Cation design in complex aqueous electrolytes for low-temperature zinc-bromine flow batteries Authors: Tao Cheng, Ming Zhao, Tianyu Li, Shuo Wang, Yanbin Yin, Xianfeng Li Published: 2026 ABSTRACT AbstractThe cation design of the multi-component dynamic concentration electrolytes that simultaneously achieve effective phase transition prevention and rapid kinetic process is challenging but vital for enhancing the performance of low-temperature zinc-bromine flow batteries. Herein, we employ NH4+ as supporting electrolyte cation to break inherent trade-off between salting-out at low temperatures and ion transport kinetic from the alkali cations (K+, Na+ and Li+). Compared to commonly used K+ in the traditional electrolyte, NH4+ achieves an improved ability to prevent the phase transition of the electrolyte and obtains better ion transport property, which is a characteristic that Na+ and Li+ do not possess. Additionally, NH4+ can significantly enhance the cycling stability by appropriately increasing the solubility of polybromides. Consequently, the zinc-bromine flow batteries supported by NH4+ not only demonstrate sustained lifespan (>2,300 cycles at 40 mA cm−2, 40 mAh cm−2, over half a year) and consistent high-rate cyclic stability (charging at 200 mA cm−2, discharging at 80 mA cm−2, >3,300 h) at room temperature; they also exhibit stable cyclic stability (over 1,600 cycles at 40 mA cm−2) at −20 °C. This work provides an effective path for the design of complex electrolytes for low-temperature aqueous zinc-based flow batteries.Subject terms: Batteries, Energy Introduction To address the escalating energy demands and mitigate environmental impacts stemming from fossil fuel reliance, renewable sources like solar and wind power have to increase their share within the electricity generation sector1. Nonetheless, the intermittent and unpredictable nature of these renewable energies poses significant challenges to their seamless integration into the grid. Consequently, large-scale energy storage systems emerge as crucial enablers to overcome these intermittency issues2. The aqueous zinc-bromine flow battery (ZBFB) stands out as a highly promising contender for the next-generation energy storage solutions due to its high theoretical energy density, cost-effectiveness, and intrinsic safety features3,4. However, ZBFBs face some challenges; their limited performance at low temperatures in particular hinders their application5,6. In aqueous ZBFBs, the electrolyte acts as an indispensable medium for storing active materials, transporting ions, and enabling electrochemical reactions3,7. It also plays a crucial role in ensuring the reliable operation of the battery at low temperatures. Generally, the electrolyte (the total molarity >5 mol L−1) consists of aqueous solvent, zinc bromide (ZnBr2), supporting electrolyte (SE) and bromine complexing agent6,8. During the charging process, bromide ions (Br−) are oxidized and complex with the complexing agent at the positive electrode, forming a denser, oil-like ionic liquid that accumulates at the bottom of the storage tank3,8. Simultaneously, zinc ions (Zn2+) are reduced to metallic zinc (Zn) deposits on the negative electrode. With the increase of charging depth, the concentration of ZnBr2 in the electrolyte decreases continuously9. The discharging process is the reverse of these reactions (Fig. 1a). Foreseeably, stable operation of ZBFBs at low temperatures encounters multiple difficulties: i) solidification of polybromide complexes at the positive electrode6; ii) phase transition of bulk electrolyte at both initial and charged states5,10; and iii) sluggish ion transport kinetic process in electrolyte5,11. Therefore, with respect to the composition of the electrolyte and the operational characteristics, designing a low-temperature electrolyte for ZBFBs is highly challenging. Fortunately, optimizing the combination and structure of complexing agents can effectively enhance the liquid-phase retention ability of polybromide complexes and then improve the battery’s performance at low temperatures6,12. However, the other issues still exist and are becoming increasingly prominent. Notably, the salts as SEs (typically the highest molarity) not only significantly affect ion conductivity, but also dominate the phase transition of the bulk electrolyte at low temperatures due to the complex interaction between anions, cations and dipolar water molecules5. As the concentration of ZnBr2 decreases during the charging process, the role of the SE becomes more significant. The introduction of various salts into aqueous electrolytes is an effective and simple strategy that significantly breaks through the low-temperature operating window and provides valuable insights for the design of low-temperature electrolytes13,14. However, research about SE salts related to the electrolyte systems with continuously changing concentrations of Zn2+ is rarely reported. Significantly, in the actual operation of battery systems, phase-transition prevention capability is not the only goal. The permissible low-temperature limit of the electrolyte and ion transport kinetics needs to be improved simultaneously, which is the relatively ideal target we should strive for11,15. Furthermore, for some common salts with good aqueous solubility, composed of cations (e.g., NH4+, K+, Na+, Li+) and anions (e.g., Cl–, Br–, ClO4–, SO42–), the hydration extent of the cations is significantly greater than that of the anions due to their much smaller ionic radii compared to those of the anions16,17. Therefore, cation design of the complex aqueous electrolytes is crucial and at the core of improving their properties. Design of supporting electrolyte cations for low-temperature ZBFBs. a Schematic diagram of ZBFBs. Fluctuations in ZnBr2 concentration elevate the risk of electrolyte phase transitions during the operation of low-temperature ZBFBs. b Schematic illustration of the role of NH4+ in the electrolyte. Due to the formation of HBs between water and NH4+, the phase-transition temperature of the electrolyte is reduced; the rapid breaking and reforming of spontaneously rotating tetrahedral NH4+–H2O HB loosen the H2O–H2O HB network and their solvation structure, enabling ZBFB electrolytes with NH4+ to achieve both low phase-transition temperatures and fast ion transport. c Schematic illustration of the role of alkali metal ions in the electrolyte. The trade-off is that smaller alkali metal ions process stronger M+–Ow interactions, which lower the phase-transition temperature, while restricting ion transport due to the tight H2O–H2O HB network and rigid ionic solvation structure. Herein, in the complex traditional electrolyte of ZBFBs, the pivotal factor at low temperatures was identified: the salting out of the SE, which triggers the initial phase transition. We employed NH4+ to alleviate this issue, and to break the trade-off between low-temperature phase-transition prevention and ion transport kinetics associated with some alkali metal ions (K+, Na+ and Li+, Fig. 1b, c). Among these cations, the NH4+ endows the electrolyte with the best ion transport property; although the capability of NH4+ to prevent low-temperature phase transition of the electrolyte is not as strong as that of Na+ and Li+, it is significantly better than that of K+ (commonly used in conventional electrolytes). NH4+ can help to loosen the hydrogen-bond (HB) structure between water molecules, thus promoting the rapid diffusion of Zn2+ and Br– in the electrolyte. Furthermore, using NH4+ can appropriately increase the concentration of polybromides in aqueous phase of the electrolyte, thereby mitigating the Zn growth into the membrane. Consequently, NH4+ not only bestows ZBFBs with long cycling and rapid charge-discharge capabilities at room temperature, but also sustains the long-term and high-current operation of ZBFBs at –20 °C. Results Electrolyte phase transition and solvation structure The low-temperature-induced phase transition of the electrolyte could result in flow channel blockage, seriously endangering the normal operation of ZBFBs. In the usual aqueous multisolute (ZnBr2, KCl, bromine complexing agent) electrolyte18, it is essential to find out the key factor of the phase transition at low temperatures. We first examined which substance would preferentially transform into a solid at low temperatures in the H2O-multisolute system. Considering that the phase transition temperature increases with decreasing ZnBr2 concentration (Supplementary Fig. 1), we chose a lower concentration of ZnBr2 