Progress in Flow Battery Research and Development

Many potential redox couples were screened by NASA (Refs. ^{22, 83}) since the first proposal of the redox flow cell concept by Thaller.⁷⁷ Out of several candidates for application as redox couples in the electrochemical energy storage system, the iron/chromium couple was selected and developed.⁸⁴ The main criteria used by NASA in the selection of iron and chromium were cost and availability. In general, the system consisted of acidified solutions of chromium [Cr(III)/Cr(II)] and iron [Fe(III)/Fe(II)], initially as unmixed reactants^{22, 83} and later as premixed solutions in order to address the issue of cross mixing of the electrolytes across the membrane.⁸⁵

In premixed solutions both the positive and negative electrolytes contained iron and chromium species as soluble salts in aqueous solutions of hydrochloric acid. The cell reactions as well as the main technical features of the iron/chromium system are summarized in Table II, while an historical overview of its development is given in Table III.

Scale-up studies of the iron/chromium RFB were conducted by a number of workers^{81, 94–97} but the system was not commercially developed at the time due to problems of low energy density for the mixed electrolyte cell, membrane fouling and the slow reaction of chromium redox species on most electrode surfaces that required expensive noble metal catalysts.⁹³

Thaller⁷⁷ discussed the possibility of employing a soluble Fe(III)/Fe(II) – Ti(IV)/Ti(III) redox system in aqueous hydrochloric acid solution for use in a redox flow battery. Preliminary size and cost estimates for bulk energy storage using such redox couples were also evaluated.⁸² The overall cost of constructing such a system compared well with that of competing energy storage systems and savings in transmission costs were also achievable. However, the system was never commercialized due to the slow kinetics of the negative electrode reaction. The technical features of the iron-titanium system are summarized in Table II.⁹⁸ The charge-discharge reactions are as follows^{86, 87}

The open-circuit potential (OCP) of this system was 1.19 V whilst operating at room temperature, with an energy efficiency varying between 44 and 50%.^{88, 89} The energy density of the system was reported to be 13.25 Wh/kg. These values were obtained for cells using lead as an electro-catalyst to enhance the kinetics of the titanium redox couple [Ti(IV)/Ti(III)] at a graphite negative electrode. The slow kinetics of this couple was also confirmed independently by other researchers.⁹⁹ Other workers⁸⁷ found that the kinetics of the titanium couple could be enhanced by impregnating the graphite negative electrode of their cell with palladium, but the cost of this would be prohibitive. Further investigations using flow cells have yet to be carried out to compare their performance with the original prototype system developed by NASA.⁸⁶ As with the Fe-Cr system, the low energy density and expensive electrode catalysts needed for the Fe-Ti cell make this system less attractive that other prospective redox couple combinations.

Organometallic redox species in acetonitrile solvent were proposed for redox flow batteries by Japanese researchers in the late-1980s.^{90, 100} These species included tris(2,2'-bipyridine) ruthenium(II) tetrafluoroborate and ruthenium(III) acetylacetonate. The former species was investigated in a redox flow cell, yielding an overall energy efficiency of 18% as shown in Table II.⁹⁰ The cell charge-discharge reactions are also given in Table II. Given the high cost of ruthenium, such a system is unlikely to become practical however and there is little justification for further research.

The iron-chlorine and tin-chlorine batteries were patented in 1985.¹⁰¹ These cells employed the Cl₋/Cl₂ couple in the positive half-cell and the Fe(II)/Fe(III) and Sn(II)/Sn(IV) couples in the negative half-cells respectively. Nozaki also reported studies of a secondary redox-flow battery (hybrid) with chromium and halogen couples giving a voltage of 1.2 V.¹⁰² In addition, an iron-chlorine redox system with graphite cloth gas electrodes was studied by Kondo (National Chemical Laboratory, Tsukuba, Japan) (Ref. ¹⁰³) while electrolytes for redox-flow batteries, prepared from ferrochromium ores, were patented by Wakabayashi (Chiyoda Chemical Engineering Co. Ltd., Japan) (Ref. ¹⁰⁴). However, none of these redox systems were considered for scale-up due to the poor electrochemical reversibility of the respective redox couples in solution.

Research on the all-vanadium redox flow battery (VRB) first began in 1984 at the University of New South Wales (UNSW), Australia under funding from the National Energy Development and Demonstration Council.^{14, 15} The VRB was first proposed by Skylllas-Kazacos and co-workers to overcome the inherent problem of cross contamination by diffusion of different redox ions across the membrane. By employing the same element in both half-cells, any cross contamination would be avoided, allowing the electrolyte life to be extended indefinitely.^{10, 14}

The VRB employs the V(II)/V(III) and V(IV)/V(V) couples in the negative and positive half-cells respectively with the following charge-discharge reactions:

Positive electrode reaction

Negative electrode reaction

The open circuit potential (OCP) of the fully charged cell is about 1.6 V when the negative and positive half-cell electrolytes comprise 2 M V(II) and 2 M V(V) respectively. The energy density for 2 M vanadium electrolytes is approximately 25 Wh/g.⁵⁰ The system has been successfully operated over a temperature range of 10–40°C.^{27, 44, 49}

Development of the vanadium redox flow battery began at the University of New South Wales in Australia where it was taken from the initial concept stage in 1984 through the development and demonstration of several 1–4 kW prototypes in stationary and electric vehicle applications during the late 1980s and 1990s.^14–63 As part of the 25 year vanadium flow battery research and development program, a wide range of research projects were undertaken, these spanning the areas of electrode screening and characterization,^15–23 electrocatalysis and carbon electrode modification and characterization,^24–26 electrolyte optimization and characterization^27–31 membrane screening, characterization and modification,^32–43 conducting plastic electrode formulation and evaluation,^44–48 additives for stabilisation of supersaturated vanadium solutions,^{49, 50} chemical regeneration,⁵¹ state-of-charge monitoring,^{52, 53} vanadium salt dissolution and electrolyte production,^{54, 55} control system development,^{52, 56} stack design and optimization^57–61 gelled electrolytes⁶² and vanadium/oxygen redox fuel cells,⁶³

A brief description of the all-vanadium redox battery's general properties and features is presented in Table IV, while its historical development is given in Table V.

Table IV. General properties and features of the all-vanadium and other vanadium based redox flow battery technologies.

