Advanced Materials - 2020 - Li - Material Design of Aqueous Redox Flow Batteries Fundamental Challenges and Mitigation

Review
www.advmat.de
Material Design of Aqueous Redox Flow Batteries:

Fundamental Challenges and Mitigation Strategies
Zhejun Li and Yi-Chun Lu*
redox flow systems to date.[7] However,

Redox flow batteries (RFBs) are critical enablers for next-generation grid-scale their high chemical cost (V2O5, $24 kg−1)
energy-storage systems, due to their scalability and flexibility in decoupling and low energy density (Ed < 50 Wh L−1,
power and energy. Aqueous RFBs (ARFBs) using nonflammable electrolytes normalized based on both side of the elec-
are intrinsically safe. However, their development has been limited by their trolytes, unless specified otherwise) limit
their wide implementation for large-scale
low energy density and high cost. Developing ARFBs with higher energy
grid storage. To boost the energy density
density, lower cost, and longer lifespan than the current standard is of of RFBs, nonaqueous RFBs with a wider
significant interest to academic and industrial research communities. Here, potential window were developed. How-
a critical review of the latest progress on advanced electrolyte material ever, nonaqueous RFBs face intrinsic chal-
designs of ARFBs is presented, including a fundamental overview of their lenges such as high cost, flammability,
and low ionic conductivity of organic
physicochemical properties, major challenges, and design strategies.
electrolytes.[8]
Assessment methodologies and metrics for the evaluation of RFB stability Here, we review recent developments of
are discussed. Finally, future directions for material design to realize practical advanced materials for electrolyte design
applications and achieve the commercialization of ARFB energy-storage in ARFBs. We further classified ARFBs
systems are highlighted. based on the type of active material: inor-
ganic ARFBs (i.e., AIRFBs, including vana-
dium-, zinc-, iron-, polysulfide-based, and
1. Introduction two other emerging systems); and organic ARFBs (i.e., AORFBs,
including viologen-, 2,2,6,6-tetramethylpiperidine-N-oxyl
The ever-increasing global population and significant lifestyle (TEMPO)-, quinone-, and N-centered heteroaromatic-molecule-
changes are accelerating global energy demands.[1] High con- based systems). We first summarize the fundamental physico-
sumption of fossil fuels leads to environmental harm and con- chemical properties of redox couples in each ARFB system and
tributes negatively toward climate change.[1,2] In recent years, then discuss critical challenges and mitigation design strategies
continuous cost reduction and broad deployment of renewable to shed light on the design guidelines for future RFBs. Finally,
energy resources have motivated the development of energy- assessment methodologies and metrics for evaluating battery sta-
storage systems to balance the fluctuation of energy supply and bility among different ARFBs are discussed.
user demands.[3] Aqueous redox flow batteries (ARFBs) are one
of the most important candidates for large-scale energy storage.
They can be applied over a wide power range (0.02–50 MW) 2. Aqueous Inorganic Redox Flow Batteries
with long lifetimes (5000–13 000 cycles) over a flexible dis-
(AIRFBs)
charge duration (0.01–10 h).[4] The most prominent advantage
of ARFBs is their intrinsic scalability, due to their unique 2.1. Vanadium-Based AIRFBs
design of decoupled power and energy.[5] In the 1970s, the first
ARFBs were launched by NASA with iron and chromium as 2.1.1. Fundamental Physicochemical Properties
the active materials (ICRFBs); however, they exhibited limited
power stability.[6] In the 1980s, Sum and Skyllas-Kazacos intro- All-vanadium redox flow batteries (VRFBs, Figure 1a) are the
duced all-vanadium RFBs, which are the most well-developed most developed and widely used RFB system to date. Applying
a vanadium element with diverse valence states (i.e., +2, +3,
Dr. Z. Li, Prof. Y.-C. Lu
+4, and +5) effectively minimizes cross contamination, facili-
Electrochemical Energy and Interfaces Laboratory tates electrolyte regeneration, and provides long-term durability
Department of Mechanical and Automation Engineering (Figure 1b–e). For example, a 200 MW/800 MWh system has
The Chinese University of Hong Kong been under construction by Dalian Rongke Power since 2016.[9]
Shatin, N. T., Hong Kong SAR 999077, China The reversible cell voltage of a VRFB is 1.26 V. The nominal
E-mail: yichunlu@mae.cuhk.edu.hk
cell voltage of VRFBs ranges from 1.15 to 1.55 V in acidic elec-
The ORCID identification number(s) for the author(s) of this article
can be found under https://doi.org/10.1002/adma.202002132.
trolytes, since the catholyte potential is pH-dependent.[10] The
redox reactions on both sides are shown in the cyclic voltam-
DOI: 10.1002/adma.202002132 metry (CV) analysis in Figure 1f.[11]
Adv. Mater. 2020, 32, 2002132 2002132 (1 of 30) © 2020 Wiley-VCH GmbH
15214095, 2020, 47, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/adma.202002132 by UFVJM - Universidade Federal Vales Jequitinhonha e Mucuri, Wiley Online Library on [19/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
www.advancedsciencenews.com www.advmat.de
Figure 1. Fundamental electrochemical properties of VRFBs. a) Schematic illustration of VRFBs during the discharge process. Reproduced with permis-
sion.[13] Copyright 2011, Wiley-VCH. b–e) Geometry-optimized structures of deprotonation reaction for pristine [VO2(H2O)3]+ in a traditional electrode
(b), mitigated deprotonation for chlorine-complexed [VO2(Cl)-(H2O)2] in a mixed acid electrolyte (c), pristine [V2+6(H2O)]2+ (d), and chlorine-com-
plexed [V(Cl)-5(H2O)]+ determined via density functional theory (DFT) calculations (e). Bond lengths are in Å. b–e) Reproduced with permission.[12]
Copyright 2014, Wiley-VCH. f) CVs of a graphite-foil electrode measured in 0.5 m VII (purple), VIiI (turquoise), VIV (blue), and VV (yellow) electrolytes
with 2 m H2SO4 at 20 mV s−1. Reproduced with permission.[11] Copyright 2015, American Chemical Society.
VO2+ (IV ) + H2O ↔ VO+2 ( V ) + 2H+ + e− (1)

