Molybdenum-assisted reduction of VO2+ for low cost electrolytes of vanadium redox flow batteries


 Efficient and affordable energy storage systems are indispensable to accomplish a successful energy transition from fossil fuels to renewable sources. Although all-vanadium redox flow batteries (VRFB) possess many distinctive advantages, much improvement in the process for electrolyte preparation is needed to overcome low productivity and complexity of the current electrolysis process. Herein, we demonstrate a simple one-pot process for the preparation of V3.5+ electrolytes from V2O5 by utilizing hydrazine monohydrate as a residue-free reducing agent and molybdenum as a homogeneous catalyst accelerating the reduction of VO2+. It is confirmed that the performance of the electrolytes prepared by the newly developed process is identical to that by electrolysis in terms of charge-discharge efficiency and capacity up to current density of 200 mA cm-2. This study can contribute to the wide spread of VRFB by providing a scalable process suitable for the mass production of V3.5+ electrolyte.


Introduction
Energy transition from fossil fuels to renewable resources is an important step to alleviate threats accompanied by climate change because most carbon dioxide, a representative greenhouse gas, is emitted in the course of electricity production 1 . Accordingly, solar and wind power plants have occupied increasing portions of distributed power generation in recent years 2-4 . It is clear this trend will be accelerated in the future, but nevertheless, many problems must be addressed in order for renewable energies to become the ultimate solution for electricity grid decarbonization 5 . Above all, the intrinsic intermittence of renewable energies makes it difficult to balance energy supply and demand, and may even destabilize the power grid 6,7 . Therefore, it is necessary to develop efficient and cost-effective energy storage systems suitable for various purposes 8 . Despite the active distribution of lithium ion batteries for stationary applications 9 , there remains an unyielding demand for development of better electrochemical energy storage systems.
The redox flow battery is one of the promising technologies for large scale and long duration stationary energy storage systems with improved safety 10 . The design and operation of the redox flow battery are flexible because the power and energy capacities can be decoupled. It has received a lot of attention because of its excellent scalability and relatively inexpensive operation and maintenance costs. As a result, various kinds of aqueous and non-aqueous flow batteries utilizing redox couples based on metal ions and organics have been reported 11,12 .
Among these, the all-vanadium redox flow battery (VRFB) is of particular interest because of its reduced occurrence of cross-contamination of electrolytes across the membrane through application of the same active species to positive and negative electrolytes 13, 14 . Another attractive point of VRFBs is their excellent sustainability 15 . Vanadium is an abundant element more commonly found in earth's crust than zinc or copper and the vanadium electrolyte is not consumed during operation, and thus, can be recycled 16 . Although significant progress toward commercialization of VRFBs has been made in the last three decades or so, further improvement is required, particularly from an economic point of view, to widen their acceptance 17 .
Compared with other key components constituting a VRFB stack such as ion exchange membranes and electrode materials, electrolytes have received relatively little attention 18, 19 . It is known that the portion of the electrolytes in the installation cost of VRFB is significant, and its importance increases as the power to energy ratio increases 20 . In the early stages of VRFB development, VOSO4 and V2O3 were used as raw materials for electrolytes, but it is now common to use V2O5 as a starting material, which is much cheaper and commercially available in large quantities. In addition, the recent trend is to prepare a so-called V 3.5+ electrolyte containing equimolar concentrations of VO 2+ and V 3+ ions and to use this as both the positive and negative electrolytes of the system at the same time. However, since V2O5 has insufficient solubility in aqueous sulfuric acid solution, the preparation of electrolyte must be preceded by reduction of VO2 + to VO 2+ in order to reach a vanadium concentration of 1.5 to 1.8 M. Various chemical reducing agents can be used in this step. On the other hand, there is no known chemical reducing agent effective in the reduction of VO 2+ to V 3+ under mild reaction conditions. Thus, the manufacture of V 3.5+ electrolyte almost entirely relies on electrolysis 21 .
The serious drawback of this process is that undesirable oxidation of VO 2+ occurs simultaneously at the anode during the reduction of VO 2+ at the cathode 22,23 . An additional process is required to chemically reduce the surplus product back into VO 2+ . According to recent techno-economic studies for VRFB, the manufacturing process cost accounts for 37 -50% of the electrolyte cost 24,25 . Therefore, it is obvious that improvement of the manufacturing process will have a great impact on the cost-effectiveness of the overall VRFB systems. For this purpose, achieving simplicity of the process is more important than anything else because costs for labor, initial investment and maintenance for equipment, as well as quality control of the manufactured electrolytes are expected to be cut down.
In this study, we develop a one-pot process using a chemical reducing agent to prepare a V 3.5+ electrolyte from V2O5. For a chemical reducing agent to be desirable, it is required not only to reduce VO2 + and VO 2+ to a lower oxidation state, but also to leave no residue in order to ensure the performance of the electrolyte. Hydrazine monohydrate is known to be oxidized in an acidic solution as follows 26 : It can be seen that N2 is the only by-product, which is inert and can be easily removed, and hydrazine monohydrate has an excellent atomic economy by emitting 4 electrons per molecule.
Surprisingly, while some patents have disclosed that hydrazine compounds could be used in the preparation of VRFB electrolytes 27,28 , detailed research has not yet been reported. This is unexpected considering that the hydrazine compound is often used as a reducing agent in the preparation of vanadium oxide nanoparticles 29, 30 . Moreover, the serendipitous discovery of the role of molybdenum in the reaction herein makes it possible to prepare V 3.5+ electrolytes at an acceptable reaction rate. Since the vanadium oxidation state (VOS) of the electrolyte can be controlled by the amount of reducing agent and the reaction rate by the molybdenum concentration, the process is concise and highly reproducible. The process developed here is suitable for mass-production of electrolytes with a consistent quality.

