Technical field The present invention relates generally to the field of regenerative fuel cell (RFC) technology. Background Art

Regenerative fuel cells, and the methods by which they are able to store and deliver electricity, have been known for many years. They are electrochemical apparatus for energy storage and power delivery. In the power delivery phase, electrochemically active species are supplied to electrodes, where they react electrochemically to produce electrochemical power. In a storage phase, electrical power is used to regenerate the electrochemically active species, which are stored. Because the electrochemically active species can be stored separately from the electrode compartments and supplied when required, the generating capacity of this equipment can be quite large.

The electrochemical reactions take place on either side of an ion transport system (such as a membrane) with selective charge carriers either being transported or exchanged by the membrane.

The fundamental chemical process in these regenerative fuel cell (RFC) systems is characterised by a chemical equation where the action proceeds in one direction in the energy storage mode of the system and in the opposite direction during the power delivery mode by the system. This chemical process can be exemplified by the following redox chemical equation, wherein the term "redox" defines reactions in which a reduction and a

complementary oxidation occur together.

+ V ⁱⁿ ( _so i) V ⁿ _(so i) + V ^V (soi) Equation 1

However the implementation of these systems in practical applications has encountered major limitations, despite what appears to be a simple direct chemical process. Practical problems have included the use of hazardous materials, poor efficiencies, system size, plugging and clogging of the flow of the electrolytes, gas formation and especially the cost of materials and the cost of equipment. These have prevented RFCs from being employed widely in industry.

There is a wide range of potential applications for energy storage technologies. Most renewable energy technologies cannot easily adjust their power output to meet fluctuating demand and therefore energy storage is important in enabling low carbon/renewable energy sources to be implemented in practice. Energy storage technologies may also be used as a remote power source, to ensure constant power supply and quality and may also be used to reduce the cost of electricity by storing energy when electricity is cheap and distributing the stored energy at peak times.

There are numerous regenerative fuel cells in various states of development, but the most developed RFC is an all-vanadium redox flow battery (VRB) that utilizes the redox reactions of Equation 1. The first study of an all-vanadium redox flow battery was performed at the University of New South Wales in 1985 (Sum, E and Skyllas-Kazacos, M., Journal of Power Sources, 1985, vol. 15 (2-3), 179-190; Sum, E., Rychcik , M. and Skyllas-Kazacos, M., Journal of Power Sources, 1985, vol. 16 (2), 85-95 ). The system developed by Skyllas- Kazacos et al utilises vanadyl sulphate dissolved in sulphuric acid in one of its half-cells. VRBs have significant advantages over other regenerative fuel cells, such as eliminating contamination issues associated with crossover between the two electrolyte compartments by using the same element in each half-cell. However, this system has a low specific energy and therefore large volumes of electrolyte are required for large-scale energy storage. Large volumes of electrolyte result in high storage costs and require the use of large amounts of expensive vanadium. The system is also characterised by low output power density. Hence larger and more cell components are needed to meet power demands, resulting in greater cost. It would be beneficial to provide a redox system that reduces or overcomes at least some of these problems and it would be especially beneficial if such a system could make use of existing equipment. Furthermore, one of the disadvantages that all regenerative fuel cells face that have a metal redox couple is that hydrogen co-evolution may occur when the metal cations are

electrochemically reduced. Most regenerative fuel cells currently in development are liquid/liquid RFCs, that is, the electrochemically active species for the anode and cathode are present in liquid anolyte and catholyte. A liquid/gas fuel cell which employs hydrogen gas and liquid bromine electrolyte has been investigated by Livshits, V et al (Electrochemistry Communications, 2006, vol. 8(8), 1358-62). Later, the hydrogen-bromine fuel cell was adapted into a RFC by EnStorage.

Although this system has demonstrated a high discharge output power, there are a number of drawbacks to using this system including a low catalyst stability and bromine gas evolution during operation. More recently WO2011/089518 has proposed a hydrogen-bromine regenerative fuel cell and also mentions using a hydrogen-iron redox system. However, due to the low standard electrochemical potential of the iron II/III redox couple (0.77 V vs SHE), the average working voltage of such a hydrogen-iron system at discharge will be even lower (indeed it is the lowest among known regenerative fuel cells) which is a significant disadvantage for practical redox battery applications. Another liquid/gas RFC which has been investigated is a vanadium/air RFC (Hosseiny, S. S., et al, Electrochemistry Communications, 2011, vol. 13, 751-754); however this system has low efficiency, low power density and poor rechargeability.

It is important to realise that regenerative fuel cells are distinct from standard fuel cells.