No.	Redox system	Electrolyte composition	Charge/discharge reaction at electrodes	OCP (V) at 100% SOC	Charge/discharge current density(mA/cm2)	Cell type	Electrode and membrane materials used	Charge/discharge Efficiency (%)	References
1	All-vanadium	1.6–2 M vanadium sulphate in sulphuric acid in both half-cells	Negative electrode: V3+ +e− → V2+Positive electrode: VO2+ +H2O − e− → VO2+ +2H+	1.6	10–130	1–5 kW bi-polar stacks	Graphite felt electrodes heat bonded on carbon-filled polyethylene conducting plastic bipolar substrates. Modified low-cost perfluorinated cation exchange membrane.	80% at 40 mA/cm2 (overall)	10, 16, 79
	All-vanadium	1.5 M vanadium sulphate + 2 M sulphuric acid at 22°C in both half-cells	As above	1.6	40	Flow cell	Sandwich-type sulfonated poly(ether ether ketone) (SPEEK)/tungstophosphoric acid (TPA)/polypropylene (PP) composite.	83% overall	105, 106
2	Vanadium-bromine	1–3 M vanadium bromide in 7–9 M HBr plus 1.5–2 M HCl in both half-cells	Positive electrode: 2VBr3 + 2e − → 2VBr2 + 2Br−Negative electrode: 2Br− + Cl− → ClBr2 − + 2e–	1.4	20	Flow cell	Nafion 112 membrane. Electrodes: carbon or graphite felt bonded onto conductive plastic sheets	74 (overall)	79, 107
3	Magnesium-vanadium	Positive half-cell: 0.3M Mn(II)/Mn(III) in sulfuric acid). Negative half-cell: V(III)/V(II) in 5 M sulphuric acid	Positive electrode: Mn(II) → Mn(III) + e−Negative electrode: V(III) + e− → V(II)	1.66	20	Flow cell	Polyacrylonitrile (PAN) based carbon felt or spectral pure graphite electrodes with Nafion 117 (DuPont, USA) membrane	63 (overall)	108
4	Vanadium-cerium	Positive half-cell: 600 ml of 0.5 M Ce(III) in 1 M H2SO4. Negative half-cell: 600 ml of 0.5 M V(III) in 1 M H2SO4	Positive electrode: Ce3+ → Ce4+ + e−.Negative electrode: V3+ + e− → V2+	1.5	22	Cylindrical flow cell	Porous Vycor glass with pore size of around 45 Å as membrane. Carbon fibers of 10 μm diameter as negative electrode filled inside cylindrical membrane. Four bundles of the carbon fibers arranged evenly around the outside of the membrane as positive electrode.	90 (coulombic)	109–111
5	Vanadium-glyoxal(O2)	Positive half-cell: 50 ml glyoxal–HCl solution of different concentration. Negative half-cell: 1–2 M V(III) + 3 M H2SO4 solution	Positive electrode: [OC]RE + H2O → [OC]OX + 2H+ + 2e− (where [OC]RE represents the organic reductive raw materials and [OC]OX represents the electro-oxidized organic products).Negative Electrode: V3+ + e → V2+	1.2	20	Flow cell	The gas diffusion layer and a PTFE sheet (Nitto Denko, 50 mm thick) were placed on each side of a Nafion115 cation exchange membrane and then hot-pressed at 150°C to form a gas diffusion layer hot-pressed separator for the BRFB. Graphite plates and porous graphite felts served as current collectors and electrodes, respectively.	66 (coulombic)	112
6	Vanadium-cystine (O2)	Positive half-cell: 0.1 M cystine dissolved in HBr aqueous solution of different concentrations. Negative half-cell: 50 ml of 1 M V(III) + 3M H2SO4	Positive electrode: RSSR + Br2 + 6H2O → 2RSO3H + 10HBr (where RSSR = L-cystine and RSO3H = L-cysteic acid) Negative electrode: V3+ + e− → V2+	1.315	20	Flow cell	GDL hot pressed separator as membrane. It employed 2.5 mm thick graphite felts (dimension: 25 × 20 mm) contacted against graphite plates that served as current collectors.	58 (overall)	113
7	Vanadium-polyhalide	Positive half-cell: 1M NaBr in 1.5M HCl. Negative half-cell: 1M VCl3 in 1.5M HCl	Positive electrode: Br− + 2Cl− → BrCl2 − + 2e− Negative electrode: VCl3 + e− →VCl2 + Cl	1.3	20	Flow cell	Glassy carbon sheets as the current-collectors and graphite felt as the electrode material in both the half-cells. Nafion 112 membrane.	83 (coulombic) 80 (voltaic)	107
8	Vanadium acetylacetonate	0.01 M V(acac)3/0.5 M TEABF4/CH3CNin both half-cells	Positive electrode:V(III)(acac)3 → [V(IV)(acac)3]+ + e−. Negative electrode: V(III)(acac)3 + e− → [V(II)(acac)3]−	2.2	2.2 (charge) 0.2 (discharge)	Stationary H-type cell	Graphite electrodes and AMI-7001 anion-exchange membrane.	47 (coulombic)	114
9	Vanadium/air system	Positive half-cell: H2O/O2. Negative half-cell: 2M V2+/V3+ solution in 3M H2SO4	Positive electrode: 2H2O → 4H+ + O2 + 4e−. Negative Electrode: V3+ + e → V2+	≈ 1 V for 8 h	24 A/m2	flow cell with oxygen gas diffusion electrode	For charging, the air side of the cell contained a membrane-electrode-assembly (MEA) that was made from a catalyst coated Ti-mesh electrode of 100 mm thickness. For discharging, the air side of the cell contained a MEA of a catalyst coated sintered porous Ti-electrode of 1.2 mm thickness. Membrane was Nafion 117.	45.7 (overall)	115, 116

Table V. Historical Overview of the All-Vanadium Redox Flow Battery.

Year	Electrode materials	Electrolyte	Membrane	Battery type	Comment	References
1986	Graphite plates	The negative and positive half-cell electrolytes consisted of 0.1 M V (III) and 0.1 M V(IV) in 2 M H2SO4 respectively	Sulphonated polyethylene anion selective material.	Stationary H-type cell and laboratory-scale flow cell	Charged and discharged at 3 mA/cm2 and gave good performance. Graphite plates not suitable under high oxidizing conditions	19
1987	Graphite negative and iridium oxide coated titanium dimensionally stable anodes as positive electrodes	0.5–2 M vanadium solution	Sulfonated polyethylene cation selective and polystyrene sulphonic acid cation selective membranes evaluated	Single redox flow cell	Dimensionally stable anode material showed best stability during short term cycling compared with graphite plates and other types of electrodes	20
Graphite feltnegative electrodes	1.5 M vanadium solution prepared from 0.1 to 2M vanadyl sulfate (VOSO4) in 2M H2SO4	Polystyrene sulfonic acid cation selective membrane	Single redox flow cell	Coulombic and voltage efficiency of 90 and 81%, respectively, over 10–90% state of charge	16
1989	6 mm thick felt electrodes of 132 cm2 surface area bonded to a graphite impregnated polyethylene plate	2 M vanadium sulphate in 2 M H2SO4	Polystyrene sulfonic acid membrane	Single redox flow cell	87% overall energy efficiency obtained using these electrodes	44
1991	Graphite felt heat bonded onto conducting plastic bipolar electrodes	1.5–2 M Vanadium sulphate in H2SO4	Selemion CMV	1 kW stack incorporating 10 cells with 1500 cm2 electrode area	90% overall energy efficiency at 30 Amp charge-discharge currents. Maximum continuous power of 1.58 kW at 120 A	57
1991	Modified graphite fibre electrodes by surface ion exchange of Pt4+, Pd2+, Au4+, Mn2+, Te4+,In3+ and Ir3+ ions	Cyclic voltametric studies in 1–2 M VOSO4 in H2SO4	N/A	Small electrochemical cell	Electrode modified by Ir3+ exhibited the best electrochemical behaviour for the various vanadium redox species.	24
1992	Thermally treated graphite felt electrodes in air atmosphere at 400°C for 30 h	2 M V(III)/2 M H2SO4 solution as the negative electrolyte, and 2 M V(IV)/3 M H2SO4 solution as the positive electrolyte	Not specified	Single redox flow cell	Over 88% energy efficiency. Studied active surface functional groups on carbon and proposed methods to increase active sites for improved electrochemical activity	25
Chemically modified graphite felt electrodes by boiling in concentrated sulphuric acid for 5 h	2 M V(III)/2 M H2SO4 solution as the negative electrolyte, and 2 M V(IV)/3 M H2SO4 solution as the positive electrolyte	Not specified	Single redox flow cell	Surface modification of graphite felt was done with concentrated sulphuric acid to increase concentration of active sites for electron transfer reactions. 91% efficiency reported	26
1992	Graphite felt on graphite plate current collectors	2 M vanadium sulphate in 3 M H2SO4	Daramic based composite ion exchange membranes	Single redox flow cell	Preparation of composite membrane using low cost microporous separator. Coulombic, voltage and energy efficiencies of 95, 85 and 83%, respectively. More than 700 cycles (4000 h), without any appreciable drop in performance	33, 34
1997	Two layer, porous electrodes comprising high surface area porous carbon fibre electrode layer at the septum side and a porous low surface carbon fiber at the bipolar plate side	Vanadium in sulphuric acid	Not specified	Flow cell with electrode dimensions 45 cm x 80 cm used in 40–50 kW stacks	Grooves in porous graphite used to reduce pressure drop. 94.1% current efficiency, 82.5% overall efficiency, 87.6% voltage efficiency, 1.07 Ω.cm2 cell resistance and 0.51 kg/cm2 pressure loss when the electrolytic solution passed through the multilayer porous electrode. Electrode design used in 40–50 kW modules for 200 kW/800 kWh VRB load-levelling system at Kashima-Kita Electric Power Station	117, 78
1997	Carbon fibre felt electrodes	2 M VOSO4 in 4 M H2SO4 solution	Cross linked anion exchange membrane by accelerated electron radiation	Single redox flow cell	Overall energy efficiency of 80% reported	118
2002	Carbon-on-gold	Electrolysis of a 1 M solution of VOSO4 in 25% H2SO4	No membrane	Membrane-less vanadium redox fuel cell	A maximum of 10% cell efficiency was achieved	119
2006	Chemically treated carbon felt	1.5M VOSO4 + 3M H2SO4	Nafion (Du Pont)	14-cell 1 kW class VRB cell	10 x 1 kW stacks integrated into 10 kW battery. Energy efficiency of more than 80%, at an average output power of 10.05 kW	120
2007	Carbon felt	2 M V(IV) in 2.5 M H2SO4 catholyte and 2 M V(III) in 2.5 M H2SO4 anolyte	Nafion/SiO2 hybrid membrane was prepared via in situ sol–gel method	Single redox flow cell	1 M active species concentration, 20 mA cm−2 current density gave an energy efficiency of nearly 80%	121
2008	Graphite felt	2 M V(IV) in 2.5 M H2SO4 catholyte and 2 M V(III) in 2.5 M H2SO4 anolyte	Nafion–[PDDA-PSS]n membrane (n = the number of multilayers)	Single redox flow cell	Maximum CE of 97.6% and EE of 83.9% achieved at charge–discharge current densities of 80 mA cm−2 and 20 mA cm−2, respectively	122
Graphite felt (electrode), an adhesive conducting layer (ACL) and a flexible graphite plate (bipolar plate)	1.5M VOSO4 + 3M H2SO4	Nafion 117 membrane	VRB Single flow cell	Energy efficiency of 81% at a charge/discharge current density of 40 mA cm−2	123
2009	Graphite felt.	1.5M VOSO4 + 3M H2SO4	Nafion 115 membrane	VRB Single flow cell	A simple mathematical model approximates reaction conditions very well. At current density of 40 mA cm−2 a cell potential of 1.65 V is achieved at 90% state of charge	105
	Two pieces of carbon felt were used as electrodes, serpentine flow fields graphite as polar plates	2.0 M V3+/V4+ + 2.5 M H2SO4 solutions	Nafion/ORMOSIL (novel Nafion/organically modified silicate) hybrid membrane	VRB Single flow cell	Energy efficiency is 87.5% with novel membrane in comparison to traditional Nafion (74%) and Nafion/SiO2 hybrid membrane (80%)	124
	Two pieces of carbon felt used as electrodes, serpentine flow fields graphite as polar plates	1 M vanadium solution in 2.5 M sulphuric acid	Nafion/organic silica modified TiO2 composite membrane prepared by in situ sol–gel method	VRB Single flow cell	Novel membrane resulted in energy efficiency of 78% in comparison to 77% for normal Nafion membrane in the all-vanadium RFB (SOC of 75%). This was constant over a cycle life nearing 100.	125
2010	Carbon felt served as electrodes, and conductive plastic plates served as current collectors	1.5 M VOSO4 in 2.0 M H2SO4	Sandwich-type sulfonated poly(ether ether ketone) (SPEEK)/tungstophosphoric acid (TPA)/polypropylene (PP) composite membrane	VRB Single flow cell	82.6% energy efficiency in comparison to the employment of a Nafion 212 membrane for more than 80 charge/discharge cycles at 35.7 mA cm−2	106
	Nitrogen-doped mesoporous carbon	3.0 M H2SO4 + 1.0 M VOSO4 solution	No membrane for CV	Cyclic voltammetry and impedance tests only	The reversibility of the redox couple is greatly improved on N-MPC (0.61 V for N-MPC vs. 0.34 V for graphite), which is expected to increase the energy storage efficiency of redox flow batteries	125
	Thermally treated graphite felt electrodes	0.02 M VOSO4 in 1 M H2SO4 solution	Undivided reactor/membrane less	Single pass flow cell	13.4% energy efficiency, which is higher than membrane less vanadium redox fuel cell (Ref. 119)	13

