Spatial separation of the electrolyte and electrode is the main characteristic of flow-battery technologies, which liberates them from the constraints of overall energy content and the energy/power ratio. The concept of a flowing electrolyte not only presents a cost-effective approach for large-scale energy storage, but has also recently been used to develop a wide range of new hybrid energy storage and conversion systems. The advent of flow-based lithium-ion, organic redox-active materials, metal–air cells and photoelectrochemical batteries promises new opportunities for advanced electrical energy-storage technologies. In this Review, we present a critical overview of recent progress in conventional aqueous redox-flow batteries and next-generation flow batteries, highlighting the latest innovative alternative materials. We outline their technical feasibility for use in long-term and large-scale electrical energy-storage devices, as well as the limitations that need to be overcome, providing our view of promising future research directions in the field of redox-flow batteries.
The increasing demand for renewable energy resources, such as solar and wind power, necessitates the development of large-scale electrical energy-storage (EES) systems, for example, for load leveling and peak shaving for the grid energy market1,2. The main purpose of EES systems is to enable the rapid and efficient storage and release of a large amount of energy on demand. Lithium-ion battery (LIB) technology is still the most mature practical energy-storage option because of its high volumetric energy density (600–650 Wh l−1 for a typical cylindrical 18650 cell3), and the use of LIBs in EES systems is continuously growing in the electric vehicle and smart-grid markets. These systems are constructed by connecting several small-sized LIB cells in series or parallel to produce the required power and energy. Despite the high energy densities, current LIB technology has limitations arising from the lithium intercalation mechanism, which occurs in the limited interstitial space4. In addition, large numbers of unit cells and battery packs that contain flammable organic electrolytes present an explosion and/or fire risk.
In an alternative technology — redox-flow batteries (RFBs; see Box 1) — the redox reaction usually occurs on the electrode surface without damaging the internal structure, thus making it more suitable for long life cycles (>10,000 cycles, 10–20 years)5. Design flexibility is an additional, strong advantage of RFBs, in which the energy (the volume of electrolyte in external tanks) and power output (the active area inside stacks) are effectively decoupled and can be adjusted to meet the demands of the end users. These properties make flow-battery technology attractive for economically viable EES applications.
There has been a rapid increase in the volume of research on flow batteries, especially for studies focused on novel redox couples and flow-type systems6. Conventional metal- or inorganic-based aqueous RFBs have successfully demonstrated reliable electrochemical performance and have undergone large-scale field testing at the commercial level7,8. However, it has been recognized that protic systems have several intrinsic disadvantages, such as low energy density (∼50 Wh l−1), low operation potential (<2 V) and water electrolysis side reactions4. To overcome these disadvantages, a growing effort has been focused on developing novel systems to increase energy density and operating voltage. This trend, which has been referred to as the ‘renaissance of the flow batteries’ (Ref. 6), is very similar to the interest in fuel-cell technologies in the early 2000s. Over the past 15 years, various fuel cells, such as alkaline, polymeric-electrolyte membrane, solid oxide and molten carbonate fuel cells, have been developed9. However, fuel-cell technology has not been commercially successful because of the lack of appropriate materials (for example, an alternative catalyst and membrane) and/or a suitable hydrogen infrastructure. The current state of development of flow-battery technologies is similar in terms of material and system development. Many studies have reported unique catalysts, membranes, redox couples and hybrid flow-type systems10,
In this Review, we discuss recent progress in the development of flow batteries, highlighting the latest alternative materials and chemistries, which we divide into two categories: conventional aqueous RFBs and next-generation flow batteries. Conventional aqueous RFBs have been reviewed in many previous publications, most of which have focused on vanadium redox-flow batteries (VRFBs). In this context, we focus on the reasons why the study of VRFBs has steadily increased and look at future research directions in this area. Despite the notable progress in next-generation flow batteries since 2010, the future of the newly developed systems based on organic, iodine, polysulfide or semi-solid materials is uncertain because of challenges that we discuss in this Review. In the final section, we highlight the main limitations and requirements of current flow batteries, presenting a realistic outlook.
Before discussing the range of materials and chemistries that have been used in flow batteries, we examine the basic components and principles of RFBs. RFBs can be classified into two main types based on the electrolyte used: that is, aqueous or non-aqueous solutions. Conventionally, the RFB system stores redox-active materials and electrolytes in two external reservoirs. However, in the current state of development, various flow batteries have evolved from other energy storage and conversion systems, leading to blurring of the traditional classification boundaries14. We refer to these kinds of systems, which are distinct from conventional RFBs containing redox couples, as flow batteries or flow-type batteries. For example, several flow batteries using lithium or zinc metal for the anode have only one external tank. More importantly, the concept of electrolyte circulation has been developed to successfully encompass several other energy storage and conversion systems such as metal–air or photochemical cells15. Therefore, in this Review, we carefully define the components and structures used in various flow batteries.
