TECHNICAL FIELD

[0001] The present disclosure relates to a membrane and a Redox Flow Battery (RFB) comprising such a membrane.

BACKGROUND

[0002] In a Redox Flow Battery (RFB), the active substances are circulated in two electrolyte flows (oxidant in posolyte, reductant in negolyte). Although RFBs have lower energy densities and rate capabilities than the state-of-the-art lithium-ion batteries, flow batteries store energy at a lower cost. An RFB can be designed according to needs directly from the choice of hardware since power and energy are decoupled, which is not the case with solid electrode batteries. The energy stored can be increased simply by increasing the volume of the storage tanks that contain the posolyte and negolyte, respectively. The power can be increased by increasing the accessible current collector area, for example by increasing the number of electrodes.

[0003] To minimize the volume of the storage tanks, while maximizing the energy density, a high concentration of reactants in solution is needed. Present-day chemistry used in the RFB exhibits solubilities of less than 2 M for vanadium-based redox chemistry (1.26 V; 25-40 Wh/kg) in sulfuric acid solution of 2-4 M. RFB technology based on vanadium electrochemistry in highly acidic medium is problematic for its cost, corrosion and environmental issues.

[0004] To improve the sustainability, aqueous organic RFB (AORFB) have been developed, using quinone electrochemistry based on proton-coupled electron transfer reactions. EP 3 447836 discloses the use of porous polymer electrodes for quinone redox reactions in a flow battery.

[0005] A distinguishing feature of RFB technology is the membrane which separates the posolyte and negolyte from each other. In addition to chemical and mechanical stabilities, an RFB membrane should possess high ionic conductivity for the compensational ions and at the same time preventing crossover of the redox components of the posolyte and the negolyte. In vanadium RFBs, the most common membrane material is Nafion™, primarily due to its high chemical and mechanical stability and high proton conductivity. However, Nafion™ exhibits poor barrier properties against crossover of the redox components in an aqueous environment. Further, Nafion™ is relatively expensive and, being a fluorosulfonic polymer, is potentially harmful to the environment.

[0006] Conventional dialysis membranes allow for size selection but without ion selectivity, also resulting in leakage of the redox components through the membrane when used in an RFB.

SUMMARY

[0007] It is an objective of the present invention to provide an RFB membrane with improved sustainability and reduced leakage of redox components (also called electroactive elements) in, especially aqueous, electrolytes.

[0008] According to an aspect of the present invention, there is provided an RFB comprising a posolyte, a negolyte, and a membrane made of a crosslinked cellulose nanomaterial, the membrane separating the posolyte and the negolyte from each other.

[0009] According to another aspect of the present invention, there is provided a membrane comprising a crosslinked cellulose nanomaterial, wherein the cellulose nanomaterial is chemically modified to exhibit, on surfaces of the nanomaterial, aldehyde groups within the range of 0.5-1 mmol/g dry nanomaterial, preferably within the range of 0.7-0.9 mmol/g, and sulfo groups in an amount providing a charge within the range of 200-1500 pmol/g dry nanomaterial, preferably within the range of 300- 1400 pmol/g, wherein at least some of the aldehyde groups have reacted with respective hydroxy groups on the surfaces to form crosslinks. The membrane may be for use in an RFB, e.g. in accordance with the previous aspect of the invention.

[0010] According to another aspect of the present invention, there is provided a method of preparing a membrane. The method comprises oxidation of cellulose fibres by means of sodium metaperiodate (NaIO ₄ ) to provide aldehyde groups on surfaces of cellulose nanofibrils (CNF) within fibre walls of the fibres. The method also comprises sulfonation of some of the aldehyde groups, e.g. by means of sodium metabisulfite (Na ₂ S ₂ O ₅ , which in solution forms the active reactant NaHSO ₃ which may take part in the sulfonation) to provide sulfo groups on the surfaces of the CNF, thus obtaining chemically modified CNF within the fibre walls. The method also comprises obtaining the chemically modified CNF in free form from the fibres by means of processing of the fibres, e.g. comprising mechanical processing by means of high-pressure homogenization. The method also comprises forming the membrane by dewatering an aqueous slurry comprising the separated chemically modified CNF, e.g. in a mould or by spraying onto a dewatering wire, preferably in a mould, e.g. by means of vacuum filtration or evaporation e.g. at elevated temperature.

[0011] By crosslinking the cellulose nanomaterial, the membrane can have improved stability in water or other solvent used in the posolyte and negolyte, preventing the membrane from disintegrating into a dispersion. The membrane can thus also retain its functionality and dimensions over time during use and the membrane can be more mechanically robust. The cellulose nanomaterial can also be chemically modified beyond the solubility limit of the modified cellulose nanomaterial. It has been found that such a crosslinked cellulose nanomaterial membrane can advantageously be used in e.g. an RFB or fuel cell.

