A Hydrogen Iron Flow Battery with High Current Density and Long Cyclability Enabled Through Circular Water Management

The hydrogen‐iron (HyFe) flow cell has great potential for long‐duration energy storage by capitalizing on the advantages of both electrolyzers and flow batteries. However, its operation at high current density (high power) and over continuous cycling testing has yet to be demonstrated. In this article, we discuss our design and demonstration of a water‐management strategy that supports high current and long‐cycling performance of a HyFe flow cell. Water molecules associated with the movement of protons from the iron electrode to the hydrogen electrode are sufficient to hydrate the membrane and electrode at a low current density of 100 mA cm−2 during the charge process. At higher charge current density, more aggressive measures must be taken to counter back‐diffusion driven by the acid concentration gradient between the iron and hydrogen electrodes. Our water‐management approach is based on water vapor feeding in the hydrogen electrode and water evaporation in the iron electrode, thus enabling high current density operation of 300 mA cm−2.


Introduction
The intermittent electricity generated from renewable energy sources, such as solar and wind, presents a challenge for a stable and reliable power grid. [1] The use of an energy storage system that has the capability to store large quantities of energy can effectively help balance the fluctuations in electricity supply and demand by modulating the storage and releasing of energy. [2] The continuing decrease in the cost of generating electricity from renewable resources, such as wind and solar photovoltaics, and the urgent need to decarbonize the economy has stimulated much research and development in low-cost stationary electrical energy storage for future societal and economic needs. [3][4][5][6][7][8] Redox flow battery (RFB) technology is a promising electrical energy storage concept because it could support a variety of critical services for grid energy applications (e.g., premium back-up power, daily energy time shift balancing, and improved grid reliability). [9][10][11][12] RFBs can enable large-scale energy conversion and storage by storing redox-active chemical species in the liquid electrolyte that can undergo reversible electrochemical reactions on demand. The most extensively researched and developed RFB system for grid energy applications is the all-vanadium redox flow battery in which electrolytes containing the same vanadium element, but at different valence states, circulate through two half-cell cycles. [13][14][15][16][17] The high cost of vanadium, however, has limited the potential to significantly reduce the system cost, so broader penetration into the grid energy storage market is constrained.
In the search for low-cost RFB systems, the hydrogen iron (HyFe) flow cell emerges as a promising candidate. [18] A HyFe flow cell pairs the hydrogen oxidation and evolution reaction (HOR/HER) in the negative half-cell with the Fe 2+ /Fe 3+ redox reaction by circulating the iron electrolyte in the positive half-cell. As shown in Figure 1a, a typical sandwich cell uses porous carbon, Pt/C catalyst coated carbon paper, and Nafion cation exchange membrane as positive and negative electrodes and the membrane, respectively. During the discharge process, the hydrogen is oxidized to protons that pass through the electrolyte membrane to the positive half-cell where iron (III) is reduced to iron (II). A reverse process takes place during the charge process. [19] In essence, the HyFe system is a hybrid flow system that operationally falls between an electrolyzer/fuel cell and a flow battery, [20] This hybrid concept capitalizes on the compressibility of hydrogen to potentially decrease the footprint of the flow system significantly while using the widely available iron resource to lower the cost. [20][21][22][23][24] A notable advantage of the HyFe flow cell is that the HOR/HER and the iron redox reaction are kinetically fast, so the system potentially could achieve high power densities on the order of 270 mW cm À2 , [22] higher than other iron-based flow batteries, e.g., the all iron flow battery (160 mW cm À2 ), [25] alkaline zinc-iron flow battery (~250 mW cm À2 ), [26] and iron-chromium flow battery (178 mW cm À2 ), [27] and even higher than commercialized VRFBs (120-160 mW cm À2 ). [28] Other advantages similar to those of typical flow batteries include the modular design, easy scalability (capacity proportional to tank size), low self-discharge, etc. [29] The hydrogen-iron (HyFe) flow cell has great potential for long-duration energy storage by capitalizing on the advantages of both electrolyzers and flow batteries. However, its operation at high current density (high power) and over continuous cycling testing has yet to be demonstrated. In this article, we discuss our design and demonstration of a water-management strategy that supports high current and long-cycling performance of a HyFe flow cell. Water molecules associated with the movement of protons from the iron electrode to the hydrogen electrode are sufficient to hydrate the membrane and electrode at a low current density of 100 mA cm À2 during the charge process. At higher charge current density, more aggressive measures must be taken to counter back-diffusion driven by the acid concentration gradient between the iron and hydrogen electrodes. Our water-management approach is based on water vapor feeding in the hydrogen electrode and water evaporation in the iron electrode, thus enabling high current density operation of 300 mA cm À2 .