Standard fuel cells consume fuel and can normally only be run in a power delivery mode; they either cannot be run in an energy storage mode (in which power is stored) or, if they can, they can only do so in a highly inefficient way. Furthermore, reversing the electrochemical reaction in a fuel cell can cause permanent damage to the catalyst. Standard fuel cells are optimised for operating in the energy generating mode only while regenerative fuel cells are optimised for the combined power delivery mode and the energy storage mode. Thus only electrochemical reactions that are readily reversible can be used in a regenerative fuel cell, while in certain fuel cells (such as direct alcohol, or direct borohydride fuel cells, or hydrogen/oxygen fuel cells) the reactions need not be reversible and indeed they are usually not. Because of these considerations, regenerative fuel cells will normally use at least one different electrochemical reaction, as compared to standard fuel cells, although where a fuel cell clearly uses half cells that both use a readily reversible redox reaction, e,g, the hydrogen-I system disclosed in "Advancements in the Direct Hydrogen Redox Fuel", Electrochemical and Solid-State Letters, 11 (2) Bl 1-B15 (2008), such a system can be used both in fuel cells and regenerative fuel cells. In addition the average operating voltage during discharge is important. A low voltage system will require either a higher number of cells in electrical series to increase the voltage, or the design of bespoke power converters to deal with low voltage-high current systems, which adds both complexity and cost to the system.

Therefore finding two redox couples for use in a regenerative fuel cell that are reversible, soluble at practical concentrations ( about 1M or above), have a suitable potential difference between the standard electrode potentials (E V) of the couples and overcome the problems in the art is a challenging task.

The present invention overcomes the above problems by providing a hydrogen gas/dissolved metal ion regenerative fuel cell where the metal is selected from vanadium, cerium, manganese or their stable and electro chemically reversible aqueous complexes. Vanadium, cerium and manganese have relatively high electrochemical redox potential:

In particularly preferred embodiments, the present invention provides a hydrogen

gas/dissolved vanadium ion regenerative fuel cell. The inventive regenerative fuel cells, and especially the hydrogen/vanadium ion system, at least partly overcome the problems of the all-vanadium RFCs (VRBs) currently in use in that costs are significantly reduced by halving the amount of expensive vanadium required. Furthermore, replacing large liquid electrolyte storage tanks with compressed gas storage vessels for hydrogen dramatically reduces the amount of space taken up by the regenerative fuel cell, which further reduces costs. Other advantages include an increase in output power of the system due to the lower overpotential of the hydrogen oxidation reaction. The preferred hydrogen/vanadium RFC provides a further advantage in that existing vanadium/vanadium RFC systems can be readily adapted to replace the vanadium anode with a hydrogen anode, thereby reducing the capital cost of installing the regenerative fuel cells of the present invention by preventing the need to install an entire system. This retro-fitting aspect therefore overcomes a substantial draw-back of using new systems for those who have already invested considerable capital in existing technology and is — important aspect of the present invention. Regenerative fuel cells can generally be distinguished from fuel cells by their "plumbing". A regenerative fuel cell has both conduits for supplying fuel to the electrodes for the power delivery phase, and also conduits for conducting the spent fuel to a store so that it can be regenerated. Often the fuel will be in the form of electrolyte that is exhausted following a power delivery phase and in this case the conduits will also be arranged to conduct exhausted (or spent) electrolyte to a store and supply it back to its half cell during an energy storage mode, e.g. by the use of appropriate pumps. In contrast, fuel cells are not set up to operate in energy storage mode to electrochemically replenish exhausted electrolyte. In the case of regenerative fuel cells having a half cell containing a gas electrode, a compressor is generally provided to compress gas generated during the energy storage mode to enable it to be collected in a compressed gas storage tank for future power delivery phases (although neither a compressor nor a storage tank is required where the electrode is an oxygen electrode since oxygen is freely available from the atmosphere). In contrast a fuel cell will generally not have such a compressor.

Disclosure of the Invention

The present invention is defined in the accompanying claims.

The present invention relates to a regenerative fuel cell (RFC), that is to say an

electrochemical apparatus configured for both energy storage and power delivery and has an anolyte compartment and a catholyte compartment separated by a membrane capable of selectively passing protons, that is to say the membrane can transfer protons e.g., by being selectively permeable to protons or is a proton exchange membrane.