Although vanadium redox couples had been previously considered for redox cell applications, they were believed to be impractical due to the very low solubility of V(V) compounds which would have restricted the concentration of the vanadium electrolyte to less than 0.5 moles/l, this being much too low for practical use. The UNSW breakthrough came when it was discovered that highly concentrated V(V) solutions could be prepared in sulphuric acid by the electrochemical oxidation of V(IV). By oxidising a 2 M vanadyl sulphate solution, it was possible to prepare a highly concentrated 2 M V(V) solution which did not precipitate over a reasonable temperature range.¹⁴ This meant that reasonable vanadium solution concentrations could be achieved for a practical flow battery system.

A second major challenge that had to be addressed during the early development was the high cost of vanadyl suphate originally used in the electrolyte production. Lower cost vanadium oxide materials could not be used due to their very low solubilities. A further milestone in the early UNSW research program therefore, was the development of a low cost process for producing vanadium electrolyte from the vanadium oxide raw material. The low solubility of the oxides meant that simple dissolution could not be used in electrolyte production, so electrolytic and chemical reductive dissolution processes were developed,⁵⁴ allowing lower cost raw materials to be employed and thereby making the VRB economically viable.

The initial system developed at UNSW had an overall energy efficiency of 71% but with further enhancements in materials and cell design, an overall energy efficiency of up to 90% was achieved with a 1 kW VRB stack in 1991.⁵⁷ These enhancements included the identification of high performance membranes with low electrical resistance to reduce ohmic losses and low vanadium permeability to maximize coulombic efficiency. In the area of electrode materials, considerable screening of electrode materials was undertaken and the kinetics of the vanadium redox couples were evaluated at different electrode surfaces. Both redox couple reactions were found to be quasi-reversible,^{18, 19} however, the use of high surface area carbon and graphite felts allowed very low current density operation, with a dramatic reduction in activation overvoltage and increased voltage efficiency.

Due to the highly oxidizing nature of V(V) ions in the fully charged positive electrolyte, there are very few materials that can be employed as positive electrodes.^{15, 20} Carbon and graphite are therefore used as both positive and negative half-cell electrode materials, but early studies showed that the electrochemical activity of carbon and graphite materials is dependent on the oxide functional groups present on the surface.^23–26 Sun and Skyllas-Kazacos proposed a mechanism for electron mediation via the surface C-O-H bonds for the vanadium oxidation and reduction reactions and identified a number of chemical and electrochemical treatment methods that could be used to increase the surface concentration of these active sites.^24–26 Later studies confirmed this and also showed that electro-oxidation of graphite felt using 3 M H₂SO₄, 0.0087 M V(IV) and 0.0087 M V(V) resulted in high voltage efficiencies of 85% at 50 mA cm₋₂ current density.¹²⁶ The improvement of the electrochemical activity was also ascribed to the increase in the COOH functional group on the felt surface.

Another critical area for the development of the VRB has been in the identification, characterization and fabrication of suitable ion exchange membranes with good stability, low resistivity and low permeability to vanadium ions. During the early development of the VRB at UNSW, very few commercial membranes could satisfy all of these requirements and only the New Selemion anion exchange membrane (Asahi Glass Japan) and the Nafion cation exchange membranes were found to provide the required chemical stability in the highly oxidising V(V) solution of the charged positive half-cell electrolyte.^{32, 37} Because of the high cost of these membranes however, the UNSW group investigated the preparation of low cost composite membranes based on Daramic separator material^33–37 and also evaluated a range of membrane pre-treatment methods to improve the performance of other lower cost membrane types.^38–43 The mechanism of water transfer across ion exchange membranes in the VRB was also investigated along with methods to reduce this by membrane modification.^{39, 42, 43}

In addition to the basic research projects in the areas of electrodes, electrolytes and membranes, during the 1990s, of the UNSW team was also involved in the design and installation a 5 kW/15 kWh VRB in a demonstration Solar House in Thailand⁶⁰ and a VRB powered electric golf cart field trial.⁶¹ Further technical development of the VRB system was undertaken by Mitsubishi Chemicals, Kashima-Kita Electric Power Corporation and Sumitomo Electric Industries in the mid to late 1990s, leading to considerable field testing and demonstrations in Japan in a range of applications (to be described in more detail later).

Since 2002, several research groups have begun significant research and development activities on the VRB in China and elsewhere.¹²⁷ These activities have expanded on the original work of Skyllas-Kazacos and co-workers and have covered the development of novel membranes,^{41, 43, 106, 121–125, 128–137} electrocatalysis,^{27, 126, 138–140} mechanistic studies of vanadium redox couples,^{31, 140–144} cell modelling and simulation studies^{105, 145–149} and stack development and demonstrations.^{10, 120, 127, 38, 150, 151} Most of the recent research activities have focussed on the development of new low cost membranes.