Although several publications have addressed HyFe flow cells, the performance of current HyFe battery technology still is facing significant challenges in terms of cyclability and high rate cycling. For instance, the highest reported charge current density of a HyFe battery is less than 300 mA cm À2 , and the capacity utilization decreases to only 10% when the charge current density is higher than 200 mA cm À2 . [20] Even worse, the cycling durability of HyFe has exhibited only~20 cycles at a low current density of 100 mA cm À2 , [22] and no data about longer cycling and high-rate cycling (e.g., >300 mA cm À2 for charge/discharge cycling testing) have been reported. Yet, the major performance deficiencies of current HyFe systems are related to a single root cause involving poor management of water transfer that supports fast redox reactions on the hydrogen electrode. Similar to the polymer electrolyte membrane (PEM) fuel cell/electrolyzer, water management in a HyFe flow cell is critical for achieving high performance levels. Unlike the traditional PEM fuel cell/electrolyzer, in which the goal of water management is to maintain membrane hydration and avoid water flooding/ drying at the two electrodes, [30][31][32] water molecule movement in a HyFe flow cell is even more complicated because of the high acid concentration in the iron electrolyte. During discharge, the water molecules tend to move to the iron side because of both the acid concentration gradient and the charge carrier (proton) movement. At the same time, the humidified hydrogen can maintain the hydration of the membrane and hydrogen electrode. Although during the charge process, water molecules associated with the movement of protons will be transported back to the hydrogen electrode. This transport of water molecules, however, is not sufficient to hydrate the membrane, especially during high current operation. During cycling, this imbalanced water transport results in reduced conductivity in both the membrane and the hydrogen electrode because of the water loss and the elevated concentration overpotential in the iron electrode due to flooding. These conditions ultimately contribute to the short cycling life and low current density demonstrated thus far. In this study, we designed and developed a new circular water management approach in which water transport to the hydrogen electrode as vapor and water evaporation in the iron electrode enable high current and long-cycling performance of the HyFe flow cell.

Result and Discussion
Cyclic voltammetry testing ( Figure 1b) was conducted on Pt/C catalyst-coated glass carbon in the iron electrolyte (0.1 M FeSO 4 + 0.25 M H 2 SO 4 ) to identify the redox pair of the HOR/HER and Fe 2+ / Fe 3+ . The average potential for HOR/HER and Fe 2+ /Fe 3+ redox is À0.705 and 0.025 V vs Hg/HgSO 4 , respectively. As a result, the HyFe flow cell has a voltage of 0.73 V. To provide a baseline, performance of the HyFe flow cell was studied first without any water management during the charge process. As shown in Figure 1c, the electrochemical performance of hydrogen iron flow cell was measured under a constant current density of 100 mA cm 2 . The charge and discharge voltage plateau are~0.707 and~0.609 V, respectively. The columbic efficiency (CE) at a current density of 100 mA cm À2 was as high as 99%-100%, indicating that the Fe 2+/3+ was highly reversible. High energy efficiency (EE) of~82% at current density of 100 mA cm À2 was achieved mostly because of the fast kinetics of anodic and cathodic active materials. It is noted that there was no substantial voltage drop or anode flooding was observed, which is typical during operation of PEM fuel cells at low current density at low operating temperatures. [33,34] We attribute this behavior to the iron concentration gradient between the iron electrode and the hydrogen electrode. This gradient provides the driving force for water-molecule transport from the hydrogen electrode to iron electrode, which prevents water flooding at the anode during the discharge process of HyFe flow cells at low current density and low operating temperatures. However, a rapid increase in the charge voltage of the HyFe flow cell was observed when the current density increased to 300 mA cm À2 , with only 18% of theoretical capacity (440 mAh) available to be charged. We performed AC impedance analysis to identify the resistance of the flow cell before and after charging the cell at a current density of 300 mA cm À2 (Figure 1d). Interestingly, there is negligible difference in ohmic and charge transfer resistance before and after charging the cell at 300 mA cm À2 as the two impedance spectra are nearly identical. Another interesting observation is that after charging the cell at 300 mA cm À2 , the electrochemical performance of the HyFe flow cell can be recovered at a lower current density of 100 mA cm À2 ( Figure S3, Supporting Information). The observed poor electrochemical performance at 300 mA cm À2 presumably is due to depletion of the water molecule in the membrane and at the hydrogen electrode. This depletion outpaces the level of water-molecule transport from the iron electrode to the hydrogen electrode. The water transport phenomenon in a HyFe flow cell is distinctively different from that in a traditional PEM electrolyzer during the charge process. Water back-diffusion and water molecules associated with the movement of protons from the oxygen electrode to the hydrogen electrode during the charge process of a traditional PEM electrolyzer usually are sufficient to maintain hydration of the membrane and the hydrogen electrode, even at a high current density of 2 A cm À2 . [35][36][37] However, in a HyFe flow cell, the natural water transport process cannot support the performance at a charge current density of 300 mA cm À2 . Additional measures must be taken to accelerate the transfer of water molecules at high charge current densities. These measures must consider both proton transport from the iron electrode to the hydrogen electrode and the high acid concentration gradient that will prevent more water transport to the hydrogen electrode, eventually leading to a dry membrane and hydrogen electrode.
