 3, 2013). 12. Foxboro, Technical Report: http://myweb.wit.edu/sandinic/Research/conductivity%20v %

20concentration.pdf (accessed April 3, 2013). 13. See, e.g., some excellent and comprehensive reviews in (a) Comprehensive Coordination

Chemistry: the Synthesis, Reactions, Properties and Application Wilkinson, G., Gillard, R. D., & McCleverty, J. A., Eds. Vols. 1–7, Oxford, England; New York: Pergmon Press, 1987; (b) Wilkins, R. G. Kinetics and Mechanism of Reactions of Transition Metal Complexes 2nd ed., VCH, New York, 1991. (c) Comprehensive Coordination Chemistry II: from Biology to Nanotechnology McCleverty, J. A. & Meyer, T. J., Eds. Vols. 1–10, Amsterdam; Boston; Elsevier Pergamon, 2004.

14. Arroyo-Curras, N. Use of Scanning Electrochemical Microscopy (SECM) in the Evaluation of Redox Couples for Flow Battery Applications, 5th Annual Center for Electrochemistry Workshop, Poster Session, February 9, 2013, Austin, TX.

Contacts Allen J. Bard: ajbard@mail.utexas.edu Brent Bennett: brent-bennett@mail.utexas.edu Jinho Chang: jinho_chang@utexas.edu Netzahualcoyotl Arroyo Curras: netz.arroyo@utexas.edu Robert Villwock: bvillwock@cm.utexas.edu

Annual Report for University of Tennessee GCEP Program: April 2012

Tom Zawodzinski, UTK PI

Alex Papandrew, Co-PI

Investigators Tom Zawodzinski, Professor, University of Tennessee-Knoxville, Chemical and Biomolecular Engineering Department Alex Papandrew, Research Assistant Professor, University of Tennessee-Knoxville, Chemical and Biomolecular Engineering Department Zhijiang Tang, Graduate Student, University of Tennessee-Knoxville, Chemical and Biomolecular Engineering Department Jamie Lawton, Post-doctoral Fellow, University of Tennessee-Knoxville, Chemical and Biomolecular Engineering Department Abstract This project has as its primary focus the development of understanding of relevant properties of membranes and separators and other materials used in redox flow batteries (RFBs), with specific emphasis on the Vanadium Redox Battery (VRB). This will lead to an understanding of development needs and directions for materials in flow batteries. During this year, we have applied our previously developed suite of experimental methods to measure uptake of species into membranes and their transport. To frame this issue, a membrane used in a VRB is exposed one each side to a concentrated aqueous acid (typically sulfuric) solution containing vanadium species in at least 2 of 4 oxidation states. The concentrations are such that several things happen in the membrane:

1. Acid is imbibed into the membrane. 2. Water is ejected from the membrane 3. Vanadium species enter the membrane from each side.

Also, substantial water can move across membranes in an operating cell. We have compiled a substantial amount of data, for several membrane types, on:

i. Acid and water uptake into membranes from bathing solutions of sulfuric acid in water.

ii. Vanadium uptake into the membrane. iii. Vanadium and other species transport across a membrane. iv. Conductivity of membranes exposed to solutions of sulfuric acid in water,

including the effects of adding vanadium ions at low concentration and in practical solutions.

In addition, we are measuring electrode kinetics in situ in the operating cell. We discovered substantial differences in the relative electron transfer rates for the positive and negative electrode reactions. Primary findings from this year include: (i) vanadium in different redox states behaves similarly with respect to uptake; (ii) under conditions in which vanadium is taken into a