the reaction. They suggested that higher polarizability of the
*Cl bridge in HCl than the *OH bridge in conventional H2SO4
E = 1.00 V versus SHE ( VSHE )
0
or HCl/H2SO4 rendered an enhanced electron transfer rate.
Roznyatovskaya et al.[14] investigated the role of phosphoric acid
V 3+ (III) + e− ↔ V 2+ (II) E 0 = −0.26 VSHE (2)
in stabilizing catholytes at high temperatures (>35 °C) through
Equations (1) and (2) show the redox reactions for the catho- nuclear magnetic resonance (NMR) spectra. They revealed that
lyte and anolyte, respectively. Vanadium cations in aqueous the formation of two phosphate-containing species effectively
solution normally exist in their hydrated forms: V(IV) and delayed the precipitation of V2O5. The redox kinetics of V2+/V3+
V(V) take the oxo-aqua form [VOy(H2O)6−y](z-2y)+ (z = 4 or 5, and VO2+/VO2+ were investigated through the Tafel relation-
Figure 1b,c), while V(II) and V(III) take the hexa-aqua cation ship. It was hypothesized that VO2+/VO2+ exhibited slower reac-
form [V(H2O)6]z+ (z = 2 or 3, Figure 1d,e) based on geometry tion kinetics due to the involvement of water and protons in the
optimization and thermodynamic calculations.[12] reaction.[17]
To increase the ionic conductivity, concentrated acid sup- The solubility and thermal stability of the VRFB catholyte
porting electrolytes have been utilized including H2SO4, has limited the cell energy density to 20–35 Wh L−1. While
HCl,[13] H3PO4,[14] and methane sulfonic acid (CH3SO3H),[15] vanadium (V) cations reach 3.0 m in 6.0 m sulfate acid below
etc. Counter anions (such as Cl−) can exchange with the H2O 30 °C, precipitation of V2O5 occurs above 40 °C at 1.8 m through
ligand in the primary solvation shell, which has been found an endothermic reaction (Equation (3)). Vanadium (II, III, and
to play a vital role in tuning the thermal and chemical sta- IV) are readily precipitated below 10 °C at 2.0 in 5.0 m sulfuric
bility of hydrated vanadium cations and their reaction path- acid[18]
ways.[12] Agarwal et al.[16] investigated the reaction mechanism
of the V2+/V3+ reaction in HCl, HCl/H2SO4, and H2SO4 elec-
trolytes, and found that a bridging ligand was formed during 2VO2+ + H2O ↔ V2O5 ( s ) + 2H+ (3)
Preparation of VRB electrolytes could use either VOSO4 glycerin, n-propyl alcohol) additives.[18] For instance, bismuth
or V2O5 as a source material via chemical or electrochemical has been used as a nontoxic catalyst for V2+/V3+ anolytes with
methods. V2O5 is cheaper than VOSO4 but exhibits a ten times a high hydrogen evolution overpotential. Li et al.[28] applied
lower solubility than VOSO4. Both methods require a precharge 10 × 10−3 m Bi3+ to a V2+/V3+ anolyte, which increased the reac-
step from a symmetric cell to acquire the desired vanadium tion rates by forming a Bi catalyst on the electrode surface and
electrolytes.[19] increased the EE by 11% at 150 mA cm−2.
Crossover of Vanadium Cations and H2O: Self-discharge of
VRFBs was mainly caused by vanadium cations crossover
2.1.2. Challenges and Mitigation Strategies driven by concentration gradients that were generated during
cell operation. Additionally, H2O migration occurred due to the
Although VRFBs demonstrate unique advantages over other hydration of vanadium cations and protons, as well as osmotic
RFB systems, the development of VRFBs has been challenged pressure due to uneven ionic strengths. To mitigate electrolyte
by the high chemical cost of vanadium, their low thermal sta- crossover, it is important to develop membranes with improved
bility, crossover of vanadium species, and limited redox reac- chemical durability in the presence of strongly acidic or highly
tion kinetics.[20] Therefore, extensive research efforts have been oxidizing electrolytes, and enhanced ionic selectivity without
devoted to improving the electrolyte composition, membrane sacrificing ion conductivity. Perfluorinated cation exchange
properties, and electrode materials. membranes (DuPont Nafion) have been the most widely
Poor Thermal Stability and Solubility: The instability of V(V) employed membranes for VRFBs to date. Nafion consists of a
above 50 °C and other V cations below 10 °C at high concen- hydrophobic tetrafluoroethylene (Teflon) backbone terminated
trations (>1.5 m) has limited the energy density and practical with hydrophilic sulfonate functional groups[29] (Figure 2a).
utilization of VRFBs.[21] Electrolyte modification strategies have The Teflon backbone provides excellent mechanical strength
been reported to improve the solubility and stability of VRFBs, and durability, while the hydrophilic channels surrounded by
such as mixing hydrochloride acid and sulfuric acid for use as sulfonated groups ensure proton transportation through the
a supporting electrolyte. Development of mixed-acid-electrolyte- Grotthuss mechanism.[29] However, Nafion membranes have
based VRFBs increased the stable temperature window (−10 to exhibited low ionic selectivity, high resistance (e.g., 0.95 Ω cm2
50 °C) under high concentrations up to 2.5 m.[13] Wang and co- for N115 vs 0.16−0.67 Ω cm2 for porous membranes[30]) and
workers [12,13] reported that the addition of chloride counter ions high costs ($600–800 m−2). Nonfluorinated membranes (e.g.,
led to chloride-complexed V(V), preventing V2O5 formation polyoxadiazole, polysulfone, sulfonated poly (ether ether
and impeding the nucleation process of V (II, III, IV) at low ketone) (SPEEK)) have effectively decreased the membrane cost
temperatures. ($40 m−2). However, their chemical stability was compromised
To address the high volatility and high causticity of chlo- due to the presence of charged functional groups.[31]
rine released from HCl electrolytes, Roe et al.[22] reported a Recently, porous membranes have been reported to exhibit
3.0 m V(V) solution stabilized by using 1 wt% H3PO4 and 2 wt% good reversibility, making them a promising alternative to
ammonium sulfate as the electrolyte. This strategy operated Nafion membranes. As shown in Figure 2b, porous nanofil-
for 90 cycles (state of charge (SoC) of 62%) with no significant tration membranes demonstrated effective vanadium cation
capacity decay at 80 mA cm−2 and an energy efficiency (EE) of exclusion via the pore size sieving effect, due to the difference
≈70%. The stabilization mechanism was attributed to the com- between the Stokes radii of vanadium cations and protons.[32]
plexing process of NH4+ or PO43− with VO2+ and their absorp- Zhang et al.[32] obtained polyacrylonitrile (PAN) porous mem-
tion on V2O5 nuclei which resulted in charge repulsion.[23] branes via the phase inversion method. Compared to VRFBs
Other additives such as organic compounds containing two or using Nafion N115 (coulombic efficiency (CE) of 93.4%, EE of
more –OH, –NH2 and –SH moieties, or surfactants, were also 82.3%), VRFBs with PAN membranes exhibited a higher CE of
reported to be effective at reducing the instability of vanadium 95% but a lower EE of 76% at 80 mA cm−2.
electrolytes.[21] Various strategies have been applied to improve the ion
Limited Redox Kinetics and Parasitic Reactions: Catholyte reac- selectivity and proton conductivity of porous membranes
tions in VRFBs suffered from slow reaction kinetics because including: 1) filling pores with hydrophilic nanoparticles
of the multiple steps involved and H2O molecule rearrange- (e.g., silica, zeolite) to reduce the pore size;[33] and 2) intro-
ment.[17] Anolyte reactions (V2+/V3+) were negatively impacted ducing charged groups to facilitate proton conduction.[30,34]
by parasitic hydrogen evolution reactions (HER, especially in Negatively charged groups, such as sulfonic acid, provided
strong acidic environments). Therefore, various strategies have additional Grotthuss pathways for proton transportation but
been proposed to facilitate the reaction kinetics and suppress were less effective at blocking vanadium cations. Positively
parasitic reactions. One such strategy is the functionalization charged N-containing groups were shown to promote proton
of carbon materials via the surface treatment of graphite felt, transportation in acidic conditions while also reducing vana-
e.g., oxidation using a strong acid or base,[24] or thermal acti- dium cation penetration.[35] For instance, advanced porous
vation,[25] which were reported to improve the EE of VRFBs.[26] membranes with a crosslinked network of sponge-like
In addition, decoration of hybrid carbon nanomaterials (e.g., poly(ether sulfone) (PES) were fabricated by introducing
graphene oxide (GO), or multiwalled carbon nanotubes positive imidazole groups on the pore walls named CMPSF-
(MWCNTs)) enhanced the redox kinetics of the VO2+/VO2+ X (where X is the crosslinking time in the imidazole solu-
reaction.[27] Another effective strategy involved employing trace tion (Figure 2c)).[34] Proton conductivity was achieved via
amounts of inorganic (e.g., Sb3+, Bi3+, In3+) or organic (e.g., the Grotthuss mechanism using the positively charged
Figure 2. Membranes for VRFBs. a) Schematics of the water channel model for a Nafion membrane. Reproduced with permission.[29] Copyright 2008,
Springer Nature. b) Schematics of a nanofiltration porous membrane design for VRFBs. Reproduced with permission.[32] Copyright 2011, The Royal
Society of Chemistry. c) Schematics of an advanced, charged, sponge-like membrane with an internal crosslinked network in VRFBs. d) VRFB perfor-
mance of CMPSF-72 and N115 at various temperatures; and e) cycle performance of CMPSF-72. c–e) Reproduced with permission.[34] Copyright 2015,
Wiley-VCH. f,g) Schematics of a sponge-like porous PBI membrane (f) and the proton conducting mechanism of a PBI membrane (g). h) The cycling
performance of VRFBs using PBI-34 and PBI-68 as membranes at different temperatures. i) A long-term stability test of VRFBs with PBI-68 at 80–120
mA cm−2. f–i) Reproduced with permission.[30] Copyright 2016, The Royal Society of Chemistry.
N-group. Blockage of vanadium was realized via the Donnan tunability and stability that could be utilized in future VRFB
exclusion effect and pore size exclusion. In this example, a applications.
99% CE and an 86% EE were demonstrated for VRFBs with
CMPSF-72 at 80 mA cm−2. The high efficiency achieved by
CMPSF-72 was consistent across a wide temperature range 2.2. Zinc-Based AIRFBs
from −5 to 50 °C (Figure 2d). The VRFBs with CMPSF-72
operated for 6000 cycles at 120 mA cm−2 with a negligible 2.2.1. Fundamental Physicochemical Properties
decay in efficiency (Figure 2e). Yuan et al.[30] further opti-
mized the chemical stability of porous membranes by using Zinc is an attractive anode material that has been widely applied
a novel heterocyclic polybenzimidazole (PBI-x, where x is in aqueous Zn-ion,[36] Zn-based hybrid flow batteries,[37] and Zn–
the membrane thickness) polymer (Figure 2f,g) containing air batteries.[38] Zinc’s intrinsic merits include: 1) a sufficiently
positive N-based groups. This strategy also enhanced the low redox potential (i.e., −0.76 VSHE for Zn2+/Zn in acidic/neu-
electronic repulsion of vanadium cations based on Donnan tral media, −1.26 VSHE for [Zn(OH)4)]2−/Zn in alkaline media);
exclusion, leading to the highest reported efficiencies over a 2) moderate reversibility with fast kinetics (k0 for [Zn(OH)4)]2−/Zn
wide operating temperature range (−5 to 50 °C, Figure 2h). is 2.5 × 10−4 cm s−1); 3) low chemical cost ($1.9 kg−1); and 4) high
VRFBs with PBI-68 exhibited an ultralong durability for volumetric capacity (e.g., volumetric capacity for highly water sol-
>13 500 cycles (Figure 2i). In summary, porous mem- uble ZnI2 is 249 Ah L−1 and ≈37.5 Ah L−1 for [Zn(OH)4)]2−).
branes drastically decreased the cost of VRFBs compared to Redox reactions of zinc are dependent on the pH of the
Nafion membranes and provided opportunities for higher aqueous electrolytes. In Figure 3a, a general Pourbaix diagram
Figure 3. Electrochemical properties of the Zn redox reactions in aqueous media. a) Pourbaix diagram of the Zn reactions. b) The distribution of zinc
species as a function of the pH of the electrolytes. b) Reproduced with permission.[42] Copyright 2016, Elsevier Ltd. c) The electrolyte conductivity,
current density, and solubility of ZnO as a function of KOH concentration. Reproduced with permission.[38] Copyright 2017, Wiley-VCH. d) CVs of
the zinc metal in ammonium chloride solutions at various pH values. Reproduced with permission.[40] Copyright 2017, The Electrochemical Society.
e) Schematic illustration of processes involved in zinc electrochemical deposition. f,g) Top-view scanning electron microscopy (SEM) images of the
Zn electrode cycled at 7.5 mA cm−2 (f) and 10 mA cm−2 (g). f,g) Reproduced with permission.[43] Copyright 2019, Wiley-VCH. h,i) SEM images of zinc-
plated washers at 40 mA cm−2 (h) and 60 mA cm−2 (i) at pH = 1. h,i) Reproduced with permission.[44] Copyright 2018, Springer Nature. j–n) SEM images
of zinc dendrite obtained from 0.7 m zincate–7 m KOH under 100 mA cm−2, 5 min (j); 60 mA cm−2, 5 min (k); 30 mA cm−2, 20 min (l); 10 mA cm−2,
0.5 min (m); 10 mA cm−2, 5 min (n). j–n) Reproduced with permission.[39] Copyright 2009, Elsevier B.V.
of Zn is presented (10−6 m Zn2+), revealing the equilibrium of ZnI2 (7.0 m ZnI2, 2/3 electrons per I−, amounting to 9.3 m
potential alongside the chemical equilibrium among various electrons). The demonstrated ZIB could stably operate over 40
Zn2+ hydrolysis products (Figure 3b). Bare Zn2+ stabilized at cycles at 10 mA cm−2 with a CE of 96.3% (Figure 4c). To further
a standard equilibrium potential at −0.763 VSHE in an acidic increase the energy density of the ZIB, which was limited by
environment (pH < 8). In neutral media (pH = 8–10), the I−/I3−, Weng et al.[46] introduced Br− as the complexing agent to
main reactions were electrochemical and chemical deposition stabilized I2 in water (Figure 4d), further boosting the energy
of the hydrolysis product Zn(OH)2. When pH > 10 (i.e., mild to 202 Wh Lcatholyte−1 (5.0 m ZnI2, 2.5 m ZnBr2, 2 electrons per
alkaline), Zn(OH)2 dissolved in the alkaline forming zincate I−, amounting to 20 m electrons). The Zn–I2 Br battery (ZIBB)
ions (i.e., HZnO2− or [Zn(OH)4)]2−). Further increasing the stably operated for 20 cycles at 10 mA cm−2 with an energy
pH of the electrolyte (pH > 13) led to stabilization of zincate at density of 70 Wh L−1 (Figure 4e,f). Xie et al.[47] made a single
−1.2 VSHE.[39] The redox reactivity of Zn2+/Zn was also related flow Zn–I2 battery (ZISFB) using 6.0 m KI–3.0 m ZnBr2. The
to the electrolyte pH. Both electrolyte conductivity and redox catholyte was static and involved I2 deposition (Figure 4g–i).
activity peaked when the concentration of KOH was in the opti- Another possible strategy for raising the energy density is to
mized range (6–7 m). However, the solubility of ZnO linearly increase the cell voltage. Zhang et al.[48] replaced the neutral
increased with pH.[38] Song et al.[40] studied the redox behavior anolyte (Zn2+/Zn) with a Zn plate in an alkaline electrolyte as
of Zn in various pH ranges (pH = 3, 7, and 10, Figure 3d). The the anode (Figure 4j), which raised the cell voltage to 1.8 V (i.e.,
potential of the zinc redox reactions decreased from −0.9 VSHE 0.4 V higher than the neutral system, Figure 4k). However, the
to −1.2 VSHE when the pH increased from 3 to 10 (in line with stability of the alkaline Zn–I2 system was limited due to 1) the
the Pourbaix diagram). A nucleation potential was observed for limited selectivity of the Nafion membrane in preventing OH−
all cases indicating the extra energy penalty required for phase transportation, and 2) serious dendrite formation of the Zn
transformation. electrode in alkaline media. The system operated for 70 cycles
Fundamental processes and issues involved in zinc electro- with an EE of 80% at a current density of 10 mA cm−2, where
chemical deposition in acidic, neutral (pH ≤ 7), and alkaline the Zn electrode was replaced every 3–4 cycles during cycling
medium (pH > 7) are depicted in Figure 3e. During the charging (Figure 4l).
process, Zn2+ (pH ≤ 7) and Zn(OH)42− (pH > 7) reduced to form Despite the promising energy density of Zn–I2 RFBs, the
Zn nuclei on electrode surfaces, acting as the center for den- high cost of ZnI2 ($43 kg−1) and limited lifespan associated
drite growth. The fast ion consumption around the tip of Zn with Zn electrodes hindered their practical applications. Alka-
deposits generated concentration gradients leading to den- line Zn–iron RFB systems were initially reported by G.B.
drite growth. The morphologies of Zn deposits were sensitive Adams et al. in 1981,[49] but were limited by the low solubility
to the current density (or overpotential) and the areal capacity of Na4Fe(CN)6 catholytes (0.2–0.5 m) and the fast capacity decay
(Figure 3f–n). When a small current (corresponding to a low of alkaline Zn anolytes. Yuan et al.[50] applied Na4Fe(CN)6 to a
overpotential) was applied, thermodynamically stable Zn with KOH supporting electrolyte, and achieved 1.0 m Na4Fe(CN)6
a higher crystallinity was observed. For example, in mild acid/ in 3.0 m KOH (Figure 5a–d) and optimized the stability
neutral Zn-ion batteries (Figure 3f,g), a plate-like crystal struc- through membrane design. To avoid using dilute ferrocya-
ture was observed at current densities <10 mA cm−2. Dendritic nide as the catholyte, Gong et al.[51] introduced an amphoteric
structures were formed when higher current densities were Zn–Fe RFB, using one anion exchange membrane (AEM)
applied (Figure 3h,i). In alkaline environments, the mechanism and one cation exchange membrane (CEM) in three electro-
was different (Figure 3j–n). The morphology of Zn deposits lytes (i.e., Zn(OH)42−/Zn|NaCl|Fe3+/Fe2+. The demonstrated
ranged from heavy spongy, dendrite, boulder, layer-like, and cell had a high theoretical voltage of 1.99 V and stably oper-
finally to a mossy structure as the current density ranged from ated over 20 cycles with a CE of 99.9% and an EE of 75.9% at
high (100 mA cm−2) to low (10 mA cm−2).[41] The decay of the Zn 80 mA cm−2. The cost of the Zn–Fe amphoteric RFB was pro-
electrode was more severe in an alkaline environment than in jected to be $150 kWh−1 (40 mA cm−2, 74% EE at system level)
a neutral environment due to Zn mossy/dendrite formation.[38] since the costs of both electrolytes were low, which meets the
2023 target of the U.S. Department of Energy (DoE). To avoid
the utilization of alkaline Zn electrolytes, which posed den-
2.2.2. Challenges and Mitigation Strategies dritic concerns, Xie et al.[52] and Selverston et al.[53] used acidic
electrolytes (i.e., 1.5 m H2SO4, 1.0 m FeCl2|| 1.0 m ZnSO4, 1.5 m
Catholyte Candidate Selection: Various high-energy-density acetate (HAc/NaAc))[52] and symmetric mild acidic electrolytes
systems with high power capabilities have been developed that (i.e., 1.6 m ZnCl2, 0.8 m FeCl2, 2.0 m NH4Cl)[53] for Zn–Fe bat-
employ Zn redox reactions as a hybrid anolyte for flow battery teries, respectively. The acidic Zn–Fe was mainly challenged by
applications. Generally, areal current densities ranging from the fast capacity decay (50 cycles) due to severe HER in acidic
10 to 200 mA cm−2 have been achieved for Zn-based AIRFBs over media.
a wide pH range (Table 1). Taking advantage of the low redox Poor Stability: Zn-based AIRFBs have been challenged by the
potential of Zn, Zn–Ce and Zn–Br AIRFBs have been devel- following Zn electrodeposition and parasitic reactions as sum-
oped with cell voltages of 2.0 and 1.85 V, respectively, which are marized in Figure 3e: 1) hydrogen evolution reaction (HER);
higher than the stability window of water. Due to the hazardous 2) electrode passivation; and 3) the formation of dendritic struc-
nature of Br2, an ambipolar Zn–I2 battery (ZIB, Figure 4a,b) was tures. As indicated in the Pourbaix diagram (Figure 3a), HER is
introduced by Li et al.,[45] which demonstrated 167 Wh Lcatholyte−1 a thermodynamically favorable process for a Zn-based electrode
(≈4 times that of conventional VRFBs) due to the high solubility although a large overpotential is required for its initiation. In
Table 1. Performance of Zn-based ARFBs.
Electrolyte composition Ecell [V] J [mA cm−2] Membrane Cycle life Ed,achieveda) CE
[Wh LCatholyte−1] EE
Zn–Br 2 m ZnBr2–3 m KCl–0.8 m MEP 1.85 20–100 Daramic HP 142[56] NA 97.2%
NA
2.25 m ZnBr2 + 0.5 m ZnCl2 + 0.8 m MEP+ Br2 ≈1.6 10–40 Nafion-SF-600 166[57] NA 94.7%
82.1%
Zn–I2 (Zn–I2Br) 5.0 m ZnI2 0.96 10 Nafion (N115) 40 (99% energy retention)[45] 167 96.3%
67.8%
7.5 m KI–3.75 m ZnBr2 1.35 20–80 Polyolefin with 500 160 97%
Or Nafion (@80 mA cm−2)[47] 81%
6.0 m KI–3.0 m ZnBr2
6 m KI–3.0 m ZnBr2 1.33 60–180 Carbon-coated 1000 (>2100 h) 80 94%
porous polyolefin (@80 mA cm−2)[55] 80%
5 m ZnI2–2.5 m ZnBr2 1.35 5–10 Two N117 20 (@10 mA cm−2)[46] 202 95%
NA
Alkaline Zn–I2 Zn plate in 6 m KOH||6 m KI-6 m I2 1.796 10–20 N117 10 (200 h) 200 100%
(@20 mA cm−2) [48] 80%
Alkaline Zn–Fe 1.0 m Na4Fe(CN)6–3 m KOH| 1.74 80–160 PBI membrane 210 (@100 mA cm−2)[50a] 34 99.48%
0.5 m Zn(OH)42−–4 m NaOH 8.90%
0.4 m Zn(OH)42−+3 m NaOH|| 1.81 35–160 PES+SPEEK (P20) 240 (@80–160 mA cm−2)[50b] 25 99.64%
0.8 m Na4Fe(CN)6+3m KOH 87.72%
@80 mA cm−2
Zn–Fe 0.3 m Na2[Zn(OH)4]–0.5 m NaCl–2.4 m NaOH|| ≈1.7 50–150 N212-FAA-3 20 (@80 mA cm−2)[51] ≈22 99.9%
0.6 m FeCl2–0.5 m NaCl–1 m HCl 75.9%
(SoC75%)
Acidic Zn–Fe 1 m ZnSO4–1.5 m HAc/NaAc|| 1.53 30 HZ115 ion exchange 50[52] NA 91.8%
1 m FeCl2–1.5 m H2SO4 membrane 71.1%
1.6 m ZnCl2–0.8 m FeCl2–2 MNH4Cl–2g L−1 PEG8000 1.2 50 Daramic 175 127, 240 h[53] NA 85%
68%
Neutral Zn–Fe 0.8 m ZnBr2–2.0 m KCl|| 1.4 20–80 PBI porous 100 (1.6 m @40 mA cm−2)[54] ≈40 ≈98%%
1.6 m FeCl2–3.2 m glycine–2.0 m KCl membrane 84%
a)Based on catholyte volume only.
mildly acidic or neutral media, the electrolysis of H2O led to neutral/mildly acidic symmetric cell systems (e.g., Zn–Br2/I2),
brittle Zn deposits (“dead Zn”) and an increased pH around the CEMs (e.g., Nafion) and porous membranes were employed.
electrode surface. These processes facilitated the hydrolysis of Nafion N115 was first applied in Zn–I2 RFBs. However, both
Zn2+ which formed an insulating ZnO layer on the electrode the power capability (5–20 mA cm−2) and the capacity reten-
surface, causing low coulombic efficiency and low reversibility tion (50 cycles) were limited by low ionic conductivity and
of Zn-based batteries. In alkaline media, [Zn(OH)4]2− easily mechanical strength of the Nafion membrane. Porous poly-
decomposed to ZnO in highly concentrated electrolytes, which olefin membranes with low costs and high ionic conduc-
led to a passivation of the electrode surface and generation of tivity are promising alternatives to Nafion for Zn–I2 RFBs.
“dead Zn” after repetitive cycling. Xie et al.[55] coated a 4 µm thick carbon layer onto a porous pol-
The driving force for dendrite growth, particularly at high yolefin membrane, filling it with oxidized I3−, which corroded
current densities (high overpotential), resulted from the con- pierced zinc dendrite and prevented short-circuiting at high
centration gradients around the nuclei due to the continuous current densities. Demonstrated “self-healing” high-energy-
consumption of Zn2+ or [Zn(OH)4]2− on the interface of the density Zn–I2 batteries (80 Wh LCatholyte−1) demonstrated stable
electrolyte and electrode. One of the major disadvantages cycle lives >1000 cycles without efficiency decay over more
caused by dendrite formation was short-circuiting due to den- than 3 months at 80 mA cm−2 with an EE of 80%. In addition,
drite penetration. To mitigate dendrite failure of Zn-based batteries with porous polyolefin membranes demonstrated a
AIRFBs, multiple strategies have been applied for membrane high Zn areal capacity >90 mAh cm–2 at 40 mA cm−2. To fur-
design: 1) interrupting the concentration gradient around the ther improve the stability of Zn–I2 batteries at lower rates, Xie
dendrite tips; and 2) increasing the mechanical or chemical sta- et al.[47] coated a Nafion layer on the porous polyolefin mem-
bility of the membrane. brane. The Zn–I2 single flow system displayed a higher CE
Here, we review membrane developments toward of 97% at 40 mA cm−2 and stably operated for 500 cycles at
addressing dendrite formation in Zn. For Zn2+ dominated 80 mA cm−2 with zero efficiency decay at 160 Wh LCatholyte−1.
Figure 4. Zn–I2 redox flow batteries. a) Schematic of ambipolar Zn–I2 batteries (ZIBs); b) charge/discharge curves for the cell with 5.0 m ZnI2 and
Nafion 115 as the membrane at 5 mA cm−2; and c) cycling performances of the Zn–I2 cell at 10 mA cm−2. a–c) Reproduced under the terms of the CC-BY
Creative Commons Attribution 4.0 International license (https://creativecommons.org/licenses/by/4.0).[45] Copyright 2015, Springer Nature. d) Sche-
matic of the Zn–I2Br batteries (ZIBBs); e) galvanostatic voltage profiles of the ZIBB systems with 5 m ZnI2 + 2.5 m ZnBr2 at a flow rate of 10 mL min−1
at 5 mA cm−2; and f) cycling retention of a ZIBB with 5 m ZnI2 + 2.5 m ZnBr2 at 10 mA cm−2 and an SoC of ≈70%. d−f) Reproduced with permission.[46]
Copyright 2017, The Royal Society of Chemistry. g) Schematic of single-flow Zn–I2 batteries (ZISFBs); h) the charge–discharge plot of a 7.5 m ZISFB at
20 mA cm−2 with a porous polyolefin–Nafion composite membrane; and i) cycling performance of the ZISFB with 7.5 m electrolyte with the composite
membrane at 20 mA cm−2. g–i) Reproduced with permission.[47] Copyright 2019, The Royal Society of Chemistry. j) Schematic of alkaline Zn–I2 bat-
teries; k) representative galvanostatic charge and discharge curves at different electrolyte concentrations at 20 mA cm−2; and l) cycling performance of
experimental alkaline Zn–I2 batteries. j–l) Reproduced with permission.[48] Copyright 2018, The Royal Society of Chemistry.
In alkaline Zn-based batteries, Yuan et al.[50b] designed a repulse the zincate ions from filling the pores of the mem-
negatively charged, nanoporous poly(ether sulfone)/sulfonated brane. The mutual repulsion between the negatively charged
poly(ether ether ketone) (PES/SPEEK) membrane to physically zincate ions and the negatively charged membrane caused
Figure 5. Zn–Fe redox flow batteries. a) Schematic of alkaline Zn-flow batteries with charge-repulsion membranes; b) cycle performance of the alkaline
Zn–Fe flow battery using a P20 membrane. a,b) Reproduced under the terms of the CC-BY Creative Commons Attribution 4.0 International License
(http://creativecommons.org/licenses/by/4.0).[50b] Copyright 2018, The Authors, published by Springer Nature. c) Schematic of a PBI membrane for
an alkaline Zn–Fe flow battery; d) the cycling performance of an alkaline Zn–Fe flow battery at 100 mA cm−2. c,d) Reproduced with permission.[50a]
Copyright 2018, Elsevier. e) Schematic of neutral Zn–Fe redox flow batteries; f) charge–discharge behavior of the flow battery at 40 mA cm−2; and
g) cycling performance of the battery with 1.6 m FeCl2. e–g) Reproduced with permission.[54] Copyright 2017, Wiley-VCH.
the dendrites to grow backward (Figure 5a). The SPEEK con- 2.3. Iron-Based AIRFBs
tent in the polymer membrane was 20% by weight (P20). The
cell stably operated for 240 cycles at 80–160 mA cm−2 with a 2.3.1. Fundamental Physicochemical Properties
CE of 99.64% and an EE of 87.72% at 80 mA cm−2 (Figure 5b).
In addition, Yuan et al.[50a] employed a sponge-like PBI mem- Various iron-based redox flow batteries have been developed in
brane which had a much higher mechanical strength than recent decades owing to the natural abundance and low cost
the Nafion membrane (i.e., the elasticity modulus of PBI of iron (¢0.13 per Ah for Fe[51]), including all-Fe,[58] Zn–Fe[50a],
was >2.9 GPa vs 94 MPa for N115) (Figure 5c,d). The alkaline S–Fe,[59] and organic-Fe.[60] The first ICRFBs were composed