To confirm that the conduction decrease caused by water transfer is the limiting factor for achieving high charge/discharge current density (high power) and to identify technical solutions for this issue, we designed several approaches for controlling water transport and mitigating its negative impact. Because drying is initiated in the hydrogen electrode, our first approach was to cycle liquid water into hydrogen electrode during the charge process. This approach successfully promoted electrochemical performance of the HyFe flow cell up to a current density of 300 mA cm À2 with charge/discharge voltage plateaus at 0.787 and 0.529 V, respectively. Further increasing the current density to 500 mA cm À2 resulted in an elevated overpotential with charge/discharge voltage plateaus at 0.851 and 0.423 V, respectively (Figure 2a).
Without the presence of the additional liquid water cycled into the hydrogen electrode, the HyFe flow cell exhibited poor electrochemical performance because of the dry membrane and hydrogen electrode at a high current density of 300 mA cm À2 . When liquid water is supplied to the hydrogen electrode, the performance of the HyFe flow cell improves significantly. An increase in the volume of the iron electrolyte was observed only after the first charge/discharge cycle, further confirming water movement to the iron electrode. However, as cycling proceeded, it is expected that the water movement would result in an accumulation of water in the positive half-cell and eventual flooding of the iron electrolyte. [38] Water flooding of the iron electrolyte would decrease the iron and proton concentrations, leading to higher overpotentials and lower energy efficiency. This postulation is confirmed by Figure 2b, which shows the cyclic performance of HyFe flow cell when liquid water is cycled in the hydrogen electrode during charge process. The first charge and discharge cycles delivered capacities of~0.425 Ah and~0.417 Ah, respectively. Significant decreases in charge/discharge capacities as cycling proceeded then were observed. After only four cycles, the charge and discharge capacities dropped to 0.382 and 0.367 Ah, respectively, which account for capacity decays of 10.1% charge and 12.0% for discharge, respectively. The voltage at 50% state of charge for each charge/discharge cycling is summarized in Table S1, Supporting Information. The voltage plateau of the HyFe flow cell with water cycling at the hydrogen electrode was around 0.787 V for the first charge cycle. There was a slight increase in at first three charge voltage plateaus, with increases to 0.789 and 0.810 V during the second and third cycles, respectively. The voltage rapidly increased tõ 1.0 V during the fourth cycle. The volume of iron solution (with a starting volume of 18 mL) also increased to 19, 20, 21, and 22 mL after the first, second, third and fourth cycle, respectively. Therefore, cycling water can maintain the hydration of the membrane and the hydrogen electrode, but at the same time, it can introduce severe water crossover to the iron electrode that without control, can rapidly decrease the iron and proton concentrations of iron electrolyte and thus lead to low energy efficiency and poor cycle life because of the reduced iron electrolyte conductivity with lower proton concentration. Based on the results described above, we explored a different approach in which water vapor was introduced into the hydrogen electrode during the charge process, thus providing sufficient hydration of the Nafion membrane while limiting the water crossover. Figure 3a shows voltage profiles of the HyFe flow cell as water vapor is cycled into the hydrogen electrode during the charge process. Similar to the flow cell with liquid water cycling, the charge current density of the HyFe flow cell with water vapor cycling also  can reach 500 mA cm À2 , and the EE and CE are around 53% and 93% (Table S2, Supporting Information), respectively, which are comparable with the flow cell with liquid-water cycling. At 300 mA cm À2 for the flow cell with water vapor cycling, the EE of 71% and CE of~98% (Table S2, Supporting Information) also are comparable with the flow cell with liquid-water cycling.