[0012] According to an alternative the cellulose nanomaterial of the RBF or membrane is further chemically modified to associate with, exhibit reaction products of, or adsorb any one of inorganic, organic acids, salts, molecular compounds, oD, 1D,2D or 3D nanoparticles, metalorganic frameworks, hetero polyacids, molecular complexes, monomers or polymers, or a combination thereof. By the expression “exhibit reaction products of” is meant that the cellulose nanomaterial is brought into contact with a liquid or gaseous media containing the species mentioned and reacts with said species.

[0013] Membrane property enhancing modification using for instance sulfuric acid decreased swelling ratio (SR) from about 340 to 10 % yielding an ionic conductivity (IC) of 22 mS cm-i. Using for instance tungsten phosphoric acid as IC-booster in a modification process called HPW-mod-T2 which resulted in a SR at 54 % while increasing IC to 44 mS cm-i at the cost of decreased mechanical integrity. RFB- performance comparison between HPW-mod-T2 and Nafion™ 212 in a single cell VRFB suggests that the CNF membrane is not as chemical stable in these environments but outperforms Nafion™ in terms of coulombic efficiency (98 vs 96 % respectively), vanadium crossover (5.57*10-8 vs 6.82*10-7 cm2 min-i) and self-discharge (161 vs no h). Also, HPW-mod-T2 displays commercially relavant voltage and energy efficiencies (84 and 82 % respectively).

[0014] It is to be noted that any feature of any of the aspects may be applied to any other aspect, wherever appropriate. Likewise, any advantage of any of the aspects may apply to any of the other aspects. Other objectives, features and advantages of the enclosed embodiments will be apparent from the following detailed disclosure, from the attached dependent claims as well as from the drawings.

[0015] Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in the technical field, unless explicitly defined otherwise herein. All references to "a/an/the element, apparatus, component, means, step, etc." are to be interpreted openly as referring to at least one instance of the element, apparatus, component, means, step, etc., unless explicitly stated otherwise. The steps of any method disclosed herein do not have to be performed in the exact order disclosed, unless explicitly stated. The use of “first”, “second” etc. for different features/ components of the present disclosure are only intended to distinguish the features/ components from other similar features/ components and not to impart any order or hierarchy to the features/ components.

BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Embodiments will be described, by way of example, with reference to the accompanying drawings, in which:

Fig 1 is a schematic sectional side view of an RFB in accordance with some embodiments of the present invention.

Fig 2 is a schematic flow chart of a method in accordance with some embodiments of the present invention.

Figs 3a and 3b are a schematic illustration two methods of membrane modifications.

Fig 4 are cross-sectional SEM-images of modified membranes.

Fig 5 is the FT-IR spectrum of three different membranes.

Fig 6 is the XRD-pattern of three different membranes.

Fig 7 is the discharge capacities of different membranes.

Fig 8 is the discharge capacities of different membranes.

Fig 9 is a comparison of different membranes and discharge capacities. DETAILED DESCRIPTION

[0017] Embodiments will now be described more fully hereinafter with reference to the accompanying drawings, in which certain embodiments are shown. However, other embodiments in many different forms are possible within the scope of the present disclosure. Rather, the following embodiments are provided by way of example so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art. Like numbers refer to like elements throughout the description.

[0018] As used herein, and in accordance with the TAPPI document “Roadmap for the Development of International Standards for Nanocellulose” of October 24, 2011, as well as the ISO TC 229 protocols, the term “cellulose nanomaterials” includes plant- based nanocellulose as well as nanocellulose from bacteria, algae, tunicates, and other sources. Examples of nanocellulose, and thus of cellulose nanomaterials, include herein mentioned cellulose nanofibrils (CNF), cellulose nanocrystals (CNC) and cellulose microfibrils (CMF).

[0019] Figure 1 illustrates an RFB 10 comprising a negolyte 2 (also known as catholyte) in a negolyte tank 6 and a posolyte 3 (also known as anolyte) in a posolyte tank 7. Each of the negolyte 2 and the posolyte 3 is pumped or otherwise allowed to circulate (e.g. as indicated by the arrows in the figure) past or otherwise in contact with respective electrodes 4 and 5 separated by a membrane 1, which membrane allows ion transport between the posolyte and negolyte through the membrane while an electrical current is induced between the electrodes (as schematically indicated by the “+” and “-“ signs in the figure). Thus, the membrane allows ion transport through the membrane while preventing the redox component(s) comprised in the negolyte 2 and posolyte 3, respectively, from passing through the membrane.

[0020] In accordance with the present invention, the membrane 1 is made of a crosslinked cellulose nanomaterial. In some embodiments, the cellulose nanomaterial of the membrane 1 comprises or consists of CNF, CNC, and/or CMF. In some preferred embodiments, the cellulose nanomaterial of the membrane 1 comprises or consists of CNF. Typically the cellulose nanomaterial consists of chemically modified CNF.

[0021] Preferably, the crosslinked cellulose nanomaterial is chemically modified to present a negatively charged environment in pores of the membrane, for allowing ion transport of positively charged ions through the membrane while preventing negatively charged or amphiphilic compounds from passing through the membrane.