as the research object in the presence of SE to simulate the electrolyte with 50% utilization rate (UR, defined as the ratio of the reduced Zn2+ concentration to the pristine Zn2+ concentration during the charging process). From the digital images of the electrolytes freezing at –10 °C (Fig. 2a), it can be found that the white substance forms in the electrolyte with simulated 50% UR, which was identified as primarily KCl by X-ray diffraction (XRD, Fig. 2a) and X-ray fluorescence (XRF, Supplementary Fig. 2d), while no significant changes were observed in the pristine electrolyte. This phenomenon reveals that the key to effectively inhibiting the phase transition of the low-temperature electrolyte lies in preventing the SE precipitation. Although reducing the molarity of KCl will prevent the solution from salting out at this temperature (Supplementary Fig. 2a), it undergoes phase transition at –20 °C, regardless of whether it is in a simulated partially charged state or a pristine state (Supplementary Fig. 2b, c), and it will decrease the ionic conductivity (Supplementary Fig. 3a). Regarding this issue, adjusting the types of anions and cations appears to be an effective and profoundly impactful strategy. For some common salts with good aqueous solubility, composed of cations (e.g., NH4+, K+, Na+, Li+) and anions (e.g., Cl–, Br–, ClO4–, SO42–), the hydration extent of the cations is much greater than that of anions because the ionic radius of the cations is much smaller than that of anions16. The hydration capability of cations primarily depends on their ionic radius and charge; the smaller the radius and the higher the charge, the stronger the interaction with polar water molecules19. Therefore, we chose to study the SE cations as our main focus. Phase transition and solvation structure of the electrolyte with different SE cations. a XRD pattern of phase change product (inset digital image) from simulated 50% UR negolyte after 24 h at −10 °C, indicating that KCl preferentially precipitates during the low temperature operation of ZBFBs. b Phase transition temperature of the electrolytes from visualized experiment (RTSO: room temperature salting out). Negolytes supported by NH4Cl/NaCl/LiCl exhibit a lower phase-transition temperature compared to negolytes with KCl. c, d Solvation structure from MD, c coordination numbers (CNs) for the first solvation layer of the SE cations with water (M+···Ow), chloride (M+···Cl−), and bromide ions (M+···Br−). d The CN curves of M+···Ow. NH4+ and K+ exhibit solvation structures loosely, whereas Na+ and particularly Li+ form comparatively tighter ones. As a low-temperature SE cation with good performance, it is necessary to ensure that it does not precipitate within a specific temperature range while achieving a high ionic conductivity. The guideline for cation design that has a greater ability to hydrate to resist crystallization with anions, compared with K+, while maintaining a loose hydrated structure to maintain high ionic mobility, should be feasible. NH4+ is a tetrahedral structure derived from the covalent bonds between the nitrogen atom and four hydrogen atoms20. Its ionic radius (1.48 Å) is larger than that of K+ (1.38 Å), and the hydration radii are similar (3.31 Å)21. In addition, the mobility (7.60 × 10−8 m2 s−1 V−1) and diffusivity (1.96 × 10−9 m2 s−1) of NH4+ are similar to those of K+ (7.62 × 10−8 m2 s−1 V−1, 1.96 × 10−9 m2 s−1, respectively) at infinite dilution in aqueous solution (25 °C)22. Considering the fast anisotropic rotation of NH4+ and continual forming and breaking of hydrogen bonds with water23–25, its function in suppressing electrolyte phase transitions and enhancing ion transport kinetics should be worth exploring in the complex electrolyte system of ZBFBs. To comprehensively investigate the mechanism, the common alkali metal ions (K+, Na+ and Li+) are used for comparison. The design concept is illustrated in Fig. 1. To evaluate the phase transition inhibition capability of these monovalent cations, visualized experiment of low-temperature phase transition was employed to examine the electrolyte with varying concentration of their chlorides, both at the pristine and simulated 50% UR state (Supplementary Fig. 4). The two-dimensional temperature maps display that the phase-transition temperatures of the electrolytes using NH4Cl, NaCl and LiCl are almost all lower than that of KCl at the specific molarity (Fig. 2b); in terms of these alkali metal ions, the smaller the ion radius, the lower this temperature. The phase transition temperature of the electrolytes with NH4+ is between those of K+ and Na+atthe pristine state. Differential scanning calorimetry (DSC) was further used to study these electrolytes. We chose the inflection points of DSC curves during the heating process, where they completely transition to the liquid phase, as the evaluation standard (Supplementary Figs. 5 and 6). The regularity shown by DSC results is basically consistent with that of the visualization experiment (Supplementary Fig. 7). It can be considered that the phase transition prevention capability of NH4+ is between K+ and Na+. It can also be found that, except for the KCl SE, the other three chlorides can maintain the electrolyte with 4 M SE in a liquid state at −20 °C, which indicates their application potential (Supplementary Fig. 4). To investigate the interaction of these cations with water and anions, molecular dynamic (MD) simulation was carried out to analyze the coordination structure of these monovalent cations (Supplementary Fig. 8). As shown in Fig. 2c, Supplementary Figs. 9 and 10, in the coordination structure of these cations, there are not only water, but also Cl– and Br– at both 298.15 K and 253.15 K, which is the result of the competition between cation-water and cation-anion interactions in this complex moderately concentrated electrolyte. The coordination structure results also indicate that the data obtained from dilute solutions can serve as an important reference, but they cannot be directly used to make judgments in this specific system. From the coordination number (CN) curves of cations with water in this complex electrolyte at both 298.15 K and 253.15 K (Fig. 2d), one can observe that the curves of NH4+ exhibit a steep slope, resembling that of K+, while the Na+ curves have a gentler slope, and the Li+ curves feature a distinct plateau. These results suggest that NH4+ possesses a looser water coordination structure than those of Na+ and Li+, which is similar to that of K+ in the electrolyte. Upon comparing the phase transition behaviors of the electrolytes with NH4+ and K+, we found that: in the pristine electrolyte, this difference primarily manifests as SE precipitation; whereas in a 50% UR electrolyte, it involves both SE precipitation and solidification of the bulk electrolyte (Supplementary Fig. 4). These phenomena are associated with cation–water interaction strength and hydrogen bond networks. Regarding cation–water interactions, NH4+ exhibits stronger binding than K+ at the minimum potential energy surface (Supplementary Fig. 11a), which should account for the higher solubility of NH4Cl at low temperatures, compared with that of KCl. Based on the structural characteristics of NH4+, this slightly stronger bonding energy is likely attributed to hydrogen bonds formed between the NH4+ and surrounding water molecules (Supplementary Fig. 11b–d). In addition, partial charges on H2O generate an ordered hydrogen bond network, raising the phase transition temperature compared with van der Waals governed chalcogen hydrides13 (Supplementary Fig. 11e). Unlike K+, NH4+ forms NH4+···H2O bonds, competing for H2O···H2O sites. In the 50% UR electrolyte, the incorporation of NH4+ markedly decreases the number of hydrogen bonds between water molecules in the bulk electrolyte (Supplementary Fig. 11f). Therefore, compared to K+, NH4+ exhibits a lower phase transition temperature in both pristine and 50% UR electrolytes. Transport kinetics of electrolyte with monovalent cations For asymmetric NH4+, the angle varies dynamically due to its random rotational motion (Supplementary Movie 1). N–H···O hydrogen bonds are continuously, rapidly disrupted and re-established. The O···N–H angle (θ) is shown in Fig. 3a. Within the range of 0° to 180°, the distance of N-H···O increases as the angle becomes larger. Therefore, within an approximate range of 0–60°, the strength of hydrogen bonds decreases rapidly with θ to reach relatively low levels (Fig. 3b). The curve of binding energy of NH4+···H2O vs. θ further confirms this trend (Fig. 3c). These results show that the shorter of N-H···O bond lengths compared to those of other metal ions (Li+, Na+ and K+) play a transient role in the interactions between ions and water. The influence of SE cations on the transport kinetics of the electrolytes. a Definition of the O···N–H angle for NH4+···H2O; the studied H atoms are shown in pink. b Studied N–H···O HB strength versus θ from DFT. c NH4+···H2O / M+···H2O binding energy versus θ from DFT. d The τ of H2O···H2O HBs and NH4+···H2O HBs. A shorter τ of H2O···H2O indicates a looser water HB network, potentially accelerating ion transport. The τ (H2O···H2O HB) in various electrolytes with different SEs: NH4Cl < KCl KCl > NaCl > LiCl. Electrochemical performance of electrolyte with different SEs. g Zn2+ and h Br2 diffusion coefficients from variation of the peak current from CV curves. i Average ITNs of SE cations. NH4+ and K+ exhibit significantly higher ion diffusion performance compared with Na+ and Li+. To reveal the effect of these cations on the ion transport, the relaxation time (τ) of water-water HBs was also investigated. A longer τ of HBs suggests a tighter HB network, thereby inhibiting ion transport. As exhibited in Fig. 3d and Supplementary Fig. 12, the τs of HBs between water molecules in the electrolyte using NH4+ are the shortest among these monovalent cations at both 25 °C (298.15 K) and –20 °C (253.15 K), indicating the HB network in this electrolyte is the loosest. The τs of NH4+···H2O HBs are also displayed in Fig. 3d. Indeed, the HBs of NH4+···H2O can form, and their τs are shorter compared with those of H2O···H2O in the NH4Cl-electrolyte. This characteristic of NH4+-H2O HBs should be beneficial for weakening the interaction of NH4+ with water, reducing the HB strength between water molecules, and then promoting the rapid diffusion of NH4+. Based on the analysis of the structures of these cations and their interactions with water, we can reasonably infer that NH4+ may act as a stirrer in the electrolyte due to the tetrahedral HBs (Supplementary Fig. 13b). We further employed nuclear magnetic resonance (NMR) spectroscopy to study the impacts of various SE cations on the entire population of water molecules. The half-peak width of the 17O-NMR can be used as an index to describe the tightness of water–water structure; the narrower half-peak width corresponds to rapidly tumbling water molecules and loose water-water structure26, implying enhanced ion mobility in the electrolyte. The electrolyte with NH4+ demonstrates the narrowest half-peak width (522 Hz, Fig. 3e, Supplementary Fig. 14), illustrating the faster ion movement in this electrolyte than those in the electrolytes with K+, Na+ and Li+. The kinematic viscosity of these electrolytes exhibits a trend similar to that of the half-peak width of the 17O-NMR, while conductivity data values demonstrate a negative correlation (Fig. 3e), which also indicate that the electrolyte with NH4+ pocesses the fastest ion transport environment. According to the Stokes–Einstein–Sutherland equation27, a reduction in viscosity necessarily leads to an increase in the diffusion coefficient. Therefore, in the NH4Cl-supported electrolyte, the ion mobility should be enhanced. To further study the influence of these monovalent cations on ion transport kinetics, the self-diffusion coefficients of the various ions were obtained from mean squared displacement (MSD) calculated using MD at 298.15 K and 253.15 K (Supplementary Fig. 15). Among the alkali metal chlorides, in the KCl-electrolyte, the self-diffusion coefficients of Cl–, Br–, Zn2+ and K+ are almost the maximum and those in the LiCl-electrolyte are the smallest (Fig. 3f), indicating the ion diffusion kinetics slow down as the ionic radius decreases due to the increasingly tight coordination structure and the network of water molecules. Interestingly, among these electrolytes with various monovalent cations, the self-diffusion coefficients of Cl–, Br–, Zn2+ and NH4+ are the greatest in the electrolyte with NH4+. To further verify the characteristics of ion transport, cyclic voltammetry (CV) curves at a series of scan rates in the solutions with the different monovalent cation chlorides were tested (Supplementary Figs. 16–18). The peak currents associated with the reduction of Zn2+-Zn and the reduction of Br2-Br– are linearly related to the square root of the scan rates (Fig. 3g, h), and the diffusion coefficient (D) can be calculated according to the Randles-Sevick equation28,29. The order of D(Zn2+) values is as follows: NH4Cl-solution (~4.42 × 10−7 cm2s−1) > KCl-solution (~3.28 × 10−7 cm2 s−1) > NaCl-solution (~2.57 × 10−7 cm2 s−1) > LiCl-solution (~2.00 × 10−7 cm2 s−1), which is consistent with the order of self-diffusion coefficients. Meanwhile, the D(Br2) values show a similar order: NH4Cl-solution (~1.90 × 10−6 cm2 s−1) > KCl-solution (~1.59 × 10−6 cm2 s−1) > NaCl-solution (~1.39 × 10−6 cm2 s−1) > LiCl-solution (~1.34 × 10−6 cm2 s−1). In addition, in both chloride and bromide solutions, among these monovalent ions, the ammonium ion has the largest ion transport numbers (ITNs, Fig. 3i, Supplementary Fig. 19), indicating that the electrolyte with NH4+ would have the lowest ohmic polarization during ZBFB operation. Electrochemical properties and zinc deposition morphologies of NH4Cl- and KCl-electrolytes Based on the above analysis, the electrolyte with NH4+ exhibits better low-temperature properties than that with K+ (most widely used in ZBFBs) and has the best ion transport kinetics compared with these alkali metal ions. Considering both room- and low-temperature properties, we believe that NH4+ is a good low-temperature electrolyte cation with comprehensive performance. Therefore, to further illustrate its better effects, the following will focus on a comparative study between NH4+ and K+. To further compare the performance of NH4+ and K+, we conducted a series of electrochemical tests within the temperature range of −10 to 30 °C. As previously mentioned, the electrolyte with 3 mol L−1 (M) KCl exhibits better phase transition resistance at low temperatures compared with that with 4 M (Fig. 2a and Supplementary Fig. 2a). Additionally, in terms of the electrolyte using KCl, 3 M is the molarity typically used in ZBFBs. 4 M NH4Cl demonstrates better resistance to phase changes compared to the 5 M NH4Cl (Fig. 2b); however, reducing the NH4Cl concentration lowers its conductivity (Supplementary Fig. 3), precluding consideration of a 3 M NH4Cl. Further comparative analysis between optimal concentrations of KCl and NH4Cl will determine the electrolyte with the achievable performance. Therefore, the electrolyte with 3 M KCl serves as the control group. The D(Zn2+) values obtained from CV curves at various temperatures were employed for further analysis (Supplementary Fig. 20). The D(Zn2+) values in the NH4Cl-electrolyte are larger than those in KCl-electrolyte in this temperature range (Fig. 4a), indicating that the NH4Cl-electrolyte at this molarity has faster Zn2+ transport properties and holds promise for better performances of ZBFBs at low temperatures. The exchange current, i0, for the Zn2+/Zn0 redox reaction can be derived from Tafel plots using the Butler-Volmer equation (Supplementary Fig. 21)28,30. The i0 can be used for assessing the electron-exchange kinetics between Zn2+ and Zn0. In the temperature range of −10 to 30 °C, all the i0 values are larger in NH4Cl-electrolyte, compared with those in KCl-electrolyte (Fig. 4b). In addition, there is a linear relationship between i0 and e1/T. The ratio of the electron transfer coefficient, k0, of Zn2+-Zn0 in NH4Cl electrolyte and KCl electrolyte can be calculated from the i0-e1/T curve at different temperatures (Fig. 4b). Obviously, the k0 (NH4Cl- electrolyte)/k0 (KCl-electrolyte) of Zn2+-Zn0 consistently exceeds 1.3 within the temperature range (Fig. 4c). The values of i0 and the ratios of electron transfer coefficients indicate that the electron exchange kinetics for Zn2+-Zn0 are faster in NH4Cl-electrolyte than in KCl-electrolyte. Characterization of electrochemical properties and zinc deposition morphologies of NH4Cl- and KCl-electrolytes. a Zn2+ diffusion coefficient in the temperature range from −10 °C to 30 °C. b, c Analysis of Tafel plot at various temperatures. b Exchange current. c Ratio of electron transfer coefficients. NH4+ supported electrolyte exhibits faster Zn2+ diffusion and reaction kinetics compared to the electrolyte with K+ from −10 °C to 30 °C. d, e The combined SEM and EDS