Evaluating bipolar redox-active molecules in non-aqueous organic redox flow batteries: trends and methods

In this review, recent advances in the development of organic bipolar redox‐active molecules (BRMs) for redox‐flow batteries (RFBs) are discussed, with special emphasis on their utilization in non‐aqueous systems. BRMs are single molecules that can be used as both anolyte and catholyte. Their unique redox chemistry enables identical components to be present in RFBs, which has distinct advantages in storage, mitigating crossover and compound degradation. Recent scientific innovations have uncovered three main strategies in preparing these bipolar species: developing stable radicals that are capable of both positive and negative redox reactions, covalently bridging electron donor‐acceptor motifs on one single molecule, and forming biredox eutectic mixtures through physically mixing active components. These approaches not only enabled symmetric flow battery design, but also resulted in electroactive materials with improved properties such as solubility, stability, and cell voltage. Nevertheless, much is still to be done. Importantly, developing BRMs that are competitive with vanadium RFBs in all performance indicators (e. g. solubility >1.0 M, voltage output >2.0 V, and cycling longevity >10 000 cycles) remains a challenge. Mechanistic studies including molecular dynamics, plausible degradation processes, and electrode/solution interface are necessary to peer into key parameters that dominate their battery performance. Accordingly, we review the current status and suggest future opportunities in these areas.

ABSTRACT In this review, recent advances in the development of organic bipolar redox‐active molecules (BRMs) for redox‐flow batteries (RFBs) are discussed, with special emphasis on their utilization in non‐aqueous systems. BRMs are single molecules that can be used as both anolyte and catholyte. Their unique redox chemistry enables identical components to be present in RFBs, which has distinct advantages in storage, mitigating crossover and compound degradation. Recent scientific innovations have uncovered three main strategies in preparing these bipolar species: developing stable radicals that are capable of both positive and negative redox reactions, covalently bridging electron donor‐acceptor motifs on one single molecule, and forming biredox eutectic mixtures through physically mixing active components. These approaches not only enabled symmetric flow battery design, but also resulted in electroactive materials with improved properties such as solubility, stability, and cell voltage. Nevertheless, much is still to be done. Importantly, developing BRMs that are competitive with vanadium RFBs in all performance indicators (e. g. solubility >1.0 M, voltage output >2.0 V, and cycling longevity >10 000 cycles) remains a challenge. Mechanistic studies including molecular dynamics, plausible degradation processes, and electrode/solution interface are necessary to peer into key parameters that dominate their battery performance. Accordingly, we review the current status and suggest future opportunities in these areas. ii ACKNOWLEDGEMENTS The text of this thesis is a reprint of the material as it appears in ChemElectroChem; permission for use has been granted by John Wiley & Sons, Inc. The co-authors listed in this publication directed and supervised research which forms the basis for the thesis. The author would like to thank Dr. Min Li for additional guidance and assistance throughout the writing process. The author would like to acknowledge financial support from the Joint Center for Energy Storage Research (JCESR), which is funded by the U.S. Department of Energy, Office of Science, Basic Energy Sciences. The author would also like to thank the Office of Undergraduate Research at the University of Utah for their support. © 2021 John Wiley & Sons, Inc iii TABLE OF CONTENTS ABSTRACT ii ACKNOWLEDGEMENTS iii 1. INTRODUCTION 1 2. TECHNICAL BACKGROUND 5 3. STRATEGIES IN DEVELOPING BIPOLAR REDOX-ACTIVE MOLECULES 9 3.1. Radicals that are Prone to both Reductive and Oxidative Events 9 3.2. Preparation of BRMs through Covalently Bridging Donor-Acceptors 18 3.2.1. Conjugating or Cross-Conjugating Linkers 18 3.2.1.1. 1,4-Diaminoanthraquinones (1,4-DAAQ) 18 3.2.1.2. Organic Complexes 21 3.2.2. Insulating Linkers 26 3.2.2.1. Ferrocene-Linker-Phthalimide 26 3.2.2.2. Ferrocene-Linker-Anthraquinone 29 3.2.2.3. Ferrocene-Linker-Fullerene 31 3.3. Formation of Bipolar Eutectic Mixtures through Physical Mixing 33 3.4. Other Alternative Strategies 36 3.5. Comparison of BRMs with Other ROMs in O-NRFBs 36 4. CONCLUSIONS 39 REFERENCES 41 iv 1. INTRODUCTION Redox flow batteries (RFBs) are a promising technique to address the growing need for renewable energy storage.1 As the energy demand increases, so too does the need to harness power from renewable sources that will not contribute to climate change. Globally, electricity production from solar and wind are increasing by an average of 10.1% and 6. % per year, respectively.2 However, a renewable energy portfolio composition greater than 20% can lead to grid destabilization.1a, 1c, 3 RFBs, which decouple the battery power and capacity, are an attractive solution for storing energy from intermittent energy sources while maintaining grid stability.1b, 3, 4 RFBs were first conceptualized in the 1970s5 and the pioneering study of the all‐vanadium redox flow battery was published in 1986.6 Since then, RFBs have been gaining exposure for their potential use in grid‐energy storage.1 The all‐vanadium redox flow battery (VRFB) is the most heavily researched RFB system to date.1c, 7 In this VRFB, related vanadium couples (V2+/V3+ and VO2+/VO2+) are used as the redox‐active species to comprise anolyte and catholyte when dissolved in an aqueous solvent. This VRFB has demonstrated solubilities up to 3 M8 and is stable up to 20,000 cycles.9 Importantly, the use of the same active element in anolyte and catholyte enables electrolyte regeneration in the event of active species crossover.10 Despite these advantages, the VRFB still experiences several challenges. First, the cost of a VRFB system is $500–$600 per KWh, 41% of which comes from the cost of membrane materials.11 Compared to the Department of Energy's recommendation of $100 per KWh, the price of a VRFB is still too high to make widely commercially available. Critically, the cell output of the VRFB is limited to the potential window of water (<1.5 V),12 preventing further improvement of the energy density. Consequently, while VRFBs demonstrate sufficient solubility and stability, their high cost and limited energy densities highlight a need to develop alternative redox‐active materials and solvent regimes. Non‐aqueous RFB systems (NRFBs) are increasingly investigated13 as an alternative to aqueous VRFBs due to their wide potential windows (up to 4 V).12 Accordingly, high cell voltages (≥2.0 V) are possible, making high energy densities achievable.12 Additionally, the use of non‐aqueous solvents opens up many possibilities in applying a wide range of cost‐effective, abundant redox‐active organic molecules (ROMs).1a, 14 These ROMs are further tunable using a molecular engineering strategy, which can improve properties such as solubility15 or redox potential.16 However, organic NRFBs (O−NRFBs) still face limitations including poor stability (often ≤200 cycles)17 and low solubility (mostly <1 M)18 for all charge states of ROMs.14 In particular, crossover is a challenge preventing the widespread use of NRFBs.19 Most NRFBs contain ROMs with unique active components, resulting in an asymmetric battery arrangement. Consequently, chemical gradients facilitate ROM crossover.20 Further, when these active species cross the cell membrane, permanent contamination occurs, leading to capacity decay.21 Therefore, to improve the battery's long‐term capacity, it is important to develop strategies to mitigate ROM crossover. An emerging strategy to address the crossover challenge in O−NRFBs is the use of bipolar redox‐active molecules (BRMs). Unlike other ROMs, BRMs are single parent molecules that can be both oxidized and reduced to facilitate positive and negative half‐ reactions during the charging/discharging process.22 As such, these molecules can be used as both anolyte and catholyte in a battery. The resulting symmetric redox flow 2 battery (SRFB) not only shares the same active elements, but has identical solution components in each half‐cell, which has important advantages. First, there is no chemical or electrochemical gradient across the membrane in the discharged state, so SRFBs can potentially be stored indefinitely without irreversible side reactions.22 Second, the use of identical redox‐active material reduces the chemical gradient of electroactive species, mitigating crossover. Even if crossover does occur, SRFBs may undergo self‐discharge instead of permanent contamination.7 Finally, it is potentially possible to recover the capacity loss due to compound degradation through regular polarity reversals, theoretically increasing the battery lifetime.22 These significant advantages make BRMs a promising development in RFB technology. In this paper, we review the state‐of‐the‐art advancements made in BRMs with emphasis on their utilization in NRFB systems. Redox‐active centers with promising performance in BRMs have been identified and are presented here, including ferrocenes, 1,4‐diaminoanthraquinones (1,4‐DAAQs), and nitronyl nitroxides (NNs). Within the framework of these redox cores, three main strategies for designing BRMs have been determined; development of stable radicals that are prone to both positive and negative redox events, covalently bridging donor‐acceptor functional groups, and formation of bipolar eutectic mixtures. Based on this analysis, we describe the main redox‐active centers employed in these BRMs and assess their relative strengths and weaknesses. Specifically, we examine how properties of these unique redox centers and their molecular engineering strategies have been used to address parameters including cell voltage, ROM solubility, and cycling stability with the purpose of improving energy density. As other ROMs have been utilized in achieving SRFBs (e. g., physical mixing of 3 catholyte and anolyte compounds), we will also examine these strategies. Finally, we discuss challenges and future opportunities of these BRMs for the wide implementation of O−NRFBs in grid‐energy storage. 