[0022] In some embodiments of the present invention, the cellulose nanomaterial is chemically modified to exhibit, on surfaces of the nanomaterial, especially within pores of the membrane, aldehyde groups [-CHO] within the range of 0.5-1 mmol/g dry nanomaterial, preferably within the range of 0.7-0.9 mmol/g, as determined by titration. For instance, the titration maybe performed by reacting the aldehyde groups of a sample of the material with a known dry weight with hydroxylamine, typically provided as hydroxylamine hydrochloride, at a specific pH (e.g. pH 4) which reaction releases a stochiometric amount of protons, lowering the pH. By performing a regular acid-base titration back to the specific pH, the number of aldehyde groups in the sample is determined. It should be noted that this relates to the amount of aldehyde groups remaining after the sulfonation. The total amount of introduced aldehyde groups prior to sulfonation is thus this remaining amount plus the amount of sulfo groups.

[0023] The aldehydes may be introduced by oxidation, e.g. of cellulose fibres or nanocellulose such as CNF, by means of sodium metaperiodate (NaIO ₄ ). In some embodiments, the amount of NaIO ₄ is within the range of 1-3 g/g dry fibre.

[0024] The aldehyde groups can contribute to the desired negatively charged environment within the membrane by facilitating introduction of acidic substituent groups, e.g. sulfo groups [-S(=O) ₂ -OH] as discussed herein, but they can also react for crosslinking as discussed below.

[0025] The crosslinking of the cellulose nanomaterial may be achieved by reaction of at least some of the aldehyde groups exhibited on the surfaces of the nanomaterial, typically with hydroxy groups [-OH] which are also exhibited on the surfaces of the nanomaterial. However, other ways of crosslinking the cellulose nanomaterial, physically and/or chemically, are also possible. When the membrane 1 is dry, all, substantially all or at least most of the aldehyde groups remaining after sulfonation may be in acetal or hemiacetal form after reaction with hydroxy groups. In water, a minor part of the crosslinks may be broken but most remain as hemiacetals at equilibrium. The crosslinking retains the structural integrity of the membrane and limits the swelling thereof in water or other solvent, e.g. when used in an RFB 10. Conveniently, the nanomaterial is crosslinked to such a degree that the membrane swells in the thickness direction within the range of 1.5-4 times, i.e. the thickness of the membrane 1 increases 1.5-4 times when immersed in an aqueous solution e.g. of the posolyte 3 and negolyte 2. In the plane of the membrane, the swelling is typically zero or negligible. As an example, a crosslinked membrane 1 with a dry thickness of 40 pm may swell e.g. two times to 80 pm or four times to 160 pm, when it would swell to several millimetres without crosslinking.

[0026] To introduce sulfo groups, at least some of the aldehyde groups exhibited on the surfaces of the nanomaterial can be sulfonated, e.g. by means of sodium metabisulfite (Na ₂ S ₂ O ₅ ). In some embodiments, the amount of Na ₂ S ₂ O ₅ is within the range of 0.5-2.5 g/g dry fibre.

[0027] In some embodiments of the present invention, the cellulose nanomaterial is chemically modified to exhibit, on surfaces of the nanomaterial, especially within pores of the membrane, sulfo groups in an amount providing a charge within the range of 200-1500 pmol/g nanomaterial, preferably within the range of 300-1400 pmol/g, as determined by conductometric titration in accordance with Katz “Determination of strong and weak acidic groups in sulphite pulps” of 1984.

[0028] Also the sulfo groups maybe negatively charged [-S(=0) ₂ -0 ] further contributing to the desired negatively charged environment within the membrane.

[0029] Alternatives to the use of Na ₂ S ₂ O ₅ for introducing the sulfo groups are envisioned. For instance, an amine containing reagent maybe used, e.g. sulfanilic acid and/or taurine. The amine can react with an aldehyde group to form imine which may then be reduced, e.g. by means of sodium borohydride, to form a stable amine, whereby the sulfo group would then be attached to the nanomaterial via the amine.

[0030] As mentioned above, each of the posolyte 3 and negolyte 2 comprises a redox component which is prevented from passing from the negolyte to the posolyte (or vice versa) by means of the membrane separating the posolyte and the negolyte from each other. The redox component may be any redox component conventionally used in RFB. In some embodiments of the present invention, one or both (preferably both) of the posolyte 3 and the negolyte 2 comprises an amphiphilic organic compound as redox component. Currently preferred amphiphilic organic compounds which may be used as redox component include amphiphilic organic compounds which maybe used as dyes, e.g. Alizarin Red S, Natural Red 4, Indigo carmine and/or Acid Blue 25, preferably Alizarin Red S which works well with the membrane of the present disclosure. In some embodiments, both the posolyte 3 and the negolyte 2 comprise a same redox component at different oxidation states, e.g. any of the amphiphilic organic compounds mentioned above such as Alizarin Red S.

[0031] The membrane 1 of the present disclosure may be especially suitable for use with aqueous solutions. Thus, both the posolyte 3 and the negolyte 2 may be aqueous.

[0032] The membrane 1 may, when dry, consists to 100% of the crosslinked cellulose nanomaterial. However, in other embodiments, the membrane 1 may comprise the cellulose nanomaterial in combination with another nanomaterial.

