Effect of Fe(III) on the positive electrolyte for vanadium redox flow battery

It is important to study the effect of Fe(III) on the positive electrolyte, in order to provide some practical guidance for the preparation and use of vanadium electrolyte. The effect of Fe(III) on the thermal stability and electrochemical behaviour of the positive electrolyte for the vanadium redox flow battery (VRFB) was investigated. When the Fe(III) concentration was above 0.0196 mol l−1, the thermal stability of V(V) electrolyte was impaired, the diffusion coefficient of V(IV) species decreased from (2.06–3.33) × 10−6 cm2 s−1 to (1.78–2.88) × 10−6 cm2 s−1, and the positive electrolyte exhibited a higher electrolyte resistance and a charge transfer resistance. Furthermore, Fe(III) could result in the side reaction and capacity fading, which would have a detrimental effect on battery application. With the increase of Fe(III), the collision probability of vanadium ions with Fe(III) and the competition with the redox reaction was aggravated, which would interfere with the electrode reaction, the diffusion of vanadium ions and the performance of VRFB. Therefore, this study provides some practical guidance that it is best to bring the impurity of Fe(III) below 0.0196 mol l−1 during the preparation and use of vanadium electrolyte.


MD, 0000-0002-6524-2743
It is important to study the effect of Fe(III) on the positive electrolyte, in order to provide some practical guidance for the preparation and use of vanadium electrolyte. The effect of Fe(III) on the thermal stability and electrochemical behaviour of the positive electrolyte for the vanadium redox flow battery (VRFB) was investigated. When the Fe(III) concentration was above 0.0196 mol l 21 , the thermal stability of V(V) electrolyte was impaired, the diffusion coefficient of V(IV) species decreased from (2.06-3.33) Â 10 26 cm 2 s 21 to (1.78-2.88) Â 10 26 cm 2 s 21 , and the positive electrolyte exhibited a higher electrolyte resistance and a charge transfer resistance. Furthermore, Fe(III) could result in the side reaction and capacity fading, which would have a detrimental effect on battery application. With the increase of Fe(III), the collision probability of vanadium ions with Fe(III) and the competition with the redox reaction was aggravated, which would interfere with the electrode reaction, the diffusion of vanadium ions and the performance of VRFB. Therefore, this study provides some practical guidance that it is best to bring the impurity of Fe(III) below 0.0196 mol l 21 during the preparation and use of vanadium electrolyte.

Experimental set-up 2.1. Preparation of electrolyte
The electrolyte with 1.6 mol l 21 VOSO 4 and 2.8 mol l 21 H 2 SO 4 presents the optimal electrochemical performance [12]. V(IV) electrolyte (1.6 M V(IV) þ 2.8 M H 2 SO 4 ) was prepared by dissolving VOSO 4 in concentrated sulfuric acid. Then V(V) electrolyte (1.6 M V(V) þ 2.8 M H 2 SO 4 ) can be obtained in a twocompartment electrolysis cell which used V(IV) electrolyte as an anolyte and H 2 SO 4 solution in the same concentration as the catholyte. The vanadium concentration in the electrolyte was analysed by the potentiometric redox titration method. The procedure of the redox titration method is to add the sulfur and phosphorus mixed acid, then add potassium permanganate solution to the electrolyte until the solution becomes purple, and finally, conduct a titration using a potentiometric titrator with ferrous ammonium sulfate standard solution until the potential of electrolyte changes; the vanadium concentration can be calculated through the titration volume and concentration of the standard solution [23]. A certain concentration (0.00893 mol l 21 , 0.0161 mol l 21 , 0.0196 mol l 21 , 0.0232 mol l 21  ignored [13].) The total iron concentration in the electrolytes was determined by ICP-AES. The ferrous iron concentration in the electrolytes was measured using the 1,10-phenanthroline spectrophotometry [24]. The ferric iron concentration in the electrolytes was calculated by mass balance. All the chemical reagents used in the experiments were analytically pure and the solutions were prepared with deionized water.

Thermal stability experiment
During the stability tests, the electrolyte with different concentrations of Fe(III) was statically stored in a temperature-controlled bath at different temperatures (258C, 408C and 508C). The samples were monitored regularly for the formation of precipitation and the time when a slight precipitation appeared was recorded. The solutions were filtered at regular intervals and titrated to determine the changes in vanadium concentration.

Electrochemical measurements
Cyclic voltammetry (CV), electrochemical impedance spectroscopy (EIS) of the V(IV) electrolyte with different amounts of Fe(III) and Linear sweep voltammetry (LSV) of 2.8 M H 2 SO 4 solution with different concentrations of Fe(III) were performed using the CHI660 electrochemical workstation (Shanghai Chenhua Instrument Co., Ltd., China). The curves of current versus potential were recorded in a threeelectrode electrochemical cell with a platinum sheet electrode (2 cm 2 ) as the counter electrode, an aqueous saturated calomel electrode (SCE) along with a salt bridge full of saturated potassium chloride solution as the reference electrode and a graphite plate (1 cm 2 ) as the working electrode. Prior to the test, the working electrode was manually polished with SiC grit paper and then washed by ultrasonic cleaning in ethanol and secondly in distilled water for 10 min, respectively. The sinusoidal excitation voltage applied to the cells in the EIS test was 5 mV with a frequency range between 0.01 Hz and 100 kHz.