image of the cross-sectional view of the interface between the zinc-deposited anode (80 mAh cm−2) and the membrane in (d) KCl-electrolyte and (e) NH4Cl-electrolyte. A clear gap exists between the membrane and Zn deposits in the NH4Cl-electrolyte, whereas Zn deposits in the KCl-electrolyte grow invasively into the membrane. f Schematic diagram indicates that membrane-invading zinc deposits obstruct the membrane, hinder ionic conduction across the membrane, and risk causing short circuits. Preventing zinc deposition within the membrane is essential for uninterrupted ion transport. As the battery reaction proceeds, “dead zinc” forms within the porous separator, severely compromising the cyclic performance of the battery. Consequently, inhibiting zinc penetration into the membrane is crucial31,32. To further reveal the advantage of NH4+, the Zn depositing behavior was investigated in the two types of electrolytes. Scanning electron microscopy (SEM) and energy dispersive spectrometry (EDS) were used to observe the cross-section at the interface between the carbon felt and membrane after the initial charging 80 mAh cm−2 (Supplementary Figs. 22–25). The sample preparation process and the morphological details are shown in Supplementary Figs. 26–28. In the KCl-electrolyte, SEM-EDS stack image in Fig. 4d displays that Zn deposits (marked by yellow) extend into the membrane (marked by green silicon staining), forming a composite region approximately 10 μm wide (highlighted in red). The accumulation of Zn in the membrane would increase ion transport resistance and easily lead to battery short circuits (Fig. 4f). As a stark contrast, Zn do not grow into the membrane in the NH4Cl-electrolyte (Fig. 4e), suggesting that the ZBFBs with NH4Cl electrolyte would possess better cyclic stability. Both NH4+ and K+ exhibit adsorption on the zinc anode (Supplementary Fig. 29), thereby flattening the zinc (002) crystal plane33,34. However, under high surface capacities (80 mAh cm−2), additional factors influencing zinc deposition at the porous membrane-zinc interface must also be considered. To explore the reason behind the difference in zinc deposition behavior, we used the iodometric method combined with ultraviolet-visible spectroscopy (UV-Vis) to quantify the effect of NH4+ and K+ on the concentration of oxidized bromine species in the simulated end-of-charge aqueous phase of the electrolyte. The Br2 concentration in the supernatant of the NH4Cl solution (~364 mM) exceeds that in the KCl solution (~248 mM) (Supplementary Fig. 30a–c), which should be derived from the higher binding energy of NH4+ to Br3− than that of K+ calculated by DFT (Supplementary Fig. 30d). The higher concentrations of oxidized bromine species in the aqueous phase and ionic liquid of the NH4Cl posolyte would appropriately increase their penetration into the porous membrane (Supplementary Fig. 30e), helping eliminate zinc dendrites and the accumulation of zinc within the porous membrane (Supplementary Fig. 30f). Although this leads to a reduction in coulombic efficiency of ZBFBs, the quantitative data indicate that the modestly higher self-discharge in the NH4Cl electrolyte removes Zn dendrites, trading an increased 10% CE loss for ~0.8 times service life extension and a higher average energy efficiency over the entire cycling test (Supplementary Fig. 30g–l). In addition, to confirm that the electrolyte’s pH does not enter the ammonia evolution region (pH >6)35, pH measurement was conducted across a range of ZnBr2 concentrations (0.05–2 M). The results indicate that the pH (2.42–3.89) remains out of the evolution range for NH4Cl concentrations of 3 M, 4 M, and 5 M (Supplementary Fig. 31a). Furthermore, we also investigated the effect of pH variations on electrochemical hydrogen evolution. In solution with zinc ions, the reduction current for zinc peaks before −1.5 V, which is significantly greater than the hydrogen evolution potential in the solutions without zinc ions (Supplementary Fig. 31b–e). In two Zn2+-free electrolytes with different pH values, the current values are below 0.01 mA from 0 to −1.5 V. These results indicate that such pH variations do not induce significant hydrogen evolution during zinc deposition. Electrochemical performances of ZBFBs at RT Combining all phase transition temperatures, MD simulations, DFT calculations, electrochemical property test and zinc deposition morphology results, NH4+ has been identified as a better candidate than K+ for SE cation in ZBFBs due to the better diffusion performance and antifreeze capability. Compared to the 4 M KCl-electrolyte, the 3 M KCl-electrolyte not only exhibits less phase change (Fig. 2a and Supplementary Fig. 2a) but also demonstrates better cycling performance (Supplementary Fig. 32). Therefore, in the characterization of electrochemical properties, the 3 M KCl electrolyte is also primarily used as the benchmark. We used 4 M NH4Cl and 3 M KCl supported electrolyte individually to assemble ZBFBs (Supplementary Figs. 33 and 34) with a traditional complexing agent (1-ethyl-1-methylpyrrolidinium bromide, MEP) at room temperature to verify the above conclusions. As expected, the ZBFB using 4 M NH4Cl-electrolyte displays a longer cycle life (above 70% energy efficiency (EE) for a duration of over 1200 h) at a high capacity (80 mAh cm−2) and high depth of charge (posolyte: average Br2 concentration ~1.35 M), while the EE of the battery using 3 M KCl-electrolyte drops below 70% after 230 h (Supplementary Figs. 33 and 34). By changing the NH4Cl concentration to 5 M, the ZBFB can achieve stable cycling performance of over 4600 h at 40 mA cm−2 and 40 mAh cm−2 (over 2390 cycles, more than half a year), while the ZBFB supported by KCl exhibited a fast decay after 750 h (Fig. 5a, Supplementary Fig. 35). Galvanostatic voltage profiles were introduced and exhibit the lower polarization of the NH4Cl electrolyte during battery operation (Fig. 5b). The midpoint voltage difference of the 1st, 100th, 200th, and 300th charge-discharge curves of ZBFB using the NH4Cl electrolyte are 0.20 V, 0.24 V, 0.24 V, and 0.24 V, respectively, which are lower than those (0.26, 0.40, 0.42, and 0.43 V) of ZBFB using KCl-electrolyte. ZBFBs employing NaCl or LiCl electrolytes became short-circuited within 50 cycles. Considering the solidification risk of oily phase with MEP at low temperatures6, choline chloride (ChCl), free from such risk, was employed as the bromine complexing agent in subsequent electrochemical studies on room- and low-temperature ZBFBs. We further evaluate the rate performance of the ZBFB from 40 mA cm−2 to 240 mA cm−2. As shown in Fig. 5c and Supplementary Figs. 36 and 37, when evaluating the top-performing monovalent cations, the ZBFB supported by NH4Cl demonstrates higher voltage efficiencies than those by KCl. Even at a relatively high current density of 240 mA cm−2, the average voltage efficiency of ZBFB with NH4Cl-electrolyte is ~57.9%, higher than that (~55.0%) with KCl-electrolyte (Supplementary Table 13). The voltage efficiency of ZBFBs assembled with NaCl and LiCl as SE (ZBFB with Na+ or Li+) is far inferior to that with NH4Cl and KCl (Supplementary Fig. 36d, Supplementary Table 13), which is consistent with the results of MD simulation and electrochemical test. Even when the concentration of the SE salt is the same (4 M), NH4Cl-electrolyte still exhibits the best rate performance among these electrolytes with different SEs (Supplementary Fig. 37, Supplementary Table 14). Moreover, given the high rate performance of the ZBFBs with NH4Cl electrolyte, we further evaluated the cycling performance of ZBFBs at a high charge current (200 mA cm−2) and a high discharge current (80 mA cm−2) (Fig. 5d, Supplementary Figs. 38 and 39). ZBFBs with NaCl or LiCl as the SE exhibited large voltage polarization and short-circuit failure within 50 h of operation. The KCl-supported ZBFB decayed and failed rapidly after 1500 hours of operation, with a cumulative electrolyte life of ~52.03 kWh L−1 (1960 cycles). In stark contrast, the NH4Cl-supported ZBFB showed stable operation with an energy density of more than 25.4 Wh L−1 for more than 3480 hours, and a cumulative electrolyte life of more than ~111.43 kWh L−1 (4200 cycles). These confirm that NH4+ is an ideal choice for long-life and high-power ZBFBs at room temperature. Electrochemical performance comparison of ZBFBs with monovalent chloride at room temperature. a Cycling stability of ZBFBs at 40 mAh cm−2, and corresponding voltage profiles with b NH4Cl, KCl, NaCl, LiCl as SE. ZBFBs with NH4Cl as SE