Introduction

With the increasing demand for renewable energy, such as solar and wind power, energy storage technologies have received intensified attention to realize their smooth output [1, 2]. Among numerous energy storage technologies, flow batteries (FBs) are considered a promising candidate because of their intrinsic features of high safety, long cycle life, and flexible design [3,4,5]. However, the traditional aqueous FBs, such as the vanadium flow battery and the iron–chromium flow battery, still suffer from low energy density due to the limited solubility (< 2 mol/L) of redox species, which has restricted their further applications [6, 7]. Among the various reactants used in FBs, halogen-based redox pairs, such as Br2/Br−, are endowed with high solubility (> 10 mol/L) and excellent electrochemical activity, which achieve a high energy density among aqueous batteries [8, 9], etc. However, in addition to Br2, higher valence states of bromine, such as Br+, also exist, and if a two-electron transfer process from Br− to Br2 and then to Br+ can be realized, the energy density of the assembled battery will be greatly improved. Recently, our group used Cl− with greater electronegativity to stabilize + 1 valence bromine to realize the reaction of Br−/Br+. However, introducing Cl− in the electrolyte may trigger the side reaction of Cl2 evolution at a high state of charge (SOC). In terms of the electrode, the appropriate diameter of Br can be intercalated in graphite to form a bromine–graphite intercalation compound (Br–GIC) [10, 11], which has been used as the cathode for the lithium-ion battery or the zinc-based battery [12,13,14]. The “water-in-salt” electrolyte or the molten hydrated electrolyte was used in these systems based on a strong interaction between the metal cation and Br− to realize the intercalation reaction. When Br− is oxidized to Br2, it is more inclined to intercalate into graphite to form Br–GIC rather than combine with the surrounding Br− ions in the electrolyte to form polybromide ions (Br3− and Br5−), which is completely different from conventional aqueous Br-based electrolytes [15, 16]. Moreover, with the help of more electronegative halide ions, such as Cl−, Br can be intercalated into graphite as Cn[BrCl]. However, using a water-in-salt electrolyte will greatly limit the energy density.

In this study, we found that in strongly acidic electrolytes, Br+ can be easily intercalated into graphite as CnBr without using “water-in-salt” electrolytes. During the intercalation process, Br− was first oxidized to Br2 and formed polybromide ions (Br3− or Br5−), which is consistent with a conventional aqueous electrolyte. Then, Br2 was further oxidized to Br+ and intercalated into graphite to form Br–GIC, which demonstrated a reversible property and a much higher redox potential (0.5 V vs. Br−/Br2). As a result, the assembled battery with Ti3+/TiO2+ as the anolyte demonstrated a 65% improvement in energy density compared with the conventional Br−/Br2 redox couple and continuously ran for more than 300 cycles, which demonstrated great potential for stationary energy storage application.

Experimental

Materials

Titanium sulfate (Ti(SO4)2, 96% mass fraction) was purchased from Shanghai Aladdin Bio-Chem Technology Co., Ltd. Hydrobromic acid (HBr, 40% mass fraction in water), H2SO4 (98%), and N-methyl-2-pyrrolidone (NMP) were purchased from Tianjin KeMiOu Chemical Reagent Co., Ltd. N-methyl-N-ethyl-pyrrolidinium bromide (MEP) was supplied by Israel Chemicals Co., Ltd. Carbon felt was supplied by Liaoyang Jingu Carbon Materials Co., Ltd. Graphite was purchased from Laixi Carbon Source Co., Ltd. Nafion 115 was supplied by Dupont. All chemicals were used as received. All electrolytes were prepared with deionized water.

Preparation of the Carbon Felt@Graphite Electrode

A schematic for preparing the carbon felt @graphite (CF@GF) electrode is shown in Fig. 1. The electrodes were prepared using a simple impregnation method. First, the graphite slurry was prepared with a mass ratio of NMP:graphite:PVDF = 100:10:1 and stirred for 12 h. Then, the carbon felt (6 × 8 cm2) (CF) was immersed in the graphite slurry for 10 s and dried in a vacuum oven at 60 ℃ for 12 h and, finally, soaked in ethanol for 5 min to enhance the hydrophilicity and dried at 60 ℃ for 2 h. The loading mass of graphite was 7.5 mg/cm2.