Charge -discharge tests
The charge -discharge tests were performed in a VRFB single dynamic cell, which was assembled by employing two pieces of PAN-based graphite felt (25 cm 2 ) as an electrode, Nafion 117 ion exchange membrane as a separator and two copper plates as current collectors. 30 ml 1.6 M V(IV) in 2.8 M H 2 SO 4 served as a negative electrolyte and 30 ml 1.6 M V(III) in 2.8 M H 2 SO 4 as a positive electrolyte were cyclically pumped into the corresponding half-cell, by two peristaltic pumps with a flow rate of 10 ml min 21 . Before the experiment, the Nafion 117 ion-exchange membrane was soaked in the 3% hydrogen peroxide solution for 1 h, then soaked in 1 mol l 21 H 2 SO 4 for 30 min at 808C and rinsed with deionized water. Each piece of graphite felt was soaked in an ethanol water solution (70%) under ultrasonication for 30 min to remove impurities in the fibres, then rinsed with deionized water and dried in an oven at 508C for 10 h. The galvanostatic charge -discharge test of the single cell was carried out using CT2001B-5 V/10A (Wuhan Land Co., China) between 0.65 and 1.65 V at a current density of 40 mA cm 22 .

Results and discussion
3.1. Effect of Fe(III) on thermal stability of V(V) As shown in table 1, the thermal stability of the V(V) electrolyte decreased greatly when the temperature increased from 258C to 508C due to the endothermic nature of the precipitation reaction of V(V) ions (1) [25]: addition was more than 0.0196 mol l 21 , as shown in figure 1. The concentration of the V(V) ion decreased gradually with the prolonged heating time, and the concentration of remaining V(V) ions in the electrolyte with 0.0232 mol l 21 Fe(III) impurity was 0.1 mol l 21 lower than that of the pristine electrolyte between 10 and 22 h, indicating that when the Fe(III) impurity content was above 0.0196 mol l 21 it significantly impaired the thermal stability of the V(V) electrolyte. As can be seen, the electrochemical activity of the V(V)/V(IV) redox couple with different concentrations of Fe(III) was changed. As the Fe(III) concentration increased from 0.00893 mol l 21 to 0.0286 mol l 21 , the DE p were clearly increased and the J pa /j pc first decreased and then increased. When the Fe(III) concentration was 0.0196 mol l 21 , the DE p was smaller and the J pa /j pc was closer to 1 compared with the other samples, which improved the reversibility of the V(V)/V(IV) redox couple to some extent [26 -28].

Cyclic voltammetry
On further study on the CV curves, a secondary oxidation peak was found at 0.4-0.6 V, as shown in figure 2. The peak currents were found to increase with the increased concentration of Fe(III). This is probably attributed to the Fe(III)/Fe(II) couple reaction. Figure      To further investigate the effect of Fe(III) on the kinetics of electrode reaction, a series of CV curves for test electrolyte, containing different concentrations of Fe(III) on the graphite electrode at various scan rates, are shown in figure 5. It presents the typical characteristics of a quasi-reversible one-electron process for the anodic and cathodic peak potentials that change gradually with the scanning rates. A plot of redox peak currents as a linear function of the square root of scan rates with different concentrations of Fe(III) further verified the quasi-reversible process for the V(V)/V(IV) redox reaction, as shown in figure 6.
Theoretically, the value of diffusion coefficient for a quasi-reversible reaction (D) is between that for a reversible (D 1 ) one and for an irreversible (D 2 ) one. For a reversible and irreversible one-step and oneelectron reaction, the peak current i p is given in equation (3.2) and equation (3.3), respectively [25][26][27][28]: For a reversible reaction: For an irreversible reaction:  royalsocietypublishing.org/journal/rsos R. Soc. open sci. 6: 181309 region represents the charge transfer process and the straight line in the low-frequency region represents the diffusion limited process [18,25,26]. Consequently, an equivalent circuit model [18] is proposed in figure 7, where R s , R i , R f , R ct and W are the solution resistance, the interface resistance, the electrolyte film resistance on the electrode surface, the charge transfer resistance and the Warburg diffusion impedance in the electrochemical process, respectively. CPE, the constant phase element, represents the electric double-layer capacitance of the electrode -solution interface. The simulation results obtained by the ZSimpWin software from fitting the impedance plots with the equivalent circuit model in figure 7 are shown in table 4.

Electrochemical impedance spectroscopy
As observed in figure 7, the semicircle in the high-frequency region had a significant change compared with the Nyquist plots of the pristine electrolyte. This indicated that the Fe(III) has a great influence on the charge transfer process. Table 4 shows that there were significant differences in R s , R i , R f , R ct for each sample and the electrolyte was above 0.0196 mol l 21 Fe(III) it exhibited a higher   electrochemical reaction and the vanadium ion diffusion became difficult due to the increased electrolyte film resistance and interface resistance resulting from too large amounts of Fe(III). In brief, below 0.0196 mol l 21 , Fe(III) had positive effects on the charge transfer reaction, and the vanadium ion diffusivity, which could promote the electrode's reaction.