exhibit continuous stable performance for durations longer than half a year, compared to <1000 h with KCl. NaCl and LiCl-supported ZBFBs developed short circuits and failed within 50 cycles. c Rate performance of ZBFBs supported by different SEs using ChCl. NH4Cl outperforms KCl in voltage efficiency at high current densities in ZBFBs as SE. CE loss occurred at 160 mA cm−2 for NaCl-supported ZBFB and at 120 mA cm−2 for LiCl. d Cycling stability of ZBFBs supported by different SEs using 1 M ChCl at high current density (charge at 200 mA cm−2, discharge at 80 mA cm−2). NH4Cl supported ZBFB operated at ~25.4 Wh L−1 for more than 3480 hours (~111.43 kWh L−1), while KCl supported ZBFB operated <1500 hours (~52.03 kWh L−1). ZBFBs employing NaCl or LiCl short-circuited rapidly and failed at such high current. Electrochemical performances of ZBFBs at −20 °C Encouraged by the phase transition prevention ability and favorable kinetic properties of the NH4+ electrolyte, as well as the high room temperature battery performance, ZBFBs supported with NH4Cl were tested at −20 °C. Other monovalent cation electrolytes were also introduced for comparison. As shown in Fig. 6a, the ZBFB with NH4Cl-electrolyte exhibits better cycling stability than others at 40 mA cm−2. The ZBFB with KCl electrolyte fails within 20 cycles due to the high phase transition temperature of the electrolyte. In addition, the midpoint voltage differences of the first charge-discharge curves are ~0.98 V and ~1.18 V for NaCl- and LiCl-electrolytes, respectively, while, in terms of NH4Cl-electrolyte, it is as low as ~0.74 V (Supplementary Figs. 40 and 41). Considering the cost of the SEs, ion transport numbers, the diffusion rates of M+/Zn2+/Br− from MD in corresponding electrolytes, the resistance to phase transition, and the battery’s average efficiency at both room and low temperatures, it can be found that NH4+ is the most suitable cation for ZBFB operations at −20 °C and room temperature (Fig. 6b). We further evaluated the long-cycle stability of the ZBFB using NH4Cl-electrolyte at −20 °C (Fig. 6c). The ZBFB supported by NH4Cl demonstrates a remarkably stable columbic efficiency, average ~96.7%, for over 1200 hours (more than 1600 cycles) at 40 mA cm−2. In the first 1000 cycles of battery operation, the ZBFB exhibits a slightly increased polarization (Fig. 6d). The midpoint voltage differences for the 1st, 100th, 500th, and 1000th charge-discharge cycles are ~0.71, ~0.75, ~0.77, and ~0.85 V, respectively. In addition, the ZBFB using NH4Cl-electrolyte has high rate performance from 20 mA cm−2 to 120 mA cm−2 at −20 °C (Fig. 6e). The average energy efficiency at 20 mA cm−2 is ~75.2%; even at a relatively high current density of 80 mA cm−2, the average energy efficiency is ~51.4%. Furthermore, the NH4Cl-electrolyte endows ZBFB with output power density of ~160.7 mW cm−2 at 200 mA cm−2 (Fig. 6f). The stable cycling performance and high-rate performance of the ZBFBs support the use of NH4+ as the SE cation, particularly at −20 °C, which used to be challenging and inaccessible for conventional ZBFBs. To further evaluate the performances of the ZBFBs with NH4Cl, we executed an extensive comparison with the recent reported studies on low-temperature flow batteries. As shown in Fig. 6g, h, the performances of ZBFBs with NH4Cl are competitive at both room and low temperatures, with respect to current density, energy density and runtime6,7,36–49. Electrochemical performance of ZBFBs at −20 °C. a Cycling stability comparison with different SEs at 40 mA cm−2. Only NH4Cl-supported ZBFBs achieve stable operation at 40 mA cm⁻² and −20 °C with energy efficiency exceeding 60%. b Comprehensive comparison of these chlorides as SEs in ZBFBs. Considering cost-effectiveness, ion transport rate, resistance to phase transition, battery efficiency at room temperature, and time to failure at low temperature, NH4Cl represents a competitive SE candidate for all-climate ZBFBs. c, d NH4Cl supported ZBFB performances at −20 °C. c Cycling stability at 40 mA cm−2 and d corresponding voltage profiles. e Rate performance from 20 to 120 mA cm−2. f Discharge polarization curves. The NH4Cl-supported ZBFB achieves a lifespan of over 1200 hours at 40 mA cm−2, maintains stability at 120 mA cm−2, and reaches a peak power density of ~160.7 mW cm−2 at −20 °C. g zinc-bromine flow battery performance comparison at room temperature. This work achieves extended cycle life in terms of high currents, enhanced surface capacity, and increased volumetric capacity density compared with other ZBFBs. h Flow batteries performance comparison at low and room temperatures. This work delivers high energy density under both room and low temperatures, with relatively low operating temperature and stable performance, establishing NH4Cl-supported ZBFBs as ideal all-climate flow batteries. For all flow batteries comparison, electrolyte volume is the sum of the posolyte and negolyte. Source data for (b, g, h) are provided in Supplementary Tables 16–18. Discussion Based on the systematic investigation conducted, employing NH4+ as the cation of the SE in ZBFBs displays a series of advantages: i) The salting-out ability of NH4+ is weaker than that of K+, resulting in a lower phase transition temperature for NH4+-based electrolytes. This effectively prevents pipeline blockages due to salt precipitation at low temperatures, thereby enhancing the reliability of battery operation in cold conditions. ii) Compared to the alkali metal ions (K+, Na+, Li+), NH4+ not only exhibits the highest diffusion coefficient but also enhances the diffusion of other ions in the electrolyte with a specific molarity (4 M), which effectively accelerates the ion transport process within the electrolyte. iii) NH4+ helps prevent the growth of zinc into the membrane, effectively mitigating the risk of battery short-circuit failures. iv) NH4Cl is among the least expensive cation chlorides. Therefore, despite the complexity of its electrolyte components, NH4+ is highly suitable for widespread application in ZBFBs. We verified that, in the traditional complex electrolyte of ZBFBs, the key determining factor triggering the initial phase transition is the precipitation of KCl salts at low temperatures. Compared with K+, NH4+ was introduced to not only significantly suppress the phase transition of the electrolyte but also effectively accelerate the ionic transport kinetics. The inherent trade-off between salting-out and ion transport at low temperatures from the monovalent alkali cations (K+, Na+ and Li+) of SEs is broken. Additionally, NH4+ can significantly enhance the battery’s cycling stability by appropriately increasing the solubility of polybromides. The ZBFBs supported by NH4+ not only demonstrate sustained lifespan (>2300 cycles at 40 mA cm−2, over half a year) and robust high-rate cyclic stability (charging at 200 mA cm−2, discharging at 80 mA cm−2, >3300 h) at room temperature; they also exhibit cyclic stability (over 1600 cycles at 40 mA cm−2) at −20 °C. This work presents an effective strategy for developing high-performance ZBFBs capable of operating in cold conditions. Methods Materials All chemicals were used as received. Zinc bromide (ZnBr2, Israel Chemicals), choline chloride (N[1,1,1,2OH]Cl, Shanghai Macklin Biochemical Co., Ltd., China), potassium chloride (KCl, Liaoning Quan Rui Reagent Co., Ltd., China), potassium bromide (KBr, Shanghai Aladdin Biochemical Technology Co., Ltd., China), ammonium chloride (NH4Cl, Tianjin Kemiou Chemical Reagent Co., Ltd., China), ammonium bromide (NH4Br, Shanghai Aladdin Biochemical Technology Co., Ltd., China), sodium chloride (NaCl, Tianjin Kemiou Chemical Reagent Co., Ltd., China), sodium bromide (NaBr, Shanghai Aladdin Biochemical Technology Co., Ltd., China), lithium chloride (LiCl, Shanghai Aladdin Biochemical Technology Co., Ltd., China), lithium bromide (LiBr, Shanghai Aladdin Biochemical Technology Co., Ltd., China), bromine (Br2, Sinopharm Chemical Reagent Co., Ltd., China), soluble starch ((C6H10O5)n, Shanghai Aladdin Biochemical Technology Co., Ltd., China), potassium iodide (KI, Shanghai Aladdin Biochemical Technology Co., Ltd., China), Iodine (I2, Shanghai Aladdin Biochemical Technology Co., Ltd., China), sodium 3-(trimethylsilyl)propane-1-sulfonate (Shanghai Aladdin Biochemical Technology Co., Ltd., China), deuterated H2O (Shanghai Aladdin Biochemical Technology Co., Ltd., China). Carbon felt (Liaoyang J-Carbon Materials Co., Ltd., China). Hydrophilic Daramic membranes (Daramic® HP Polypore (Shanghai) Membrane Products