In accordance with standard terminology in the field of regenerative fuel cells, the terms "anode" and "cathode" are defined by the functions of the electrodes in the power delivery mode. To avoid confusion, the same terms are maintained to denote the same electrodes throughout the two modes of operation (power deliver and energy storage) of the RFC. The terms "anolyte" and "catholyte" will be used to denote the electrolyte in contact with the "anode" and "cathode". In a power delivery mode, an electrochemically active species is oxidised at the anode and an electrochemically active species is reduced at the cathode to form reacted (or "spent") species. In the energy storage mode, electrochemical system is reversed and the "spent" cathode species is electrochemically oxidised at the cathode to regenerate the corresponding electrochemically active species. It will be appreciated that the redox reaction at the hydrogen gas anode will not produce any "spent" species in the power delivery mode as the

electrochemically active species (hydrogen gas) is transformed into protons that are dissolved in the electrolyte. Protons are selectively passed by the membrane separating the anode and cathode compartments from the anode side of the membrane to the cathode side of the membrane. The electrons produced during the oxidation of the hydrogen gas at the anode during the power delivery mode are collected by a current collector. However, any unreacted hydrogen gas may be transferred away from the anode compartment by one or more conduits and returned to a gas source vessel (which may be pressurised or unpressurised). In the energy storage mode, protons are selectively passed by the membrane separating the anode and cathode compartments from the cathode side of the membrane to the anode side of the membrane and protons are reduced at the anode to regenerate the hydrogen gas, which forms the anode electrochemically active species.

Therefore, in accordance with the present invention there is provided a regenerative fuel cell that is capable of operating in a power delivery mode in which it generates electrical power by the reaction of an electrochemically active species at an anode and the reaction of a different electrochemically active species at a cathode and that is also capable of operating in an energy storage mode in which it consumes electrical power to generate said electrochemically active species, the cell comprising:

• a reversible hydrogen gas anode, in an anode compartment;

• a reversible cathode in a cathode compartment;

• a membrane separating the anode compartment from the cathode compartment, which membrane is capable of selectively passing protons;

• conduits configured to supply electrochemically active species to the anode and to the cathode in said power delivery mode, and to carry generated electrochemically active species away from the anode and away from the cathode in said energy storage mode;

wherein the redox reaction at the anode is: 2Ff + 2e ^~ H ₂ (gas)

and the redox reaction at the cathode is one selected from:

(i) V ⁵⁺ + e V ⁴⁺ ,

(ii) Ce ⁴⁺ + e ^" Ce ³⁺ , and

(iii) Mn ³⁺ + e ^" Mn ²⁺

For the cathode side of the regenerative fuel cell, the electrochemically active species is present in liquid catholyte supplied to the catholyte electrode compartment. In respect of the anode, the electrochemically active species is hydrogen and the anolyte electrode is a gas permeable electrode.

The electrochemically active species present in the catholyte may be stored in a first vessel ready for passing to the cathode compartment in the power delivery mode. For the anode, the electrochemically active species is hydrogen gas, and will generally be in the form of a pressurised gas source.

During the power delivery mode, the spent catholyte is collected in a vessel, which may be the first vessel or a separate second vessel. During the energy storage mode, the spent electrolyte may be supplied from the vessel to the cathode compartment, where the electrochemically active species is regenerated. The first and second vessels (if provided) may be different compartments of a single container.

The regenerative fuel cell may additionally include a pump which allows the liquid catholyte to be circulated through the conduits between the storage vessel and the cathode

compartment. As described above, the regenerative fuel cell may include a compressor that allows the hydrogen gas to be stored at pressure in a source vessel exterior to the

electrochemical cell. The system can operate without a compressor, provided the gas storage tank is sufficiently large to accommodate the generated gas. For example, 100 m of 1M of vanadium electrolyte requires around 1200, 120 and 12 m storage tanks for hydrogen at 1, 10 and 100 bar pressure respectively. The regenerative fuel cell may comprise a means for circulating the hydrogen gas through the conduits between the storage vessel and the anode compartment, e.g. a pump or a fan. The regenerative fuel cell may also additionally include a dryer which dries the hydrogen gas before it is stored in the source vessel. The regenerative fuel cell may also be equipped with a hydrogen expander-generator to deliver electricity as a result of compressed gas expansion.