Jia et al.¹⁰⁶ synthesized a novel sandwich-type composite membrane based on sulfonated poly (fluorenyl ether ketone) (SPEEK). The SPEEK/tungstophosphoric acid/polypropylene (SPEEK/TPA/PP) composite membrane consisted of a film of polypropylene (PP) between two layers of SPEEK/TPA composite membranes. They compared its properties and performance against Nafion 212 and found that the SPEEK/TPA/PP composite membrane exhibits the lowest diffusion coefficient for V(IV) ions under the reported test conditions, while a VRB single cell using the SPEEK/TPA/PP composite membrane gave a higher energy efficiency compared with Nafion 212. The long-term stability of this membrane was not however, reported.

New membrane materials based on SPEEK- SiO₂ composites have also been evaluated and proton conduction comparable to that of Nafion N117 and significantly lower V(IV) ion permeation were reported.¹³³ Again the long-term stability of this material has yet to be verified. Many of the more recently synthesized hydrocarbon or composite membranes designed for VRB applications have not been extensively studied with regard to their long-term chemical stability and in most studies, battery cycling performance is only reported for a short number of cycles¹²² making it difficult to assess their true potential for commercial application. In the interim therefore, New Selemion and Nafion continue to be used in early production systems. In the case of New Selemion, are excellent long-term performance has been demonstrated and the costs are reasonable. On the other hand, Nafion membranes are still very expensive, but offer very high chemical stability in the highly oxidising V(V) electrolyte.

Despite the significant progress in the development of the VRB for commercial application therefore, a number of challenges still remain and these will be discussed further in later sections.

Several systems have been developed over the years based upon the use of one half of the all-vanadium redox flow battery. These systems have been summarized briefly in Table IV. The previous review paper⁶⁸ discussed the vanadium-bromine system and the vanadium-polyhalide systems. Other systems have been reported since 2006 and these are covered in the present review.

Vanadium-polyhalide

The vanadium-polyhalide and vanadium bromide batteries were also invented at UNSW by Skyllas-Kazacos and coworkers.^{10, 127} The cells employ the V(II)/V(III) couple and the Br₋/Br_{3 −} couple in the negative and positive half-cells respectively with the following cell reactions

Positive Half-Cell Reactions

Negative Half-Cell Reactions

Preliminary studies were carried out with a 3–4 M vanadium-bromide solution in the negative half-cell and a 8–10 M HBr solution in the positive half-cell by Skyllas-Kazacos¹⁰⁷ followed by evaluation of membrane materials.¹⁵² For this concentration of active ions, it was possible to reach energy densities up to 50 Wh kg₋₁.^{10, 127, 152} This cell showed rapid loss of capacity however due to the transfer of vanadium ions across the membrane into the positive half-cell solution because of the large difference in ionic strength between the two half-cell solutions. To overcome this osmotic pressure effect, vanadium bromide was added to both half-cells, giving rise to the current G2 (second generation) V-Br cell technology that employs the same electrolyte in both half-cells. As with the all-vanadium battery, the G2 V-Br also overcomes the problem of cross contamination, but the higher solubility of vanadium halides compared with vanadium sulphate salts, allows much higher energy densities to be achieved. This technology was also patented in 2008.¹⁵³

Further development of the V-Br technology was carried out by UNSW and V-Fuel Pty Ltd between 2005 and 2010 leading to the identification of highly stable, low cost membranes and electrode materials for the cell, in addition to the evaluation of bromine complexing agents such as tetrabutylammonium bromide, N-ethyl-N-methylpyrrolidiniumbromide (MEP), and N-ethyl-N-methylmorpholiniumbromide (MEM) to prevent the formation of bromine vapor during charge.¹²⁶ A feature of the G2 V-Br is the formation of a two-phase electrolyte system in which the bromine complexes separate out into an organic phase during charging, the stability of which is a function of temperature and state-of-charge. Unfortunately the current complexing agents are too expensive for commercial application, so commercialisation of the G2 V-Br will be dependent upon the successful development of improved, low cost complexing agents that produce stable bromine complexes over a wide temperature (0–50°C) and SOC ranges.

Vanadium-cerium

Vanadium-cystine

Other vanadium based redox flow systems

The sulphide-polysulphide system was first patented in 1983, opening up the future for research in the polysulphide-bromine redox flow battery.¹⁵⁴ This system was found to be attractive for RFB applications due to abundance of the electrolyte, reasonable cost of chemicals and high solubility in aqueous media.⁶⁸ The polysulphide-bromine redox flow battery, often referred to as the Regenesys cell, has a nominal open-circuit cell potential of 1.5 V and cell energy efficiencies of 60–65% depending on operating conditions. The cell operating temperature is typically between 20 and 40°C.⁶⁸ Table VI summarizes the battery operating conditions briefly, while Table VII briefly describes the historical evolution of the technology.

Technical challenges with this system have included:^{68, 165}

• (a)  
  cross-contamination problems of both electrolyte solutions over a period of time;
• (b)  
  The difficulty in maintaining electrolyte balance;
• (c)  
  The possibility of deposition of sulphur species on the membrane; and
• (d)  
  The need to prevent H₂S(g) and Br₂(g) formation.

Most of the development of the polysulphide-bromine system was carried out by Innogy in the 1990s and considerable advances were made with stack design and fabrication. Numerical modelling of the polysulfide-bromide (PSB) system revealed that mass transport overpotentials at the bromide electrode limit the performance during discharge.¹⁶⁶ The model showed that significant drift in conditions could occur due to self-discharge and electro-osmotic effects. Careful electrolyte management was suggested to ensure reliable operation of the polysulphide-bromine RFB system. Because of the complexity of the electrolyte management system, however, it was decided to restrict the application of the polysulfide-bromide RFB to MW-scale installations where the electrolyte maintenance costs would not be prohibitive. A separate mathematical model incorporating capital and operating costs to predict the technical and commercial performance of the polysulphide-bromine RFB at a 120MWh/15MW utility-scale storage plant for arbitrage applications revealed a net loss of US$0.0073/kWh at an optimum current density of 500 Am₋₂ and an energy efficiency of 64%,¹⁶⁷ indicating the need for further cost reduction. Furthermore, unlike the V-Br cell, the polysulphide-bromine cell has not utilised complexing agents to bind any bromine produced at the positive electrode during charging, and this has often been seen as a considerable safety risk with this technology. Hence, considerable research is still required to ensure that this system overcomes current techno-economic constraints and safety concerns in order to become a widespread commercialized technology.

These systems were very briefly mentioned in the previous review paper.⁶⁸ Two systems have been proposed as a means of utilizing excess depleted actinides for energy storage purposes. One involved the neptunium couples Np₃₊/Np₄₊ and NpO_{2 +}/NpO_{2 2+} in aqueous solution¹⁶⁸ while the other considered the use of uranium {U(IV)/U(III) and U(VI)/U(V)} couples in organic solvents.^169–173 An open circuit potential of around 1 V was estimated for the all-uranium couples complexed by a range of β-diketone ligands.¹⁷⁴ Besides this, charge/discharge test data of the all-uranium redox flow battery have not been provided as yet. The same is the case for the all-neptunium redox couple system, although theoretical calculations have revealed that an all-neptunium battery can produce energy efficiencies ranging from 40 to 99.1%,^{175, 176} with 99.1% efficiency obtainable at 70 mA/cm₂. Some more details on the system are given in Table VII.