Fig. 1
figure 1

Schematic of CF@GF electrode preparation

Full size image

Characterizations

The positive electrolytes with different SOCs were characterized by Raman spectroscopy (NanoWizard Ultra Speed & inVia Raman, 532-nm laser source excitation, 10% excitation intensity, 30-s exposure time). The X-ray diffraction (XRD) patterns of graphite were recorded on a Rigaku Ultima IV X-ray diffractometer, which was operated at 40 mA and 40 kV using Cu Kα radiation (λ = 1.541 Å, 10°/min, 10°–60°). The cyclic voltammetry measurements were taken in a three-electrode system using a Gamry 1000 electrochemical workstation. The working electrode of carbon felt and CF@GF was fabricated by cutting them into a wafer with a diameter of 5 mm and a thickness of 2 mm. A saturated calomel electrode (0.242 vs. SHE) was used as the reference electrode, and a graphite plate with an area of 9 cm2 (3 cm × 3 cm) was used as the counter electrode. The galvanostatic intermittent titration technique (GITT) was used to evaluate the diffusion behavior of Br- in an electrolyte and graphite. The assembled battery is charged or discharged at a constant current for 10 s and then left to stand for 1 min. The voltage changes of ΔEt induced by each pulse and the steady-state voltage changes of ΔEs after relaxation were measured, and the diffusion coefficient variation in bromine from the electrolyte to the graphite was evaluated in terms of (ΔEs/ΔEt)2 [17]. Electrochemical impedance spectroscopy (EIS) tests were performed on a Gamry Interface 1000 with the amplitude set at 5 mV and the frequency at 10–2–106 Hz. For the full battery test, the anode was used as the reference electrode, and the counter electrode with the cathode served as the working electrode.

Flow Battery Measurements

The assembly of the single-flow battery was consistent with previous reports [18], the anolyte was circulated through the pump at a flow rate of approximately 60 mL/min, and the catholyte was sealed in the electrolyte cavity. Nafion 115 was used as the membrane, and the graphite-coated carbon felt was used as the positive electrode. The effective area of the electrode was 6 × 8 cm2, with a 2-mm titanium plate and graphite plate used as the positive and negative current collectors, respectively. These components were fixed by two stainless steel end plates with a 30% compression ratio of electrodes. For battery evaluation, the anolyte was in excess compared to the catholyte to ensure a full charge of the positive electrode. The capacity and energy density were calculated based on the volume of the catholyte only. The battery test was performed on a Neware battery test system (5 V 12 A, Neware Corp., Shenzhen, China).

Results and Discussion

Electrochemical Measurements

To investigate the intercalation process of Br, a cyclic voltammogram of the 50 mmol/L HBr + 1 mol/L H2SO4 electrolytes was evaluated on graphite (GF) and glass carbon (GC) electrodes (Fig. 2a). The redox process of Br2/Br− (0.83 V vs. SCE) could be observed on the GC and GF electrodes. As the voltage increased, a new oxidation peak appeared at 1.6 V on the inert GC electrode, and the peak current was proportional to the scan rate (Fig. S1), indicating that the intercalation process was adsorption controlled. Meanwhile, no corresponding reduction peak was observed, possibly because of the formation of irreversible Br+ products such as hypobromous acid (HBrO) [19]. For the GF electrode, a new redox pair with an oxidation peak at 1.47 V vs. SCE and a reduction peak at 1.12 V vs. SCE indicated the reversible intercalation/deintercalation of Br in graphite. Moreover, both reduction peak currents demonstrated a good linear relationship with the square root of the scan rate, indicating diffusion-controlled behavior (Fig. 2b). On the basis of the above results, graphite-loaded carbon felt (namely, CF@GF) with two redox processes (0.83 V vs. SCE and 1.33 V vs. SCE) (Fig. 2c) served as the cathode for further Ti/Br flow battery evaluation. The morphology detection results by scanning electron microscope show that the graphite was coated on the carbon fiber (Fig. S2).