[0033] The membrane 1 may, when dry, have a thickness within the range of 5-100 pm, which maybe a suitable thickness for an RFB 10. The dryness of the membrane will be dependent for instance on the ambient temperature and relative humidity (RH%). However as a general definition and as recognized by the skilled person the word “dry” means substantially free from moisture or liquid; not wet or moist, but not anhydrous. As an example, when a membrane is treated or conditioned at 50% RH and 23°C th moisture content is approximately 10-12 wt-%.

[0034] The membrane 1 may, when dry, have a density within the range of 1.2-1.6 g/cm3, preferably within the range of 1.4-1.5 g/cm3, which is close to the density of pure cellulose.

[0035] Figure 2 schematically illustrates some embodiments of the method of the present invention. The method is for preparing a membrane 1, e.g. for use in an RFB 10 as discussed herein, or in e.g. a fuel cell.

[0036] Cellulose fibres are oxidized Si by means of NaIO ₄ to provide aldehyde groups on surfaces of CNF within fibre walls of the fibres. Then, some (but not all) of the thus provided aldehyde groups are sulfonated S2, e.g. by means of addition of Na ₂ S ₂ O ₅ which disintegrates, in contact with water to NaHSO ₃ or by addition of NaHSO ₃ , to provide sulfo groups on the surfaces of the CNF, thus obtaining chemically modified CNF within fibre walls. Then, the chemically modified CNF are obtained S3 in free form (i.e. separated from each other instead of bound in a fibre) by means of processing of the fibres, e.g. comprising mechanical processing by means of high-pressure homogenization. Then, the membrane is formed S4 by dewatering an aqueous slurry comprising the separated chemically modified CNF, e.g. in a mould or by spraying onto a dewatering wire, preferably in a mould, to obtain the membrane in dry form. The mould may e.g. be on a filter, allowing the water to be removed by filtration, e.g. by means of vacuum or only gravity. Alternatively, the water may be removed from the mould by evaporation, e.g. at an elevated temperature, e.g. within the range of 4O-8o°C, such as within the range of 4O-6o°C.

[0037] To form CNF-membrane the fibrils extracted from e.g. wood are fully separated by diluting with a solvent to about 0.2 wt%, usually water, which is done through a dispersive process. These dispersions are subsequently used to assemble very dense films which is done by completely removing the solvent from the CNF dispersion using e.g. casting or filtration. The removal of water allows the nanofibrils to come into close proximity and form a dense network, through capillary forces, held together by strong hydrogen bonds. This membrane can in theory be used as an IEM in a similar way as Nafion™ but instead its hydrophobic fluorinated regions the fibril itself would be used as a volume excluding media making it inaccessible to water and the conduction of ions will happen in the hydrated voids in between the CNFs. Instead of a hydrophobic PTFE part, the fibril volume will act as a liquid displacing media while the interfibrillar space along with surface charges will act as conducting channels. The IC within these channels would be governed by the total concentration of fixed charges in the membrane but also how these charges are distributed. Creating a dense enough CNF- membrane structure to inhibit percolating channels is a challenge but techniques such as interfibrillar cross-links in CNF-based IEMS have also successfully been demonstrated to work in organic RFBs.

[0038] According to one alternative embodiment the membrane may be further chemically modified by exposing the membrane, either as a never dried filter cake or as a dried membrane to a medium, being either a liquid solvent or a gaseous media, comprising solubilized or dispersed species e.g. any one of inorganic, organic acids, salts, compounds, other nanoparticles (oD, 1D,2D or 3D) such as nanocarbons, clays, metalorganic frameworks, hetero polyacids, molecular complexes, monomers or polymers, or combinations thereof, or as a vaporized/gaseous species. The membrane, or more specifically the cellulose nanomaterial, associates with, reacts with and/ or adsorbs these aforementioned solubilized or dispersed species. Examples of such species are listed below as non-limiting examples. Inorganic acids, i.e. mineral acids, such as any one of sulphuric acid, hydrochloric acid, perchloric acid, boric acid and phosphoric acid. Organic acids, i.e. any acid containing a carboxyl group and organic compound, such as any one of citric acid, butane tetracarboxylic acid, acetic acid. Salts, for instance metal salts, such as metal chlorides. Examples are iron chloride and calcium chloride. Compounds, for instance zirconoyl chloride (octahydrate), imidazole derivatives, e.g. guanazol, diamines (ethylen diamin), triamines (ethanol triamine), metal organic fram works (MOFS) such as Zeolitic imidazolate frameworks (ZIFs). The ISO definition of nano-objects includes as nano-objects nanoparticles (nanoscale in all the three dimensions), nanofibers (nanoscale in two dimensions), and nanosheets or nanolayers (nanoscale only in one dimension) that include graphene and MXenes. Examples of spices of nanoparticles therefore includes for instance; oD maybe silica nanoparticles, 1D maybe carbon nanotubes, ,2D maybe graphene or clays, such as montmorillonite or 3D having a nanoscale in all three dimensions. Hetero polyacids maybe any one of tungsten phosphoric acid, tungstosilicic acid, molybdosilicic acid, phosphotungstic acid. Complexes may include iron-ethylenediamin tetra acetic acid. Monomers may include dopamine, methacrylate, vinyl sulfonate, vinyl phosphonate. Polymers may include polystyrene sulfonate, polydopamine, polydiallyldimethyl ammoniochloride, polyacrylic acid.