Zn–Fe flow battery using the PBI membrane exhibited a higher of a Cr2+/Cr3+ anolyte (E0 = −0.41 VSHE) and a Fe2+/Fe3+ catho-
CE and EE than for N115 (i.e., a CE of 99.97% and an EE of lyte (E0 = 0.77 VSHE) in hydrochloric acid with an overall cell
88.11% vs a CE of 99.2% and an EE of 78.83% for N115) over an voltage of 1.18 V. Development of ICRFBs has been limited by
extended lifetime of 240 cycles, due to fast OH− transportation the sluggish reaction kinetics of Cr2+/Cr3+, parasitic HER, and
via the Grotthuss mechanism and high mechanical stability the crossover of active materials. Hybrid all-iron redox flow
(Figure 5d). batteries, featuring low toxicity and low material cost, were
developed with ferrous cations as the starting materials for both addition, an increased size of the Fe complex resulted in lower
sides crossover rates. As shown in Figure 6b, organic ligands like
citrate, glycine, and cyanide were reported to effectively com-
Fe2+ ↔ Fe3+ + e− , E 0 = 0.77 VSHE (4) plex with free Fe2+ and Fe3+ cations.[58,62] Among these organic
ligands, glycine was superior for enhancing the complexing sol-
Fe2+ + 2e− ↔ Fe, E 0 = −0.44 VSHE (5)
ubility (>2.0 m FeCl2–4.0 m glycine in 2.0 m NaCl[54]) at a sim-
Equations (4) and (5) correspond to the catholyte and anolyte, ilar potential as for Fe2+/Fe3+ (≈0.77 VSHE, Figure 5e–g). Ferri/
respectively. Combining a 1.21 V cell voltage and a high solu- ferrocyanide redox couples functioned well under neutral/
bility of FeCl2/FeCl3, an energy density of 76 Wh L−1 was alkaline media at 0.36 VSHE, where they exhibited high stability
expected for the all-iron RFBs. However, the all-iron RFBs were and fast kinetics (0.25 cm s−1 vs Fe2+/Fe3+ at 1.2 × 10−4 cm s−1).
limited by the intrinsic conflict of the selected pH range. As Robb et al.[61] introduced an innovative ICRFB by applying a
shown in the Pourbaix diagram in Figure 6a, the redox poten- 1,3-propylenediaminetertaacetic acid (PDTA) chelated CrPDTA/
tial and the type of the most stabilized Fe-containing species KCrPDTA anolyte (−1.1 VSHE, which was 700 mV lower than for
were dependent on the pH of the aqueous electrolytes. The Cr3+/Cr2+) with a [Fe(CN)6]3−/4− catholyte. This demonstrated a
equilibrium potential of Fe2+/Fe was ≈0.6 V lower than that of high voltage of 2.13 V with negligible H2 evolution and capacity
HER, especially for low pH values (e.g., pH = 0), which makes loss over 100 cycles at 0.1 A cm−2. An alternative Fe-based redox
the HER side reactions thermodynamically favorable in acidic active material, iron-triethanolamine (Fe[(TEOA)OH]−/2− or
Fe-based RFB systems. On the contrary, Fe3+ easily underwent Fe-TEA), exhibited the lowest potential among all of the Fe
hydrolysis when pH > 2, making it inappropriate for pairing complexes (i.e., −0.86 VSHE) with a solubility of 0.8 m in alka-
with Fe2+/Fe. line environments, making it a suitable anolyte for ARFBs.[58,63]
Introducing metal chelates (e.g., Fe, Cr, V, Ce, etc.) enabled Arroyo-Currás et al.[63] coupled a Fe-TEA anolyte with a
the manipulation of the electrolyte stability, redox potential, Co(mTEA) catholyte in 5 m NaOH to give an overall cell voltage
and solubility.[58,61] The iron–organic complex was found to of 0.93 V. A flow cell with 0.5 m active species was demonstrated
effectively suppress iron cation hydrolysis, facilitating Fe3+/ for 30 cycles and a 4% crossover at 30 mA cm−2. Gong et al.[58]
Fe2+ kinetics as well as manipulating the redox potential. In developed an all-iron RFB based on Fe[(TEOA)OH]−/2− as the
Figure 6. a) Simplified potential–pH Pourbaix diagrams of iron compounds. Reproduced with permission.[73] Copyright 2018, Elsevier B.V. b) Redox
pairs of iron complexes that have been tested for RFB applications. c) Water solubility diagram of the ferrocyanide and ferricyanide salts; d) extended
1000 cycle stability of the 0.9 m (NH4)4[Fe(CN)6]–(SPr)2 V symmetric AORFB. c,d) Reproduced with permission.[60] Copyright 2018, Elsevier Inc.
anolyte and [Fe(CN)6]3+/4+ as the catholyte with a cell voltage up Another challenge for iron–organic compounds has been
to 1.34 V. The cell displayed stable cycling for 110 cycles without poor stability. Recently, Luo et al.[68] revisited the stability of
noticeable capacity decay at 40 mA cm−2. [Fe(CN)6]3−/4− in different pH mediums, and revealed that
K3Fe(CN)6/K4Fe(CN)6 in strong alkaline suffers from the CN−/
OH− ligand exchange reaction. However, the instability of ferri/
2.3.2. Challenges and Mitigation Strategies ferrocyanide in alkaline media has been debated. Páez et al.[69]
reported that the intrinsic stability of ferrocyanide was more
Challenges for Ligand-Free Fe/Fe2+ Anolytes and Fe2+/Fe3+ stable than previous reports. They suggested that the imbal-
Catholytes: The redox reaction of Fe2+/Fe3+ is associated with the ance of the cell system (e.g., SoC[70]) may have contributed to
hydrolysis of Fe2+/3+ cations, generating precipitates like Fe2O3 the degradation. Cazot et al.[71] suggested that a reversible dimi-
(Figure 6a).[64,73] Therefore, pH management is critical for long- nution of kinetics and membrane degradation was the major
term stability. For instance, Xie et al.[52] utilized a 1.5 m HAc/ reason for the battery instability. Ferrocene/ferrocenium deriva-
NaAc buffer electrolyte to maintain pH values in the range of tives were prone to undergo bimolecular decomposition under
2–6 to circumvent HER in acidic Zn–Fe RFBs (1–1.5 m H2SO4). high concentrations, especially when exposed to oxygen.[72] Beh
The cell operated for 202 cycles until the capacity decayed to et al.[72] designed bis((3-trimethylammonio)propyl)-ferrocene
zero, which was attributed to crossover of the active materials. dichloride (BTMAP-Fc) as the catholyte for ARFBs (1.9 m sol-
The challenge of using an Fe/Fe2+ anolyte in RFBs is the for- uble in water, 0.39 VSHE), which achieved a higher chemical sta-
mation of solid Fe inside the cell stack, which compromises bility and reduced permeability compared to previous reports
scalability. Additionally, the sluggish kinetics and parasitic HER due to the higher charge density per molecule.
have required the use of catalysts to facilitate Fe2+ reduction
and suppress H2 generation. Finally, cross contamination of
cations (i.e., Fe2+/3+) through CEMs has been demonstrated to 2.4. Polysulfide-Based AIRFBs
accelerate capacity decay. Considering AEMs typically exhibit
high resistance, developing symmetric electrolytes with porous 2.4.1. Fundamental Physicochemical Properties
membranes is a promising future avenue of research (e.g.,
symmetric Fe/V,[65] symmetric Zn/Fe[53]). Sulfur-based materials have been widely applied in energy-
Challenges for Iron–Organic Complex as an Electrolyte: The sol- storage applications including metal–sulfur batteries,[74] solar
ubility of iron–organic complexes in aqueous electrolytes limits cells,[75] and redox flow batteries,[74] due to their reversible elec-
the energy density of ARFBs. Ferri/ferrocyanide [Fe(CN)6]3−/4− trochemical reactivity, high solubility, and earth abundance. In
is the most widely used iron chelate ion as a catholyte in aqueous media, polysulfides (Sn2−) with a short chain length
ARFBs. It was first used in Zn/K4Fe(CN)6 in 1979. Yuan et al.[50] (1 ≤ n < 4) were observed to be highly soluble (e.g., K2S, 8.8 m
reported that the low solubility of K4Fe(CN)6 or Na4Fe(CN)6 was in 1 m KOH), while long-chain polysulfides and elemental sulfur
due to the “common ion effect” in its corresponding hydroxide were less soluble or insoluble in water.[76] Polysulfides could be
electrolyte. Therefore, they applied Na4Fe(CN)6 in a KOH sup- stabilized in alkaline media from the formation of H2S in neu-
porting electrolyte, which increased the solubility to 1.0 m in tral and acidic environments. The equilibrium potential of poly-
3.0 m KOH (Figure 5a–d). Luo et al.[60] used cation modulation sulfide was pH-dependent for pH < 11.5, but remained consistent
to enhance the solubility of ferri/ferrocyanide. They reported at –0.51 VSHE when pH > 11.5 according to the reported Pourbaix
that the solubility of [Fe(CN)6]3+/4+ increased from 0.56 m diagram.[77] The reaction of polysulfide in water is as follows[78]
in Na-type to 1.3 m in NH4-type (Figure 6c). They also mixed
0.9 m (NH4)2[Fe(CN)6] with a 0.9 m 1,10-bis(3-sulfonatopropyl)- S22− + 2H2O + 2e − ↔ 2HS− + 2OH− ; E 0 = −0.51VSHE (6)
4,40-bipyridinium ((SPr)2 V) anolyte to form a symmetric
cell. The cell operated for 1000 cycles in 1100 h with 100% Aqueous polysulfide promises a theoretical volumetric
capacity retention (12.05 Ah L−1) and an average EE of 62.6% at capacity of 353.8 Ah L−1, resulting from its multiple electron
40 mA cm−2 (Figure 6d). transference and high solubility. The CV analysis of poly-
Metallocenes such as ferrocene/ferrocenium (Fc/Fc+) are sulfide on gold electrode is shown in Figure 7b. The voltage gap
a class of compounds whose metal ion center (Fe, Co, etc.) between the oxidation and reduction reactions was as high as
is bound to two aromatic cyclopentadienyl anions (C5H5−). 381 mV, indicating its sluggish kinetics in aqueous media. Poly-
Fc/Fc+ and its derivatives are Fe2+/Fe3+ centered redox cou- sulfide was easily disproportionated in the electrolyte. Li et al.[79]
ples that undergo reversible one-electron transfer at 0.4 VSHE. applied in operando UV–vis spectra to evaluate the chemical
Most derivatives of ferrocene are hydrophobic, leading to very reversibility of polysulfide species including S22−, S42−, and S2−.
low solubility in aqueous electrolytes. To address this chal- The electrochemical reactions and chemical disproportionation
lenge, Hu et al.[66] introduced hydrophilic electron-withdrawing of polysulfide were found reversible after one CV cycle.
ammonium substituents to form FcNCl, which increased the Developing cost-effective, sulfur-based ARFBs have received
volumetric capacity to 107.2 Ah L−1 in water and 80.4 Ah L−1 significant attention in recent years due to the low material cost
in 2.0 m NaCl. In addition, Yu et al.[67] introduced ferrocene of sulfur ($0.29 kg−1). Polysulfide–polyhalide ARFBs were devel-
bis(propylsodiumsulfite) (Fc-SO3Na), which reached a solubility oped by Regenesys Ltd. in 1991.[80] For polysulfide–bromide
of 2.5 m in water (67 Ah L−1). The 1.5 m flow battery, combining redox flow batteries (PSBs), Br− is oxidized to Br2 and is con-
a Zn anode, stably operated for 60 cycles with no capacity currently complexed to form Br3− during the charging process,
fading. while polysulfide is reduced to a lower order species. During
Figure 7. a) A schematic illustration of a PSIB; b) CV analyses of K2S2–KCl (blue) and KI–KCl (red) at 5 mV s−1; and c) cycling retention of 4 m KI|3 m
K2S2–PSIB cells using N115 and N117 as the separator (SoC 80%) at 15 mA cm−2. a–c) Reproduced with permission.[79] Copyright 2016, Elsevier Ltd.
d) CV analysis in 1.0 m NaCl–0.05 m Na2S2 on GC and on Co–GC at 20 mV s−1; e) voltage curves of Fe/S RFB at different current densities; and f) the
cycling capacities and efficiencies with respect to the cycle number. d–f) Reproduced with permission.[59] Copyright 2015, The Electrochemical Society.
g) Schematic of the air-breathing S–O2 battery during the discharging process; h) polarization curve of the S–O2 cell at 55 °C under a 1 mL min−1
fluid flow rate; and i) voltage–time curves at room temperature and 0.325 mA cm−2 for 40 cycles, corresponding to 960 h of continuous operation.
g–i) Reproduced with permission.[82] Copyright 2017, Elsevier Inc.
the discharging process, the electrochemical reactions reverse polysulfide, the PSIB demonstrated a much lower chemical cost
on each electrode. Despite a high reversible cell voltage (1.35 V), ($85.4 kWh−1) than comparable VRFBs ($152.0–154.6 kWh−1)
the intrinsically hazardous nature and limited solubility of Br−/ with a similar energy density of 43.1 Wh L−1. Wei et al.[59] reported
Br2 (0.21 m) limit wide application of PSBs. Yi and co-workers a low-cost polysulfide–iron ARFB by applying [Fe(CN)6]4−/3− as
conducted electrode design for a polysulfide-based redox flow the catholyte. The voltage gap between [Fe(CN)6]4−/3− and poly-
battery,[81] and demonstrated a PSB prototype that operated for sulfide (S22−/S42−) was 0.91 V (Figure 7d–f). Stable cell per-
50 cycles over 600 h with a steady energy efficiency at an SoC formances were obtained for 1.0 m K3Fe(CN)6–1.0 m Na2S2 at
ranging from 20–25%.[81]Analogous to PSBs, Li et al.[79] demon- 20 mA cm−2 achieving ≈7.5 Ah L−1 for 100 cycles with a CE
strated a polysulfide–iodide ARFB (PSIB) by replacing Br2/Br− of 99% and an EE of 74% (Figure 7e–f). However, the energy
with highly soluble I−/I3−, which demonstrated faster kinetics density of the S/Fe RFB was limited by the low solubility of
and a lower vapor pressure (i.e., I2 < Br2) (Figure 7a). The PSIB [Fe(CN)6]4−/3− (1.0 m) and Na2S2 in a neutral environment
stably operated for 50 cycles with no capacity decay at 80% SoC (2.1 m). The neutral Na2S2 anolyte used in this work facili-
at 15 mA cm−2 (Figure 7c). Due to the high solubility of both tated the hydrolysis of sulfide anions, leading to battery decay.
polysulfide and iodide (i.e., 3.3 m K2S2–1.0 m KOH and 6 m Li et al.[82] further decreased the cost of sulfur-based RFBs by
KI, with capacity limited by KI) and the low chemical cost of introducing an air-breathing S–O2 RFB (Figure 7g). The system
demonstrated an ultralow installed cost compared to current for 50 cycles at 1.5 mA cm−2 with a CE of 98.6% and an EE
energy conversion and storage technologies, based on the cost of ≈50% with a volumetric capacity of ≈26.8 Ah Lcatholyte−1
model developed by Darling et al.[8b] The air-breathing S–O2 (50% SoC). The PSI battery operated for 100 cycles (>1600 h)
battery consisted of 1 m Li2S4–1.0 m LiOH as the anolyte and at 0.5 mA cm−2 and 50% SoC (≈12 Ah Lcatholyte−1) with a CE of
0.5 m Li2SO4–0.5 m H2SO4 as the catholyte, sandwiched by a 98.7% and an EE of 80% (Figure 8b,c). The power capability
solid-state electrolyte (LiSICON) as a separator to avoid mixing was compromised by the low ionic conductivity of the SSE
of the alkaline anolyte and acidic catholyte. However, its power (LATP: ≈1.0 × 10−4 S cm−1 and NASICON: ≈1.0 × 10−3 S cm−1).
capability was lower than that of conventional aqueous poly- They demonstrated a peak power of 2.8 and 3.9 mW cm−2 for
sulfide-based RFBs (i.e., the peak power was only 3.2 mW cm−2 PSI and PSB, respectively, which was about 1–2 orders of mag-
at 6 mA cm−2 in Figure 7h) due to the low ionic conductivity of nitude lower than that of conventional ARFBs. In addition, the
the solid-state electrolyte and the slow reaction kinetics of both instability of the SSE in highly corrosive polyhalide electrolytes
electrolytes. The cell operated at 0.325 mA cm−2 at an elevated led to capacity decay. Future design strategies are required to
temperature (55 °C) for 40 cycles over 960 h (Figure 7i). enhance the stability of polysulfide-based ARFBs without sacri-
ficing their power capability.
Limited Kinetics of Polysulfide Electrochemical Reactions: The
2.4.2. Challenges and Mitigation Strategies sluggish electrochemical reaction of polysulfide limits the
power density and energy efficiency of polysulfide AIRFBs.
Crossover of Polysulfide Species: Polysulfide ARFBs suffer from Therefore, it is critical to develop active catalysts to promote
severe polysulfide crossover resulting from concentration gra- charge transfer kinetics of polysulfide in aqueous media.
dients and limited ionic selectivity of CEMs (e.g., Nafion). In Zhao et al.[81a] applied nickel foam pretreated with Na2S4 as
addition, crossover of active materials promoted water migra- the electrode in a PSB. The cell obtained an EE of 77.5% at
tion and further accelerated capacity decay. To alleviate the 40 mA cm−2 owing to the catalyst NiSx which formed on the
fast capacity decay of polysulfide-based RFBs, Li et al.[79] and current collector. Several metal sulfides such as CuS, CoS, and
Ma et al.[20b] applied two layers of membranes (i.e., one N117 PbS have been reported to be effective electrocatalysts for poly-
and one N115) for a PSIB system however, this compromised sulfide conversion reactions. For example, Fan et al.[85] investi-
the voltage efficiency (VE) and electrolyte utilization. Su gated transition metal dichalcogenides such as MoS2, WS2, and
et al.[83] replenished the electrolyte every 50 cycles to recover Cu-doped MoS2 as catalyst candidates for polysulfide reactions
the capacity, achieving 200 cycles with three replenishments. and identified that the Cu-doped MoS2 catalyst exhibited fast
Gross et al.[84] applied a solid state electrolyte (SSE, Na+-ion kinetics and high stability over the cycling process. Ma et al.[20b]
conductor NASICON (Na3Zr2Si2PO12) and a Li+-ion conductor applied highly active nanostructured CoS2/CoS heterojunction
LATP (Li1+x+yAlxTi2−xP3−ySiyO12)) for a polysulfide–halide (Br−/ electrocatalysts for polysulfide/iodide ARFBs (Figure 8d). The
Br2 and I−/I3−) RFB (Figure 8a). The PSB cell stably operated peak power of the PSIB with a CoS2/CoS electrode reached
Figure 8. a) Schematic of a polysulfide–polyiodide battery with a Na+ solid-state electrolyte (SSE); b) galvanostatic cycling, with the retention of cycling
efficiencies (CE, VE, and EE); and c) discharge polarization curves. a–c) Reproduced with permission.[84] Copyright 2019, American Chemical Society.
d) Scheme of the SIFB cell configuration; e) the galvanostatic cycling energy efficiency of the SIFBs at 20 mA cm−2; and f) polarization curves of the
SIFBs at 50% SoC. d–f) Reproduced under the terms of the CC-BY Creative Commons Attribution 4.0 International License (https://creativecommons.
org/licenses/by/4.0).[20b] Copyright 2019, The Authors, published by Springer Nature.
86.2 mW cm−2, which was much higher than that of a graphite type [P2W18O62]6− anion (Figure 9f–i), which achieved 17.2
felt electrode (8.9 mW cm−2) (Figure 8f). The authors suggested electron transfer (96% charge) at 100 × 10−3 m and a pH = 1.5
that the CoS2/CoS n–n heterojunction with an uneven charge (Figure 9g). As shown in Figure 9h, [P2W18O62]6− demonstrated
distribution enhanced the absorption strength of the charged four reversible waves within the range of –0.6–0.6 VSHE. The
ions and promoted charge transfer kinetics of both I−/I3− and Li6[P2W18O62]–H2SO4|HBr–Br2 battery stably operated for
S2−/Sx2−. The cell with the CoS2/CoS catalyst operated for 20 cycles delivering 210 Ah L−1 (0.5 m system) at 100 mA cm−2
60 cycles at 20 mA cm−2 and 50% SoC with an 84% capacity with a CE of 98% (Figure 9i). Further investigations on sup-
retention and an EE of 71.6% (Figure 8e). pressing parasitic HER and reducing viscosity under high con-
centration are required to achieve a higher cycling stability. In
addition, a less corrosive catholyte with a comparable volumetric
2.5. Other Emerging AIRFBs capacity to Br−/Br2 should be developed to achieve higher sta-
bility and energy density. A similar system was demonstrated
2.5.1. Polyoxometalate (POM)-Based AIRFBs by Feng et al.,[92] which applied H3PW12O40 (PTA) as an anolyte
and HI as a catholyte (Figure 9d). PTA has a solubility of 0.72 m
POMs are a class of metal-oxide polyanion clusters formed by in water and undergoes a reversible 4-electron redox reaction.
linkage of d0 metal-centered polyhedral and oxygen atoms. They As shown in Figure 9e, three waves involving 1, 1, and 2 elec-
have been widely applied in energy research including for electrons were observed at 0.222, –0.033, and –0.348 VSHE, respec-
trocatalysis, photocatalysis, proton conversion, and recharge- tively. The authors mixed the positive and negative couples and
able batteries (e.g., Li/Na-ion and redox flow batteries).[86] limited the charge voltage to improve the CE and cycling sta-
Symmetric POM-Based AIRFBs: Pratt et al.[87] first intro- bility, which had been limited by: 1) the imbalance of osmotic
duced a vanadium-substituted tungsten-based Keggin POM pressure of the catholyte and anolyte; 2) I− permeation; and
K6H[A-α-SiV3W9O40] (ASi) as both the catholyte and anolyte 3) parasitic HER. The 1.6 m HI–0.25 m PTA|1.1 m HI–0.25 m
for ARFBs (i.e., a symmetric cell). The solubility of ASi was PTA system stably operated for 700 cycles with a capacity reten-
0.45 m in water and could be doubled by exchanging K+ with tion of 99.81% per day, a CE of 99.6%, and an EE of 80.1% at
Li+.[87] Two-step, two-electron redox reactions occurred in the 100 mA cm−2 (charge cutoff voltage = 0.84 V). However, the
anolyte on the WVI↔WVI/V center at 0.3 and 0.7 VSHE, respec- overall energy density of the PTA–HI RFB (2.06 Wh L−1) was
tively. One-electron and two-electron redox reactions occurred low due to the low discharge cell voltage (0.2 and 0.5 V). The
in the catholyte on the VV↔VVI center at 1.3 and 1.7 VSHE, reduction potential of the Keggin-type heteropolyacid (HPA)
respectively. Overall, a symmetric cell with a theoretical cell was found to increase with the increasing electronegativity of
voltage of 1.0 V and a total 3 electron transfer was constructed. the counter-cations (e.g., H (χ = 2.2) > Zn(χ = 1.65)) and heter-
The cell operated at 20 × 10−3 m in 0.5 m H2SO4 for 100 cycles oatoms (e.g., P (χ = 2.19) > Si (χ = 1.9)), or the decreasing elec-
with a CE < 97% and an EE of 50%. The same research group tronegativity of polyatoms (e.g., W (χ = 2.36) > Mo (χ = 2.16) > V
further investigated the influences of heteroatoms (phospho- (χ = 1.63)).[93] Based on these trends, Friedl et al.[94] first intro-
rous vs silicon) and substitution patterns (corner- vs edge- duced a POM material, i.e., [PV14O42]9− (PV14), that could store
sharing) on the CE by comparing A-α-PV3W9O406− (AP), multiple electrons, which was suitable for ARFB applica-
B-α-PV3W9O406− (BP), and P2V3W15O629− (WD).[88] They found tions. They developed an asymmetric POM-based RFB with
that a Si heteroatom and edge-sharing offered higher revers- [SiW12O40]4− (SiW12) as the anolyte (E0 of 0.01, –0.21, –0.37 VSHE,
ibility than a P heteroatom and corner-sharing, respectively. 4e− involved) and PV14 as the catholyte (E0 of 0.6 VSHE, 4e−
In addition, they revealed that the Keggin-type POMs demon- involved), which cycled for 155 cycles over 14 days. In summary,
strated higher stability than the Wells–Dawson type due to their while POMs exhibit high structural diversity, they are limited by
lower charge density, which resulted in higher stability in the their cycling stability and intrinsic solubility. Rational designs of
reduced form.[88] The size of the POM molecules also affected POMs with desirable properties for RFBs are needed to further
the charge density. Pratt et al.[89] compared cis-V2W4O194− develop high-energy-density and stable POM-based ARFBs.
(VW) and (SiFe3W9(OH)3O34)2(OH)311− (SiFeW), and found
that SiFeW, which had a larger molecule size, exhibited more
stable cycling due to its lower charge density. Recently, Liu 2.5.2. Redox Targeting AIRFBs
et al.[90] improved symmetric POM-based ARFBs by developing
an “all-H6[CoW12O40] RFB” (Figure 9a). The α-Keggin-type Dual-Molecule Redox Targeting AIRFBs: The energy density
[CoW12O40]6− anion had a solubility of 1.0 m in H2O at 25 °C. of conventional RFBs is limited by the solubility of the redox
The catholyte redox reaction occurred on the CoII↔CoIII center active species in the electrolyte. Wang and co-workers[95] pro-
at an E0 value of 1.103 VSHE. The anolyte redox reactions took posed the concept of redox-targeting-based redox flow batteries
place on the WVI↔WV center at –0.074 VSHE and –0.191 VSHE, (RTRFB), which utilize dissolved redox mediators (RMs) to
respectively (for a two-step, two-electron process, with a total of wire the solid battery electrode materials in the tank. The first
4 electrons) (Figure 9b). The all-H6[CoW12O40] RFB exhibited a reported, non-aqueous RTRFB full cell[95c] employed two types
CE of 99% and an EE of 86% at 25 mA cm−2 with an energy of reactions that occurred on each electrode: 1) electrochemical
density of 15.4 Wh L−1 for 30 cycles at a volumetric ratio of 4:1 reactions of the RMs which determined the cell voltage and
(catholyte:anolyte) to balance electrons (Figure 9c). power capability; and 2) redox targeting chemical reactions
Asymmetric POM-Based AIRFBs: Chen et al.[91] reported between the RMs and solid electrode materials in the reservoir
an asymmetric POM-based ARFB based on a Dawson which decided the energy capacity. The volumetric capacities of
Figure 9. Performance of POM-based ARFBs. a) Schematic of an all-H6[CoW12O40] RFB; b) CV analysis of H6[CoW12O40]; and c) the battery perfor-
mance of the asymmetrical 0.8 m all-H6[CoW12O40] RFB with a volume ratio of 4:1. a–c) Reproduced with permission.[90a] Copyright 2016, Wiley-VCH.
d) Schematic of a PTA|HI RFB; e) CV analyses of PTA and HI–HClO4. d,e) Reproduced with permission.[92] Copyright 2019, Elsevier B.V. f) Schematic
of a Li6[P2W18O62]|HBr redox flow battery; g) relationship between polyoxometalates concentration, solution pH, and the number of electrons; h) CV
analyses of Li6[P2W18O62]–H2SO4 (blue) and HBr–H2SO4 (orange) at 10 mV s−1; and i) cycling performance of Li6[P2W18O62]|HBr at 0.1 A cm−2. f–i) Repro-
duced with permission.[91] Copyright 2018, Springer Nature.
the cathode and anode were 613 and 603 Ah L−1, respectively, identical potential to the corresponding solid material (e.g.,
0 0
based on their condensed states, which amounted to a high LiFePO4, E RM = E LiFePO 4
)
energy density of 500 Wh L−1 (ten times that of VRBs). A pair
of RMs with appropriate redox potentials were employed for RT aRM+
RM ↔ RM+ + e − ; E RM = E RM
0
+ ln (7)
both the anode (cobaltocene (CoCp2, 2.10 V) > TiO2 (1.82 V) > F aRM
bis(pentamethylcyclopentadienyl)cobalt (CoCp*2, 1.67 V)) and
the cathode (FcBr2 (3.78 V) > LiFePO4 (3.45 V) > Fc (3.4 V)). LiFePO4 + RM+ ↔ FePO4 + RM + Li+ ; E LiFePO4 (8)
During the discharging process, Fc in the catholyte was reduced 0 RT
on the positive electrode and pumped to the tank, which was = E LiFePO + ln a Li+
4
F
then reoxidized by LiFePO4. Concurrently, CoCp2 in the anolyte
was oxidized on the negative electrode and subsequently The potential difference between the RM and LiFePO4 was
reduced by the solid TiO2 solid electrode in the reservoir. A
high energy density was demonstrated within the two-molecule RT a +
∆E = ln RM (9)
redox system however, a high voltage loss was observed leading F aRMaLi +
to a low VE (42–91%.
Single-Molecule Redox-Targeting AIRFBs: To improve the VE where E0 is the standard potential, a is the activity, T is tem-
and drive full utilization of solid materials in the reservoir, perature, R is the gas constant, and F is the Faraday constant.
Zhou et al.[96] developed a Nernstian-potential-driven RTRFB, As shown in Figure 10a, electrons were transferred from
in which single molecules were applied for both the charging p-type LiFePO4 to the LUMO of RM+ during charge. The reac-
and discharging processes of each electrode. For single-mole- tions were reversed during the discharging process. Based on
cule redox-targeting (SMRT) reactions, the RM exhibited an this mechanism, 1-ferrocenylmethyl-3-methylimidazolium
Figure 10. Working mechanism and performance of redox targeting ARFBs. a) Energy diagram and charge transfer of the SMRT reactions. b) CV anal-
yses of FcIL with/without FePO4/LiFePO4 (1:1); and c) voltage profiles of flow cells with FcIL–LiFePO4. a–c) Reproduced with permission.[96] Copyright
2017, Elsevier Inc. d) CV analyses of PB and [Fe(CN)6]3−; e) voltage profiles before and after adding PB; and f) prolonged cycling. d–f) Reproduced with
permission.[97b] Copyright 2019, Elsevier Inc.
bis(trifluoromethanesulfonyl)amide, a ferrocene-grafted 3. Aqueous Organic Redox Flow Batteries