A stack-type flow battery, similar in configuration to conventional fuel cells, is probably the design that is most closely approaching commercial applicability. The main components of the stack cell are the negative and positive electrodes, bipolar plates, current collectors and membranes. In conventional RFBs, the redox-active materials are supplied from external tanks and undergo redox reactions on the electrode surfaces. Carbon-based materials (for example, carbon felts or carbon papers) are most widely used as the basic electrode to provide redox-active sites. The bipolar plates can be used to not only protect against direct contact between the electrolyte and metal current collectors, but also to enable series construction for scaling up. Notably, there are two external reservoirs containing a cathode electrolyte and an anode electrolyte (also referred to as the catholyte and the anolyte). The membrane, which is usually located at the centre of each stack cell, separates the two different electrolytes. In addition, the membrane selectively transfers charge-carrier ions, thereby maintaining the charge balance in both the catholyte and the anolyte. Electrolyte circulation is a unique feature of flow batteries that is realized through pumps connected to the stack and the continuous supply of catholyte and anolyte to the system. Most practical aqueous and non-aqueous RFBs based on metal or metal–ligand complex redox couples use the same stack design. Recently reported flow batteries (referred to hereafter as next-generation flow batteries), break away from the traditional frameworks in terms of their unique architectures and use of novel materials. Typical energy storage and conversion systems, such as LIBs, solar cells and metal–air cells, can be combined with flow batteries using the concept of design flexibility4,16. A timeline of the development of flow batteries is provided in Fig. 1.
Notably, the use of an extendable storage vessel and flowable redox-active materials can be advantageous in terms of increased energy output. Lithium-metal-based flow batteries have only one electrolyte tank with either a non-aqueous or an aqueous catholyte, containing metal oxide, iodine, polysulfide or organic active materials4. In some cases, ceramic separators are used to protect the lithium metal from the aqueous catholyte17. The concepts of pumpless or membraneless flow batteries have also been reported. These concepts, which are still in the proof-of-concept stage, use magnetic-field-controlled transfer18, gravity-induced properties19 or a laminar-flow electrochemical cell20. More recently, photochemical flow batteries, in which titanium- or iron-based photocatalysts are used, have been intensively studied to directly store the generated solar energy16. Two different configurations have been studied for these flow batteries. The first example is a photochemical cell and a RFB that are physically connected by an external electrolyte circuit21,22. In this case, the photocharged redox-active materials are moved into the RFB for electrochemical discharge. The other configuration is a single-stack design in which the photocharge cell and the electrochemical part are co-located23. Metal–air batteries are promising, environmentally friendly EES systems that can also be combined with flow systems24,25. These systems have only one electrolyte tank in the negative electrode side for electrolyte circulation, whereas the other side uses oxygen from the air. In this case, the flowing electrolyte can mitigate metal dendrite growth and other side reactions, as discussed later.
The combination of flow batteries and other energy storage and conversion mechanisms can lead to synergistic increases in electrochemical performance and a reduction in capital costs. For the purpose of storing energy by simply holding redox-active materials in an external reservoir, the flow-battery concept addresses the limitations of traditional static-type energy devices through the spatial separation of the electrolyte and the electrode, which are often packaged in the same compartment in static electrode batteries and therefore subject to space constraints. Rather than discussing traditional RFBs in detail, we focus on next-generation flow-battery designs.
Conventional redox-flow batteries
Feasible aqueous catholytes and anolytes. The operating voltage of aqueous RFBs is highly restrained by the water splitting region. Thus, appropriate redox-active materials should be carefully selected using Pourbaix diagrams, considering chemical stability as well as pH dependence. Notably, most RFBs are operated in highly concentrated acidic conditions because most metal ions precipitate at high pH, forming insoluble hydroxides11.
Early RFBs based on all-vanadium, Fe/Cr and Zn/Br couples have encountered commercialization barriers. Fe/Cr RFBs were invented at the National Aeronautics and Space Administration Lewis Research Center in the 1970s2. This design uses a ferric/ferrous (Fe2+/Fe3+) couple as the catholyte and a chromic/chromous (Cr2+/Cr3+) couple as the anolyte in a halide solution and has a working voltage of 1.18 V. However, several critical issues, including poor electrochemical activity of the Cr2+/Cr3+ redox couple, hydrogen evolution, a high working temperature (∼65 °C) and capacity decay by severe cross-contamination, have hampered practical use. RFBs using a Zn/Br couple were also first demonstrated in the 1970s26. In contrast with one-phase, all-liquid flow batteries, this system is a phase-transition-based RFB concept, known as a two-phase hybrid system. Unfortunately, the degree of deposition on the zinc anode limits the overall capacity. Moreover, the use of toxic bromine and a poor cycle life resulting from the formation of zinc dendrites have hindered commercialization of this RFB system.