[0039] Method-wise two different types of procedures can be used: The first being a chemical treatment of membrane filter cake pre-drying, referred to as Modification Type 1 (mod-Ti), as shown in Fig. 3a, where the wet filtration cake is treated in a solution for 17 h and then dried. The second branch was the post-drying treatment of membranes primarily aimed at modifying the membrane bulk, which was called Modification Type 2 (mod-T2), as shown in Fig. 3b, where the membrane is dried prior to treatment in a solution for 17 h and then dried again.

[0040] Description of Trials and results for Mod-Ti and Mod-T2

Never dried fibres supplied by Nordic Paper, H ₂ SO ₄ , HC1, KC1, Dopamine hydrochloride, Alizarine S Red, tungsten phosphoric acid hydrate, pDADMAC, sodium periodide, sodium bisulphite, hydroxyl hydrochloride, (3- Aminopropyl)triethoxysilane(APTES) Durapore® 0.65pm PVDF membrane were all purchased from Sigma Aldrich. Vanadium redox mix of 1.6 M V(III)/V(IV) in 2M H ₂ SO ₄ (Gesekkschaft fur Elektrometallurgie mbH, GfE). Nafion™ 212 and Sigracet 29 BA carbon paper both from FuelCellStore. CNFs were produced by the sequential oxidation and sulfonation of the never dried pulp using sequential reactions with periodate and sodium bisulphite. The periodate oxidation selectively introduced aldehydes to the C2 and C3 carbons and these aldehydes were subsequently partially converted to sulfonic acids by the bisulphite reaction. The modified fibres were subsequently homogenized in order to liberate the individual fibrils from the fibres. The liberation was carried out by passing the fibres two times through the homogenizer at approximately 15 g/L using 400pm and 200pm sized chambers for the first pass at 500 bars and 200pm and 100 pm sized chambers for the second pass at 1500 bars. The number of aldehydes after homogenization was determined to be 1 mmol/g using hydroxylamine hydrochloride and the number of sulfonic acids to be between 200-400 peq/g as determined by colloidal titration using a Stabino. The formed gel was tested for dry-eight and stored in a fridge at 4 C° until further use. The CNF-slurry was diluted with Milli-Q water to desired concentration followed by three dispersive steps using an Ultra Turrax for 10 min at ~15.OOO rpm then ultra-sonication for 20 min at 50% amplitude in an ice bath and finally centrifuging for 60 min at 4500 rpm. The supernatant was collected and the dry content was measured and the nano yield, i.e. the ratio of initial to final concentration, was determined. For higher concentrated dispersion (> 2 g/1) the viscosity increased which required increased Ultra Turrax speed (~2o.ooo rpm) and additional magnetic bar-stirring was applied in both steps.

Membranes were prepared by dewatering the CNF dispersion, and collection of the CNF filter cake collected and completely dried in a Rapid Kothen sheet dryer Siemens Simatic OP7 at 70 °C for 15 min. Lastly, the membranes were carefully removed from the PVDF-filters. Since the active filter area was defined by the vessel itself, different membrane thicknesses were possible to prepare by varying the amount of dispersion.

Finally the following modifications were performed according to Fig. 3a and 3b.

A-Mod-Ti

Wet filter cake (Ti) straight from the pressure vessel was put in the fridge at 4 C° for around 2 h to let it solidify to facilitate the removal of the PVDF filter from the filter cake. The cake was submerged in 0.5 M H ₂ SO ₄ for at least 12 h before drying in Rapid Khoten at 70 C° for 15 min.

A-mod-T2

The pristine membrane was submerged in 0.5 M H ₂ SO ₄ for at least 12 h before drying in Rapid Khoten at 70 C° for 10 min.

HPW-mod-T2 Tungsten phosphoric acid hydrate, CAS 1343-93-7, was dissolved in 0.5 H ₂ SO ₄ to a concentration of 5 wt%. Wet filter cake (Ti) or dried membrane (T2) was submerged in HPW-solution for at least 12 h before drying in Rapid Khoten at 70 C for 15 min (Ti) or 10 min (T2). The IC can be increased by a number of ways and ideally a membrane should exhibit a zero-resistance permselectivity. A high density of fixed charges is the most straight forward way, however, for CNF-based membranes an increased charge density means a more destructive chemical pre-treatment which weakens membranes as mentioned before. Compared to Nafion™, which typically exhibits low levels of hydration compared to CNF-based membranes the problem in these is not due to low IC, but instead a low permselectivity. To solve these problems the idea is to decrease swelling of CNF-membranes and incorporate solid ion conducting nano-particles that blocks unwanted permeation while still maintaining a high conductivity. One such particle is the heteropoly acid called phosphotungstenic acid which is a metal complex with a so-called keggin-structure. The general formula of this structure is H ₃ PW ₁₂ 0 _4o (HPW) which forms a semi-spheric metal complex with a width of around 1 nm. These metal complexes have been commercially used in fuel cell membranes to maintain conductivity at low levels of hydration at elevated temperatures. The proton conduction mechanisms associated with the HPW happens both on surface of the structure by proton hopping and also straight through by quantum tunneling.