The electrochemical reactions may take place at a discrete anode and cathode or, particularly in the case of the gas anode, they will generally take place at the gas separation

membrane/catalysed porous gas electrode interface and so it may not always be easy to identify a discrete anode and cathode and the main manifestations of the anode and the cathode may simply be the anodic and cathodic current collectors, which facilitate the supply of electrons to an electrode from an external circuit and the removal of electrons from an electrode to the external circuit (in the power delivery mode, the anodic current collector will transfer electrons away from the anode to an external circuit, and the cathodic current collector will supply electrons from the external circuit to the cathode. In the energy storage mode, this will be reversed). The anode is a porous gas electrode and the cathode may be a porous or non-porous electrode, although porous electrodes are preferred. Examples of suitable electrodes are well known in the art. Catalysed porous carbon electrodes are particularly preferred in the present invention, especially when the electrochemically active species present in the cathode half cell is V ⁵⁺ (and the respective "spent" species is V ⁴⁺ )., Examples of catalysed porous carbon electrodes include catalysed carbon paper, cloth, felt and composites. In certain embodiments, the cathode comprises one or more catalysed carbon papers. The carbon may be graphitic, amorphous, or have glassy structure. In a particularly preferred embodiment of the invention, the anode is a catalysed electrode and the cathode is a non-catalysed electrode. Examples of other suitable electrodes include corrosion resistant metals (or metal alloys), such as titanium or alloys thereof, in form of meshes felts or foams. As the redox reaction at the cathode does not usually require catalysis, having a regenerative fuel cell whereby only one of the electrodes is catalysed may allow the production costs of the regenerative fuel cell to be significantly reduced; it is possible, but not necessary, to use some non-noble metal catalyst and this would also keep the cost down as compared to the use of noble metal catalysts. The catalyst used in the anode may be of noble metals such as for example platinum, palladium, iridium, ruthenium, rhenium, rhodium, osmium or combinations thereof, including alloys for example a platinum/ruthenium alloy or binary catalyst such as PtCo, PtNi, PtMo etc. or ternary catalyst PtRuMo, PtRuSn, PtRuW etc. or chalcogenides/oxides as RuSe, Pt-MoOx etc. Snme binary/ternary or other than pure precious metal catalysts can be more tolerant to probable catalytic poisoning as results of catholyte species crossover. The amount of precious metal can be significantly reduced as compared to normal catalysed fuel cell gas electrodes since RFCs are not required to operate at the same high power mode as conventional fuel cells.

The electrochemically active species present in the anode half cell is hydrogen gas. Therefore, the redox reaction which takes place at the anode is:

2H ⁺ + 2e ^~ H ₂ (gas)

The hydrogen gas is stored externally to the anode compartment in a container, which may be a pressurised gas source vessel. The hydrogen gas may be supplied to the anode compartment by one or more conduits in the power delivery mode and may be carried away from the anode compartment by one or more conduits in the energy storage mode.

The electrochemically active species present in the cathode half cell may be one of V ⁵⁺ , Ce ⁴⁺ or Mn ³⁺ , and the respective "spent" species are V ⁴⁺ , Ce ³⁺ or Mn ²⁺ . Therefore the redox reaction which takes place at the cathode is selected from:

(i) V ⁵⁺ + e V ⁴⁺ ,

(ii) Ce ⁴⁺ + e ^" Ce ³⁺ , and

(iii) Mn ³⁺ + e Mn ²⁺ .

In a preferred embodiment, the redox reaction which takes place at the cathode is:

V ⁵⁺ + e —— V ⁴⁺

The Regenerative Fuel Cells of the present invention provide the advantages /overcome the problems discussed above, most notably (a) the reduction of costs as compared to the all- vanadium RFCs (VRBs) currently in use by halving the amount of expensive metal ions, and especially vanadium, required, (b) a reduction in the amount of space taken up by the regenerative fuel cell resulting from the replacement of large liquid electrolyte storage tanks in VRBs with compressed gas storage vessels for hydrogen, (c) an increase in output power of the system due to the lower overpotential of the hydrogen oxidation reaction and (d) in the case the preferred hydrogen/vanadium RFC, it can be retro-fitted to existing VRBs which reduces the capital investment needed to replace VRBs.

It will be appreciated that although the electrochemically active species present in the cathode half cell is referred to as a free cation e.g. M ⁿ⁺ , it may be present in the catholyte solution as any stable positively charged complex, for example, an oxide complex, such as V0 ₂ ⁺ and 2_ _| _

VO . The nature of the positively charged complex will depend on the materials used to produce the catholyte solution. For example, when the electrochemically active species present in the cathode half cell is vanadium, the liquid catholyte may be prepared using tetravalent vanadium oxide (V0 ₂ ), vanadyl sulphate (VOSO ₄ ) or pentavalent vanadium oxide (V2O5), Ce ^m 2(S0 ₄ )3/ Ce ^IY (S0 ₄ ) ₂ ; MnS0 ₄ . The electrolytes will generally be aqueous.