Results
One-pot preparation of V 3.5+ electrolytes.
It has been reported that the reduction of VO2 + to VO 2+ proceeds easily in acidic medium with various reducing agents such as oxalic acid 22 expected to be applicable to the reduction of VO2 + because it has a standard redox potential of -0.23 V in an acidic solution, which is much lower than that of VO2 + /VO 2+ (1.0 V vs standard hydrogen electrode). The color of V2O5 changed from yellow to dark gray immediately with moderate N2 evolution. This slight exothermic reaction seems to reflect the reduction of VO2 + to VO 2+ , but due to limited solubility of V2O4 in water, it was not possible to obtain a homogeneous solution even after prolonged agitation or temperature elevation. Fig. 1a shows the temperature change inside the reactor measured while adding 4 moles of concentrated sulfuric acid over a period of 1 h. Addition of sulfuric acid evoked vigorous N2 evolution and liberated considerable reaction heat in the first 25 min, while the reaction mixture became homogeneous. Roughly, the amount of sulfuric acid added up to this point was 1.7 moles, which is consistent with that required for formation of VOSO4. After that, a slight decrease in temperature was observed despite the continuous addition of sulfuric acid. UV-Vis spectra measured at 30 min after start of the addition of sulfuric acid indicated that all vanadium was converted to VO 2+ (Fig. 1b). It is worthy of note that three different V2O5 materials of industrial grade (99.5% min) were examined in these experiments, and no significant difference was observed until this point. In order to increase the temperature of the reaction mixture to 100 °C, external heating of the reactor was initiated. When the addition of sulfuric acid was completed, the VOS was checked again and the corresponding UV-vis spectra are shown in Fig. 1c. An interesting result that can be found in this figure is that the existence of V 3+ ions is evident in one of the three electrolytes tested 39-41 . That is, the absorbance of VO 2+ (760 nm) was decreased and that of V 3+ (401 nm) was appreciable in the case of V2O5-C. While maintaining the temperature inside the reactor at 100 ℃, the changes of VOS were further tracked by UV-Vis spectroscopy ( Fig. 1d). Although there was a significant difference in reaction rates, the appearance of V 3+ by the reduction of VO 2+ was obvious in all electrolytes. In one experiment (V2O5-C), it was even possible to produce the targeted V 3.5+ electrolyte after 10 h. To elucidate the reason for this encouraging observation, the V2O5 raw materials were analyzed by X-ray fluorescence spectroscopy and a clear relationship between the reaction rate of VO 2+ reduction and the molybdenum content could be found (Supplementary Table 1). As a validation experiment, 1 mol% of MoO3 (16 mmole) with respect to total vanadium was added to V2O5-A as a molybdenum source and the electrolyte was prepared with the same procedure described above. It was again confirmed that the V 3.5+ electrolyte could be prepared by a simple one-pot synthesis with hydrazine monohydrate as the sole reducing agent in the presence of molybdenum. In this case, only 2 h was needed to reach a VOS of 3.50 ( Fig. 1d and e).