membrane, it has a major deleterious effect on conductivity; vanadium selectively partitions into the membrane under conditions in which Donnan exclusion pertains; however, for practical situations at high acid concentration, the partitioning favors protons and the effect of vanadium is somewhat mitigated; (iii) small differences are observed in mobility of different vanadium oxidation states; (iv) while V(2/3) redox kinetics are slower than those of (V(4/5), the kinetics of both are sufficiently facile that kinetic effects are minimal in a cell with sufficient surface area of porous carbon electrodes. Finally, some preliminary comparisons between different membrane types and PFSA membranes of different EW have been made. Introduction In VRBs, the membrane separator is a key contributor to the overall cell resistance and plays a critical role in preventing chemical ‘shorting’ of the cell due to intermixing of the various redox active species present. Understanding and controlling these effects is critical to realizing high performance in VRBs. For example, at the onset of this project, results from the literature suggested that cell areal specific resistance on the order of 6 ohm-cm2 was typical. By contrast, present day PEM fuel cells exhibit a resistance of <0.05 ohm-cm2. This 100-fold difference is at least partly attributable to the membrane. In our hands, improving contact resistance lowered the cell ASR by roughly an order of magnitude. The remaining difference is due to the membranes. Lowering ASR through the use of alternative membranes, based partly on information reported below, has led to a present-day minimum ASR of 0.13 ohm-cm2 for a VRB, with a concomitant improvement in cell performance as measured by peak power. This project has as its primary focus the development of understanding of relevant properties of membranes and separators and other materials used in redox flow batteries (RFBs), with specific emphasis on the Vanadium Redox Battery (VRB). Our emphasis was initially oriented toward two goals: (1) supporting the activities of Prof. Meyers at UT-Austin by providing experimental data on transport of species in the battery materials as a function of composition, suitable for use in models; and (2) using this same data to draw conclusions concerning ‘best’ material approaches to improve on several aspects of VRB performance. Regarding the first of these targets, Professor Meyers has since moved on and, though we discussed direct collaboration between UTK and remaining graduate students, the timing was poor (UT-A graduate student was essentially finished and could not assimilate our information into his work). We are exploring adjusting our specific chemical focus to more effectively collaborate with UT-A. Our vanadium flow battery work has proceeded. Under the GCEP program, we have developed (and continue to develop) a suite of experimental methods to reveal the somewhat complex physical chemistry of membranes exposed to concentrated electrolyte solutions that are typically used in these devices. We have also carried out some studies of electrode kinetic of materials used in cells. This work has formed the underpinnings for some advances in performance of VRBs achieved in our lab and at Oak Ridge National Laboratory.

Background Redox flow battery development efforts, and particularly those associated with vanadium batteries, have proliferated world-wide. Most of these efforts are focused on developing commercial products. Projects like ours, which seek to provide the basic understanding of processes occurring in the cells leading to improvement of components, cost reduction and performance enhancement, are less common. Much of what has been done on vanadium redox batteries prior to the last year has lacked rigor. An additional development in our labs at UTK, predating and not covered under the GCEP activity, has focused on improving power density of the VRB via cell design, improved membranes and control of mass transport in electrode materials, leading to spectacular gains across the entire range of operating current density. We can currently operate a VRB cell at a peak power density exceeding 2.5 W/cm2. As noted above, roughly half of this improvement can be related to improvements in membranes. Our cell design is available for our use in testing materials such as membranes and electrodes, the primary focus of our GECP activity, and electrodes. In the next phase of this project, we expect to deploy and, if needed, adapt our hardware and component approaches to support the demonstration of alternative chemistry under development at UT-A for redox flow batteries. Results In this first phase of the activity, we developed a number of experimental methods to measure uptake of species into membranes and their transport. To frame this issue, a membrane used in a VRB is exposed one each side to a concentrated aqueous acid (typically sulfuric) solution containing vanadium species in at least 2 of 4 oxidation states. The concentrations are such that several things happen in the membrane:

1. Acid is imbibed into the membrane. 2. Water is ejected from the membrane 3. Vanadium species enter the membrane from each side.