Co., Ltd, China), ~900 µm thick, with an average pore size of around 0.1 µm, a porosity of about 58%50, and no selectivity. All electrolytes in this work were prepared in ambient laboratory conditions at room temperature (25 °C ± 3 °C). Ultrapure water produced by a water purification system (PURELAB flex 3, ELGA LabWater, USA; conductivity ~18.2 MΩ) was used as the solvent. During preparation, the solutions were stirred using a PTFE-coated stir bar and stored in glass bottles. Visualized experiment of the low-temperature phase transition of the electrolyte During the visualization experiment, each sample was held at a specified temperature for more than three hours to ensure complete phase transition. The UR is defined as the ratio of the reduced Zn2+ concentration to the pristine Zn2+ concentration during the charging process. The temperature was continuously decreased throughout the test duration. Characterizations The X-ray diffraction measurements were obtained using an X-ray diffractometer (D8 ADVANCE ECO; RIGAKU, Japan). The melting points of the negative electrolyte at the pristine state and at 50% usage rate state were characterized by Differential Scanning Calorimetry (DSC, NETZSCH STA 449 F3) from −125 to 25 °C with the temperature change rate of 10 °C min−1. X-ray fluorescence (XRF) was carried out on an Almelo Zetium X-ray Fluorescence Spectrometer (Netherlands) for elemental semi-quantitative analysis on solids. The white solid depicted in Fig. 2a was derived from the electrolyte after sustaining it at −10 °C for over 24 hours, followed by filtering and drying. XRF analysis of this solid is presented in Supplementary Fig. 2d, where the amount of substance of the potassium element was normalized to 1 for other elements. A METTLER TOLEDO SevenDirect SD30 (Switzerland) was employed to conduct electrical conductivity tests on various electrolytes. The kinematic viscosity is measured by a rotational viscometer (LC-NDJ-1T, China), and characteristic data are listed in Supplementary Table 4. A METTLER TOLEDO SevenExcellence Multiparameter (Switzerland) was employed to measure pH of various electrolytes. NMR spectroscopy was carried out on a Bruker AVANCE III HD 400 MHz equipment (Germany) to determine the effect of SE cations on the charge distribution of water. A liquid sample with a volume of about 400 μL was prepared via the coaxial double tube where the inner tube was filled with deuterated H2O and the outer tube was filled with the sample. The deuterated reagent was used as the field frequency lock for NMR tests. The field emission scanning electron microscopy (FE-SEM, JEOL 6360LV, Japan) images were employed to characterize the morphologies of the ‘Daramic membrane – zinc deposition – Carbon felt’ ternary complex. Ultraviolet−visible (UV−vis) spectroscopy was measured to determine the concentration of Br2 in ZBFB positive electrolyte supernatant based on the iodine-starch color reaction. UV-vis data were collected on TU-1901 UV-vis spectrophotometer. A KI-starch solution with a concentration of 2.5 g / 500 mL starch and 1.25 g/500 mL KI was prepared. An iodine stock solution for the iodine-starch colorimetric assay was prepared with 5 mM I2 and 50 mM KI in a solvent comprising ultrapure water and ethanol at a ratio of 1:1. The colorimetric solution was prepared as indicated in the Supplementary Table 10 and perform UV-Vis measurements. For I2 concentrations ranging from 10.08 to 70.56 μM, a standard curve consistent with Lambert-Beer’s law is achieved, with an R² of 0.9999. The standard curve is shown in Supplementary Fig. 30b. The concentration of Br2 in the solution was determined using the following protocol: The supernatant of the ZBFB positive electrolyte was diluted to 1/40 of their respective original concentrations using a solvent comprising ultrapure water and ethanol at a ratio of 1:1. A mixture comprising 200 μL of the diluted solution, 1.8 mL of ultrapure water, and 8 mL of KI-starch solution was prepared. The absorbance of this mixture was measured via UV spectroscopy, and the concentrations of the positive electrolyte supernatant were quantified based on the standard curve. Electrochemical characterization Metal electrodes were stored dry under sealed conditions and polished before each use using 3000-grit or finer sandpaper. Prior to use, the glassy carbon electrode was polished with a chamois cloth to renew the surface. After polishing, all electrodes were rinsed with ultrapure water and ethanol and dried with a gas stream. CV at room temperature The room temperature tests of CV curves at a series of scan rates were used to measure the apparent diffusion coefficient of Zn2+/Br2. All of them were performed using a three-electrode cell configuration. Glass carbon (GC) (diameter(Ø) 3.0 mm, Ada Hengsheng Technology Co., Ltd.) as the working electrode under ambient atmosphere. A zinc plate (25 × 30 × 0.6 mm) was used as the counter electrode when measuring D(Zn2+). A platinum plate (10 × 10 × 0.1 mm) was used as the counter electrode when measuring D(Br2). A Saturated Calomel Electrode (SCE, pre-soaked in saturated KCl solution, Ada Hengsheng Technology Co., Ltd.) was used as the reference electrode. CV tests were conducted on an electrochemical workstation (CHI604, CH Instruments, Inc.) at a scan rate of 40, 60, 80, 100, 120 mV s−1. For CV measurements used to determine D(Zn2+), the potential was scanned from −0.2 V to −1.6 V and back to −0.2 V. For CV measurements used to determine D(Br2), the potential was scanned from 0.85 V to 0.2 V and back to 0.85 V. The relationship between the peak current and scan rate can be described as Where peak current ip (A), electron transfer number in electrode reactions n, Faraday’s constant F (96485.33 C mol−1), the electrode’s true surface area A (cm−2), diffusion coefficient of reactants D0 (cm2 s−1), scan rate ν (V s−1), initial concentration of reactants C0* (mol cm−3), gas constant R (8.413 J mol K−1) and thermodynamic temperature T (K). The composition of the electrolyte is shown in Supplementary Table 5. Ion transport numbers Ion transport numbers (ITNs) were measured to determine the ability of diffusing through the Daramic membrane of different ZBFB SE cations. A concentration cell was utilized to determine the zero-current voltage (V0) with Nernst relationship51,52. The device comprises an H-shaped cell, featuring two half-cells partitioned by a Daramic membrane (Fig. 3i), which is macroporous (~0.1 µm pore size) devoid of ion selectivity ability. The I-V curves of the device were collected with Gamry Interface 1000 or Gamry Reference 3000. Ion transport numbers were derived from the zero-current voltage on the I-V line treated by the Nernst relationship: Where zero-current voltage V0 (V), ion transport number t, Boltzmann constant k (1.380649 × 10−23 J K−1), thermodynamic temperature T (K), elementary charge e (1.602176634 × 10−19 C), the ratio of high concentration to the low concentration ΔC, the average activity coefficient in the high concentration solution and low concentration solution γ±high and γ±low, respectively. Subscript “+” and “−” represent cation and anion, respectively. The composition of the electrolyte is shown in Supplementary Table 5. Activity coefficients for high and low concentration solutions are listed in Supplementary Table 6-7. Mean values and standard deviations are used to represent the ITNs for different salts. All error bars represent standard deviations from at least three independent tests. Cyclic voltammetry at various temperatures The tests at different temperatures of CV curves at a series of scan rates were used to measure the apparent diffusion coefficient of Zn2+. CV tests were conducted on an electrochemical workstation (CHI604, CH Instruments, Inc.) with a three-electrode cell configuration where a GC (diameter(Ø) 3.0 mm, Ada Hengsheng Technology Co., Ltd.) was used as the working electrode, a zinc plate (25 × 30 × 0.6 mm) was used as the counter electrode and a platinum plate (10 × 10 × 0.1 mm) was used as the reference electrode. The scan rate was from 60 mV s−1 to 140 mV s−1, the temperature series was −10 °C, 0 °C, 10 °C and 30 °C. The potential was scanned from 0 V to −1.6 V, then to −1.95 V, and finally returned to 0 V. The composition of the electrolyte is shown in Supplementary Table 5. Steady-state polarization curve (Tafel plot) at various temperatures The tests at different temperatures of Tafel plots were used to measure the energy barrier ΔG≠ and standard rate constant k0 of Zn2+ ↔ Zn at the equilibrium potential with