Zinc–bromine hybrid flow battery: effect of zinc utilization and performance characteristics

Abstract

In order to achieve maximum efficiency and long lifetime of a zinc–bromine flow battery (ZBB), the deposition and dissolution of zinc during the charging and discharging processes, respectively, need to be in balance. In view of this, the percentage utilization of zinc during the discharge process was investigated in a zinc–bromine redox flow cell through a potentio/galvanodynamic polarization test and electrochemical impedance spectroscopy. The cell employed carbon–plastic composite electrodes and 98% pure zinc bromide electrolyte solution. The zinc–bromine cells were charged at various current densities of 10, 20 and 30 mA cm⁻², and the deposited Zn during charging was compared with the dissolved zinc during the discharge process and found to be 39, 41 and 39%, respectively. A 10% increase in the zinc utilization factor was observed, and this resulted in a 17% increase in the Faradaic efficiency when 99.9% pure zinc bromide electrolyte solution was used. At the same time, 50% utilization of zinc resulted in a further 20% increase in Faradaic efficiency in a cell with non-porous graphite electrodes and 98% pure zinc bromide salt solution. To qualify the nature of the Zn deposit, XRD analysis was carried out along with other spectral studies. The basal plane of zinc (002) was observed at a lower intensity peak, whereas the plane (101) showed preferential growth in the (101) direction. From these spectra, grain size and texture coefficient (TC) were also calculated in order to realize better Zn utilization in a zinc–bromine hybrid flow cell.

---

Introduction

Redox flow batteries are known to be the most compatible with grid energy storage systems. In general, redox flow batteries theoretically have no performance loss because they rely on the electrochemical reaction between electrolytes and not between the electrodes and electrolyte. Furthermore, one of the major advantages of this system is its 100% discharge capacity and it can be left in a completely discharged state without damage, which indicates that its rated capacity is its real capacity. Another advantage is decoupled energy and power, as with fuel cells.¹ The technical data of one of the promising system, i.e. a zinc–bromine redox flow battery, is 75–80% cycle efficiency, 80–85 W h kg⁻¹ energy density and a unit design lifetime of 11–14 years.² This shows promise for use as an energy storage system for onsite power generation.

In spite of these favorable features of the zinc–bromine redox flow battery, systematic studies of different aspects of its electrochemistry are scarce. Moreover, more studies were carried out on bromine electrochemistry and its utilization^3,4 than on zinc because of the fast reaction kinetics of zinc. Regardless of the fast reaction kinetics of zinc when compared to the bromine reaction, its reversibility is open to discussion, which is also considered to be one of the deciding factors for the overall cell performance characteristics. Hence, in the present study, we have focused on the zinc electrochemical deposition/dissolution reaction in a zinc–bromine redox flow cell to realize improvements in the cell performance. First, we have conducted simple quantitative analyses of zinc deposition and dissolution in terms of capacity. Second, we explored the effect of substrate material and addition of complexing agents on the zinc deposition/dissolution reaction with respect to improvements in the cell performance.

Experimental

The hardware used in this work was a frameless carbon–plastic composite electrode with a geometric area of 50 cm² on both sides. A 500 μm thick activated carbon active layer coating was applied on the bromine electrode. A microporous polyethylene membrane (Daramic) with thickness 0.6 mm was used as separator. The temperature was maintained at 25 °C. The zinc electrolyte used in the work was 2 M zinc bromide (Alfa Aesar, 98% and 99.9% purity) in 3 M NaCl (Alfa Aesar) as supporting electrolyte in order to eliminate ionic migration (i.e., no potential drop in the solution) and 0.8 M N-ethyl-N-methylpyrrolidinium bromide as complexing agent. The quantity of electrolyte used for circulation was 250 ml for each side. The electrolyte flow rate was 20 ml min⁻¹ on both sides. Peristaltic pumps were used for electrolyte circulation. A test cell was assembled with the respective components as shown in Fig. 1.

Fig. 1 Single cell assembly setup.

Charging was carried out at a constant current density using Aplab testronic model 9313, India and discharging through load box model KM 0121, India. The charging time was 15 min and the cut-off voltage during discharge at the same current density was 0.5 V. Characterization studies were carried out using a powder X-ray diffractometer Philips XRD ‘X’ PERT PRO with Cu Kα (α = 1.5418 Å) as source at a scan rate of 5° min⁻¹. The surface morphology of the deposits was examined by scanning electron microscopy (SEM, Vega-3 TESCAN). Electrochemical impedance spectroscopy (EIS) (10⁻² Hz to 10⁵ Hz, amplitude = 5 mV) and cyclic voltammetry studies were carried out with a VersaSTAT 3 instrument (Princeton Applied Research, USA). The cycle life tests were performed in a cell controlled with a BT-2000 workstation (Arbin instruments, USA). The XRD intensities of (hkl) planes were used to calculate the absolute texture coefficient (ATC) values through the following procedure.⁵

The intensities of the seven peaks were used to calculate the intensity ratio.