Currently, the most widely studied aqueous RFB is the all-vanadium system (VRFB), developed in the 1980s27. This system incorporates vanadium in four different oxidation states (V2+/V3+ for the anolyte and VO2+/VO2+ for the catholyte, with sulfuric acid in each half), thus mitigating capacity decay resulting from cross-contamination and yielding high efficiency27. When each electrolyte is mixed, the initial oxidation states of vanadium can be recovered via a ‘rebalancing’ process. For more practical use, intensive efforts have focused on increasing the volumetric energy density of VRFBs, which is related to the saturation and solubility limits of vanadium ions in the supporting electrolyte. The properties of vanadium electrolytes are affected by the operating temperature, state of charge and composition7. In terms of stability, VO2+ ions with yellow-coloured dioxovanadium are readily precipitated in the form of solid V2O5 at elevated temperatures (>40 °C)28. Thus, several modifications of vanadium species using organic and inorganic additives and a mixed acid electrolyte have been developed to prevent the precipitation of VO2+ ions and enhance cycling stability29,
Materials for electrodes and electrocatalysts. Redox reactions in RFBs occur on the surface of electrodes with circulating catholytes and anolytes. Among these reactions, vanadium redox reactions in VRFBs have been the most widely studied with the goal of increasing electrochemical performance7,33. In general, because of their excellent chemical stability in concentrated acidic conditions and their high electrical conductivity, carbon-based electrodes, such as carbon felts and carbon papers, have been widely used to provide redox-active sites10. However, low electrochemical activity, poor kinetic reversibility of carbon substrates and undesired water electrolysis from increased overpotentials impede the vanadium redox reactions34,35. Intensive efforts have focused on overcoming these issues. These efforts have included performing various surface treatments (for example, electrochemical oxidation, hydrothermal carbonization, thermal activation, acid treatment and corona discharge) and coating or depositing electrocatalysts onto the surface of the electrodes (Fig. 2a). Such surface modifications can enhance the electrochemical activity and wettability of carbon-based electrodes. The main factors that improve catalytic activities are related to increasing the number of active sites and the electron conductivity. In this context, several electrocatalysts for VRFBs have been reported. These materials are divided into two categories: metal-based36,
Recently, many review papers providing excellent summaries of reported electrocatalysts for VRFB electrode materials have been published10,34,41, and there has been an increasing number of studies on the vanadium redox reaction mechanism at the electrocatalyst surface34. For carbon-based electrodes, it was initially suggested that the oxygen functional groups, C=O and COOH, were the main active sites42. Intensive effort has since been devoted to increasing the number of redox-active sites and to understand the reaction kinetics of vanadium redox pairs. However, there is continuing controversy regarding the main active sites and the rate-determining step for vanadium redox reactions10. In this regard, fundamental studies need to be pursued, not only to improve electrochemical performance, but also to identify the exact reaction mechanism.
More practically, recent studies have shown that there are substantial differences among commercial carbon felt electrodes43. In particular, a polyacrylonitrile-based carbon felt demonstrated excellent electrochemical performance compared with other substrates, which originated from the different processing procedure and precursors, as well as differences in morphology of the carbon felt electrodes. Thus, when assessing the electrochemical performance, the origin of the carbon felt materials should be taken into consideration.
Materials for membranes. The membrane or separator has a key role in RFBs by preventing cross-contamination between the catholyte and anolyte and by selectively allowing transport of charge-carrier ions. Early RFBs used a perfluorosulfonated cation-exchange membrane, Nafion, which originated from proton-exchange membrane fuel-cell technology44. In subsequent RFBs, membranes have been adopted from other technologies, such as fuel cells, water treatment systems and water electrolysis. Notably, research into membranes for VRFBs has intensified since 2010. Research trends for IEMs based on perfluorinated (with organic and inorganic resins) and non-fluorinated materials have been summarized in many review articles12,13,45. The authors of these reviews emphasized the importance of developing non-fluorinated IEMs because of the high cost and low selectivity of perfluorinated IEMs. Thus, non-fluorinated IEMs, such as sulfonated poly(arylene thioether ketone ketone), sulfonated poly(ether ether ketone) and sulfonated Diels–Alder polyphenylene, have been studied widely33. However, these IEMs still provide low ion conductivity with low mechanical strength and chemical stability12. To overcome these problems, various inorganic materials, such as ZrO2 (Ref. 46), SiO2 (Ref. 47), Al2O3 (Ref. 48), TiO2 (Ref. 49), MoS2 (Ref. 50), carbon nanotubes51 and graphene52,
Alternatively, porous polymer separators have been used to effectively separate the redox-active ions by pore size exclusion. This concept is widely used in commercial LIBs44. Initially, polyacrylonitrile nanofiltration and polyvinylidene difluoride (PVDF) ultrafiltration membranes were used as porous separators in VRFBs55,56 (Fig. 2h). In an effort to identify low-cost porous separators with high ionic conductivity for commercial viability, researchers are now reporting promising separators, such as a nanoporous polytetrafluoroethylene/silica composite44. Although this porous separator showed good capacity retention during cycling, the Coulombic efficiency (∼95% at a current density of 40 mA cm−2) needs to be improved to a level comparable to that of commercialized LIBs (>99.5%). Recently, great progress has been made using porous polybenzimidazole (PBI) separators that demonstrate ultrahigh selectivity and chemical stability57,
A range of membrane fabrication methods are being introduced in the RFB field to replace the conventional perfluorinated IEM with low-cost porous separators. In general, as the IEM thickness increases, crossover can be significantly alleviated; however, this inevitably increases the membrane resistance and decreases the ionic conductivity. By contrast, porous membranes with ion-exchange groups have the substantial advantages of low cost and high power density owing to the lower internal resistance that results from the membranes being thin. It is noteworthy that crossover occurs again when the membranes become very thin. In addition, the cycle life of the RFBs is closely related to the chemical stability of the membranes, and only some candidates can be used in VRFBs because of highly oxidizing VO2+ ions and concentrated acidic conditions. Therefore, membranes for aqueous RFBs operating in concentrated acid or base conditions should be carefully selected by considering the chemical stability, mechanical strength, permselectivity and capital cost.
Next-generation flow batteries
The limitation of metal ion-based aqueous RFBs encourages researchers to refocus on non-aqueous or all-organic flow-battery technologies63. In this field, a range of novel redox couples and design concepts (for example, based on organic compounds, polysulfides, iodine or semi-solid active materials) have been extensively developed for new flow batteries. Although proof-of-concept high-performance devices have been reported, follow-up studies are still required to stabilize and optimize the developed systems for scaling up. In this section, we summarize novel redox-active materials and design concepts for next-generation flow batteries and discuss their limitations.