A major obstacle in the development of actinide-based RFB, is the use of radioactive redox species that is likely to encounter significant consumer resistance. Special precautionary measures and a thorough investigation will therefore need to be conducted to evaluate their safety and environmental implications before commercialization. For example, the high radioactivity of neptunium has limited the practical evaluation of the all-neptunium redox flow battery so that only theoretical estimations of energy efficiencies are available by means of mathematical modeling.¹⁷⁶

The latest redox flow battery chemistries that are currently being developed are summarized in Table VIII. The zinc-nickel hybrid system appears to give an energy efficiency of 86% (Ref. ¹⁸¹) comparable to the all-vanadium RFB, followed by the tiron/Pb redox flow battery.¹⁸⁰ The zinc-nickel hybrid system utilises the Zn(II)/Zn and Ni(III)/Ni(II) redox couples. Since this system uses a single electrolyte and produces solid products at the electrodes during charging, it does not require a membrane, so its cost is likely to be less than most conventional redox flow battery systems.¹⁸² Theoretically, the deposition/dissolution of zinc on inert metal current collectors can be cycled endlessly.¹⁸⁵ However, that is not possible practically due to formation of zinc dendrites during charging. Researchers have studied the morphology of zinc dendrites and found that at higher electrolyte flow rates (> 15 cm s₋₁) good cycle life for the battery can be obtained at 100% depth of discharge.¹⁸⁶ Complete discharge is however a critical requirement for the long-term prevention of dendrites, and this produces an operational restriction on any cell employing the Zn₂₊/Zn couple. In addition, the cycle life of the zinc-nickel single flow battery is dependent on the stability of nickel oxide electrodes in the presence of zinc ions in the electrolyte that lowers the discharging capacity of the nickel oxide electrode. Cheng et al found however, that in concentrated KOH electrolytes containing 20 g l₋₁ LiOH, addition of 0.4 M ZnO to the electrolyte actually enhanced the stability of the nickel oxide electrodes during a cell cycling.¹⁸⁵ The potential for Zn dendrite formation will however be the major technical challenge for the zinc-nickel flow cell and electrolyte additives that can prevent dendritic growth during partial discharge operation is an area for further research and development.

In the case of the tiron-Pb redox flow battery studies, it is interesting to note that tiron (4,5-dibenzoquione-1,3-benzenedisulfonate) was investigated in aqueous environment whereas a similar aromatic species (rubrene) was investigated in organic media⁷⁶ and gave very poor electrochemical performance. It may be interesting to investigate tiron in organic media and compare its performance with that of the aqueous system to assess its suitability as active species in the positive electrolyte of a RFB. On top of that, a preliminary understanding of its electrode reaction mechanism in both acidic aqueous solutions and organic solvents may be attempted in an undivided redox flow battery similar to the reactor reported by Chakrabarti and co-workers.^{12, 13, 178, 187–189}

An all-chromium redox cell was investigated by Bae and co-workers,^{93, 178} building on an original proposal by Chen and co-workers.¹⁹⁰ The static, H-type cell employed chromium-EDTA complex as redox species in HCl media and an energy efficiencies of 15% was reported⁹³ whereas for the same redox species in a flowing undivided battery, poorer efficiencies of 7% were obtained.¹⁷⁸ Although static H-type cells are unlikely to produce good performance because of poor cell geometry, it should be mentioned that the all-vanadium redox species showed much better performance when tested in similar cell designs.^{183, 187–189} However, recent studies on chromium acetylacetonate redox couple complexes in H-type glass cells gave comparable charge/discharge performance to vanadium acetylacetonate in acetonitrile.^{114, 184} Overall efficiencies of 20% or less were obtained with these organic based systems similar to those achieved with an all-ruthenium redox flow battery (Fig. 2) using acetonitrile as the solvent.^{12, 187–189} High cost and low efficiencies have however limited the application of organic solvents for the redox flow battery.

Zoom In Zoom Out Reset image size

Methylimidazolium iron chloride molten salt system has also been considered for redox flow battery applications.¹⁹¹ It was predicted that if a sodium chloride-sodium electrode was combined with this EMICl–FeCl₂–FeCl₃ molten salt, a high energy density per unit volume may be expected. Since Na(I)/Na couple in EMICl–AlCl₃ system has the formal potential of − 2.15 V at room temperature,¹⁹² the electromotive force of approximately 2 V can be expected for the Na/EMICl–FeCl₂–FeCl₃ battery. Although this battery appears to have the advantage of a low operation temperature and a long cycle life compared with Na–S and Zebra cells,¹⁹³ further work with this system appears to be lacking in the literature as focus has been more towards the all-vanadium and polysulfide-bromide systems over the years. One reason for the lack of activity in the area of ionic liquids for flow batteries is the fact that these materials are known to be sensitive to air and moisture, making their handling difficult in large-scale commercial applications. Although other ionic liquids that are not as sensitive to air and moisture may be found, these materials also tend to be quite expensive and are unlikely to be economically viable for these types of applications compared to the lower cost aqueous systems. Given the large electrochemical window of many ionic liquids however, the possibility of using redox couples that fall outside the decomposition potential of water, may open the way to the development of high voltage flow cells that offer much higher power and energy densities than current aqueous systems. Further investigation of such couples could therefore prove fruitful as long as practical systems can be shown to offer better performance, cell voltage and cycle life than the VRB and PSB systems to offset the high costs of these electrolytes.

The electrochemical behavior of the Fe(III)/Fe(II)–triethanolamine(TEA) complex redox couple in alkaline medium and the influence of the concentration of TEA were investigated recently.¹⁹² A change of the concentration of TEA mainly produces the following two results:

• (1)  
  With an increase of the concentration of TEA, the solubility of the Fe(III)–TEA can be increased to 0.6 M, and the solubility of the Fe(II)–TEA is up to 0.4 M.
• (2)  
  In high concentration of TEA with the ratio of TEA to NaOH ranging from 1 to 6, side reaction peaks on the cathodic main reaction of the Fe(III)–TEA complex at low scan rate can be minimized.¹⁹²

The electrode process of Fe(III)–TEA/Fe(II)–TEA was shown to be electrochemically reversible with higher reaction rate constant than the non-complexed species.⁹³ Constant current charge–discharge showed that applying anodic active materials of relatively high concentrations facilitates the improvement of cell performance. The open-circuit potential of the Fe–TEA/Br₂ cell with the Fe(III)–TEA of 0.4 M concentration, after full charging, is nearly 2 V and is about 32% higher than that of the all-vanadium batteries, while the energy efficiency is comparable at approximately 70%.⁹³ Although the active material concentrations used to date have been too low for practical application, further optimization of the electrolyte composition may establish its potential for future commercialization.

The chemically regenerative redox fuel cell is a type of fuel cell that employs redox couple solutions as electron mediators for the fuel and oxidant reactions. A chemically regenerative redox fuel cell is thus a type of flow cell since it contains two redox couples which are circulated past the electrodes, and after electrochemical reaction at the electrodes, the solutions are passed into regeneration reactors where they are re-reduced or re-oxidized by the reductant and oxidant respectively. After the regeneration step the solutions are once more circulated past the electrodes, and the process proceeds. Most of the early work on redox fuel cells was reported by Kummer and Oei^194–196 whose work has shown the advantages and the limitations of the redox fuel cell. These workers investigated a wide range or redox couples, the main criterion for selection being the feasibility to regenerate the charged species using hydrogen and/or oxygen for the negative and positive half-cell reactants respectively, while also attaining the required power density for electric vehicle applications.^{195, 197} Other workers evaluated different membranes for redox fuel cells^{119, 198} and regeneration reactants.^{199, 200} The main attraction of this concept is the possibility of avoiding catalysts at the electrode surface and of using simple (inexpensive) electrode materials. Although hydrogen was the original fuel of choice, the concept offers a freedom of choice of fuels. The main disadvantage of using hydrogen for the regeneration of the negative half-cell active species is the relatively high reversible potential for the hydrogen couple that limits the range of redox couples that can be used in the negative half-cell. On the other hand, early studies^194–196 also showed that the oxidative regeneration of the positive half-cell couple using air or oxygen, is also kinetically slow, requiring relatively expensive catalysts that negate the main purpose of this approach.

More recently, a group of researchers from the University of British Columbia and the National Research Council of Canada has been working on two new approaches for a direct liquid redox fuel cell (DLRFC) in which the air cathode of a regular direct methanol liquid fuel cell is replaced with a metal-ion redox couple over a carbon cathode. For example, in a Fe-methanol fuel cell, methanol is used as the fuel and Fe₃₊ is used as oxidant. When the Fe₃₊ is depleted, the Fe₂₊ is passed through a separate regeneration cell where it is reacted with oxygen gas at the anode to reform the Fe₃₊ reactant for the fuel cell. In the DLRFC described by Ilicic et al.²⁰¹ however, spontaneous redox couple regeneration is achieved by simply substituting the methanol anolyte with an air stream on the anode side. The methanol anode then becomes an air cathode, which reverses the direction of electron flow and regenerates the Fe₃₊ oxidant in the DLRFC.