The mechanical integrity, SR and IC are important properties to consider when optimizing the membranes for systems such as the VRFB. Results show that anti- swelling modifications such as the surface-blocking (D-Mod-T2) and bulk modifications (A-mod-T2) both outperformed pristine membranes in wet integrity and SR but the IC decreased as seen in Table 1. The IC-boosting modification (HPW-mod) resulted in higher conductivity despite keeping a low swelling at the cost of structural integrity.

[0041] Table 1. Membrane characteristics including tensile properties, SR, and IC (EIS 1.)

A-mod-Ti modification with an SR at 12.5% shows promising results in both wet strength and modulus, 39.0 MPa and 1.95 GPa, along with quite low conductivity (EIS 1), 9.9 mS cm-i, Table 1. However, the Ti modification method turned out to be a somewhat delicate process and removing the filter cake from the PVDF filter often resulted in damage to the membrane, hence, leading to the discontinuation of Ti modifications. The most promising anti-swelling modification turned out to be A-mod- T2 that only exhibits an SR of 10.1 % in H ₂ SO ₄ compared to 338 % of pristine membranes, Table 1. The dry mechanical integrity of the A-mod-T2 shows a decrease in both tensile strength and young’s modulus of 68 % (29.9 MPa) and 57 % (1.5 GPa) respectively compared to pristine membranes. As mentioned, from a VRFB applicability point of view, the dry strength is not as relevant as the strength of the VRFB environment (H ₂ SO ₄ ). Wet strength (1M H ₂ SO ₄ ) shows that the A-mod-T2 increases the tensile strength tenfold and the modulus almost 100 times, Table 1. Interestingly, when taken out of the H ₂ SO ₄ solution, the liquid form pearls on the surface and the effect seem to be stable, demonstrating the same behavior after weeks of submersion and even after cycling in VRFB. Submersion in Milli-Q water, on the other hand, resulted in blister-like structures appearing all over the membrane within less than an hour. The reason for these effects, including the higher wet integrity, could be connected to higher interfibrillar crosslinking and the blisters could indicate that there is a heterogeneous membrane structure with more and less hydrophilic areas. IC measurements of A-mod-T2 show a poor conductivity (4.38 mS cm-2) in these initial measurements, Table 1.

[0042] The HPW-mod-T2 increases IC reaching 59.22 mS cm-2 and while exhibiting relatively low SR at 54 %. The lower swelling and retained IC suggest the presence of other transport mechanisms than simple bulk water conduction and prior studies have shown that the kegging structure of the HPW particles facilitates proton transport. Tensile strength test show extraordinarily low mechanical integrity both wet and dry which is a known occurring trade-off phenomena when modifying for higher conductivity, Table 1. Moreover, there is a notable swelling in x-y-plane, not seen among the other modifications, which was noted when drying caused shrinking and subsequent cracking of the membrane. The investigation of XRD-results indicates a dramatic decrease in crystallinity which suggests extensive fibril degradation that in turn would explain why the swelling becomes more prominent in all 3 dimensions. The thickness of dry HPW-mod-T2 increases with around 35 % after modification, which suggests that HPW occupies space during the treatment and seem to exclude water. The content of HPW in the HPW-mod-T2 membranes is calculated to be slightly above 30 wt% and when decreasing the content down to 20 wt%, the SR halves and both dry tensile strength and young’s modulus increases from 14.0 to 44.7 MPa and 0.74 to 1.32 GPa respectively. However, the IC of 20 wt% HPW is around a third (21 mS cm-1) compared to 30 wt% (59 mS cm-i), Table 1.

Visually the HPW-mod-T2 modification is opaque, has a light brown undertone most likely because of some degree of hydrolyzation and a greyish tint most probably from the HPW since it was not observed in the A-mod. Other observations of HPW-mod-T2 include the same blistering effect seen in A-mod-T2 which indicates that it is important to keep these at a low pH to prevent this phenomenon. On the other hand, the solution- pearling-effect of A-mod-T2 was absent in HPW-mod-T2 and the surface was uniformly wet after submersion in H ₂ SO ₄ . However, upon submersion, the HPW-mod-T2 membranes show an interestingly metallic/ mother of pearlish appearance. Water contact angle results suggest that the membranes are fairly hydrophilic which rules out the possibility that the reflective effect is due to an air gap on the membrane surface and instead is due to some reflective effect of the HPW or air/ nitrogen is trapped inside.

[0043] Pristine, A-mod-T2 and HPW-mod-T2 comparison.