More importantly, only a negligible increase in the volume of iron electrolyte could be observed after one cycle, showing that water vapor cycling into the hydrogen electrode during the charge process can sufficiently hydrate the Nafion membrane and hydrogen electrode while limiting water crossover. Figure 3b and Figure S4, Supporting Information, show the charge/discharge profiles, CE, and EE of our HyFe flow battery at a current density of 300 mA cm À2 . Almost no voltage decay during the charge and discharge modes could be observed over six cycles. Furthermore, CE is maintained at around 96%-97% during six cycles with only negligible capacity decay ( Figure S4, Supporting Information). A high CE and stable discharge/ charge capacities during cycle testing at 300 mA cm À2 indicates that water vapor cycling during the charge process is an efficient way to hydrate the membrane and hydrogen electrode, while minimizing water transfer to maintain cycle durability. We also studied cycle performance and water transport behavior of the HyFe flow cell at a high current density of 500 mA cm À2 . Similar to the study at 300 mA cm À2 , water vapor was supplied to the hydrogen electrode during the charge process. Voltage profiles at different cycles at the high current density of 500 mA cm À2 are shown in Figure S5, Supporting Information. Voltages at the charge and discharge plateaus were~0.865 and~0.493 V, respectively. Almost no voltage decay was observed within six cycles, indicating that water vapor cycling in the hydrogen electrode during charge process can keep the membrane and hydrogen electrode hydrated at the high operation current density of 500 mA cm À2 . However, there is a decrease in charge/discharge capacities as cycling proceeds. We observed a slight decrease in capacity during the first five cycles, and the capacity decrease became more severe during the sixth cycle. We attribute this capacity decay to increased water crossover from water vapor from the hydrogen electrode to the iron electrode at a higher current density. This water crossover dilutes the iron electrolyte concentration and increases the concentration polarization, leading to the reduced capacity utilization. We therefore proceeded to further optimize water management for the HyFe flow cell at the high current density of 500 mA cm À2 . We noted a slight increase in the volume of iron electrolyte for the flow cell with water vapor cycling into the hydrogen electrode during charge process after only six cycles. We then carried out further studies on long-cycling performance of our HyFe flow cell with a starting solution of 40 mL of 1 M FeSO 4 + 2.5 M H 2 SO 4 . We kept the water reservoir temperature at 90°C to sustain the water vapor supply to the hydrogen electrode during the charge process. However, we kept the entire HyFe flow cell at room temperature. Figure 4a shows the CE and EE results of our hydrogen iron flow battery at a 300 mA cm À2 current density. The CE was kept maintained around 96%-97% during 100 test cycles. We achieved EEs as high as 70% at 300 mA cm À2 in the first 15 cycles. After 15 cycles, the EE started to decrease, dropping to 64% at the 50th cycle from~70% at the 1st cycle mainly because of the increased concentration polarization resulting from water crossover.  After 50 cycles, we performed a water evaporation process on the iron electrolyte reservoir to restore the proton and Fe 2+ /Fe 3+ redox couple concentrations and, thus, ameliorate the overpotential increase due to both the conductivity loss and the increased concentration polarization (The detailed water evaporation process is described in the experimental section). Both discharge capacity and EE were recovered at the 51st cycle, as shown in Figure 4a,b. The discharge capacity delivered 0.91 Ah at the first cycle and then decreased to 0.82 Ah after 50 cycles. After the water evaporation process, the discharge capacity recovered to 0.87 Ah at the 51st cycle and then decreased to 0.8 Ah at 100th cycle. The capacity retention of~88% after 100 cycles demonstrated the excellent cycle stability of our HyFe flow cell. Voltage profiles at different cycles are shown in Figure 4c,d. The voltage at 50% state of charge for each charge/discharge profile are summarized in Table S3, Supporting Information. The voltage plateau is~0.757 V at the first charge cycle and increases to 0.764 and 0.781 V at the 25th and 50th cycles, respectively. After the 50 cycles for the water evaporation process, the plateau voltage decreased to 0.769 V and then increased to 0.785 V after 100 cycles. Only~28 mv (~3.7%) decay was observed for the charge mode after 100 cycles. For the discharge mode, the voltage of plateau could maintain at~0.570 V during the cycling test, and almost no voltage decay was observed after 100 cycles. To the best of our knowledge, this is the first time a high current density HyFe flow cell has been successfully cycled for more than 100 cycles.