a three-electrode cell configuration. A GC (diameter(Ø) 3.0 mm, Ada Hengsheng Technology Co., Ltd.) was used as the working electrode, a zinc plate (25 × 30 × 0.6 mm) was used as the counter electrode and a platinum plate (10 × 10 × 0.1 mm) was used as the reference electrode. A scan rate of 10 mV s−1 was used. The potential was scanned within ±300 mV (vs. Zn). Tafel curves data were collected using an electrochemical workstation (CHI604, CH Instruments, Inc.). The temperature series was −10 °C, 0 °C, 10 °C, 20 °C and 30 °C. When the current i significantly exceeds exchange current i0, the electrochemical equilibrium is substantially disrupted, causing the electrode potential to deviate markedly from the equilibrium potential and placing the electrode in a highly polarized state. The Tafel formula is as follows Where gas constant R (8.413 J mol K−1), thermodynamic temperature T (K), electron transfer number in electrode reactions n, Faraday’s constant F (96485.33 C mol−1), apparent transfer coefficient α and β, exchange current i0 (A) and current (i). The exchange current i0 is related to the reaction rate of Zn2+ ↔ Zn30: According to the Butler-Volmer model of electrode reaction kinetics28: Where exchange current i0 (A), Faraday’s constant F (96485.33 C mol−1), the electrode’s true surface area A (cm−2), initial concentration of reactants C0 (mol m−3), temperature dependent rate constant kT (cm K s−1 V−1), standard rate constant k0 (cm K s−1 V−1), energy barrier ΔG≠ (kJ mol−1), gas constant R (8.413 J mol K−1), thermodynamic temperature T (K). The energy barrier and standard rate constant can be calculated by measuring the exchange current density at different temperatures. The composition of the electrolyte is shown in Supplementary Table 5. LSV at room temperature The tests of LSV were used to measure the Hydrogen evolution window of NH4Cl electrolyte. LSV tests were conducted on an electrochemical workstation (CHI604, CH Instruments, Inc.) with a three-electrode cell configuration where a GC (diameter(Ø) 3.0 mm, Ada Hengsheng Technology Co., Ltd.) was used as the working electrode, a platinum plate (10×10×0.1 mm) was used as the counter electrode and a Saturated Calomel Electrode (SCE, pre-soaked in saturated KCl solution, Ada Hengsheng Technology Co., Ltd.) was used as the reference electrode. The scan rate was 10 mV s−1. For electrolytes containing Zn2+, the potential was scanned from 0 V to −1.3 V. For Zn2+-free electrolytes, the potential was scanned from 0 V to −2.0 V. DFT calculations Density functional theory for solvation free energy and binding energy Density Functional Theory (DFT calculations were consulted to optimize the geometry and frequency of the molecule structures based on B3LYP functional, 6-311 g(d) basis set in the Gaussian 16 packages53. The optimized geometries are provided in Supplementary Dataset 1. The HOMO/LUMO energies of MBr3/MBr5 complexes were calculated by DFT. The energy of the studied N–H···O hydrogen bond in NH4+···H2O was calculated at the B2PLYP/def2-QZVP level, with each configuration generated by rotating a specified angle from the geometry-optimized complex. The solvation free energy (ΔGsolv) is calculated based on the following formula: where Esolv and Egas refer to electronic energy with the solvent and without the solvent, respectively, and the solvent is water in this work. Esolv and Egas are calculated by M052x/6-31 g(d) calculation level after geometry optimization and frequency, and incorporated Grimme’s D3 dispersion correction to account for long-range interactions54. The binding energy (ΔE) was computed at the B2PLYP/def2-QZVP level following geometry optimization and vibrational-frequency analysis, according to the following equation: Where E(AB), E(A) and E(B) represent the energy of complexes AB, fragment A, and fragment B, respectively. DFT calculations for adsorption energy of M+ on Zn (002) plane DFT calculations for adsorption energy were carried out using the Vienna Ab initio Simulation Package (VASP 6.2.0)55. Exchange-correlation potential energies were treated with the Perdew–Burke–Ernzerhof functional within the generalized gradient approximation56. The interaction between valence electrons and ionic cores was described by the projector augmented-wave (PAW) method57, with a plane-wave cutoff of 450 eV. Geometry relaxation was continued until the residual force on each atom was below 0.05 eV Å−1, and the electronic self-consistent convergence threshold was set to 10−5 eV. A 2 × 2 × 1 k-point grid was used during geometry optimization, and a denser 4 × 4 × 1 grid was adopted for electronic-structure calculations. The Zn substrate was represented by a four-layer 4 × 4 slab containing 64 atoms. The lattice parameters were a = b = 10.65 Å, c = 22.42 Å, with α = β = 90° and γ = 120°. The adsorption energy of a monovalent cation on Zn (002) (Eads) was defined as: Where Etotal is the total energy of the adsorption system, Esurface is the energy of the bare Zn (002) surface, and EM+ is the energy of the isolated cation. The results are summarized in Supplementary Fig. 29. MD simulations The simulation structure was constructed by Packing Optimization for Molecular Dynamics Simulations (Packmol)58. Moltemplate package was used to generate the topology files for LAMMPS simulations59. To obtain the accurate charge of choline cation, the restrained electrostatic potential (RESP) charges for the choline cation were calculated using following method prior to their integration into MD simulations. Molecular geometries were first optimized at the B3LYP-D3(BJ)/def2-SVP level of theory. Subsequently, single-point energy was calculated at the B3LYP-D3(BJ)/def2-TZVP level to obtain accurate electrostatic potential (ESP) data. Multiwfn software was then utilized to derive RESP charges from the computed ESP data60. Unless stated otherwise, the box size and number of particles in the model are established following the protocol outlined in Supplementary Table 2-3. MD simulations were performed using the LAMMPS package61. The forced field parameters of Zn2+, K+, NH4+, Na+, Li+, Cl−, Br− and Ch+ were consulted from OPLS_AA force fields62. OPC3 was chosen as the water model63. The velocity-verlet integral method was used with 1 fs integration time step64. The electrostatic interactions and vdW cutoff radii were set to 12 Å. The long electrostatic interactions were calculated by Particle-particle particle-mesh65 solver method with a precision of 1 × 10−4. Simulation system was initially equilibrated for 5 ns in NPT by Nosé-Hoover thermostat/barostat algorithm with annealing temperature from 373.15 K to 298.15 K at 1 bar and continued calculating another 10 ns at 298.15 K after energy minimization66,67. Once the system was fully equilibrated, the data from another 5 ns or more were used for statistical analysis in the NVT ensemble, which was controlled by the Berendsen thermostat algorithm68. The first and last frames of the trajectories used for data analysis were provided in Supplementary Dataset 2. The radial distribution functions (RDF) and CN curves were consulted after MD simulations. The CN in the first solvation shell is defined as the CN value at the first minimum of g(r) following its initial maximum. Additional details are provided in the Supplementary Methods. Number of hydrogen bonds Hydrogen bond analysis was conducted using the VMD plug-in “Hydrogen Bonds”69, with one frame extracted every 500 fs from the MD trajectory over 5 ns, totaling 10,000 frames. The calculation of HB is based on the geometrical configuration that the distance between hydrogen-bond-donor (HBD) and the hydrogen-bond-acceptor (HBA) is <3.5 Å and the angle of HBD–H···HBA is <30°. HBD and HBA are usually atoms such as O, N, etc. The model is constructed using a 1 M ZnBr2 + 4 M MCl + 1 M ChCl electrolyte within a 50 Å × 50 Å × 50 Å simulation box, containing 2794 H2O, 376 Cl−, 150 Br−, 301 M+ (K+ or NH4+), 75 Zn2+, and 75 Ch+. Relaxation time The relaxation time τ is determined using the autocorrelation functions of corresponding existence functions. For hydrogen bond relaxation time, the function h(t) is dependent on both the cutoff angle and cutoff distance. The existence of HB is determined by geometrical configuration that the distance of hydrogen-bond-donor (HBD) and the hydrogen-bond-acceptor (HBA) is <3.5 Å and the angle of HBA···H–HBD is <30°. HBD and HBA are usually atoms such as O, N, etc. h(t) equals 1 if the conditions for a HB between the HBD and the HBA are satisfied, otherwise h(t) is 0. The autocorrelation