When inspecting the results from modified membranes the extremes along with the reference (pristine) were chosen to proceed to further i.e. the least swelling (A-mod-T2) and the highest ion conducting membrane at low swelling (HPW-mod-T2). By interpreting these results there is an evident trade-off between swelling and conductivity that Nafion™ seems to have dissociated from, Table 2. With this in mind, the IC of Nafion™ 212s in H ₂ SO ₄ is considered a close-to-maximum value for a low swelling membrane and is therefore used as a comparence for proton permselective conductivity.

Further characterizations included EIS 2, SEM, XRD, FT-IR, oxidation stability, vanadium permeability water uptake and in the next section, VRFB performance.

The IC, using EIS 2, confirms the trends seen using EIS 1, however, the values differ, Table 2. Interestingly, A-mod-T2 exhibits a lot higher water uptake than expected when comparing to its SR. For the HPW-mod-T2 it is the other way around, high SR and low WU. The reason behind the large gap between Nafion™ WU and the cellulose-based membranes is due to the difference of hydrophilicity of the membrane matrix for the two types. For Nafion™, the matrix is highly hydrophobic while the CNF matrix is more prone to hydration.

[0044] Table 2. Membrane characteristics including avg. thickness, IC, and water uptake.

[0045] SEM images were taken of membranes pre and post swelling to understand hydration and other modification effects and are shown in Fig. 4. Fig. 4 shows cross- section SEM-images of membranes swelled in 1M HC1 with subsequent freeze-drying A: Pristine. B: A-mod-T2. C: HPW-Mod-T2. Concerning the swelling, the most profound observation in the cross-sections of the wet and freeze-dried samples is that A-mod-T2 and HPW-mod-T2 exhibit delamination, Fig. 4. The cause of delamination is most probably due to confined water expanding fast when submerged in liquid nitrogen. Pristine membranes have no sign of delamination and at higher magnification it seems like the individual layers are more swelled than the other two which suggests that the fibrils in pristine membrane are not as restricted. It could also be due to the share volume of liquid in the Pristine membrane which would result in a slower freezing hence giving the fibrils more time to move. Otherwise, the membranes show laminated structures which which is a result of dead-end filtration but, as seen at the higher magnified image of HPW-mod-T2 (C in Fig. 4), the laminated structure is almost gone in comparison to the other two. This could mean that the modification has disintegrated the fibrils and be the reason these membranes exhibit very low mechanical integrity.

[0046] The FT-IR spectrums of the three membranes was analyzed using literature on similar material, and are shown in Fig. 5. The broad peak around 3300 cm-1 is a result of the -OH stretching and vibrating which also correlates to the peak at 1647 cm-i which is the bending motion of the same -OH groups. The broadening of the 1647 cm-1 peaks, seen for A-mod-T2 and HPW-mod-T2, is due to increased hydration of the -OH groups. The peak slightly higher than 3324 cm-i(to the left), is the intramolecular H- bonds of cellulose which is least pronounced for the HPW-mod-T2 suggesting a lower share of cellulose crystallinity. The peak at 2895.5 cm-1 is due to symmetric and antisymmetric vibration of -CH ₂ groups. The largest deviation is seen at 1021 cm-1 together with the peak at 1105 cm-i which is due to the C ₁ -O-C ₄ glucoside bonds that links the monomers, in the case of HPW-mod-T2 spectrogram, these peaks are more or less absent, indicating severe hydrolyzation of the glucose chains.

[0047] The XRD-pattern of pristine, A-mod-T2 and HPW-mod-T2 is presented in Fig. 6. There is a slight intensity decrement after A-mod-T2 treatment, which indicates that the crystallinity decreases slightly. However, when looking at the HPW-mod-T2 there is almost no crystallinity detected which means that there has been a radical change of the fibril intramolecular composition.

[0048] Summary of material characteristics

The pristine membranes displays good dry mechanics but swells a lot in aqueous solutions, probably to the point of being more a gel rather than a solid structure. The fact that it becomes a gel and not fully dissolve argues for interfibrillar crosslinking, probably due to van der Waals forces and H-bonds but also some due to a share of hemiacetalic bonds. However, as a result of extensive swelling, the effective pore size will be too large to yield good selectivity. The A-mod-T2 have low SR but quite a high WU which is quite contradictory. One explanation could be a faulty method of WU measurement which, as seen in Water uptake, does not consider any leaching during hydration which could have been avoided by measuring the dry membrane before hydration. On the other hand, if the measurement is correct, the A-mod-T2 should have a very low water content in the dry state together with a porous structure with interfibrillar bonds strong enough to withstand the internal osmotic pressure when being submerged in solution. The opposite is true for HPW-mod-T2 that has a higher SR than A-mod-T2 but a lower WU which could mean that the HPW nano particle occupies space inside the pores which could be the reason why these membranes become 46 % thicker compared to pristine. With this said, SEM, FT-IR and XRD analysis indicate extensive material alteration which suggests general CNF degradation for HPW-mod-T2. This might also explain the reason for the noticeable increased swelling in the x-y-plane, as mentioned in HPW-mod. If the CNF degrades it will result in the loss of the laminated structure which ultimately diminishes the x-y-restriction seen in pristine and A-mod-T2. Based on these observations of the membrane modifications, one pore of four fibrils can be described in the following: A-mod-T2 is swelled in the A-solution and during drying the fibrils slip back and when the last water leaves it forms some kind of strong interfibrillar bonds that holds up better against the capillary forces when swelling in electrolyte compared to pristine membranes. HPW- mod-T2 is similar to A-mod-T2 except for the HPW particles that are being introduced during swelling and packed tight during drying.