ionic liquid (FcIL+/FcIL) has been applied as a mediator for (AORFBs)
LiFeIIPO4/FeIIIPO4 (3.43 VLi/Li+, Figure 10b). The half-cell
demonstrated an increased VE of 95% with a high solid mate- Organic molecules are promising electrolyte candidates for
rial utilization ratio of 95% at 0.025 mA cm−2. novel ARFBs due to their large earth abundance and high tun-
Aqueous RTRFBs based on SMRT have been developed with ability through functionalization. Electron-donating groups
an enhanced rate capability.[97] A full redox-targeting RTRFB (e.g., –OH, –OCH3) can negatively shift the redox potential
was demonstrated using [Fe(CN)6]4−/3− and S2−/S22− as the which however, leaves the reaction center susceptible to elec-
redox mediators for the catholyte (LiFePO4, LFP) and anolyte trophilic substitution.[99] Electron-withdrawing groups (e.g.,
(LiTi2(PO4)3, LTP), respectively.[97a] The cell stably operated for –SO3H) can positively shift the redox potential which however,
55 cycles (99.1% capacity retention) at 5 mA cm−2 with 76 Ah L−1 reduces the stability of the reaction core and makes it subject to
at a discharge potential of 0.7 V. Wang et al. further decreased nucleophilic attack by OH−/H2O.[100] Polar groups enhance the
the cost of the solid materials by applying Prussian blue solubility of organic molecules in water.[101] Bulky groups (e.g.,
(Fe4[Fe(CN)6]3, PB).[97b] PB has two redox reactions centered at trimethylammonium, phosphonate group, etc.) prevent bimo-
0.25 V (PB↔PW (Prussian white, reduced form of PB)) and lecular decomposition of organic species.[72,102]
0.92 V (PB↔BG (Berlin green), both vs Ag/AgCl), as shown
in Figure 10d–f, which coincide with the RM [Fe(CN)6]4−/3− and
Br−/Br2 redox reactions, respectively. The [Fe(CN)6]3−/Br−–PB 3.1. Viologen-Based AORFBs
full cell operated for >40 cycles at 20 mA cm−2, 41.2 Wh L−1, and
a peak power of 118 mW cm−2. The authors coupled K3Fe(CN)6– 3.1.1. Fundamental Physicochemical Properties
PB with [Zn(OH)4]2− as the anolyte, such that the cell deliv-
ered 1.73 V during the discharging process at 20 mA cm−2 for Viologen (Vi2+) and its derivatives (Figure 11a) are highly soluble
500 h over 180 cycles along with a CE of 99.7% and an EE of (3.0 m in 1.5 m NaCl, 43.2 Ah L−1)[101] in aqueous electrolytes.
87.5% (Figure 10e,f). The energy density of this SMRT system Methyl viologen (3.1a, MVi2+) and ethyl viologen (3.1b, EVi2+)
amounted to 97.4 Wh L−1 with a peak power of 264.6 mW cm−2, are the two most commonly used viologen-based compounds
indicating that this is a promising direction for high-energy- in AORFBs. The solubility and hydrophilicity of viologen com-
density systems. Future work on using intrinsically present pounds were found to decrease with increasing size of substit-
RMs for redox-targeting RFBs will further improve the voltage uent group on the N atoms.[101] For instance, the solubility of
efficiency and simplify the system operation.[98] MViCl2 with a methyl group on N (3.0 m) was higher than that
Figure 11. Viologen as the anolyte for aqueous redox flow batteries. a) Multiple derivatives of viologen and the challenges involved for its application in
aqueous RFBs; b) electrochemical reactions and the chemical decomposition mechanism of viologen; c) CV analyses of 3.1d (−0.35 V and −0.72 V) and
NMe–TEMPO (1.0 V) in 0.5 m NaCl aqueous electrolyte; and d) extended 50-cycle data of the 3.1d/NMe–TEMPO AORFB at 60 mA cm−2. c,d) Reproduced
with permission.[102b] Copyright 2017, Elsevier Inc. e) CV analyses of 3.1j (−0.44 VSHE) versus NMe–TEMPO in 0.5 m NaCl solution (1.0 VSHE, red); and
f) cycling stability of 0.1 m 3.1j|0.2 m NMe–TEMPO in 2.0 m NaCl at 40 mA cm−2. e,f) Reproduced with permission.[104] Copyright 2018, Wiley-VCH.
of EViI2 with an ethyl group (2.0 m), both of which were higher propyl]-4,40-bipyridinium trichloride, 3.1c, [(Me)(NPr)V]Cl3 and
than that of 4,4-dibenzyl bipyridinium dichloride (BViCl2) 1,10-bis[3-(trimethylammonio) propyl]4,40-bipyridinium tetra-
(40 × 10−3 m) which had the largest substituent group. Viologen bromide, 3.1d, [(NPr)2V] Br4) and revealed that the reversibility
takes two one-electron steps upon reduction (Figure 11b). The of Vi/Vi0 was enhanced with an improved capacity utilization.
reduction of Vi2+ to Vi+ was shown to be highly reversible at They combined 0.05 m 3.1d, [(NPr)2V] Cl4–2.0 m NaCl with
a potential of −0.5 VSHE with fast kinetics (e.g., MVi, k0 of 0.1 m NMe-TEMPO–2.0 m NaCl to make a full cell, which oper-
2.8 × 10−4 cm s−1) and a high diffusion coefficient (e.g., MVi, D ated for 50 cycles at 60 mA cm−2 with two cell voltage plateaus
of 2.57 × 10−5 cm2 s−1).[101] Liu et al.[101] reported a 0.5 m MViCl2| at 1.35 and 1.72 V, respectively (Figure 11c,d). Liu et al.[103] intro-
4-OH-TEMPO flow cell with an energy density of 8.4 Wh L−1 at duced a hydroxyl group (–OH) to make (1,1′)-di(2-ethanol)-4,4′-
60 mA cm−2. The cell stably operated for 100 cycles at 71.5% at bipyridinium dibromide (3.1h, (2HO-V)Br2), which successfully
its theoretical capacity with an EE of 62.1%. The reduction of increased the solubility of Vi0 from <10 × 10−3 m to 2.1 m in water.
Vi+/Vi0 occurred at −0.7 VSHE, which was less reversible com- Consequently, the reaction kinetics and diffusion coefficient of
pared to Vi2+/Vi+ due to the low solubility of Vi0 (<10 × 10−3 m). Vi/Vi0 were improved (k0 >0.312 cm s−1; D = 3.99 × 10−6 cm2 s−1)
such that they were comparable to Vi2+/Vi+ (k0 >0.356 cm−1;
D = 5.19 × 10−6 cm2 s−1). Luo et al.[104] further linked the two
3.1.2. Challenges and Mitigation Strategies pyridinium moieties of viologen by a π-conjugated framework
hiazolo[5,4-d]thiazole (TTz), 4,4′(thiazolo[5,4-d]thiazole-2,5-diyl)
Limited Utilization of the Second Electron: The insoluble bis(1-(3(trimethylammonio)propyl)pyridin-1-ium) tetrachloride
nature of neutral charged Vi0 in aqueous environments limits (3.1j, [(NPr)2TTz]Cl4), which successfully reduced the voltage
the full utilization of viologen. Introducing an ionic func- gap between the two charge–transfer steps (i.e., −0.38 VSHE
tional group (e.g., hydrophilic ammonium and sulfonate and −0.5 VSHE, Figure 11e). The demonstrated AORFB (0.1 m
groups) can effectively enhance the polarity and hydrophilicity 3.1j–2.0 m NaCl|0.2 m NMe-TEMPO–2.0 m NaCl) demonstrated
of Vi0. DeBruler et al.[102b] introduced one and two ammo- stable cycling for 300 cycles at 40 mA cm−2 with 99.97% capacity
nium groups into MVi (1-methyl-10-[3-(trimethylammonio) retention per cycle, a 100% CE and a 70% EE (Figure 11f).
Bimolecular Decomposition: Viologen and its derivatives density (227 mW cm−2) and energy density (30.35 Wh L−1,
are subject to bimolecular decomposition, particularly under 1.51 V) for the 1.5 m (SPr)2V|Br− AORFB. For long-term sta-
high concentration. As summarized in Figure 11b, the mono- bility, they demonstrated a 1.0 m (SPr)2V|Br− RFB which exhib-
cation radical MVi+ is prone to dimerize and is subsequently ited a 100% capacity retention for 50 cycles (75 h) with an EE
disproportionate to MVi2+ and MVi0 (low solubility), leading of 80% and a CE of 99.12% at 40 mA cm−2. A summary of
to irreversible capacity decay. Numerous charged functional the reported works of viologen-based AORFBs can be found
groups that exhibit steric effects have been reported to avoid in Table 2.
the dimerization of Vi+.[72,102a] For instance, viologen derivatives
with two bulky quaternary ammonium moieties were reported
to improve viologen stability in battery applications.[72,102b] 3.2. TEMPO-Based AORFBs
Combining viologen with a ferrocene-based catholyte, Beh
et al.[72] demonstrated a full cell with an unprecedentedly low 3.2.1. Fundamental Physicochemical Properties
capacity decay rate of 0.0057% per cycle and 0.1% per day at a
high concentration of 1.3 m. TEMPO (3.2a) is an N-oxyl-containing compound that has
Limited Selection of Membrane and Catholyte: Viologen-based been widely applied as an antioxidant in biology, a mediator
redox flow batteries have a limited selection of membranes and in polymerization, as a catalyst, and a charge storage mate-
catholytes. Active viologen ions are positively charged (Vi2+/Vi+), rial.[108] However, TEMPO is not soluble in aqueous media,
which prevents the use of CEMs and some highly concentrated, which limits its use in ARFBs.[109] However, functionalization
negatively charged catholytes (e.g., halogen, KI, or NH4Br, >5 m). of hydrophilic functional groups that used TEMPO increased
Liu and co-workers[105] introduced a negatively charged sul- the solubility to >1.0 m, making TEMPO-based materials
fonate-functionalized viologen 1,1′-bis(3-sulfonatopropyl)- promising candidates for ARFBs (Figure 12a,b). Liu et al.[101]
4,4′bipyridinium (3.1f, (SPr)2V) as the anolyte, combined applied 4-hydroxy2,2,6,6-tetramethylpiperidin-1-oxyl (3.2b,
with KI as the catholyte. This strategy allowed the use of 4-OH-TEMPO or TEMPOL) as the catholyte and coupled it
CEMs (e.g., Nafion 212 or Selemion CSO), which are more with viologen to make an organic ARFB. The AFRB exhibited
conductive than AEMs. They demonstrated a full cell 0.5 m a solubility of 2.1 m in H2O and 0.5 m in 1.5 m NaCl with one
(SPr)2V–2.0 m KCl|2.0 m KI–2.0 m KCl using a low-cost Sele- electron transfer from a nitroxide radical to oxoammonium
mion CSO CEM, which reached a 67% EE and a 98.8% capacity salts (TEMPO+) at 0.8 VSHE (Figure 12c(i)). The reaction rate
retention (99.99% per cycle) for 100 cycles at 60 mA cm−2. Luo constant k0 was 2.6 × 10−4 cm s−1 and the diffusion coefficient
et al.[105b] further demonstrated a full cell using Nafion as the D was 2.95 × 10−5 cm2 s−1.[101] The 0.5 m 3.2b|MVi full cell stably
membrane and NH4Br as the catholyte, which had a higher operated for 100 cycles (89% capacity retention) at 60 mA cm−2
redox potential than KI. The cell demonstrated a high power with a CE of 99% and an EE of 62.5%.[101]
Table 2. Summary of performance of viologen derivatives as the anolyte in ARFBs.
Sol. [m] k0 [cm s−1] E0 [VSHE] Catholyte Ecell [V] J [mA cm−2] Cycle life (decay rate) Ed, achieved CE
D [cm2 s−1] Pa) [mW cm−2] [Wh L−1] EE
3.1a 3.0 m in 2.8 × 10−4 −0.45 4-OH-TEMPO 1.25 60 100 (0.5 m, 0.12%/cycle 8.4 99.4%
MVi2+ 1.5 m NaCl 2.57 × 10−5 NA 3.5%/day)[101] 62.1%
FcNCl 1.06 60 700 (0.5 m, 0.01%/cycle)[66] 7 98%