Reaction mechanism of molybdenum-assisted electrolyte preparation.
Separate experiments were conducted to explain the role of molybdenum in the preparation of The oxidation of hydrazine monohydrate: The reduction of Mo 6+ to Mo 5+ : The oxidation and reduction of Mo 5+ /Mo 6+ and VO 2+ /V 3+ : Overall reaction: The effect of MoO3 on the one-pot preparation of V 3.5+ electrolytes was investigated more thoroughly, the results of which are represented in Fig. 2. The concentration of MoO3 in the electrolytes was varied from 1 mM to 16 mM. As can be found in Fig. 2a As shown in Fig. 2b, the reduction of VO 2+ with hydrazine monohydrate can be assumed to be a second-order reaction, and the reaction rate constant can be estimated from a linear relationship between the reaction time and the concentration of the reactant (Fig. 2c). The linear relationship was satisfactory until the conversion of hydrazine monohydrate reached approximately 70 to 80%. Fig. 2d shows that the molybdenum concentration in the electrolytes measured by ICP-OES was linearly proportional to the amount of MoO3 used in the reaction.
From this, we can understand that MoO3 is completely dissolved in the electrolyte and plays a role as a homogeneous catalyst, which is beneficial to secure the reproducibility of the electrolyte manufacturing process (Supplementary Fig. 3). Due to the simplicity of the process, scale-up of electrolyte production can be accomplished without many difficulties ( Supplementary Fig. 4). The effect of molybdenum on the physicochemical properties of the electrolytes, such as kinematic viscosity and electrical conductivity, was not appreciable (Supplementary Table 2). In addition, it was confirmed that all electrolytes had the target composition of 1.6 M vanadium and 4.0 M sulfate.