The first approach uses mixed-reactant operation that involves supplying a mixed methanol Fe₂₊/Fe₃₊ redox electrolyte only to the carbon cathode. Spontaneous methanol crossover supplies the fuel to the anode. This approach eliminates problems associated with the oxygen diffusion electrode and has the potential to significantly improve the cost, compactness, and volumetric and gravimetric power densities of the cell. The second approach is the in situ regeneration of the redox couple by supplying air to the methanol anode that then becomes an air cathode, which reverses the direction of electron flow and regenerates the redox couple on the other electrode.^{201, 202}

Hybrid flow batteries are distinguished from conventional redox flow batteries by the feature that at least one redox couple species is not fully soluble and may be either a metal or a gas. A number of hybrid flow cells were listed in Tables VI and VIII, but the most widely known of these is the zinc-bromine battery. The underlying concept of the zinc-bromine battery was first proposed more than 100 years ago, but in the mid 1970s and early 1980s, Exxon and Gould pioneered the initial designs for practical application. The zinc-bromine hybrid system, ranging in size from 50 to 400 kWh, is capable of storing energy for 2–10 h at efficiencies of 70% or higher¹⁵⁸ Coulombic and voltage efficiencies were reported to be around 90 and 85% respectively, whereas the energy density is around 65–75 Wh kg₋₁ (Ref. ¹⁵⁹) The system is briefly summarized in Table VI and a schematic of the zinc-bromine hybrid system is given in Fig. 3.¹²⁷

Zoom In Zoom Out Reset image size

In order to optimise the zinc-bromine battery, various mathematical models have been used to describe the system.^203–206 The problems with the zinc-bromine battery include high cost electrodes, material corrosion, dendrite formation during zinc deposition on charge, high self-discharge rates, unsatisfactory energy efficiency and relatively low cycle life⁶⁸ Another disadvantage of this system is slow kinetics of the bromine/bromide couple that causes polarization and loss in voltage efficiency. To overcome this, high surface area carbon electrode on the cathode side is normally used to reduce the effective current density, however, the active surface area of the carbon eventually decreases and oxidation of the carbon coating occurs.⁶⁸

In zinc-bromine hybrid systems, the energy storage and power of the battery are not fully decoupled as the energy storage capacity will depend on the thickness and morphology of the metallic layer formed. A porous separator is often used between the positive and negative electrodes to avoid the reduction of dissolved bromine during charge⁶⁸ however bromine cross-over is still an important issue, as is the problem of dendritic growth and shorting. Electrolyte additives (e.g., quaternary ammonium salts) can be used to complex any dissolved bromine that has been inadvertently transported through the membrane to the zinc half-cell.²⁰⁷ However, its performance does not match that of the all-vanadium RFB as yet and is mainly being developed for smaller applications up to 500 kWh. The main reason for the limited attention has been the operational limitations associated with the need to prevent zinc dendrites. Despite extensive work to optimise the design of electrolyte channels and manifolds to minimise shunt currents, zinc dendrites can still form after extended cycling, causing channel blockage and shorting through the separator. This is typically overcome by regular full discharge to completely strip all of the zinc from the negative plates to eliminate sites where dendrites can grow. This requirement creates undesirable operational restrictions, so further work on electrolyte additives to inhibit dendritic growth is an area that could yield considerable benefits for the future implementation of this technology.

Despite these problems with zinc plating, however, the very negative potential of the Zn₂₊/Zn couple continues to make this half-cell attractive for flow cell applications. The zinc-cerium hybrid redox flow battery is one such system that combines the very negative zinc couple with a very positive cerium couple to yield a cell voltage of more than 2 V and an OCP that is higher than any of its commercial competitors.¹⁶³ The zinc-cerium has been under development since the early 1990s by Electrochemical Design Associates Inc.¹⁵⁹ and some of its properties are given in Table VII. The testing and development of this system has been undertaken by Plurion Systems Limited. This hybrid flow cell involves the use of salts of zinc and cerium in an organic solvent.^{68, 208} It has some similarity to the zinc-bromine system in that the negative half-cell redox couple involves a solid zinc phase. It also uses an environmentally benign organic acid as the solvent (not degraded by cross-membrane migration) and a common electrolyte system in both half-cells.

Mathematical modelling to understand the redox nature of the Ce(IV)/Ce(III) redox couple has also been conducted for a batch system comprising an electrochemical reactor and an electrolyte circuit.²⁰⁹ The batch recycle system consisted of a pumped flow through divided FM01-LC parallel-plate electrochemical reactor (64 cm₂ projected electrode area) and a well mixed tank (3600 cm₃). Unfortunately, significant differences between experimental and predicted values were found at long electrolysis times. This was partly attributable to the presence of solvated species and complex formation involving Ce(III) and Ce(IV) species, which modified the actual concentration of Ce(III) and Ce(IV) from the predicted values.²⁰⁹ This has limited the energy density of the zinc-cerium system to date, however, further research may help to address this limitation, allowing it to compete commercially with the all-vanadium RFB.

Other zinc hybrid systems (such as zinc-nickel) have been developed recently and will be discussed in later sub-sections of this paper.

The system differs from the traditional lead-acid battery since it uses a highly soluble form of the Pb(II) species that is supplied as a aqueous acid electrolyte for both the negative and positive half-cells reactions.⁶⁸ It also differs from conventional redox flow batteries since it uses a single electrolyte and involves the formation of solid products at the two electrodes during charging, so that no separator or membrane is necessary. This reduces the cost and design complexity of the batteries significantly⁶⁸ Some properties of the system are described in brief in Table VI, while a schematic representation of the system is given in Fig. 4.

Zoom In Zoom Out Reset image size

The electrode reactions involve the conversion of the soluble species into solid Pb and PbO₂ phases at the negative and positive electrodes respectively during charging and re-dissolution during the discharging cycles. The deposition of solid phases on the electrodes during charging introduces complexities to the electrode reactions that may reduce the performance of the battery if the metal grows across the inter-electrode gap and short circuits the battery.⁶⁸ Dissolution and deposition of lead is fast and no overpotentials are usually required, however as with conventional lead-acid batteries, hydrogen evolution is observed during the charge cycle at high state-of-charge, thus reducing storage capacity.²¹⁰ These cells have been studied in several electrolytes; percholoric acid, hydrochloric acid, hexafluorosilicic acid, tetrafluoroboric acid and most recently in methanesulphonic acid.^{68, 161, 210–212}

The structure of lead deposits (approximately 1 mm thick) formed in conditions that are met at the negative electrode during the charge/discharge cycling of a soluble lead-acid flow battery was examined in some detail by Pletcher et al.²¹³ The quality of the lead deposit could be improved by appropriate additives and the preferred additive was shown to be the hexadecyltrimethylammonium cation, C₁₆H₃₃(CH₃)₃N₊, at a concentration of 5 mM. In the presence of this additive, thick layers with acceptable uniformity could be formed over a range of current densities (20–80 mA cm₋₂) and solution compositions.²¹³ While electrolyte compositions with lead(II) concentrations in the range 0.1–1.5 M and methanesulfonic acid concentrations in the range 0–2.4 M have been investigated, the best quality deposits are formed at lower concentrations of both species. Surprisingly, the acid concentration was more important than the lead(II) concentration; hence a possible initial electrolyte composition for an efficient system was postulated to be 1.2 M Pb(II) + 5 mM C₁₆H₃₃(CH₃)₃N₊ without added acid.²¹² The system was predicted to be cycled between 0.2 M Pb(II) + 2 M CH₃SO₃H at top of charge to 1.15 M Pb(II) + 0.1 M CH₃SO₃H at bottom of charge. Also the current density was expected to be up in the range of 20–80 mAcm₋₂ for suitable operation of the system.²¹⁴

Extensive cycling of the soluble lead flow battery has revealed unexpected problems with the reduction of lead dioxide at the positive electrode during discharge.²¹⁴ This has led to a more detailed study of the PbO₂/Pb₂₊ couple in methanesulphonic acid. The variation of the phase composition measured by XRD and deposit structure measured using SEM have been defined as a function of current density, Pb₂₊ and H₊ concentrations, deposition charge and temperature as well as the consequences of charge cycling.²¹⁴ Pure α-PbO₂, pure β-PbO₂ and their mixtures can be deposited from methanesulphonic acid media successfully. The α-phase deposits as a more compact, smoother layer, which is well suited to charge cycling. While the anodic deposition of thick layers of PbO₂ is straightforward, their reduction is not; the complexities are explained by an increase in pH within the pores of the deposit. The results suggest that operating the battery at lead(II) concentrations < 0.3M and elevated temperatures should be avoided.²¹⁴