Lithium-ion batteries with flow systems. Commercial LIBs consist of cylindrical, prismatic and pouch configurations, in which energy is stored within a limited space3. Accordingly, to effectively increase energy-storage capacity, conventional LIBs have been combined with flow batteries. The simplest approach is the use of conventional cathode and anode materials but making them flowable (Fig. 3a). Since the first demonstration of the semi-solid concept, various active materials, such as LiCoO2, LiNi0.5Mn1.5O4, LiNi1/3Co1/3Mn1/3O2, LiFePO4, Li4Ti5O12, silicon and graphite have been used in semi-solid flow batteries (SSFBs)64,
Redox targeting reactions were proposed to avoid the use of a solid suspension and large amounts of conductive carbon70,71 (Fig. 3b). Similar to the SSFB, conventional cathode and anode materials are used in this lithium RFB. These electrode materials are immobile and are kept in the external reservoir; for example, porous TiO2 pellets as an anode material are located in an electrolyte tank. This system circulates only two pairs of redox mediators dissolved in the electrolyte. Redox couples, such as bis(pentamethyl-cyclopentadienyl)cobalt [Co(Cp*)2)] and cobaltocene [Co(Cp)2], and dibromoferrocene (FcBr2) and ferrocene (Fc), chemically oxidize and reduce the solid active materials step by step, thus participating in a chemical lithiation–delithiation process71,
Lithium metal with aqueous catholytes. In contrast to conventional LIBs operated in aprotic electrolytes, a proof-of-concept model for lithium aqueous flow batteries used an Fe(CN)63−/Fe(CN)64− redox couple as an alkaline cathode75,76 (Fig. 3d). The use of aqueous catholytes is advantageous because it increases the electrolyte volume in terms of safety and capital cost77. With respect to the reaction kinetics, Li+ diffusion in water-soluble redox couples is much faster than in intercalation cathode materials76. In addition, this kind of hybrid electrolyte system can achieve a high operating potential of >3.5 V. A high-performance lithium–iodine hybrid flow battery was developed that did not use heavy-metal-based cathodes17,78,
Lithium–sulfur batteries with flow systems. From 2013, lithium–sulfur based flow batteries have been intensively studied for large-scale energy storage18,82,
For solution-state active materials, the cell is operated within a narrow voltage window in the solution regime, which is more suitable for a flow system because a real liquid phase is used. This strategy can avoid detrimental effects from solid discharge products; however, the use of sulfur is severely limited. In systems with dispersed solid-state active materials, a voltage range of 1.6–2.5 V is used, which is similar to that of common lithium–sulfur batteries; this leads to greater utilization of the sulfur cathode. Flowable cathodes, such as polysulfide with carbon black86 or sulfur-impregnated carbon-black85, have been proposed, in which carbon additives were used to improve the electrical networks of the low-conductivity sulfur catholyte. Regarding the volumetric energy density, a sulfur-impregnated carbon composite achieved a catholyte volumetric capacity of 294 Ah l−1, which was calculated using the density of sulfur (2.07 g cm−3) without the supporting electrolyte volume. The redox targeting method, mentioned previously, was also applied in lithium–sulfur flow batteries, in which bis-(pentamethyl-cyclopentadienyl)chromium (CrCp*2) and bis-(pentamethyl-cyclopentadienyl)nickel (NiCp*2) were selected as redox mediators83. In this system, a sulfur-based copolymer was located in the external tank, but a conductive agent, such as carbon black, was not used. Significant capacity loss was observed because of the permeation of mediators and dissolved polysulfide through the membrane. More recently, the concept of a multiple-redox, semi-solid–liquid catholyte based on solid sulfur, solid carbon and liquid lithium iodide was reported82. Interestingly, the liquid lithium iodide electrolyte increased the reversible volumetric capacity and the utilization of sulfur, which can be explained by the suppression of polysulfide dissolution by highly concentrated lithium iodide (5 M). Although a volumetric capacity of 550 Ah l−1 was achieved in this system, the practical volumetric density based on the volume of the tank was not described. Only the density of the cathode materials and lithium that participated in the actual discharge reaction were considered. Recent fundamental work to develop more practical lithium–polysulfide RFBs for large-scale applications has considered the solubility of the short-chain polysulfide, which is generally insoluble in organic electrolytes89. Remarkably, deposition of the short-chain polysulfide was suppressed in a tailored dimethyl sulfoxide-based electrolyte with a high concentration of lithium salt (3 M of lithium bis(trifluoromethanesulfonyl)imide), resulting from the strong ionic association force of the Tf− anions. However, the discharge capacity decreased by ∼15% after only 50 cycles at a current density of 0.25 mA cm−2.
The phase of the sulfur catholyte between soluble and insoluble states is dependent on the voltage range and supporting electrolyte. Although it seems that the use of polysulfide with a liquid phase is more practical, the intrinsic problem of internal self-discharge should be addressed for applications in large-scale EES systems. A recently reported polysulfide-blocking microporous polymer membrane is a viable approach84, which highlights the importance of the membrane separator.