125 60%
3.1d 1.3 m in 2.2 × 10−2 −0.39 (first) BTMAP-Fc 0.75 50 500 (0.75 m, 0.0011%/cycle, 5.3 99.9%
[(NPr)2V]4+ 2.0 m NaCl 3.3 × 10−6 −0.78 (second) 60 0.033%/day)[72] ≈65%
0.309, 3.9 × 10−6 (first) NMe-TEMPO 1.35 and 60 500 (0.5 m, 0.005%/cycle) [106] ≈3 100%
0.305, 3.8 × 10−6 (second) 1.72 128.2 60%
3.1f 2.0 m in 0.28 −0.43 KI 1.0 60 100 (0.5 m, 0.01%/cycle)[105a] ≈3.6 ≈100%
(SPr)2V water 3.26 × 10−6 92.5 67%
(NH4)4 0.8 40 1000 cycles,1100 h, 0% decay (0.9 7.5 ≈100%

Fe(CN)6 72.5 m symmetric electrolyte)[60] 62.6%
3.1j 1.1 m in 0.28 −0.44 NMe-TEMPO 1.44 40 300 (0.1 m, 0.03%/cycle)[104] 3.86 100%
[(NPr)2TTz]Cl4 2.0 m NaCl 3.15 × 10−6 NA 70%
3.1h 2.0 m in >0.356, 5.19 × 10−6 (first) −0.36 (first) KBr-MEP 1.49 and 40 50 (0.5 m, porous polyolefin)[103] 18.2 99.1%
(2HO-V) Br2 H2O >0.312, 3.99 × 10−6 (second) −0.76 (second) 1.89 204 80%
3.1l NA (9 ± 2) × 10−5 −0.29 PolyTEMPO 1.19 60 95 (0.24%/cycle)[107] 8 99.8%