By using these ratio values, texture coefficients for all seven planes were calculated through the equation TC_(hkl) = R_(hkl)/R_{(std hkl)} and ATC through the relation

Results and discussion

In order to calculate the amounts of zinc deposited during the charging and dissolved during the discharging process, the test cell was dismantled and the cathode was removed as soon as possible after the 15 min charging process. After careful washing in triply distilled water and drying in a vacuum oven at 60 °C for 10 min, the cathode containing the zinc deposit was weighed. Thus, the calculated amounts of deposited zinc at current densities 10, 20 and 30 mA cm⁻², respectively, were 0.136, 0.264 and 0.303 g. The amounts of zinc dissolved at the same load current density values until the cut-off voltage of 0.5 V were 0.054, 0.117 and 0.118 g, respectively, which in terms of weight percentages amount to 39, 41 and 39. The interesting observation of these results is that almost 60 wt% of the deposited zinc remained unutilized irrespective of the load currents. This may be due to concomitant utilization of bromine at the anode or other reasons. Assuming that bromine is utilized concomitantly and considering the experimental results obtained at 10 mA cm⁻² current density, the amount of deposited zinc for 15 min charging time was 0.136 g. The theoretically calculated amount of zinc deposition for the same current density and time was 0.153 g. The difference being 0.017 g (11%) of zinc, which is ascribed to the reason that bromine crossover during the charging process favors the self-discharge chemical reaction, namely, Zn + Br₂ ↔ ZnBr₂. There are two points that have to be considered before making conclusions: (1) the flow pattern of the reactive species and (2) the presence of a complexing agent in both catholyte and anolyte. In the present study, the flow pattern adopted was front-flow; hence, bromine diffusion would not be as large as with a back-flow pattern during the charging process⁶ and the small amount of bromine crossing over will be complexed by the complexing agent in the catholyte. This was substantiated when the above result was compared with the 0.009 g of zinc (6%) that was obtained by using 99.9% pure zinc bromide salt solution.

In a single charge–discharge cycle at a constant current density of 10 mA cm⁻² until the cut-off voltage 0.5 V, the amounts of zinc utilized were 0.054 g and 0.072 g and the cut-off voltage was reached within 6.5 and 9 min of the discharge time for 98% and 99.9% pure zinc bromide salt solutions, respectively, as shown in Fig. 2a and b. The Faradaic efficiency for the 99.9% pure zinc bromide salt solution was approximately 1.4 times higher than that for the 98% pure solution. This is because of utilization of zinc from the 99.9% pure zinc bromide salt solution, as calculated from the weight result, which was 10 wt% higher (49.96 wt%) than that for 98% purity electrolyte (39.79 wt%).

Fig. 2 Discharge curves at a constant current density of 10 mA cm⁻²; cell with carbon–plastic composite electrodes and 98% purity zinc bromide electrolyte (a), cell with carbon–plastic composite electrodes and 99.9% purity zinc bromide electrolyte (b), cell with non-porous graphite plate electrodes and 98% purity zinc bromide electrolyte (c).

As shown in Fig. 2a and b, the slopes of the curves in the internal resistance (IR) region for both the electrolytes are the same, i.e. 0.023, whereas in the diffusion-limiting region of the 98% purity electrolyte the slope is 0.66 times greater than that of the 99.9% purity electrolyte. From the above observations, it is concluded that the internal resistance is not affected by impurities present in the electrolyte but the zinc deposition/dissolution processes are affected.

We must admit that no precise conclusion can be drawn from the above discussion until more specific experimental characterization has been carried out; hence, EIS studies were performed. Accordingly, the electrodes were charged galvanostatically at 20 mA to make 20% of the theoretical capacity of the zinc active material. The impedance measurements were made at the same discharge current. The impedance data as shown in Fig. 3 reveal that the middle-frequency semicircle is related to charge transfer and the electrochemical double layer, representing the kinetics of the zinc dissolution reaction and a typical Warburg semi-infinite linear diffusion characteristic at the low-frequency part.

Fig. 3 Nyquist plot of Zn–Br₂ flow cell containing 98% (a) and 99.9% (b) pure ZnBr₂ in graphite and composite electrodes.