The electrochemically active species in the cathode half cell is present in liquid electrolyte. Acidic electrolytes are well known in the art and any standard acidic electrolytes may be used in accordance with the present invention. Preferred electrolytes include sulphuric acid, which may be an aqueous solution of concentrated sulphuric acid, methanesulfonic acid (MSA) or trifluoromethanesulfonic acid (TFSA), or mixtures thereof, most preferably sulphuric acid. Due to the high electrochemical potential of redox couples such Ce and Mn, the use of organic acid electrolyte will be preferable in order to minimise oxygen generation during energy storage mode (charging). The use of any other strong acid is not prohibited if the acid can form soluble metal cations but not reduce or oxidise the catholyte. The concentration of the electrochemically active species in the catholyte determines the power and energy density of the regenerative fuel cell. Therefore, the concentration of electrochemically active species in the catholyte is preferably at least 0.2M, and more preferably greater than 0.5M, e.g.

greater than 1M. The maximum practical concentration of the electrochemically active species will generally be governed by its solubility in the electrolyte as precipitation from the electrolyte becomes an increasing problem at higher concentrations, and the presence of precipitated materials in the cell is preferably avoided since it interferes with the flow of the electrolyte and the functioning of the regenerative fuel cell in question. The membrane separating the anode compartment from the cathode compartment is a membrane capable of selectively passing protons (hydrogen ions), which means that the membrane may be a proton exchange membrane or a membrane which is permeable to protons. The membrane may be a porous separator, such as a microporous membrane.

Alternatively, the membrane may be a hybrid of both cation and anion conductors. Preferably, the membrane is a proton exchange membrane. Proton exchange membranes are well known in the art, for example, the Nafion™ ion exchange membrane produced by DuPont. Although the Nafion™ membrane has good proton conductivity and good chemical stability, it has a number of disadvantages including a high permeability to vanadium cations and high cost. Therefore, in other preferred embodiments, the membrane is substantially impermeable to metal cations, for example vanadium, cerium and manganese cations.

Brief Description of Drawings

Figure 1 is a schematic sectional view of a liquid/gas regenerative fuel cell of the invention (the terms "liquid" and "gas" denoting the phases of the electrochemically active material supplied to the two electrodes).

Figure 2 is a charge/discharge curve for a vanadium/hydrogen regenerative fuel cell, using 0.23M vanadium.

Figure 3A shows charge/discharge curves for a vanadium/hydrogen regenerative fuel cell, using 0.23M vanadium at variable currently densities. Figure 3B shows charge/discharge curves for a vanadium/hydrogen regenerative fuel cell, using 0.23M vanadium at variable flow rates

Figure 4A shows a power density curve for a vanadium/hydrogen regenerative fuel cell, using 0.23M vanadium at a flow rate of 50 ml/min.

Figure 4B shows a power density curve for a vanadium/hydrogen regenerative fuel cell, using 0.23M vanadium at a flow rate of 100 ml/min.

Figure 5 shows a power density curve for a vanadium/hydrogen regenerative fuel cell, using 1.0M vanadium at a flow rate of 100 ml/min; the graph shows a combined I-V curve (left) and a power curve (right).

Figure 6 shows the performance of a vanadium/hydrogen regenerative fuel cell with untreated carbon papers in 1 M of V(IV)/V(V) solution at 100 % SOC and at variable flow rates.

Figure 7 shows charge/discharge curves for a vanadium/hydrogen regenerative fuel cell, using untreated carbon papers in 1 M of V(IV)/V(V) at 50 ml/min hydrogen flow rate, 100 ml/min vanadium flow rate and 40 mA/cm current density. Description of Embodiments and Examples

Figure 1 shows a schematic of a regenerative fuel cell in which the electrochemically active materials used to generate power are (a) hydrogen gas (supplied to the anode) and (b) dissolved metallic ions in a liquid catholyte (supplied to the cathode).

In the power delivery mode, the liquid catholyte containing the electrochemically active species M ⁿ⁺¹ is pumped by a pump (11) from a compartment of a catholyte storage container (12A), through a conduit (12B) and into the catholyte compartment (9), where it is reduced at a cathode (2) according to the half reaction:

M ⁿ⁺¹ + e ^" _► M ⁿ

The catholyte containing the spent electrolyte species M ⁿ is then carried away from the catholyte compartment through a second conduit (1) to the catholyte storage container (12A), where it is stored in a compartment separate from the fresh catholyte compartment.