4 2. TECHNICAL BACKGROUND As shown in Scheme 1, a RFB mainly consists of an electrochemical cell and two connected storage tanks. Anode and cathode materials are dissolved in solvents containing supporting electrolytes to form redox‐active solutions called anolyte and catholyte, respectively. The anolyte and catholyte are separated in two half‐cells by a selective membrane,1b, 3, 4 which is designed to impede crossover of electroactive species while allowing transport of ions (X+ and Y−) from supporting electrolytes to maintain charge balance.20 Pumps are used to flow the electrolyte solutions between the electrochemical cell and the external storage tanks. Scheme 1. Schematic illustration of a redox‐flow battery. During RFB charging, the catholyte is oxidized in one half‐cell, while the anolyte is reduced in the other half‐cell. Electrons pass through an external circuit to convert electrical energy into chemical energy. During battery discharging, the reverse process occurs. Chemical energy is converted into electrical energy at the electrochemical cell as 5 the catholyte is reduced and the anolyte is oxidized. These charged species are pumped to external electrolyte reservoirs for storage. The separation of the electrochemical cell and the electrolyte storage tanks is the defining feature of the RFB and allows for the independent scaling of its components. This decoupling of power and capacity is a key advantage compared to other enclosed battery systems (e. g., lithium‐ ion batteries).1b, 3, 4 In addition to their scalability, RFBs are attractive due to their potentially low cost1b, 4 and achievable long lifetime.23 For example, the capital cost of lithium‐ion batteries range from $850–$5,000 per KWh,23 well over the $180 per KWh reached in some RFBs.24 Additionally, lithium‐ion batteries have a lifetime of 5–10 years, while the lifetime of RFBs potentially extends beyond 10 years.23 RFBs are further advantageous in their ability to use a wide range of redox‐active materials in aqueous or non‐aqueous systems.13a, 25 The combination of ROMs with non‐ aqueous solvents, in particular, has afforded a variety of RFBs with distinct advantages in achieving high energy density.1a, 16, 26 While O−NRFBs enable the development of energy‐dense systems, the energy density is often limited by the solubilities of ROMs, which must exceed 1 M in the non‐ aqueous solvent to compete with organic aqueous RFBs (O−ARFBs).12, 16 However, at ROM concentrations above 0.5 M, O−NRFBs experience high viscosity (>10 cP) and low solution conductivity (<5 mS/cm).18 Therefore, battery cycling is often limited to low concentrations of active species (≤0.5 M), which inhibits energy density. Another significant challenge to O−NRFB performance is the compatibility of ion‐exchange membranes (IEMs) in non‐aqueous solvents.19 While IEMs are widely used in O−ARFBs, they exhibit ionic conductivity one order of magnitude lower (0.2–0.5 6 mS/cm) in O−NRFBs.27 This decreased conductivity reduces the potential for high power densities in non‐aqueous regimes. Consequently, size‐exclusion membranes are often used for O−NRFBs. While this separator type is a cheaper alternative to IEMs, small molecule flow cells using this membrane only accessed a maximum of 50% discharge capacity.28 Additionally, the size‐exclusion membrane exhibited faster capacity decay than the IEM due to higher rates of crossover.28 Because crossover significantly reduces the capacity of the RFB, it is an important challenge to address in O−NRFBs. The crossover performance of electroactive materials can be evaluated by their permeabilities (P) using a H‐cell.29 As depicted in Scheme 2a, one chamber of the H‐cell, termed the retentate side, is filled with ROM solution; while the permeate side only contains supporting electrolyte. The volumes of the retentate and permeate sides are set to be equal for the balance of osmotic pressure. A bridge between the chambers can further eliminate this pressure‐driven crossover. After sealing the H‐cell with a desired membrane in between, both sides of the H‐cell are stirred continuously, and cyclic voltammetry (CV) is conducted at a series of time intervals. The concentration of ROM that has permeated can be calculated from the calibration curve of the peak current vs. concentration. The permeability (P) is then derived from the following equation shown in Scheme 2b. 7 Scheme 2. a) H‐cell set up for permeability measurements. b) Equation applied to derive permeability of a system. These permeability measurements can be employed to analyze whether ROM structure (e. g., ionic or neutral,29a size29b, 29c), or synthesized membranes30 reduce the crossover of active species. Notably, this set up simulates the scenario where a RFB has distinct ROMs in each compartment, and only neutral ROMs are present. In practical, however, when ROMs are charged, migration of these active species driven by potential gradient may become non‐trivial. Nevertheless, it is safe to conclude that when both chambers contain identical solution components, ROM crossover is significantly mitigated. Therefore, application of BRMs provides a paradigm shift in resolving crossover. These BRMs will be discussed in the following sections. 8 3. STRATEGIES IN DEVELOPING BIPOLAR REDOX-ACTIVE MOLECULES Recent advancements in BRMs center on strategic screening of a variety of redox‐ active motifs to improve battery performance, including cell voltage, ROM solubility, and cycling stability. As applied to diverse redox‐active moieties, three main strategies in developing BRMs were identified: (i) development of stable radicals that can be involved in both positive and negative redox reactions; (ii) covalently linking electron donor‐ acceptor pairs through either conjugating or insulating spacers; and (iii) physically mixing two electroactive species with or without supporting electrolytes to form a eutectic mixture. Accordingly, we will discuss how these approaches are used to promote the performance of BRMs in O−NRFBs. Examples include molecular engineering of linkage to enhance solubility, choice of two electroactive moieties to power cell output, and incorporation of bulky backbones to mitigate crossover. Regarding the stability of BRMs, we will examine the cycling efficiencies of BRMs, such as coulombic efficiency (CE), voltage efficiency (VE), and energy efficiency (EE), as well as capacity retention. In parallel, other alternative approaches in achieving SRFBs, such as a solution mixture of catholyte and anolyte, have also been discussed. Finally, we compare these BRMs to other ROMs to assess their relative strengths and weaknesses. In light of that, challenges and future opportunities are suggested. 3.1. Radicals that are Prone to both Reductive and Oxidative Events Radicals possess singly occupied molecular orbitals (SOMO) that render them applicable in both reductive and oxidative electron transfer processes. Most radicals, however, are inherently highly reactive, resulting in a plethora of decay pathways 9 including radical coupling and propagating to other spin‐paired molecules.17 Radical stabilization, therefore, is crucial. Incorporation of electron‐withdrawing or electron‐ donating groups, conjugation, or steric hindrance near the radical centers have been shown to be effective stabilization methods.31 To be employed as BRMs, neutral radicals are preferable materials as charged radicals are often relatively short‐lived states and are difficult to synthesize. To date, C‐centered and heteroatom‐centered (e. g., N, O, S) are the most studied neutral radicals. Note that in the application of BRMs, not only the neutral state needs to be stable, but also the reduced and oxidized states. This requirement further restrains the number of neutral radicals that could be employed in SRFBs. For example, 2,2,6,6‐tetramethylpiperidine‐1‐oxyl (TEMPO) is the most‐researched neutral radical. Nevertheless, the insufficient stability of aminoxyl anions limits their applications to be used only as a cathodic material.32 Some neutral radicals, however, are stabilized in all three oxidation