Lithium–organic flow batteries. Because of their structural diversity and tunable electrochemical properties, organic electrode materials have long been studied in various battery fields96, in particular, for LIBs, for which the reaction mechanisms and the material properties of the organic compounds have been described in many reviews97,
Lithium–organic flow batteries are attractive as cost-effective EES systems. The aforementioned lithium-based flow batteries that are based on heavy metals, metal complexes or toxic halogens have drawbacks (in particular, the solubility and availability of the redox couples) that hinder their widespread use as large-scale EES systems. The use of organic redox-active materials allows the solubility and operating voltage to be tuned. Among these materials, carbonyl- and nitroxide-based catholyte redox couples have been widely reported: for example, tailored anthraquinone102, 2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPO)103 and 4-methoxy-2,2,6,6-tetra-methylpiperidine-1-oxyl (MeO-TEMPO)104. Despite the short history of organic redox-active materials in flow batteries, remarkable properties have been accomplished: for example, high discharge voltage (>3.9 V)105, high volumetric energy density (∼126 Wh l−1)103 and high solubility (∼2.5 M)104. However, the use of lithium metal as the anode inevitably presents safety issues similar to other systems that use lithium metal, and pretreatment (such as the formation of a passivation layer or polymer coating) is necessary to prevent dendrite growth. In this context, recently reported approaches using lithiated graphite106 (Fig. 3f) and an organic anolyte107 with low redox potential look highly promising.
RFBs based on organic redox couples. Metal-based RFBs inevitably encounter issues related to the scarcity of mineral resources and their rising prices. For example, when the use of VRFBs increases and vanadium resources are limited, the current capital cost of approximately US$450 per kilowatt will significantly increase, which further inhibits the realization of the US Department of Energy (DOE) target cost of less than $150 per kilowatt108. Therefore, inexpensive, metal-free, organic redox-active materials have become widely studied in the flow-battery field. Flow batteries with organometallic materials, such as metallocenes, are good examples of the transition from metal-based to organic-based and metal-free systems109,110. However, most metal–ligand complexes have intrinsic limitations, such as complicated preparation processes or low solubility of the redox couples111. The concept of a non-aqueous, all-organic RFB was first reported in 2011 (Ref. 112). A system configuration of the all-organic RFB is presented in Fig. 4a, which is similar to that of the conventional RFB device, except for the anolyte and catholyte composition. These systems use 0.1 M TEMPO and 0.1 M N-methyl-phthalimide dissolved in acetonitrile with NaClO4 salt as the catholyte and anolyte112. Since this concept was proposed, several types of all-organic chemistries have been reported, including 2,5-di-tert-butyl-1,4-bis(2-methoxyethoxy)benzene, quinoxaline derivatives, polythiophene conducting polymers, 4-oxo-TEMPO and camphorquinone113,
Organic-based aqueous RFBs have been considered as an alternative for grid-scale EES systems118,119. In 2014, an innovative metal-free, organic–inorganic aqueous RFB was reported120. The technology was based on 9,10-anthraquinone-2,7-disulfonic acid (AQDS) and bromine redox couples in sulfuric acid. Notably, it was first proved that a high power density of 600 mW cm−2 could be achieved at a high current density of 1,300 mA cm−2 without any metal redox couples. Although this battery showed capacity retention of 99.2% at a current density of 500 mA cm−2 after 10 cycles, its long-term cycle stability over a prolonged time was not mentioned. Further research121,122 to determine the cycle performance of this battery demonstrated stable capacity retention from 250 to 1,000 mA cm−2 for 40 cycles. However, because of concern about the use of toxic bromine, it was replaced with non-toxic ferricyanide as the positive redox couple (Fe(CN)63−/Fe(CN)64−)123. This novel flow battery was operated in an alkaline solution of potassium hydroxide, with highly soluble hydroxylated anthraquinones on the negative electrode side. A high peak power density of 700 mA cm−2 was observed at 45 °C, which is comparable to that in previous works that used AQDS and bromine redox couples. With respect to the cycle stability, the flow cell showed ∼10% capacity loss after 100 cycles and a volumetric energy density of 6.8 Wh l−1. Unfortunately, in this study, typical charge and discharge curves (that is, voltage as a function of capacity) to confirm the electrochemical behaviour at specific capacities were not provided. In 2015, a low-cost and sustainable all-organic aqueous RFB was reported that operated in benign NaCl solution with methyl viologen and 4-hydroxy-TEMPO as the anolyte and catholyte108 (Fig. 4b). Although 3.0 M methyl viologen anolyte can achieve a specific capacity of 43.2 Ah l−1, the overall performance was limited by the low solubility of the 4-hydroxy-TEMPO catholyte (<0.5 M) in the supporting electrolyte, and the volumetric energy density was only 8.4 Wh l−1. It is important to note that continuous capacity loss is observed in most RFBs that use low-molecular-weight organic compounds. For large-scale EES applications, these issues should be addressed by identifying the specific reason for the capacity decay.