The anode and at least part of the anolyte compartment (8) are formed by a porous gas flow electrode and hydrogen is supplied from a pressurised gas source vessel (7) through a conduit (13), to the anode / anode compartment (8), where the hydrogen is oxidised to protons (H ⁺ ) according to the half reaction:

H ₂ ► 2H ⁺ + 2e ^"

and the current is collected by a current collector (4). A proton exchange membrane (3) separates the anolyte and catholyte compartments (8 & 9) and selectively passes the protons from the anolyte to the catholyte side of the membrane (3) to balance the charge, thereby completing the electrical circuit. Any unreacted hydrogen is carried away from the anolyte compartment (8) by a second conduit (5) and returned to the pressurised gas source vessel (7).

In the energy storage mode, the system is reversed so that the electrochemically active species M ⁿ is pumped from the catholyte storage container (12A), through the conduit (1) to the catholyte compartment (9), where the spent electrolyte species M ⁿ is oxidised at the cathode (2) to form the electrochemically active species M ⁿ⁺¹ . The resulting regenerated electrolyte is transferred away from the catholyte container (9) by the pump (11), through the second conduit (12B) to the catholyte storage container (12A). Meanwhile, protons at the anolyte side of the proton exchange membrane (3) are catalytically reduced at the porous gas anode (4) to hydrogen gas; the hydrogen is transferred away from the porous anode (4) through the conduit (5) and compressed by the compressor (6) before being stored in the pressurised gas source vessel (7). It will be appreciated that the species M ⁿ⁺¹ /M ⁿ may be any of V ⁵⁺ /V ⁴⁺ , Ce ⁴⁺ /Ce ³⁺ , Mn ³⁺ /Mn ²⁺ , preferably V ⁵⁺ /V ⁴⁺ .

Examples Example 1

A 25 cm active area regenerative vanadium fuel cell (RVFC) was produced and tested as set out below. Details of the particular components used in the above cell are set out below: Serpentine flow channel plates were CNC-machined from highly conductive polymer composite Electrophen (Bac2) purposely produced for fuel cell bipolar plate applications. Plastic end plates were cut from PTFE sheet and were sandwiched between aluminium plates to guarantee even pressure distribution across the cell. A commercially available HiSPEC™ M-200 Class MEAs (Johnson Matthey Fuel Cells) with 25 cm of catalytic active area were used in all experiments [Johnson Matthey http://wwwjmfuelcells.com/products/measl. To minimise iR losses and provide good electric conductivity gold plated copper current collectors were inserted between end and flow channel plates. Finally, Viton rubber gaskets were placed between different cell components to ensure the system be properly sealed during the operation.

Preparation of Vanadium Sulphate Solution

Vanadium catholyte solutions of 0.23M and 1M were prepared by dissolving corresponding amounts of vanadyl sulphate (Sigma-Aldrich) in 5M concentrated sulphuric acid. A Masterflex easy-load peristaltic pump and Masterflex Chem-Durance tubing were used to pump the vanadium catholyte through the cell. Typically VRBs use vanadium concentrations in the order of 1.5M - 2.0M. However, a 0.23M solution was chosen as it would suffice to give an initial indication of the cells performance. Using a lower concentration also allows for shorter charge and discharge times at low current densities. This solution was used for all experiments, except where indicated otherwise.

Charge and Discharge Cycle

Standard cycle

The following procedure details the standard steps taken when performing a charge/discharge cycle. The vanadium and hydrogen flow rates remained constant throughout the procedure.

1. The system was discharged to a target voltage of 0.45V using the current density at which the cycle was to be performed. If the system's state of charge (SOC) was below this target, the system was charged to a point above the target SOC and then discharged to 0.45V.

2. The open circuit voltage (OCV) of the system was measured for 5 minutes.

3. The system was charged at the desired current density until the upper voltage cut-off limit was reached.

4. The OCV of the system was measured for 5 minutes.

5. The system was discharged at the desired current density until the lower voltage cutoff limit was reached.

6. The OCV of the system was measured for 5 minutes.

Cycle between set capacities

1. The system was discharged to a target voltage of 0.45V using the current density at which the cycle was to be performed. If the system's SOC was below this target the system was charged to a point above the target SOC and then discharged to 0.45V.

2. The time t (in seconds) required to reach the desired capacity for a particular current is calculates using equation 1.1. n · F · C · V Q

^{1 ~} I ^~ 3.6 - 1 (1.1)

Where n- number of electrons, F- Faraday number, C- species concentration [mole/L],

V- total solution volume [L], I- current [A], Q- capacity [mAh]

3. The OCV of the system was measured for 5 minutes.

4. The system was charged at the desired current density for the time calculated in step 2 or until the upper voltage cut-off limit was reached.