states and have revealed their potential in BRM applications.17, 32a In this section, we will highlight these radicals, particularly the methods applied in improving their stability and other critical physicochemical properties. Nitronyl nitroxides (NNs) are O‐centered neutral radicals. Unlike TEMPO, the πconjugated fraction through the iminium fragment stabilizes the aminoxyl anions, giving rise to good stability in their reduced states. In addition to the improved stability, NNs often possess high solubilities in non‐aqueous solvents.33 The molecule 2‐phenyl‐4,4,5,5‐ tetramethylimidazoline‐1‐oxyl‐3‐oxide (PTIO, BRM 1), for example, is a commercially available NN radical (Figure 1a). Duan et al. have pioneered the study of 1 in O−NRFBs without any purification (>98.0% purity).33a Voltammetric analysis confirmed the 10 chemical and electrochemical reversibility of both redox reactions as well as a cell voltage of 1.73 V (Figure 1b). A key advantage of BRM 1 is its high solubility of 2.6 M in acetonitrile (MeCN), as determined by UV‐vis and electron spin resonance (ESR) microscopy. Nevertheless, despite the high solubility of the neutral state, the charged species proved to be significantly less soluble. Galvanostatic cycling tests of 0.1 M 1 at a current density of 20 mA/cm exhibited 96% CE, 75% VE, and 72% EE over 35 cycles (Figure 1e). When this concentration was increased to 0.5 M and cycled for 15 cycles, these efficiencies dropped to 90%, 67%, and 60%, respectively (Figure 1f). Cycling at concentrations >0.5 M was not possible due to the increased viscosity and rate of crossover. Importantly, all cycling tests demonstrated severe capacity decay. In particular, approximately 75% capacity decay occurred during a long‐term cycling test of 100 cycles. This confirms the instability of the charged species, particularly at high concentrations. As a result, the achievable energy density was limited to 9 Wh L−1, compared to its theoretical value of 11.6 Wh L−1. While the energy density of BRM 1 is comparable to that of BRMs 9 and 10, it is much lower than most BRMs. Because the energy density is limited by the poor solubility and stability of the charged species, molecular engineering and electrolyte optimization are necessary to achieve solubilities closer to the high solubility of the neutral state. 11 Figure 1. Strategic engineering of BRMs employing radicals. a) Structure and the redox chemistry of BRM 1. b) Cyclic voltammogram of 0.1 M BRM 1 over 500 cycles in 1.0 M tetrabutylammonium hexafluorophosphate (TBAPF6) in MeCN, recorded at 100 mV/s. c) Electron spin resonance (ESR) spectra of original PTIO, PTIO+, PTIO−, and a mixture of PTIO+ and PTIO− in a 1:1 volume ratio. PTIO+ and PTIO− were prepared by charging a flow cell containing 5.0 mM original PTIO in 1.0 M TBAPF6. Charged PTIO does not have unpaired electrons so is not ESR active. d) FTIR spectra of 0.5 M PTIO, PTIO+, and PTIO− in 1.0 M TBAPF6 in MeCN. e) Bulk electrolysis evaluation of 0.1 M BRM 1 over 35 cycles in 1.0 M TBAPF6 in MeCN at 20 mA/cm2. f) Bulk electrolysis evaluation of 0.5 M BRM 1 over 15 cycles in 1.0 M TBAPF6 in MeCN at 20 mA/cm2. Reproduced with permission from Ref. [33a]; copyright (2016) Royal Society of Chemistry. Concurrently with demonstrating the use of BRM 1 in symmetric O−NRFBs, Duan et al. have introduced Fourier transform infrared spectroscopy (FTIR) to monitor the state of charge (SOC) of RFBs.33a SOC diagnostics promote the safe, cost‐effective operation of RFBs as they detect risks including ROM decomposition, electrolyte imbalance, gas evolution, overcharge, and ROM crossover. Figure 1c shows ESR spectra of BRM 1’s three charged states as well as a mixture of its positive and negative states, demonstrating that charged BRMs neutralize to a discharged state upon crossover. When FTIR spectra of the neutral and charged states of BRM 1 were observed and cross‐ 12 referenced to ESR spectra, FTIR was found to successfully distinguish between the three charged states (Figure 1d). Additionally, the spectra exhibited low interference from supporting electrolyte and high dependence on BRM concentration while also surpassing ESR in accessibility, cost, and response time. As such, FTIR is proven to be a potentially attractive technique for SOC diagnostics in RFBs. While 1 has been successfully implemented as single parent molecules in O−NRFBs, it is also possible to combine PTIO units using a linker strategy to improve solubility and capacity. This strategy has been evaluated in the molecule 4,4′‐[oxybis(2,1‐ ethanediyloxy‐2,1‐ethanediyloxy)]bisnitronylnitroxide (BRM 2), which was synthesized by covalently linking two identical PTIO units with a tetraethylene glycol chain (Figure 2a).33b By using two redox‐active units, it was possible to double the theoretical capacity compared to the single redox‐active unit. Importantly, the glycol chain was able to increase the solubility of the neutral molecule to 3.8 M in MeCN, the highest solubility reported of the BRMs. However, despite exhibiting the highest neutral‐state solubility, BRM 2 demonstrated the second lowest energy density at 4.1 Wh L−1. This common challenge encountered in BRM 1 and 2 highlights that the solubilities of the charged states must be carefully examined, as they may limit the cycling performance and energy density of the battery. 13 Figure 2. a) Chemical structure of BRM 2. b) Charge/discharge stability of BRM 2 at a current density of 1 mA/cm2. The experiments were performed using a 5 cm2 static cell with 16 mM BRM 2 in 0.5 M TBAPF6/MeCN. c) Cycling performance when increasing the concentration of BRM 2 to 100 mM. Reproduced from Ref. [33b]; copyright (2017) The Authors. Charge/discharge experiments in a static electrochemical cell employing 0.016 M BRM 2 found improved performance compared to 1, with only 20 % capacity fade over 75 cycles (Figure 2b). This high stability was maintained at current densities of 3–5 mA cm−1. Additionally, cycling efficiencies of 95% CE, 86% VE, and 82% EE were observed. However, the cycling performance declined significantly at higher concentrations. Cycling tests of 0.1 M 2 in a pumped flow battery exhibited 40% capacity fade after 20 cycles (Figure 2c). Further, an increase in electrolyte viscosity impaired mass transport, resulting in a 20% material utilization and steep voltage profile. Bulk electrolysis of 0.5 M 2 was not possible as the active species precipitated out of the solution. This poor performance at high concentrations is attributed to the instability of the charged species, which significantly limits the energy density. The need for molecular engineering and electrolyte optimization is reemphasized to address this challenge. In particular, functionalization of the molecule with electron‐donating groups and conjugation has the potential to stabilize charged species.46 Very recently, Moutet et al. reported the use of helical carbenium ions, dimethoxyquinacridiniums (DMQA+, BRM 3), as both anolyte and catholyte in an O−NRFB.34 As depicted in Figure 3a, BRM 3 (C+) can be either electrochemically reduced or oxidized to form a neutral radical (C.) or a radical dication (C.2+). As aforementioned, it is critical to develop radical stabilization methods as they are highly 14 active. Interestingly, instead of functionalization of the radical center, Moutel et al. uncovered that solvents played an important role in their stability. For example, both reduced and oxidized forms of BRM 3 showed electrochemical reversibility in dichloromethane (DCM), but the reduced radicals (C.) were stable only up to a few minutes. When N,N‐dimethylformamide (DMF) was applied, the oxidized radical dications (C.2+) turned out to be irreversible. However, in MeCN, both electroactive species exhibited good reversibility and stability at the time scale of CV (Figure 3b). A cell output of 2.21 V could be achieved when both radical species are applied as electroactive electrolytes, demonstrating a high‐voltage O−NRFB. Although the mechanism behind the role of solvents in radical stability remains to be discussed, these findings indicate that solvent engineering might provide an opportunity to regulate radical performance. Figure 3. a) illustration of the redox chemistry of BRM 3 as both catholyte and anolyte. b) cyclic voltammogram of BRM 3 in 0.1 M TBAPF6/MeCN. c) The set‐up of H‐cell used for charge/discharge experiments. d) Corresponding potential and current profile of 12th to 14th cycles. e) Capacity and efficiency performance of the H‐cell cycling. The inset shows in the cyclic voltammograms of working and counter chambers before and after cycling. Reproduced with permission from Ref. [34]; copyright (2020) American Chemical Society. 