PolyVi (7.6 ± 0.9) × 10−7 75%
a)Peak power
Figure 12. TEMPO and its derivatives as the catholyte candidate for ARFBs. a) Multiple derivatives of TEMPO used in aqueous media; b) summary
of the redox potential and volumetric capacity of TEMPO derivatives in the presence of supporting salts; c) scheme of the electrochemical reaction of
4-OH-TEMPO as well as the N-oxyl centered compounds; d) cyclic voltammogram of TMAP-TEMPO versus BTMAP-Vi; and e) long-term galvanostatic
cycling of a 0.1 m TMAP-TEMPO|BTMAP-Vi redox flow battery. d,e) Reproduced with permission.[114] Copyright 2019, Elsevier Inc.
3.2.2. Challenges and Mitigation Strategies attached a sulfonic acid group on the 4-position of the piperi-
dine ring (i.e., 3.2d, TEMPO-4-sulfate) and demonstrated a
Low Energy Density: The low solubility of TEMPO, espe- Zn|3.2d full cell, which exhibited a stable capacity (0.92 Ah L−1)
cially in the presence of supporting salts (0.5 m in 1.5 m over 1000 cycles at 40 mA cm−2 with a high discharge potential
NaCl), limited the energy density of TEMPO-based ARFBs.[101] of 1.5 V (≈300 mV higher than that of viologen-based systems).
Various functionalization strategies have been applied to Poor Stability: The chemical decomposition of TEMPO and
increase the solubility and cell voltage as summarized in its derivatives led to capacity decay in ARFBs.[113] Liu et al.[114]
Figure 12b. Janoschka et al.[110] introduced a hydrophilic and reported the use of 4-[3-(trimethylammonio)propoxy]-2,2,6,6
electron-withdrawing trimethylammonium chloride group tetramethylpiperidine-1-oxyl (3.2f, TMAP-TEMPO) as a catho-
(–N(CH3)3+Cl−) as a 4-position substituent (i.e., N,N,N-2,2,6,6- lyte, which demonstrated a higher water solubility (4.62 m) and
heptamethylpiperidinyl oxy-4-ammonium chloride, 3.2c, lower chemical decomposition rate than 4-OH-TEMPO, NMe-
TEMPTMA or NMe-TEMPO), which successfully increased TEMPO, and TEMPO-sulfate due to the steric hindrance and
the solubility (i.e., 2.3 m in 1.5 m NaCl, 61 Ah L−1) and redox coulombic repulsion of the four peripheral methyl groups. By
potential (i.e., 0.95 VSHE vs 0.8 VSHE for TEMPOL). They dem- coupling with a viologen anolyte, the full cell demonstrated
onstrated a 2.0 m TEMPTMA|MVi system that exhibited full a concentration-independent temporal capacity retention
capacity retention over 100 cycles at 80 mA cm−2. rate of 99.38% per day and 99.993% per cycle for 1000 cycles
To increase the overall cell voltage, Zn2+/Zn was considered (Figure 12d,e). Isolating charged functional groups protected
as an alternative anolyte to viologen-based anolytes. However, the redox couple from dimerization.
the AEMs used for TEMPO-based catholytes were not compat- The crossover of TEMPO led to fast capacity decay in
ible with Zn2+ ions.[111] Chang et al.[111] reported a glycidyltri- AORFBs. Connecting bipolar redox couples by covalent bond
methylammonium cation (GTMA+) grafted 4-OH-TEMPO (3.2e, was an effective strategy for the mitigation of capacity fade
g+-TEMPO) which could couple with a Zn2+/Zn anolyte using a caused by cross contamination. Hagemann et al.[115] reported
PBI membrane. However, capacity decay was accelerated after combined molecules (combi-molecules) of 3.2g TEMPO
140 cycles due to the crossover of g+-TEMPO. Winsberg et al.[112] (catholyte)-phenazine (anolyte), via a triethylene glycol
linker,[116] and TEMPO (catholyte)-viologen (anolyte), via a 4-benzoquinone (3.3a, BQ), naphthoquinone (3.3b, NQ), and
benzyl linker (i.e., 1-(4-(((1-oxyl-2,2,6,6-tetramethylpiperidin- anthraquinone (3.3c, AQ). Their derivatives are classified by
4-yl)oxy)carbonyl)benzyl)-1′-methyl-[4,4′-bipyridine]-1,1′-diium- their solubility in aqueous electrolytes as: acidic (pH < 7, mostly
chloride, 3.2h, VIOTEMP),[117] in a symmetric cell configura- with –SO3H functional groups), alkaline (pH > 10, mostly with
tion. The 10 × 10−3 m 3.2g symmetric electrolyte operated for –OH, –COOH and –PO3H2 functional groups), and neutral
over 1800 cycles without capacity decay with a CE up to 98.3%. (pH = 7) environments (Figure 13a).
Another effective strategy was the development of large The redox reaction of quinone is a single two-electron
polymers as the active species to reduce crossover. Janoschka reaction involving proton coordination in aqueous
et al.[118] developed multiple TEMPO-based polymers (i.e., 3.2i, environments.[123b,127] The Nernst equation of the equivalent
p(TEMPO-co-METAC),[107,119] 3.2j, p(TEMPO-co-zwitterion),[115] potential is as follows[123a,127a]
and 3.2k, p(TEMPO-co-PEGMA)[120]), which were selectively
blocked via pore-size exclusion using a low-cost dialysis mem-
RT Hm BQ 
m −n
brane (molecular-weight cut-off (MWCO) of 60 000 g mol−1). 3.2j BQ + mH+ + ne− ↔ Hm BQm −n ; E = E 0 − ln m
(10)
was synthesized by grafting a zwitterionic [(2-(methacryloxy) nF [BQ] H+ 
ethyl)dimethyl-(3-sulfopropyl)]ammonium hydroxide as the sol-
ubilizing comonomer,[115] which exhibited a high redox potential where [HmBQm−n], [BQ], and [H+] are the activities. Quan
(0.93 VSHE), a high solubility (20 Ah L−1 in 1.5 m NaCl vs et al.[127a] investigated the electrochemical reactions of BQ and
10 Ah L−1 for 3.2i and 2.39 Ah L−1 for 3.2k), and a low dynamic AQS in aqueous solutions with different pH values (Figure 13b).
viscosity (17 mPa s at 20 °C, 10 Ah L−1 vs 17 mPa s for 3.2i). In acidic solutions, a 2e−/2H+ redox reaction took place to
The ARFB that used 3.2j with MVi demonstrated 8.75 Ah L−1 give hydroquinone (BQH), which led to a theoretical slope of
and stably cycled for 125 cycles with a 0.29% capacity loss per –59 mV pH−1 in the Pourbaix diagram. In neutral environments,
cycle at 8 mA cm−2. To date, polymer-based ARFBs suffer from a 2e−/1H+ reaction occurred with a slope of –30 mV pH−1. In
a low power capability and limited energy density. Future work unbuffered, alkaline environments, the redox potential was
on increasing their hydrophilicity and solubility is needed to pH-independent because H+ was not involved.[123b,127b,128] For
develop high-performance, polymer-based ARFBs. example, the Pourbaix diagram of 3.3t, AQDS (Figure 13c) fol-
Potential N-Oxyl Compounds Beyond TEMPO: Other classes lowed the theoretical slope of the potential–pH curve.[123b]
of N-oxyl radicals, such as imidoxyl species (e.g., phthalimide The redox potential of quinone has been manipulated
N-oxyl, 3.2l, PINO), are also promising candidates for high by the type and number of functional groups on the aro-
potential catholytes for RFBs. PINO was first reported by matic ring.[99,129] DFT calculations on the frontier orbitals of
Tian et al.[121] in a semi-aqueous RFB based on an acidic water– HOMO–LUMO,[129,130] the Gibbs free energy change,[131] and
acetonitrile (ACN) binary electrolyte. The redox potential of the Hammett parameters of different functional groups[131]
PINO was as high as 1.3 VSHE (500 mV higher than TEMPO) offered a quantitative understanding on how functional
thanks to the higher OH bond strength of N-hydroxyimides groups affect the redox potentials (E0) of quinone. As sum-
(≈88 kcal mol−1 for N-hydroxyphthalimide (NHPI)) relative to marized in Figure 13f,[37,99,100,123b,127b,132] –SO3H functionaliza-
hydroxylamines (≈69–71 kcal mol−1 for TEMPO). In addition, it tion on BQ and AQ offered the highest redox potential com-
displayed excellent reaction kinetics (k0 = 1.27 × 10−2 cm s−1), pared to their siblings that were functionalized with –OH
diffusion properties (D = 7.31 × 10−6 cm2 s−1), and reversibility in and –COOH. To reduce the redox potential of AQDS as the
the binary electrolyte. A full cell coupling PINO with 9,10-anth- anolyte, Gerhardt et al.[99] suggested numerous methods
raquinone-2,7-disulfonic acid (AQDS) demonstrated 0.20 Ah L−1 to tune the functional groups including: 1) removing one
under 3 mA cm−2 for 20 cycles. Other N-oxyl species, such as electron-withdrawing group (e.g., anthraquinone-2-sulfonic
2-phenyl-4,4,5,5-tetramethylimidazoline-1oxyl-3-oxide (PTIO), acid (3.3u, AQS)); 2) adding electron-donating groups (e.g.,
were reported as ambipolar molecules for symmetric cell in 1,8-dihydroxy9,10-anthraquinone-2,7-disulphonic acid (3.3s,
nonaqueous redox flow batteries (Figure 12c(ii)[122]). However, DHAQDS)); and 3) distancing electron-withdrawing groups
they suffered from low stability and high crossover. Future from the redox center (e.g., 1,4-dihydroxyanthraquinone2,3-
investigations focused on the enhancement of the aqueous dimethylsulfonic acid, 3.3r, DHAQDMS). In addition, 2,6-dihy-
solubility and chemical stability of N-oxyl compounds would droxyanthraquinone (3.3 k, 2,6-DHAQ) was reported as a
improve the long-term stability of N-oxyl-based AORFBs. promising anolyte (E0 of –0.7 VSHE) and demonstrated a high
power capability (0.4 W cm−2).[132a] When combined with fer-
rocyanide in a full cell, it stably operated for 100 cycles at
3.3. Quinone-Based AORFBs 100 mA cm−2 with an EE of 84%. The cell voltage was further
improved by Lin et al.[132a] from 1.2 to 1.35 V by adding –OH and
3.3.1. Fundamental Physicochemical Properties –CH3 groups to the aromatic ring (i.e., 1,5-dimethyl-2,6-DHAQ
(3.3l, DMAQ) and 2,3,6,7-tetrahydroxy-AQ (3.3m, THAQ)).
Quinones are a class of carbonyl compounds that contain two
carbonyl groups (CO) in an aromatic ring.[123] Quinone and its
derivatives store charges via an “ion-coordination” mechanism 3.3.2. Challenges and Mitigation Strategies
in which CO groups are reduced to yield oxygen anions coor-
dinated by guest cations (e.g., H+ in aqueous, Li+,[124] Na+,[124a,125] Low Solubility: Introducing polar functional groups (e.g.,
Zn2+[126]). Quinones that have been applied in ARFBs include 1, –OH, –COOH, –SO3H, –PO3H2, etc.) has been reported as an
Figure 13. a) Benzoquinone (3.3a, BQ), naphthoquinone (3.3b, NQ), anthraquinone (3.3c, AQ) and their derivatives that have been utilized in aqueous
redox flow batteries; b) electrochemical reactions of BQ in aqueous solutions. c,d) Relationships between electrochemical potential of 3.3t, AQDS,
and solution pH (c) or the number of –OH groups (d); and e) CV analysis of 3.3s and 3.3t. c–e) Reproduced with the permission.[123b] Copyright 2014,
Springer Nature. f) Theoretical volumetric capacity and standard voltage of the reported quinone derivatives.[37,99,100,123b,127b,132,133]
effective strategy to enhance the solubility of organic molecules for 100 cycles at 100 mA cm−2 with 94.7% capacity retention.
via increasing the molecules’ polarity and decreasing their sol- Jin et al.[127b] developed a pH-independent water-miscible
vation energy (ΔGsolv).[131] For instance, 2-hydroxy-1,4-naphtho- anthraquinone with poly(ethylene glycol)-based groups as the
quinone (3.3b, 2-HNQ) exhibited limited solubility in aqueous anolyte (i.e., 1,8-bis(2-(2-(2-hydroxyethoxy)ethoxy) ethoxy)anthra-
media (i.e., 0.4 m in alkaline). Wang et al.[132c] reported a deriva- cene-9,10-dione (3.3p, PEGQ)). This cell demonstrated a high
tive of 2-HNQ, named 2-hydroxy-3-carboxy-1,4-naphthoquinone solubility of 2.2 m in aqueous media. A full AFRB using 7 mL
(3.3i, 2,3-HCNQ), by introducing hydroxyl and carboxyl groups 1.5 m 3.3p|150 mL 0.31 m K4Fe(CN)6 was demonstrated to
to 2-HNQ. The solubility increased to 1.2 m in 1.0 m KOH at operate for 220 cycles (18 days) with a CE of 99.90% and a
20 °C. Employing N212, a flow cell that consisted of 0.5 m temporal fade rate of 0.5% per day and 0.043% per cycle using
2,3-HCNQ and 0.4 m K4Fe(CN)6 demonstrated stable cycling a sequential potential step method (Figure 14a). Combining
Figure 14. Degradation mechanisms of quinone derivatives. a) Long-term galvanostatic cycling of 7 mL 1.5 m 3.3p, PEGAQ|150 mL 0.31 m K3Fe(CN)6/
K4Fe(CN)6; and b) charge–discharge curve under potentiostatic conditions. Electrolytes comprised 6 mL of 1.5 m 3.3p in 0.5 m KCl and 12 mL of 1.51 m
1:1 mixed potassium/sodium ferrocyanide and 0.01 m potassium ferricyanide in DI water. a,b) Reproduced with permission.[127b] Copyright 2019, American
Chemical Society. c) Long-term cycling of 0.5 m 3.3o||0.4 m K3Fe(CN)6/0.1 m K4Fe(CN)6 at pH 9. Reproduced with permission.[132d] Copyright 2019,
Wiley-VCH. d) Nucleophilic addition mechanism of benzoquinone derivatives; and e) multiple degradation mechanisms of anthraquinone derivatives.
3.3p with a concentrated catholyte (i.e., 1.51 m 1:1 mixed potas- Anchoring electron-withdrawing groups, such as sulfonic acid
sium/sodium ferrocyanide and 0.01 m potassium ferricyanide), groups (–SO3H), on benzoquinone effectively increased the redox
Jin et al. achieved a record 25.2 Wh L−1 energy density compared potential of BQ to >0.8 VSHE, making these derivatives promi
to previously reported organic flow batteries that used potentio- sing catholytes for ARFBs (Figure 13f). However, the resulting
static methods (Figure 14b). Lastly, the selection of electrolyte electron deficiency in the ring promoted nucleophilic addition
counter cations has been shown to tune the solubility of qui- of H2O or OH− (in alkaline environments).[100] For example,
none and its derivatives. Carretero-González et al.[134] reported Michael addition (1,4-addition) was found to take place on 1,2-
using organic cations (e.g., tetrakis (hydroxyethyl)-ammonium BQDS by H2O nucleophilic addition in both acidic and neutral
[N(CH2CH2OH)4]+ or tetrakis (hydroxymethyl)-phosphonium environments (Figure 14d[135]). To mitigate the reactivity of 1,2-
[P(CH2OH)4]+) to increase the solubility of organic molecules BQDS, Hoober-Burkhardt et al.[135a] synthesized 3,6-dihydroxy-
compared to their sodium forms (Na-type). Hu et al.[140] found 2,4-dimethylbenzenesulfonic acid, (3.3g, DHDMBS), to mini-
that the solubility of AQDS increased from 0.58 m in Na-type mize the amount of open positions on the phenyl ring by
(AQDSNa2) to 1.9 m in NH4+ type (AQDS(NH4)2). The demon- methyl group substitution. This strategy reduced the chance of
strated AQDS(NH4)2–NH4I system at 0.75 m was reported to H2O attack. In alkaline media, hydroxide ions acted as stronger
achieve 12.5 Wh L−1 for 300 cycles (15 days) at 60 mA cm−2. nucleophiles to deprotonate methyl groups (Figure 14d), which
Chemical Degradation of Quinone: The four major chemical led to the decomposition of BQ derivatives in alkaline battery
degradation pathways of quinone and its derivatives include: systems. For instance, 2,5-dihydroxy-1,4-benzoquinone (3.3d,
1) nucleophilic addition (or Michael addition, Figure 14d); DHBQ) was reported to be highly soluble in aqueous media
2) dimerization (Figure 14e(i,ii); 3) electrophilic substitution (>4.0 m in alkaline) with a sufficiently low redox potential
(Figure 14e(iii)); and 4) γ-hydroxybutyrate cleavage (Figure 14e(iv)). (Figure 13f).[133] However, due to the Michael addition of OH−
on 3- and 6-positions, the flow cell only operated for 150 cycles undergo a disproportionation reaction to 2,6-dihydroxyan-
with a decay rate of 0.24% per cycle. To enhance the stability of throne (DHA3−), which irreversibly formed a dimer (DHA)2
the DHBQ-based cell, the authors polymerized DHBQ to block (Figure 14e(ii)).[138] To avoid decomposition of DHAQ, Goulet
vulnerable positions and achieved 99.96% per cycle retention for et al.[138] suggested the restriction of SoC and the aeration of the
400 cycles.[133] Michael addition was also prone to occur at the anolyte. The battery stability was improved by 2 orders of mag-
C–H position next to the functional group 3.3i in alkaline media. nitude (i.e., from 5.6% per day to 0.042% per day) by combining
Tong et al.[136] constructed a bislawsone structure with two naph- these two strategies. Recently, Zhao et al.[139] employed in situ
thoquinones linked by their 3-position. 3.3j, bislawsone demon- 1H NMR to detect the degradation products anthrone (DHA3−)
strated a higher stability than 3.3i and operated for 600 cycles at and anthrol (DHAL3−) and revealed that the extent of decompo-
100–300 mA cm−2 with 98.3% utilization of the anolyte capacity. sition was dependent on the reduction potential, highlighting
Carney et al.[137] reported that 2-electron transfer of AQDS that decomposition was an electrochemical process rather than
could only be achieved at low concentrations (<1 × 10−3 m). At a chemical reaction with water. Representative quinone-based
higher concentrations, only transfer of 1 to 1.5 electrons could AORFBs are summarized in Table 3.
be achieved depending on the pH environment (Figure 14e(i)).
When coupling with Br−/Br2,[123b] Br could replace the H atom
and attach to an aromatic carbon on the outer rings of AQ, 3.4. Nitrogen-Centered Heteroaromatic-Molecule-Based
Figure 14e(iii).[99] To mitigate this electrophilic substitution reac- AORFBs
tion, an electron-withdrawing group (e.g., –SO3H) could be intro-
duced to raise the activation barrier or to protect the aromatic 3.4.1. Fundamental Physicochemical Properties
ring from Br substitution. Kwabi et al.[132b] introduced bulky
functionalization groups and developed 4,40-((9,10-anthraqui- Quinoxaline is a simple aza-aromatic molecule which has two
none-2,6-diyl)dioxy)dibutyrate (3.3n, 2,6-DBEAQ), which dem- heteroatoms (Figure 15a(i)). Milshtein et al.[141] reported that
onstrated 1.1 m solubility in pH = 14, and successfully decreased quinoxaline possessed a low E0 of –0.8 VSHE in alkaline envi-
the calendar decay rate to 0.05% per day. However, this mole ronments and demonstrated a solubility of up to 4.0 m (3.4a).
cule was subject to decomposition through γ-hydroxybutyrate Leung et al.[142] reported a rechargeable quinoxaline-air redox
cleavage by both hydroxide and carboxylate as nucleophiles flow battery which operated over 20 cycles with an average CE
(Figure 14e(iv)).[132b] Aziz et al.[132d] designed (((9,10-dioxo- of 81% and an EE of 25%. Phenazine has three fused aromatic
9,10-dihydroanthracene-2,6-diyl) bis(oxy))bis(propane-3,1-diyl)) rings (Figure 15a(ii)) and is insoluble in water. Functionaliza-
bis(phosphonic acid) (3.3o, 2,6-DPPEAQ) by replacing the car- tion could simultaneously enhance its solubility and tune its
boxylic groups with bulky phosphonate acid groups, which were redox potential. Similar to quinone, the redox reactions of
much weaker nucleophiles and stable under milder, less basic phenazine are pH-dependent processes involving two elec-
conditions (pH = 9). A 0.5 m 3.3o|ferri/ferrocyanide system trons and two protons. After screening multiple derivatives of
was demonstrated with a record stability of 0.00036% per cycle phenazine via DFT calculations, Hollas et al.[143] reported the
and 0.014% per day for AORFBs at 100 mA cm−2 for 480 cycles use of 7,8-dihydroxyphenazine-2-sulfonic acid (3.4b, DHPS)
(Figure 14c). The reduction product DHAHQ was found to as an anolyte for ARFBs, which exhibited a high solubility
Table 3. Summary of flow cell performance of quinone-based AORFBs.
Quinone and Qtheoretical [Ah L−1] k0 [cm s−1] E0 [VSHE] Catholyte Ecell [V] J [mA cm−2] Cycle life (decay rate) Ed,achieved CE
derivatives (Sol. [m]) D [cm2 s−1] Pa) [mW cm−2] [Wh L−1] EE
3.3t, AQDS 80.4 7.2(5) × 10−3 0.21 HBr/Br2 ≈0.81 200 10 (1 m, 1%/cycle, 19.49 99.2%
(1.5 m (pH0)) 3.8(1) × 10−6 600 ≈2.4%/day) [123b] NA
3.3k, DHAQ 32 NA −0.7 Fe(CN)64− 1.2 100 100 (0.5 m,0.1%/cycle)[132a] 8.04 99%
(0.6 m (pH = 14)) 4.8 × 10−6 400 84%
3.3i, 2,3-HCNQ 64.3 2.07 × 10−3 −0.73 K4Fe(CN)6 1.02 100 100 (0.5 m, 0.05%/cycle, ≈4 ≈1 00%
(1.2 m (pH = 14)) 3.44 × 10−6 225 3.4%/day)[132c] 68.8%
3.3d, DHBQ >429 2.12 × 10−3 −0.72 K4Fe(CN)6 1.21 100 150 (0.5 m, 5.87 99%
(>8 m (pH = 14)) 3.66 × 10−6 300 0.24%/cycle)[133] 65%
3.3q, AQDS(NH4)2 102 0.077 −0.2 I−/I3− 0.865 60 300 (0.75 m, 15 days,[140] 12.5 100%
(1.9 m (pH = 7)) 4.55 × 10−6 91.5 no decay) 70%
3.3n, 2,6-DBEAQ 58.9 NA −0.52 Fe(CN)64− 1.05 50 250 (0.5 m, 0.05%/day 2.61 >99%,
(1.1 m (pH = 14)) 1.58 × 10−6 240 0.001%/cycle)[132b] 88%
3.3o, 2,6-DPPEAQ 40.2 NA −0.47 (pH9) K4Fe(CN)6K3Fe(CN)6 1.0 100 480 (0.5 m, 0.014%/day 1.58 99.9%
(0.75 m (pH = 9)) 1.37 × 10−6 160 0.00036%/cycle)[132d] 65%
3.3p, PEGAQ 118 6.14 × 10−3 −0.43 (pH7) K4Fe(CN)6K3Fe(CN)6 1.0 50 220 (0.5%/day 25.2 99.90%
(2.2 m) 2.94 × 10−6 180 0.043%/cycle)[127b] NA
a)Peak power
Figure 15. Flow cell performance of RFBs based on aza-aromatic organic compounds. a) Nitrogen-centered heteroaromatic molecules for aqueous
redox flow batteries. Multiple aza-aromatic organic molecules and their redox reactions including class i: quinoxaline, class ii: phenazine, class
iii: phenothiazine, and class iv: pteridine; b) capacity and efficiency for a cell composed of 3.4b, DHPS||K4Fe(CN)6/K3Fe(CN)6 at 20 to 100 mA cm−2;
and c) long-term cycling of 0.1 m DHPS at 50 mA cm−2. b,c) Reproduced with permission.[143] Copyright 2018, Springer Nature. d) Long-term cycling
performance of 3.4c, BHPC||K4Fe(CN)6 at 100 mA cm−2 for 1305 cycles. Reproduced with permission.[144] Copyright 2020, American Chemical Society.
(1.8 m) in 1.0 m NaOH with a redox potential of –0.86 VSHE. the additive used in the electrolyte (i.e., H2O + AA (volume
They demonstrated a 1.4 m (2.8 m electron) full cell coupled ratio of 3:1) containing 3.5 m H2SO4), which effectively reduced
with Fe(CN)63−/4−, which operated over 500 cycles at 67 Ah L−1 the lattice energy of MB precipitates via the strong interac-
(10.2 Wh L−1) with 0.0195% capacity decay per cycle and 0.169% tion between AA and 3.4d. Additionally, common hydrotropic
per day (Figure 15c). Recently, Wang et al.[144] introduced an agents including nicotinamide, urea, and caffeine enhanced the
electron-donating phenyl group adjacent to the hydroxyl group water solubility of organic molecules. For instance, FMN–Na
in phenazine to further enhance the stability. As a result, (3.4e) exhibited ultralow solubility in acidic environments (i.e.,
benzo[a]hydroxyphenazine-7/8-carboxylic acid (3.4c, BHPC) 10 × 10−3 m in 1.0 m H2SO4, pH 0.8), neutral environments (i.e.,
demonstrated a low decay rate (0.014% per cycle and 0.08% 50 × 10−3 m in 1.0 m KCl), and alkaline environments (i.e., 100 ×
per day) at a 1.0 m electron concentration for 1300 cycles when 10−3 m in 1.0 m KOH). Orita et al.[147] introduced 3 m nicotina-
coupled with a Fe(CN)63−/4−catholyte, achieving an energy den- mide as a hydrotropic agent to stabilize 3.4e in alkaline environ-
sity of 7.3 Wh L−1 (Figure 15d). ments (1.0 m KOH) and increased its solubility to ≈1.5 m.
Substituting one of the nitrogen atoms in phenazine with Chemical Stability: Flavin was reported to undergo photo-
an electron-withdrawing sulfur atom (positively charged), i.e., induced electron transfer, alkaline catalyzed hydrolysis, and
phenothiazine (Figure 15a(iii)), significantly increased its redox dimerization in aqueous media, leading to fast capacity decay in
potential by increasing the electron affinity of the molecules. redox flow batteries.[147] Alloxazine is subject to a ring opening
For instance, methylene blue (3.4d, MB) involves two-electron reaction in aqueous media via the addition of H2O to amidic
and one-proton transfer processes which led to a pH-dependent carbonyls, followed by hydrolysis to redox-inactive species.
redox potential in acidic media. The solubility of MB was Lin et al.[148] conducted theoretical calculations of various alloxa-
1.8 m in acetic acid (AA) and H2SO4. MB exhibited fast reac- zine derivatives to predict their redox potential and the equi-
tion kinetics (k0 = 0.32 cm s−1), which was one of the highest librium constant of the reversible hydration of carbonyl groups
among organic molecules. Zhang et al.[145] reported a flow cell (Khyd). They revealed that electron-donating functional groups,
system that coupled MB with a vanadium (II) anolyte as a full such as hydroxyl groups, protected the redox center against
cell. A 2.4 m electron concentrated system of V(II)|MB was hydrolysis and simultaneously decreased the redox potential
demonstrated, which operated for 160 cycles with a decay rate of the alloxazine anolyte. The performances reported in related
of 0.025% per cycle and 0.52% per day. works are summarized in Table 4.
Another important class of aza-aromatic anolyte candidates
was heterocyclic molecules that contained a 1,3,5,8-tetraazan-
aphthalene (pteridine), which is composed of fused pyrimidine 4. Characterization Techniques for ARFBs
and pyrazine rings[146] (Figure 15a(iv)). Flavins (isoalloxazinic
form) and lumichrome (alloxazinic form) are common tauto- 4.1. Electrochemical Characterization: Galvanostatic versus
meric pairs of pteridine derivatives that involve two-electron and Potentiostatic Cycling
two-proton processes in aqueous media (Figure 15a(iv)). Flavin
is a highly water-soluble derivative and operates two-electron Stability is one of the most important performance metrics of
reactions through the formation of flavin semiquinone radical ARFBs. It is imperative to develop effective assessment meth-
intermediates (Figure 15a(iv)). Orita et al.[147] reported a flow odologies and metrics to evaluate the stability of different
cell that applied the sodium salt of flavin mononucleotide (3.4e, ARFBs.[10] A typical electrochemical characterization method
FMN–Na (–0.53 VSHE)) and Fe(CN)63−/4− (0.5 VSHE) at 1.4 m, for evaluating the stability of ARFBs is galvanostatic cycling
which operated for 200 cycles at 50–80 mA cm−2. Lin et al.[148] under a designated current density and SoC. This technique
reported using the alloxazinic form of lumichrome that has a evaluates the efficiencies (CE, EE) and stability of full cell RFBs.
diazabutadiene double bond as an anolyte, which demonstrated The stability metrics usually include cycle life (i.e., capacity
a low redox potential of –0.7 VSHE. By grafting a carbolic acid retention per cycle over the total number of cycles) and cal-
group onto insoluble alloxazine, the solubility of the alloxazine endar life (i.e., capacity retention per day over the total cycle
7/8-carboxylic acid (3.4f, ACA) increased to 2.0 m in 1.0 m KOH, time). Since the most critical degradation mechanism can be
which was equivalent to 108 Ah L−1. When combined with very different in different ARFBs, using single metrics to com-
Fe(CN)63−/4− as a full flow cell at a 1.0 m electron concentration, pare different ARFBs could lead to biased conclusions. A more
operation for 400 cycles with a capacity decay rate of 0.02% per balanced comparison can be made by considering both cycle
cycle at 100 mA cm−2 was demonstrated. life and calendar life.
Degradation mechanisms of ARFBs can be categorized
into: 1) the intrinsic irreversibility/instability of active species;
3.4.2. Challenges and Mitigation Strategies 2) crossover of active species and water migration; 3) increase
of membrane resistance; 4) insufficient capacity utilization; and
Low Solubility: Quinoxaline (3.4a) exhibited a low solu- 5) formation of dendrites on metal electrodes in hybrid ARFBs.
bility (0.5 m) in alkaline media, which limited its volumetric For AIRFBs that employ small molecules (e.g., polysulfide,
capacity for ARFB applications.[142] Milshtein et al.[141] reported halogen, and vanadium), time-dependent crossover largely
that Cl− effectively enhanced the solubility of quinoxaline in limits the stability of the batteries, which makes calendar life an
alkaline environments to 4.0 m. Similarly, both 3.4d, MB and important evaluation metric when comparing stability among
3.4e, FMN–Na demonstrated low water solubilities within their these AIRFBs. For hybrid AIRFBs (e.g., Zn-based, Fe-based),
chemically stable pH range. Zhang et al.[145] introduced AA as dendrite formation during repetitive stripping/deposition
Table 4. Electrochemical performance of AORFBs based on N-containing heteroaromatic compounds.
Active species k0 [cm s−1] Qtheoretical [Ah L−1] Counter Ecell [V] J [mA cm−2] Cycle life (decay rate) Ed,achieved CE
(E0 [VSHE]) D [cm2 s−1] (Solubility [m]) electrolyte Pa) [mW cm−2] [Wh L−1] EE
3.4b, DHPS NA 96.48 (1.8) [Fe(CN)6]3−/4− 1.4 100 500 (1.4 m, 0.0195%/cycle, 9.46 ≈100%
(−0.86) (catholyte) NA 0.69%/day)[143] 82%
3.4c, BHPC 1.55 × 10−3 83.1 (1.55) K4Fe(CN)6 1.27 100 1300 (0.5 m, 0.014%/cycle, 7.3 99.2%
(−0.78) 3.52 × 10−6 (catholyte) 430 0.08%/day)[144] 80%
3.4e, FMN–Na 5.3 × 10−3 81 (1.5 m in 1.0 m KOH and K4Fe(CN)6 1.03 50–80 100 (0.24 m, 0.01%/cycle, 3.2 99%
(−0.53) 1.3 × 10−6 3.0 m nicotinamide (NA)) (catholyte) 160 0.16%/day) [147] ≈80% (@80 mA cm−2)
3.4f, ACA 1.2 ± 0.2 × 10−5 108 (2 m in 1 m KOH) [Fe(CN)6]3−/4− 1.2 100 400 (0.5 m, 0.02%/cycle, 7.25 99.7%
(−0.62) NA vs. (catholyte) 350 1.8%/day)[148] 63%
3.4d, MB 0.32 96.5 (1.8 m in V(II) 0.83 80 160 (1.2 m, 0.025%/cycle, ≈9.8 ≈100%
(0.57) 2.05 × 10−6 H2O+acetic acid +H2SO4) (anolyte) NA 0.52%/day)[145] 76%
a)
Peak power
largely limits their battery stability, which makes cycle life an 4.2. Material Characterization
important metric when comparing stability among hybrid
AIRFBs. In addition, since energy and power are coupled in 4.2.1. Spectroscopic Characterizations of Electrolytes
hybrid AIRFBs, stability under a specified areal capacity is also
a vital parameter when comparing hybrid systems that use Spectroscopic characterizations of electrolytes probe the chemical
solid electrodes. stability and electrochemical reversibility of redox reactions. UV–
For AORFBs, Goulet et al.[70] revealed that capacity fading vis spectroscopy is suitable for active materials that absorb ultra-
was independent of charge–discharge cycle numbers for most violet and visible radiation (200–750 nm) with high precision for
organic and organometallic redox active species. Instead, it was concentration measurements.[61,90a,148] For example, the solubility
a time-denominated exponential decay phenomenon in which of H6[CoW12O40] could be accurately determined by UV–vis spec-
the time constant was SoC-dependent. To properly evaluate the troscopy compared to standard solutions.[90a] Zhou et al.[96] applied
intrinsic stability of electrolyte materials without interference operando UV–vis spectroscopy to understand the SMRT reaction
from other factors, new testing protocols are recommended. mechanism. By measuring the evolution of absorbance intensity
Goulet et al.[70] suggested an unbalanced, compositionally sym- at 630 nm for RM FcIL+, they verified the working mechanism
metric flow cell to probe the chemical and electrochemical of the SMRT-based LiFePO4 cell. Additionally, Li et al.[79] applied
stability based on the potentiostatic charge/discharge process. operando UV–vis along with CV tests of I−/I3− and S22−/S2− redox
This method enabled the evaluation of the intrinsic stability of pairs. They revealed reversible evolution of the UV–vis spectra
redox active materials by minimizing reactant crossover, elimi- during the oxidation and redox processes for both redox pairs.
nating the influence of membrane resistance, and allowing NMR has been widely applied to investigate the chemical
precise control of SoC. One example was demonstrated for stability of electrolyte materials in ARFBs. Gerhardt et al.[99]
3.3n, 2,6-DBEAQ AORFB.[132b] The temporal fade rate of 3.3n, applied NMR spectra to investigate the influence of bromine
2,6-DBEAQ was only <0.01% per day (over 26 days) which was exposure for various quinone derivatives (i.e., AQS, DHAQRS,
one of the most stable AORFBs tested by the symmetric cell and ARS). After 24 h of exposure to bromine, the aromatic
method. proton disappeared for both DHAQDS and ARS from the
Considering the advantages provided by the potentiostatic proton NMR results, indicating electrophilic attacks by Br in
charge/discharge method, adding a potential holding step these two derivatives. Zhao et al.[139] employed in situ NMR
after the galvanostatic charge/discharge measurement was setups to unravel the electrolyte degradation mechanism in real
suggested for RFBs.[127b,132b,d] This method avoided capacity time. Friedl et al.[94] applied in situ 51V NMR to detect the sta-
loss due to insufficient capacity utilization and the fluctuation bility of PV14 upon reduction and revealed that PV14 redox reac-
of membrane resistance during high current tests.[127b] For tions involved the transfer of 5 electrons to 14 vanadium ions.
instance, to investigate the stability of 2,6-DBEAQ in AORFB,
Kwabi et al.[132b] set a potential holding step at 1.4 V after galva-
nostatic charge and 0.6 V after discharge, respectively, until the 4.2.2. Membrane Assessment
magnitude of the current density fell below 2 mA cm−2. Under
these conditions, the flow cell demonstrated a capacity decay The choice of membrane is a critical component for deter-
rate of 0.05% per day (in total 5 days) and 0.001% per cycle mining the power capability and stability of RFBs. Water uptake
(250 cycles). To evaluate the stability of 3.3p, PEGAQ under (WU) reflects the hydrophilicity of the ion exchange membrane
full capacity utilization, which was not achievable in the gal- (IEM), which could be measured via the weight change of dry
vanostatic test, Jin et al.[127b] applied a sequential potential step (Wd) and wet states (Ww) [50b,149]
method for the flow cell tests, and achieved a high utilization
of 95.7%. The capacity decay rates were reported to be 0.5% per Ww − Wd
WU = × 100% (17)
day (18 days) and 0.043% per cycle (220 cycles). Wd
Ionic conductivity (σ) could be obtained through electro- 5.2. AORFBs