Also, substantial water can move across membranes in an operating cell. To describe uptake and transport in such a system, we need to develop quantitative methods to determine the water, acid and vanadium content in the membrane as well as the oxidation state(s) of the vanadium, all as a function of concentration of the solution to which the membrane is exposed. We then need to determine transport rates of species under the same conditions. This is truly a daunting task. The transport of the individual species is coupled to that of other species. Accordingly, we have broken the overall large measurement problem down into a series of simpler measurements and started to develop methods for obtaining accurate subsets of data. These measurements are:

i. Determination of acid and water uptake into the membrane from bathing solutions of sulfuric acid in water. We have largely completed this effort for

standard membranes and are now using it to compare different materials. Recent highlights presented below.

ii. Determination of vanadium uptake into the membrane, one oxidation state at a time, from solutions. We have expanded our studies beyond vanadium (IV) to compare the relative behavior of different oxidation states.

iii. Determination of vanadium and other species transport across a membrane, one species at a time. This is critical because of the effect of cross-over and water transfer on capacity fade and cell imbalancing in the device. We have expanded our studies beyond vanadium (IV) to compare the relative behavior of different oxidation states.

iv. Determination of conductivity of membranes exposed to solutions of sulfuric acid in water, including the effects of adding vanadium ions to the sulfuric acid solutions (and thus into the membranes).

For our initial work, the membrane employed was typically Nafion or another PFSA. In what follows, we expand upon our studies reported last year to include further analysis of acid uptake data. However, the primary focal point of the work is a more detailed consideration of the effects of vanadium uptake into the membranes on conductivity and transport, with a comparison of the effects of different oxidation states. A. Acid and Water Uptake The acid and water uptake by the membrane are the background for all other measurements of membrane properties that we report. Acid and water uptake into the membrane were determined in the following manner. A membrane sample was immersed in a solution of sulfuric acid in water of a given concentration. The membrane was removed after equilibration, surface-blotted, weighed and then immersed in pure water. This serves to extract acid into the water. After the membrane was removed, the solution is titrated to determine the acid content. ICP-AES analysis was used to determine the concentration of acid and, if present, vanadium in the membrane. The membrane is dried and weighed. A simple mass balance then reveals water and acid contents in the equilibrated membrane. In Figure 1, we show the results of this experiment for Nafion 117. In Figure 2, we show the water uptake by the membrane as a function of water activity, in comparison to water uptake from the vapor phase. This is plotted in two different ways in the figure. First, the absolute water content versus number of membrane-derived sulfonates is higher for membranes exposed to solutions. However, renormalizing to include the number of acid molecules results in data that are largely similar to previously reported uptake curves. In Figure 3, we show that substantiall sulfuric acid is indeed taken into the membrane. Finally, in Figure 4, we show the results of gas pycnometry measurements of the density of membranes exposed to sulfuric acid solutions. The increasing density is related to the loss of water from the membrane, a correlation shown previously by our group. In Figure 5, we estimate the effect on total acid content in the membrane by calculating the amount of dissociated protons derived from the sulfuric acid. The simple linear relation indicates that all sulfuric acid is expected to be mono-dissociated based on its acid equilibrium constant. The overall picture that is emerging from these studies is one in which high concentrations of acid in the bathing solution leads to lowered water content and significant excess proton content.

Figure 1. Water and sulfuric acid uptake by Nafion 117 in aqueous sulfuric acid solutions within concentration range from 0 to 17.4mol/kg. The presence of acid can lead to lowered water content in Nafion, and allow sulfuric acid to overcome the Donnan potential and enter the membrane.

Figure 2. Water uptake in Nafion 117 vs. water activity in equilibrium between membrane and sulfuric acid solution. The water content dependence on water activity in sulfuric acid is very similar to that in water vapor.

Figure 3. Sulfuric acid concentration in the membrane versus sulfuric acid concentration in bathing solution. The ratio of membrane sulfuric acid to environmental sulfuric acid is roughly 1:2.5.

Figure 4. Nafion 117 density variation with sulfuric acid concentration in bathing solution. The increasing density with acid solution concentration indicates that the membrane de-swells in acid solution.