15 The viability of BRM 3 as anolyte and catholyte was further evaluated via H‐cell electrolysis experiments (Figure 3c). Figure 3d shows the cycling parameter employed – one‐electron charge/discharge at 90% state of charge (SOC) was conducted via a constant current followed by a constant voltage galvanostatic charging. Significantly, BRM 3 showed capacity retention >90% over 550 cycles, and ∼100% CE over 800 cycles (Figure 3e). Post analysis through CV showed a slight change in the reduction peak of C., indicating the high stability of BRM 3 as anolyte. The CV of the counter chamber was performed to reveal the stability of 3 as catholyte. As shown in the inset of the Figure 3e, the characteristic peaks of the 3 as catholyte was lost after cycling, suggesting the poor solubility of C.2+. However, it is worth noting that due to the fact that the counter chamber is only used to complete electric circuit, and any reactions may occur to balance the current observed at the working electrode, the results obtained from analyzing the auxiliary side may not be accurate. In fact, in the subsequent two‐electron charge/discharge stress experiments where in each cycle, C. was sequentially oxidized to C+, and C.2+ and then reduced to form C., the ∼100% capacity retention over 80 cycles indicated the good stability of BRM 3 as both catholyte and anolyte. Therefore, cycling of BRM 3 as catholyte in the working chamber is suggested to gain more insights into the stability of the radical dication (C.2+). Another potential improvement is the solubility of these redox‐active species. C+ and C. were reported to reach saturation at 32.4 and 149.1 mM in pure MeCN. This low solubility impedes its usage in energy‐dense battery applications. As revealed in the BRM 1 and 2, incorporation of polar glycol chains might be an effective approach to enhance solubility. 16 It is worth mentioning that in theory, any electroactive materials capable of both positive and negative redox events can be employed as BRMs. However, for practical applications, a wide span in their two half‐cell potentials is necessary. A lack of a wide potential gap introduces perils to achieving high energy density and power output. Examples in this regard include verdazyl radical 3‐phenyl‐1,5‐di‐p‐tolylverdazyl (BRM 4)35 and pyridinium derivatives.36 BRM 4 is a N‐centered neutral radical. Similar to NNs, the extended π conjugation delocalizes the charges on the N‐atom and stabilizes its anionic state. Regardless of the high chemical and electrochemical stability demonstrated at all states of charge, the small separation between its anodic and cathodic peaks resulted in a cell voltage of only 0.98 V (Figure 4a). This low potential limits the energy density to 0.52 Wh L−1, significantly lower than is necessary for commercial RFBs. This trend is also observed for pyridinium derivatives – C‐centered radicals.36 Figure 4b displays the redox behavior of a 4‐benzoylpyridine derivate (BRM 5).36b Although the redox potential varies with supporting electrolytes, the voltage gap was consistently within ∼0.6 V. Consequently, these materials are recommended to be applied in O−NRFBs as two‐ electron storage anolytes rather than as BRMs. 17 Figure 4. Examples of BRMs with low cell potentials, recommended instead as anolyte materials. a) Structure and cyclic voltammogram of BRM 4 in 0.5 M TBAPF6 in MeCN at 100 mV/s. Reproduced with permission from Ref. [35]; copyright (2019) Elsevier Inc. b) Structure and cyclic voltammogram of 5.0 mM BRM 5 in 0.5 M electrolyte and MeCN at 100 mV/s. Reproduced with permission from Ref. [36b]; copyright (2017) American Chemical Society. 3.2. Preparation of BRMs through Covalently Bridging Donor-Acceptors As for spin‐paired molecules, an individual functional group is hardly capable of both positive and negative reactions. Hence, to develop bipolar species, the combination of electron donor‐acceptor functional groups, preferably with a large potential gap, is warranted. This can be achieved through either covalently linking two electroactive moieties or physically mixing to form a eutectic mixture. Because these two approaches apply different chemistries, we will discuss them in section 3 and section 4, respectively. When donor‐acceptor pairs are covalently linked in one molecule, the choice of linkage employed has become a critical factor in the overall performance as it may dominate electronic interactions between electron donors and acceptors. We, therefore, classify this group of materials into two structural variations: BRMs with conjugating or cross‐conjugating linkers, and BRMs with insulating linkers. 3.2.1. Conjugating or Cross-Conjugating Linkers 3.2.1.1. 1,4-Diaminoanthraquinones (1,4-DAAQ) 1,4‐DAAQs consist of electron‐donor (amine) and electron‐acceptor (quinone) functional groups, thus enabling both positive and negative half‐reactions. Structurally, the aromatic base of the 1,4‐DAAQ stabilizes the doubly charged quinone and amines through charge delocalization, facilitating electrochemical accessibility of two reduced 18 and two oxidized states.37 The existence of two redox couples allows for two‐electron transfer reactions, which opens up a way to double the energy density of RFBs. It is further advantageous to have an odd number of oxidation states to prevent low battery potentials at low states of charge.22 Many 1,4‐DAAQs are commercially available at low cost. In particular, the dye disperse blue 134 (BRM 6)22 is a cheap, abundant material with five reversible oxidation states in MeCN. As determined by CV (Figure 5a),38 the cell potentials for the inner and outer redox couple are 1.76 V and 2.72 V, respectively, with high corresponding energy densities of 47 Wh mol−1 and 120 Wh mol−1. However, cycling tests of the inner redox couple exhibited voltage efficiencies under 50% as a result of a measured series resistance of 100 ohms, owing to large electrode separation (≥1 cm) and low supporting electrolyte concentration (100 mM). As such, a voltage drop of ≥0.4 V was obtained. Within the voltage limits of ±3 V imposed in the galvanostatic experiment, this high resistance suggested that only the inner redox couple was accessible, negating the energy density benefits of the outer redox couple. The RFB is further limited by the poor solubility of BRM 6 in organic solvents. For example, its neutral form was sparingly soluble in MeCN (less than 10 mM), and in toluene, it saturated at ∼200 mM. With the presence of supporting electrolyte (e. g., 0.1 M tetrabutylammonium perchlorate, TBAP), the maximum solubility was suppressed to about 100 mM in 3:2 MeCN/toluene mixture.22 This low concentration significantly reduces the energy densities achievable. As a result, while 6 demonstrates the promise of 1,4‐DAAQs as bipolar electrolyte material with high cell potential and energy density, its solubility needs improvement in order to be feasible in NRFBs. 19 Figure 5. Strategic engineering of 1,4‐DAAQ‐based BRMs. a) Cyclic voltammogram of 0.01 M BRM 6 in 0.1 M TEATf2N in dimethoxyethane (DME), recorded on a platinum disk working electrode at 50 mV/s. b) Cyclic voltammogram of 0.01 M BRM 7 in 0.1 M TEATf2N in DME, recorded on a platinum disk working electrode at 50 mV/s. c) Bulk electrolysis evaluation of the catholyte for BRM 6 over 100 cycles in DME (full symbols) and MeCN (hollow symbols) at current density of +/−0.21 A/dm2. Upward triangles (▴) represent charge capacities, downward triangles (▾) represent discharge capacities, and squares (▪) represent coulombic efficiencies. Dotted lines indicate the theoretical capacity for one (2.68 Ah/L) and two (5.36 Ah/L) electron transfer reactions. d) Corresponding bulk electrolysis evaluation of the anolyte for BRM 7. Adapted with permission from Ref. [38]; copyright (2020) American Chemical Society. To address the low solubility of 1,4‐DAAQs, Geysens et al. have incorporated oligo(ethylene oxide) chains in the structure, forming 1,4‐bis((2‐(2‐(2‐ methoxyethoxy)ethoxy) ethyl)amino)anthracene‐9,10‐dione (BRM 7) (Figure 5b).38 The addition of these polar chains lowers the melting point of the molecule to 25 °C by increasing the entropy of dissolution, which in turn increases its solubility. Studies by 20 UV‐vis spectroscopy revealed that the solubility of BRM 7 was 2.2 M in its pure state and >1 M in both MeCN and DME solvents, exhibiting a significant improvement from BRM 6. Consequently, without sacrificing cell potentials, BRM 7 demonstrated an increase of energy densities to 49 Wh mol−1 and 122 Wh mol−1 for the inner and outer redox couples, respectively. Bulk electrolysis of 7 was conducted and its behavior as catholyte and anolyte was observed, as shown in Figure 5c and Figure 5d, respectively. Cycling of the inner redox couple found 20% capacity fade over 100 cycles. Bulk electrolysis of the outer redox pair, however, led to fast capacity fade due to the quasi‐ reversibility of the second oxidation reaction. The difference in the reversibility of the two oxidation states also indicates that the doubly oxidized intermediate species is more prone to side reactions, either with electrolytes or to internal decomposition events. In fact, both 1,4‐DAAQs have exhibited poor cycling performance of the outer redox pair, demonstrating a key challenge when accessing high oxidation states. However, since the position and length of incorporated ether groups have shown an influence on the chemical reversibility of ROMs,39 it is possible to further improve the stability of the highly charged 1,4‐DAAQ derivatives through molecular genomic screening of a variety of oligoglycol chains. 