It has been demonstrated that extended cycling of the soluble lead acid battery in a 10 × 10 cm parallel plate cell is possible, with greater than 100 cycles achievable under some conditions.²¹⁵ Eventual failure is, however, inevitable if the battery is operated under conditions where solids are allowed to accumulate continuously on the two electrodes. Failure usually results from: (a) shorting of the electrodes owing to lead dendrites formation largely around the edges of the negative electrode plate and (b) poor adhesion of PbO₂ to the positive electrode surface leading to particles in the electrolyte and loss of active material.²¹⁵

Extended cycling of the battery can lead to problems due to an imbalance in the coulombic efficiency of the negative and positive charging reactions that produce deposits of Pb and PbO₂ on the electrodes.¹⁶⁰ Periodic addition of hydrogen peroxide to the electrolyte largely overcomes several operational problems seen during extended cycling. It is shown that this treatment greatly extends the number of cycles that can be achieved with a reasonable energy-, voltage-, and charge efficiency of 54–66, 71, and 77–91%, respectively.¹⁶⁰ Further research and development is necessary before the soluble lead acid battery can be considered for commercial scale projects.

A further extension of the chemically regenerative redox fuel cell is the hybrid redox fuel cell concept that employs a redox couple electrolyte for the negative half-cell, but the positive half-cell electrolyte is replaced by air or oxygen. Like a conventional fuel cell, air or oxygen is fed through a gas diffusion electrode in the positive half-cell, but in contrast to the fuel cell, the negative half-cell reactant comprises a soluble redox couple electrolyte, as illustrated in Fig. 5. By replacing the positive half-cell electrolyte reaction with an air or oxygen gas diffusion electrode, the total electrolyte volume is reduced by half, effectively doubling the energy density of the system.

Zoom In Zoom Out Reset image size

The vanadium-oxygen redox fuel cell (VOFC) concept was initially proposed by Kaneko and co-workers in 1992 (Refs. ^{115, 116}) and first evaluated at UNSW by Menictas and Skyllas-Kazacos in 1997.^{63, 216} In this project the performance of the VOFC over a range of temperatures and using different types of membranes and air electrode assemblies was evaluated. Despite early problems with the membrane electrode assemblies that saw separation of the membrane due to swelling and expansion during hydration, with improved fabrication techniques, this problem was minimized and it was possible to operate a 5-cell VOFC system for a total of over 100 h without any deterioration in its performance.⁶³

With renewed interest in electric vehicles, the VOFC concept has recently received further attention with a range of reports emerging from the Fraunhofer Institute²¹⁷ and Twente University in the Netherlands²¹⁸ Hosseiny et al.²¹⁸ reported the effective quadrupling of the energy density of the VCFC relative to the conventional all-vanadium redox battery by the use of 4 M vanadium solutions in the negative half-cell. The elimination of the positive half-cell reactant also eliminates the problem of thermal precipitation of V(V) species at high temperatures that currently restricts the upper temperature range of the VRB to 40°C. Without a threat of thermal precipitation, the researchers were able to operate the cell at 80°C, thereby achieving 4 M solubilities for the V(II) and V(III) species in the sulphuric acid negative electrolyte. This is a major advance in the development of a high energy density VOFC that could find application in electric vehicles, however, an issue that would need to be addresses is the requirement to maintain high temperatures in the negative electrolyte tanks and half-cells to prevent precipitation of the vanadium ions on cooling to room temperature. The evaluation of alternative supporting electrolytes that can give high solubilites for V(II) and V(III) ions at room temperature is therefore an area that requires further attention.

The first 1 kW prototype Fe-Cr systems was developed in 1980 by NASA.⁹¹ The Fe-Cr redox flow battery was installed and tested in a photovoltaic system but results were not sufficiently satisfactory for consideration of a full-scale prototype. Further work in Japan in the 1980s did however lead to the development of a 10 kW Fe-Cr redox battery prototype with an 80% energy efficiency and 300 life cycles, as demonstrated by Shimizu and co-workers (Kansai Electrical Power Co., Amagasaki, Japan).²¹⁹ Operations involving the catholyte and anolyte circulation rates (in a 10 kW Fe-Cr redox-flow battery) to save energy, and a method of rebalancing were described by Nakamura (Mitsui Ltd., Japan).²²⁰ Other similar prototype systems were developed and tested by NASA (Ref. ⁸⁵) as well as in Japan⁹⁵ for different applications while a 0.1 MW pilot scale unit was evaluated in the mid 1990s in Spain.⁸¹ Commercial development of the Fe-Cr battery was however abandoned because of problems of cross-contamination between anolyte and catholyte, poor energy efficiencies due to hydrogen evolution at the negative electrode and fouling of the ion-exchange membranes.⁶⁸ Hence, this system was largely ignored during the late-1990s and early 2000s. In the late 2000s however, the Fe-Cr battery was revisited by a US-based technology company, Deeya Energy,²²¹ when the world prices for vanadium briefly peaked at close to 4 times historical average prices and the company saw the Fe-Cr system as a potentially lower priced product than the VRB.

The all-vanadium redox flow battery has to date shown the greatest potential for large-scale energy storage applications with long cycle life and high energy efficiencies of over 80% in large installations.^15–20 This technology has already been applied in a MW- scale and several kW scale projects,^222–231 with many practical demonstrations covering a range of stationary and mobile applications in countries such as Japan, Europe, Australia and the USA.^{79, 222–225} One of the main advantages of the vanadium redox flow battery that distinguishes it from most other flow battery systems is the use of the same element in both half-cells that prevents cross contamination and a theoretically indefinite electrolyte life. It also exhibits a low cost for large storage capacities; cost per kWh decreases as energy storage capacity increases and typical projected battery costs for eight or more hours of storage are as low as US$150/kWh.¹²⁷

Since 1993 a number of field trials of the vanadium battery were been undertaken both by UNSW as well as in Thailand and Japan. In collaboration with UNSW Centre for Photovoltaic Devices and Systems and licensee Thai Gypsum Products Ltd., a vanadium battery storage system was installed in a demonstration Solar House in Thailand.^{10, 60, 61}

The solar energy system included 2.2 kW of installed solar cells and a 12 kWh vanadium battery. The original battery had 12 cells giving a system voltage of 16.8 V and used 200 l of each half-cell electrolyte in the two reservoirs.⁶⁰ A 48 V, 36 cell stack was later constructed in the laboratory and tested with a 4 kW inverter and specially designed battery controller, prior to installation in the demonstration Solar House in Thailand. The 48 V batteries replaced the original 12-cell stack for long-term field trials. The microprocessor controller built by the UNSW Centre for Photovoltaic Systems and Devices, was designed to optimise the efficiency of the battery for this application.^{60, 127} This included the use of a pump control system that only switched on the pumps if the current exceeded a pre-set value. At lower loads, the pumps would only turn on for a few minutes at a time when the stack voltage dropped below 1 V, thereby allowing the electrolyte within the cell stack to be replenished. This simple on-off pump controller significantly reduced the pumping energy losses so as to maximise the overall energy efficiency of the system.