Fig. 2
figure 2

a Cyclic voltammogram of the 50 mmol/L HBr + 1 mol/L H2SO4 electrolytes at a scan rate of 20 mV/s on the graphite and glass carbon electrode, respectively. b Linear fitting curves of 50 mmol/L HBr + 1 mol/L H2SO4 electrolytes on the graphite electrode with peak reduction current versus the square root of the scan rate. c Cyclic voltammogram of the 50 mmol/L HBr + 1 mol/L H2SO4 electrolytes at a scan rate of 0.5 mV/s on the CF@GF electrode. d Charge–discharge plots of the assembled battery using CF and CF@GF as the positive electrode, respectively

Full size image

For the battery tests, 0.5 mol/L HBr + 0.5 mol/L Ti(SO4)2 + 2 mol/L H2SO4 + 0.05 mol/L MEP was used as the electrolyte, where MEP is commonly used in bromine-based FBs to alleviate Br2 diffusion [20, 21], and 2 mol/L H2SO4 was added to the electrolyte to inhibit oxygen evolution at high voltage (Fig. S3). In Fig. 2d, the flow battery assembled with carbon felt exhibited a large polarization at the end of the first charging plateau, rising rapidly to the cutoff voltage (1.6 V), indicating that further oxidation of Br2 to HBrO suffered from serious polarization (Fig. 2d). However, using CF@GF as the electrode, a second reversible charge–discharge process could be observed, which indicated the reversibility of Br intercalation/intercalation. Compared with Br−/Br2, the Br intercalation process demonstrates a 0.5 V higher potential and a 65% improvement in energy density (from the initial 11.3 W·h/L to 18.7 W·h/L). Upon further increasing the cutoff voltage from 1.6 to 1.7 V, the Coulombic efficiency of the battery decreased from 98 to 94%, indicating the possible occurrence of oxygen evolution side reactions (Fig. S4).

Properties of Br–GIC

To confirm Br–GIC formation at the second plateau, in situ Raman spectra were recorded at different SOCs. The CF@GF electrode exhibited two vibrational absorption peaks at 1355 cm−1 and 1585 cm−1, which correspond to the D-band and G-band of carbon, respectively [22]. From the test results, no obvious change occurred below 50% SOC, while the splitting of the G-band could be observed above 50% SOC, which may be ascribed to a structural change after Br intercalation. In addition, a new stretching vibration absorption peak of Br at 245 cm−1 was observed on the electrode above 50% SOC, which further indicated Br–GIC formation (Fig. S5). Since only weak van der Waals forces exist between the graphite layers, Br intercalation would lead to an increase in the interlayer spacing, which was manifested as a shift in the XRD peaks of the CF@GF electrode (Fig. 3b). According to the Bragg diffraction formula, the interlayer spacing of graphite increased from an initial value of 3.35–3.45 Å after 50% SOC. Moreover, Br intercalation in the graphite would inevitably lead to a decrease in Br2 in the solution, which corresponds to a decrease in Br2 absorption intensity (310 cm−1) in the Raman spectra (Fig. S6). These test results suggested that Br–GIC formation occurred at the second charging plateau.

Fig. 3
figure 3

a In situ Raman spectra on the CF@GF electrode at different states of charge (SOCs) of the battery. b X-ray diffraction patterns of the CF@GF electrode at different SOCs. c The charge–discharge plots of the in situ battery

Full size image

To investigate the diffusion behavior of Br in the solution and graphitic phases, GITT was used (Fig. 4a). During the charge–discharge process, the voltage could reach a rapid equilibrium after each current pulse. (ΔEs/ΔEt)2 was used to measure the variation in the diffusion coefficient, as shown in Fig. 4b. The diffusion of Br could be divided into three regions: electrolyte diffusion, transition diffusion, and solid diffusion. The diffusion rate of Br in the electrolyte underwent no obvious change with increasing SOC, while the diffusion rate of Br in the graphite phase gradually decreased with increasing SOC, which could be attributed to the decreasing bromine intercalation sites in graphite. Br underwent an independent transition diffusion from the liquid phase (electrolyte) to the solid phase (graphite), which should be attributed to the structural change from Br2 to Br+. This transitional behavior can also be observed in the self-discharge of the battery, where a small plateau was observed between the first and second discharge plateaus (Fig. S7).