Polymer-based RFBs can be a good alternative, in which the redox-active organic molecules can be branched in the polymer backbone by a copolymerization process124,
Metal–air flow batteries. Metal–air cells are a promising technology that use oxygen as a nearly unlimited natural resource and provide a high theoretical energy density127. Traditional metal–air batteries include a metal plate or foil anode, a separator and an air electrode for the oxygen reduction and evolution reactions (ORR and OER). In these systems, compared with the fast kinetics of anodic reactions, sluggish cathodic reactions, such as the ORR and OER, govern the overall electrochemical reaction as a rate-determining step, accompanying highly reactive electrocatalysts on the air electrode128. These devices usually contain a certain amount of electrolyte in a static cell. Therefore, when operating the cell, by-products are continuously deposited on the surface of the metal anode or the air electrode. These by-product deposits can clog electrode pores and thus reduce the number of active reaction sites129. In addition, although the supply of air is nearly infinite, the energy output is limited by the size of the metal anode. Combining with flow systems can overcome the aforementioned challenges and harness the intrinsic advantages of metal–air batteries. This can be achieved by two approaches: circulating the anode materials in the form of a slurry or a solution of the metal and ions (Fig. 5a); and supplying fresh electrolyte to increase utilization of the metal anode while eliminating by-product precipitation and preventing side reactions on the air electrode (Fig. 5b).
The energy density of VRFBs can be readily doubled by replacing the VO2+/VO2+ redox couple in the catholyte with oxygen molecules130,
Zinc–air cells were originally developed as primary batteries; however, in recent research, bifunctional catalysts for the ORR and OER have been introduced to test their rechargeable properties137, and rechargeable zinc–air flow batteries are in the early stages of development25,138. Mechanically rechargeable zinc–air cells have a long history dating back 40 years139. To avoid the electrochemical charging process in these systems, spent zinc metal is physically replaced with fresh anodes, or a zinc slurry dispersed in alkaline electrolyte is hydraulically refueled by the pump140,141. However, current technologies related to bifunctional catalysts offer the opportunity for charging primary zinc–air cells to some degree, thus making it a secondary battery. Therefore, it makes sense that the circulation of only alkaline electrolyte can be considered. Fortunately, it is well known that electrolyte flow greatly affects the structure of zinc deposits142,143 (Fig. 5c,d). Recently, an electrically rechargeable zinc–air flow battery was reported by circulating potassium hydroxide electrolyte. This system provided a high peak power density of 270 mW cm−2 at a current density of 460 mA cm−2; however, the charge and discharge performance needs to be improved. Finding a suitable bifunctional catalyst, a separator to prevent dendrite formation and a good flow-cell design remain a challenge.
The lithium–air battery, which has a much higher theoretical energy density than other batteries, is considered to be one of the most promising candidates for EES systems144. Since its development145 in 1996, extensive research has been carried out to promote the reaction kinetics and electrochemical reversibility of the oxygen electrode. However, discharge by-products, such as insoluble Li2O2, in the aprotic electrolyte accumulate at the oxygen electrode (known as pore clogging) and block the active reaction sites, thus leading to poor electrochemical performance146. Using protic electrolytes, cathode clogging can be prevented in lithium–air batteries. However, the soluble by-product LiOH can also cause precipitation and is not easily decomposed during the charge process147.
Several systems combining lithium–air batteries with flow systems have been demonstrated. The previously discussed flow concepts used in other batteries, such as redox targeting24, a flowing electrolyte148 and a semi-solid catholyte149, have been tested in lithium–air batteries. For example, the redox-targeting method was used to achieve reversible formation and decomposition of Li2O2, whereby the cathode material was stored in an external tank and redox mediators flowed between the tank and the cell24. This system prevented the deposition of discharge product (Li2O2) inside the cell by remotely forming it in the tank, thereby avoiding the passivation and pore clogging of the electrode. A recently reported semi-solid catholyte consisting of conductive carbon and electrolyte demonstrated dramatically increased discharge capacity with low overpotentials, because the solid-phase by-product during the discharge process was removed from the electrode materials by the flowing catholyte149. Despite such progress in addressing the problem of Li2O2 deposition, the electrochemical performance still needs to be improved.
Solar rechargeable RFBs. Similar to air in a metal–air battery, solar energy has great potential as an almost unlimited energy source16,150. Recent research has focused on the direct storage of energy generated from photovoltaic cells using approaches such as hybrid systems that combine hydrogen production with fuel cells, and solar rechargeable cells. However, sluggish water oxidation in the hydrogen production process and the physically limited space for solid electrodes with slow intercalation reactions have impeded the broad application of solar EES systems. In contrast to the systems mentioned here, liquid-phase redox reactions occurring in RFBs have relatively fast reaction kinetics, which enable higher solar-to-chemical conversion efficiencies.
The VRFB, a representative flow battery, integrated with a photoelectrochemical (PEC) cell was proposed151 in 2012. This technology takes advantage of the fast electrochemical kinetics of V2+/V3+ and VO2+/VO2+ redox pairs. Vanadium electrochemical cells include the photoanode, a platinum counter electrode and a Ag/AgCl reference electrode. All cells were tested in static mode without the flow of electrolyte, and a Nafion ion-exchange membrane was used. Typical photocatalysts, such as TiO2 (Refs 152,153), WO3–TiO2 hybrids154, TiO2 nanobelts155 and multiwalled carbon nanotube/CdS hybrids156, were used to generate photoelectrons and holes, which are located in the catholyte reservoir. The generated hole photoelectrochemically oxidizes the VO2+ ions to VO2+ ions, while the photoelectron moves along the external circuit reducing V3+ ions to V2+ ions. Unfortunately, a low concentration of V3+ and VO2+ redox species (∼0.01 M) dissolved in the 3 M H2SO4 electrolyte was used for photocharge tests and showed a low conversion efficiency153. In addition, considering the acidic environment of vanadium electrolytes, it is difficult to guarantee the long-term stability of metal-based photoanodes.