Charge -discharge test
The charge-discharge experiments were performed in a battery using a positive electrolyte with different concentrations of Fe(III) to investigate the changes of performance for the VRFB. Figure 8 shows the charge-discharge curves of VRFB with different concentrations of Fe(III) at the discharge current density of 40 mA cm 22 . VRFB without Fe(III) presents the capacity of 0.99 Ah, which is larger than that (0.92 Ah) with the Fe(III) concentration of 0.0196 mol l 21 and that (0.93 Ah) with the Fe(III) concentration of 0.0286 mol l 21 . The reason for the difference might be that Fe(III) is competitive with vanadium ions for the adsorption on the electrode surface and redox reaction, which results in the side reaction and capacity fading. Figure 9 shows the coulombic efficiency (CE), voltage efficiency  Table 4. Parameters resulting from fitting the impedance plots with the equivalent circuit model.

Mechanism discussion
In view of the results discussed above, it was shown that Fe(III) has an influence on the vanadium ion diffusion, the electrode interface behaviour, the charge transfer reaction, the electrolyte impedance and the battery performance. The electrode process of V(V)/V(IV) redox reaction mainly includes [29][30][31][32]: The C-OH functional groups behave as active sites for the oxidation of VO 2þ on the surface of the electrode. The first step in the reaction involves an ion exchange process between VO 2þ ions transported from the bulk of the catholyte and the H þ ions of the phenolic functional groups on the carbon surface. Next, an oxygen atom is transferred from the C-O functional group to the VO 2þ ions to form VO 2 þ on the surface of the electrode, while an electron is transferred from VO 2þ to the electrode along the C-O-V bond. In the final step, the oxidation reaction is terminated by an ion exchange between the VO 2 þ formed on the electrode surface and the H þ ion in the electrolyte. During the discharge, the redox reactions occur in the opposite directions. Therefore, the possible reasons for the influence of Fe(III) on the electrode process of V(V)/V(IV) redox reaction are as follows, as illustrated in figure 10.

Collision and competition
The collision probability of vanadium ions with Fe(III) increases due to the larger radius of Fe 3þ (0.63 Å ) [33] compared to that of V 4þ (0.46 Å ), V 5þ (0.355 Å ) [34]. Thus, the diffusion resistance of vanadium ions increased when the concentration of Fe(III) was above 0.0196 mol l 21 in anolyte. Furthermore, the valence of Fe(III) is lower than that of vanadium ions; with the Fe(III) addition, the electrode surface status changes slightly and Fe(III) is competitive with vanadium ions for the adsorption on the electrode surface and redox reaction, which results in the charge imbalance on the electrode surface and alters the electric double-layer at the interface. Therefore, the electrolyte film resistance and the interface resistance increased with the addition of Fe(III). The increase of the collision probability and the competition charge-discharge can interfere with the electrode reaction, the diffusion of vanadium ions and the performance of VRFB. When the electrolyte has a lower concentration (below 0.0196 mol l 21 ) of Fe(III), the traces of Fe(III) may facilitate dispersion of vanadium ions for the synergy of coulombic repulsion and steric hindrance [25,26] and may interfere in the association of vanadium ions to a certain degree, which results in the corresponding improvement of electrode reaction activity.

Conclusion
The thermal stability tests showed that the Fe(III) impurity content above 0.0196 mol l 21 significantly impaired the thermal stability of the V(V) electrolyte. The CV measurements indicated that better reaction kinetics were achieved by adding 0.0196 mol l 21 Fe(III), resulting in the improvement of the V(IV) diffusion coefficient, and better reversibility of the electrode reaction compared with the pristine electrolyte. The EIS tests showed that positive electrolyte exhibited a higher electrolyte resistance and charge transfer resistance when the Fe(III) impurity concentration was above 0.0196 mol l 21 . The charge-discharge tests showed that positive electrolyte with Fe(III) has little effect on the efficiency of VRFB, but Fe(III) could result in the side reaction and capacity fading, which will have a detrimental effect on battery performance.
Moreover, when the concentration of Fe(III) was above 0.0196 mol l 21 , the increase in the collision probability of vanadium ions with Fe(III) and Fe(III) is competitive with vanadium ions for the adsorption on the electrode surface and redox reaction can interfere with the electrode reaction, the diffusion of vanadium ions and the VRFB performance.
As discussed above, understanding the influence of Fe(III) on the positive electrolyte is important, and it is best to control the concentration of Fe(III) impurity below 0.0196 mol l 21 during the preparation and use of vanadium electrolyte.
Ethics. Research ethics: We were not required to complete an ethical assessment prior to conducting our research. Animal ethics: We were not required to complete an ethical assessment prior to conducting our research.
Permission to carry out fieldwork. No permissions were required prior to conducting our research. Data accessibility. The datasets supporting this article have been uploaded as part of the electronic supplementary material.