chemical impedance spectroscopy (EIS). The in-plane conduc-
tivity can be calculated by[50b] The major challenge pertaining to AORFBs (e.g., viologen-based,
N-heteroaromatic-molecule-based, and quinone-based AORFBs)
L (18)
σ= is their low chemical stability which leads to a short lifespan.
RS In addition, their energy densities are still far from satisfac-
where L is the distance between electrodes, R is the membrane tory (1.58–25.45 Wh L−1 with an average value of ≈10 Wh L−1,
resistance, and S is the cross-sectional area. Tables 2–4) due to either low solubility or reduced instability at
Ion permeability quantifies the ionic selectivity of the high concentrations. Several organic molecules (e.g., 3.3t, AQDS,
membrane, which determines the crossover of active mate-
3.4c, BHPC) have demonstrated high power capabilities (peak
rials. Permeability (P) could be measured in H-type cells power in the range of 100–600 mW cm−2) thanks to their fast
(i.e., concentrated and blank electrolytes with supporting salts kinetics, which are several orders of magnitude faster than those
balancing the ionic strength, separated by the membrane) via of inorganic materials (i.e., 10−1 to 10−2 cm s−1vs 2.2 × 10−5 cm s−1
the following equation[50b] for V2+/V3+). Finally, applying computational approaches such
as DFT and machine learning[150] to rationally screen and design
dc t P
V = A ( c 0 − c t ) (19) organic molecules with high solubility, fast kinetics, and high
dt L stability is an important direction of future research to further
where V is the volume of both sides in the H-cell, A and L are expedite the development of low-cost and long-lasting ARFBs for
the area and thickness of membrane, respectively, and c0 and ct practical applications.
are the concentration of ions in the concentrated electrolyte and
blank sides after testing, respectively. The self-discharge perfor-
mance of a full cell can be measured by measuring the open- Acknowledgements
circuit voltage of the battery after charging to a certain SoC.
The authors greatly acknowledge the financial support from the Research
Grant Council (RGC) of the Hong Kong Special Administrative Region,
China (Project No. T23-601/17-R) and Innovation and Technology
5. Conclusion and Perspectives Commission of the Hong Kong Special Administrative Region, China
(Project No. ITS/063/18).
5.1. AIRFBs
VRFBs are the most mature ARFB system with a high stability Conflict of Interest
and power capability. However, their development is limited
The authors declare no conflict of interest.
by the cost of vanadium and the limited choice of membranes.
Further developments focusing on enhancing the energy den-
sity and power density are critical to minimize the cost of
VRFBs. Zn-based hybrid RFBs (e.g., Zn–I2, Zn–Fe, etc.) are Keywords
promising candidates due to the low potential of zinc. However, aqueous redox flow batteries, energy storage, redox active materials,
both the poor scalability of Zn anodes and dendrite formation redox reactions
have critically challenged the durability of Zn-based systems.
Extensive efforts have been made toward developing functional Received: March 27, 2020
porous membranes[50,54] for long cycle life Zn-based ARFBs Revised: June 22, 2020
(e.g., decay rate of 0.0069% per cycle for Zn–Fe ARFB).[50a] Fur- Published online: October 22, 2020
ther improvements on the temporal stability should be made to
improve the competitiveness of Zn-based AIRFBs.
For polysulfide-based AIRFBs (e.g., PS/halogen and PS/
[1] D. Larcher, J.-M. Tarascon, Nat. Chem. 2015, 7, 19.
Fe), the low cost and high solubility of polysulfide promises [2] N. Kittner, F. Lill, D. M. Kammen, Nat. Energy 2017, 2, 17125.
competitive material costs. However, they suffer from a short [3] a) W. A. Braff, J. M. Mueller, J. E. Trancik, Nat. Clim. Change 2016,
lifespan and limited power capability. The fast capacity decay 6, 964; b) B. Dunn, H. Kamath, J.-M. Tarascon, Science 2011, 334,
highlights a need for the development of membranes with 928.
enhanced ionic selectivity and conductivity to reduce crossover [4] O. Schmidt, S. Melchior, A. Hawkes, I. Staffell, Joule 2019, 3, 81.
and support high currents. In addition, the sluggish kinetics of [5] a) G. L. Soloveichik, Chem. Rev. 2015, 115, 11533; b) J. W. Choi,
the sulfur redox reaction hinder power improvement of poly- D. Aurbach, Nat. Rev. Mater. 2016, 1, 16013.
sulfide-based AIRFBs. Further developments focusing on effec- [6] L. H. Thaller, US 3996064 1976.
tive and low-cost catalyst and membrane designs with excellent [7] a) E. Sum, M. Skyllas-Kazacos, J. Power Sources 1985, 15, 179;
b) M. Skyllas-Kazacos, M. Rychick, R. Robins, US 4786567, 1988.
ionic selectivity and low resistance are urgently needed. Finally,
[8] a) X. Wei, W. Pan, W. Duan, A. Hollas, Z. Yang, B. Li, Z. Nie,
the energy density of polysulfide-based AIRFBs is limited by J. Liu, D. Reed, W. Wang, V. Sprenkle, ACS Energy Lett. 2017, 2,
the catholyte. To take full advantage of its high capacity, it is 2187; b) R. M. Darling, K. G. Gallagher, J. A. Kowalski, S. Ha,
critical to develop novel catholyte materials with high redox F. R. Brushett, Energy Environ. Sci. 2014, 7, 3459.
potentials (>0.5 VSHE) and comparable volumetric capacities [9] Z. Huamin, IFBF 2017-Int. Flow Battery Forum, Manchester, UK 2017.
(≈200 Ah L−1). [10] D. G. Kwabi, Y. Ji, M. J. Aziz, Chem. Rev. 2020.
[11] N. Pour, D. G. Kwabi, T. Carney, R. M. Darling, M. L. Perry, Y. Shao- [42] A. L. Zhu, D. P. Wilkinson, X. Zhang, Y. Xing, A. G. Rozhin,
Horn, J. Phys. Chem. C 2015, 119, 5311. S. A. Kulinich, J. Energy Storage 2016, 8, 35.
[12] M. Vijayakumar, Z. Nie, E. Walter, J. Hu, J. Liu, V. Sprenkle, [43] Q. Yang, G. Liang, Y. Guo, Z. Liu, B. Yan, D. Wang, Z. Huang,
W. Wang, ChemPlusChem 2014, 80, 428. X. Li, J. Fan, C. Zhi, Adv. Mater. 2019, 31, 1903778.
[13] L. Li, S. Kim, W. Wang, M. Vijayakumar, Z. Nie, B. Chen, J. Zhang, [44] L. N. Bengoa, P. Pary, P. R. Seré, M. S. Conconi, W. A. Egli,
G. Xia, J. Hu, G. Graff, Adv. Energy Mater. 2011, 1, 394. J. Mater. Eng. Perform. 2018, 27, 1103.
[14] N. V. Roznyatovskaya, V. A. Roznyatovsky, C.-C. Höhne, M. Fühl, [45] B. Li, Z. Nie, M. Vijayakumar, G. Li, J. Liu, V. Sprenkle, W. Wang,
T. Gerber, M. Küttinger, J. Noack, P. Fischer, K. Pinkwart, J. Tübke, Nat. Commun. 2015, 6, 6303.
J. Power Sources 2017, 363, 234. [46] G.-M. Weng, Z. Li, G. Cong, Y. Zhou, Y.-C. Lu, Energy Environ. Sci.
[15] Z. He, J. Liu, H. Han, Y. Chen, Z. Zhou, S. Zheng, W. Lu, S. Liu, 2017, 10, 735.
Z. He, Electrochim. Acta 2013, 106, 556. [47] C. Xie, Y. Liu, W. Lu, H. Zhang, X. Li, Energy Environ. Sci. 2019, 12, 1834.
[16] H. Agarwal, J. Florian, B. R. Goldsmith, N. Singh, ACS Energy Lett. [48] J. Zhang, G. Jiang, P. Xu, A. Ghorbani Kashkooli, M. Mousavi,
2019, 4, 2368. A. Yu, Z. Chen, Energy Environ. Sci. 2018, 11, 2010.
[17] M. Gattrell, J. Qian, C. Stewart, P. Graham, B. MacDougall, Electro- [49] J. McBreen, J. Electroanal. Chem. Interfacial Electrochem. 1984, 168,
chim. Acta 2005, 51, 395. 415.
[18] C. Choi, S. Kim, R. Kim, Y. Choi, S. Kim, Renewable Sustainable [50] a) Z. Yuan, Y. Duan, T. Liu, H. Zhang, X. Li, iScience 2018, 3, 40;
Energy Rev. 2017, 69, 263. b) Z. Yuan, X. Liu, W. Xu, Y. Duan, H. Zhang, X. Li, Nat. Commun.
[19] N. Roznyatovskaya, J. Noack, H. Mild, M. Fühl, P. Fischer, 2018, 9, 3731.
K. Pinkwart, J. Tübke, M. Skyllas-Kazacos, Batteries 2019, 5, 13. [51] K. Gong, X. Ma, K. M. Conforti, K. J. Kuttler, J. B. Grunewald,
[20] a) B. Hu, J. Luo, C. DeBruler, M. Hu, W. Wu, T. L. Liu, in Encyclopedia K. L. Yeager, M. Z. Bazant, S. Gu, Y. Yan, Energy Environ. Sci. 2015,
of Inorganic and Bioinorganic Chemistry (Ed: R. A. Scott), Wiley 2019, 8, 2941.
https://doi.org/10.1002/9781119951438.eibc2679; b) D. Ma, B. Hu, [52] Z. Xie, Q. Su, A. Shi, B. Yang, B. Liu, J. Chen, X. Zhou, D. Cai,
W. Wu, X. Liu, J. Zai, C. Shu, T. Tadesse Tsega, L. Chen, X. Qian, L. Yang, J. Energy Storage 2016, 25, 495.
T. L. Liu, Nat. Commun. 2019, 10, 3367. [53] S. Selverston, R. F. Savinell, J. S. Wainright, J. Electrochem. Soc.
[21] L. Cao, M. Skyllas-Kazacos, C. Menictas, J. Noack, J. Energy Storage 2017, 164, A1069.
2018, 27, 1269. [54] C. Xie, Y. Duan, W. Xu, H. Zhang, X. Li, Angew. Chem., Int. Ed.
[22] S. Roe, C. Menictas, M. Skyllas-Kazacos, J. Electrochem. Soc. 2016, 2017, 56, 14953.
163, A5023. [55] C. Xie, H. Zhang, W. Xu, W. Wang, X. Li, Angew. Chem., Int. Ed.
[23] T. D. Nguyen, L. P. Wang, A. Whitehead, N. Wai, G. G. Scherer, 2018, 57, 11171.
Z. J. Xu, J. Power Sources 2018, 402, 75. [56] Y. Yin, S. Wang, Q. Zhang, Y. Song, N. Chang, Y. Pan, H. Zhang,
[24] B. Sun, M. Skyllas-Kazacos, Electrochim. Acta 1992, 37, 2459. X. Li, Adv. Mater. 2019, 32, e1906803.
[25] B. Sun, M. Skyllas-Kazacos, Electrochim. Acta 1992, 37, 1253. [57] R. Kim, H. G. Kim, G. Doo, C. Choi, S. Kim, J. H. Lee, J. Heo,
[26] T. Liu, X. Li, H. Zhang, J. Chen, J. Energy Storage 2018, 27, 1292. H. Y. Jung, H. T. Kim, Sci. Rep. 2017, 7, 10503.
[27] a) Q. Deng, P. Huang, W. X. Zhou, Q. Ma, N. Zhou, H. Xie, [58] K. Gong, F. Xu, J. B. Grunewald, X. Ma, Y. Zhao, S. Gu, Y. Yan, ACS
W. Ling, C. J. Zhou, Y. X. Yin, X. W. Wu, Adv. Energy Mater. 2017, 7, Energy Lett. 2016, 1, 89.
1700461; b) W. Li, Z. Zhang, Y. Tang, H. Bian, T. W. Ng, W. Zhang, [59] X. Wei, G.-G. Xia, B. Kirby, E. Thomsen, B. Li, Z. Nie, G. G. Graff,
C. S. Lee, Adv. Sci. 2016, 3, 1500276. J. Liu, V. Sprenkle, W. Wang, J. Electrochem. Soc. 2016, 163,
[28] B. Li, M. Gu, Z. Nie, Y. Shao, Q. Luo, X. Wei, X. Li, J. Xiao, A5150.
C. Wang, V. Sprenkle, Nano Lett. 2013, 13, 1330. [60] J. Luo, B. Hu, C. Debruler, Y. Bi, Y. Zhao, B. Yuan, M. Hu, W. Wu,
[29] K. Schmidt-Rohr, Q. Chen, Nat. Mater. 2008, 7, 75. T. L. Liu, Joule 2019, 3, 149.
[30] Z. Yuan, Y. Duan, H. Zhang, X. Li, H. Zhang, I. Vankelecom, Energy [61] B. H. Robb, J. M. Farrell, M. P. Marshak, Joule 2019, 3, 2503.
Environ. Sci. 2016, 9, 441. [62] K. L. Hawthorne, J. S. Wainright, R. F. Savinell, J. Electrochem. Soc.
[31] a) Z. Yuan, X. Li, Y. Duan, Y. Zhao, H. Zhang, J. Membr. Sci. 2015, 2014, 161, A1662.
488, 194; b) Z. Yuan, X. Li, Y. Zhao, H. Zhang, ACS Appl. Mater. [63] N. Arroyo-Currás, J. W. Hall, J. E. Dick, R. A. Jones, A. J. Bard, J.
Interfaces 2015, 7, 19446; c) Z. Yuan, X. Li, J. Hu, W. Xu, J. Cao, Electrochem. Soc. 2015, 162, A378.
H. Zhang, Phys. Chem. Chem. Phys. 2014, 16, 19841. [64] R. Tolouei, J. Harrison, C. Paternoster, S. Turgeon, P. Chevallier,
[32] H. Zhang, H. Zhang, X. Li, Z. Mai, J. Zhang, Energy Environ. Sci. D. Mantovani, Phys. Chem. Chem. Phys. 2016, 18, 19637.
2011, 4, 1676. [65] a) W. Wang, S. Kim, B. Chen, Z. Nie, J. Zhang, G.-G. Xia, L. Li,
[33] H. Zhang, H. Zhang, X. Li, Z. Mai, W. Wei, Energy Environ. Sci. Z. Yang, Energy Environ. Sci. 2011, 4, 4068; . b) W. Wang, Z. Nie,
2012, 5, 6299. B. Chen, F. Chen, Q. Luo, X. Wei, G. G. Xia, M. Skyllas-Kazacos,
[34] Y. Zhao, M. Li, Z. Yuan, X. Li, H. Zhang, I. F. Vankelecom, Adv. L. Li, Z. Yang, Adv. Energy Mater. 2012, 2, 487.
Funct. Mater. 2016, 26, 210. [66] B. Hu, C. DeBruler, Z. Rhodes, T. L. Liu, J. Am. Chem. Soc. 2017,
[35] W. Lu, Z. Yuan, Y. Zhao, H. Zhang, H. Zhang, X. Li, Chem. Soc. 139, 1207.
Rev. 2017, 46, 2199. [67] J. Yu, M. Salla, H. Zhang, Y. Ji, F. Zhang, M. Zhou, Q. Wang, Energy
[36] B. Tang, L. Shan, S. Liang, J. Zhou, Energy Environ. Sci. 2019, 12, Storage Mater. 2020, 29, 216.
3288. [68] J. Luo, A. Sam, B. Hu, C. DeBruler, X. Wei, W. Wang, T. L. Liu,
[37] Z. Yuan, Y. Yin, C. Xie, H. Zhang, Y. Yao, X. Li, Adv. Mater. 2019, 31, Nano Energy 2017, 42, 215.
1902025. [69] T. Páez, A. Martínez-Cuezva, J. Palma, E. Ventosa, ACS Appl.
[38] J. Fu, Z. P. Cano, M. G. Park, A. Yu, M. Fowler, Z. Chen, Adv. Mater. Energy Mater. 2019, 2, 8328.
2017, 29, 1604685. [70] M.-A. Goulet, M. J. Aziz, J. Electrochem. Soc. 2018, 165, A1466.
[39] X. G. Zhang, in Encyclopedia of Electrochemical Power Sources (Ed: [71] M. Cazot, G. Maranzana, J. Dillet, F. Beille, T. Godet-Bar,
J. Garche), Elsevier, Amsterdam, The Netherlands 2009, p. 454. S. Didierjean, Electrochim. Acta 2019, 321, 134705.
[40] Y. Song, J. Hu, J. Tang, W. Gu, Y. Fu, X. Ji, J. Electrochem. Soc. 2017, [72] E. S. Beh, D. De Porcellinis, R. L. Gracia, K. T. Xia, R. G. Gordon,
164, D230. M. J. Aziz, ACS Energy Lett. 2017, 2, 639.
[41] R. Wang, D. Kirk, G. Zhang, J. Electrochem. Soc. 2006, 153, C357. [73] Y.-S. Kim, J.-G. Kim, J. Ind. Eng. Chem. 2019, 70, 169.
[74] S. Yun, S. H. Park, J. S. Yeon, J. Park, M. Jana, J. Suk, H. S. Park, b) C. DeBruler, B. Hu, J. Moss, X. Liu, J. Luo, Y. Sun, T. L. Liu,
Adv. Funct. Mater. 2018, 28, 1707593. Chem 2017, 3, 961.
[75] N. Li, Y. Wang, D. Tang, H. Zhou, Angew. Chem., Int. Ed. 2015, 54, [103] W. Liu, Y. Liu, H. Zhang, C. Xie, L. Shi, Y. G. Zhou, X. Li, Chem.
9271. Commun. 2019, 55, 4801.
[76] N. Li, Z. Weng, Y. Wang, F. Li, H.-M. Cheng, H. Zhou, Energy [104] J. Luo, B. Hu, C. Debruler, T. L. Liu, Angew. Chem., Int. Ed. 2018,
Environ. Sci. 2014, 7, 3307. 57, 231.
[77] C. G. Anderson, Miner. Eng. 2016, 92, 248. [105] a) C. DeBruler, B. Hu, J. Moss, J. Luo, T. L. Liu, ACS Energy Lett.
[78] R. Demir-Cakan, M. Morcrette, J.-B. Leriche, J.-M. Tarascon, 2018, 3, 663; b) J. Luo, W. Wu, C. Debruler, B. Hu, M. Hu, T. L. Liu,
J. Mater. Chem. A 2014, 2, 9025. J. Mater. Chem. A 2019, 7, 9130.
[79] Z. Li, G. Weng, Q. Zou, G. Cong, Y.-C. Lu, Nano Energy 2016, 30, [106] B. Hu, Y. Tang, J. Luo, G. Grove, Y. Guo, T. L. Liu, Chem. Commun.
283. 2018, 54, 6871.
[80] A. Price, S. Bartley, S. Male, G. Cooley, Power Eng. J. 1999, 13, 122. [107] T. Janoschka, N. Martin, U. Martin, C. Friebe, S. Morgenstern,
[81] a) P. Zhao, H. Zhang, H. Zhou, B. Yi, Electrochim. Acta 2005, 51, H. Hiller, M. D. Hager, U. S. Schubert, Nature 2015, 527, 78.
1091; b) H. Zhou, H. Zhang, P. Zhao, B. Yi, Electrochim. Acta 2006, [108] J. E. Nutting, M. Rafiee, S. S. Stahl, Chem. Rev. 2018, 118, 4834.
51, 6304. [109] X. Wei, W. Xu, M. Vijayakumar, L. Cosimbescu, T. Liu, V. Sprenkle,
[82] Z. Li, M. S. Pan, L. Su, P.-C. Tsai, A. F. Badel, J. M. Valle, S. L. Eiler, W. Wang, Adv. Mater. 2014, 26, 7649.
K. Xiang, F. R. Brushett, Y.-M. Chiang, Joule 2017, 1, 306. [110] T. Janoschka, N. Martin, M. D. Hager, U. S. Schubert, Angew.
[83] L. Su, A. F. Badel, C. Cao, J. J. Hinricher, F. R. Brushett, Ind. Eng. Chem., Int. Ed. 2016, 55, 14427.
Chem. Res. 2017, 56, 9783. [111] Z. Chang, D. Henkensmeier, R. Chen, ChemSusChem 2017, 10,
[84] M. M. Gross, A. Manthiram, ACS Appl. Energy Mater. 2019, 2, 3445. 3193.
[85] L. Fan, I. I. Suni, J. Electrochem. Soc. 2019, 166, A1471. [112] J. Winsberg, C. Stolze, A. Schwenke, S. Muench, M. D. Hager,
[86] Y. Zhang, J. Liu, S.-L. Li, Z.-M. Su, Y.-Q. Lan, EnergyChem 2019, 1, U. S. Schubert, ACS Energy Lett. 2017, 2, 411.
100021. [113] D. L. Marshall, M. L. Christian, G. Gryn’ova, M. L. Coote,
[87] H. D. Pratt, N. S. Hudak, X. Fang, T. M. Anderson, J. Power Sources P. J. Barker, S. J. Blanksby, Org. Biomol. Chem. 2011, 9, 4936.
2013, 236, 259. [114] Y. Liu, M.-A. Goulet, L. Tong, Y. Liu, Y. Ji, L. Wu, R. G. Gordon,
[88] H. D. Pratt III, T. M. Anderson, Dalton Trans. 2013, 42, 15650. M. J. Aziz, Z. Yang, T. Xu, Chem 2019, 5, 1861.
[89] H. D. Pratt, W. R. Pratt, X. Fang, N. S. Hudak, T. M. Anderson, [115] T. Hagemann, M. Strumpf, E. Schröter, C. Stolze, M. Grube,
Electrochim. Acta 2014, 138, 210. I. Nischang, M. D. Hager, U. S. Schubert, Chem. Mater. 2019, 31,
[90] a) Y. Liu, S. Lu, H. Wang, C. Yang, X. Su, Y. Xiang, Adv. Energy 7987.
Mater. 2017, 7, 1601224; b) Y. Liu, H. Wang, Y. Xiang, S. Lu, J. Power [116] J. Winsberg, C. Stolze, S. Muench, F. Liedl, M. D. Hager,
Sources 2018, 392, 260. U. S. Schubert, ACS Energy Lett. 2016, 1, 976.
[91] J. J. Chen, M. D. Symes, L. Cronin, Nat. Chem. 2018, 10, 1042. [117] T. Janoschka, C. Friebe, M. D. Hager, N. Martin, U. S. Schubert,
[92] T. Feng, H. Wang, Y. Liu, J. Zhang, Y. Xiang, S. Lu, J. Power Sources ChemistryOpen 2017, 6, 216.
2019, 436, 226831. [118] T. Janoschka, S. Morgenstern, H. Hiller, C. Friebe,
[93] I. K. Song, M. A. Barteau, J. Mol. Catal. A: Chem. 2004, 212, 229. K. Wolkersdörfer, B. Häupler, M. Hager, U. S. Schubert, Polym.
[94] J. Friedl, M. V. Holland-Cunz, F. Cording, F. L. Pfanschilling, Chem. 2015, 6, 7801.
C. Wills, W. McFarlane, B. Schricker, R. Fleck, H. Wolfschmidt, [119] T. Hagemann, J. Winsberg, M. Grube, I. Nischang, T. Janoschka,
U. Stimming, Energy Environ. Sci. 2018, 11, 3010. N. Martin, M. D. Hager, U. S. Schubert, J. Power Sources 2018, 378,
[95] a) Q. Huang, H. Li, M. Gratzel, Q. Wang, Phys. Chem. Chem. Phys. 546.
2013, 15, 1793; b) F. Pan, J. Yang, Q. Huang, X. Wang, H. Huang, [120] J. Winsberg, T. Janoschka, S. Morgenstern, T. Hagemann,
Q. Wang, Adv. Energy Mater. 2014, 4, 1400567; c) C. Jia, F. Pan, S. Muench, G. Hauffman, J. F. Gohy, M. D. Hager, U. S. Schubert,
Y. G. Zhu, Q. Huang, L. Lu, Q. Wang, Sci. Adv. 2015, 1, e1500886; Adv. Mater. 2016, 28, 2238.
d) J. Ye, L. Xia, C. Wu, M. Ding, C. Jia, Q. Wang, J. Phys. D: Appl. [121] Y. Tian, K. H. Wu, L. Cao, W. H. Saputera, R. Amal, D. W. Wang,
Phys. 2019, 52, 443001; e) R. Yan, Q. Wang, Adv. Mater. 2018, 30, Chem. Commun. 2019, 55, 2154.
1802406. [122] W. Duan, R. S. Vemuri, J. D. Milshtein, S. Laramie, R. D. Dmello,
[96] M. Zhou, Q. Huang, T. N. P. Truong, J. Ghilane, Y. G. Zhu, C. Jia, J. Huang, L. Zhang, D. Hu, M. Vijayakumar, W. Wang, J. Liu,
R. Yan, L. Fan, H. Randriamahazaka, Q. Wang, Chem 2017, 3, R. M. Darling, L. Thompson, K. Smith, J. S. Moore, F. R. Brushett,
1036. X. Wei, J. Mater. Chem. A 2016, 4, 5448.
[97] a) J. Yu, L. Fan, R. Yan, M. Zhou, Q. Wang, ACS Energy Lett. [123] a) Y. Ding, G. Yu, Angew. Chem., Int. Ed. 2016, 55, 4772;
2018, 3, 2314; b) Y. Chen, M. Zhou, Y. Xia, X. Wang, Y. Liu, Y. Yao, b) B. Huskinson, M. P. Marshak, C. Suh, S. Er, M. R. Gerhardt,
H. Zhang, Y. Li, S. Lu, W. Qin, X. Wu, Q. Wang, Joule 2019, 3, 2255. C. J. Galvin, X. Chen, A. Aspuru-Guzik, R. G. Gordon, M. J. Aziz,
[98] Y. Zhou, G. Cong, H. Chen, N.-C. Lai, Y.-C. Lu, ACS Energy Lett. Nature 2014, 505, 195.
2020, 5, 1732. [124] a) C. Han, H. Li, R. Shi, T. Zhang, J. Tong, J. Li, B. Li, J. Mater.
[99] M. R. Gerhardt, L. Tong, R. Gómez-Bombarelli, Q. Chen, Chem. A 2019, 7, 23378; b) Y. Jing, Y. Liang, S. Gheytani, Y. Yao,
M. P. Marshak, C. J. Galvin, A. Aspuru-Guzik, R. G. Gordon, Nano Energy 2017, 37, 46; c) Y. Liang, Y. Jing, S. Gheytani, K. Y. Lee,
M. J. Aziz, Adv. Energy Mater. 2017, 7, 1601488. P. Liu, A. Facchetti, Y. Yao, Nat. Mater. 2017, 16, 841; d) Z. Luo,
[100] a) B. Yang, L. Hoober-Burkhardt, F. Wang, G. K. Surya Prakash, L. Liu, Q. Zhao, F. Li, J. Chen, Angew. Chem., Int. Ed. 2017, 56,
S. R. Narayanan, J. Electrochem. Soc. 2014, 161, A1371; b) B. Yang, 12561.
L. Hoober-Burkhardt, S. Krishnamoorthy, A. Murali, G. K. S. Prakash, [125] X. Wang, Z. Shang, A. Yang, Q. Zhang, F. Cheng, D. Jia, J. Chen,
S. R. Narayanan, J. Electrochem. Soc. 2016, 163, A1442. Chem 2019, 5, 364.
[101] T. Liu, X. Wei, Z. Nie, V. Sprenkle, W. Wang, Adv. Energy Mater. [126] Q. Zhao, W. Huang, Z. Luo, L. Liu, Y. Lu, Y. Li, L. Li, J. Hu, H. Ma,
2016, 6, 1501449. J. Chen, Sci. Adv. 2018, 4, eaao1761.
[102] a) S. Jin, E. M. Fell, L. Vina-Lopez, Y. Jing, P. W. Michalak, [127] a) M. Quan, D. Sanchez, M. F. Wasylkiw, D. K. Smith, J. Am. Chem.
R. G. Gordon, M. J. Aziz, Adv. Energy Mater. 2020, 10, 2000100. Soc. 2007, 129, 12847; b) S. Jin, Y. Jing, D. G. Kwabi, Y. Ji, L. Tong,
D. De Porcellinis, M.-A. Goulet, D. A. Pollack, R. G. Gordon, [137] T. J. Carney, S. J. Collins, J. S. Moore, F. R. Brushett, Chem. Mater.
M. J. Aziz, ACS Energy Lett. 2019, 4, 1342. 2017, 29, 4801.
[128] K. Wedege, E. Drazevic, D. Konya, A. Bentien, Sci. Rep. 2016, 6, 39101. [138] M. A. Goulet, L. Tong, D. A. Pollack, D. P. Tabor, S. A. Odom,
[129] J. E. Bachman, L. A. Curtiss, R. S. Assary, J. Phys. Chem. A 2014, A. Aspuru-Guzik, E. E. Kwan, R. G. Gordon, M. J. Aziz, J. Am.
118, 8852. Chem. Soc. 2019, 141, 8014.
[130] D. P. Tabor, R. Gómez-Bombarelli, L. Tong, R. G. Gordon, [139] E. W. Zhao, T. Liu, E. Jónsson, J. Lee, I. Temprano, R. B. Jethwa,
M. J. Aziz, A. Aspuru-Guzik, J. Mater. Chem. A 2019, 7, 12833. A. Wang, H. Smith, J. Carretero-González, Q. Song, C. P. Grey,
[131] S. Er, C. Suh, M. P. Marshak, A. Aspuru-Guzik, Chem. Sci. 2015, 6, Nature 2020, 579, 224.
885. [140] B. Hu, J. Luo, M. Hu, B. Yuan, T. L. Liu, Angew. Chem., Int. Ed.
[132] a) K. Lin, Q. Chen, M. R. Gerhardt, L. Tong, S. B. Kim, L. Eisenach, 2019, 58, 16629.
A. W. Valle, D. Hardee, R. G. Gordon, M. J. Aziz, Science 2015, [141] J. D. Milshtein, L. Su, C. Liou, A. F. Badel, F. R. Brushett, Electro-
349, 1529; b) D. G. Kwabi, K. Lin, Y. Ji, E. F. Kerr, M.-A. Goulet, chim. Acta 2015, 180, 695.
D. De Porcellinis, D. P. Tabor, D. A. Pollack, A. Aspuru-Guzik, [142] P. Leung, D. Aili, Q. Xu, A. Rodchanarowan, A. A. Shah, Sustain-
R. G. Gordon, M. J. Aziz, Joule 2018, 2, 1894; c) C. Wang, Z. Yang, able Energy Fuels 2018, 2, 2252.
Y. Wang, P. Zhao, W. Yan, G. Zhu, L. Ma, B. Yu, L. Wang, G. Li, [143] A. Hollas, X. Wei, V. Murugesan, Z. Nie, B. Li, D. Reed, J. Liu,
J. Liu, Z. Jin, ACS Energy Lett. 2018, 3, 2404; d) Y. Ji, M. A. Goulet, V. Sprenkle, W. Wang, Nat. Energy 2018, 3, 508.
D. A. Pollack, D. G. Kwabi, S. Jin, D. Porcellinis, E. F. Kerr, [144] C. Wang, X. Li, B. Yu, Y. Wang, Z. Yang, H. Wang, H. Lin, J. Ma,
R. G. Gordon, M. J. Aziz, Adv. Energy Mater. 2019, 9, 1900039. G. Li, Z. Jin, ACS Energy Lett. 2020, 5, 411.
[133] Z. Yang, L. Tong, D. P. Tabor, E. S. Beh, M.-A. Goulet, [145] C. Zhang, Z. Niu, S. Peng, Y. Ding, L. Zhang, X. Guo, Y. Zhao,
D. De P orcellinis, A. Aspuru-Guzik, R. G. Gordon, M. J. Aziz, Adv. G. Yu, Adv. Mater. 2019, 31, 1901052.
Energy Mater. 2018, 8, 1702056. [146] J. Hong, M. Lee, B. Lee, D. H. Seo, C. B. Park, K. Kang, Nat.
[134] J. Carretero-González, E. Castillo-Martínez, M. Armand, Energy Commun. 2014, 5, 5335.
Environ. Sci. 2016, 9, 3521. [147] A. Orita, M. G. Verde, M. Sakai, Y. S. Meng, Nat. Commun. 2016,
[135] a) L. Hoober-Burkhardt, S. Krishnamoorthy, B. Yang, A. Murali, 7, 13230.
A. Nirmalchandar, G. K. S. Prakash, S. R. Narayanan, J. Elec- [148] K. Lin, R. Gómez-Bombarelli, E. S. Beh, L. Tong, Q. Chen, A. Valle,
trochem. Soc. 2017, 164, A600; b) A. Murali, A. Nirmalchandar, A. Aspuru-Guzik, M. J. Aziz, R. G. Gordon, Nat. Energy 2016, 1,
S. Krishnamoorthy, L. Hoober-Burkhardt, B. Yang, G. Soloveichik, 16102.
G. K. S. Prakash, S. R. Narayanan, J. Electrochem. Soc. 2018, 165, [149] Y. Cui, X. Chen, Y. Wang, J. Peng, L. Zhao, J. Du, M. Zhai, Polymers
A1193. 2019, 11, 1482.
[136] L. Tong, M.-A. Goulet, D. P. Tabor, E. F. Kerr, D. De Porcellinis, [150] a) J. Bao, V. Murugesan, C. J. Kamp, Y. Shao, L. Yan, W. Wang,
E. M. Fell, A. Aspuru-Guzik, R. G. Gordon, M. J. Aziz, ACS Energy Adv. Theory Simul. 2019, 3, 1900167; b) B. Sanchez-Lengeling,
Lett. 2019, 4, 1880. A. Aspuru-Guzik, Science 2018, 361, 360.
Zhejun Li received her B.E. degree at Tianjin University in 2014 and her Ph.D. degree at The
Chinese University of Hong Kong in 2019, under the supervision of Prof. Yi-Chun Lu. She is
currently a postdoctoral research associate for Prof. Yi-Chun Lu. Her research interests mainly
focus on materials design and mechanistic investigation of high-energy-density batteries.
Yi-Chun Lu received her B.S. degree in materials science & engineering from National Tsing
Hua University, Taiwan in 2007 and earned her Ph.D. degree in materials science & engineering
from Massachusetts Institute of Technology in 2012. She worked as a Postdoctoral Fellow in
the Department of Chemistry at the Technische Universität München, Germany in 2013. She
is currently an associate professor of mechanical and automation engineering at The Chinese
University of Hong Kong. Her research interest centers on fundamental redox chemistry and
developing functional materials for clean energy storage and conversion.