3.2.1.2. Organic Complexes Organic complexes often possess large extended π‐systems and donor‐acceptor groups, enabling multiple oxidation events. Porphyrins, for example, are a natural pigment containing an aromatic π‐conjugated macrocycle. Ma et al. evaluated the redox chemistry of a porphyrin ((5,10,15,20‐Tetraphenylporphyrin, H2TPP, BRM 8) by DFT 21 calculations, and revealed that for the highest occupied molecular orbital/singly occupied molecular orbital (HOMO/SOMO) at different redox states, the aromatic p‐conjugated framework in H2TPP enabled charge delocalization within the structure.40 This stability indicates the possibility of two one‐electron oxidation and reduction processes as depicted in Figure 6b. Cyclic voltammetric analysis confirmed this high stability and reversibility (Figure 6a). Remarkably, the potential gap between the 1/1’ and 4/4’ peaks makes a voltage output of 2.83 V achievable. Another advancement is the preparation of a highly conductive ion‐selective membrane by utilizing a commercial poly(vinylidene fluoride) (PVDF) porous membrane as the substrate, and Y‐zeolite as the sieves. As shown in Figure 6c–e, the pore size of the zeolite is right between the size of H2TPP, TBA+ and ClO4−. Accordingly, due to the size exclusion, only ClO4− freely diffuses between the catholyte and anolyte compartments. The crossover of H2TPP could be further suppressed by applying a symmetric cell architecture. As a result, BRM 8 observed a capacity retention of 99.98% per cycle for 200 cycles. This extraordinary stability has suggested the synergistic effects of the good chemical/electrochemical stability of a molecule (chemical structure), the size‐exclusion approach (membrane), and the symmetric structure of a battery (cell design) on enhancing cycling longevity. Notably, this cycling stability was also seen at temperatures between 20 °C and −40 °C (Figure 6f), indicating the great potential of utilizing BRM 8 for energy storage in cold regions. 22 Figure 6. a) Structure and cyclic voltammogram of 1 mM BRM 8 in 0.1 M TBAP/DCM at 20 mV/s. b) The redox chemistry of BRM 8 as catholyte and anode. c) Illustration of the ion‐selectivity of the Y‐PVDF ion‐selective membrane. d) Structure of the Y‐zeolite used in the preparation of Y‐PVDF membrane. e) The high‐resolution transmission electron microscopy (HRTEM) and selected‐area electron (SAED) images of the Y‐zeolite. f) Bulk electrolysis at varying temperatures of 200 mg/mL 8 in 1 M TBAP and DCM over 200 cycles at 1 mA/cm2. Adapted with permission from Ref. [40]; copyright (2018) Wiley‐ VCH. Regardless of the outstanding stability, the solubility of BRM 8 was limited to only 10 mM in DCM, and was even less soluble in other organic solvents such as MeCN and propylene carbonate (PC). Consequently, a suspension of the BRM 8 with 5 wt% Ketjen Black was utilized, enabling higher concentrations (200 mg/mL) and an energy density of 24.68 Wh L−1. However, the suspension of the electrolyte greatly increased the solution viscosity and ohmic resistance. As a result, less than 50% VE and EE were observed. In conclusion, although BRM 8 demonstrated superior capacity retention at a wide range of temperatures, its low solubility needs to be addressed. This is possible by 23 adding a metal center to the organic ligands of the complex, forming a metal coordination complex (MCC).41 While it is beyond the scope of this paper to explore the effects of the metal centers applied for MCCs in O−NRFBs, it is important to note that the choice of metal center significantly impacts the electrochemistry of MCCs.42 As such, these organic complexes can be strategically engineered to enhance properties such as viscosity, ohmic resistance, and solubility. In recent years, other alternative molecules in this category have also been reported.43 An interesting example is the polythiophene microparticle (BRM 9).44 Depicted in Figure 7a, BRM 9 is another π‐aromatic organic molecule. Through n‐doping or p‐doping, polythiophene undergoes reduction or oxidation (Figure 7b). Importantly, BRM 9 is conductive, which has the potential to improve ionic conductivity of non‐ aqueous systems. However, this promise is limited by the low conductivity of the neutral species, leading to 34.5% capacity utilization during bulk electrolysis of 9 in 1.0 M TEABF4/PC. Charge and discharge capacities were also observed to decrease during cycling, possibly resulting from BRM 9 sedimentation (Figure 7c). Another peril is the poor chemical and electrochemical reversibility. As shown in Figure 7b, this inferior electrochemical performance resulted in unstable cell voltage, ranging from 1.0 V to 3.0 V. Therefore, although conductive microparticles open an avenue to prepare functional BRMs, the particle aggregation together with the low conductivity of neutral species require further investigation of this type of material. 24 Figure 7. a) Schematic illustration of a redox flow battery employing polythiophene microparticles as bipolar electroactive species. b) Electrochemical performance of BRM 9 revealed by cyclic voltammetry: (left) n‐doping and (right) p‐doping. c) Charge/discharge results of a flow cell using BRM 9 suspensions (0.1 eq. L−1 of thiophene, Ketjen black 2 g L−1) at 0.5 mA/cm2 in 1.0 M TEABF4/PC solution. Reproduced with permission from Ref. [44]; copyright (2016) Royal Society of Chemistry. The present series of BRMs demonstrates the effectiveness of incorporating extended π‐systems into redox units, particularly in improving stability. A potential pitfall of this strategy is that if the molecules are not functionalized from natural compounds, preparing such artificial structures can be very synthetically challenging. To date, this concept has been illustrated by Fornari et al.45 They outlined 35 electron donor‐ acceptor pairs on two aromatic rings and further connected on a compact molecular core. The redox potentials were computed and compared to their single‐redox counterparts – 25 the molecules that only possess one of the redox motifs. Although the selected electrolytes are aqueous, the structure‐property relationships discovered suggest the critical role that electronic interaction plays in an aromatic framework. Because this electronic interaction is within the aromatic system, the principle in theory also applies to non‐aqueous electrolytes. Such studies demonstrate the possibility of designing conjugated BRMs with large cell output. Consequently, aromatic π‐conjugated BRMs have great potential in achieving both high stability and voltage. 3.2.2 Insulating Linkers In contrast to aromatic π‐conjugated BRMs, connecting two redox‐active moieties with insulating linkers is not only synthetically practical, but also offers many possibilities in structural variations as a result of a wide range of electron‐donors or ‐ acceptors being available. In addition, the two redox‐active units can be either identical31b or varied.44 The insulating linkers often have additional advantages in improving solubility.31b, 44 Thus, in this section, we will highlight advancements made in this area of interest. 3.2.2.1 Ferrocene-Linker-Phthalimide Ferrocenes are electron‐donor moieties that undergo one‐electron oxidation at 0.597 V vs Ag/Ag+. They show promising redox activity in O−NRFBs due to their high stability and reversibility.47 Hwang et al. have done pioneering work in combining ferrocene with phthalimide groups for applications in BRMs.46a, 46b While phthalimide itself demonstrates a large reduction potential (−2.07 vs Ag/Ag+), the redox functionality 26 is limited by the instability of the ketyl anion radical, which causes an irreversible oxidation event.48 However, the substitution of the hydrogen atom at the N‐site of the phthalimide with an electroactive ferrocene has improved the chemical stability of the reduced species.46a Together with the additional redox reactivity presented by ferrocene, the synthesized N‐ferrocenylphthalimide (BRM 10) has become a viable material for SRFBs.46a This combination of ferrocene and phthalimide exemplifies the combi‐ molecule strategy, in which two distinct redox‐active groups are covalently linked. As shown in Figure 8a, the oxidation peak of phthalimide was present in BRM 10, indicating the enhanced chemical stability of the ketyl radical after functionalization. Further cyclic voltammetric analysis confirmed the chemical reversibility of the phthalimide redox reaction, as the peak current ratio was consistently found to be ∼1 and independent of scan rate. Based on these scan‐rate dependent cyclic voltammograms, the rate constant was calculated to be 3.14–3.37×10−4 cm s−1, indicating an electrochemically quasi‐reversible reaction. 