Electrochemical properties of electrolytes.
We investigated the electrochemical properties of electrolytes containing molybdenum using cyclic voltammetry (CV) and electrochemical impedance spectroscopy (EIS). There is a lot of controversy about the effect of impurities in electrolyte 48, 49 . It is not rare to find that an impurity showing a negative effect in one study led to an improvement in performance in another study.
High purity electrolytes may be desirable to ensure long-term durability of VRFB systems.
However, if one considers the economics of the entire system, it is recommended to identify and distinguish between impurities that have a relatively insignificant effect and those that cause fatal performance degradation 50 . Accordingly, the influence of molybdenum on the electrochemical performance of electrolytes should be carefully evaluated.
CV was performed at scan rates ranging from 2 mV s -1 to 30 mV s -1 (Supplementary Fig. 5).
As an example, voltammograms of the positive (VO 2+ /VO2 + ) and negative (V 2+ /V 3+ ) electrolytes measured at a scan rate of 5 mV s -1 are shown in Fig. 3a and b, respectively. The results for electrolyte prepared by chemical reduction with oxalic acid and electrolysis (Ox-EL) were also included for comparison. In the case of positive electrolytes, peak potential separation (ΔEp = Ec -Ea) and anodic and cathodic peak currents (ia and ic) showed little difference between each electrolyte when the molybdenum concentration was lower than 4 mM (Fig. 3a). Increased peak separation and decreased peak currents were observed with increasing molybdenum. On the other hand, for negative electrolytes, the effect of molybdenum was more appreciable. As the molybdenum concentration increased from 1 mM to 16 mM, ΔEp increases from 0.23 V to 0.35 V, and ia and ic decreases from 2.17 mA to 1.40 mA and -2.41 mA to -1.61 mA, respectively, were observed (Fig. 3b). The effect of molybdenum on the electrochemical performance was further examined by CV at different scan rates. From Fig. 3c and d, an increase in ΔEp with increasing scan rate can be found, which is consistent with many studies showing that the redox reaction of vanadium ions on the electrode is a quasi-reversible reaction 51-53 . As the scan rate increases, ΔEp increases because a larger overpotential is required due to the enhanced polarization effect, which suggests that the reaction kinetics come into competition with mass transfer. In addition, we can find that ia and ic deviated from linearity in plots of these versus the square root of the scan rate for both positive (Fig. 3e) and negative ( Fig. 3f) electrolytes, especially with increase of the scan rate or amount of molybdenum in the electrolytes, which provides additional evidence of quasi-reversibility of the redox reaction.
According to the aforementioned CV results, molybdenum has a larger effect on the reversibility of the redox reaction of negative electrolytes than positive electrolytes.
Nevertheless, it is worthwhile to mention that the CV behavior of Ox-EL containing a negligible amount of molybdenum (1.9 ppm) is almost the same as electrolytes prepared by molybdenum-catalyzed one-pot synthesis. For negative electrolyte, similar values of ΔEp, ia and ic with electrolyte containing 16 mM of molybdenum were observed. mM, but the radius of the semicircle increased slightly at 8 mM, with a more evident increase at 16 mM (Fig. 4a). However, the slope of the linear line of the low frequency region was independent of the molybdenum concentration. In contrast, for negative electrolytes, the radius of the semicircle increased consistently according to the molybdenum concentration (Fig. 4b).
Therefore, charge transfer between electrode and electrolyte may be influenced to some extent by molybdenum in these cases. to Ox-EL when the molybdenum concentration was increased to 8 mM and 16 mM, respectively (Fig. 4c). A more pronounced effect of molybdenum can be observed in Fig. 4d, in which estimated Rct values for negative electrolytes are presented. Compared to electrolytes containing less than 2 mM of molybdenum, a more than 2-fold increase of Rct could be found when 16 mM of molybdenum was added to the electrolyte. It was also found that the increase of Rct was linearly proportional to the molybdenum concentration. Other parameters provided by the equivalent circuit, CPE and Warburg coefficients are listed in Supplementary Table 3.

VRFB single cell performance.
In order to examine the performance of electrolytes, a charge/discharge cycle test was conducted using a VRFB cell at different current densities of 50, 100, 150 and 200 mA cm -2 .
For each current density, five charge/discharge cycles were carried out as shown in Supplementary Fig. 6, and the charge and discharge voltages of the fifth cycle versus capacity are represented in Fig. 5 for comparison. The performance of all electrolytes seemed to be identical in the viewpoint of overpotential and charge/discharge capacities when the current density was not too high. Supplementary Fig. 7 shows the average coulombic (CE), voltage (VE), and energy (EE) efficiencies of five charge/discharge cycles at different current densities.
CE usually increases with current density since the charging/discharging occurs at a faster rate, and thus, the permeation of ions through the separator decreases. CE reflects the decrease in capacity caused by the crossover of vanadium ions, which is mainly governed by the physical properties of the ion exchange membrane 55 , and may not be influenced much by the existence of molybdenum in electrolytes. Therefore, it can be expected that all electrolytes exhibit similar performance in terms of CE. VE is a factor that reflects the resistance of the electrode 56 . As the current density increased, a decrease in VE due to an increase in overpotential was observed for all electrolytes. However, when comparing VE at the same current density, it is difficult to find a consistent trend resulting from the molybdenum contained in the electrolyte. All electrolytes showed a similar discharge energy, as shown in Supplementary Fig. 8. The longcycle performance of V 3.5+ electrolytes was evaluated by carrying out 200 cyclecharge/discharge experiments at the current density of 150 mA cm -2 . There was a slight variation in EE in the early stage of charge/discharge cycles, but eventually, all electrolytes revealed the same efficiencies (Fig. 5e). In addition, the discharge capacity retention, which is defined as the percentage of discharge capacity of the Nth cycle compared to that of first cycle, showed similar tendency for all electrolytes (Fig. 5f). Accordingly, it can be concluded that molybdenum provides little effect on the quality of the electrolyte up to 16 mM, the amount considered in this study.
Recently, it has been reported that MoO3 deposited on carbon electrodes can act as an electrocatalyst enhancing the reversibility of VO 2+ /VO2 + and V 2+ /V 3+ redox couples 57, 58 . This was responsible for the improvement of voltage efficiency of the VRFB cell. In one study 57 , it was also shown that addition of 5 mM of Na2MoO4 directly to the electrolytes modified the surface of the electrodes during cycling and produced similar effects on VRFB cell performance as MoO3 deposited on electrodes. In our experiments, the deposition of molybdenum compounds after cycling was not observed, and the dissolved molybdenum appeared to slightly worsen the reversibility of redox couples, though not seriously. Therefore, the enhancement of VRFB cell performance via electrode modification with MoO3 particles observed in the literature may be attributed to the increased surface wettability of the electrodes more than to electrocatalytic effects, for which further studies are required to understand this controversial phenomenon.