In the case of pure electrolyte, the values of charge transfer resistance decrease due to the increase in the amount of zinc metal, while the value of capacitance increases. On the other hand, in the case of impure electrolyte, the value of charge transfer resistance increases due to a decrease in the amount of zinc metal, while the value of double-layer capacitance decreases. Fig. 3(a) shows an electrode material (composite/graphite)-independent, high-frequency response with 98% pure electrolyte, indicating that grains are less sensitive to fast switching of an applied alternating electric field⁷ due to an impurity effect. On the other hand, Fig. 3(b) exhibits an electrode material-dependent, high-frequency response with 99.9% pure electrolyte, due to less grain boundary contribution. For both composite and non-porous graphite electrodes in 98% pure electrolyte, the low-frequency spur as shown in Fig. 3(a) becomes larger, indicating an increased grain boundary effect caused by more aggregation.⁸

A cycle test of the test cell with a carbon–plastic composite electrode and two different purity electrolytes was performed to investigate the variation of Faradaic efficiency versus cycle number, as shown in Fig. 4. As the number of cycles increases, the discharge time also increases as seen in Fig. 4(a). The difference in Faradaic efficiency between the first and sixth cycle for 99.9% pure electrolyte is ∼10%, whereas for 98% pure electrolyte it is ∼16% (Fig. 4(b)). This can be ascribed to the better utilization of zinc in the absence of impurities during the charge–discharge cycle. In general, the deposition/dissolution process is hindered by the impurities present in the 98% pure electrolyte; hence, the generated ions are smaller in number during the dissolution process, whereas in the 99.9% pure electrolyte, more ions are generated due to the absence of hindrance from impurities, leading to rapid dissolution, which is confirmed by lower R_ct values.

Fig. 4 Charge–discharge curves of Zn–Br₂ flow cell containing 99.9% pure ZnBr₂ as electrolyte (a) and Faradaic efficiency as a function of cycle number for both 98% and 99.9% pure ZnBr₂ (b).

Fig. 5a and b show the X-ray diffraction patterns of the zinc electrodeposits obtained from cells with 99.9% and 98% pure zinc bromide salt solution, respectively. The sharpness of the diffraction peaks indicates zinc crystallization and it exhibits a hexagonal structure. The observed ‘d’ value is in good agreement with the standard values for zinc deposition (ASTM File no.1:*40831Zn). Note that no impurity peak was observed. This showed that the reflection from the (101) plane was more predominant compared to other peaks with an intensity ratio (I₀₀₂/I₁₀₁) of 0.33 : 1, approximating to the ratio for randomly oriented powder particles,⁹ which is almost the same for both electrolytes. We have chosen the 101 line to calculate the grain size as it is prominent in all the patterns. The average crystallite size of the zinc deposit estimated from the FWHM of the (101) plane using Scherrer's formula is 124 and 103 nm for 98% and 99.9% pure zinc bromide salt solutions, respectively. Fig. 6 shows the variation in ATC of the planes (002), (100), (101), (102), (103), (110) and (004) for the Zn deposit obtained from both zinc bromide salt solutions. The higher ATC values of the planes in 99.9% pure zinc bromide salt solution with a prominent (004) plane are indicative of the better deposition and higher crystalline nature of the zinc deposit grown. Fig. 7a and b show the surface morphology of the zinc deposits obtained from both electrolyte solutions. It is evident that the surface morphology of the zinc deposit from 98% pure (Fig. 7b) zinc bromide salt solution predominantly contains a leaf-like structure. Even though the difference in morphology is evident, it is difficult to exactly explain the gain in the Faradaic efficiency other than via the role of impurities in the deposition/dissolution process of zinc.¹⁰

Fig. 5 XRD patterns for Zn deposit obtained from the cell made with carbon–plastic composite electrodes and 98% purity zinc bromide electrolyte (a) and 99.9% purity zinc bromide electrolyte (b).

Fig. 6 Absolute texture coefficient values of the Zn crystallographic planes for deposits obtained from a cell with carbon–plastic composite electrodes from 98% and 99.9% purity zinc bromide electrolyte.

Fig. 7 SEM images of the zinc deposit from a cell with carbon–plastic composite electrodes and 98% purity zinc bromide electrolyte (a) and 99.9% purity zinc bromide electrolyte (b).