5. The OCV of the system was measured for 5 minutes. 6. The system was discharged at the desired current density for the time calculated in step 2 or until the lower voltage cut-off limit was reached.

7. The OCV of the system was measured for 5 minutes. The upper voltage cut-off of 1.7V was selected due to the risk of corrosion to the carbon paper at voltages above 1.9V (Joerissen et al, Journal of Power Sciences, 2004, vol. 127(1-2), 98-104) and excessive oxygen evolution due to water oxidation. The lower voltage cut-off of 0.45V was selected due to the reduction of V(IV) ions to V(III) ions at voltages below 0.35V (Pourbaix diagram at pH of 0).

System testing

Charge/discharge cycle at 0.23M V(V)/VQV)

A number of charge/discharge cycles were performed to analyse the behaviour of the RVFC.

2

Figure 2 shows the charge discharge behaviour at a current density of 8mA/cm . The system gives a smooth charge/discharge curve. The open circuit potential (OCV) for the vanadium/hydrogen system can be calculated from the Nernst equation (Equation 1.2).

*cell ZCBII ^m [vfli]EH ⁺ 3P¾ ^S ( ^L2 >

The reactions in the power delivery mode are:

Anode: H ₂ (g)→ 2H ⁺ + 2e .

Cathode: 2V0 ₂ ⁺ + 4H ⁺ + 2e ^~ →2V0 ²⁺ + 2H ₂ 0.

Total: H ₂ (g) + 2V0 ₂ ⁺ + 2H ⁺ → 2V0 ²⁺ + 2H ₂ 0.

The OCV at the charged state for the cycle in figure 2 is 1.099V. Using the Nernst equation, the percentage of V(V) species over V(IV) species was calculated as 99%. A higher concentration of V(V) species would be expected at the charged state and the high OCV would indicate a very high SOC. The OCV following discharge is 0.9213V which corresponds to a V(V) concentration of 7%. The decreased V(V) concentration is expected as the solution is discharged, and the low concentration would indicate a deep discharge. Charge/discharge cycle at 0.23M V(V)/V(TV) at different current densities

The system was charged to a target capacity of 264mAh, as this represented a 36% utilisation of the maximum theoretical capacity of 740mAh. The hydrogen and vanadium flowrates were maintained at 50mL/min throughout each cycle (Figure 3 A). From the results it was observed that the overpotential for both charge and discharge steps increased as the current density increased. Overpotentials are due to losses associated with ohmic resistance, charge transfer and mass transport phenomena. At the start of each charge/discharge step the losses are most likely predominated by charge transfer processes while at the end it is mostly mass transport limitations that contribute to the overpotential.

Charge/discharge cycle at 0.23M V(V)/V(TV) at different flow rates

System parameters that affect the utilisation of redox species are current density, voltage window and electrolyte flow rate. It has been observed that as the current density increases, utilisation of electrolyte decreases. Current density also determines the operating power of the system. There is a trade -off between electrolyte utilisation and power output. The effect of vanadium flow rate on capacity utilisation was studied at a current density of 14mA/cm2 (Figure 3B). The maximum power point at 75% SOC was chosen as it is assumed to be a good representative point for the system.

At a vanadium flow rate of 50mL/min, the system had a capacity utilisation of 16.2% (: table 1).

Decreasing the flow rate to 30mL/min resulted in a decrease in capacity utilisation to 9.5%. However, increasing the flow rate to 70mL/min resulted in an increase in the capacity utilisation to 31.6%. A 40% decrease in flow rate yielded a 41% decrease in the utilisation, while a 40% increase yielded a 95% increase in the utilisation. This shows the importance of mass transport and concentration polarisation losses to the system. Further increasing the flow rate to lOOmL/min marginally increased the capacity utilisation to 32.7%, and achieved higher efficiencies.

Table 1 : Capacity utilisation for a wide range of vanadium flow rates

Cycle Number [-] _m ^7 _\%] - 30mL/min 9.5

3 - 50mL/min 16.2

- 70mL/min 31.6

- lOOmL/min 32.8

Power density for 0.23M VCVWflV at a flow rate of 50 ml/min

A power curve was measured for the system at a vanadium flow rate of 50mL/min and at three states of charge (SOC) (Figure 4A). For an SOC of 75% the system reached a maximum power point of 7.7m W/cm 2 at a current density of 14mA/cm 2.