Advanced Materials - 2020 - Li - Material Design of Aqueous Redox Flow Batteries Fundamental Challenges and Mitigation

Uploaded by

Copyright:

Available Formats

Advanced Materials - 2020 - Li - Material Design of Aqueous Redox Flow Batteries Fundamental Challenges and Mitigation

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Advanced Materials - 2020 - Li - Material Design of Aqueous Redox Flow Batteries Fundamental Challenges and Mitigation

Uploaded by

Copyright:

Available Formats

Review

Material Design of Aqueous Redox Flow Batteries:

redox flow systems to date.[7] However,

VO2+ (IV ) + H2O ↔ VO+2 ( V ) + 2H+ + e− (1)

Table 1. Performance of Zn-based ARFBs.

a)Based on catholyte volume only.

bis(trifluoromethanesulfonyl)amide, a ferrocene-grafted 3. Aqueous Organic Redox Flow Batteries

Table 2. Summary of performance of viologen derivatives as the anolyte in ARFBs.

FcNCl 1.06 60 700 (0.5 m, 0.01%/cycle)[66] 7 98%

(NH4)4 0.8 40 1000 cycles,1100 h, 0% decay (0.9 7.5 ≈100%

3.1l NA (9 ± 2) × 10−5 −0.29 PolyTEMPO 1.19 60 95 (0.24%/cycle)[107] 8 99.8%

Table 3. Summary of flow cell performance of quinone-based AORFBs.

Table 4. Electrochemical performance of AORFBs based on N-containing heteroaromatic compounds.

Ionic conductivity (σ) could be obtained through electro- 5.2. AORFBs

You might also like

Advanced Materials - 2020 - Li - Material Design of Aqueous Redox Flow Batteries Fundamental Challenges and Mitigation

Uploaded by

Copyright:

Available Formats

Advanced Materials - 2020 - Li - Material Design of Aqueous Redox Flow Batteries Fundamental Challenges and Mitigation

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Advanced Materials - 2020 - Li - Material Design of Aqueous Redox Flow Batteries Fundamental Challenges and Mitigation

Uploaded by

Copyright:

Available Formats

Review

Material Design of Aqueous Redox Flow Batteries:

redox flow systems to date.[7] However,

VO2+ (IV ) + H2O ↔ VO+2 ( V ) + 2H+ + e− (1)

Table 1. Performance of Zn-based ARFBs.

a)Based on catholyte volume only.

bis(trifluoromethanesulfonyl)amide, a ferrocene-grafted 3. Aqueous Organic Redox Flow Batteries

Table 2. Summary of performance of viologen derivatives as the anolyte in ARFBs.

FcNCl 1.06 60 700 (0.5 m, 0.01%/cycle)[66] 7 98%

(NH4)4 0.8 40 1000 cycles,1100 h, 0% decay (0.9 7.5 ≈100%

3.1l NA (9 ± 2) × 10−5 −0.29 PolyTEMPO 1.19 60 95 (0.24%/cycle)[107] 8 99.8%

Table 3. Summary of flow cell performance of quinone-based AORFBs.

Table 4. Electrochemical performance of AORFBs based on N-containing heteroaromatic compounds.

Ionic conductivity (σ) could be obtained through electro- 5.2. AORFBs

You might also like

VO2+ (IV ) + H2O ↔ VO+2 ( V ) + 2H+ + e− (1)