Fig. 4
figure 4

a GITT test of the flow battery using CF@GF as the cathode. b (ΔEs/ΔEt)2 versus the SOC derived from the GITT test. ΔEs is the steady-state voltage change, and ΔEt is the voltage change during the constant current pulse

Full size image

To further investigate the effect of graphite on the charge transfer behavior of the cathode, EIS was evaluated based on a three-electrode system and a full battery, respectively (Fig. 5). From the Nyquist plot of the three-electrode system test in Fig. 5a with an electrolyte composition of 50 mmol/L HBr + 1 mol/L H2SO4, the ohmic impedance of the electrode increased after graphite was loaded, probably because the use of a binder (PVDF) reduced the conductivity of GF. Moreover, the increasing charge transfer resistance of Br−/Br2 was mainly due to the inhibited kinetics of Br2/Br− on the graphite surface [23, 24]. For the full battery tests at different SOCs, the impedance of charge transfer resistance (Br−/Br2) decreased with increasing SOCs below 50%, mainly due to an oxidation state (Br3− or Br5−) that was conducive to electrode surface adsorption and charge transfer. However, the charge transfer resistance of Br2/Br–GIC increased with increasing SOC above 50%, which is mainly due to a gradual decrease in the active sites of graphite as the Br intercalation reaction proceeded, which intensified the charge transfer impedance. In addition, the increased ohmic impedance after 50% SOC may be caused by the reduced conductivity of graphite after Br intercalation.

Fig. 5
figure 5

a The electrochemical impedance spectroscopy (EIS) tests of the 50 mmol/L HBr + 1 mol/L H2SO4 electrolytes on the CF and CF@GF electrodes, respectively. b EIS tests of the full battery at different SOCs

Full size image

Battery Performance

Figure 6a demonstrates the efficiencies of the titanium–bromine flow battery using CF@GF as the cathode. At a current density of 30 mA/cm2, the assembled battery continuously ran for more than 300 cycles with a Coulombic efficiency of more than 97% and an energy efficiency of approximately 80%. As the current density increased from 20 to 80 mA/cm2, the Coulombic efficiency of the battery increased from 96 to 99%, which is mainly due to the limited crossover of Br2. However, the voltage efficiency decreased from 82 to 72%, which is due to the intensified ohmic and electrochemical polarization. Figure 6c demonstrates a trend of capacity decay during a cycle, particularly in the early stage, which is mainly due to the damage to the graphite structure when bromine was extracted from the graphite at high potential. In this regard, the Raman spectra of graphite after cycling show that the D-band intensity increased with the cycling, indicating the increase in defects in graphite. In addition, XRD showed that with the increase in cycling, the diffraction peak of graphite shifted to a low angle, indicating that the interlayer spacing increased (Fig. S8). Therefore, the complete structure of graphite is essential for exploiting the capacity in the second plateau. Upon loading defect-rich graphite onto carbon felt as the cathode, such as fluorinated graphite or FSN-2, the second charge–discharge plateau was not even fully observable (Fig. S9).

Fig. 6
figure 6

a Efficiencies of the flow battery at 30 mA/cm2 using CF@GF as the cathode. The electrolyte was 0.5 mol/L HBr + 0.5 mol/L Ti(SO4)2 + 2 mol/L H2SO4 + 0.05 mol/L MEP. b Efficiencies at different current densities of the flow battery using CF@GF as the cathode. c Capacity retention of the flow battery during cycling. d Raman spectra of the graphite after different cyclings. The graphite was scraped from the CF@GF electrode after battery cycling

Full size image

Conclusion

In summary, a novel two-electron transfer reaction based on Br intercalation was developed and used in a bromine-based flow battery. In strongly acidic electrolytes, Br+ can intercalate into graphite to form Br–GIC with a redox potential of 0.5 V vs. Br−/Br2, and the extent of the reaction is closely related to the integrity of the graphite structure. Different from that of Br−/Br2 in an electrolyte, the diffusion rate of Br intercalation is suppressed as the SOC increases. As a result, using graphite-loaded carbon felt as an electrode, the charge–discharge curve demonstrates two plateaus, which achieve 65% higher energy density compared with Br−/Br2. Most importantly, the assembled battery can continuously run for more than 300 cycles at 30 mA/cm2. Combined with a two-electron transfer property, excellent kinetics, and much higher redox potential, the Br intercalation/deintercalation redox couple is crucial for improving the energy density of aqueous batteries with promising application prospects.