Discussion
In this study, we have reported the simplest process ever described for the preparation of high performance V 3.5+ electrolyte. It was found that hydrazine monohydrate is quite effective not only at the reductive dissolution of V2O5, but also at the reduction of VO 2+ to V 3+ with the assistance of molybdenum. Consequently, it is possible to prepare V 3.5+ electrolyte by a onestep process without using a complex electrolysis process. The molybdenum concentration in the electrolytes is coincident with the amount of MoO3 used as a catalyst. Molybdenum slightly deteriorates the reversibility of redox reaction of the negative electrolyte by increasing charge transfer resistance, but this seems not to be a critical demerit according to the charge/discharge test conducted using a VRFB single cell. Conversely, the electrochemical performance of positive electrolytes was found to be less sensitive to the molybdenum concentration.
When considering the mass production of V 3.5+ electrolytes, the current one-pot process is attractive because only one jacketed reaction vessel equipped with an agitator and a condenser is necessary, which is common equipment that can be found in ordinary chemical plants.
Moreover, no consumable materials are necessary in the process developed herein. In contrast, electrolysis requires regular replacement of electrode materials, and catalytic processes consume expensive Pt catalyst 35 . Impurities in V2O5 depend on the source of the vanadium and the manufacturing process. At present, considerable amounts of V2O5 are recycled from spent hydrodesulfurization catalysts 59 . In this case, 50 ~ 150 ppm of molybdenum is usually contained in V2O5 raw materials. Therefore, no additional MoO3 is actually required in order to prepare V 3.5+ electrolyte from V2O5. All aforementioned aspects are favorable for low-cost V 3.5+ electrolyte production (Supplementary Note 1).

Preparation of electrolytes.
A series of V 3.5+ electrolytes were prepared in the presence of MoO3 as a catalyst. Total Electrochemical analysis of electrolytes.
The electrochemical performance of electrolytes was evaluated using an electrochemical and negative (V 2+ /V 3+ ) electrolytes were prepared by mixing VO2 + and V 2+ electrolytes at a volume ratio of 5:1 or 1:5, respectively. Cyclic voltammograms were obtained at different scan rates. At least three scans were carried out for each experiment and the last scan was taken.
Electrochemical impedance spectroscopy measurements were carried out at open-circuit potential with an a.c. amplitude of 10 mV from 20 kHz to 0.2 Hz, and data were analyzed using a Randles equivalent circuit so as to calculate charge transfer resistances.

Data availability
The authors declare that data supporting the findings of this study are available within the article and its Supplementary Information and also are available from the corresponding author upon reasonable request.       Figure 1 Electrolyte preparation using hydrazine monohydrate as a chemical reducing agent. a Temperature pro les during the addition of H2SO4. b, c UV-Vis spectra measured after addition of half of (b) and the whole amount of (c) H2SO4. d VOS of electrolytes prepared from different V2O5 raw materials with or without MoO3 as a function of reaction time. e UV-Vis spectra of the electrolyte prepared from V2O5-A and 16 mM of MoO3 at different reaction times at 100 °C.  Cyclic voltammograms (CV) of electrolytes. a, b CVs measured at a scan rate of 5 mV s-1 for V4.5+ (a) and V2.5+ (b) electrolytes. c, d Peak potential separation (ΔEp) versus scan rate for V4.5+ (c) and V2.5+