A solar rechargeable flow battery based on the I−/I3− redox couple, which is widely used in dye-sensitized solar cells (DSSCs), was proposed in 2013 for storing solar energy at large scales157,158. This device included a non-aqueous iodine catholyte with a dye-sensitized TiO2 photoanode, an aqueous anolyte, a platinum-based counter electrode and a ceramic separator. Because a typical DSSC uses a liquid charge-storage electrode, various redox-active anodes, such as [Fe(C10H15)2]+/[Fe(C10H15)2] (Ref. 158), Li2WO4/Li2+xWO4 (Ref. 157) and quinoxaline (C8H6N2/C8H6N2−)159, have been introduced to replace the solid-phase anode used in the PEC system with a flowable liquid-phase anolyte. However, the conversion efficiency of these systems was only 0.15–1.2%, which is largely because of the high-overpotential ceramic solid separators used. The aqueous lithium–iodine RFB that was discussed earlier can also be photo-recharged by a DSSC in a single device23 (Fig. 6a). It was recently demonstrated that, with an external voltage input, the photo-assisted charging process partially reduced the initial charge voltage from 3.6 to 2.9 V by saving 20% of the charging energy (Fig. 6b). Although iodine-based solar rechargeable flow cells show great promise, the use of corrosive halogen elements and brittle ceramic separators with low conductivity remains a challenging issue.
A more ideal approach for large-scale applications is to use aqueous organic-based RFBs as photogenerated EES systems (Fig. 6c). In two recent studies21,22, water-soluble quinone derivatives were used for solar rechargeable cells. As previously mentioned, the fast reaction kinetics and the electrochemical reversibility of functionalization of anthraquinone molecules have already been proven for organic-redox-couple-based RFBs120. Consequently, a solar rechargeable system based on an alkaline RFB with a haematite photoanode (α-Fe2O3) has been proposed. This system includes AQDS and ferrocyanide redox pairs in a NaOH solution21. During photocharging, the Fe(CN)64− and AQDS were oxidized and reduced to Fe(CN)63− and AQDS2− by the generation of electron–hole pairs, whereby the incident solar-to-chemical energy conversion efficiency was about 0.05–0.08%. A more advanced solar rechargeable flow cell was later developed by combining a dual-silicon PEC cell with a AQDS/bromine RFB22. In particular, two different photoelectrodes (carbon/TiO2/Ti/n+p-Si and Pt/p+n-Si) were integrated in a PEC cell in which Br− and AQDS were photocharged to Br3− and AQDSH2, respectively (Fig. 6d). Subsequently, these photocharged species (Fig. 6e) were transferred and discharged in the RFBs, maintaining a discharge capacity of 462 mAh l−1 after 10 photocharge and discharge cycles, with an overall photon–chemical–electricity energy conversion efficiency of ∼3.2%. The development of direct solar RFBs is now emerging; thus, fundamental and systematic studies in this area will probably lead to great success in the EES field.
Challenges facing flow batteries
Recent progress in the research and development of flow batteries has focused on two major aspects: improving system performance (for example, energy and power densities) by finding novel materials and chemistries, including redox couples, electrodes, electrocatalysts and membranes, while maintaining conventional RFB configurations; and integrating flow batteries with other energy storage and conversion systems to achieve a synergistic effect. To overcome the limitations of conventional metal-ion-based aqueous RFBs, great effort has been expended to improve the solubility and operating voltage by introducing a range of metal coordination complexes, semi-solid materials, and organic and inorganic redox couples with aqueous or non-aqueous electrolytes. However, when considering scaling up, which was an initial reason for developing RFBs, it is still difficult to find appropriate materials and systems.
Non-aqueous RFBs with a metal-centred ligand complex can avoid water electrolysis, achieving high operating voltages. However, their extremely low solubility and instability in air are unappealing properties, compared with the common aqueous RFBs. These issues appear in most non-aqueous organic-based RFBs, which have low volumetric energy densities (<10 Wh l−1) and operating voltages (<2 V). Molecular modification of metal complexes or organic materials can increase the solubility in non-aqueous electrolytes; however, this inevitably causes poor ion-transfer kinetics with a corresponding decrease in the operating current density. More importantly, the storage of a flammable organic electrolyte in large external reservoirs can lead to cost and safety concerns, making this system unsuitable for a large-scale EES technology. Accordingly, organometallic coordination complexes in aqueous solution can resolve these scale-up and safety concerns.
Lithium-based flow batteries have been shown to be technologically simple, high-energy and high-voltage systems. Semi-solid or redox targeting methods use conventional cathode and anode materials, thus benefiting from the high energy densities that can be achieved. However, the technical feasibility of a semi-solid system for use in long-term and large-scale energy-storage devices should be carefully evaluated (for example, assessing issues associated with the circulation of a high-viscosity slurry or sedimentation). With respect to redox-targeting methods that only circulate redox mediators, several flow batteries using this concept have demonstrated unprecedentedly high volumetric energy densities (∼500–670 Wh l−1; calculated from the density of the active materials)72,82, which are comparable to those in conventional LIBs.