In 1993, a consortium comprising Mitsubishi Chemicals and Kashima-Kita Power Corporation of Japan licensed the UNSW vanadium battery technology for stationary uses and for the next 5–6 years spent several million US dollars per annum to scale up the technology for large-scale load-levelling and solar energy storage applications.^{10, 127} Kashima-Kita Electric Power Corporation employs vanadium rich Venezuelan pitch as the fuel for electricity generation, thus producing a high vanadium content fly-ash as a waste product. An efficient chemical process was developed to extract the vanadium from the fly-ash which is then used to produce a low-cost vanadium electrolyte for the vanadium redox flow battery. A 3 m₃ d₋₁ electrolyte production plant was commissioned in early 1996.^{10, 127}

In 1997, a 200 kW/800 kWh grid-connected vanadium battery was commissioned at the Kashima-Kita Electric Power station in Japan where it underwent long-term testing as a load-levelling system. By the beginning of 1998, it had already undergone 150 charge-discharge cycles and was continuing to show high energy efficiencies of close to 80% at current densities of 80–100 mA cm₋₂.^{10, 127}

Since 1999, Sumitomo Electric Industries (SEI) in Japan has completed more than 20 medium to large VRB demonstration systems in a wide range of applications including wind energy storage, emergency back-up power and load leveling, demonstrating overall energy efficiencies as high as 80% and up to 270,000 charge-discharge cycles.^{64–66, 223}

In 2001, a vanadium energy storage system (VESS) incorporating a 250 kW/520 kWh VRB was established in South Africa²²⁵ using six 40 kW stacks produced by Sumitomo Electric Industries. Pinnacle VRB also installed a 250 kW/1 MWh system for Hydro Tasmania in Australia for wind energy storage and the replacement of diesel fuel in 2003 (Refs. ^{10, 127, 225}) while a 250 kW/2 MWh was installed in the USA in 2004 by VRB Power for voltage support and rural feeder augmentation.^{10, 225} In 2005, Sumitomo Electric Industries installed a 4 MW/6MWh system at Subaru Wind Farm in Japan for wind energy storage and wind turbine output power stabilization. The latter system was reported to give overall round trip energy efficiency of 80% with cycle life of over 270,000 cycles over 3 years of testing.²²³ In addition a vanadium battery powered electric golf cart was field tested at UNSW, using 40 l of 1.85 M all-vanadium RFB; a driving range of 17 km off-road was obtained,²³² which suggests that the energy density of an optimised all-vanadium RFB could approach that of lead-acid, with the added advantage of rapid recharging by electrolyte replacement.²² Subsequent studies with a 3 M stabilised vanadium electrolyte gave a driving range of 31.5 km with partly filled electrolyte tanks and showed that up to 54 km could be achieved if the tanks were filled to their maximum capacity.²³² Careful temperature control was however required to avoid vanadium precipitation at temperatures above 35 or below 15°C.

Early VRB stack development in China was initiated by Zhang and co-workers at the Dalian Institute of Chemical Sciences where a one kW vanadium battery stack was designed and tested in 2006. Coulombic, voltage and energy efficiencies of 85.9, 91.1 and 78.3%, respectively, were obtained at a current density of 60 mAcm₋₂, with a maximum average output power of 1.35 kW at a discharge current density of 85 mAcm₋₂.²²⁹ The 1 kW modules were subsequently integrated into a 10 kW battery with a configuration of 4 × 2 (serial × parallel) and an overall energy efficiency of more than 80%, at an average output power of 10.05 kW (current density 85 mA cm₋₂) was achieved.

In July 2009, Chinese National Grid announced the launch of the Zhangbei storage building, China's first comprehensive demonstration project, which includes 75 MW of projects in energy storage. Technologies such as all-vanadium, lithium and the Japanese sodium- sulphur batteries have been included in several demonstration projects for energy storage. However, for wind energy storage, the all-vanadium redox battery was found desirable.^225–227

A significant number of commercial VRB systems are now being delivered to customers for a wide range of applications by Prudent Energy in China⁶⁴ and Cellstrom GmbH in Austria.⁶⁷ In both companies, the focus to date has been on the manufacture of 2–5 kW power rating and these are being integrated into a range of products for small to medium-scale applications (up to 100 kW). In recent years however, a significant market for energy storage products in the MW range has been emerging, so the focus now needs to be on scale-up and production engineering to achieve the required cost structure for these markets. Although Sumitomo Electric Industries successfully engineered and demonstrated several MWh scale VRB systems based on 40–50 kW stack modules, these were custom-made and therefore too expensive for commercial implementation.

Several groups are now reporting scale-up efforts to produce 20–50 kW stack modules to address the MW-scale smart grid market.^228–231 Huamin and co-workers at the Dalian Institute of Chemical Physics and Rongke Power Co., Ltd in China, have described their 20 kW stack module that has been shown to operate at 80 mA.cm₋₂ with an overall energy efficiency of 80%.²³⁰ These stack modules have been incorporated into a 260 kW subsystem (Fig. 6) with plans to integrated these into a 5MW VRB for installation at a 30–50 MW wind farm during 2011.

Zoom In Zoom Out Reset image size

On the other hand, other developers are staying with smaller 5–10 kW stack module and integrating these into larger units off-site.⁶⁴

In 2010, the US Department of Energy funded the demonstration of a 1 MW/8MWh vanadium redox battery for load levelling trials at the Painesville Municipal Power Station in Ohio²³³ and this project will include the development of 10–20 kW stacks for mass production.

Like all redox flow cell chemistries that employ different elements in each half-cell, problems of cross contamination and solution chemistry maintenance were serious limitations for the polysulphide-bromine system that could not be addressed in small installations. For this reason, target applications were for very large utility scale projects ranging from 10 to 100 MW with 8–12 h of duration. The former Innogy Technologies using the trade name of Regenesys Ltd. developed the polysulphide-bromine redox battery for these target applications and began installation and commissioning of a 12 MW test facility at Little Barford, UK in the early 2000's.^{68, 233} Figure 7 shows the interior of the Little Barford facility showing the stream of 100 kW stacks developed by Innogy.

Zoom In Zoom Out Reset image size

The Regenesys technology had been tested at laboratory scale and was in the process of being proven at pilot plant scale. Development of the 100 kW XL module was started in parallel with full validation of the design concepts under test in the smaller reactors. Little Barford was the first demonstration of the Regenesys Technology at utility scale. The plant design was for 120 stack modules to operate with 1800 m₃ of each electrolyte. The plants intended power output was to be 12MW (peak output of 15 MW) with an energy capacity of 120 MWh. The balancing system for the Regenesys Technology was in its early days of development and was unproven at plant scale. The original concept was to move the prototype balancing system being built at the OTEF test facility to Little Barford after it had been proven at scale. The OTEF balancing system encountered many problems however, as knowledge of the chemistry improved resulting in the Regenesys system becoming more complex than first envisaged. A number of other design and commissioning problems were also encountered and the plant was never properly commissioned or tested.

In 2002, Innogy was acquired by the German multi-utility RWE group of companies and under RWE Innogy's ownership the Regenesys energy storage technology was progressed to its first full-scale demonstration plant and into the commercialisation phase. In 2003, however, RWE decided that this did not fit with RWE's core business so a decision was made to sell the technology and business. In 2004 the Regenesys²³⁵ system was acquired by VRB power systems Inc. in Canada but no further development has been undertaken to date.

Unfortunately there are still several technical issues related to the commercialization of the polysulphide-bromine redox battery.²³⁶ Firstly, the preparation cost of carbon felt-based electrodes is considerably high, while the activated carbon-based electrode demonstrates energy efficiency less than 60%. In addition, the synthesis methods of sodium polysulfide from molten sodium and sulphur or H₂S are very complex and expensive, so they are not suitable for large-scale production. Furthermore, the present commercialized cation exchange membrane does not show 100% cation selectivity, so anions can permeate through and cause crossover contamination during long term operation with the need for regular removal of sulphate from the positive half-cell solution and replenishment of sodium sulphide in the negative electrolyte.²³⁵

Several projects were undertaken by ZBB Energy Corporation to evaluate the ability of the zinc-bromine system for solar energy storage.²³⁷ In the first project, a 50 kW rooftop PV system was installed in parallel with a 50 kW/100 kWh battery system at a commercial facility in New York. The second project was a 250 kW/500 kWh utility system that was installed on a remote utility circuit in New South Wales, Australia. This system was meant to complement an existing 20 kW PV concentrator system, support the remote line, and offer enhanced reliability. The battery was charged by the solar array during the day in order to provide reliable night time power to remote area property owners.²³⁷

Two other companies that are currently commercialising the zinc-bromine battery are Premium Power based in Massachusetts²³⁸ and Redflow based in Queensland Australia.²³⁹ Each have modular units that deliver up to 500 and 30 kWh of electricity storage respectively, and are designed as integrated power generation units complete with inverters and power conditioning equipment within a transportable trailer for easy installation. However, further scale up and full commercialization of the technology is yet to become a reality possibly due to the high cost of the bromine complexing agents and problems related to zinc deposition at the cathode during charge that can lead to dendrites and short-circuiting across the separator.