Conclusion
In this article, we describe three different approaches (i.e., no water, liquid water, and water vapor are cycled into hydrogen electrode during charge process) taken to optimize water management during the charge process. We demonstrated that water vapor introduced into the hydrogen electrode during the charge process at high current densities (i.e., 300-500 mA cm À2 ) provided sufficient hydration of the Nafion membrane while limiting water crossover. We also demonstrated a high reversible-capacity hydrogen iron flow battery that when employing water vapor cycling during the charge process and water evaporation process during discharge, achieved a high current density of 300 mA cm À2 with after 50 cycles. Water management, which is important for maintaining stable cycling, can be achieved relatively easily by feeding water vapor into the hydrogen electrode and using the water evaporation process in the iron electrode. For future larger devices, this process can be engineered into an automatic water management system based on the specifics of the device. The electrochemical cycling performance of a HyFe flow battery, including a high EE of~70% at 300 mA cm À2 current density, no voltage decay for the discharge mode,~3.7% voltage decay for the charge mode, and good capacity retention of~88% over 100 cycles, were successfully demonstrated.

Experimental Section
All the HyFe flow battery tests were performed on a typical sandwich single cell (homemade) that uses Pt/C catalyst (40 wt% Pt on carbon black, Fuel Cell Store, USA) coated carbon paper, Nafion 211 cation exchange membrane (Chemours, USA) and carbon electrode (SGLCarbon GmbH, Germany) as anode, electrolyte membrane, and cathode, respectively. The active area of the cell is 10 cm 2 , and the graphite bipolar plate is composed of serpentine flow fields. Before the battery assembly process, the fresh Nafion 211 membrane was treated in warm water (80°C) 1 hr. The electrolyte solutions were prepared with deionized water using FeSO 4 (analytical, Sigma, USA) and H 2 SO 4 (96%, Sigma, USA). Cyclic voltammetry (CV) testing was conducted using the three-electrode method in a hydrogensaturated iron electrolyte solution (0.1 M FeSO 4 + 0.25 M H 2 SO 4 ) at a scan rate of 5 mv s À1 . For the three-electrode setup, Pt/C catalyst-coated glass carbon (diameter: 2 mm; CH Instruments Inc., USA), graphite rod (Fuel Cell Store, USA), and Hg/HgSO 4 electrode (CH Instruments Inc., USA) were used as the work, counter, and reference electrodes, respectively. An 18-20 mL volume of iron electrolyte solution (1 M FeSO 4 + 2.5 M H 2 SO 4 ) is circulated through the cell at a flow rate of 40 mL min À1 . During the charge process, hydrogen is generated by applying a voltage across the two electrodes of the sandwich cell, while a peristaltic pump (Cole-Parmer, USA) is used to circulate the iron electrolyte though the positive half-cell. For the discharge process, the humidified hydrogen flows though the hydrogen electrode, while the iron electrolyte is still circulated though the positive half-cell. Alternating current (AC) impedance measurements are carried out between hydrogen electrode and iron electrode to identify the resistance of the flow cell. CV and AC impedance measurements are performed using a CHI 760 electrochemical workstation (CH Instruments Inc., USA).
Flow battery performance was evaluated using battery testing system (CT3001A, Landt Instruments, USA). The flow cell was charged ranging from 100 to 500 mA cm À2 at the cutoff voltage of 1.2 V at room temperature. Three different approaches were used to optimize water management during the charge process: 1) no water introduced to the hydrogen electrode; 2) water cycling at the hydrogen electrode; and 3) water vapor (90°C) cycling to the hydrogen electrode. For all the discharge processes at 100, 300, and 500 mA cm À2 , humidified hydrogen will flow though the hydrogen electrode. The test conditions for three different approaches are summarized in Table 1. For approach 2 ( Figure S1, Supporting Information), the liquid water was circulated into the hydrogen electrode during the charge process, while, during discharge, the humidified hydrogen instead of liquid water was circulated into the hydrogen electrode. For water vapor approach, a water reservoir that connected with the hydrogen electrode was heated to provide water vapor to humidify the hydrogen as shown in Figure S2, Supporting Information.
To investigate the long cycle performance over longer durations, the cycle test was performed at current density of 300 mA cm À2 with 40 mL 1.0 M FeSO 4 and 2.5 M H 2 SO 4 . During the long-cycling testing, water reservoir that connected with hydrogen electrode was always heated to provide water vapor and humidify the hydrogen. After the 50th cycle test, the iron electrolyte was heated to 50°C, and pure nitrogen gas was flowed through the headspace of the iron electrolyte reservoir to evaporate water. This water evaporation process on the iron electrolyte can maintain the concentration of the iron and proton. After the volume of iron electrolyte decreased to its original volume of~40 mL, another 50 cycles of charge and discharge were performed.