Figure 5. Excess proton in membrane is generated by the first dissociation of sulfuric acid in the membrane. Membrane proton content is equal to the total content of sulfonic acid group plus sulfuric acid in membrane.

B. Vanadium uptake During this year, we have studied much more extensively the uptake and effects of vanadium uptake into the membrane. Generally, we can consider the observed a vanadium uptake to be a resultant of the partitioning of vanadium relative to protons into the membrane. The uptake of polyvalent cations into membranes is known to have a strong effect on the conductivity of the membrane. Thus, it is very important to understand how much vanadium is present in the membrane. This is illustrated in Figure 6 for low total ionic strength situations.

Figure 6. V3+ and water contents in Nafion 117 after being equilibrating in solutions of 0.1 mol∙dm-3 total Cl- background, with comparison to VO2+ and water contents in Nafion 117 equilibrated in a similar condition. V3+ presence in Nafion cannot significantly change membrane’s water content. Sulfonate in Nafion has slightly higher affinity to V3+ than VO2+. To take this a bit further, we show in Figure 7 the uptake of water and acid from solutions of vanadium in 5M acid. The vanadium concentration and valence state have minimal effects on the uptake of acid and water by the membrane. Figure 8 shows the vanadium ion concentration in the membrane under these conditions. Unlike the case for low concentrations, vanadium ions (all valence states qualitatively similar) enter the membrane as a minority component even up to high concentration. Thus, the high acid concentration flouts the preferential partitioning of vanadium into the membrane.

Figure 7. Water and sulfuric acid contents in Nafion after equilibration in solutions of 0 to 2 mol∙dm-3 vanadium ions (V3+, VO2+ and VO2

+) and 5 mol∙dm-3 total sulfate. Water and sulfuric acid contents in Nafion are not influenced by the valence state of vanadium in equilibrium.

Figure 8. Contents of vanadium ions of different valence state, in Nafion 117 after equilibration in solutions of 0 to 2 mol∙dm-3 vanadium ions (V3+, VO2+ and VO2

+) and 5 mol∙dm-3 total sulfate. V3+, VO2+ have higher presence in Nafion than VO2

+. The content of each vanadium ion is respectively nearly proportional to vanadium ion concentration in equilibrating environment.

C. Vanadium Transport To measure vanadium transport, we have developed and implemented several methods. First, we are acquiring a standard diffusion cell, with two compartments separated by a membrane, to allow transport. In a second configuration, we are using a standard test cell to determine vanadium cross-over with and without current. In this case, our detector is an EPR instrument. EPR active species are obtained from V(II) and V(IV). We can access other oxidation states (V(III) and V(V)) by quantitatively converting these species in the effluent into V(II) and V(IV). We have constructed a flow-through cell to feeding solution on the receiving side of the membrane into the instrument. The EPR is a highly sensitive detector and allows us to detect relatively minute rates of transport. We have also developed a flow-through visible spectroscopy apparatus for similar studies. Several significant issues arise related to obtaining meaningful transport data in any experiment. First, the concentrated solutions typically used in the VRB yield an extremely complex transport problem. In principle, we would need to determine 21 independent transport coefficients describing pure species transport plus all needed pair-wise coupling coefficients. Second, it is virtually impossible to isolate a single transported species. Indeed, substantial water transport is observed in any diffusion experiment. To mitigate these factors, we are taking several steps. First, we are ruthlessly simplifying our experiments. We are looking at transport of one vanadium species at a time. We also are developing a systematic framework, based on non-equilibrium thermodynamics, to assess these simplifications. Some initial work was carried out, in the manner of the literature, to suppress the water transport across the membrane by maintaining a constant ionic strength on each side of the membrane through the additional of MgSO4 to the ‘receiving’ side of the membrane. However, this adds further complexity because Mg2+ can likely partition into the membrane. On the other hand, even the simplest experiment to probe transport of a single species leads is more complicated than revealed at first glance. If we study transport of VO2+ through the membrane by preparing a membrane separated cell with acid on one side and acid plus VO2+ on the other, we must choose between a constant ionic strength experiment (maintaining constant sulfate on the two sides) which will suppress water transport versus a constant acid concentration experiment that suppresses acid transport. We compare these two cases in an EPR-detected diffusion experiment, the results of which are shown in Figures 9 and 10. In Figure 9, we show the raw EPR signal intensity data for a transport experiment in which acid content is constant and the vanadium to acid ratio is kept constant. At a glance, a series of curves with different rates of increase of signal intensity is seen. However, closer inspection reveals that the maximum rate is obtained for the most dilute vanadium solution, with systematically decreasing transport rates with increasing vanadium concentration! Thus the higher the concentration gradient of vanadium is, the lower the flux. Obviously, this flouts a simple Fickian treatment (though the diffusion versus time data are well represented by a Fickian diffusion