The observed impurities through EDAX analysis (inset figure) are aluminum, copper and iron. The wt% of these impurities in 98% purity electrolyte is 0.64, 1.11 and 0.39, respectively, and in 99.9% purity electrolyte is 0.00, 0.89 and 0.26, respectively. The presence of aluminum in 98% purity electrolyte appears to hinder the zinc deposition/dissolution process.¹¹ The effect of the impurity is thus seen to be important when using carbon–plastic composite plates as electrode and future study will investigate the effect of other impurities using cyclic voltammetry studies. The absence of XRD peaks due to Al and other impurities indicates that the impurities might be segregated in the non-crystalline region in the grain boundary.¹²

Fig. 8 shows a typical cyclic voltammogram, at a scan rate of 5 mV s⁻¹, corresponding to zinc deposition on a carbon composite electrode from a solution containing 2 M zinc bromide in 3 M NaCl as supporting electrolyte and 0.8 M N-ethyl-N-methylpyrrolidinium bromide as complexing agent. No peak is observed during the cathodic scan but two crossovers at −1.091 V and −1.025 V are observed owing to the formation of a new phase involving a nucleation process and equilibrium potential of the Zn²⁺/Zn redox couple, respectively,¹³ along with an anodic peak at −0.819 V. The observed crossover peak values fit well with the observed results on glassy carbon electrode,¹⁴ while the anodic peak is shifted by 100 mV towards the positive side. The nonappearance of the cathodic peak is due to the effect of pH, as described elsewhere.¹⁵

Fig. 8 Cyclic voltammogram obtained from carbon–plastic composite electrode with 2 M zinc bromide in 3 M NaCl as supporting electrolyte and 0.8 M N-ethyl-N-methylpyrrolidinium bromide as complexing agent at a potential scan rate of 5 mV s⁻¹.

When nonporous graphite electrodes are used in the cell with 98% pure zinc bromide salt solution, the cell exhibited twice the Faradaic efficiency when compared with the cell using a carbon–plastic composite electrode and 99.9% pure zinc bromide salt solution, as shown in Fig. 2c. The amount of reacted zinc is 0.111 g in 11 min, which is almost equal to the theoretically calculated value of 0.112 g for the same time period, 99.9 wt% of zinc utilization. From the slopes of the curves in the IR (0.013) and diffusion-controlled region (0.51239), as shown in Fig. 2c, it can be ascertained that the electrode material has more effect in the IR region (lower slope than carbon–plastic composite electrode) than in the diffusion controlled region (same slope). From the above observations, one can conclude that the difference in internal resistance and Faradic efficiency is due to the difference in resistivity of the electrode materials and the active surface area difference between the non-porous graphite and carbon–plastic composite electrode materials,¹⁶ respectively.

Fig. 9 shows X-ray diffraction patterns of the zinc electrodeposits obtained from a cell with non-porous graphite electrodes and 98% pure zinc bromide salt solution. All peaks are accounted for with zinc metal as was observed for a carbon–plastic composite electrode. The predominant plane is again (101) and the average crystalline size of the zinc deposit estimated from FWHM is 103 nm, as was observed for a carbon–plastic composite electrode with 99.9% pure electrolyte. Even though the observed intensity of the basal and pyramidal planes is suppressed, the intensity ratio (I₀₀₂/I₁₀₁) is 0.32 : 1, which is almost the same as that obtained for carbon–plastic composite electrodes. The order of reduction in ATC values of the planes to that of the composite electrode is (002), (101), (100), (102), (103), (110) and (004), as shown in Fig. 10. Note that the preferential growth is along the (004) plane. The surface morphology is as shown in Fig. 11a and b. Zinc deposition on non-porous graphite produced a more compact deposit with a flake- and leaf-like structure. Thus, the change of substrate did not change the morphology but affected its compactness. The effect of impurity predominates when the electronic conductivity and active surface of the substrate are low, as with carbon composite electrode, whereas this is not appreciable when substrate electronic conductivity and active surface are as high as those of a non-porous graphite electrode. Apart from this fact, the compactness of the deposit also has an impact by improving total conductivity from that of the loose structure.

Fig. 9 XRD pattern obtained for Zn deposit obtained from cell with nonporous graphite electrodes and 98% pure zinc bromide electrolyte.

Fig. 10 Absolute texture coefficient values of the Zn crystallographic planes for deposit obtained from cell with nonporous electrodes and 98% pure zinc bromide electrolyte.

Fig. 11 SEM pictures of Zn deposit obtained from cell with nonporous electrodes and 98% purity zinc bromide electrolyte (a) and Zn deposit obtained from cell with composite carbon electrodes and 98% purity zinc bromide electrolyte (b).