27 Figure 8. Strategic engineering of ferrocene‐linker‐phthalimide BRMs. a) Voltammetric analysis of the chemical reversibility of phthalimide, ferrocene/phthalimide mixture, and BRM 10. For all the cyclic voltammograms, 10 mM of electroactive samples dissolved in 1.0 M TBABF4/1,3‐dioxolane was employed with scan rate of 100 mV/s. Reproduced with permission from Ref. [46a]; copyright (2018) Elsevier Inc. b) Comparison of the solubility and working voltage for BRM 10 (FcPI) and BRM 11 (α‐FcEtPI). c) Nyquist plot of the coin‐cell utilizing 0.1 M or 0.6 M α‐FcEtPI as a single redox couple with the equivalent circuit shown in the inset. d) The equivalent series resistance (ESR) and charge transfer resistance (Rct) obtained from impedance data (left) and viscosity at 0.1 M and 0.6 M as a function of concentration (right). Adapted with permission from Ref. [46b]; copyright (2019) Elsevier Inc. Symmetric cycling tests employing 0.1 M BRM 10 with 1.0 M tetrabutylammonium tetrafluoroborate (TBABF4) in 1,3‐dioxolane in an H‐type coin cell found ∼24% decay in discharge capacity over 50 cycles. High cycling efficiency was observed through VE (90.1–96.5%), EE (88.3–92.7%), and CE (96.4–97.3%). Finally, it is worth noting that coupling phthalimide and ferrocene provides a 1.94 V RFB. However, despite this large voltage output, the energy density of the battery is limited to 7.80 Wh L−1. This low energy density is mainly attributed to the low solubility of 10. Because the structures of the phthalimide and ferrocenyl moieties are conducive to well‐ ordered packing in the solid‐state,49 the resulting increase in intermolecular interactions lowers the solubility of the molecule.50 Consequently, a maximum solubility of 0.3 M in 1.0 M TBABF4/1,3‐dioxolane was achieved for 10, while only 0.1 M was applied for cycling. Therefore, although BRM 10 demonstrates a high cell potential with good energy efficiency, it requires improvement in solubility in order to achieve high energy density. To inhibit molecular packing of solids for solubility improvement,51 BRM 10 was modified to design N‐(α‐ferrocenyl)ethylphthalimide (BRM 11).46b The α‐methyl group 28 used to link the phthalimide and ferrocenyl moieties was able to reduce intermolecular interactions and therefore increase the solubility (Figure 8b). As a result, the measured solubility of BRM 11 was found to be 0.81 M – a nearly 3‐fold enhancement in solubility compared to its parent molecule, BRM 10. This significant improvement in solubility resulted in an equally substantial increase in energy density to 21.06 Wh L−1 while maintaining a similar working voltage of 1.98 V. Importantly, 0.6 M of 11 was employed for cycling in 1.0 M TBABF4/1,3‐dioxolane and experienced <1% capacity fade over 50 cycles with a CE of 97.8%. This stable cycling at 0.6 M is an impressive feature of the battery. However, this high concentration comes at the cost of great ohmic resistance and cell polarization (Figure 8c–d). As the concentration of 11 was increased from 0.1 M to 0.6 M, a corresponding increase in viscosity by a factor of 1.8 was measured, impeding the ionic conductivity of the electrolyte. As a result, the ohmic resistance of the solution and separator (series resistance) increased by a factor of 1.6 and cell polarization of 0.35 V occurred. Increases in viscosity and resistance are common challenges of RFB operation at high concentrations.18, 26 Accordingly, while this derivatized molecule improves upon the solubility of 10 by a factor of ∼3, the concentration‐dependent conductivity and viscosity highlight that applying concentrated electroactive fluids in O−NRFBs to achieve high energy density needs to be carefully examined. 3.2.2.2. Ferrocene-Linker-Anthraquinone In addition to phthalimide, ferrocenes have been linked with anthraquinone moieties, whose two redox couples have the potential to increase the energy density (Figure 9a).52 Cyclic voltammetry found 1‐(ferrocenylmethyl‐amino)‐anthraquinone 29 (BRM 12) to be quasi‐reversible with cell potentials of 1.42 V and 2.06 V for the inner and outer redox pairs, respectively (Figure 9b). However, cyclic voltammograms of the outer redox pair demonstrated poor cycling over 70 scans. Accordingly, only the inner redox pair was selected for bulk electrolysis. As shown in Figure 9c, symmetric cycling of 0.01 M 12 in DMF with 0.1 M tetramethylammonium bis(trifluoromethanesulfonyl)imide (TEATFSI) demonstrated high cycling efficiency with 90.8% CE, 90.1% VE, and 81.8% EE over 100 cycles at 2 mA cm−2. While energy density was not reported, a peak power density of 16 mW cm−2 was observed at 100% SOC (Figure 9d). Nevertheless, a discharge capacity of ∼45% was observed after 100 cycles due to side reactions of the negatively charged species. Furthermore, UV‐vis spectroscopy found BRM 12 to have a maximum solubility of 0.017 M in DMF and 0.016 M in 0.1 M TEATFSI/DMF, among the lowest of the BRMs employing ferrocene. As such, an improvement of cycling stability and solubility is necessary to improve the longevity and energy density of the molecule. This may be done through strategic optimization of the linker chain. 30 Figure 9. a) Illustration of the redox chemistry of BRM 12. b) Cyclic voltammograms of 5 mM BRM 12 in 0.1 M TEATFSI/DMF at the scan rate of 100 mV s−1. c) Cycling performance of 0.01 M BRM 12 over 100 cycles at the current density of 2 mA cm−2. A home‐designed flow battery was used with Daramic AA‐250 as membrane (250 μm in thickness). d) Polarization curves of the flow battery at different SOC. Reproduced with permission from Ref. [52]; copyright (2020) Elsevier Inc. 3.2.2.3. Ferrocene-Linker-Fullerene In line with the incorporation of phthalimide and anthraquinone, another application of ferrocenes is anchoring these redox‐active moieties onto bulky molecules to address ORM crossover. The Friedl et al. have demonstrated this concept by attaching 1–4 ferrocene units to a fullerene cage (Figure 10a, b, BRM 13–BRM 16).46c The large diameter (7.09 Å) of fullerenes make them attractive for use in RFBs, as the combination of bulky ROMs with a size‐exclusion membrane can be an effective strategy for reducing crossover.29c Fullerenes are further advantageous in their potential for high energy density. The fullerene derivatives were observed to be soluble in a variety of solvents and exhibited particularly high solubility in ortho‐dichlorobenzene (O−DCB) (300±22 mg/mL for 14), with a general trend of higher solubility upon addition of more ferrocene units. As such, BRMs 14–16 offer a significantly higher energy density (44–80 WhL mol−1) than BRMs 10 and 11, despite a lower cell voltage of 1.64 V. Energy density data was not provided for BRM 13, which exhibited the lowest solubility due to the least amount of ferrocene functionalization. 31 Figure 10. Cyclic voltammograms of a) 13 and b) 14, 15, and 16 in 0.1 M TBABF4/ODCB. c) Bulk electrolysis evaluation for symmetric and asymmetric cells. Black square (▪) stands for the cells with 1 mM of fullerene derivative 13 as catholyte and anolyte; red circle (•) represents the cells with 1 mM of fullerene derivative 14 as catholyte and 1.5 mM of indene‐C60 bis‐adduct (ICBA) as anolyte; blue triangle (▴) indicates the cells with 1 mM of fullerene derivative 15 as catholyte and 2 mM of ICBA as anolyte. d) Corresponding normalized capacity. Bulk electrolysis was performed in ortho‐dichlorobenzene (O−DCB) with 0.1 M TBABF4. Adapted with permission from Ref. [46c]; copyright (2018) American Chemical Society. To evaluate the cell chemistries of BRMs 14–16, bulk electrolysis employing three different catholyte‐anolyte couples was conducted. Figure 10c shows the discharge capacity of each individual cell design, and Figure 10d displays their normalized capacities. From these, it can be concluded that (i) bulk electrolysis of a symmetric cell employing 1 mM 14 in 0.1 M TBABF4 /O−DCB exhibited ∼43% discharge capacity loss over 100 cycles. Low voltage efficiency was observed during cycling due to the large size of the supporting electrolyte, which slowed the electrolyte transfer needed to balance 32 the charge. (ii) Asymmetric cells employing 1 mM of 15 and 16, respectively, as catholyte and 1.5 mM of indene‐C60 bis‐adduct (ICBA) as anolyte demonstrated a higher capacity than the symmetric 14 cell due to a higher degree of functionalization of the ferrocene units. However, the cells cycling 15 and 16 experienced a faster capacity fade than the cell employing 14, owing to a higher rate of crossover in an asymmetric setup. To improve the rate performance, a solvent system consisting of DMF and 0.1 M LiCl was utilized for cycling 1 mM BRM 16 in an asymmetric cell. A current of 1 mA was possible in DMF, representing a 10‐fold improvement from a cell employing O−DCB. Furthermore, a current density of 1200 mA cm−1 is potentially possible, a great improvement upon the maximum 0.07 A cm−1 possible in VRFBs.53 However, HPLC and mass spectroscopy studies found only traces of 16 after 100 cycles, indicating poor cycling stability in DMF. Additionally, the solubility of BRM 16 was an order of magnitude smaller in DMF (0.017 M) than in O−DCB (0.12 M). Therefore, while BRMs 13–16 show promise in mitigating crossover and providing high energy density, optimization of the electrolyte system as well as solubility is necessary to make these BRMs viable in RFBs. 3.3. Formation of Bipolar Eutectic Mixtures through Physical Mixing Supplementing radicals and covalently linking electron donor‐acceptors to prepare BRMs, a third promising strategy has been demonstrated that employs a eutectic electrolyte in which electroactive species are physically mixed with or without solvent or salts (Figure 11a). The interactions of the anionic and/or cationic components reduce the lattice energy of the mixture, which has the ability to lower the melting point of the 33 mixture below the individual melting points of the separate species (Figure 11c).54 This strategy enables the preparation of highly concentrated redox‐active materials by physically mixing components, thereby making great energy densities readily accessible. Several eutectic mixtures have employed organic molecules as the electroactive species, such as viologen, TEMPO, and