function C(t) provides the probability of HBD–H···HBA HB still existing until time τ. The decay of the autocorrelation function can be quantified by calculating the concerned time scale of the decay. An increase in relaxation time indicates that the hydrogen bond persists longer from formation to dissociation. Equation (11) results in an autocorrelation function that has a decaying character, and which is then fitted with a stretched exponential function. τ1 is relaxation time and τ2 is the stretching parameter. For the HB cutoff time in different electrolytes, the autocorrelation functions show good fitting concerning the stretched exponential function. For the HB between water molecules and HB between NH4+ and water, curves fitted with the stretched exponential function are shown in Supplementary Fig. 12. Mean squared displacement MSD curves were obtained after MD simulations. The self-diffusion coefficient D was derived according to the Einstein equation: Potential of mean force The potential of mean force (PMF) reflects the kinetic barrier of breaking the atom pair A-B. Here PMF was used to determine the kinetic barrier of water molecules in the fi rst solvation shell of Mn+. The PMF is related to the RDF using the equation: The dissociation energy, the kinetic barrier of breaking the atom pair A-B, is obtained using the following equation: Averaged noncovalent interaction (aNCI) aNCI analysis is employed to delineate the weak noncovalent interactions within fluctuating environments, effectively representing van der Waals interactions, hydrogen bonds, and steric repulsion60,70. The aNCI method illustrates weak interaction areas by mapping the RDG function’s isosurface and overlaying the sign (λ2) function in distinct colors to classify the types of interactions. The average RDG function is determined using the following formula: where ρr¯ is the average density, ∇ρr¯ is the average density gradient, and λ2 is the second largest eigenvalue of the electron density Hessian matrix. Both aRDG and λ2 are calculated by averaging multiple frames. Batteries tests The ZBFB was assembled by using a Daramic membrane between two carbon felt electrodes, clamped by two graphite plates (without flow channels) under ambient atmosphere. The Daramic membrane is a macroporous membrane (~0.1 µm pore size) devoid of ion selectivity ability. The active area of the electrode is 6 × 6 cm2. All of these components were fixed between two stainless steel plates (Supplementary Fig. 42). The electrolyte was cyclically pumped through the corresponding electrodes in airtight pipelines. The cycling tests were conducted by Arbin (LBT21084, America) or NEWARE (CT-4008T-5V12A, China). The battery test at room temperature utilized a magnetic pump at both positive and negative side with a flow rate of ~50 mL min−1. The charge process was controlled by the charge time to keep a constant charge capacity, while the discharge process ended with a cut-off voltage. Error bars shown in Fig. 5c represent the sample standard deviation (SD). The battery test at low temperature (<0 °C) utilized a peristaltic pump at both positive and negative side with a flow rate of ~50 mL min−1. All components of the low-temperature flow battery, including peristaltic pumps, electrolyte storage tanks, and single cell, are placed within a high and low temperature test chamber (TEMI880), ensuring an accurate temperature for the ZBFB. The battery’s electrolyte composition, operating temperature, charge and discharge current, charging time, discharge cut-off voltage, electrolyte volume on one side (electrolyte volume of the positive and negative electrodes are the same), UR (based on the volume of one side) are listed in Supplementary Tables 11 and 12. Low-temperature power density tests were performed with a NEWARE CT-4008T-5V12A (China). The active area of the electrode is 3 × 3 cm2. Peristaltic pumps on the positive and negative sides maintained a flow rate of ~50 mL min−1, and each side of the battery held 10 mL of electrolyte. Power density was tested by first charging the electrolyte to 33.58% UR, followed by charging at a current of 40 mA cm−2 to achieve an extra area capacity of 0.58% UR. Discharge tests were then conducted at currents ranging from 20 mA cm−2 to 240 mA cm−2 to reach an area capacity of 0.58% UR. Ultimately, the charging and discharging processes were conducted alternately, resulting in the determination of power density across all discharge current ranges. Last but not least, in our work, both energy density and volume capacity are calculated based on the aggregated volumes of both posolyte and negolyte. Unless specified, the volume of the posolyte is the same as that of the negolyte. The negolyte was contained in polypropylene (PP) bottles, whereas the posolyte was contained in glass bottles. Current density and areal capacity were calculated by dividing the total current or total discharged capacity by the electrode active area. Volumetric capacity and volumetric energy density were calculated by dividing the discharged capacity or discharged energy by the electrolyte volume. At least two cells were measured for each electrochemical test. When compared with other works, both energy density and volume capacity are based on the aggregated volume of posolyte and negolyte, and cycle times are calculated from the maximum durations reported in these articles. Comprehensive comparison of monovalent chloride as SEs in ZBFBs The data and metrics used in Fig. 6b are detailed below: (a) Cost: unit prices on April 25, 2024 from Alibaba (International price); molar cost is converted from per-kilogram/ton values in Supplementary Table 15 and Supplementary Dataset 3. (b) M+/ Zn2+/ Br− migration rates: mean self-diffusion coefficients obtained from MSD analysis of MD simulations at 298.15 K and 253.15 K. (c) Pristine electrolyte phase transition temperature: temperatures from Fig. 2b, c. (d) RT Ave EE: average energy efficiency at room temperature, determined by the average energy efficiency over all current densities in the room temperature rate test (Supplementary Fig. 36). (e) LT failure cycle: the cycle number at which failure first appears under the Fig. 5a conditions (40 mA cm−2, 25 mAh cm−2). (f) RT Ave t+: room temperature average t+, obtained by averaging the M+ mobility values for MCl and MBr from Fig. 3i. Optimal concentration of SE in ZBFB electrolytes The optimal concentrations of the different SEs were determined by first ensuring the lowest phase-transition temperature for both the pristine and 50% UR electrolytes, then selecting the composition exhibiting the highest ionic conductivity among eligible candidates (Supplementary Fig. 3b, c). Supplementary information Supplementary information Description of Additional Supplementary Files Dataset 1 Dataset 2 Dataset 3 Supplementary movie 1 Transparent Peer Review file Source data Source data Acknowledgements The authors acknowledge the financial support from the National Natural Science Foundation of China (Grant No. 22278392 to Y.Y.), CAS Strategic Leading Science & Technology Program (A) (XDA0400201 to X.L.), and the Youth Innovation Promotion Association, CAS (2022184 to Y.Y.). Author contributions X.L. and Y.Y. conceived the project. X.L. and Y.Y. supervised the research. T.C. contributed to data and schematic visualization, material preparation, battery fabrication and operation, characterization, and data analysis. M.Z. contributed to materials preparation and characterization. T.C., T.L., and S.W. performed DFT calculation. T.C. and T.L. performed MD simulations. T.C., Y.Y. and X.L. wrote the paper. All authors engaged in discussions related to the manuscript. Peer review Peer review information Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. [A peer review file is available. Data availability All relevant data are included in the paper and its Supplementary Information. Source data are provided with this paper. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Yanbin Yin, Email: yinyanbin@dicp.ac.cn. Xianfeng Li, Email: lixianfeng@dicp.ac.cn. Supplementary information The online version contains supplementary material available at 10.1038/s41467-026-71846-6. References Associated Data Supplementary Materials Supplementary information Description of Additional Supplementary Files Dataset 1 Dataset 2 Dataset 3 Supplementary movie 1 Transparent Peer Review file Source data Data Availability Statement All relevant data are included in the paper and its Supplementary Information. Source data are provided with this paper.