The ATC values of zinc crystallographic planes on a non-porous graphite substrate after the charge and discharge processes are shown in Fig. 12. The difference in ATC values of the zinc deposit after the charge and discharge processes are in the order (101), (002), (102), (100), (103), (004) and (110), whereas the order of difference in ATC values over the carbon composite electrode is (103), (002), (101), (102), (100), (110) and (004). On both substrate materials the preferential orientation of the crystallites after discharge is (004). This observation is due to the presence of graphite materials as a filler to form a continuous conductive network through the composite plate.¹⁷ Similarly, on both substrate materials the basal plane (002) took second place in the order, which indicates that the preferential orientation of the covered surface is (004) basal plane parallel to the substrate surface with highest binding energy. Hence, the total energy involved in the breaking of bonds for dissolution is higher and because of that the substrate is electrochemically less reactive.¹⁸ Conversely, with reference to the (002) plane, there must be some crystallites present in the deposited zinc surfaces that are oriented at angles apart from that parallel to the substrate plane resulting in electrochemical reaction.

Fig. 12 Absolute texture coefficient values of the zinc crystallographic planes for deposit obtained from cells with carbon composite electrode (a) and non-porous graphite electrode (b) with 98% pure electrolyte solution after charge–discharge process.

Conclusion

The effect of zinc utilization and the performance characteristics in a Zn–Br₂ hybrid flow battery have been reported. It was observed that simultaneously changing the quality of the electrolyte and the electrode substrate material resulted in an increase in the surface coverage with randomly oriented crystallographic (101) planes and, as a result, the cell exhibited higher Faradaic efficiency. The enhanced ATC values of the planes in 99.9% pure zinc bromide salt solution, with a prominent (004) plane, are evident due to the better deposition and higher crystalline nature of the elemental zinc. Furthermore, EIS study indicates that a 98% pure electrolyte favors the grain boundary effect more than a 99.9% pure electrolyte. Changing the substrate material to graphite plate increases the Faradaic efficiency even without changing the quality of the electrolyte, resulting in 99.9 wt% zinc utilization.

Acknowledgements

Authors Arulraj, Dheenadayalan and Ragupathy acknowledge CSIR-TAPSUN project (NWP 56) for providing financial support. Dr Vijayamohanan K. Pillai is acknowledged for his constant encouragement and support.

References

1. S. Kauff and K. Bolwin, J. Power Sources, 1992, 38, 303–315.
2. H. Chen, T. N. Cong, W. Yang, C. Tan, Y. Li and Y. Ding, Prog. Nat. Sci., 2009, 19, 291–312.
3. K. T. Cho, P. Ridgway, A. Z. Weber, S. Haussener, V. Battaglia and V. Srinivasan, J. Electrochem. Soc., 2011, 159, A1806–A1815.
4. E. Lancry, B. Z. Magnes, I. Ben-David and M. Freiberg, ECS Trans., 2013, 53, 107–115.
5. R. Ramanauskas, P. Quintana, L. Maldonado, R. Pomés and M. A. Pech-Canul, Surf. Coat. Technol., 1997, 92, 16–21.
6. S. Ravindra Vaman, Investigation of transport processes and electrode kinetics in a zinc bromine battery, Doctoral dissertations, University of Connecticut, 1991, Paper AAI9215396.
7. D. He, M. Bligh and T. David Waite, Environ. Sci. Technol., 2013, 47, 9148–9156.
8. N. Hirose and A. R. West, J. Am. Ceram. Soc., 1996, 79, 1633–1641.
9. L. M. A. Monzon, L. Klodt and J. M. D. Coey, J. Phys. Chem. C, 2012, 116, 18308–18317.
10. H. Nakano, S. Oue, K. Yanai, H. Fukushima, K. Nishina and N. Sogabe, J. MMIJ, 2011, 127, 267–271.
11. F. W. Harris, 57th General Meeting of the American Electrochemical Society, St. Louis, Mo., 1930.
12. J. H. Lee and B. O. Park, Thin Solid Films, 2003, 426, 94–98.
13. R. Greef, R. Peat, L. Peter, D. Pletcher and J. Robinson, Instrumental Methods in Electrochemistry, Ellis Horwood, Chichester, 1985, ch. 9, pp. 283–316.
14. L. H. Mendoza-Huízar, C. H. Rios-Reyes and M. G. Gómez-Villegas, J. Mex. Chem. Soc., 2009, 53, 243–247.
15. L. Zhang, Q. Lai, J. Zhang and H. Zhang, ChemSusChem, 2012, 5, 867–869.
16. B. C. H. Steele and A. Heinzel, Nature, 2001, 414, 345–352.
17. R. Yeetsorn, M. W. Fowler and C. Tzoganakis, Nanocomposites with Unique Properties and Applications in Medicine and Industry, ed. John Cuppoletti, 2011, ch. 16, pp. 317–344.
18. J. C. Scully, The Fundamentals of Corrosion, Pergamon Press, Oxford, 2nd edn, 1990, pp. 1–226.

---