Reports of novel flow batteries often neglect the total volume of the catholyte and anolyte and focus only on the redox-active materials themselves. The salt or supporting electrolyte is not a removable option and is necessary for the operation of flow batteries. At this stage, the volumetric energy density should be carefully defined for objective and practical comparisons among many flow batteries. When describing cathode and anode materials in flow batteries, the terminology of catholyte and anolyte is usually used because they are dissolved or exist in an electrolyte that can be circulated. Accordingly, the volumetric energy density of conventional RFBs is generally obtained by dividing the total energy by the sum of catholyte and anolyte volumes. This approach, which considers the energy contribution from the specific catholyte and anolyte volumes, enables a fair comparison to be made between traditional and next-generation flow batteries. Other lithium-based flow batteries typically use a catholyte based on organometallic complexes, halogen elements or organic redox-active materials with a lithium-metal anode, and most studies have focused on the development of these catholyte materials. Thus, the next step is to find a promising anolyte material, which will be a key challenge for developing viable flow batteries.
Large quantities of active materials are needed to store the generated energy in grid-scale EES systems. Vanadium and lithium metals are not abundant resources, and therefore sodium and zinc are being considered as alternative materials for use in flow batteries. Sodium-ion batteries, in particular, have attracted much attention as a post-LIB technology. Several sodium-based flow batteries using NaxNi0.22Co0.11Mn0.66O2 and NaTi2(PO4)3 semi-solids160, a liquid sodium-alloy anode161 or ambient-temperature molten sodium with VRFBs162 have been reported. Although this research is in its initial stages, it has great potential in terms of economic sustainability. For novel zinc-based systems, aqueous zinc–polyiodide RFBs have demonstrated a high volumetric energy density of 167 Wh l−1 with 5.0 M ZnI2 electrolyte. A zinc–polymer hybrid flow battery that uses poly(TEMPO) and ZnCl2 solution has also been proposed163,164. Novel copper- and iron-based flow batteries, such as all-copper165,166 or all-iron167,168 RFBs, have also been investigated and might prove to be economically competitive.
Aqueous flow batteries can provide a rapid response time and good flowability of the catholytes and anolytes with minimum pump loss, thus facilitating the storage of the generated energy. Substantial progress has been made in this field by using organic redox couples with fast reaction kinetics, and extremely high peak power densities of 600–700 mW cm−2 have been achieved. This high power density is helpful to reduce the stack size; however, the volumetric energy densities (7–16 Wh l−1) are still lower than those of conventional metal-ion-based aqueous RFBs108, which inevitably increases the external tank size. In other words, there is a trade-off between high energy and high power that must be considered when designing redox couples and supporting electrolytes. In addition, one major advantage that contributes to the long service life of the metal-ion-based aqueous RFBs is their ability to recover from capacity loss through well-designed rebalancing processes. At this stage, one critical challenge facing aqueous RFBs that use organic redox couples is their cycle life. The flexibility and tunability of the organic redox molecules also create concerns about the stability and longevity of their structure. For this reason, detailed capacity decay mechanisms and recovering methods need to be carefully studied and developed.
The reaction kinetics are highly dependent on the nature of the electrode surface169. Indeed, most kinetic studies have been reported in the VRFB field by using a range of electrode materials; however, the reaction kinetics are still not well understood6,34. Thus, fundamental studies of newly reported redox-active materials should be systemically performed to understand the electrode kinetic and reaction mechanism.
The concept of flowing electrolytes in flow batteries can complement the function of solar cells and metal–air cells. Solar rechargeable flow batteries show great promise by converting solar energy and storing photogenerated clean energy in external reservoirs, which is considered the most effective way to realize future EES systems. However, there are still numerous technical problems, such as the low conversion efficiency and energy density, because research integrating these two systems is still at the proof-of-concept stage. Thus, intensive development efforts are needed to identify suitable photoanodes and RFB systems. In the case of metal–air batteries, there are intrinsic limitations, such as dendritic growth and precipitation of insoluble by-products. Fortunately, hybrid metal–air flow batteries have exhibited great potential to solve these issues, thus enhancing the possible use of metal anodes with high power densities. For example, an aluminium–air battery has shown a high theoretical energy density of 8,100 Wh kg−1 (oxygen excluded); however, problems associated with the precipitation of solid by-products, such as Al2O3 and Al(OH)3, self-discharge by hydrogen evolution and a naturally formed thin layer of aluminium oxide in the static cell must be solved. When combined with a flow system, the aluminium–air flow battery may create a synergistic effect, demonstrating unprecedented high energy density and high power density.
We have provided an in-depth review of the state of the art of flow-battery technologies and their limitations. The key requirements for the future development of these systems are summarized in Fig. 7. Finding a suitable candidate for a next-generation, large-scale EES system has triggered an explosion of interest in flow-battery technologies. The level of interest strongly resembles what was referred to as the ‘renaissance of the fuel cell’ in the early 2000s. In this time of burgeoning new technologies, we should carefully evaluate the strengths and weaknesses of each system with a balanced view. In practice, the volumetric energy density based on the total volumes of catholyte and anolyte, and the peak power density, which are both directly connected to the capital cost of the system, should be considered. Finally, on the basis of this Review, we believe that flow-battery technologies, with their high level of design flexibility, have great potential for large-scale EES systems in our future power grid.
This work was supported by the 2016 Research Fund (1.160033.01) of Ulsan National Institute of Science and Technology (UNIST). W.W. acknowledges the financial support from the US Department of Energy (DOE) Office of Electricity Delivery and Energy Reliability (OE) under Contract No.57558 and US DOE Office of Advanced Research Projects Agency-Energy (ARPA-E) through Award DE-AR0000686. Pacific Northwest National Laboratory (PNNL) is operated by Battelle for the DOE under Contract DE-AC05-76RL01830.