Power density for 0.23M VfVyVQV) at a flow rate of 100 ml/min

The power curve was measured for a flow rate of lOOmL/min and at three different SOCs (Figure 4B). At 100% SOC a maximum power point of 13.3mW/cm was reached at a current density of 24mA/cm 2. At an SOC of 75% the maximum power reached was 10.7m W/cm 2 at a current density of 21mA/cm . This represents a 39% increase in the maximum power when compared with the power reached at a flow rate of 50mL/min. Operating at higher vanadium flow rate improves system performance by reducing mass transport limitation and increasing capacity utilisation.

Power density for 1.0 M VfWVflV) at a flow rate of 100 ml/min

The concentration of vanadium was increased from 0.23M to 1.0M in order to investigate the effect of increasing the vanadium concentration, an important consideration for large-scale systems. A typical large-scale VRB system would have a vanadium concentration of 1.5 - 2.0M. Solubility and precipitation of V ⁵⁺ becomes an issue at concentrations above 2.0M. A 1.0M solution was chosen as it is a substantial increase in the concentration from the original system and brought it closer to the 1.5M concentration that would be used at large-scale.

The power curve for the 1.0M system was measured at a vanadium flow rate of lOOmL/min and at SOCs of 100%, 75% and 50% (Figure 5). The maximum power reached at 100% SOC was 54.3mW/cm 2 at a current density of 84mA/cm 2. The maximum power point for an SOC of 75% was also at a current density of 84mA/cm 2 with a specific power of 48.4m W/cm 2. If the maximum power point determined at 75% SOC for the 0.23M system is scaled up linearly to 1.0M, a power of 46.5mW/cm 2 at a current density of 91mA/cm 2 would be achieved. This indicates a remarkably linear relationship between maximum power and vanadium concentration. In this work we have found that our non-optimised RVFC has demonstrated surprisingly high power density comparing to current commercial VRB technology. For example the discharge power density of common VRB system with 2M vanadium solution and average working voltage of 1.3 V is around 65 mW/cm (Rychcik , M. and Skyllas- Kazacos, M., Journal of Power Sources, 1988, vol. 22, 56-67) or 32 mW/cm for 1 M vanadium solution which is 40 % less than that of our non-optimised RVFC. This result further reinforces the potential of an optimised RVFC system in terms of power output as compared with the current state-of-the-art all-vanadium systems. An energy efficiency of 30.9% was achieved for the 1.0M system, when run at a current density of 72mA/cm (just below the maximum power point) and vanadium flow rate of lOOmL/min. This compares with an energy efficiency of 33.5% for the 0.23M system under similar conditions (Table 2). The 1.0M solution has lower overall energy efficiency primarily due to its lower voltage efficiency; however, it does achieve higher coulombic efficiencies.

Table 2: Comparison of efficiencies for the 0.23M and 1.0M systems.

Vanadium Efficiency

Concentration Coulombic Voltage Energy

[M] [%] [%] [%]

0.23M 74.2 45.1 33.5

1.0M 86.2 35.9 30.9 Example 2

In attempt to improve the performance of the hydrogen/vanadium regenerative fuel cell set out in Example 1 , the HiSPEC M-200 Class MEA was replaced by commercial GDL

(available from Johnson Matthey, see

http://www.alfa om/eri/GP100w.pgm?DSSTK=045357 with a Pt loading of 0.5 mg/cm for the hydrogen side, a NAFION 117 membrane, and two as-received (untreated) Toray carbon papers for the vanadium side (available from Johnson Matthey, see

http://ww¾ ^r .al.fa.co ^, m/en,/GP100w.pgm?DSSTIv=045356). The performance of the system with two untreated carbon papers is shown in Figure 6. At 100 ml/min flow rate the regenerative fuel cell achieved a peak power density of 106.7 mW/cm 2 , almost double of the peak power density for the previous version (54.3 mW/cm 2 ). Increasing the flow rate to 150 and 200 ml/min only slightly improved the performance further (113 and 114 mW/cm respectively), indicating that mass transport was not a significant contributor to voltage loss under these conditions.

Figure 7 demonstrates the charge/discharge characteristics of the hydrogen/vanadium regenerative fuel cell of Example 2. In comparison to the performance of the cell with the commercial MEA configuration (Example 1), the efficiencies were improved and utilisation of the 1M vanadium solution was higher. The voltage efficiency was 69%. The couloumbic efficiency was 87%. The energy or overall efficiency was 60 %. The capacity utilisation was 70% (initial capacity of 1M solution was 3127 mAh, the discharge capacity was 2190 mAh).

The improvement appears to have been achieved due to two main factors: no PTFE treated carbon papers were used thus the wettability of the carbon electrodes was enhanced.

Furthermore, the use of two carbon papers instead of one increased the total volume for bulk electrochemical redox reaction for vanadium.