phthalimide‐based electrolytes.54, 55 Figure 11. Strategic engineering of BRMs employing a eutectic mixture. a) Preparation strategy for BRM 17. b) Structure of BRM 17 and demonstration of eutectic working principle. c) Phase diagram of BRM 17, demonstrating eutectic strategy of increasing concentration. d) Cyclic voltammogram of 0.05 M 17 with TBABF4, collected at a scan rate of 50 mV/s. e) Bulk electrolysis evaluation of 0.1 M BRM 17 over 500 cycles at 60 mA/cm2. f) Cycling efficiencies of 1.0 M BRM 11 at 60 mA/cm2. Reproduced with permission from Ref. [56]; copyright (2019) Wiley‐VCH. Zhang et al. prepared a biredox eutectic electrolyte by physically mixing N‐butyl‐ phthalimide and 1,1‐dimethylferrocene (BRM 17) (Figure 11a–c).56 While BRM 17 contains the same redox‐active species as BRMs 10 and 11, the eutectic strategy enables a concentration of 3.5 M, a more than 4‐fold improvement upon 11 and an almost 12‐fold improvement on 10 (Figure 11a). Accordingly, an energy density of 84.4 Wh L−1 is 34 possible. Cyclic voltammograms of the individual components in 0.1 M TBABF4/MeCN measured a peak ratio of ∼1, confirming chemical stability of both anodic and cathodic species, as well as small peak separations indicating the electrochemical reversibility. A cyclic voltammogram of 0.05 M 17 determined a cell potential of 1.8 V (Figure 11d). As shown in Figure 11e, bulk electrolysis of 0.1 M 17 detected a 28% capacity fade over 500 cycles at a current density of 60 mA cm−2. This demonstrates a much longer cycling stability than all other BRMs studied in O−NRFBs. While a 93% CE was observed, a 51% EE was calculated during cycling. In a further flow cell test employing 1.0 M 17, VE and EE decreased from 50% to 45% and 66% to 54%, respectively, over 20 cycles (Figure 11f). This low cycling efficiency, along with an observed decay in peak current, is likely due to the crossover of electroactive materials. At 1.0 M cycling, a viscosity of 5.1 mPa s and ionic conductivity of 18.5 mS cm−1 was observed. When the concentration increased to 2.0 M, the material utilization ratio decreased significantly from the high ratio at 1.0 M (96%) as a result of increased viscosity (16.1 mPa s) and corresponding low ionic conductivity (3.3 mS cm−1). The limited solubility of the TBABF4 supporting salt (<2 M) contributed to the decreased ionic conductivity at higher concentrations and caused electromigration at high current density. Therefore, while eutectic mixtures promise high solubility and good cycling stability, further improvement of both the solubility of the supporting electrolyte and the ion selectivity of the separator are suggested. It is also worth mentioning that a eutectic mixture employing a single electroactive species with supporting electrolyte is another possibility for achieving high material utilization with low viscosity, as additional supporting salt is not needed.57 35 3.4. Other Alternative Strategies With the rapid development of O−NRFBs, the past decade has seen a surge of interest in developing SRFBs to address active‐species crossover.32a, 43a, 58 Aside from preparing BRMs, one alternative strategy is utilizing a mixture of catholyte and anolyte. In fact, due to the ease of preparation, such approach has been commonly used in the literature.59 However, the effectiveness of these mixed electrolytes as compared to BRMs in battery performance remains to be discussed. For example, the solubility of BRMs was computed and found to be higher than their single redox counterparts,45 yet this still requires experimental evidence. Because the presence of supporting electrolytes decreases the solubility limits of ROMs in solution,26 adding catholyte into anolyte or vice versa in theory also decreases the maximum concentration of ROMs. Further, it is reported that for many ROMs, the solubility of their neutral states often does not translate to the solubility of their reduced or oxidized forms.33a, 39 Accordingly, the actual solubility of BRMs compared to mixed electroactive species at different oxidation states is still under scrutiny. 3.5 Comparison of BRMs with Other ROMs in O-NRFBs Over the last decade, the availability of a large number of organic electroactive molecules and the versatility of molecular engineering have prompted a surge of interest in developing robust ROMs for RFB applications.1a, 14, 60 BRMs, as a distinct type of ROM, have received considerable attention as a result of their unique property of bearing both catholyte and anolyte functional groups. This feature enables the development of symmetric flow batteries, which is critical in mitigating the crossover of active species. 36 Recent studies have further illustrated other added benefits, indicating their great potential in advancing ROMs. For instance, the fact that mixing reduced and oxidized PTIO species resulted in the formation of their neutral state33a suggests that such a BRM enables ‘cell self‐rebalancing’ when crossover occurs. As such, cross‐contamination is eliminated. In addition, coupling symmetric cell design with a size‐exclusion membrane40 showed synergistic effects in addressing crossover. Importantly, the three main strategies developed – stable radicals that are prone to both positive and negative redox reactions, covalently linked donor‐acceptors, and bipolar eutectic mixtures – provide opportunities for developing versatile BRMs. Regardless of these advances, the development of BRMs is still in its early stage and many questions remain to be answered. We highlight the key examples reported and summarize their relative advantages and weaknesses. As shown in Table 1, the cell potential for most BRMs is still within 2.0 V. Although BRM 7 has a voltage output of 2.83 V, the solubility is limited to only 10 mM in DCM. The energy density of O−NRFBs has been reported up to 200 W h L−1.61 Clearly, there is still much to improve in both solubility and cell voltage. In addition, a second‐electron transfer is achievable, but cycling of these high oxidation states showed significant capacity decay.38 Permeability is an important factor in evaluating crossover,29a, 29c yet such studies are often absent in the literature. Further, mechanistic studies including charge transport dynamics, decomposition processes and products are still rare. Notably, due to the complicated structure‐property relationship of ROMs, improvement of one physicochemical property is often achieved at the cost of sacrificing other merits. 37 Therefore, recent advances have seen great potential in utilizing BRMs for O−NRFBs, yet there remain important opportunities for growth. Table 1. Summary of some key example BRMs reported. 38 4. CONCLUSIONS The renaissance of renewable energy harvesting has prompted remarkable innovation in RFBs. O−NRFBs, in particular, are advantageous in grid energy storage as a result of their wide potential windows and a wide range of abundant electroactive molecules. Among the ROMs developed, BRMs, single molecules that can be used as both anolyte and catholyte, stand out as a distinct class of electroactive species. Their unique redox chemistry enables identical components to be present in RFBs, which has distinct advantages in mitigating crossover and degradation. In recent years, a large variety of BRMs have been reported in the literature, signifying the critical role of molecular engineering in improving BRM performance. Specifically, three main strategies are developed: preparation of stable radicals that enable both positive and negative redox chemistry, covalently connecting donor‐acceptor functional groups in one single molecule, and formation of bipolar eutectic mixtures. We, therefore, highlight key advances and exemplify the impact on improving BRM performance such as cell voltage, solubility, and cycling stability. The present series of BRMs exhibit promising properties and lay an important foundation in symmetric cell designs. Despite this promise, for wide implementation of BRMs in O−NRFBs, the need to develop a robust structure that features high solubility (>1.0 M), voltage output (2∼3 V), and cycling longevity (10,000 cycles based on vanadium redox‐flow batteries) is still a challenge. Although many molecules have been screened, their properties are often only partially improved. Moreover, not all crucial properties (e. g., solubility, redox potential, kinetics, stability) are reported. While solubility is a major hurdle for O−NRFBs to realize high energy density, this evaluation is often absent or only examined in pure 39 organic solvents. As the presence of supporting electrolytes can significantly suppress solubility, the data obtained in pure solvents does not necessarily translate to their dissolution behavior with salts. Another potential pitfall is that solubility varies at different charge states. Accordingly, a systematic evaluation of the maximum concentration of neutral and charged species with supporting electrolytes is necessary. As for stability examination, bulk electrolysis in H‐cells is widely performed, yet flow cell testing is often absent. Considering experimental conditions (e. g., O2/H2O level, stir/flow rate and cutoff voltage) are vital, it is also imperative to promote standardization in experimental protocols. To further stretch the research boundaries of BRMs, the development of robust computational modeling to predict structure‐property relationships is of importance. Mechanistic studies (e. g., molecular dynamics, decompositions, and electrode‐solution interface) are additionally essential as they can gain insights into this complex battery system and may inspire crucial factors that revolutionize control of key physicochemical properties. In summary, BRMs have shown great promise for promoting high energy density in O−NRFBs, yet much is still to be done to encourage their widespread application. 