coefficient value) and, indeed, the result indicates coupling of the acid and vanadium flux. Figure 10 shows the derived diffusion constants as a function of composition for the constant ionic strength and constant acid cases, plotted as a function of vanadium concentration. From these plots, it appears that the most important determining factor for vanadium diffusion rate is the acid concentration. Incidentally, this effect is much larger than any effect that would be expected from viscosity changes.

Interestingly, this simple dependence on acid concentration also provides a possibility that the cross-over rate in the cell could be controlled by acid concentration. For the specific case of vanadium, this would be difficult because the acid concentration also plays a key role in the solubility of vanadium species. D. Conductivity Studies The primary practical result of uptake of acid and vanadium by the membrane is an overall lowering of conductivity. This is illustrated by Figure 11 which shows the conductivity vs. vanadium concentration in the bathing solution of 5M total sulfate.

Figure 11. Nafion 117 conductivity when equilibrated in electrolyte solutions of 0 to 2 mol∙dm-3 vanadium ions (V3+, VO2+ and VO2

+) and 5 mol∙dm-3 total sulfate/bisulfate background. The conditions used are similar to the practical compositions. Roughly a factor of 2 decrease in conductivity is observed for Nafion 117 relative to the same membrane in pure water. This result is similar for all valence states. To put this into perspective, this loss corresponds to roughly 150 mV extra loss due to membrane iR in cell operating at 1 A/cm2, a >10% efficiency loss in the cell. In Figures 12, we show data that highlight differences in the behavior of vanadium III and IV in the membrane. Simply put, the proton mobility is influence strongly by VO2+ but not by V3+ in the membrane. We are investigating the reasons for this.

Figure 12. Membrane conductivity loss caused by V3+ or VO2+ presence in the membrane.

Progress In this initial phase of the project, we developed and deployed a series of experimental methods to study the needed composition and transport properties of membranes under conditions used in VRBs. We have continued to carry out these detailed studies to reveal several important features of uptake and transport. Key findings include:

(i) Acid concentration in the vanadium solutions plays a critical role in all uptake and transport properties. Higher acid concentration discriminates against vanadium uptake and cross-over. This may provide a handle to control the cross-over rates.

(ii) The presence of different valence states in the membrane has little impact on the bulk parameters at a given concentration of vanadium.

(iii) However, the intrinsic proton mobility appears to be lower in the presence of VO2+ while V3+ has little or no influence.

(iv) Vanadium cross-over rates are strongly dependent on experimental details, particularly the choice of constant ionic strength vs. constant acid concentration condition.

Future Plans We will complete the initial plan of fully characterizing the vanadium effects on membranes and will expand slightly our testing of electrodes. However, our future work topics will be adjusted to coincide more closely with those of UT-Austin’s current directions, including direct collaboration through testing of their chemistry in our cells. Contacts Tom Zawodzinski: tzawodzi@utk.edu Alex Papandrew: apapandr@utk.edu