[0049] The membranes were run one time each in the VRFB with the exception of HPW-mod-T2 that were run three times since its results seemed most promising. Results from 30 cycle experiments show that the CNF-membranes indeed are quite comparable to Nafion™ 212, Table 3. Nafion™ has a higher IC than A-mod-T2 and HPW-mod-T2 which is why it has the highest voltage efficiency, also this confirms the very low permselectivity for pristine CNF-SO ₃ which had the highest IC while at the same time having the lowest marks on efficiency. HPW-mod-T2 outperformed Nafion™ with a higher coulombic efficiency hence a very low-capacity loss over the 30 cycles while remaining stable for 100 cycles. [0050] Table 3. Mean values over 30 cycles in VRFB at 100 mA/cm2 current density and a 20 ml/min flow.

A noteworthy difference is observed between the first cycles discharge capacities with 37 and 24 Ah/1 for HPW-mod-T2 B#2.2 and A-mod-T2 B#i respectively, showing discharge capacity of 30 cycles visualizing the differences between batch #1 (B#i) and #2 (B#2). The reason is not straightforward but are most likely due to different activation processes of electrolyte, electrode and the membrane. Another reason could be differentiating electrolyte concentration. The first round of VRFB cycling test, with membranes manufactured from CNF batch #1, show an overall greater capacity loss compared to subsequent runs with HPW-membranes made from batch #2, Fig. 7. This could be due causing the membranes to be less dense but can also be battery associated variations. The 30 cycles comparison of HPW-mod-T2 from batch #2 against Nafion™ 212 show that the CNF-based membrane has superior columbic efficiency (98.4 vs 95.7 %) and lower discharge capacity loss (5.9 vs 15.8 %) respectively Fig. 8 and Table 3. This indicates a better selectivity for the HPW-mod-T2 and further tests of self-discharge (161 h) and permeability rate of VO2+ (5.57 *10-8 cm-2 min-1) which are both very good results when comparing to other studies.

[0051] The rate performance test was done to see how the membrane reacts to increasing discharge currents to derive the regenerative ability after stressing the membrane which can give clues about membrane properties. The results indicate that the HPW-mod-T2 has larger capacity drops for each increment of current density which is due to higher ohmic losses and transport limitations which is in accordance with the EIS results. However, when comparing the regenerative ability, HPW-mod-T2 regained 93.2 % while Nafion™ 212 regained 90.8 % of the discharge capacity when comparing the cycles 6-10 and 26-30 at 50 mA/ cm2. The reasons for this are mostly due to the lower vanadium crossover but it is also an indication of the surprisingly high chemical stability HPW-mod-T2. This is disclosed in Fig. 9 comparing Nafion™ and HPW-mod- T2. In this test, the HPW-shows higher capacity drops compared to Nafion™ but show a better regeneration when comparing cycle 11-15 against 26-30 (91 % and 93 % for Nafion™ and HPW respectively).

[0052] The anti-swelling modification A-mod-T2 successfully decreased the SR from 338 % to 10.1% in 1M H ₂ SO ₄ and exhibited good wet tensile properties but poor ion- conductivity. The HPW-mod-T2 on the other hand, displayed almost double the IC of A- mod-T2 (from 23.0 to 44.4 mS cm-1) at the cost of poor mechanical integrity. The HPW- mod-T2 was comparable to Nafion™ 212 VRFB performance on most points showing good voltage efficiency and formidably low VO 2 -permeability hence exhibiting the lowest discharge capacity loss and high coulombic efficiency.

[0053] Membrane property enhancing modification using sulfuric acid decreased swelling ratio (SR) from about 340 to 10 % yielding an ionic conductivity (IC) of 22 mS cm-i. Using tungsten phosphoric acid as IC-booster in a modification process called HPW-mod-T2 which resulted in a SR at 54 % while increasing IC to 44 mS cm-1 at the cost of decreased mechanical integrity. RFB-performance comparison between HPW- mod-T2 and Nafion™ 212 in a single cell VRFB suggests that the CNF membrane is not as chemical stable in these environments but outperforms Nafion™ in terms of coulombic efficiency (98 vs 96 % respectively), vanadium crossover (5.57*10-8 vs 6.82*10-7 cm2 min-i) and self-discharge (161 vs no h). Also, HPW-mod-T2 displays commercially relavant voltage and energy efficiencies (84 and 82 % respectively).

[0054] The present disclosure has mainly been described above with reference to a few embodiments. However, as is readily appreciated by a person skilled in the art, other embodiments than the ones disclosed above are equally possible within the scope of the present disclosure, as defined by the appended claims.