40 REFERENCES 1. a) J. A. Luo, B. Hu, M. W. Hu, Y. Zhao, T. L. Liu, ACS Energy Lett. 2019, 4, 2220– 2240; b) G. L. Soloveichik, Chem. Rev. 2015, 115, 11533– 11558; c) Z. Yang, J. Zhang, M. C. Kintner-Meyer, X. Lu, D. Choi, J. P. Lemmon, J. Liu, Chem. Rev. 2011, 111, 3577– 3613. 2. J. Conti, P. Holtberg, J. Diefenderfer, A. LaRose, J. T. Turnure, L. Westfall, USDOE Energy Information Administration (EIA), Washington, DC, United States, 2016. 3. A. Z. Weber, M. M. Mench, J. P. Meyers, P. N. Ross, J. T. Gostick, Q. H. Liu, J. Appl. Electrochem. 2011, 41, 1137– 1164. 4. C. Ponce de León, A. Frías-Ferrer, J. González-García, D. A. Szánto, F. C. Walsh, J. Power Sources 2006, 160, 716– 732. 5. M. Bartolozzi, J. Power Sources 1989, 27, 219– 234. 6. M. Skyllas-Kazacos, M. Rychcik, R. G. Robins, A. G. Fane, M. A. Green, J. Electrochem. Soc. 1986, 133, 1057– 1058. 7. K. Lourenssen, J. Williams, F. Ahmadpour, R. Clemmer, S. Tasnim, J. Energy Storage 2019, 25, 100844. 8. S. Roe, C. Menictas, M. Skyllas-Kazacos, J. Electrochem. Soc. 2015, 163, A5023– A5028. 9. H. R. Jiang, J. Sun, L. Wei, M. C. Wu, W. Shyy, T. S. Zhao, Energy Storage Mater. 2020, 24, 529– 540. 10. M. Ulaganathan, V. Aravindan, Q. Y. Yan, S. Madhavi, M. Skyllas-Kazacos, T. M. Lim, Adv. Mater. Interfaces 2016, 3, 1500309. 11. C. Ding, H. Zhang, X. Li, T. Liu, F. Xing, J. Phys. Chem. Lett. 2013, 4, 1281– 1294. 12. R. M. Darling, K. G. Gallagher, J. A. Kowalski, S. Ha, F. R. Brushett, Energy Environ. Sci. 2014, 7, 3459– 3477. 13. a) K. Gong, Q. R. Fang, S. Gu, S. F. Y. Li, Y. S. Yan, Energy Environ. Sci. 2015, 8, 3515– 3530; b) W. Wang, Q. T. Luo, B. Li, X. L. Wei, L. Y. Li, Z. G. Yang, Adv. Funct. Mater. 2013, 23, 970– 986; L. Su, M. Ferrandon, J. A. Kowalski, J. T. Vaughey, F. R. Brushett, J. Electrochem. Soc. 2014, 161, A1905– A1914. 14. Y. Ding, C. K. Zhang, L. Y. Zhang, Y. G. Zhou, G. H. Yu, Chem 2019, 5, 1964– 1987. 15. X. L. Wei, L. Cosimbescu, W. Xu, J. Z. Hu, M. Vijayakumar, J. Feng, M. Y. Hu, X. C. Deng, J. Xiao, J. Liu, V. Sprenkle, W. Wang, Adv. Energy Mater. 2015, 5, 1400678. 16. J. A. Kowalski, L. Su, J. D. Milshtein, F. R. Brushett, Curr. Opin. Chem. Eng. 2016, 13, 45– 52. 17. C. G. Armstrong, K. E. Toghill, Electrochem. Commun. 2018, 91, 19– 24. 18. J. L. Zhang, R. E. Corman, J. K. Schuh, R. H. Ewoldt, I. A. Shkrob, L. Zhang, J. Phys. Chem. C 2018, 122, 8159– 8172. 19. S. H. Shin, S. H. Yun, S. H. Moon, RSC Adv. 2013, 3, 9095– 9116. 41 20. L. Su, R. M. Darling, K. G. Gallagher, W. Xie, J. L. Thelen, A. F. Badel, J. L. Barton, K. J. Cheng, N. P. Balsara, J. S. Moore, J. Electrochem. Soc. 2015, 163, A5253. 21. M. Pugach, M. Kondratenko, S. Briola, A. Bischi, Appl. Energy 2018, 226, 560– 569. 22. R. A. Potash, J. R. McKone, S. Conte, H. D. Abruna, J. Electrochem. Soc. 2016, 163, A338– A344. 23. M. Skyllas-Kazacos, M. H. Chakrabarti, S. A. Hajimolana, F. S. Mjalli, M. Saleem, J. Electrochem. Soc. 2011, 158, R55– R79. 24. T. B. Liu, X. L. Wei, Z. M. Nie, V. Sprenkle, W. Wang, Adv. Energy Mater. 2016, 6, 1501449. 25. a) A. Hollas, X. L. Wei, V. Murugesan, Z. M. Nie, B. Li, D. Reed, J. Liu, V. Sprenkle, W. Wang, Nat. Energy 2018, 3, 508– 514; b) S. R. Narayan, A. Nirmalchandar, A. Murali, B. Yang, L. Hoober-Burkhardt, S. Krishnamoorthy, G. K. S. Prakash, Curr. Opin. Electrochem. 2019, 18, 72– 80. 26. M. Li, Z. Rhodes, J. Cabrera-Pardo, S. Minteer, Sustain. Energy Fuels 2020. 27. Y. Y. Lai, X. Li, Y. Zhu, ACS Appl. Polym. Mater. 2020, 2, 113– 128. 28. E. C. Montoto, G. Nagarjuna, J. S. Moore, J. Rodriguez-Lopez, J. Electrochem. Soc. 2017, 164, A1688- A1694. 29. a) J. Chai, A. Lashgari, X. Wang, C. K. Williams, J. Mater. Chem. A 2020; b) K. H. Hendriks, S. G. Robinson, M. N. Braten, C. S. Sevov, B. A. Helms, M. S. Sigman, S. D. Minteer, M. S. Sanford, ACS Cent. Sci. 2018, 4, 189– 196; c) S. E. Doris, A. L. Ward, A. Baskin, P. D. Frischmann, N. Gavvalapalli, E. Chenard, C. S. Sevov, D. Prendergast, J. S. Moore, B. A. Helms, Angew. Chem. Int. Ed. 2017, 56, 1595– 1599; Angew. Chem. 2017, 129, 1617– 1621. 30. A. Shrestha, K. H. Hendriks, M. S. Sigman, S. D. Minteer, M. S. Sanford, Chem. Eur. J. 2020, 26, 5369– 5373. 31. a) B. H. Tang, J. T. Zhao, J. F. Xu, X. Zhang, Chem. Sci. 2020, 11, 1192– 1204; b) Y. P. Wang, M. Frasconi, J. F. Stoddart, ACS Cent. Sci. 2017, 3, 927– 935. 32. a) V. V. Sentyurin, O. A. Levitskiy, T. V. Magdesieva, Curr. Opin. Electrochem. 2020, 24, 15– 23; b) T. Sukegawa, I. Masuko, K. Oyaizu, H. Nishide, Macromolecules 2014, 47, 8611– 8617. 33. a) W. T. Duan, R. S. Vemuri, J. D. Milshtein, S. Laramie, R. D. Dmello, J. H. Huang, L. Zhang, D. H. Hu, M. Vijayakumar, W. Wang, J. Liu, R. M. Darling, L. Thompson, K. Smith, J. S. Moore, F. R. Brushett, X. L. Wei, J. Mater. Chem. A 2016, 4, 5448– 5456; b) T. Hagemann, J. Winsberg, B. Haupler, T. Janoschka, J. J. Gruber, A. Wild, U. S. Schubert, NPG Asia Mater. 2017, 9, e340. 34. J. Moutet, J. M. Veleta, T. L. Gianetti, ACS Appl. Energ. Mater. 2020, 4, 9– 14. 35. G. D. Charlton, S. M. Barbon, J. B. Gilroy, C. A. Dyker, J. Energy Chem. 2019, 34, 52– 56. 36. a) C. S. Sevov, R. E. Brooner, E. Chenard, R. S. Assary, J. S. Moore, J. Rodriguez-Lopez, M. S. Sanford, J. Am. Chem. Soc. 2015, 137, 14465– 14472; b) K. H. Hendriks, C. S. Sevov, M. E. Cook, M. S. Sanford, ACS Energy Lett. 2017, 2, 2430– 2435. 37. V. Vatanen, J. A. Pedersen, J. Chem. Soc. Perkin Trans. 2 1996, 2207– 2212. 42 38. P. Geysens, Y. Li, I. Vankelecom, J. Fransaer, K. Binnemans, ACS Sustainable Chem. Eng. 2020, 8, 3832– 3843. 39. N. H. Attanayake, J. A. Kowalski, K. V. Greco, M. D. Casselman, J. D. Milshtein, S. J. Chapman, S. R. Parkin, F. R. Brushett, S. A. Odom, Chem. Mater. 2019, 31, 4353– 4363. 40. T. Ma, Z. Pan, L. Miao, C. Chen, M. Han, Z. Shang, J. Chen, Angew. Chem. Int. Ed. 2018, 57, 3158– 3162; Angew. Chem. 2018, 130, 3212– 3216. 41. R. W. Hogue, K. E. Toghill, Curr. Opin. Electrochem. 2019, 18, 37– 45. 42. J. Kucharyson, 2017. 43. a) V. V. Sentyurin, O. A. Levitskiy, T. V. Magdesieva, Curr. Opin. Electrochem. 2020; b) J. Winsberg, T. Hagemann, S. Muench, C. Friebe, B. Haupler, T. Janoschka, S. Morgenstern, M. D. Hager, U. S. Schubert, Chem. Mater. 2016, 28, 3401– 3405. 44. S. H. Oh, C. W. Lee, D. H. Chun, J. D. Jeon, J. Shim, K. H. Shin, J. H. Yang, J. Mater. Chem. A 2014, 2, 19994– 19998. 45. R. P. Fornari, M. Mesta, J. Hjelm, T. Vegge, P. de Silva, ACS Materials Lett. 2020, 239– 246. 46. a) S. Hwang, H. S. Kim, J. H. Ryu, S. M. Oh, J. Power Sources 2018, 395, 60– 65; b) S. Hwang, H.-s. Kim, J. H. Ryu, S. M. Oh, J. Power Sources 2019, 421, 1– 5; c) J. Friedl, M. A. Lebedeva, K. Porfyrakis, U. Stimming, T. W. Chamberlain, J. Am. Chem. Soc. 2018, 140, 401– 405. 47. R. R. Gagne, C. A. Koval, G. C. Lisensky, Inorg. Chem. 1980, 19, 2854– 2855. 48. D. W. Leedy, D. L. Muck, J. Am. Chem. Soc. 1971, 93, 4264. 49. C. J. McAdam, B. H. Robinson, J. Simpson, Organometallics 2000, 19, 3644– 3653. 50. G. Natta, P. Corradini, Il Nuovo Cimento (1955–1965) 1960, 15, 40– 51. 51. H. S. Kim, T. Yoon, Y. Kim, S. Hwang, J. H. Ryu, S. M. Oh, Electrochem. Commun. 2016, 69, 72– 75. 52. Y. Zhen, C. Zhang, J. Yuan, Y. Zhao, Y. Li, J. Power Sources 2020, 480, 229132. 53. P. Zhao, H. M. Zhang, H. T. Zhou, J. Chen, S. J. Gao, B. L. Yi, J. Power Sources 2006, 162, 1416– 1420. 54. C. Zhang, L. Zhang, G. Yu, Acc. Chem. Res. 2020, 7454– 7459. 55. C. K. Zhang, L. Y. Zhang, Y. Ding, X. L. Guo, G. H. Yu, ACS Energy Lett. 2018, 3, 2875– 2883. 56. C. Zhang, Y. Qian, Y. Ding, L. Y. Zhang, X. Guo, Y. Zhao, G. Yu, Angew. Chem. Int. Ed. 2019, 58, 7045– 7050. 57. C. K. Zhang, Z. H. Niu, Y. Ding, L. Y. Zhang, Y. G. Zhou, X. L. Guo, X. H. Zhang, Y. Zhao, G. H. Yu, Chem 2018, 4, 2814– 2825. 58. a) T. Hagemann, J. Winsberg, A. Wild, U. S. Schubert, Electrochim. Acta 2017, 228, 494– 502; b) C. G. Armstrong, R. W. Hogue, K. E. Toghill, J. Power Sources 2019, 440, 227037. 59. a) W. T. Duan, J. H. Huang, J. A. Kowalski, I. A. Shkrob, M. Vijayakumar, E. Walter, B. F. Pan, Z. Yang, J. D. Milshtein, B. Li, C. Liao, Z. C. Zhang, W. Wang, J. Liu, J. S. Moore, F. R. Brushett, L. Zhang, X. L. Wei, ACS Energy Lett. 2017, 2, 1156– 1161; b) X. L. Wei, W. T. Duan, J. H. Huang, L. Zhang, B. 43 Li, D. Reed, W. Xu, V. Sprenkle, W. Wang, ACS Energy Lett. 2016, 1, 705– 711; c) J. H. Huang, Z. Yang, M. Vijayakumar, W. T. Duan, A. Hollas, B. F. Pan, W. Wang, X. L. Wei, L. Zhang, Adv. Sustainable Syst. 2018, 2, 1700131. 60. Y. Ding, C. Zhang, L. Zhang, Y. Zhou, G. Yu, Chem. Soc. Rev. 2018, 47, 69– 103. 61. a) X. Q. Xing, Q. H. Liu, W. Q. Xu, W. B. Liang, J. Q. Liu, B. G. Wang, J. P. Lemmon, ACS Appl. Energ. Mater. 2019, 2, 2364– 2369; b) K. Takechi, Y. Kato, Y. Hase, Adv. Mater. 2015, 27, 2501– 2506. 44 Name of Candidate: Julia Case Birth date: June 16, 1999 Birth place: Silver Spring, MD Address: 71 S Elizabeth Street Apt #1 Salt Lake City, UT 84102 45